Datasets:

florentgbelidji

/

ncbi_extracted_running

like 0

PMC3868388

Environment and Scheduling Effects on Sprint and Middle Distance Running Performances Amal Haı¨da1,2*, Fre´de´ric Dor1, Marion Guillaume1, Laurent Quinquis1, Andy Marc1, Laurie-Anne Marquet1, Juliana Antero-Jacquemin1, Claire Tourny-Chollet2, Franc¸ois Desgorces1,3, Geoffroy Berthelot1,3, Jean-Franc¸ois Toussaint1,3,4 1 IRMES (biomedical Research Institute of Sports Epidemiology, Paris, France), INSEP (Institut National du Sport de l’Expertise et de la Performance), Paris, France, 2 Universite´ de Rouen, Faculte´ des Sciences du Sport et de l’Education Physique CETAPS (Centre d’Etude Transformations des Activite´s Physiques et Sportives), Mont- Saint-Aignan, France, 3 Paris Descartes University, Sorbonne, Paris Cite´, Paris, France, 4 Hoˆtel-Dieu Hospital, CIMS (Centre d’Investigations en Me´decine du Sport), AP-HP (Assistance Publique-Hoˆpitaux de Paris), Paris, France Abstract Purpose: Achievement of athletes’ performances is related to several factors including physiological, environmental and institutional cycles where physical characteristics are involved. The objective of this study is to analyse the performance achieved in professional sprint and middle-distance running events (100 m to 1500 m) depending on the organization of the annual calendar of track events and their environmental conditions. Methods: From 2002 to 2008, all performances of the Top 50 international athletes in the 100 m to 1500 m races (men and women) are collected. The historical series of world records and the 10 best annual performances in these events, amounted to a total of 26,544 performances, are also included in the study. Results: Two periods with a higher frequency of peak performances are observed. The first peak occurs around the 27.15th 60.21 week (first week of July) and the second peak around 34.75th 60.14 week (fourth week of August). The second peak tends to be the time of major international competitions (Olympic Games, World Championships, and European Championships) and could be characterized as an institutional moment. The first one, however, corresponds to an environmental optimum as measured by the narrowing of the temperature range at the highest performance around 23.2563.26uC. Conclusions: This is the first study to demonstrate that there are two performance peaks at a specific time of year (27th and 34th weeks) in sprint and middle distance. Both institutional and ecophysiological aspects contribute to performance in the 100 m to 1500 m best performances and define the contours of human possibilities. Sport institutions may take this into account in order to provide ideal conditions to improve the next records. Citation: Haı¨da A, Dor F, Guillaume M, Quinquis L, Marc A, et al. (2013) Environment and Scheduling Effects on Sprint and Middle Distance Running Performances. PLoS ONE 8(11): e79548. doi:10.1371/journal.pone.0079548 Editor: Alejandro Lucia, Universidad Europea de Madrid, Spain Received July 19, 2013; Accepted September 23, 2013; Published November 20, 2013 Copyright:  2013 Haı¨da et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: The authors thank the Centre National de De´veloppement du sport and the Ministry of Health, Youth and Sports for financial assistance. The funders had no role in study design, data gathering, data collection and analysis, decision to publish, or preparation of the manuscript. The corresponding author had full access to all data in the study and had final responsibility for the decision to submit for publication. Competing Interests: The authors have declared that no competing interests exist. * E-mail: amal.haida@insep.fr Introduction Since the beginning of the modern Olympic era (1896), the best performance (BP) are in a process of exponential growth which now seems to have reached its limits [1]. Performance is often understood as a very broad term which involves many components such as : psychomotor abilities, flexibility and joint stiffness, muscle strength and power [2]. Athletes, like any living organism, depend on physiological regulations that respect the nycthemeral, seasonal or vital cycles [3]. There are variations in physiological factors such as maximum oxygen consumption (VO2 max) or concentra- tions of melatonin on the basis of seasonal rhythms [4]. This seasonal rhythmicity has been demonstrated for certain factors such as mood [5], lung function [6] and the core body temperature. It is also observed in the physical activity of the general population, which tends to be higher during summer [7– 9]. Chan and colleagues (2006) [9] found a significant change in physical activity in the general population, as number of steps walked per day being related to temperature, precipitation and wind speed. There is a limited amount of research that has investigated effects of seasonality in sports on sprint athletes’ performances. Yet the annual schedule of events seems to be a contributing factor to performance. Comparison of track and field world records (WR) shows that performance prevails in summertime. The influence of environmental parameters on physiology (ecophysiology) partly determines the evolution of human performance [10,11]. Marathon optimal performances are set at a temperature around 10u. This performance dependency on PLOS ONE | www.plosone.org 1 November 2013 | Volume 8 | Issue 11 | e79548 temperature occurs not only for elite-standard athletes but for all participants also [12–14]. The objective of this study is to compare the date and temperature of the BP in sprint and middle distance races (100 m to 1500 m) for men and women during the annual calendar of international competitions, and observe their evolution over the Olympic era in order to assess the environment and scheduling effects on sprint and middle distance running performances. Methods Data Collection From 2002 to 2008, all performances of the Top 50 international athletes in running events ranging from 100 m to 1500 m races for men and women were collected from the official website of the International Association of Athletics Federations (IAAF) [15]. For each event, data collection includes: full name of the athlete, the completion date and place of the competition: 23,746 performances are collected, 11,813 from males and 11,933 from females. Performances are divided into five categories defining the performance as a percentage in relation to the BP obtained at the event. The percent categories (PC) used were: [95– 96%], [96–97%[, [97–98%[, [98–99%], [99–100%]. Within each group, performances were collected according to the competition type: (i) major competitions (the Olympic Games (OG), World Championships (WC), European Championships (EC) and Amer- ican selections (US)), (ii) the international circuit represented by the Golden League and (iii) other meetings. Date, place and name of the athlete when WR were set for the same distances between 1952 and 2010 are collected (181 WR) and the performances of the top 10 male and female 100 m race are gathered from 1891 to 2008, representing 2,617 performances. Temperatures for each city, at the time of the competition, are recorded from 97 to 100 PC in 100 m, 200 m, 400 m, 800 m and 1500 m. In order to improve resolution, half PC are defined to study temperature density: [97–97.5%], [97.5–98%], [98–98.5%], [98.5–99%], [99–99.5%], [99.5–100%]. Temperature data are collected from the weather underground website [16]. The total number of performances collected for this study is 26,544. Statistical Analysis Distribution of performance by PC. The performance data from 2002 to 2008 is based on the distribution of performance per week of the year depending on the PC and the type of competition: mathematical analysis and modeling are done using Matlab. To estimate the two dates when the greatest numbers of performances occur, two functions are adjusted using the least squares method: the double Gaussian and double Lorentzian functions. For each PC, the best-fitted function is selected on the basis of adjusted R2 and the mean square error (RMSE) (See Materials and Methods S1, Figure S1, Tables S1 and S2). The two dates of peak performance are estimated using the elected model for each PC (Inert Figure 1). The proportion of performances in the two peaks is estimated by computing the area under the curve (proportion of performances) of each elected model and for each PC (See Materials and Methods S1, Figure S2, Table S2). Distribution of WR Effect size for One-Way ANOVA is Cohen’s d and is evaluated with Cohen’s conventional criteria [17]. It is used to study the stability of the WR mean date by decades (1952–1959, 1960– 1969, 1970–1979, 1980–1989, 1990–1999, 2000–2010). Statistical significance is considered at p,0.05. Temperature For the temperature, the statistical analyses are done on R, Version 3.0.0 (R Core Team, Vienna, Austria, 2013) and results are expressed as a mean 6 standard deviation. Fisher test is used to compare the dispersions between the different PC with a value of p,0.05 considered significant. We estimate the density of temperature degrees for each of the PC over a homogeneous mesh of 5*6 nodes. The resolution used is of 7.5uC in the x-axis (temperature) and 0.59 percent in the y-axis PC. Results Distribution of Performance by PC and Competition The distribution of weekly performances for each PC (95 to 100 PC) over the competition calendar shows two high frequency periods (Figure 1). The estimated dates of the two peak performances are constant within all PC (on mean 27.15th 60.21 week for peak 1 and 34.75th 60.14 for peak 2) (Figure 1, Inset). Reaching the highest level, the areas under the curves of both peaks converge toward the same 50% value (See Figure S2). The number of performances during major competitions (OG/ WC/US/EC) increases from 16.7% for the 95 PC to 25.7% for the 99 PC. The performances recorded during the Golden League increase from 7.7% for the 95 PC to 29.1% for the 99 PC. Conversely, the number of performances in the other competitions decreases from 75.6% to 45.1% (Figure 1) (See Table S3). Distribution of WR The mean distribution of WR date by decade from 1952 to 2010 is concentrated at the 206.09th day 646.17. The variability of WR date decreases considerably. In the first period (1952– 1959), SD is 64.34, in the last period (2000–2010), SD is 35.09. However the mean day remains stable throughout the period (p = 0.29) (Figure 2), with a large effect size (d = 0.98). Influence of Temperature on Performance The analysis of the distribution of PC according to temperature shows a restriction in the thermal interval when reaching the highest performance level. This interval narrows from 10–32uC at 97 PC of the BP to 20–27uC for the 100 PC with a mean temperature of 23.2364.75uC. Subdividing the data into PC, the mean temperature is 23.1364.80uC for the PC [97 to 97.5[, 23.4964.88uC for the PC [97.5 to 98[, 23.2364.92uC for the PC [98 to 98.5[, 22.8964.56uC for the PC [98.5 to 99[, 22.6363.72uC for the PC [99 to 99.5[and 23.2563.26uC for the BP [99.5 to 100]. Figure 3 highlights the narrowing of the temperature range at the BP interval. The peak value of the density mesh is 362 temperature values at 23uC and at 97.59%. The density decreases in both dimensions (temperature and PC) from this point confirming the mean temperature value stated above (Inset, Figure 3). Top 10 Sprinters from 1891 to 2008 There is no evolutionary trend in the completion date on the 100 m performances throughout the modern Olympic era. Since 1891, men accomplish their best performance around July 10th (650 days) id est during the 28th week and since 1921, women perform best around July 20th (644 days) id est during the 29th week (Figure 4). Environment and Scheduling in Track Races PLOS ONE | www.plosone.org 2 November 2013 | Volume 8 | Issue 11 | e79548 Discussion Our study is the first to our knowledge to analyze the exhaustiveness of the best performers in sprint and middle distance races in relation to temperature. Previous studies have mostly analysed seasonality in the rhythms of daily life [2,18] or in marathon runners [12] but no studies have demonstrated effects of seasonality through environmental or institutional conditions on performance in sprint. Two yearly performance peaks are observed for all levels in this study. The first peak corresponds to the 27th week of the year (first week of July) suggesting an environmental optimum for sprint events. The second peak occurs at the 34th week (fourth week of August), which is related to the main sporting events such as: Olympic Games, European and World Championships. As seen in Figure 1, both peak dates are stable throughout all performance categories. Cultural Peak at the 34th Week The impact of major international competitions corresponds to the performance peak in August. The calendar scheduling for world championships or Olympic Games can be considered as an institutional attractor. IAAF hosts competitions taking place outdoor between February and October. Major competitions such as the World Championships, European championships and Olympic Games are usually scheduled in August whereas the international circuit of the Golden League covers the whole period between June and September. National federations plan their own schedules proposing competitions that allow their athletes to qualify for the major competitions. This study highlights the existence of a cultural peak (second peak) occurring at the same times as the major international events. Globally, top athletes focus on the same goal: to be the most physically and mentally fit for this time of the year (Figure 1). This second peak corresponds to the athlete’s own planning for major competitions, which is a result of long term training, Figure 1. Number of performances per week and per percent category (PC) by (i) major competitions (Olympic Games (OG), World Championships (WC), American selections (US), European championships (EC)), (ii) Golden League and the others meetings (iii). INSET: Dates (week) of peaks performance modeling by PC. doi:10.1371/journal.pone.0079548.g001 Figure 2. A. Distribution of world records (WR) date (day) in 100 m, 200 m, 400 m, 800 m and 1500 m running events from 1952 to 2010. B. Mean distribution of world records (WR) date (day) in 100 m, 200 m, 400 m, 800 m and 1500 m running events by decade from 1952 to 2010. The mean date is: 206,09 th day. doi:10.1371/journal.pone.0079548.g002 Environment and Scheduling in Track Races PLOS ONE | www.plosone.org 3 November 2013 | Volume 8 | Issue 11 | e79548 technical analysis, strategic choice, awareness of physical and psychological limits [19–21]. Although, training and preparation are essential to reach a BP at a specific moment, environmental factors will allow the achievement of the highest level of performance. Thermal Peak at the 27th Week The analysis of WR (Figure 2) and the top 10 BP in 100 m sprint (men and women) illustrates this first peak (Figure 4). Numerous studies have demonstrated effects of environmental conditions on the performance of marathon runners [22,23]. Marathon requires a number of appropriate environmental conditions for thermoregulation of any runner, elite or amateur. The humidity, barometric pressure, dew point, and temperature are all essential in the quest of achieving optimal performance [24]. A recent study analysed the impact of environmental param- eters on the performance of marathon running. It established a distribution of performances depending on temperature, observed regardless of the athlete’s level. This distribution function defines the field limits of the human possibilities [12]. The impact of temperature and season on biological parameters is largely documented in the literature [24–26]. In this present study, the results show a distribution for top performance in sprint and middle-sprint where the effective temperature range decreases with performance level (10–32uC at the bottom (97 PC of the BP); 20–27uC at the top (100 PC of the BP)) (Figure 3). Competitions are mainly organized in the northern hemisphere. The range of temperatures collected from the different host cities was large: ranging from 10 to 38uC but the mean temperature when achieving the BP is 23.2364.75uC. The standard deviations decrease progressively with increasing level, but all categories remain centered on the 23.23uC value. This suggests a very regulated process at all performance levels. The effects of temperature on biological parameters. All biological structures and processes (human or not) are affected by temperature in thermodynamical regulations [27]. Perfor- mance depends on physiological responses to exercise perfor- mance in an interaction between body temperature and environmental temperature [26]. Performance decreases pro- gressively as the environmental heat stress increases [25]. As with other biological rate processes, muscle function is strongly influenced by temperature. Specifically, muscle contraction rates (the rates of both force development and relaxation) are accelerated by an increase in temperature in both invertebrates and vertebrates [28,29]. Fundamental biological functions like metabolic activity synchronize with the rhythmic phases of environmental change such as temperature. For gradually intensity increasing aerobic exercise the plasma concentration of certain ions (K+, Ca2+) and lactic acids appear differently when muscular exercise takes place at thermal neutrality (21uC) in comparison to exercise performed at 0uC [30]. At the favorable season, body temperature and metabolic rates increase and so does growth rate. Mammal growth depends on Figure 3. Percent category (PC) depending on temperature: comparison of temperature at different level from 97 PC to 100 PC in 100 m, 200 m, 400 m, 800 m and 1500 m. Respectively, in each half PC the mean are 23.12uC, 23.49uC, 23.23uC, 22.89uC, 22.63uC and 23.25uC and the median are 23.00uC, 23.00uC, 22.00uC, 22.00uC, 21.00uC and 22.50uC. INSET: Temperature density (ie. number of recorded temperatures) per PC computed over a mesh. The maximal density is computed at 23uC and 97.59% and progressively decreases as PC increase (due to the decrease in performance number). The density decreases as temperature increases or decreases from the maximal density (due to the effect of temperature on performance). doi:10.1371/journal.pone.0079548.g003 Environment and Scheduling in Track Races PLOS ONE | www.plosone.org 4 November 2013 | Volume 8 | Issue 11 | e79548 Figure 4. Relation between day of the performance and year in men and women. A. Average day of the achievement of the performance in the top 10 at the 100 m men since 1891. For all years combined, the average day is the 192.76th 649.77. B. Average day of the achievement of the performance in the top 10 at the 100 m women since 1921. For all years combined, the average day is the 202.52th644.0. doi:10.1371/journal.pone.0079548.g004 Environment and Scheduling in Track Races PLOS ONE | www.plosone.org 5 November 2013 | Volume 8 | Issue 11 | e79548 seasonal variation even for their bones structure [31]. Climates and seasons have a marked influence on human biology [18] including mental abilities [32], sexual activity [33] or territorial conflicts [34]. Temperature and mortality. Many chronobiological health aspects depend on season and temperature cycles. Affective disorders show a predictable onset in the fall/winter months and, reversely, a reduction in the spring/summer period [32]. Large-scale population studies have shown seasonal variations in mortality rates in different parts of the world peaking during the cold winter months [35,36]. Relations between mortality and cardiovascular disease (CVD) in the winter months have been reported for many countries and might be partly explained by seasonal changes of risk factors. Cardiac death also depend on the season even after adjustment for age, cholesterol, blood pressure, and body mass index [36,37]. Several studies have reported the existence of optimal ranges of air temperatures [38,39]. Specifi- cally, cold weather has been reported to be associated with increased risk of death from cardiovascular causes and respiratory infections [39–41]. The mortality rate is lower on days in which the maximum temperatures range between 20–25uC [38]. This means that survival rate is highest at this temperature range. Our results show a mean temperature of 23.2364.75uC for the BP which is converging with the temperature of the lowest mortality rates. Therefore, both survival capability and physiolog- ical capacities of the human are optimal at 20–25uC. The two peaks of performance change its distribution in function of the performance level (See Figures S1 and S2). However, the first peak which corresponds to a ‘‘thermal peak’’ persists even at the highest level of competition. This demonstrates that despite the presence of an institutional attractor, represented by the major competitions, the environmental attractor remains omnipresent with an ideal temperature period for maximal performance. The adequacy of the thermal peak is as important as the cultural peak at the highest level. Conclusion The range of possible combinations of environmental and institutional components is narrowed for the top performers. For sprint and middle-sprint races, when progressing toward the highest levels of performance, the importance of the institutional component regularly increases with a balanced effect for the top performers. The novelty of this study is that, environmental conditions must be taken into account in order to achieve maximal speed. This field of possibilities reveals an ideal temperature to achieve optimal performance. Calendars for major competitions should take this into account in order to increase the probability of breaking the next records. Supporting Information Figure S1 Date of peak performance modeling for 95% to 99% categories. The Double Lorentzian (continuous line) and Double Gaussian functions (broken line) are adjusted to each percent category. Although the two models differ in the estimate of the tails, they roughly provide the same estimates of x01 and x02 (maximum difference is around 0.5 week). (EPS) Figure S2 Area under the curve (AUC) for the elected functions and both peaks in each PC. Equations (10) and (11) are used to estimate p1 and p2. The AUC of both peaks converge toward a unique value as the PC increase (50%). The proportion of performances: The estimation of the area under the curve for the two peaks shows that, when increasing the PC, the proportion of performances in each peak progressively converge to the same value, from 93.67% (peak 1) vs. 6.33% (peak 2) for PC = 95% to 50.65% (peak1) vs. 49.35% (peak2) with PC = 99% (Figure S2). (EPS) Table S1 Statistics of the two models. For each percent category, the adjusted R2, rMSE and sse are given. Statistics of the elected function are mentioned in bold. (DOC) Table S2 Results of the two models. For percent category and model, the estimated date of peak (x01, x02), value of peak (f(x01), f(x02)), the total proportion of performance (area under the curve), and p1, p2 are given. Results of elected function are given in bold. (DOC) Table S3 Number of performances per depending on the type of competition and the percent category. N is the number for different percent category (PC) and its equivalent percentage. On the overall performance analyzed, 2,347 were conducted during major competitions and 2,093 during the Golden League. The other 8,079 performances were done in other competitions (OG: Olympic Games; WC: World championships; US: American selections; EC: European Championships). (DOC) Materials and Methods S1 The two models (double Gaussian and double Lorentzian functions) are present- ed. The methods for estimating the dates of the two peaks and the area under the curve are described. (PDF) Acknowledgments We thank the Centre National de De´veloppement du sport and the Ministry of Health, Youth and Sports for financial assistance. We thank the National Institute of Sports Expertise and Performance teams for their full support. We thank Mrs Katrine Okholm Kryger and Ms Maya Dorsey for their helpful comments for their critical reading of the Article. Author Contributions Conceived and designed the experiments: JFT AH F. Dor GB LQ F. Desgorces. Analyzed the data: F. Dor MG GB AH AM. Wrote the paper: AH F. Dor GB JFT. Reviewed the paper: KOK MD LAM JAJ CTC JFT. References 1. Berthelot G, Thibault V, Tafflet M, Escolano S, El Helou N, et al. (2008) The citius end: world records progression announces the completion of a brief ultra- physiological quest. PLoS ONE 3: e1552. doi:10.1371/journal.pone.0001552. 2. Atkinson G, Drust B (2005) Seasonal rhythms and exercise. Clin Sports Med 24: e25–34, xii–xiii. doi:10.1016/j.csm.2004.11.001. 3. Magnanou E, Attia J, Fons R, Boeuf G, Falcon J (2009) The timing of the shrew: continuous melatonin treatment maintains youthful rhythmic activity in aging Crocidura russula. PLoS ONE 4: e5904. doi:10.1371/journal.pone.0005904. 4. Lincoln GA, Johnston JD, Andersson H, Wagner G, Hazlerigg DG (2005) Photorefractoriness in mammals: dissociating a seasonal timer from the circadian-based photoperiod response. Endocrinology 146: 3782–3790. doi:10.1210/en.2005-0132. 5. Boivin DB, Czeisler CA, Dijk DJ, Duffy JF, Folkard S, et al. (1997) Complex interaction of the sleep-wake cycle and circadian phase modulates mood in healthy subjects. Arch Gen Psychiatry 54: 145–152. Environment and Scheduling in Track Races PLOS ONE | www.plosone.org 6 November 2013 | Volume 8 | Issue 11 | e79548 6. Spengler CM, Shea SA (2000) Endogenous circadian rhythm of pulmonary function in healthy humans. Am J Respir Crit Care Med 162: 1038–1046. 7. McCormack GR, Friedenreich C, Shiell A, Giles-Corti B, Doyle-Baker PK (2010) Sex- and age-specific seasonal variations in physical activity among adults. J Epidemiol Community Health 64: 1010–1016. doi:10.1136/jech.2009.092841. 8. Hamilton SL, Clemes SA, Griffiths PL (2008) UK adults exhibit higher step counts in summer compared to winter months. Ann Hum Biol 35: 154–169. doi:10.1080/03014460801908058. 9. Chan CB, Ryan DAJ, Tudor-Locke C (2006) Relationship between objective measures of physical activity and weather: a longitudinal study. Int J Behav Nutr Phys Act 3: 21. doi:10.1186/1479-5868-3-21. 10. Desgorces F-D, Berthelot G, El Helou N, Thibault V, Guillaume M, et al. (2008) From Oxford to Hawaii ecophysiological barriers limit human progression in ten sport monuments. PLoS ONE 3: e3653. doi:10.1371/journal.pone.0003653. 11. Marino FE, Lambert MI, Noakes TD (2004) Superior performance of African runners in warm humid but not in cool environmental conditions. J Appl Physiol 96: 124–130. doi:10.1152/japplphysiol.00582.2003. 12. El Helou N, Tafflet M, Berthelot G, Tolaini J, Marc A, et al. (2012) Impact of environmental parameters on marathon running performance. PLoS ONE 7: e37407. doi:10.1371/journal.pone.0037407. 13. Cheuvront SN, Haymes EM (2001) Thermoregulation and marathon running: biological and environmental influences. Sports Med 31: 743–762. 14. Kenefick RW, Cheuvront SN, Sawka MN (2007) Thermoregulatory function during the marathon. Sports Med 37: 312–315. 15. International Association of Athletics Federations. Available: http://www.iaaf. org/. Accessed 26 September 2012. 16. Weather Underground website. Internet weather service. Available: http:// www.wunderground.com/history/. Accessed 26 September 2012. 17. Field A (2009) Discovering Statistics Using SPSS. SAGE Publications. 857 p. 18. Foster RG, Roenneberg T (2008) Human responses to the geophysical daily, annual and lunar cycles. Curr Biol 18: R784–R794. doi:10.1016/ j.cub.2008.07.003. 19. Steinacker JM, Lormes W, Kellmann M, Liu Y, Reissnecker S, et al. (2000) Training of junior rowers before world championships. Effects on performance, mood state and selected hormonal and metabolic responses. J Sports Med Phys Fitness 40: 327–335. 20. Costill DL, Thomas R, Robergs RA, Pascoe D, Lambert C, et al. (1991) Adaptations to swimming training: influence of training volume. Med Sci Sports Exerc 23: 371–377. 21. Mujika I (2010) Intense training: the key to optimal performance before and during the taper. Scand J Med Sci Sports 20 Suppl 2: 24–31. doi:10.1111/ j.1600-0838.2010.01189.x. 22. Ely MR, Martin DE, Cheuvront SN, Montain SJ (2008) Effect of ambient temperature on marathon pacing is dependent on runner ability. Med Sci Sports Exerc 40: 1675–1680. doi:10.1249/MSS.0b013e3181788da9. 23. Montain SJ, Ely MR, Cheuvront SN (2007) Marathon performance in thermally stressing conditions. Sports Med 37: 320–323. 24. Ely MR, Cheuvront SN, Roberts WO, Montain SJ (2007) Impact of weather on marathon-running performance. Med Sci Sports Exerc 39: 487–493. doi:10.1249/mss.0b013e31802d3aba. 25. Galloway SD, Maughan RJ (1997) Effects of ambient temperature on the capacity to perform prolonged cycle exercise in man. Med Sci Sports Exerc 29: 1240–1249. 26. Atkinson G, Reilly T (1996) Circadian variation in sports performance. Sports Med 21: 292–312. 27. Somero GN (2002) Thermal physiology and vertical zonation of intertidal animals: optima, limits, and costs of living. Integr Comp Biol 42: 780–789. doi:10.1093/icb/42.4.780. 28. Bennett AF (1985) Temperature and muscle. J Exp Biol 115: 333–344. 29. Josephson KR (1984) Contraction dynamics of flight and stridulatory muscles of tettigoniid insects. J Exp Biol 108: 77–96. 30. Therminarias A, Quirion A, Pellerei E (1987) Effets de l’exposition au froid sur les variations plasmatiques du lactate et des ions induites par un exercice musculaire ae´robie intense. Science & Sports 2: 1–8. doi:10.1016/S0765- 1597(87)80039-4. 31. Koehler K, Braun H, Achtzehn S, Hildebrand U, Predel H-G, et al. (2012) Iron status in elite young athletes: gender-dependent influences of diet and exercise. Eur J Appl Physiol 112: 513–523. doi:10.1007/s00421-011-2002-4. 32. Magnusson A (2000) An overview of epidemiological studies on seasonal affective disorder. Acta Psychiatr Scand 101: 176–184. 33. Re´gnier-Loilier A, Rohrbasser J-M (2011) Y-a-t-il une saison pour faire des enfants? Population & socie´te´s: 1–4. 34. Hsiang SM, Meng KC, Cane MA (2011) Civil conflicts are associated with the global climate. Nature 476: 438–441. doi:10.1038/nature10311. 35. Healy JD (2003) Excess winter mortality in Europe: a cross country analysis identifying key risk factors. J Epidemiol Community Health 57: 784–789. 36. The Eurowinter Group (1997) Cold exposure and winter mortality from ischaemic heart disease, cerebrovascular disease, respiratory disease, and all causes in warm and cold regions of Europe. Lancet 349: 1341–1346. doi:10.1016/S0140-6736(96)12338-2. 37. Ghebre MA, Wannamethee SG, Rumley A, Whincup PH, Lowe GDO, et al. (2012) Prospective study of seasonal patterns in hemostatic factors in older men and their relation to excess winter coronary heart disease deaths. J Thromb Haemost 10: 352–358. doi:10.1111/j.1538-7836.2012.04617.x. 38. Ballester F, Corella D, Pe´rez-Hoyos S, Sa´ez M, Herva´s A (1997) Mortality as a function of temperature. A study in Valencia, Spain, 1991–1993. Int J Epidemiol 26: 551–561. 39. Rocklo¨v J, Forsberg B (2008) The effect of temperature on mortality in Stockholm 1998–2003: a study of lag structures and heatwave effects. Scand J Public Health 36: 516–523. doi:10.1177/1403494807088458. 40. Eccles R (2002) An explanation for the seasonality of acute upper respiratory tract viral infections. Acta Otolaryngol 122: 183–191. 41. Barnett AG, Sans S, Salomaa V, Kuulasmaa K, Dobson AJ (2007) The effect of temperature on systolic blood pressure. Blood Press Monit 12: 195–203. doi:10.1097/MBP.0b013e3280b083f4. Environment and Scheduling in Track Races PLOS ONE | www.plosone.org 7 November 2013 | Volume 8 | Issue 11 | e79548

Environment and scheduling effects on sprint and middle distance running performances.

11-20-2013

Haïda, Amal,Dor, Frédéric,Guillaume, Marion,Quinquis, Laurent,Marc, Andy,Marquet, Laurie-Anne,Antero-Jacquemin, Juliana,Tourny-Chollet, Claire,Desgorces, François,Berthelot, Geoffroy,Toussaint, Jean-François

eng

PMC7068447

International Journal of Environmental Research and Public Health Brief Report Race Analysis of the World’s Best Female and Male Marathon Runners Véronique Billat 1,2,*,† , Damien Vitiello 1,†, Florent Palacin 1, Matthieu Correa 1 and Jean Renaud Pycke 1,3 1 Université de Paris, EA3625-Institut des Sciences du Sport Santé de Paris (I3SP), 75015 Paris, France; damien.vitiello@parisdescartes.fr (D.V.); palacinflorent@gmail.com (F.P.); matthieucorrea37@gmail.com (M.C.); jeanrenaud.pycke@univ-evry.fr (J.R.P.) 2 Véronique Billat, Université de Paris, EA 3625-Institut des Sciences du Sport Santé de Paris (I3SP), 1 rue Lacretelle, 75015 Paris, France 3 Université d’Evry Val d’Essonne, UMR8071—CNRS-Laboratoire de Mathématiques et Modélisation d’Evry (LaMME), 91037 Evry, France * Correspondence: veroniquelouisebillat@gmail.com; Tel.: +33-(0)786117308 † Equally contributing authors. Received: 23 December 2019; Accepted: 9 February 2020; Published: 13 February 2020



 Abstract: Background: Beyond the difference in marathon performance when comparing female and male runners, we tested the hypothesis that running strategy does not different according to sex. The goal of the present study is to compare the running strategy between the best female and male marathon performances achieved in the last two years. Methods: Two aspects of the races were analyzed: (i) average speed relative to runner critical speed (CS) with its coefficient of variation and (ii) asymmetry and global tendency of race speed (i.e., the race’s Kendall τ). Results: The females’ best marathons were run at 97.6% ± 3% of CS for the new record (Brigid Kosgei, 2019) and at 96.1% ± 4.4% for the previous record (Paula Radcliffe, 2003). The best male performances (Eliud Kipchoge, 2018 and 2019) were achieved at a lower fraction of CS (94.7% ± 1.7% and 94.1% ± 2.3% in 2018 and 2019, respectively). Eliud Kipchoge (EK) achieved a significant negative split race considering the positive Kendall’s τ of pacing (i.e., time over 1 km) (τ = 0.30; p = 0.007). Furthermore, EK ran more of the average distance below average speed (54% and 55% in 2018 and 2019, respectively), while female runners ran only at 46% below their average speed. Conclusions: The best female and male marathon performances were run differently considering speed time course (i.e., tendency and asymmetry), and fractional use of CS. In addition, this study shows a robust running strategy (or signature) used by EK in two different marathons. Improvement in marathon performance might depend on negative split and asymmetry for female runners, and on higher fractional utilization of CS for male runners. Keywords: running strategy; critical speed; endurance; performance; health 1. Introduction For 20 years, marathon racing has gained popularity given that it is one of the rare sporting events in which elite and non-elite runners compete at the same time, despite some athletes completing the race in twice the time of others. Although males are nearly able to finish the race in under two hours (Eliud Kipchoge (EK): 2 h 01 min 39 s in Berlin, 2018, and 2 h 02 min 37 s in London, 2019), the milestone for female runners of 2 h 15 min was broken by Brigid Kosgei (BK) (2 h 14 min 04 s) in 2019 during the Chicago marathon, 16 years after the previous world record of Paula Radcliffe (PR) at the London marathon in 2003. Beyond speculation on the future of performance or on the comparison of relative performance in males and females [1,2], the goal of this study is to test the hypothesis that running strategy does Int. J. Environ. Res. Public Health 2020, 17, 1177; doi:10.3390/ijerph17041177 www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2020, 17, 1177 2 of 6 not differ according to sex. We consider that both female and male marathoners might benefit from a training program based on their perception to improve their performance. Indeed, this may allow them to have a more positive asymmetry, a lower coefficient of variation of speed, and to run at a higher percentage of their critical speed (CS) during the entire race. Therefore, this study analyzed two aspects of the races: (i) the average runner’s speed relative to their CS [3–5], with the coefficient of variation and (ii) the tendency of the pace and its asymmetry [6]. 2. Materials and Methods To achieve this study, pacing (i.e., time per distance) run by EK (35 years old, Berlin 2018 and London 2019), BK (25 years old, Chicago 2019) and PR (29 years old, London 2003) were examined. Data were retrieved from the World Athletics website on 14th October 2019. A computation of the average speed per distance was achieved by dividing distance per time unit. 2.1. Critical Speeds (CS) The average speed relative to runner CS, with the coefficient of variation, was calculated. CS was calculated from the runner’s personal best performances in the 3000 m and half marathon (run in less than 1h for EK, and 1 h 04 min 28 s and 1 h 05 min 40 s for BK and PR, respectively). The CS was calculated using the following equation [7]: Dlim = α + β tlim (1) Dlim = distance; α = constant reserve; β = critical speed; tlim = record time. 2.2. Global Tendency of Pace and Its Asymmetry Here, the trend in speed time series (i.e., Kendall’s τ non-parametric rank correlation coefficient) [8] and the pacing design (i.e., asymmetry characteristics of the race) [6] were compared. Here is the equation of Kendall’s τ: τ = 2 / n(n − 1) X i<j K(vi, vj) (2) vi = ith value of a speed; vj = jth value of a speed; i < j = i indicates a period of time prior to j; sum being performed over the n(n − 1)/2 distinct unordered couples of indices {i, j}, so that τ takes values in between −1 and 1. We sought to establish running strategy or signature in real race format for the same runner (EK), and for male and female official best marathon performances. 3. Results 3.1. Average Marathon Speed Relative to Marathoners’ Critical Speed and Coefficient of Speed Variation The male and female marathon races were run at different percentages of the CS (Table 1). Int. J. Environ. Res. Public Health 2020, 17, 1177 3 of 6 Table 1. Average marathon speed relative to male and female marathoners’ critical speed and coefficient of speed variation. Speed Critical Speed (km) Speed (% Critical Speed) Skewness (% km below Mean Speed) Pace / Trend Athlete Place / Year Time Mean SD Variation Coefficient Mean SD Variation Coefficient Kendall’s t p-Value E. Kipchoge London /2019 2 h 02 min 37 s 20.67 0.51 2.48% 21.6 94.1 2.3 2.44% 55 −0.0289 0.8000 E. Kipchoge Berlin /2018 2 h 01 min 39 s 20.78 0.35 1.68% 21.6 94.7 1.7 1.80% 54 0.3000 0.0069 P. Radcliffe London /2003 2 h 15 min 25 s 18.89 0.30 1.59% 19.6 96.0 4.4 4.58% 32 0.0050 0.7227 B. Kosgei Chicago /2019 2 h 14 min 04 s 18.82 0.65 3.45% 19.29 97.6 3.4 3.45% 46 0.0955 0.4891 Data are presented in means ± SD and percentages. Indeed, the best female marathon performances were run at 97.6% ± 3% of CS for the new record (BK, 2019) and 96.1% ± 4.4% for the previous record (PR, 2003), while the best male performance (i.e., EK, Berlin 2018) was run at a lower fraction of CS (94.7% ± 1.7%). 3.2. Speed Trend and Asymmetry Characteristics EK (male athlete) achieved a large negative split race considering pace with a positive Kendall’s τ (0.30; p = 0.007) and ran more of the average distance below average speed (Figure 1) (54%). By contrast, there were no trends in previous and current marathon world records of BK and PR (female athletes) (Figure 2). Their races were not optimal in regard to speed asymmetry as BK (66%) and PR ran 46% of the distance above their average speed. Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 3 of 6 Table 1. Average marathon speed relative to male and female marathoners’ critical speed and coefficient of speed variation. Speed Critic al Spee d (km) Speed (% Critical Speed) Skewne ss (% km below mean speed) Pace / Trend Athlete Place / Year Time Mean SD Variatio n Coefficie nt Mea n SD Variatio n Coeffici ent Kendall' s t p- value E. Kipchoge London / 2019 2h02 min37 s 20.67 0.5 1 2.48% 21.6 94.1 2.3 2.44% 55 -0.0289 0.8000 E. Kipchoge Berlin / 2018 2h01 min39 s 20.78 0.3 5 1.68% 21.6 94.7 1.7 1.80% 54 0.3000 0.0069 P. Radcliffe London / 2003 2h15 min25 s 18.89 0.3 0 1.59% 19.6 96.0 4.4 4.58% 32 0.0050 0.7227 B. Kosgei Chicago / 2019 2h14 min04 s 18.82 0.6 5 3.45% 19.29 97.6 3.4 3.45% 46 0.0955 0.4891 Data are presented in means ± SD and percentages. 3.2. Speed Trend and Asymmetry Characteristics EK (male athlete) achieved a large negative split race considering pace with a positive Kendall’s τ (0.30; p = 0.007) and ran more of the average distance below average speed (Figure 1) (54%). By contrast, there were no trends in previous and current marathon world records of BK and PR (female athletes) (Figure 2). Their races were not optimal in regard to speed asymmetry as BK (66%) and PR ran 46% of the distance above their average speed. Figure 1. Speed trend and asymmetry characteristics in the world’s fastest male marathoner. The curve (dotted line) represents the time course of Eliud Kipchoge’s running speed during the London marathon in 2019. The curve (dashed line) represents the time course of his running speed during the Berlin marathon in 2018. These curves represent the percentage of the mean speed achieved in each distance unit during the entire marathon. Figure 1. Speed trend and asymmetry characteristics in the world’s fastest male marathoner. The curve (dotted line) represents the time course of Eliud Kipchoge’s running speed during the London marathon in 2019. The curve (dashed line) represents the time course of his running speed during the Berlin marathon in 2018. These curves represent the percentage of the mean speed achieved in each distance unit during the entire marathon. Int. J. Environ. Res. Public Health 2020, 17, 1177 4 of 6 Int. J. Environ. Res. Public Health 2020, 17, x FOR PEER REVIEW 4 of 6 Figure 2. Speed trend and asymmetry characteristics in the world’s fastest female marathoners. The curve (solid line) represents the time course of running speed during the previous world record marathon reached by Paula Radcliffe during the London marathon in 2003. The curve (dotted line) represents the time course of running speed during the new world record reached by Brigid Kosgei during the Chicago marathon in 2019. These curves represent the percentage of the mean speed achieved in each distance unit during the entire marathon. 4. Discussion To the best of our knowledge, this is the first study to compare the running strategy between the best male and female marathoners, based on a previous statistical analysis jointly using the trend and asymmetry of the race [6]. In this study, it was demonstrated that EK, PR, and BK (i) did not use the same running strategy and (ii) did not use the same fraction of their respective CS. Finally, EK achieved his two best performances using the same running signature on two different marathon routes (2 h 01 min 39 s in Berlin 2018, and 2 h 02 min 37 s in London 2019). This study demonstrated that females ran at a constant pace considering their Kendall’s τ was between −0.05 and 0.05 [6]. It was also highlighted that world records are broken using a running strategy based on running speeds below median speed, unlike popular marathoners who run more distance at speeds above the median [9]. This may be due to the fact that lower level runners (>2 h 20 min) run at too high a target that they cannot maintain beyond the 26th km, where the average speed (i.e., final performance) is reduced and is therefore lower than the median speed. These results may ask questions of the perception of physiological load, the plan of action to produce speed variations (or not), and the strategy of the U-race, which is conducive to performance but which requires a high reserve of power. This latter point may imply training protocols based on running acceleration and deceleration [10]. Therefore, high intensity interval training using both positive and negative acceleration to attain VO2max at a wide range of speeds may allow an increase of endurance capacity and anaerobic power, which are required to achieve top performances in the marathon [11–13]. The present work also examined the newly established marathon world records (in 2019, for both EK and BK) using an assessment of the race speed asymmetry [6]. We also examined pacing variability and confirmed that this was run at and below 3% in marathons [14]. The speed coefficient of variation of EK was within 1.7%, indicating a relative pacing strategy. In addition, the asymmetry of the speed time series of the runners was also analyzed. Our results suggest the possibility of Figure 2. Speed trend and asymmetry characteristics in the world’s fastest female marathoners. The curve (solid line) represents the time course of running speed during the previous world record marathon reached by Paula Radcliffe during the London marathon in 2003. The curve (dotted line) represents the time course of running speed during the new world record reached by Brigid Kosgei during the Chicago marathon in 2019. These curves represent the percentage of the mean speed achieved in each distance unit during the entire marathon. 4. Discussion To the best of our knowledge, this is the first study to compare the running strategy between the best male and female marathoners, based on a previous statistical analysis jointly using the trend and asymmetry of the race [6]. In this study, it was demonstrated that EK, PR, and BK (i) did not use the same running strategy and (ii) did not use the same fraction of their respective CS. Finally, EK achieved his two best performances using the same running signature on two different marathon routes (2 h 01 min 39 s in Berlin 2018, and 2 h 02 min 37 s in London 2019). This study demonstrated that females ran at a constant pace considering their Kendall’s τ was between −0.05 and 0.05 [6]. It was also highlighted that world records are broken using a running strategy based on running speeds below median speed, unlike popular marathoners who run more distance at speeds above the median [9]. This may be due to the fact that lower level runners (>2 h 20 min) run at too high a target that they cannot maintain beyond the 26th km, where the average speed (i.e., final performance) is reduced and is therefore lower than the median speed. These results may ask questions of the perception of physiological load, the plan of action to produce speed variations (or not), and the strategy of the U-race, which is conducive to performance but which requires a high reserve of power. This latter point may imply training protocols based on running acceleration and deceleration [10]. Therefore, high intensity interval training using both positive and negative acceleration to attain VO2max at a wide range of speeds may allow an increase of endurance capacity and anaerobic power, which are required to achieve top performances in the marathon [11–13]. The present work also examined the newly established marathon world records (in 2019, for both EK and BK) using an assessment of the race speed asymmetry [6]. We also examined pacing variability and confirmed that this was run at and below 3% in marathons [14]. The speed coefficient of variation of EK was within 1.7%, indicating a relative pacing strategy. In addition, the asymmetry of the speed time series of the runners was also analyzed. Our results suggest the possibility of establishing a world Int. J. Environ. Res. Public Health 2020, 17, 1177 5 of 6 record (i.e., Berlin 2018) by running below the average pace, thanks to a very fast start and finish, in the classical format of the marathon. Finally, the marathon should not be a constant speed race [15] and new strategies of running must be addressed to allow better performance in the future. As VO2max is not sufficient to distinguish high level marathon runners (2 h 11 min–2 h 16 min) from elite marathon runners (less than 2 h 11 min) [16], this parameter may be used in parallel with average running speed and CS to propose a new pacing strategy adapted to the level of marathoners. Indeed, this study highlighted that average speed represents 94% of CS. The value is close to personal best one-hour record, meaning that the runner might be able to double his time limit just by decreasing his speed by 5%. Finally, another way of optimizing the running strategy might be the development of runners’ anaerobic capacity and their ability to use a higher fraction of their CS during the marathon. 5. Practical Applications - A new training program based on athletes’ perception could be introduced to better adapt their speed while running a marathon. - New training methods in marathon running could optimize the running strategy to allow the marathoner to have a positive asymmetry and a lower coefficient of variation of speed during the race. - The CS is a crucial parameter that might be improved by new training methods since a high fraction of the CS is achieved in both female and male marathon world records. - A performance target for coaches and athletes should be the maintenance of constant pace just below average pace and at 95% of the CS. 6. Conclusions Due to the already high fraction of CS achieved in marathons by females and also for the world’s best performance by EK, this parameter and the average speed achieved in each km might be two parameters to consider in elaborating a new running strategy to improve performance in marathon and short events, as already reported [16]. In addition, optimizing the running strategy with positive asymmetry and lower coefficient of variation of speed might allow marathoners to run more slowly at almost a constant pace, just below the average pace, and at 95% of CS in accordance with their higher speed reserve. However, future research on other marathoners is necessary to confirm difference in running strategy between females and males and to verify consistency of running signature for a single marathon. Author Contributions: Conceptualization, D.V., V.B.; Methodology, D.V., J.R.P., V.B.; Software, M.C., F.P.; Validation, D.V., V.B.; Formal Analysis, M.C., J.R.P.; Investigation, M.C., F.P.; Resources, V.B.; Data Curation, F.P.; Writing—Original Draft Preparation, D.V., M.C., V.B.; Writing—Review & Editing, D.V.; Visualization, M.C., J.R.P.; Supervision, D.V., V.B.; Project Administration, D.V., V.B. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Acknowledgments: The research team would like to thank our Australian, Eddy C., for his proofreading of the manuscript. Conflicts of Interest: The authors declare no conflict of interest. References 1. Hunter, S.K.; Joyner, M.J.; Jones, A.M. The two-hour marathon: What’s the equivalent for women? J. Appl. Physiol. 2015, 118, 1321–1323. [CrossRef] [PubMed] 2. Joyner, M.J.; Ruiz, J.R.; Lucia, A. The two-hour marathon: Who and when? J. Appl. Physiol. 2011, 110, 275–277. [CrossRef] [PubMed] 3. Jones, A.M.; Burnley, M. Oxygen uptake kinetics: An underappreciated determinant of exercise performance. Int. J. Sports Physiol. Perform. 2009, 4, 524–532. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2020, 17, 1177 6 of 6 4. Jones, A.M.; Burnley, M.; Black, M.I.; Poole, D.C.; Vanhatalo, A. The maximal metabolic steady state: Redefining the ’gold standard’. Physiol. Rep. 2019, 7, e14098. [CrossRef] [PubMed] 5. Jones, A.M.; Vanhatalo, A. The ‘Critical Power’ concept: Applications to sports performance with a focus on intermittent high-intensity exercise. Sports Med. 2017, 47, 65–78. [CrossRef] [PubMed] 6. Billat, V.L.; Carbillet, T.; Correa, M.; Pycke, J.R. Detecting the marathon asymmetry with a statistical signature. Phys. A. 2019, 515, 240–247. [CrossRef] 7. Ettema, J.H. Limits of human performance and energy production. Eur. J. Appl. Physiol. 1966, 22, 45–54. [CrossRef] 8. Kendall, M.G.; Stuart, A. The Advance Theory of Statistics; Griffin and Co: London, UK, 1963. 9. Billat, V.L.; Palacin, F.; Correa, M.; Pycke, J.R. Racing strategy affects the sub-elite marathoner’s cardiac drift and performance. Front. Psychol. 2019, 10, 3026. 10. Billat, V.; Brunel, N.J.B.; Carbillet, T.; Labbé, S.; Samson, A. Humans are able to self-paced constant running accelerations until exhaustion. Phys. A 2018, 506, 290–304. [CrossRef] 11. Niel, R.; Ayachi, M.; Mille-Hamard, L.; Le Moyec, L.; Savarin, P.; Clement, M.J.; Besse, S.; Launay, T.; Billat, V.L.; Momken, I. A new model of short acceleration-based training improves exercise performance in old mice. Scand. J. Med. Sci. Sports 2017, 27, 1576–1587. [CrossRef] [PubMed] 12. Morton, R.H.; Billat, V. Modelling decremental ramps using 2- and 3-parameter "critical power" models. J. Sports Sci. 2013, 31, 731–735. [CrossRef] [PubMed] 13. Billat, V.; Petot, H.; Karp, J.R.; Sarre, G.; Morton, R.H.; Mille-Hamard, L. The sustainability of VO2max: Effect of decreasing the workload. Eur. J. Appl. Physiol. 2013, 113, 385–394. [CrossRef] [PubMed] 14. Haney, T.A., Jr.; Mercer, J.A. A description of variability of pacing in marathon distance running. Int. J. Exerc. Sci. 2011, 4, 133–140. [PubMed] 15. Maron, M.B.; Horvath, S.M.; Wilkerson, J.E.; Gliner, J.A. Oxygen uptake measurements during competitive marathon running. J. Appl. Physiol. 1976, 40, 836–838. [CrossRef] [PubMed] 16. Billat, V.L.; Demarle, A.; Slawinski, J.; Paiva, M.; Koralsztein, J.P. Physical and training characteristics of top-class marathon runners. Med. Sci. Sports Exerc. 2001, 33, 2089–2097. [CrossRef] [PubMed] © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Race Analysis of the World's Best Female and Male Marathon Runners.

02-13-2020

Billat, Véronique,Vitiello, Damien,Palacin, Florent,Correa, Matthieu,Pycke, Jean Renaud

eng

PMC6040767

Código de Proyecto: Fecha de Presentación/Versión: A completar por CEI-UCJC FORMULARIO DE SOLICITUD DE EVALUACIÓN POR EL COMITÉ ETICO DE INVESTIGACIÓN (CEI-UCJC) Datos del Investigador Principal Nombre Carlos Balsalobre Fernández Facultad/Escuela Facultad de Formación de Profesorado y Educación, Universidad Autónoma de Madrid Departamento Educación Física, Deporte y Motricidad Humana Datos contacto carlos.balsalobre@icloud.com Datos del Estudio de Investigación Título del Proyecto The effects of nitrate supplementation via beetroot juice on the running economy, neuromuscular performance and running mechanics in elite middle and long distance runners / Los efectos de la suplementación de nitratos a través del zumo de remolacha en la economía de carrera, el rendimiento neuromuscular y la mecánica de carrera en corredores de media y larga distancia de élite (ACRONIMO: BEET-RUN) La investigación incluye: Seleccionar Apartado a completar Solo seres humanos X Apartado A Muestras biológicas Apartado B Organismos modificados genéticamente Apartado C Animales de experimentación Apartado D Documentación que se adjunta: X Formulario solicitud de evaluación X Apartados A – B – C – D (tachar los que no procedan) X COPIA DEL PROYECTO X Hoja de información a los participantes X Consentimiento informado a emplear X Compromiso escrito del Investigador Responsable  Código de Proyecto: Fecha de Presentación/Versión: A completar por CEI-UCJC Apartado A. Investigación en seres humanos sin toma de muestras biológicas ¿Qué grupos de participantes se han establecido? (indicar todos: enfermos, controles sanos menores, discapacitados…) Describir Adultos sanos X Adultos enfermos Niños sanos Niños enfermos Discapacitados Mujeres embarazadas Mujeres lactantes Población en riesgo de exclusión social Otras poblaciones incapaces de expresar su consentimiento Grupos étnicos ¿Por qué se han seleccionado dichos grupos? Porque la investigación tiene como objetivo analizar un determinado tipo de suplementación nutricional (mediante zumos de remolacha) en el rendimiento en un grupo de corredores de élite. ¿Qué método de disociación de datos se va a utilizar? (según Ley 14/2007 de investigación biomédica) Seleccionar el que proceda Codificación (disociación reversible) X Anonimización (disociación irreversible) Describir brevemente el procedimiento empleado para llevar a cabo dicha disociación Los deportistas se numerarán de menor a mayor por orden alfabético. Posteriormente, cada variable que se mida tendrá un acrónimo, y en función de si dicha variable se ha medido en el momento pre o post intervención, se utilizarán las siglas PRE o POST. Por ejemplo, para identificar el valor de consumo de oxígeno pre-intervención del deportista ordenado en la lista en la posición 11, se utilizará el término “VO.11.Pre”  Código de Proyecto: Fecha de Presentación/Versión: A completar por CEI-UCJC ¿Se van a aplicar métodos invasivos para la toma de datos? En caso afirmativo, describa brevemente como se va a llevar a cabo dicho procedimiento estableciendo claramente las medidas relacionadas con la evitación del daño y la cualificación del personal que va a llevarlo a cabo. NO PROCEDE. No se usarán métodos invasivos En caso de producirse daño, ¿qué procedimiento paliativo/curativo se prevé realizar? ¿se cuenta con algún tipo de aseguramiento/compensación del daño? ¿Cuál? Si es que no explicar por qué no se ha establecido. NO PROCEDE. No se usará métodos invasivos ¿Se ofrecen incentivos o compensaciones a los sujetos por su participación en los experimentos? Indique su naturaleza y cuantía. NO  Código de Proyecto: Fecha de Presentación/Versión: A completar por CEI-UCJC APPLICATION FORM FOR THE ETHICS COMMITEE APPROVAL (CEI-UCJC) PRINCIPAL INVESTIGATOR NAME Carlos Balsalobre Fernández FACULTY Facultad de Formación de Profesorado y Educación, Universidad Autónoma de Madrid Departament Educación Física, Deporte y Motricidad Humana CONTACT INFORMATION carlos.balsalobre@icloud.com INFORMATION OF THE TRIAL TITLE The effects of nitrate supplementation via beetroot juice on the running economy, neuromuscular performance and running mechanics in elite middle and long distance runners / Los efectos de la suplementación de nitratos a través del zumo de remolacha en la economía de carrera, el rendimiento neuromuscular y la mecánica de carrera en corredores de media y larga distancia de élite (ACRONIMO: BEET-RUN) RESEARCH INCLUDES: SELECT SECTION TO COMPLETE HUMANS X SECTION A BIOLOGICAL SAMPLES SECTION B GENE MODIFICATIONS SECTION C EXPERIMENTS WITH ANIMALS SECTION D ATTACHED DOCUMENTATION: X APPLICATION FORM X SECTIO A – B – C – D (REMOVE WHAT DON’T PROCEED) X COPY OF THE APPLICATION FORM X INFORMATION FOR PARTICIPANTS ABOUT THE STUDY X INFORMED CONSENT X DECLARATION OF THE PRINCIPAL INVESTIGATOR  Código de Proyecto: Fecha de Presentación/Versión: A completar por CEI-UCJC SECTION A. INVESTIGATIONS WITH HUMAN BEINGS WITHOUT COLLECTING BIOLOGICAL SAMPLES WHAT TYPE OF PARTICIPANTS ARE USED? SELECT HEALTHY ADULTS X UNHEALTHY ADULTS HEALTHY KIDS UNHEALTHY KIDS DISABLE PEOPLE PREGNANT WOMA BREASTFEEDING WOMEN POPULATION AT RISK OF SOCIAL EXCLUSION OTHER POPULATIONS ETHNICAL SUBGROUPS WHY DID YOU SELECT THESE GROUPS? Because the trial aims to analyze certain supplementation (beetroot juice) in the performance of elite runners WHAT METHODS OF DISSOSIATION WILL YOU USE? SELECT CODIFICATION (reversible) X ANONIMATION (irreversible) DESCRIBE THE PROTOCOL TO CONDUCT THAT DISSOSIATION Athletes will be labeled with increasing numbers in an alphabetic order. Then, each variable will have an acronym, with PRE or POST depending of the moment at which it was measured. For example, to identify the pre-intervention VO2max of the athlete number 11, the term VO2Max.11.PRE will be used.  Código de Proyecto: Fecha de Presentación/Versión: A completar por CEI-UCJC WILL YOU USE INVASIVE METHODS TO COLLECT THE DATA? NO IN THE EVENT OF PRODUCING DAMAGE, WHAT TREATMENT WILL YOU USE? N/A. No invasive methods will be used WILL YOU OFFER ANY REWARD TO THE SUBJECTS FOR THEIR PARTICIPATION IN THE EXPERIMENT? NO

The effects of beetroot juice supplementation on exercise economy, rating of perceived exertion and running mechanics in elite distance runners: A double-blinded, randomized study.

07-11-2018

Balsalobre-Fernández, Carlos,Romero-Moraleda, Blanca,Cupeiro, Rocío,Peinado, Ana Belén,Butragueño, Javier,Benito, Pedro J

eng

PMC6193581

RESEARCH ARTICLE Diagnoses and time to recovery among injured recreational runners in the RUN CLEVER trial Benjamin Mulvad1, Rasmus Oestergaard NielsenID2*, Martin Lind1, Daniel Ramskov2,3 1 Division of Sports Traumatology, Department of Orthopedics, Aarhus University Hospital, Aarhus, Denmark, 2 Section for Sports Science, Department of Public Health, Aarhus University, Aarhus, Denmark, 3 Department of Physiotherapy, University College of Northern Denmark, Aalborg, Denmark * roen@ph.au.dk Abstract Purpose The purpose of the present study was to describe the incidence proportion of different types of running-related injuries (RRI) among recreational runners and to determine their time to recovery. Methods A sub-analysis of the injured runners included in the 839-person, 24-week randomized trial named Run Clever. During follow-up, the participants reported levels of pain in different ana- tomical areas on a weekly basis. In case injured, runners attended a clinical examination at a physiotherapist, who provided a diagnosis, e.g., medial tibial stress syndrome (MTSS), Achilles tendinopathy (AT), patellofemoral pain (PFP), iliotibial band syndrome (ITBS) and plantar fasciopathy (PF). The diagnose-specific injury proportions (IP) and 95% confidence intervals (CI) were calculated using descriptive statistics. The time to recovery was defined as the time from the first registration of pain until total pain relief in the same anatomical area. It was reported as medians and interquartile range (IQR) if possible. Results A total of 140 runners were injured at least once leading to a 24-week cumulative injury pro- portion of 32% [95% CI: 26%; 37%]. The diagnoses with the highest incidence proportion were MTSS (IP = 16% [95% CI: 9.3%; 22.9%], AT (IP = 8.9% [95% CI: 3.6%; 14.2%], PFP (IP = 8% [95% CI: 3.0%; 13.1%]. The median time to recovery for all types of injuries was 56 days (IQR = 70 days). Diagnose-specific time-to-recoveries included 70 days (IQR = 89 days) for MTSS, 56 days (IQR = 165 days) for AT, 49 days (IQR = 63 days) for PFP. Conclusion The most common running injuries among recreational runners were MTSS followed by AT, PFP, ITBS and PF. In total, 77 injured participants recovered their RRI and the median time PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 1 / 11 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Mulvad B, Nielsen RO, Lind M, Ramskov D (2018) Diagnoses and time to recovery among injured recreational runners in the RUN CLEVER trial. PLoS ONE 13(10): e0204742. https://doi.org/ 10.1371/journal.pone.0204742 Editor: Manoj Srinivasan, The Ohio State University, UNITED STATES Received: December 11, 2017 Accepted: September 13, 2018 Published: October 12, 2018 Copyright: © 2018 Mulvad et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: Data are from the RUN CLEVER study whose authors may be contacted at DAR@ucn.dk and relevant data from the injured participants can be found in the supplementary files. Funding: The Danish Rheumatism Association (https://www.gigtforeningen.dk/) provided a DKK 75.000 grant for this study. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. to recovery for all types of injuries was 56 days and MTSS was the diagnosis with the lon- gest median time to recovery, 70 days. Introduction Running is a very popular type of exercise and the number of runners worldwide has grown over the past decades [1]. Among recreational runners, the most supported motives are to keep healthy, to maintain stamina and to reduce weight or avoid increasing their weight [2]. Running contributes to a range of health-related benefits such as lowering overall body fat, optimizing the composition of fat molecules in the blood, lowering the resting heart rate and improving the overall cardiovascular fitness [3]. In general, runners have a 25–40% reduced risk of premature mortality and live approximately 3 years longer than non-runners [4]. Owing to the health benefits and because of the considerable interest in running illuminating barriers to continued running deserves to be a key public health priority. In Denmark, it has been estimated that 5% of the adult population, equivalent to 260,000 individuals, suffer from a running-related injury (RRI) on a yearly basis [5]. Running is hence the sports activity that contributes with most annual sports injuries in Denmark. When evalu- ated in a population of runners, 1-year injury incidence proportions have been reported in the range from 43.2% to 84.9% in different types of runners [6]. Running injuries were the most common reason for permanently dropping out of a running regime among males, and the third-most common reason among females according to a 10-year prospective cohort study [7]. Direct economic costs of running-related injuries range from 0.3% to 4.6% of national healthcare expenditure [8]; and some injured runners come to suffer from permanent physical disability making them unable to exercise due to pain or discomfort [9,10]. Indeed, the combi- nation of mental and physical consequences increases the likelihood of lapsing into a sedentary lifestyle during and after injury recovery. Running-related injuries usually occur in the lower extremity [11]. Some of the most fre- quent diagnoses amongst runners are patellofemoral pain (PFP), iliotibial band syndrome (ITBS) and plantar fasciosis (PF), with proportions in relation to all injuries ranging between 10–17%, 4–8%, and 5–8%, respectively [12,13]. Commonly, runners receive a referral to a physiotherapist for treatment purposes [14]. Here, many runners are concerned with the time to recovery. To provide answers, insights into diagnose-specific time-to-recoveries are needed. Unfortunately, there is a literature gap concerning the time to recovery for classical running- related injuries such as PFP, ITBS and PF. Among novice runners, the median time to recovery of all types of RRIs has been estimated to approximately 10 weeks with diagnose-specific recov- eries ranging between 26 days to 174 days [13]. Still, no study has investigated the time to recovery among injured recreational runners. Consequently, the purposes of the present study were to describe the incidence proportion of different types of running-related injuries among recreational runners, engaged in the Run Clever trial [15], and to determine their time to recovery measured in days. Materials and methods The present paper presents a sub-analysis of the injured participants from the Run Clever trial. The Run Clever trial was a randomized 24-week follow-up intervention study including recre- ational runners. The intervention was two different running schedules, the main outcome was RRIs and the participants were followed by weekly questionnaires. The two running schedules Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 2 / 11 Competing interests: The authors have declared that no competing interests exist. were founded on the same framework, 3 running sessions per week, and an identical 8 weeks preconditioning period followed by 16 weeks of intervention. The intervention training period was organized in cycles of 4 weeks with progression. One group, the intensity training group, had a fixed running volume but the amount of hard pace was increased during the cycles of progression. The other group, the volume training group, focused on increasing the total run- ning volume per week but only performed at an easy or moderate pace. The original purpose was to compare overall risk of injury between progression in running intensity and running volume [15]. The Run Clever trial was approved by The Ethics Committee Northern Denmark and the Danish Data Protection Agency (N-20140069). Prior to recruitment, on January 23rd 2015, the trial was registered at Clinicaltrials.gov under registration number: NCT02349373. Healthy persons between 18 and 65 years of age were eligible for inclusion in the Run Clever trial. They had to be recreational runners free of injury in their lower extremities in the past 6 months. A recreational runner was defined as a person who had been running 1 to 3 weekly sessions for at least 6 months. The approach of recruiting participants and further crite- rions for inclusion or exclusion of the Run Clever trial are described in detail elsewhere [15]. The sub-sample included in the present study, were participants included in the Run Clever trial who sustained at least one RRI during the follow-up period. At baseline, each participant was provided access to an internet-based training diary. After being registered in the diary, the participants received weekly automated e-mails including a link to an online questionnaire on injury-related pain. The questionnaire contained questions regarding symptoms of overuse or injuries based on the Oslo Sports Trauma Research Center Questionnaire (OSTRC) [16]. The OSTRC was modified with two additional questions and an additional option of answers to adapt it for the Run Clever Trial. When discomfort or an injury was registered in the OSTRC questionnaire, the participant informed on their pain in different anatomical areas, and the options were the “foot”, “ankle”, “front of lower leg”, “calf”, “knee”, “thigh”, “hamstrings”, “groin”, “glutes”, “hip” and “lower back”. The questionnaires were dis- tributed as e-mails every Sunday to the participants’ e-mail address. The participants had to complete it whether or not suffering an injury, hereby getting information of any experienced pain the previous week. In case no response was received during the Sunday, a reminder e- mail was sent to the participant the following Monday (the day after). In line with most recent scientific work, a RRI was defined as any physical pain or com- plaints from muscles, joints, bones or tendons of the lower extremities or back as a result of running [17]. It had to reduce the training performance such as distance, frequency, intensity or pace for at least 7 days [18]. When a participant reported a RRI via the weekly injury-ques- tionnaire, an appointment with a certified physiotherapist, who was part of a study-specific diagnostic team, was made. The physiotherapist performed the clinical examinations in their respective clinics, generally within a week, and used a standardized examination procedure [13]. The physiotherapist made the standardized examination of the foot, ankle, lower leg, knee, thigh, hip or back and compared their findings with standardized, non-validated diag- nostic criterions for different diagnoses [13]. The diagnosis was based on the medical history and objective findings. When the physiotherapist had completed an examination, the diagno- sis (e.g., medial tibias stress syndrome (MTSS), Achilles tendinopathy (AT)) and date of exam- ination was registered and reported to the database. No treatment or plans of rehabilitation was delivered, only a few pieces of advice at the most. However, the participant was allowed to search for treatment and receive treatment elsewhere. The definition of time to recovery was based on the responses in the weekly OSTRC-scores on pain as well as the diagnostic examination by the physiotherapist. The date of examination and diagnosis provided by the physiotherapist were compared to the responses from the weekly OSTRC-scores to identify if pain reported via the OSTRC in the affected anatomical Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 3 / 11 site corresponded with the anatomical location of the diagnosis provided by the physiothera- pists. Based on this, the time to recovery was defined as the time from the first registered pain in a specific anatomical area until total pain relief in the same anatomical area. Date of recov- ery was defined as the date total pain relief occurred, which, then, was followed by at least three weeks without pain in the relevant anatomical site. If a participant was pain-free for a week but reported pain the following two weeks in the same anatomical location, the partici- pant was still classified as being injured. However, if new pain arose in the same anatomical site after three weeks without pain, it was considered as a new injury. If a participant sustained two different RRIs or more during the follow-up period, only the first injury was included in the analysis. The injured runners were excluded from the analyses on time to recovery if they did not meet the following eligibility criteria: (i) the injury had to recover before at end of 24-week fol- low-up, (ii) they had to answer at least ten of the weekly administered questionnaires, (iii) their pain had to be registered in the same anatomical location as the one registered by the physiotherapist, (iv) they needed to register pain (e.g., in some cases, no pain was registered at all), (v) they had to register a date of injury occurrence or (vi) the time to recovery had to be plausible compared to the diagnosis (e.g., we found pain for one week following a broken leg unreliable). The Kaplan-Meier estimator was used to calculate the proportion of injury-free Run Clever participants as a function of weeks. As these methods takes into account censoring, the propor- tion of injured participants after 24-week follow-up is not number of injured runners divided by the total sample size as the latter approach assumes complete follow-up for all runners. Data on time to recovery was evaluated using histograms and 95% prediction intervals to decide if it was normally distributed. As this was not the case, non-parametric statistics were used to present time to recovery as medians and IQRs. At least five recovered injuries were required to include these calculations. The data-management and analyses presented are per- formed using STATA/SE version 14 and Microsoft Excel 2010. Results A total of 839 runners participated the Run Clever Trial of whom 521 (62%) were female and 318 (38%) were male. The mean age was 39.2 (±10.0) years. 140 sustained at least one RRI dur- ing the follow-up period. A Kaplan-Meier graph visualizing the proportion of injury-free run- ners as a function of follow-up time is presented in Fig 1 showing that 32% [95% CI: 26; 37] of the population sustain injury over the 24 weeks. Of these, 28 injured runners were excluded since they did not meet the requirements for inclusion to the analyses (Fig 2). Among the remaining 112 injured runners, 82 (73%) were female and 30 (27%) were male, and their mean age was 41.4 years (minimum: 21 years, maximum: 63 years). A total of 1225 injury question- naires were distributed to injured participants of which 1064 (87%) were returned successfully. The most common RRI was MTSS reported among 18 incident cases (16% [95% CI: 9.3; 22.9]). This was followed by AT (n = 10; 8.9% [95% CI: 3.6; 14.2]), PFP (n = 9; 8% [95% CI: 3.0; 13.1]), ITBS (n = 8; 7.1% [95% CI: 2.4; 11.9] and PF (n = 8; 7.1% [95% CI: 2.4; 11.9]. In total, these five diagnoses account for 47% of the injuries. The remaining incident cases were classified within 20 other diagnosis-groups (Table 1). At the end of follow-up 35 participants remained injured. Therefore, a total of 77 incident cases recovered from their RRIs before the end of follow-up and were included in the analyses on time to recovery (Table 2). The overall median time to recovery was 56 days (IQR = 70) regardless the injury diagnoses. In the diagnose-specific recoveries, the shortest median time to recovery was observed among participants sustaining PF with 35 days (IQR = 70). As Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 4 / 11 opposed to this, MTSS had the longest median time to recovery with 70 days (IQR = 89). Eight participants suffered two running-related injuries, and none suffered from three or more inju- ries during follow-up. Discussion During the 24-week follow-up in the Run Clever trial, 32% of the recreational runners sus- tained at least one RRI. Compared with previous research this seems similar to the incidence proportion 25.9% of the novice runners in a study by Buist et al. suffering from a RRI during the 8-week observation period [19]. Moreover, Taunton et al. found an incidence proportion of RRI to be 29.5% during the 13-week training protocol before the Vancouver Sun Run [20]. Finally, in a systematic review on injuries among different types of runners, incidence propor- tions of RRIs were reported in the range between 20% to 80% [6]. However, these differences should be interpreted with caution because of different injury definitions and different dura- tions of follow-up across studies. The overall median time to recovery across RRI diagnoses Fig 1. Kaplan-Meier graph. Kaplan-Meier graph visualizing the proportion of injury-free runners as a function of follow-up time. The results revealed 32% [95% CI: 26; 37] of the runners sustained injury over the 24 weeks. https://doi.org/10.1371/journal.pone.0204742.g001 Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 5 / 11 Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 6 / 11 was 56 days among the recreational runners analyzed. Previously, the median time to recovery among novice runners has been found to exceed 70 days [13]. MTSS was the RRI diagnosis with the highest incidence proportion followed by AT, PFP, ITBS, and PF. Interestingly, these diagnoses are also among the five most common diagnoses found in previous research [12, 13, 21]. Collectively, the five diagnoses accounted for almost half the injuries sustained (47%) in the present study. This is also similar to previous studies revealing these injuries to target 42.6%, 51.8% and 41% of the injured runners, respectively [12,13,21]. Consequently, across various studies it is not uncommon that almost half the RRIs are distributed between these five diagnoses. The RRI diagnosis with the longest recovery time was medial meniscus injury followed by hamstring injury. However, the most incident RRI diagnoses, MTSS, AT, PFP, ITBS and PF in the present study were also among the top 10 RRI with the longest recovery time. A strength of the present study is the weekly status updates, which reduced the risk of recall bias and information problems. Furthermore, the diagnostic approach, encompassing a Fig 2. Flowchart. Flowchart visualizing the flow of runners sustaining injuries during the Run Clever trial. https://doi.org/10.1371/journal.pone.0204742.g002 Table 1. Incident cases, incidence proportion, and characteristics of the 25 different diagnoses of running-related injuries. Injuries are presented in descending order starting with the most frequent. n = number. The total count is presented. Non-recovered injuries are the number of RRIs still sustained at the end of follow-up. The incidence proportion of the injuries and their related confidence interval, CI 95%, are presented in percentages. The distribution of gender is presented as the number of females with each diagnosis. y = years. Furthermore, the mean age, stated in years. Diagnosis Incident cases, n (Non-recovered injuries, n) Incidence proportion in % (95%CI) Gender female (n) Mean Age (y) Medial tibial stress syndrome (MTSS) 18 (5) 16.07 (9.3; 22.9) 13 35 Achilles tendinopathy (AT) 10 (3) 8.93 (3.6; 14.2) 7 43 Patellofemoral pain (PFP) 9 (2) 8.04 (3.0; 13.1) 8 37 Iliotibial band syndrome (ITB) 8 (2) 7.14 (2.4; 11.9) 8 34 Plantar fasciopathy (PF) 8 (3) 7.14 (2.4; 11.9) 5 43 Gastrocnemius injury 8 (1) 7.14 (2.4; 11.9) 2 49 Gluteus medius tendinopathy 7 (2) 6.25 (1.8; 10.7) 7 42 Medial meniscus injury 7 (4) 6.25 (1.8; 10.7) 6 47 Hamstring injury 6 (2) 5.36 (1.2; 9.5) 6 37 Soleus injury 5 (0) 4.46 (0.6; 8.3) 1 47 Ankle distortion 3 (1) 2.68 3 32 Greater Trochanter Bursitis 3 (1) 2.68 3 46 Patellar tendinopathy 3 (1) 2.68 2 31 Quadriceps injury 3 (1) 2.68 1 42 Psoas major injury 2 (1) 1.79 1 46 Peroneus tendinopathy 2 (2) 1.79 1 42 Pes anserine injury 2 (2) 1.79 1 52 Adductor injury 1 (0) 0.89 1 54 External coxa saltans 1 (0) 0.89 1 40 Flexor hallucis longus tendinitis 1 (0) 0.89 1 44 Hallux valgus 1 (0) 0.89 1 40 Mortons neurom 1 (0) 0.89 1 45 Sacroiliac joint injury 1 (0) 0.89 0 43 Lower back injury 1 (1) 0.89 1 49 Stress fracture collum femoris 1 (1) 0.89 1 45 Total 112 (35) 100 82 41 https://doi.org/10.1371/journal.pone.0204742.t001 Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 7 / 11 standardized physical examination performed by a study-specific diagnostic team of physio- therapists, ensured a greater certainty of accurate injury diagnosis as well as exact date of injury occurrence. Very few comparable studies exist, but an interesting finding is the time to recovery among the recreational runners sustaining MTSS of median 70 days. Since, comparable recovery times of 72 days in a study on novice runners [13], 82 days among infantry recruits in the Brit- ish army [22], and 58 days among 15 military recruits from the Royal Dutch army [23] have been reported. However, differences in the populations investigated and definitions of recovery should be considered. The main reason for the discrepancy in definition of injury recovery between the present study and the previous DANORUN study also including runners by Nielsen et. al, stems from the different ways the data was collected [13]. The electronical database facilitated more frequent and standardized follow-up in the Run Clever trial allowing for a better evalua- tion of the levels of pain and symptoms. Furthermore, the altered definition of injury recovery enabled to avoid runners being labeled injury-free though they participated in running with injuries. Table 2. Incident cases recovered and their time to recovery. Diagnoses related to median time to recovery presented in decreasing order. When no median time to recovery is available, number of incident cases recovered is listed in decreasing order. Only recovered injuries are included in the table and the total count of recovered RRIs is presented. Min = minimum time to recovery. Max = maximum time to recovery. Q1 = 25th percentile of time to recovery. Q3 = 75th percentile of time to recovery. Interquartile ranges are presented with minimum and maximum time to recovery as well as breakdown points of 25%; all numbers are represented in days. For diagnosis with only one incident case present, the time to recovery is listed in the “min” category.  = mean time (instead of median time) to recovery presented. Diagnosis Incident cases recovered, n Median time to recovery in days Min Q1 Q3 Max Medial meniscus injury 3 89 70 105 Hamstring injury 4 74 14 140 Medial tibial stress syndrome (MTSS) 13 70 21 37 126 238 Gluteus medius tendinopathy 5 56 42 42 84 91 Iliotibial band syndrome (ITB) 6 56 14 39 105 168 Achilles tendinopathy (AT) 7 56 7 42 207 245 Patellofemoral pain (PFP) 7 49 14 28 91 119 Soleus injury 5 49 14 42 70 70 Gastrocnemius injury 7 49 7 10,5 70 91 Plantar faschiopathy (PF) 5 35 35 35 105 301 Ankle distortion 2 21 28 Quadriceps injury 2 21 70 Greater Trochanter Bursitis 2 35 70 Patellar tendinopathy 2 35 133 Psoas major injury 1 7 Sacroiliac joint injury 1 28 Flexor hallucis longus tendinitis 1 42 External Coxa saltans 1 56 Hallux valgus 1 98 Mortons neurom 1 133 Adductor injury 1 154 Peroneus tendinopathy 0 Pes anserine injury 0 Lower back injury 0 Stress fracture collum femoris 0 Total 77 56 7 35 105 301 https://doi.org/10.1371/journal.pone.0204742.t002 Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 8 / 11 Still, some limitations exist. Firstly, in total, 35 participants did not recover their RRIs before the end of follow-up. For instance, only 3 of the 7 runners with medial meniscal injured recovered. For these three runners, the median time-to-recovery was 89 days. How- ever, if the remaining four runners had been followed until recovery it is likely the case that the median time-to-recovery would have been longer. This underestimation targets many diagnose-specific recovery-times as the proportion of individuals with medial meniscus injury, MTSS, ITB and AT who became injury-free ranged from 42.3%–70%, respectively. Further, comparing time to recovery in the current study, to recovery times from the study by Nielsen et al. [13] a considerable difference in the diagnoses specific maximum values reported becomes evident. A reason for this may be the definition of recovery in the current study including a margin of three consecutive pain-free weeks was different that the one used in other studies. Consequently, an extended follow-up time would have been preferred to reduce the loss of data. Secondly, the diagnostic approach was standardized to reduce the risk of subjective infor- mation bias regarding the diagnosing for which reason every injury was diagnosed on the basis of a physical examination and the injured runner’s anamnesis. Making a diagnosis adher- ing to the guidelines was not always possible, which makes the objectivity less solid. Thirdly, the definition of recovery is complex. The RRI was deemed to be recovered after three succes- sive weeks without any pain during running in the related anatomical site, but no physical examination or test was performed to make sure full recovery was attained. Moreover, the experience of pain might be diverse in different injuries so that the three-week distinction might be undiscriminating. Despite various limitations in the present study, the results may be of interest for both researchers and clinicians dealing with RRIs. The present study is a prospective analysis of data obtained from the Run Clever trial in which information on new injury onset and exact diagnosing were very important and as proper as possible. However, a major drawback was the lack of continually follow-up on the accuracy on the information submitted by the injured participants. Conclusion The cumulative incidence proportion of injured participants in the Run Clever trial was 32%. The injuries were classified across 25 different diagnoses with MTSS, AT, PFP, ITBS and PF as the most frequent ones. Altogether, these five diagnoses accounted for 47% of all injuries. The median time to recovery for all types of injuries was 56 days. MTSS was the diagnosis with the longest median time to recovery of 70 days. Supporting information S1 Dataset. A STATA.dta file. (DTA) Acknowledgments The authors wish to acknowledge the physiotherapists who made a priceless contribution by willingly and free of charge, diagnosing injured participants. The Danish Rheumatism Associ- ation (https://www.gigtforeningen.dk/) provided a DKK 75.000 grant for this study. The funder had no role in study design, data collection and analysis, decision to publish, or prepa- ration of the manuscript. Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 9 / 11 Author Contributions Conceptualization: Rasmus Oestergaard Nielsen, Daniel Ramskov. Data curation: Benjamin Mulvad, Rasmus Oestergaard Nielsen, Martin Lind, Daniel Ramskov. Formal analysis: Benjamin Mulvad, Rasmus Oestergaard Nielsen. Funding acquisition: Daniel Ramskov. Investigation: Martin Lind. Methodology: Rasmus Oestergaard Nielsen. Project administration: Daniel Ramskov. Software: Rasmus Oestergaard Nielsen. Supervision: Martin Lind. Writing – original draft: Benjamin Mulvad. Writing – review & editing: Rasmus Oestergaard Nielsen, Martin Lind, Daniel Ramskov. References 1. Pilgaard M, Rask S editors. Danskernes motions- og sportsvaner 2016 (In Danish). 1st ed. Copenha- gen, Denmark: Danish Institute of Sports Studies; 2016. 2. Nielsen RO, Videbaek S, Hansen M, Parner ET, Rasmussen S, Langberg H. Does running with or with- out diet changes reduce fat mass in novice runners? A 1-year prospective study. J Sports Med Phys Fit- ness 2016 Jan-Feb; 56(1–2):105–113. PMID: 25766050 3. Hespanhol Junior LC, Pillay JD, van Mechelen W, Verhagen E. Meta-Analyses of the Effects of Habitual Running on Indices of Health in Physically Inactive Adults. Sports Med 2015 Jul 16; 45(10):1455–1468. https://doi.org/10.1007/s40279-015-0359-y PMID: 26178328 4. Lee DC, Brellenthin AG, Thompson PD, Sui X, Lee IM, Lavie CJ. Running as a Key Lifestyle Medicine for Longevity. Prog Cardiovasc Dis 2017 Jun–Jul; 60(1):45–55. https://doi.org/10.1016/j.pcad.2017.03. 005 PMID: 28365296 5. Bueno AM, Nielsen RO. Hvad er omfanget af løbeskader i Danmark? (In Danish). Dansk Sportsmedicin 2017; 2(21):42–45. 6. Kluitenberg B, van Middelkoop M, Diercks R, van der Worp H. What are the Differences in Injury Propor- tions Between Different Populations of Runners? A Systematic Review and Meta-Analysis. Sports Med 2015 Aug; 45(8):1143–1161. https://doi.org/10.1007/s40279-015-0331-x PMID: 25851584 7. Koplan JP, Rothenberg RB, Jones EL. The natural history of exercise: a 10-yr follow-up of a cohort of runners. Med Sci Sports Exerc 1995 08; 27(8):1180–1184. PMID: 7476063 8. Ding D, Kolbe-Alexander T, Nguyen B, Katzmarzyk PT, Pratt M, Lawson KD. The economic burden of physical inactivity: a systematic review and critical appraisal. Br J Sports Med 2017 Oct; 51(19):1392– 1409. https://doi.org/10.1136/bjsports-2016-097385 PMID: 28446455 9. Sumilo D, Stewart-Brown S. The causes and consequences of injury in students at UK institutes of higher education. Public Health 2006 Feb; 120(2):125–131. https://doi.org/10.1016/j.puhe.2005.01.018 PMID: 16260012 10. Plugge E, Stewar-Brown S, Knight M, Fletcher L. Injury morbidity in 18-64-year-olds: impact and risk factors. J Public Health Med 2002 Mar; 24(1):27–33. PMID: 11939379 11. Lopes AD, Hespanhol Junior LC, Yeung SS, Costa LO. What are the Main Running-Related Musculo- skeletal Injuries?: A Systematic Review. Sports Med 2012 Oct 1; 42(10):891–905. https://doi.org/10. 2165/11631170-000000000-00000 PMID: 22827721 12. Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, Zumbo BD. A retrospective case- control analysis of 2002 running injuries. Br J Sports Med 2002 Apr; 36(2):95–101. https://doi.org/10. 1136/bjsm.36.2.95 PMID: 11916889 13. Nielsen RO, Ronnow L, Rasmussen S, Lind M. A prospective study on time to recovery in 254 injured novice runners. PLoS One 2014 Jun 12; 9(6):e99877. https://doi.org/10.1371/journal.pone.0099877 PMID: 24923269 Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 10 / 11 14. Videbaek S, Jensen AV, Rasmussen S, Nielsen RO. Do General Medical Practitioners Examine Injured Runners? Int J Sports Phys Ther 2017 Jun; 12(3):450–457. PMID: 28593099 15. Ramskov D, Nielsen RO, Sorensen H, Parner E, Lind M, Rasmussen S. The design of the run Clever randomized trial: running volume, -intensity and running-related injuries. BMC Musculoskelet Disord 2016 Apr 23; 17:177-016-1020-0. 16. Clarsen B, Myklebust G, Bahr R. Development and validation of a new method for the registration of overuse injuries in sports injury epidemiology: the Oslo Sports Trauma Research Centre (OSTRC) over- use injury questionnaire. Br J Sports Med 2013 May; 47(8):495–502. https://doi.org/10.1136/bjsports- 2012-091524 PMID: 23038786 17. Yamato TP, Saragiotto BT, Lopes AD. A consensus definition of running-related injury in recreational runners: a modified Delphi approach. J Orthop Sports Phys Ther 2015 May; 45(5):375–380. https://doi. org/10.2519/jospt.2015.5741 PMID: 25808527 18. Nielsen RO, Parner ET, Nohr EA, SOrensen H, Lind M, Rasmussen S. Excessive progression in weekly running distance and risk of running-related injuries: an association which varies according to type of injury. J Orthop Sports Phys Ther 2014 Oct; 44(10):739–747. https://doi.org/10.2519/jospt.2014.5164 PMID: 25155475 19. Buist I, Bredeweg SW, Bessem B, van Mechelen W, Lemmink KA, Diercks RL. Incidence and risk fac- tors of running-related injuries during preparation for a 4-mile recreational running event. Br J Sports Med 2010 Jun; 44(8):598–604. https://doi.org/10.1136/bjsm.2007.044677 PMID: 18487252 20. Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, Zumbo BD. A prospective study of running injuries: the Vancouver Sun Run "In Training" clinics. Br J Sports Med 2003 06; 37(3):239–244. https://doi.org/10.1136/bjsm.37.3.239 PMID: 12782549 21. Macintyre J, Taunton JE, Clement D. Running injuries: a clinical study of 4,173 cases. Clin J Sport Med 1991; 1:81–87. 22. Sharma J, Greeves JP, Byers M, Bennett AN, Spears IR. Musculoskeletal injuries in British Army recruits: a prospective study of diagnosis-specific incidence and rehabilitation times. BMC Musculoske- let Disord 2015 May 4; 16:106-015-0558-6. 23. Moen MH, Bongers T, Bakker EW, Zimmermann WO, Weir A, Tol JL, et al. Risk factors and prognostic indicators for medial tibial stress syndrome. Scand J Med Sci Sports 2012 Feb; 22(1):34–39. https://doi. org/10.1111/j.1600-0838.2010.01144.x PMID: 20561280 Running-related injuries PLOS ONE | https://doi.org/10.1371/journal.pone.0204742 October 12, 2018 11 / 11

Diagnoses and time to recovery among injured recreational runners in the RUN CLEVER trial.

10-12-2018

Mulvad, Benjamin,Nielsen, Rasmus Oestergaard,Lind, Martin,Ramskov, Daniel

eng

PMC7826783

International Journal of Environmental Research and Public Health Article Can an Incremental Step Test Be Used for Maximal Lactate Steady State Determination in Swimming? Clues for Practice Mário C. Espada 1,2, Francisco B. Alves 3,4, Dália Curto 3, Cátia C. Ferreira 1,5, Fernando J. Santos 1,2,3 , Dalton M. Pessôa-Filho 6,7 and Joana F. Reis 3,4,*



 Citation: Espada, M.C.; Alves, F.B.; Curto, D.; Ferreira, C.C.; Santos, F.J.; Pessôa-Filho, D.M.; Reis, J.F. Can an Incremental Step Test Be Used for Maximal Lactate Steady State Determination in Swimming? Clues for Practice. Int. J. Environ. Res. Public Health 2021, 18, 477. https://doi.org/ 10.3390/ijerph18020477 Received: 11 December 2020 Accepted: 3 January 2021 Published: 8 January 2021 Publisher’s Note: MDPI stays neu- tral with regard to jurisdictional clai- ms in published maps and institutio- nal affiliations. Copyright: © 2021 by the authors. Li- censee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and con- ditions of the Creative Commons At- tribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). 1 Polytechnic Institute of Setúbal, Department of Science and Technology, 2914-514 Setubal, Portugal; mario.espada@ese.ips.pt (M.C.E.); catia.ferreira@ese.ips.pt (C.C.F.); fernando.santos@ese.ips.pt (F.J.S.) 2 Quality of Life Research Centre, 2040-413 Rio Maior, Portugal 3 Faculdade de Motricidade Humana, Universidade de Lisboa, 1499-002 Cruz Quebrada-Dafundo, Portugal; falves@fmh.ulisboa.pt (F.B.A.); daliacurto@fmh.ulisboa.pt (D.C.) 4 Interdisciplinary Center for the Study of Human Performance (CIPER), Faculdade de Motricidade Humana, Universidade de Lisboa, 1499-002 Cruz Quebrada-Dafundo, Portugal 5 Training Optimization and Sports Performance Research Group (GOERD), Faculty of Sport Science, University of Extremadura, 10003 Cáceres, Spain 6 Department of Physical Education, São Paulo State University (UNESP), Bauru 17033-360, Brazil; dalton.pessoa-filho@unesp.br 7 Institute of Bioscience, Graduate Program in Human Development and Technology, São Paulo State University (UNESP), Rio Claro 13506-900, Brazil * Correspondence: joanareis@fmh.ulisboa.pt; Tel.: +351-21-414-9100 Abstract: We aimed to compare the velocity, physiological responses, and stroke mechanics between the lactate parameters determined in an incremental step test (IST) and maximal lactate steady state (MLSS). Fourteen well-trained male swimmers (16.8 ± 2.8 years) were timed for 400 m and 200 m (T200). Afterwards, a 7 × 200-m front-crawl IST was performed. Swimming velocity, heart rate (HR), blood lactate concentration (BLC), stroke mechanics, and rate of perceived exertion (RPE) were measured throughout the IST and in the 30-min continuous test (CT) bouts for MLSS determination. Swimming velocities at lactate threshold determined with log-log methodology (1.34 ± 0.06 m·s−1) and Dmax methodology (1.40 ± 0.06 m·s−1); and also, the velocity at BLC of 4 mmol·L−1 (1.36 ± 0.07) were not significantly different from MLSSv, however, Bland–Altman analysis showed wide limits of agreement and the concordance correlation coefficient showed poor strength of agreement between the aforementioned parameters which precludes their interchangeable use. Stroke mechanics, HR, RPE, and BLC in MLSSv were not significantly different from the fourth repetition of IST (85% of T200), which by itself can provide useful support to daily practice of well-trained swimmers. Nevertheless, the determination of MLSSv, based on a CT, remains more accurate for exercise evaluation and prescription. Keywords: well-trained swimmers; maximal lactate steady state; lactate threshold; continuous test; incremental test; performance markers 1. Introduction Performance enhancement in sport is closely and decisively related to accuracy in identifying exercise intensities domains toward the optimization of daily training. Specifi- cally, the determination of the boundaries (thresholds) that separate the individual training zones can induce optimal adaptations in athletes and should be periodically assessed to evaluate training effects [1]. Over the years, this has been challenging for athletes, coaches, and researchers largely because the time spent in testing procedures may interfere with the training routines, and fundamentally, the high costs with equipment prevents its widespread use. Nevertheless, the combination of high accuracy with less time demanding Int. J. Environ. Res. Public Health 2021, 18, 477. https://doi.org/10.3390/ijerph18020477 https://www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2021, 18, 477 2 of 13 or invasive procedures sustains the drive for continuing research on the applied physiology of sports training [2]. Lactate threshold (LT) was originally defined as the onset of blood lactate above resting values from a blood lactate velocity curve obtained in an incremental step test (IST) [3]. It represents the first increase of a metabolic acidosis and oxidative capacity of athletes and is a strong determinant of performance within populations of similar maximum oxygen uptake [4]. It is also an important reference point when setting training intensities for endurance athletes, having previously been considered by some researchers as a good surrogate of maximal lactate steady state (MLSS) [5], which by itself represents not only a boundary intensity between the heavy and severe intensity domains but also a biomechanical boundary beyond which stoke length (SL) becomes compromised over time [6]. However, multiple definitions and methods to determine LT have arisen in the literature. It has been considered as the initial rise in BLC above rest, the onset of a fixed blood lactate accumulation ranging from 2.0 to 4.0 mmol L−1, or with curve fitting procedures [7]. These methods seem to be correlated, but their relationship with the boundaries that separate the intensity domains are affected by the methodology chosen, without a clear physiological support for the preference of one of them [8]. The swimming velocity at LT (vLT) is one of the most frequently used indices to assess the swimming endurance capacity [9,10] and several methods are utilized for its calculation [11]. More specifically, the log-log methodology (vLTlog-log), the velocity at which lactate increased exponentially when the log blood lactate concentration (BLC) is plotted against the log swimming velocity [12], Dmax methodology (vLTDmax) as the maximal perpendicular distance of the lactate curve from the line connecting the start with the endpoint of the lactate curve [13]. On the other hand, V4 is the the swimming velocity eliciting a lactate concentration of 4 mmol.L−1 through linear interpolation and has been associated with the onset of blood lactate accumulation (OBLA) [14]. More specifically, a 7 × 200-m IST is used to identify aerobic training intensity domains and subsequent changes during a year-round training plan [15]. Heck et al. [5] were pioneers with respect to the MLSS concept, first considered to occur at a fixed BLC of 2.2 mmol L−1 [16], but more often to 4 mmol L−1, or defined as OBLA [5]. Later, it was observed that the absolute BLC at MLSS velocity (MLSSv) varied considerably between individuals and between exercise modalities [17]. However, it is considered to be the best predictor of aerobic endurance performance [18], defined as the highest constant exercise intensity that can be sustained while maintaining equilibrium between the processes of blood lactate accumulation and elimination [1]. It is noteworthy that very recently Jones et al. [19] outlined concerns with the arbitrariness of the definition of, and the procedures for evaluating MLSS, indicating that progress in these fields has been slowed by the disagreement over definitions and procedures, and by a fixation with the behavior of a single biomarker, BLC. The methodology associated with MLSS determination separates it from most other lactate parameters. The major methodological difference is that it requires several exercise bouts of longer duration (30-min) to be performed in different days at a constant exercise intensity, a constant intensity test (CT), a method that is very time-consuming and demand- ing. MLSS is attained when, in a CT lasting at least 30-min, the BLC does not increase more than 1.0 mmol L−1 after the 10th testing minute [20] and Billat et al. [18] stated that to achieve a true MLSS it is necessary to have four or five prolonged exercise sessions of up to 30-min duration. Over the last 10 years, studies related to MLSS in swimming are scarce, namely those using unimpeded swimming (without oxygen uptake breath by breath apparatus), and the large majority present MLSSv below 1.30 m·s−1, which can preclude the application of those results to well-trained or high-level swimmers. Some examples are studies with twelve adult middle-distance and long-distance male swimmers which found a MLSSv of 1.22 ± 0.05 m·s−1 [6], with ten male swimmers and a MLSSv of 1.17 ± 0.11 m·s−1 [21] and with seventeen long-distance swimmers with a MLSSv of 1.09 ± 0.14 m·s−1 [9]. More Int. J. Environ. Res. Public Health 2021, 18, 477 3 of 13 recently, evaluating twenty well-trained competitive swimmers MLSSv was determined at 1.29 ± 0.05 m·s−1 [22] and Nikitakis et al. [23] observed that MLSSv was higher in adolescents compared to children (1.297 ± 0.056 m·s−1 vs. 1.083 ± 0.065 m·s−1). To overtake the time-consuming limitations of the CT, researchers developed an at- tempt to determine MLSS from a BLC curve obtained during a swimming IST [11,24] fact that is somewhat controversial in the scientific community, namely the interchangeably use of ISTs and CTs to provide useful indexes of aerobic potential [25]. Despite LT and MLSS being regularly considered fundamental physiological concepts in sport [17] most of the studies were conducted in the laboratory using ergometers in the context of running and cycling [26]. Therefore, research in swimming is scarce compared to other sports because of the swimming pool constrains. In an individual sport where results and medals are regularly decided by hundredths of a second, the accuracy of training prescription becomes extremely relevant. Additionally, from a practical point of view, swim coaches and sport sci- entists require accurate methods that allow them to evaluate the progress of their swimmers and to fine point the training prescription with minimal training time interference. Thus, it is necessary to have an in-depth understanding of assessment protocols and associated swimmer responses, namely from a physiological and stroke mechanics perspective in order to select the most suitable evaluation protocols and interpret their results. Therefore, the purpose of this study is to determine if the velocities associated with the lactate parameters determined from a single IST is equivalent to MLSSv determined from several CT. Additionally, we intent to ascertain if the stroke and physiological parameters, representative of MLSS, are similar to those obtained during the IST. 2. Materials and Methods 2.1. Study Design Athletes performed a total of four visits to the water training facility within a 10-day period. On the first visit, all athletes provided written consent to participate in this study, as well as performed an anthropometric and body composition evaluation. Afterwards, the athletes performed a maximal 400-m front crawl (T400) in order to use the average velocity between 50 and the 350 m as an estimate of the maximal aerobic velocity (MAV) [27]. After one week, swimmers performed the IST. In days 3 and 4, the continuous swimming velocity MLSS tests (CT) were completed, first at 90% of MAV and in day 4 at 95% of MAV. Figure 1 illustrates the experimental protocol. Figure 1. Schematic representation of the experimental protocol. IST, incremental step test; MLSS, maximal lactate steady state test (continuous test); MAV, maximal aerobic velocity. 2.2. Participants Fourteen male competitive swimmers volunteered for this study (mean ± SD; 16.8 ± 2.8 years, 1.78 ± 0.05 m, 66.5 ± 7.2 kg and 10.2 ± 2.6% of body fat). The in- clusion criteria were: (1) regularly competing at national level for at least three years and (2) time in 400-m front crawl below 4:35-s. The exclusion criteria were: (1) swimmers with <14 years of age; and (2) swimmers injured three months before experimental protocol. Subjects trained regularly at competitive level for at least eight years (seven to eight swim sessions and 3–4 gym sessions per week the months before data collection with a mean Int. J. Environ. Res. Public Health 2021, 18, 477 4 of 13 swimming volume of 40-km per week) and took no drugs or medicine during the study. Mean performance in 400-m front crawl swimming was determined the week before testing (4:22 ± 0:11-s), corresponding to 81% of the short course world record. All swimmers were familiar with the swimming pool exercise testing procedures. The swimmers were instructed to refrain from intense training sessions at least 24 h before the experimental ses- sions and to retain their normal nutritional habits. All subjects or their parents/guardians (when appropriate) signed an informed consent form prior to participation in the research. The study was approved by the local University Ethical Committee in Human Research from São Paulo State University (UNESP—CAAE:02402512.7.0000.5398) and conducted in accordance with the 1975 Declaration of Helsinki. 2.3. Procedures Tests were conducted at similar time of the day (±2 h) for each swimmer in order to minimize the circadian effect on performance [28] and in separate days (with at least 24 h of rest between tests) in a 25-m swimming pool with the water temperature at 28.2 ◦C. Body composition was assessed with Tanita BC-543 (Tokyo, Japan) and all tests were swum in front crawl. A standardize warm-up of 600-m aerobic swim of low to moderate intensity was completed in every testing session. During the IST and CT, swimming velocity was controlled through a visual pacer (TAR. 1.1, GBK-electronics, Aveiro, Portugal), with flashing lights on the bottom of the pool, helping swimmers to keep up the predetermined swimming velocity. Split times over 50-m were determined and used by two investigators positioned at 7.5 and 17.5-m of the swimming pool to control athletes’ swimming pace. Within a 10-day period, each subject was asked to complete the following tests. 2.3.1. Maximal Lactate Steady State Subjects performed, in different days, 30 min constant swimming velocity at 90 and 95% of MAV. Each swimmer was asked to maintain the pre-established swim pace for as long as possible. The test was interrupted when the swimmer could no longer match the required swimming velocity. Each subject was stopped 30-sec every 400-m for blood sample collection determined in fingertip using the Lactate Pro portable analyzer (Arkray, Kyoto, Japan). MLSS was defined as the highest BLC that increased by no more than 1 mmol.L−1 during the final 20-min of a 30-min CT [29]. When this criterion was not accomplished, the test was stopped. MLSSv was the swimming velocity associated with MLSS. Rate of perceived exertion (RPE) was determined in a 6 to 20 scale [30] by verbal indication of the swimmers in the 30-sec stop every 400-m, while blood sample was being collected. According to the proposal of Craig and Pendergast [31], stroke rate (SR) was cal- culated for each cycle using the equation (SR = 60/stroke duration) and expressed in cycles per minute (cycles min−1). Stroke length (SL) was determined with the equation (SL = V/SR/60) and expressed in meters per cycle (m-cycle−1). SR was measured from three stroke cycles taken in the middle of the pool for every 50 m, SR was measured from three stroke cycles taken in the middle of the pool for every 50 m and averaged based on 100-m distance during the last 20-min swim for the CT tests and the last 100 m of each step for the IST. The average of heart rate (HR) values collected during the last 20 min of MLSS were determined with Polar Sport Tester (S410), with frequency every 5 s during tests. 2.3.2. Incremental Step Test Swimmers T200 (performance time in 200-m front crawl) was assessed in formal com- petition with a maximum distance of 2 months for the determination of the IST swimming velocities. Afterwards, in a separate session, swimmers completed 7 × 200-m front crawl IST [10]. All steps started each 5-min, the first one at 70% of T200 and the subsequent with a 5% increment. At rest, immediately after each step and at the end of the tests, RPE and BLC were recorded. HR, SR, and SL were measured throughout the test. Int. J. Environ. Res. Public Health 2021, 18, 477 5 of 13 The BLC values were registered, and the results were plotted against the respective swimming velocities using Lactate-E software [32]. LT was determined according to the log-log methodology (LTlog-log), the velocity at which lactate increased exponentially when the log BLC is plotted against the log swimming velocity [12]. LT was also measured according to Dmax methodology (LTDmax) as the maximal perpendicular distance of the lactate curve from the line connecting the start with the endpoint of the lactate curve [13]. vLTDmax and vLTlog-log were the swimming velocities associated to both LT method- ologies. V4 was considered as the swimming velocity eliciting a lactate concentration of 4 mmol L−1 through linear interpolation [14]. Vmax was assumed as the swimming velocity performed in the last repetition of the IST (100% T200). 2.4. Statistical Analysis The data are expressed as the mean ± standard deviation (SD). The normality of the distributions was assessed with the Shapiro–Wilk test, parametric statistical procedures were selected. Linear regression models between swimming velocities in continuous test (MLSSv) and IST (V4, vLTDmax and vLTlog-log) were computed. Trendline equation, determination coefficient (R2), and standard error of estimation (SEE) were calculated. Comparisons of swimming velocities were evaluated using standardized differences with combined variance, derived from the M and SD of each variable, with 95% confidence intervals. Paired-samples t-test was used to compare swimming performance markers using standardized differences with combined variance, derived from the M and SD of each variable, with 95% confidence intervals. The statistical limits for the effect sizes Cohen’s d [33] were trivial (0–0.2), small (0.2–0.6), moderate (0.6–1.2), large (1.2–2), very large (2–4), and extremely large (>4) [34]. The variance analysis (ANOVA) was used to verify the differences between swimming velocities, the magnitude of the differences was evaluated by eta square (small 0.01 ≤ η2 p < 0.06), moderate (0.06 ≤ η2 p < 0.15), or large (η2 p ≥ 0.15) [33]. The post-hoc Bonferroni test was also performed in order to verify which pairs of means were significantly different (p < 0.05). Bland–Altman plot [35] was used to assess the agreement between MLSSv, vLTlog-log, vLTDmax, and V4 showing the bias and the limits of agreement. Also, the concordance correlation coefficient (CCC) was performed using the Lin [36] approach with MedCalc® v11.1.1.0 (2009) software. The CCC (ρc) contains a measurement of precision ρ and accuracy (ρc = ρ Cb): where ρ is the Pearson correlation coefficient, which measures how far each observation deviates from the line of best-fit and is a measure of precision, and Cb is a bias correction factor that measures how far the best-fit line deviates from the 45◦ line through the origin and is a measure of accuracy. Data analysis was performed using the Statistical Package for Social Sciences (SPSS 25.0, SPSS. Inc., Chicago, IL, USA). 3. Results All the fourteen swimmers were able to perform the 30-min constant swimming at 90% of MAV within the criteria established to assume the MLSS and stopped their CTs at an intensity equal to 95% of MAV because of exhaustion. MLSSv (1.36 ± 0.06 m·s−1), was significantly lower compared to Vmax (1.53 ± 0.07 m·s−1; 112% MLSSv) and MAV (1.51 ± 0.07 m·s−1; 111% MLSSv). Vmax and MAV were not significantly different (p > 0.05). vLTDmax (1.40 ± 0.06 m·s−1; 103% MLSSv), V4 (1.36 ± 0.07 m·s−1; 100% MLSSv), and vLTlog-log (1.34 ± 0.06 m·s−1; 99% MLSSv) were not significantly different to MLSSv. In Table 1 is presented the comparative analysis between the different swimming velocities. Int. J. Environ. Res. Public Health 2021, 18, 477 6 of 13 Table 1. Comparative analysis between different swimming velocities. Anova One-way Tests M ± SD F p η2p Post-Hoc Test Bonferroni Vmax (m·s−1) 1.53 ± 0.07 22.20 0.00 0.58 MAV (1.000); vLTDmax (0.000); MLSSv (0.000); V4 (0.000); vLTlog-log (0.000) MAV (m·s−1) 1.51 ± 0.07 Vmax (1.000); vLTDmax (0.000); MLSSv (0.000); V4 (0.000); vLTlog-log (0.000) vLTDmax (m·s−1) 1.40 ± 0.06 Vmax (0.000); MAV (0.000); MLSSv (1.000); V4 (1.000); vLTlog-log (0.452) MLSSv (m·s−1) 1.36 ± 0.06 Vmax (0.000); MAV (0.000); vLTDmax (1.000); V4 (1.000); vLTlog-log (1.000) V4 (m·s−1) 1.36 ± 0.07 Vmax (0.000); MAV (0.000); vLTDmax (1.000) MLSSv (1.000); vLTlog-log (1.000) vLTlog-log (m·s−1) 1.34 ± 0.06 Vmax (0.000); MAV (0.000); vLTDmax (0.452) MLSSv (1.000); V4 (1.000) M ± SD, mean ± standard deviation; Vmax, swimming velocity performed in the last repetition of the incremental step test; MAV, maximal aerobic velocity; vLTDmax, swimming velocity associated to lactate threshold determined from Dmax methodology; MLSSv, swimming velocity associated to maximal lactate steady state; V4, swimming velocity eliciting a lactate concentration of 4 mmol.L−1; vLTlog-log, swimming velocity associated to lactate threshold determined from log-log methodology. Note: Anova one-way F, p and η2 p related to all swimming velocities. Significant differences between swimming velocities (p < 0.05) are observed with post-hoc test. Regression analysis between MLSSv and V4 revealed adjusted r2 value of 0.81 with a SEE of 0.027, in spite of standardized residuals remaining within the 95% confidence interval limits, indicating a fairly good estimation model. The linear regressions with r2 and SEE values between MLSSv, V4, vLTlog-log, and vLTDmax are presented in Figure 2. Figure 2. Linear regression of MLSSv on V4, vLTlog-log, and vLTDmax with standard error of estimate (SEE). The agreement between MLSSv and V4 is shown in Figure 3. The 95% limits of agreement (Loa) ranged from −0.059 to 0.066. Although the bias was 0.004 and there was no relation between the difference and the mean of the parameters there were somewhat wide limits of agreement (±5.1%). The bias between MLSSv and vLTlog-log was −0.016 and the Loa ranged between −0.039 and 0.072, representing a variation of ±4.1% of MLSSv. There was not a significant trend between the difference and the mean of the two measures. The MLSSv was overestimated by the vLTDmax with a bias of −0.038 with somewhat wide Loa ranging between −0.092 and 0.017, representing a variation of ±4.1% of MLSSv. There was not a significant trend between the difference and the mean of the two measures. Int. J. Environ. Res. Public Health 2021, 18, 477 7 of 13 Figure 3. Bland-Altman plot showing the bias and limits of agreement between MLSSv, vLTDmax, and vLTlog-log. CCC of the methods are shown in Table 2, where <0.90 indicates a poor strength of agreement between the methods. Table 2. Concordance correlation coefficient between MLSSv and the three lactate indexes determined in the incremental step test. CCC Precision Accuracy V4 (m·s−1) 0.88 0.90 0.98 vLTlog-log (m·s−1) 0.86 0.89 0.96 vLTDmax (m·s−1) 0.74 0.90 0.83 CCC, concordance correlation coefficient; V4, swimming velocity eliciting a lactate concentration of 4 mmol.L−1; vLTlog-log, swimming velocity associated to lactate threshold determined from log-log methodology; vLTDmax, swimming velocity associated to lactate threshold determined from Dmax methodology. In IST, it was observed that swimmers tend to neglect SL with the purpose of maintaining the pre-establish swimming velocity throughout all the 7 × 200-m repeti- tions. Above 85% T200, a stroking efficiency breakpoint was observed in all well-trained swimmers, SR and SL in the fourth repetition in the IST (85% of T200) (respectively 33.88 ± 3.89 cycles min−1 and 2.50 ± 0.32 m cycle−1) were significantly different (p < 0.01) compared to the fifth repetition, at 90% T200 (respectively 36.77 ± 3.20 cycles min−1 and 2.40 ± 0.23 m cycle−1). SL in the third IST repetition (80% of T200 = 2.59 ± 0.31 m cycle−1) was significantly higher compared to the fourth and also SL at MLSSv (2.54 ± 0.33 m cycle−1; p < 0.01). 85% T200 (1.36 ± 0.05 m·s−1) was not significantly different from MLSSv (p > 0.05), ES was trivial (0.07), and a close relationship between stroke and physiological markers was observed between both, presented in Table 3. Table 3. Mean and standard deviation of performance markers during MLSS test and fourth repetition in the incremental step test, at 85% T200. Variable MLSSv 85% T200 t p Cohen’s d HR (beats.min−1) 174.2 ± 7.0 169.5 ± 6.2 1.841 0.077 −0.70 SR (cycles.min−1) 32.76 ± 4.07 33.88 ± 3.89 −0.749 0.461 0.28 SL (m.cycle−1) 2.54 ± 0.33 2.50 ± 0.32 0.269 0.790 −0.10 BLC (mmol.L−1) 4.84 ± 1.53 4.83 ± 0.94 0.015 0.988 −0.01 RPE (6–20 scale) 13.50 ± 1.50 13.28 ± 0.72 0.479 0.637 −0.18 MLSSv, swimming velocity associated to maximal lactate steady state; T200, performance time in 200-m front crawl; HR, heart rate; SR, stroke rate; SL, stroke length; BLC, blood lactate concentration; RPE, rate of perceived exertion. ES are considered trivial (0–0.2), small (0.2–0.6), moderate (0.6–1.2), large (1.2–2), very large (2–4) and extremely large (>4) (Cohen’s d). In all cases p > 0.05 represent no significant statistical differences. Int. J. Environ. Res. Public Health 2021, 18, 477 8 of 13 Although not significantly different, in HR the p value and Cohen’s d revealed values close to significant differences. MLSS ranged between 2.6 and 7.1 mmol L−1 and mean LTD-max (5.1 ± 0.7 mmol L−1; range 3.9 and 6.2) was not significantly different from MLSS, however, LTlog-log (3.8 ± 0.7 mmol.L−1; range 2.6 and 4.5) was lower (p < 0.01). Mean RPE during MLSS test was 13.5 ± 1.5, associated to “somewhat hard” with values ranging from 11 to 16 and in the fourth repetition in IT (85% of T200) from 12 to 15. 4. Discussion The purpose of this study was to determine if the velocities associated with different lactate parameters determined from IST are equivalent to MLSSv determined from a CT. Additionally, we intent to ascertain if the stroke and physiological parameters, represen- tative of MLSS, are similar to those obtained during the IST. The first main finding in the present study was that MLSS can be determined in well-trained swimmers with only two to three attempts of 30-min constant swimming velocity performed in different days assuming the first swimming test at 90% MAV, the swimming velocity at which athletes participating in our study achieved the MLSS. Second, although it is interesting to speculate if an IST provides reliable indicators of MLSS, we verified that from a practical perspective, daily training, some outputs may be useful for swimmers and coaches, but an accurate evaluation of MLSS is only possible through the traditional methodology, a CT. Although there was not a significant statistical difference between MLSSv, V4, vLTlog-log, and vLTDmax, there was a poor CCC (<0.90) between the MLSSv and the other three lactate indexes, which precludes the use of these measures interchangeably. V4 did not present a bias when compared with MLSSv determined in a CT, however, Bland–Altman analysis showed somewhat wide limits of agreement (±5.1%), which in a practical point of view can represent meaningful differences. For example, a swimmer with a MLSSv of 1.36 m·s−1 can have a V4 of 1.41 or 1.32 m·s−1, which represents a difference of 4 sec each 100 m. Regarding vLTlog-log, it underestimated MLSSv by 0.02 m·s−1 with limits of agreement of ±4.1%. Conversely vLTDmax was consistently higher than MLSSv by 0.04 m·s−1, with Bland–Altman analysis also showing limits of agreement of ±4.1%. Previously, a MLSSv of 1.22 ± 0.09 m·s−1, representing 88.9 ± 3.3% of MAV, was found in eleven male well-trained competitive swimmers [37]. Also, Baron et al. [29] in ten well- trained competitive swimmers showed that MLSS corresponded to the velocity a swimmer spontaneously chooses during the first 15 min of a 2-h test. These authors also observed a MLSSv of 1.22 ± 0.14 m·s−1, corresponding to 86.5 ± 5.1% of MAV (1.41 ± 0.12 m·s−1). Later, Espada, and Alves [38] also observed a MLSSv of 1.34 ± 0.06 m·s−1 corresponding to 89.7 ± 1.7% of MAV. However, another study conducted with twelve middle-distance and long-distance male swimmers showed that the stroke parameters and BLC were significantly different between MLSSv (1.22 ± 0.05 m·s−1; 88.6 ± 1.1% of MAV) and 102.5% MLSSv (1.25 ± 0.04 m·s−1; 91.3 ± 1.1% MAV) [6]. In the present study, swimmers were able to perform 30-min at 90% of MAV, which can be related with the level of the swimmers participating in our study since to our best knowledge this is the first study to compare the speed, physiological and stroke parameters determined from the IST and CT methods in well-trained swimmers with MLSSv above 1.35 m·s−1. However, in previous research some swimmers were able to complete 30-min at 90% MAV, although, without meeting MLSS criteria. For example Dekerle et al. [37] reported that five swimmers could complete the 30-min swim but increased their BLC values by more than 1 mmol L−1 between the 10th min (4.4 ± 1.6 mmol L−1) and the 30th min (5.9 ± 1.9 mmol L−1) and in Pelarigo et al. [6] study, the 102.5% MLSS intensity was maintained without exhaustion in the 30-min test but the criteria to be considered MLSS was also not accomplished. Int. J. Environ. Res. Public Health 2021, 18, 477 9 of 13 Lower MLSS values were found in studies conducted with swimmers, which can be attributed to the lower MLSSv values compared to our study. Also lower values were evident in rowers (3.1 ± 0.5 mmol L−1) but higher in cyclists (5.4 ± 1.0 mmol L−1) and speed skaters (6.6 ± 0.9 mmol L−1) [17]. In cycling, Van Schuylenbergh et al. [39] found a significant correlation between MLSS workload and V4, which led to the indication that LT (determined from Dmax methodology) was closely correlated with MLSS power (r = 0.72). Though, these authors point out that the validity of MLSS predicted from an IST must be verified by a 30-min constant load test. In swimming, a significant correla- tion between LT and V4 (r = 0.90; p < 0.01) [40] was found and a study conducted with five male long-distance swimmers and eight triathletes showed that vLT (determined from Dmax methodology), although higher, was not significantly different than MLSSv (1.18 ± 0.08 m·s−1 and 1.13 ± 0.08 m·s−1, respectively) [41]. Conversely, Fernandes et al. [9] determined V4 and V8 (swimming velocity at 8 mmol.L−1) by linear interpolation or ex- trapolation of the lactate BLC vs. velocity curve in a IST (respectively 1.20 ± 0.15 and 1.30 ± 0.17 m·s−1), with both being significantly higher than MLSSv (1.09 ± 0.14 m·s−1) in seventeen long-distance swimmers. Although previous research reported that the 7 × 200 protocol is sensible to different training regimens [42] our results seem to confirm that the parameters derived from this protocol cannot be used interchangeably with the MLSS gold standard determination protocol. Furthermore, our results also confirm that it is impossible to link the true MLSS to a fixed lactate concentration as it was previously pointed [5], because MLSS ranged from 2.6 to 7.1 mmol.L−1. It should be noted that MLSS in swimming is affected by brief interruptions in exercise that are necessary for blood sampling. On the other hand, it was recently indicated that 30 to 45-sec passive recovery between 10 × 200-m swimming repetitions enables steady BLC, oxygen uptake and HR similar to MLSS [23]. However, we must acknowledge that the dynamic interaction between the rates of muscle lactate production, lactate efflux from muscle to blood, and lactate clearance/metabolism both within muscle and from the blood by other organs [43], means that a steady-state in BLC need not imply the existence of a bioenergetic steady-state in contracting skeletal muscle [19]. These authors indicated that BLC per se, is neither an appropriate nor a sufficiently sensitive metric to enable a confident assessment of whether a specific velocity or power output may be sustainable in a metabolic steady-state muscle. Pelarigo et al. [6], found a MLSSv of 1.22 ± 0.05 m·s−1 and a T200 of 1.45 ± 0.05 m·s−1 which means that MLSSv corresponded to around 84% of T200. Additionally, another study indicated that the decrease in SL started above 85% of MAV [37] and Fernandes et al. [9] observed that SR was different across the 7 × 200-m particularly after the 4th repetition. It was previously stressed that MLSS could represent an intensity to develop aerobic endurance and perform technical work of very-high-standard quality [29,37], fact that was confirmed by Pelarigo et al. [6] research where the MLSS (3.28 ± 0.97 mmol.L−1) was significantly lower than BLC at 102.5% MLSSv (4.59 ± 1.36 mmol.L−1) and the SR was maintained in MLSSv between 10th and 30th minute and significantly increased at 102.5% MLSS, contrary to SL, maintained during the 30 min swam at MLSS but significantly decreased at 102.5% MLSSv. Our study revealed that during IST, the SR and SL at 85% of T200, the fourth repetition, were closely related to those observed at MLSS throughout the 30 min CT and tend to repre- sent a boundary of the swimming efficiency, showing that the transition from the heavy to the severe intensity domain is not only related to swimming velocity, RPE and BLC, but also to stroke parameters. The values we measured in MLSS (SR 32.8 ± 4.1 cycles.min−1/SL 2.54 ± 0.33 m.cycle−1) in well-trained swimmers were also higher compared to previous studies, such as Dekerle et al. [37] (27.7 ± 2.2 cycles.min−1; 2.64 ± 0.32 m.cycle−1) or Pelarigo et al. [6] (30.9 ± 3.4 cycles.min−1; 2.47 ± 0.2 m.cycle−1), fact that we consider associated to the level of swimmers participating in the different studies. Int. J. Environ. Res. Public Health 2021, 18, 477 10 of 13 Baron et al. [44] verified that during exercise performed at MLSS, exhaustion occurred while physiological reserve capacity still existed, but in association with an increase in the RPE, as predicted by the central governor model. This research team added that exercise termination may be induced by an integrative homoeostatic control of the peripheral phys- iological system specifically to ensure the maintenance of homeostasis. Demello et al. [45] indicated that LT occurs at a feeling of “somewhat hard” and “hard” from the RPE per- spective, with values ranging from 12.9 to 13.6. Our results are in line with these previous observations, although the CT values ranged from 11 to 16 and in the 4th repetition, in IST (85% of T200) from 12 to 15, fact that led us to agree with previous literature stating that the use of an absolute RPE value to prescribe exercise intensity is unwise [46] because of the fairly large between-subject variability. Also, Potteiger and Weber [47] investigated RPE during incremental and constant intensity exercise and concluded that RPE cannot be used as a particularly accurate marker of exercise intensity. This study presents some limitations since we did not consider the maturation of the swimmers or their distance specialty, which may be relevant to the training methodological framework. We also did not evaluate the potential differences regarding sexes since we only tested male athletes, neither compared results in short- and long-course swimming pools, factors that make it impossible to generalize our results to the whole swimming community. Tracking individual responses during the swimming process is crucial for training prescription and adjustments as inter-individual differences are significant in well- trained athletes. Future research should consider different swimmers’ level, gender, and the comparison between CT and IST in long-course swimming pool (50-m), with swimming velocities increment of 2.5%, as implemented by Pessôa-Filho et al. [48]. The possible measurement of gas exchange could be useful to better understand the physiological, metabolic, and stroke parameters pathways associated to CT and IST, as well as more accurately measure LT and MLSS, understanding the athletes holistic fatigue associated to the gold standard, not only from the BLC perspective. 5. Conclusions From a practical application perspective, the main findings of this study show that in well-trained swimmers MLSSv can be estimated with a maximum two to three 30-min constant swimming bouts performed in different days starting at 90% of MAV and elevating or decreasing that exercise intensity in subsequent bouts. This procedure is more practical and less time consuming than protocols previously suggested which could be important for coaches and athletes in daily swimming practice. Moreover, the 90% of MAV or 85% of T200 may be considered aerobic power zones, where high-quality technique training may occur. In fact, stroke parameters achieved during MLSSv are very closely related to the fourth repetition of the IST and represent not only a physiological, but a mechanical boundary above which athletes achieve fatigue and the swimming technique starts to deteriorate. Performing an IST in well-trained swimmers can be very useful for practice because it provides several useful indicators, nevertheless, this methodology should be used with caution. Also, the level of the swimmers seems to decisively influence the obtained results, this fact should be carefully examined in future research. V4, vLTDmax, and vLTlog-log were not statistically different from MLSSv, nevertheless, it is our understanding that both the IST and CT in well-trained swimmers provide useful indexes of aerobic potential, but they cannot be used interchangeably for MLSSv determi- nation. The direct determination of MLSSv, with CT testing procedure, remains the more accurate for evaluation and prescription of swimming training and for research purposes. Int. J. Environ. Res. Public Health 2021, 18, 477 11 of 13 Author Contributions: Conceptualization: M.C.E., J.F.R. and F.B.A.; methodology: M.C.E., J.F.R. and F.B.A.; formal analysis, F.B.A.; investigation, M.C.E., J.F.R. and F.B.A.; supervision: F.B.A. and J.F.R.; data curation: M.C.E., D.C., J.F.R., C.C.F. and F.J.S.; writing—original draft preparation: M.C.E., J.F.R., C.C.F. and D.M.P.-F.; writing—review and editing: M.C.E., D.C., J.F.R., C.C.F. and D.M.P.-F.; visualization: F.B.A.; funding acquisition: M.C.E., F.J.S. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by Foundation for Science and Technology, I.P., Grant/Award NumberUIDB/04748/2020. M.C.E. also acknowledge the financial support from Polytechnic Institute of Setúbal. D.M.P.F. would like to thank São Paulo Research Foundation—FAPESP (PROCESS 2016/04544-3) for the partial support. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the local University Ethical Committee in Human Research from São Paulo State University (UNESP—CAAE:02402512.7.0000.5398/2013). Informed Consent Statement: Informed consent was obtained from all subjects or their parents/ guardians (when appropriate) involved in the study. Data Availability Statement: The data that support the findings of this study are available from the corresponding and first authors (joanareis@fmh.ulisboa.pt and mario.espada@ese.ips.pt), upon reasonable request. Acknowledgments: We would like to express our gratitude to the swimmers for their time and effort and the swimming teams for making both their infrastructures and staff available for the study. Conflicts of Interest: The authors declare no conflict of interest. References 1. Jones, A.M.; Carter, H. The effect of endurance training on parameters of aerobic fitness. Sports Med. 2000, 29, 373–386. [CrossRef] [PubMed] 2. Espada, M.C.; Reis, J.F.; Almeida, T.F.; Bruno, P.M.; Vleck, V.E.; Alves, F.B. Ventilatory and Physiological Responses in Swimmers Below and Above Their Maximal Lactate Steady State. J. Strength Cond Res. 2015, 29, 2836–2843. [CrossRef] [PubMed] 3. Wasserman, K.; Whipp, B.J.; Koyl, S.N.; Beaver, W.L. Anaerobic threshold and respiratory gas exchange during exercise. J. Appl. Physiol. 1973, 35, 236–243. [CrossRef] [PubMed] 4. Coyle, E.F. Integration of the physiological factors determining endurance performance ability. Exerc. Sport Sci. Rev. 1995, 23, 25–63. [CrossRef] [PubMed] 5. Heck, H.; Mader, A.; Hess, G.; Mucke, S.; Muller, R.; Hollmann, W. Justification of the 4 mmol−1 lactate threshold. Int. J. Sports Med. 1985, 6, 117–130. [CrossRef] 6. Pelarigo, J.G.; Denadai, D.S.; Greco, C.C. Stroke phases responses around maximal lactate steady state in front crawl. J. Sci. Med. Sport 2011, 14, 168.e1–168.e5. [CrossRef] 7. Jamnick, N.A.; Botella, J.; Pyne, D.B.; Bishop, D.J. Manipulating graded exercise test variables affects the validity of the lactate threshold and VO2peak. PLoS ONE 2018, 30, e0199794. 8. Poole, D.C.; Rossiter, H.B.; Brooks, G.A.; Gladden, L.B. The anaerobic threshold: 50+ years of controversy. J. Physiol. 2020. [CrossRef] 9. Fernandes, R.J.; Sousa, M.; Machado, L.; Vilas-Boas, J.P. Step length and individual anaerobic threshold assessment in swimming. Int. J. Sports Med. 2011, 32, 940–946. [CrossRef] 10. Pyne, B.D.; Lee, H.E.; Swanwick, K.M. Monitoring the lactate threshold in world ranked swimmers. Med. Sci. Sports Exerc. 2001, 33, 291–297. [CrossRef] 11. Faude, O.; Kindermann, W.; Meyer, T. Lactate threshold concepts. How valid are they. Sports Med. 2009, 39, 469–490. [CrossRef] 12. Beaver, W.L.; Wasserman, K.W.; Whipp, B.J. Improved detection of lactate threshold during exercise using a log-log transformation. J. Appl. Physiol. 1985, 59, 1936–1940. [CrossRef] [PubMed] 13. Cheng, B.; Kuipers, H.; Snyder, A.C.; Jeukendrup, A.; Hesselink, M. A new approach for the determination of ventilatory and lactate thresholds. Int. J. Sports Med. 1992, 13, 518–522. [CrossRef] [PubMed] 14. Sjodin, B.; Jacobs, I. Onset of blood lactate accumulation and marathon running performance. Int. J. Sports Med. 1981, 2, 23–26. [CrossRef] [PubMed] 15. Arsoniadis, G.G.; Botonis, P.G.; Nikitakis, S.I.; Kalokiris, D.; Toubekis, A.G. Effects of successive annual training on aerobic endurance indices in young swimmers. Open Sport Sci. J. 2017, 10, 214–221. [CrossRef] Int. J. Environ. Res. Public Health 2021, 18, 477 12 of 13 16. Priest, J.W.; Hagan, R.D. The effects of maximum steady state pace training on running performance. Br. J. Sports Med. 1987, 21, 18–21. [CrossRef] 17. Beneke, R.; von Duvillard, S.P. Determination of maximal lactate steady state in selected sport events. Med. Sci. Sports Exerc. 1996, 28, 241–246. [CrossRef] 18. Billat, V.L.; Sirvent, P.; Py, G.; Koralsztein, J.P.; Mercier, J. The concept of maximal lactate steady state: A bridge between biochemistry, physiology and sport science. Sports Med. 2003, 33, 407–426. [CrossRef] 19. Jones, A.M.; Burnley, M.; Black, M.I.; Poole, D.C.; Vanhatalo, A. The maximal metabolic steady state: Redefining the ‘gold standard’. Physiol. Rep. 2019, 7, e14098. [CrossRef] 20. Beneke, R. Methodological aspects of maximal lactate steady state–implications for performance testing. Eur. J. Appl. Physiol. 2003, 89, 95–99. [CrossRef] 21. Pessôa Filho, D.M.; Greco, C.C.; Denadai, B.S. Tether-power at maximal lactate steady-state and endurance indexes of swimming performance. Rev. Bras. Med. Esporte 2014, 20, 359–365. [CrossRef] 22. Machado, M.V.; Borges, J.P.; Galdino, I.S.; Cunha, L.; Sá Filho, A.S.; Soares, D.C.; Andries Junior, O. Does critical velocity represent the maximal lactate steady state in youth swimmers? Sci. Sports 2019, 34, e209–e215. [CrossRef] 23. Nikitakis, I.S.; Paradisis, G.P.; Bogdanis, G.C.; Toubekis, A.G. Physiological Responses of Continuous and Intermittent Swimming at Critical Speed and Maximum Lactate Steady State in Children and Adolescent Swimmers. Sports 2019, 7, 25. [CrossRef] [PubMed] 24. Arratibel-Imaz, I.; Calleja-González, J.; Emparanza, J.I.; Terrados, N.; Mjaanes, J.M.; Ostojic, S.M. Lack of concordance amongst measurements of individual anaerobic threshold and maximal lactate steady state on a cycle ergometer. Phys. Sportsmed. 2016, 44, 34–45. [CrossRef] [PubMed] 25. Borszcz, F.K.; Ferreira, A.T.; Costa, V.P. Is the Functional Threshold Power Interchangeable With the Maximal Lactate Steady State in Trained Cyclists? Int. J Sports Physiol. Perform. 2019, 14, 1029–1035. [CrossRef] [PubMed] 26. Jakobsson, J.; Malm, C. Maximal Lactate Steady State and Lactate Thresholds in the Cross-Country Skiing Sub-Technique Double Poling. Int. J. Exerc. Sci. 2019, 12, 57–68. 27. Lavoie, J.M.; Montpetit, R.R. Applied physiology of swimming. Sports Med. 1986, 3, 165–189. [CrossRef] 28. Atkinson, G.; Reilly, T. Circadian variation in sports performance. Sports Med. 1996, 21, 292–312. [CrossRef] 29. Baron, B.; Dekerle, J.; Depretz, S.; Lefevre, T.; Pelayo, P. Self selected speed and maximal lactate steady state in swimming. J. Sports Med. Phys. Fit. 2005, 45, 1–6. 30. Borg, G.A. Psychophysical bases of perceived exertion. Med. Sci. Sports Exerc. 1982, 14, 377–381. [CrossRef] 31. Craig, A.; Pendergast, D. Relationships of stroke rate, distance per stroke, and velocity in competitive swimming. Med. Sci. Sports 1979, 11, 278–283. [CrossRef] [PubMed] 32. Newell, J.; Higgins, D.; Madden, N.; Cruickshank, J.; Einbeck, J.; McMillan, K.; McDonald, R. Software for calculating blood lactate endurance. J. Sports Sci. 2007, 25, 1403–1409. [CrossRef] [PubMed] 33. Cohen, J. Statistical Power Analysis for the Behavioral Sciences, 2nd ed.; Lawrence Erlbaum Associates: Hillsdale, NJ, USA, 1988. 34. Hopkins, W.G.; Marshall, S.W.; Batterham, A.M.; Hanin, J. Progressive Statistics for Studies in Sports Medicine and Exercise Science. Med. Sci. Sports Exerc. 2009, 41, 3–13. [CrossRef] [PubMed] 35. Bland, J.M.; Altman, D.G. Statistical methods for assessing agreement between two methods of clinical measurements. Lancet 1986, 1, 307–310. [CrossRef] 36. Lin, L.I. A concordance correlation coefficient to evaluate reproducibility. Biometrics 1989, 45, 255–268. [CrossRef] 37. Dekerle, J.; Nesi, X.; Lefevre, T.; Depretz, S.; Sidney, M.; Marchand, F.H.; Pelayo, P. Stroking parameters in front crawl swimming and maximal lactate steady state speed. Int. J. Sports Med. 2005, 26, 53–58. [CrossRef] [PubMed] 38. Espada, M.; Alves, F. Critical velocity and the velocity at maximal lactate steady state in swimming. In Biomechanics and Medicine in Swimming XI; Kjendlie, P.L., Stallman, R.K., Cabri, J., Eds.; Norwegian School of Sports Sciences: Oslo, Norway, 2010; pp. 194–196. 39. Van Schuylenbergh, R.; Vanden Eynde, B.; Hespel, P. Correlations between lactate and ventilator thresholds and the maximal lactate steady state in elite cyclists. Int. J. Sports Med. 2004, 25, 403–408. [CrossRef] 40. Toubekis, A.G.; Tsami, A.P.; Tokmakidis, S.P. Critical Velocity and Lactate Threshold in Young Swimmers. Int. J. Sports Med. 2006, 27, 117–123. [CrossRef] 41. Oliveira, M.F.; Caputo, F.; Lucas, R.D.; Denadai, B.S.; Greco, C.C. Physiological and stroke parameters to assess aerobic capacity in swimming. Int. J. Sports Physiol. Perform. 2012, 7, 218–223. [CrossRef] 42. Pugliese, L.; Porcelli, S.; Bonato, M.; Pavei, G.; La Torre, A.; Maggioni, M.A.; Bellistri, G.; Marzorati, M. Effects of manipulating volume and intensity training in masters swimmers. Int. J. Sports Physiol. Perform. 2015, 10, 907–912. [CrossRef] 43. Stainsby, W.N.; Brooks, G.A. Control of lactic acid metabolism in contracting muscles and during exercise. Exerc. Sport Sci. Rev. 1990, 18, 29–63. [CrossRef] [PubMed] 44. Baron, B.; Noakes, T.D.; Dekerle, J.; Moullan, F.; Robin, S.; Matran, R.; Pelayo, P. Why does exercise terminate at the maximal lactate steady state intensity? Br. J. Sports Med. 2008, 42, 828–833. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2021, 18, 477 13 of 13 45. Demello, J.; Cureton, K.J.; Robin, J.; Boineau, E. Ratings of perceived exertion at lactate threshold in trained and untrained men and women. Med. Sci. Sports Exerc. 1987, 19, 354–362. [CrossRef] [PubMed] 46. Grant, S.; McMillan, K.; Newell, J.; Wood, L.; Keatley, S.; Simpson, D.; Leslie, K.; Fairlie-Clark, S. Reproducibility of the blood lactate threshold, 4 mmol.l−1 marker, heart rate and ratings of perceived exertion during incremental treadmill exercise in humans. Eur. J. Appl. Physiol. 2002, 87, 159–166. [PubMed] 47. Potteiger, J.A.; Weber, S.F. Rating of perceived exertion and heart rate as indicators of exercise intensity in different environmental temperatures. Med. Sci. Sports Exerc. 1994, 26, 791–796. [CrossRef] 48. Pessôa Filho, D.M.; Alves, F.B.; Reis, J.F.; Greco, C.C.; Denadai, B.S. VO2 kinetics during heavy and severe exercise in swimming. Int. J. Sports Med. 2012, 33, 744–748. [CrossRef]

Can an Incremental Step Test Be Used for Maximal Lactate Steady State Determination in Swimming? Clues for Practice.

01-08-2021

Espada, Mário C,Alves, Francisco B,Curto, Dália,Ferreira, Cátia C,Santos, Fernando J,Pessôa-Filho, Dalton M,Reis, Joana F

eng

PMC4892259

rsif.royalsocietypublishing.org Research Cite this article: Rankin JW, Rubenson J, Hutchinson JR. 2016 Inferring muscle functional roles of the ostrich pelvic limb during walking and running using computer optimization. J. R. Soc. Interface 13: 20160035. http://dx.doi.org/10.1098/rsif.2016.0035 Received: 14 January 2016 Accepted: 7 April 2016 Subject Category: Life Sciences–Engineering interface Subject Areas: biomechanics Keywords: musculoskeletal model, inverse dynamics, forward dynamics, OpenSim, static optimization, computed muscle control Author for correspondence: Jeffery W. Rankin e-mail: jrankin@rvc.ac.uk Electronic supplementary material is available at http://dx.doi.org/10.1098/rsif.2016.0035 or via http://rsif.royalsocietypublishing.org. Inferring muscle functional roles of the ostrich pelvic limb during walking and running using computer optimization Jeffery W. Rankin1, Jonas Rubenson2,3 and John R. Hutchinson1 1Structure and Motion Laboratory, Department of Comparative Biomedical Sciences, The Royal Veterinary College, Hawkshead Lane, Hatfield, Herts, UK 2Department of Kinesiology, Pennsylvania State University, University Park, PA, USA 3School of Sport Science, Exercise and Health, The University of Western Australia, Perth, Western Australia, Australia JWR, 0000-0002-6639-8280; JRH, 0000-0002-6767-7038 Owing to their cursorial background, ostriches (Struthio camelus) walk and run with high metabolic economy, can reach very fast running speeds and quickly execute cutting manoeuvres. These capabilities are believed to be a result of their ability to coordinate muscles to take advantage of specialized passive limb structures. This study aimed to infer the functional roles of ostrich pelvic limb muscles during gait. Existing gait data were combined with a newly developed musculoskeletal model to generate simulations of ostrich walking and running that predict muscle excitations, force and mech- anical work. Consistent with previous avian electromyography studies, predicted excitation patterns showed that individual muscles tended to be excited primarily during only stance or swing. Work and force estimates show that ostrich gaits are partially hip-driven with the bi-articular hip– knee muscles driving stance mechanics. Conversely, the knee extensors acted as brakes, absorbing energy. The digital extensors generated large amounts of both negative and positive mechanical work, with increased magnitudes during running, providing further evidence that ostriches make extensive use of tendinous elastic energy storage to improve economy. The simulations also highlight the need to carefully consider non-muscular soft tissues that may play a role in ostrich gait. 1. Introduction Ostriches (Struthio camelus) walk and run with high metabolic economy [1–3], can reach very fast running speeds [4,5], and quickly execute cutting (turning) manoeuvres [6]. The ability to achieve such impressive performance is thought to largely arise from morphological specializations within the pelvic limbs as result of their cursorial and secondarily flightless evolutionary background. Like other birds, ostriches use three-dimensional limb joint motions during locomotion [6–8] and have specialized passive structures at the hip, including bony stops (e.g. the antitrochanter), which play an unclear role during move- ment [9–14]. The distal limb muscles are also highly specialized, consisting of extremely long tendons that cross mobile metatarsophalangeal (MTP) joints. Experimental studies of these features in ostriches and other birds support the inference that they improve gait performance and economy [2,15–18]. However, these adaptations also contribute to the extremely complex ostrich pelvic limb musculoskeletal structure, which consists of more than 30 muscles—the majority of which are multiarticular—that cross joints with multiple degrees of freedom (DOF). As a result, little can be intuitively inferred about specific functional roles that individual pelvic limb muscles perform in ostriches (or many other birds) during gait. Obtaining the data required to & 2016 The Authors. Published by the Royal Society under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original author and source are credited. determine muscle function is further limited owing to the numerous challenges associated with the required experimental techniques (e.g. electromyography (EMG), sonomicrometry, tendon buckles). To date, these factors have obscured how ostriches and other birds successfully meet the biomechanical demands of walking and running. During a movement, the functional role of a muscle– tendon unit (MTU) can be established based on a combination of muscular force generation and muscle and tendon length trajectories [19–21]. If an MTU generates high force and posi- tive power (concentric contraction) during the movement, then energy is added to the system and the MTU can be classi- fied as a ‘motor’. In contrast, an MTU that generates high force but negative power (eccentric contraction) removes energy from the system and acts as a ‘brake’. In some cases, an MTU may generate high forces but produce very little positive or negative power (i.e. no length change) during the move- ment. In this case, the MTU has not added or removed energy from the system and acts as a joint stabilizer or ‘strut’. Last, an MTU may generate high force and switch from nega- tive to positive power production. In this case, the net energy provided to the system is again near zero. However, the MTU has undergone a systematic change in length and likely acts as a ‘spring’, storing energy from an earlier portion of the movement that can be released later. To define an MTU’s functional role(s) in this study, muscle excitation timing is first used to classify whether or not a muscle primarily contrib- utes to ‘stance’ (i.e. when the foot is in contact with the ground) or ‘swing’ (i.e. no foot–ground contact) movements, when possible [22,23]. Following this classification, specific muscle roles (i.e. motor, brake, strut or spring) during stance and swing are then determined using MTU force and length. These roles can then be used to infer how individual muscles contribute to the overall mechanical energy flow during gait. Because the aforementioned difficulties associated with experimental approaches limit their usefulness, an alternative approach is to use realistic, detailed musculoskeletal models and simulations. The first simple ostrich model was developed over 35 years ago by Alexander et al. [4] to estimate muscle and bone stress during running. More recently, two-dimensional ostrich models have been developed to investigate postural effects on running joint mechanics [5] and to validate running posture [24] and maximal speed [25] predictions for various extinct taxa. Until very recently, only a single model of loco- motion has included muscle geometry, which was limited to six muscles [25]. However, we have just published a highly detailed musculoskeletal model of an ostrich’s pelvic limbs, building on prior efforts [26]. Similar approaches have been successfully used to address many questions in human gait: providing insights into muscle function [27–29] and form–function relationships [30,31]. Like most animal musculoskeletal systems, the ostrich pelvic limb has many more muscles than DOF. As a result, multiple muscle excitation patterns exist to produce identical joint mech- anics. Knowing how to correctly ‘parse’ the different muscle contributions to the net joint mechanics during movement is critical to understanding muscle functional roles. Two distinct approaches have been used to overcome this major challenge: static and dynamic optimization [32–34]. Static optimization (SO) addresses each instant in time as an independent data point, reducing computational cost but ignoring time-dependent quantities such as activation–deactivation dynamics and tendon strain energy. Dynamic optimization techniques can account for these time-dependent quantities, but incur a high computational cost. There remains considerable debate over which (if either) is more suitable than another for studying muscle function during movement, in large part because a gold standard (i.e. empirical dataset) is not readily available for comparison. For example, Anderson & Pandy [35], after simulating half-gait cycles of human walking, suggested that static and dynamic optimization solutions were ‘practically equivalent’, but qualified their state- ment and provided scenarios in which dynamic optimization may be necessary. Later comparisons between the two approaches in other human movements have been inconclusive in determining a preferred technique for predicting muscle activity [36–38]. Because of the large number of differences that exist between humans and ostriches in both limb mor- phology and gait mechanics [2], determining how sensitive muscle functional roles (and by extension structure–function relationships) between these two techniques during ostrich gait could help future comparative research focused on movement in different species. The primary purpose of this study was to determine the functional roles that individual pelvic limb muscles have in ostriches during walking and running. Existing biomechanical data were combined with a newly developed, detailed ostrich musculoskeletal model [39] to generate computer simulations that estimate MTU excitation, length and force during the two gaits. A secondary purpose was to assess how sensitive muscle functional roles are to choice of optimization approach (static versus dynamic) using a model that widely diverges in morphology from humans and a higher speed movement than those investigated previously. These two purposes are linked, because methodological assumptions of static versus dynamic analysis [5,25,35] might influence biological conclusions about the functions of particular muscles, which can be tested by achieving these two major aims. 2. Methods A detailed musculoskeletal model of the ostrich pelvic limb [39] was combined with experimental data obtained from a represen- tative walking and running trial [2,8,39] within OpenSim [40] to generate six different simulations (three for each motion, table 1). Two simulations (WSO, RSO) were performed using OpenSim’s SO routine [41]. Two additional simulations (WCMCC, RCMCC) were then generated using OpenSim’s computed muscle control (CMC) routine [42]. The final two simulations (WCMCR, RCMCR) were generated using CMC, but tendons were constrained to be rigid in order to provide a direct comparison with the SO sol- ution, which did not incorporate tendon dynamics, whereas the other two CMC simulations (WCMCC, RCMCC) did. The simulations estimated MTU excitation patterns, force and length, which were used to infer muscle function. Details of the musculo- skeletal model, optimization framework and experimental data are given below. 2.1. Musculoskeletal model The original musculoskeletal model was created using muscle and tendon architecture, digitized muscle paths and computed tomography (CT) scan data collected via dissection [39]. The left pelvic limb was generated by mirroring the right-side segments, joint definitions and muscle tendon paths about the sagittal plane. The model consisted of nine rigid body segments representing the pelvis and left and right-side femur, tibiotarsus, tarsometatarsus and pes (figure 1). The original model’s segment mass and inertia values were scaled using the original ostrich’s rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 2 body mass (65.3 kg) and mass of the bird that provided the experimental data (78.7 kg; see §2.3). Each pelvic limb had 8 DOF representing the hip (3 DOF), knee (3 DOF), ankle (1 DOF) and MTP (1 DOF) joints. In the orig- inal model, both the ankle and MTP joints were modelled as 3 DOF (ball-and-socket) joints. However, minimal long axis rotation and ad/abduction have been observed in the avian ankle and MTP during walking and running [7,8,43,44] and these DOFs were constrained to match experimental mid-stance values. The pelvis moved freely relative to the ground (i.e. three translational and three rotational DOFs). Model segments were driven by a combination of musculo- tendon and idealized joint (coordinate) actuators (figure 1). Thirty-four of the 35 musculotendon actuators from the original model were retained on the right side, which represented the major muscles in the ostrich pelvic limb (FCLA was removed due to its very low maximum force [39]). Musculotendon actua- tors were modelled using a Hill-type model that included intrinsic force–length–velocity relationships [45]. Because walking and running are everyday activities and critical to survival, it is likely that MTU properties are tuned so force and power gener- ation are near optimal during these movements [46,47]. However, many muscles in the original model did not reflect this, with nor- malized fibre lengths exceeding the physiological operating range of 0.5–1.5 optimal fibre lengths in some postures. In the original model, tendon slack lengths (Ltsl) were estimated based on joint range of motion [39,48], which may not reflect tuning for major activities like gait. To correct for this inconsistency, the original model’s Ltsl were systematically adjusted, so that muscle fibre lengths operated over a more optimal range (i.e. 0.75–1.25 optimal fibre length) in the joint ranges of motion defined by the exper- imental gait kinematics. New Ltsl were within 10% of the original model values for all actuators except for M. iliotibialis (ILa, ILp, 19%) and M. femorotibialis intermedius (FMTIM, 19%). Maximum isometric forces were scaled using the mass ratio between the original model and experimental subject (table 2). For all musculotendon actuators, maximum contraction velocity was set to 14 Lfopts21 [49]. Excitation–activation dynamics were represented by a first-order differential equation with acti- vation and deactivation time constants of 10 and 15 ms. As the left side’s movement was assumed to be symmetric with the right side (see §2.3 Experimental data), the model was simplified by having the left side’s joints actuated by eight idealized torque actuators—one for each DOF. Six additional actuators were used to compensate for residual forces and moments at the pelvis during the motion and eight torque actuators—one for each DOF in the right limb—were used to compensate for mechanical work that could not be satis- fied by the muscles alone (reserve actuators). Each optimization was tasked with minimizing the use of these reserve actuators, ensuring that, at each joint, the required joint moments were satisfied primarily through muscle force. 2.2. Simulations Three simulations were generated from the experimental walking data. Each simulation used the same experimental data and mus- culoskeletal model as inputs, but used a different optimization framework to estimate MTU excitation, force and length changes. Three additional simulations were then generated from the exper- imental running data using the same model and optimization frameworks (table 1). Simulations were first generated using the SO routine included in OpenSim. SO determines MTU excitation patterns by optimizing a predetermined objective criterion subject to the biomechanical constraints associated with the motion. The objec- tive criterion used here minimized muscle activation squared, (b) (a) Figure 1. Musculoskeletal model at mid-stance during running. The arrow (blue) indicates the direction and location (centre of pressure) of the ground reaction force. Muscle–tendon actuators (red lines) of the left limb were replaced by idealized joint actuators. (a) Sagittal view. (b) Frontal view. Table 1. Names and description of the six simulations performed. Simulations were performed for either a walking or running motion (rows) using three different optimization frameworks (columns). simulation motion static optimization computed muscle control (rigid tendon) computed muscle control (compliant tendon) walking WSO WCMCR WCMCC running RSO RCMCR RCMCC rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 3 Table 2. Muscle–tendon actuator properties. Optimal fibre lengths and pennation angles are from the original model by Hutchinson et al. [39] but provided for reference. abbreviation muscle name maximum isometric force (Fiso, N) optimal fibre length (Lfopt, m) tendon slack length (Ltsl, m) pennation angle (88888) IC M. iliotibialis cranialis 889 0.174 0.0451 0 ILa M. iliotibialis lateralis (cranial part) 1265 0.174 0.2432 0 ILp M. iliotibialis lateralis (caudal part) 1265 0.174 0.3099 0 AMB1 M. ambiens, ventral (pubic) head 971 0.039 0.1648 10 AMB2 M. ambiens, dorsal (iliac) head 1793 0.044 0.3941 15 FMTL M. femorotibialis lateralis 1434 0.088 0.1746 15 FMTIM M. femorotibialis intermedius 1706 0.084 0.1863 25 FMTM M. femorotibialis medialis 1089 0.089 0.0603 30 ILFBa M. iliofibularis (cranial part) 1254 0.176 0.2134 0 ILFBp M. iliofibularis (caudal part) 1254 0.176 0.2733 0 ITCa M. iliotrochantericus caudalis (cranial part) 897 0.064 0.0469 25 ITCp M. iliotrochantericus caudalis (caudal part) 897 0.064 0.038 25 IFE M. iliofemoralis externus 479 0.025 0.0667 25 ITM M. iliotrochantericus medius 181 0.058 0.0241 0 ITCR M. iliotrochantericus cranialis 330 0.053 0.0488 10 IFI M. iliofemoralis internus 410 0.041 0.0533 0 FCM M. flexor cruris medialis 1109 0.036 0.435 35 FCLP M. flexor cruris lateralis pars pelvica 544 0.24 0.2449 0 ISF M. ischiofemoralis 419 0.033 0.0816 15 PIFML M. puboischiofemorales medialis þ lateralis 816 0.089 0.1669 15 OM M. obturatorius medialis 3124 0.055 0.1651 25 CFP M. caudofemoralis pars pelvica (et caudalis) 1125 0.108 0.215 15 GL M. gastrocnemius pars lateralis 1836 0.12 0.5818 20 GIM M. gastrocnemius pars intermedius 798 0.125 0.507 15 GM M. gastrocnemius pars medialis 3124 0.094 0.5957 20 FL M. fibularis longus 2270 0.081 0.9633 20 FDL M. flexor digitorum longus 1130 0.048 1.0366 20 FPPD3 M. flexor perforans et perforatus digitorum 3 1154 0.025 1.0737 30 FPD3 M. flexor perforans digitorum 3 3210 0.017 1.02 35 FPD4 M. flexor perforans digitorum 4 1434 0.026 1.004 20 FHL M. flexor hallucis longus 469 0.04 1.0939 25 EDL M. extensor digitorum longus 833 0.049 0.8512 30 TCf M. tibialis cranialis (femoral head) 686 0.045 0.4791 25 TCt M. tibialis cranialis (tibial head) 686 0.045 0.4215 25 rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 4 summed across all muscles at each time step [33] J ¼ min X 34 m¼1 a2 m, ð2:1Þ where am is the activation level of the mth muscle. The time step was set to 0.005 s and MTU excitation, force and length time his- tories were obtained over the gait cycle. MTU force calculations included intrinsic muscle force–length–velocity relationships [45]. Because each time step is solved independently within the SO framework, there is neither energy transfer between time steps (e.g. tendon energy storage and return) nor muscle excitation–activation dynamics. Passive fibre force generation is also ignored, and tendons are assumed rigid with all MTU length changes occurring in the muscle fibres. The second optimization framework used to generate simu- lations was OpenSim’s CMC routine [42]. CMC is a hybrid forward–inverse approach, with muscle excitations for each time step determined using the same objective criterion as the SO routine. However, like purely forward dynamic simulations, the model state from a previous time step (e.g. joint angles, muscle activation level, tendon strain) influences the optimal solution for the current step. Because time steps are linked, this approach incorporates muscle excitation–activation dynamics and non-rigid tendon characteristics. Passive muscle fibre force generation is also accounted for. In order to reduce the potentially confounding factors of differ- ent tendon and muscle models when directly comparing between SO and CMC, a third optimization framework was implemented. This approach was identical to the previous CMC framework, with the exception that, like SO, a rigid tendon model was implemented, and muscle passive force generation was removed. Using rigid tendons eliminates tendon–muscle fibre dynamics and partially negates the ability of a forward dynamics optimization to account for time-dependent muscle interactions (e.g. tendon energy storage and return). As a result, using this framework would not be a realistic choice under normal circumstances. However, eliminating these potentially confounding factors allows for a more direct comparison between the SO and CMC frameworks. 2.3. Experimental data Experimental data for a representative walking (1.2 ms21; 0.66 duty factor) and running trial (3.5 ms21; 0.40 duty factor) were taken from a single adult bird (78.7 kg) of a previously collected dataset [2,8,39]. Three-dimensional segment and joint kinematics were calculated from retro-reflective marker clusters located on the pelvis, right-side femur, tibiotarsus and tarsometatarsus, and a single marker on digit III. Marker locations were recorded at 200 Hz using high-speed video (Peak Performance; Centennial, CO). Ground reaction forces were simultaneously collected using a Kistler force plate (model 9865E, Kistler, Winterthur, Switzerland). Data were filtered in OpenSim using a low-pass fre- quency of 10 Hz. Because only right-side data were collected experimentally, left-side motion and force data were estimated by mirroring the right-side data about the sagittal plane and phase-shifting the data 1808 to generate a complete gait cycle. 2.4. Analysis In each simulation, muscle excitation onset and offset timings were determined from the predicted muscle excitation patterns, with muscles considered to be excited when the values exceeded a 0.1 (i.e. 10% of maximum excitation) threshold. A period of excitation was then determined by first identifying the onset time as the closest previous time step where excitation fell below 0.05. Offset time was then identified as the first subsequent point that excitation fell below 0.05. Stance (i.e. foot in contact with the ground) and swing phases were identified and timing values were used to group muscles into ‘stance’ or ‘swing’ groups. Predicted muscle excitation onset and offset times were then normalized to the entire gait cycle and compared with existing avian EMG data [22,23] as a form of indirect validation. MTU force and length time histories were used to generate com- parisons among the six simulations. First, average muscle forces were calculated as the mean force value during stance and swing. An ‘integrated activation’ (iAct) value was also calculated for the two phases. To calculate iAct, the stance and swing phases were first normalized to per cent phase. The activation trajectory was then integrated over the entire phase to generate a single activity value ranging from 0 (no activity) to 100 (maximally active over the entire phase). Net MTU work was calculated for each muscle from the instantaneous MTU force and velocity values over the entire gait cycle. Positive and negative work were calculated for stance and swing by integrating only the positive and negative por- tions of the power curves of each MTU within each phase. Muscles were grouped based on anatomical location, creating seven distinct groups: (i) hip rotators (ITCa, ITCp, ITCR, ITM), (ii) biarticular hip–knee (ILa, ILp, ILFBa, ILFBp, FCLP, FCM), (iii) knee extensors (FMTL, FMTIM, FMTM), (iv) gastrocnemius (GL, GIM, GM), (v) digital flexors (FDL, FHL, FL, FPPD3, FPD3, FPD4), (vi) ankle flexors (EDL, TCf, TCt) and (vii) other (proximal) muscles (OM, IFE, IFI, ISF, PIFML, CFP, AMB1, AMB2, IC). To evaluate the influence that reserve actuators may have had on simulation results, average and peak reserve actuator values were compared with the peak net joint torques (obtained via OpenSim’s inverse dynamics analysis). Reserve actuator work was also calculated from the actuator torque and joint angle trajec- tories, analysed in the same manner as MTU work and then compared with the total amount of mechanical work generated by the muscles in each corresponding simulation. In addition, for the CMC simulations, which were not explicitly constrained to follow the experimental joint kinematics, root mean square (RMS) differences between the experimental and simulation joint kinematics were calculated for the entire movement. 3. Results The three optimization frameworks were able to successfully generate simulations of walking and running, with all six simulations generating a solution. In the CMC simulations, peak errors in simulated joint trajectories were within 28 of experimental angles and RMS errors well below 0.18 (see electronic supplementary material, table S1). 3.1. Reserve actuators In all six simulations, average reserve actuator values remained below 10% of the inverse dynamics moment with the exception of hip ad–abduction, knee ad–abduction and ankle flexion– extension (table 3, average reserve torque). Knee ad–abduction was below 10% for all simulations but WCMCC (15%). Hip ad– abduction had by far the highest average reserve actuator values, accounting for up to 90% of the inverse dynamics moment. Average ankle flexion–extension moments were con- sistent between all simulations, ranging from 9.1 to 16.3%. Peak reserve actuator values were more variable across the different simulations. Peak knee rotation and knee flexion– extension reserve values fell below 10% of the inverse dynamics torques in all simulations except for WCMCC. Peak hip flexion– extension reserve values were below 10% in all but RSO (12%) and RCMCR (15%). Peak hip rotation reserve actuator values all fell below 15%. Ankle flexion–extension and MTP flexion– extension peak reserve values were high in most of the simulations. The hip ad–abduction reserve actuator was highest in all six simulations (table 3). rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 5 Even though the hip ad–abduction reserve actuator had the highest average and peak reserve actuator values, its con- tribution to limb mechanical work over the gait cycle was small (less than 6% of total muscle work) in all simulations (table 3 and figure 2). Knee ad–abduction reserve actuator work was consistently positive, with values ranging from 2.89 (2%; WCMCR, WSO) to 18.09 J (13%; WCMCC). The highest net values were generated by the ankle and MTP reserve actuators, with magnitudes reaching 31.71 J (24%; table 3 and figure 2). The other reserve actuators had low net mechanical work (less than 5%) over the simulation. 3.2. Muscle excitation and activation Muscle timing data were similar across all six simulations, with the majority of muscles having a single excitation period that occurred primarily in either stance or swing (figure 3). The major hip, knee and ankle extensors (e.g. M. flexor cruris lateralis pars pelvica, FCLP; M. femorotibialis, FMTIM; M. gastrocnemius, GL), many hip rotators (e.g. Mm. iliotrochan- tericus, ITCp, ITCr) and the digital flexors (M. flexor digitorum longus, FDL) were primarily excited during stance. The uniarti- cular hip extensors, M. caudofemoralis pars pelvica (CFP) and M. puboischiofemoralis (PIFML) were excited from mid- to-late swing through mid-stance. Owing to their large origin sites, the M. iliotibialis lateralis and M. iliofibularis were parti- tioned into cranial and caudal regions in the model. In both muscles, the caudal portions (ILp, ILFBp) tended to be excited during stance whereas the cranial portions (Ila, ILFBa) were excited during swing (figure 3). The hip and ankle flexors (e.g. M. iliotibialis cranialis, IC; M. tibiocranialis, TC) were primarily excited during swing. In both running and walking ISF is not excited. IFE, IFI and FHL are only excited during the running simulations. Although no ostrich EMG data are available for direct com- parison, simulation results compare favourably to previous Table 3. Average and peak moments as well as net mechanical work generated by the reserve actuators for each of the six simulations. Shaded columns are for the three walking simulations. Moment values are presented in Nm and parenthetical values indicate the per cent of the inverse dynamic analysis joint torque. Work values are presented in joules (J) and parenthetical values are percentages relative to the total muscle–tendon unit mechanical work generated in each simulation. Positive values indicate hip/knee extension, adduction and medial rotation, and ankle/MTP flexion moments. Positive/negative mechanical work indicates energy being added/removed from the limb. degree of freedom WSO WCMCR WCMCC RSO RCMCR RCMCC average reserve torque in Nm (%) hip flexion–extension 20.8 (,1) 20.9 (,1) 22.6 (2) 23.7 (1) 24.3 (2) 21.8 (,1) hip ad–abduction 47.7 (77) 43.7 (71) 57.9 (94) 37.1 (77) 32.7 (68) 28.4 (59) hip rotation 3.8 (4) 3.2 (3) 3.3 (3) 20.3 (,1) 0.2 (,1) 1.0 (,1) knee flexion–extension 0.5 (,1) 0.8 (,1) 11.3 (9) 0.5 (,1) 1.0 (,1) 1.8 (1) knee ad–abduction 0.1 (,1) 1.1 (,1) 18.8 (15) 24.2 (2) 25.6 (2) 4.3 (2) knee rotation 20.5 (1) 20.5 (1) 22.0 (5) 20.5 (,1) 20.1 (,1) 0.3 (,1) ankle flexion–extension 9.4 (14) 6.5 (10) 6.1 (9) 11.2 (16) 9.2 (13) 9.6 (14) MTP flexion–extension 23.1 (4) 22.1 (3) 4.0 (5) 29.4 (6) 212.5 (8) 27.0 (4) peak reserve torque in Nm (%) hip flexion–extension 23.1 (3) 23.7 (3) 28.0 (7) 232.9 (12) 239.1 (15) 210.9 (4) hip ad–abduction 130.7 (212) 138 (224) 133.5 (217) 170.3 (353) 127.4 (263) 112.4 (233) hip rotation 13.5 (14) 13.4 (14) 10.9 (11) 29.3 (5) 14.6 (8) 12.0 (6) knee flexion–extension 2.4 (2) 2.9 (2) 86.0 (69) 8.9 (5) 12.2 (7) 8.7 (5) knee ad–abduction 12.3 (10) 13.7 (11) 129.2 (104) 243.4 (16) 277.3 (29) 27.0 (10) knee rotation 1.8 (4) 21.8 (4) 219.1 (43) 25.7 (10) 22.9 (5) 2.2 (4) ankle flexion–extension 32.7 (49) 29.5 (44) 19.4 (29) 66.7 (97) 66.9 (97) 46.8 (68) MTP flexion–extension 211.7 (15) 29.2 (12) 45.1 (57) 267.9 (43) 291.0 (58) 256.3 (36) net mechanical work (J) hip flexion–extension 0.44 (,1) 0.27 (,1) 20.28 (,1) 0.91 (,1) 0.39 (,1) 20.19 (,1) hip ad–abduction 7.50 (6) 5.08 (4) 27.32 (5) 22.65 (1) 26.05 (3) 24.96 (3) hip rotation 0.83 (,1) 0.12 (,1) 0.18 (,1) 0.45 (,1) 0.01 (,1) 20.49 (,1) knee flexion–extension 20.03 (,1) 20.25 (,1) 6.11 (5) 22.51 (1) 23.61 (2) 0.48 (,1) knee ad–abduction 2.89 (2) 2.89 (2) 18.09 (13) 6.16 (2.8) 7.67 (3.5) 4.89 (3) knee rotation 20.06 (,1) 20.14 (,1) 21.37 (1) 0.26 (,1) 0.16 (,1) 0.16 (,1) ankle flexion–extension 14.71 (11) 7.30 (5) 28.34 (6) 22.31 (10) 12.13 (6) 8.36 (5) MTP flexion–extension 2.50 (2) 0.80 (,1) 31.71 (24) 16.06 (7) 18.7 (8.5) 26.8 (4) rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 6 comprehensive studies of guinea fowl limb muscle activity (figure 3; [22,23]). Except for small timing changes that are to be expected owing to comparisons being performed between different avian species, the simulated muscle excitation pat- terns were consistent with the empirical data: most muscles had a single period of EMG activity in either the stance or swing phase. Nonetheless, there were a few notable exceptions. Similar to EMG recordings [23], CFP was excited during mid- stance. However, either an additional period of excitation or an extended single period occurred during late swing in the simulations that was not evident in the EMG data. The CFP may have been preferentially used to slow down hip flexion and assist in hip extension prior to foot strike. Digital flexor and ankle extensor onset times occurred in early stance in the simulations, but EMG recordings suggest an earlier onset during late swing (e.g. FPD4, FDL, GL). Last, EMG recordings for the ITCR suggest that this muscle is excited during swing. However, the simulations consistently excited ITCR during mid-stance, likely to oppose the high hip lateral rotation moment. Instead, ITCp was excited during both mid-swing and stance in the simulations, whereas EMG data indicate that this muscle only has a single excitation period beginning in late-swing through stance. The ITCa, ITCR and ITCp are all medial hip rotators and discrepancies could be owing to comparing different species. This will remain uncertain until ostrich EMG data become available, even though EMG patterns in avians measured to date generally are conservative [23,50]. When averaged across all muscles, iAct was always greater during stance than swing in both gaits, with the smal- lest difference occurring in WCMCC (21.2 versus 16.7). The running simulations also consistently required more muscle activity than during walking (e.g. RCMCC, 21.5; WCMCC, 19.6). In both gaits, the PIFML and CFP muscles were active during both phases. However, stance phase iAct was much larger during running than walking (figure 4). The medial hip rotators ITCa, ITCp, ITCR and ITM and the lateral hip rotator OM had similar activity levels in all simulations, with the medial rotators primarily active during stance and OM active during swing. Conversely, many of the biarticular muscles crossing the hip and knee (i.e. ILp, ILFBp, FCLP, FCM) had noteworthy changes in iAct between the two gaits (figure 5). Even though muscle activity primarily occurred during stance for both gaits, iAct values for ILFBp, FCLP and FCM were markedly lower in the walking motion. Similar to their excitation patterns, ILa and ILFBa had notable iAct values during both the stance and swing phases in running (figure 5). AMB1 and AMB2 had similar activity levels during swing in both gaits, but had increased activity during stance in running. The IC, a hip flexor and knee extensor, had consistent iAct values across all simulations, which were highest during swing. In both gaits, the uniarticular knee extensors FMTL and FMTIM had larger iAct values during stance than swing, whereas the converse was true for FMTM (figure 6). Knee extensor iAct values differed greatly between simulations, with the CMC compliant tendon simulations (i.e. WCMCC and RCMCC) producing higher values during swing com- pared with the other four simulations. The major ankle extensors (Mm. gastrocnemius: GL, GIM, GM) had higher integrated muscle activity during stance in the running simu- lations. Ankle flexor (TCf, TCt) iAct was comparable between running and walking (e.g. figure 6, TCf: WSO versus RSO). However, the CMC simulations consistently estimated higher overall ankle flexor activity than the SO simulations, with the greatest differences occurring during swing in the CMC rigid tendon simulations. Digital flexor (FPPD3, FPD3, FPD4, FDL) muscle activity occurred almost exclu- sively during stance (figure 7). Differences in FPPD3, FPD3 stance −20 0 0 20 40 −40 −20 0 20 40 −40 −20 0 0 20 40 −40 −20 0 20 40 −40 −20 0 0 20 40 −40 −20 0 20 40 −40 swing gait cycle negative work (J) positive work (J) net work (J) RSO RCMCR RCMCC WSO WCMCR WCMCC hip extension hip abduction hip rotation knee extension knee abduction knee rotation ankle extension MTP extension hip extension hip abduction hip rotation knee extension knee abduction knee rotation ankle extension MTP extension hip extension hip abduction hip rotation knee extension knee abduction knee rotation ankle extension MTP extension negative work (J) positive work (J) net work (J) hip extension hip abduction hip rotation knee extension knee abduction knee rotation ankle extension MTP extension sum hip extension hip abduction hip rotation knee extension knee abduction knee rotation ankle extension MTP extension sum hip extension hip abduction hip rotation knee extension knee abduction knee rotation ankle extension MTP extension sum Figure 2. Positive, negative and net mechanical work generated by the reserve actuators in each simulation. Positive/negative work indicates energy ( joules) added to/removed from the limb during the movement. Sum: total of all reserve actuators. rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 7 and FPD integrated activity occurred between the two gaits, with running simulations consistently having higher values. The digital extensor EDL was primarily active during swing but did have a small amount of activity during stance. 3.3. Muscle force and work Average muscle forces tended to follow the same trends as activation, but there was higher variability between optimiz- ation frameworks, with the compliant tendon simulations using CMC (RCMCC, WCMCC) regularly generating larger forces than the other simulations (figures 8–11). Among all the uniarticular hip muscles, the medial hip rotators and the hip extensors (PIFML, CFP) had the greatest forces during stance (figure 8). During swing, the PIFML and CFP had large forces in the compliant tendon CMC simulations. The lateral hip rotator OM consistently had larger forces in running. Except for the AMB1 and AMB2 muscles—which clearly generated more force during running—the biarticular hip–knee muscles had similar amounts of force in both gaits (figure 9). Swing phase forces were consistent across simu- lations and movements, with the IC, AMB1 and AMB2 muscles generating the largest average forces. The uniarticu- lar knee extensors FMTL and FMTIM and the digital flexor FL had the greatest forces during stance (figures 10 and 11). The GM and GL had large average stance forces in running, but much lower values in walking. The ankle flexors (TCf, TCt) had small forces during both stance and swing in the 0 25 50 75 100 CFP PIFML AMB1 IC ILa ILp ILFBa ILFBp FCLP ITCp ITCR FMTIM TC GL FDL FPD4 Marsh Gatesy stance swing per cent gait cycle RSO RCMCC Figure 3. Example simulated muscle excitation timings during running. Blue (dark grey) and green (light grey) bars indicate periods of excitation for the RSO and RCMCC solutions, respectively. For comparison, onset and offset timing obtained from EMG studies of guinea fowl during slower [20] (Gatesy, 1.0 m s21, hatched bars) and faster [21] (Marsh, 1.5 m s21, striped bars) running are provided. Owing to differences in stance and swing times between the studies, stance and swing phases were normalized to 50% of the gait cycle. Zero per cent (0%) of gait cycle indicates the beginning of the stance phase. The other four simulations had similar excitation patterns. 0 25 50 75 100 swing 0 25 50 75 100 ITCa ITCp ITCR ITM OM IFE IFI ISF PIFML CFP stance integrated activation (iAct) integrated activation (iAct) RSO RCMCR RCMCC WSO WCMCR WCMCC Figure 4. Integrated muscle activation values of the uniarticular hip muscles during the swing and stance phases for each of the six simulations. Solid bars, running simulations; striped bars, walking simulations. rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 8 CMC simulations, with the compliant tendon simulations generating the highest average forces (figure 10). Digital flexor muscles’ forces had a clear distinction between stance and swing, with much smaller swing forces compared with stance (figure 10). The digital extensor EDL primarily generated force during swing. Total MTU mechanical work had similar patterns between walking and running (figure 12). The hip rotators (ITCa and ITCp), knee extensors (FMTL and FMTIM), AMB2, FL and FPPD3 consistently produced negative work, whereas many of the biarticular hip extensors (e.g. ILFB, FCLP, FCM), the hip flexor IC, and ankle extensor (GL) generated positive work in the simulations. In contrast, the mechanical work gen- erated by the ankle flexors TCf and TCt varied greatly between simulations, with no clear pattern. The remaining muscles tended to generate little positive or negative net mechanical work (figure 12). The total amount of positive and negative muscle work generated during swing was much lower than that genera- ted during stance (figure 13). There were increases in both positive and negative mechanical work generated by the M. gastrocnemius, digital flexors and ankle flexors in WCMCC and RCMCC relative to the other simulations. During stance, the biarticular hip–knee muscles generated the majority of the positive work in both gaits, amounting to more than twice their negative work (figure 13). The digital flexors generated large amounts of both positive and negative work, with similar amounts of negative work predicted by all six simulations. However, the amount of positive work generated by the digital flexors increased dramatically in 0 25 50 75 100 swing 0 25 50 75 100 stance AMB1 AMB2 IC ILa ILp ILFBa ILFBp FCLP FCM integrated activation (iAct) integrated activation (iAct) RSO RCMCR RCMCC WSO WCMCR WCMCC Figure 5. Integrated muscle activation values of the biarticular muscles crossing the hip and knee during the swing and stance phases for each of the six simulations. Solid bars, running simulations; striped bars, walking simulations. 0 25 50 75 100 swing 0 25 50 75 100 stance FMTL FMTIM FMTM GL GIM GM TCf TCt RSO RCMCR RCMCC WSO WCMCR WCMCC integrated activation (iAct) integrated activation (iAct) Figure 6. Integrated muscle activation values for the uniarticular and biarticular knee and ankle muscles during the swing and stance phases for each of the six simulations. Solid bars, running simulations; striped bars, walking simulations. rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 9 RCMCC and WCMCC simulations. On the other hand, the knee extensors generated a large amount of negative work and very little positive work. The gastrocnemius group generated very little work in walking, but consistently produced a small amount of positive work in running. 3.4. Muscle functional roles To act as a motor that drives motion, muscles must produce force during concentric contractions and generate positive work. In both gaits, the muscles identified as motors were the same (table 4). The IC and AMB2 provided much of the energy required during swing, whereas the biarticular hip and knee muscles (ILFBa, ILFBp, FCM, FCLP) and lateral gas- trocnemius (GL) provided energy during stance (figures 11 and 12 and table 3). In contrast, the hip rotators (ITCa, ITCp, ITM, ITCR), FMTM, AMB1, ankle flexors (TCf, TCt, EDL) and uniarticular hip extensors (PIFML, CFP) all acted as struts, generating moderate to high forces but little positive or negative work. Furthermore, the digital flexors acted primarily as springs during stance, first absorbing energy (negative work) in early stance and then generating positive work during late stance (figure 13 and table 4). Finally, the FDL also generated force during an eccentric contraction in early stance, resulting in net negative limb work (i.e. a brake). Likewise, the knee extensors FMTM and FMTL acted as brakes, absorbing energy from the limb during stance (figure 13 and table 4). A few differences in functional roles between gaits were evident. During walking, the IL and GM acted as brakes and absorbed energy from the limb during swing stance EDL FDL FHL FL FPPD3 FPD3 FPD4 RSO RCMCR RCMCC WSO WCMCR WCMCC 0 25 50 75 100 0 25 50 75 100 integrated activation (iAct) integrated activation (iAct) Figure 7. Integrated muscle activation values for the muscles crossing the metatarsophalangeal (MTP) joint during the swing and stance phases for each of the six simulations. All of these muscles are either biarticular (ankle–MTP) or multiarticular (knee–ankle–MTP). Solid bars, running simulations; striped bars, walking simulations. swing stance average force (N) average force (N) RSO RCMCR RCMCC WSO WCMCR WCMCC 0 400 500 600 0 400 600 ITCa ITCp ITCR ITM OM IFE IFI ISF PIFML CFP 200 200 300 100 100 300 500 Figure 8. Average muscle force values of the uniarticular hip muscles during the swing and stance phases for each of the six simulations. Solid bars, running simulations; striped bars, walking simulations. rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 10 stance. However, these muscles acted primarily as struts during running, generating force but very little work. Muscles with a second excitation period during running did not alter the functional roles of the comparable excitation periods between the two gaits. Instead, the additional excitation periods added an additional role to the muscle during the movement. The AMB1 and AMB2 had additional roles as a strut and brake, respectively, during stance in running, whereas the ITCa and ILFBa had additional roles of strut and brake, respectively, during swing. 4. Discussion Combining detailed musculoskeletal models and simulations with empirical data allows for the estimation of quantities that can greatly enhance our understanding of specific functional roles during dynamic movements [28,29,51]. Although anatom- ical and EMG studies can provide insight into muscle classification relative to gait events (e.g. stance versus swing phase), a detailed understanding of a muscle’s functional role(s) requires additional quantities that are not readilyobtained using experimental techniques. The musculotendon force and mechanical work data generated in this study enable the deter- mination of specific muscle mechanical roles such as motor, brake, strut or spring during gait [19–21]. These roles provide important information regarding how energy flows through the limb and generates the required external work during move- ment. Muscle functional roles were also mainly insensitive to optimization approach or gait type (table 4). However, there were some subtle differences between the SO and computed muscle control compliant tendon (CMCC) simulations, especially among muscles with long tendons that were classified as mechanical springs (table 4). These swing stance average force (N) average force (N) 0 450 750 900 0 300 450 900 AMB1 AMB2 IC ILa ILp ILFBa ILFBp FCLP FCM RSO RCMCR RCMCC WSO WCMCR WCMCC 600 750 300 150 150 600 Figure 9. Average muscle force values of the biarticular muscles crossing the hip and knee during the swing and stance phases for each of the six simulations. Solid bars, running simulations; striped bars, walking simulations. swing stance average force (N) average force (N) 0 300 600 900 1200 0 300 600 900 1200 FMTL FMTIM FMTM GL GIM GM TCf TCt RSO RCMCR RCMCC WSO WCMCR WCMCC Figure 10. Average muscle force values for the uniarticular and biarticular knee and ankle muscles during the swing and stance phases for each of the six simu- lations. Solid bars, running simulations; striped bars, walking simulations. rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 11 differences were most evident in the digital flexors (FL, FPPD3, FPD3, FPD4) during running, where the magnitude of the net mechanical work produced by these muscles was lower in CMCC than SO (figure 12). On the other hand, the amount of negative and positive work generated by these muscles in CMCC was higher than SO (figure 13). Ideal mechanical springs have zero net mechanical work; all absorbed energy is stored and returned. An MTU acting in a spring-like fashion will exhibit high amounts of positive and negative work but have a low net mechanical work. Although the digital flexors exhibited these spring-like characteristics in both optimization approaches, the CMCC simulations more clearly indicated that the muscles were acting as springs. Using CMCC may be more helpful in other situations, where functional roles are not as easily identified. For example, the ankle flexor group produced close to zero net mechanical work during stance in all simulations (figure 13). The total negative and positive mechanical work varied greatly between simulations, however. Positive and negative mechanical work were near zero in the SO simulations, defining these muscles as struts during stance. However, the positive and negative values were many times greater in the CMCC simulations, resulting in a func- tional role of a spring for these muscles (figure 13). Based on their anatomical features (i.e. short muscle fibres and long tendons), it is likely that the ankle extensor MTUs truly do act as springs as suggested by the CMCC simu- lations. Interestingly, the computed muscle control simulations incorporating a rigid tendon (CMCR) generated results similar to the SO simulations. Thus, the incorporation average force (N) average force (N) 0 600 900 1200 1500 0 600 900 1200 1500 EDL FDL FHL FL FPPD3 FPD3 FPD4 RSO RCMCR RCMCC WSO WCMCR WCMCC 300 300 swing stance Figure 11. Average muscle force values for the muscles crossing the metatarsophalangeal (MTP) joint during the swing and stance phases for each of the six simulations. All of these muscles are either biarticular (ankle–MTP) or multiarticular (knee–ankle–MTP). Solid bars, running simulations; striped bars, walking simulations. ITCa ITCp ITCR ITM OM IFE IFI ISF PIFML CFP IC ILa ILp ILFBa ILFBp FCLP FCM AMB1 AMB2 FMTL FMTIM FMTM GL GIM GM TCf TCt EDL FDL FHL FL FPPD3 FPD3 FPD4 −50 −30 −20 −10 0 10 net work (J) –10 0 10 20 30 50 net work (J) RSO RCMCR RCMCC WSO WCMCR WCMCC −40 40 Figure 12. Net mechanical work for each musculotendon unit over an entire gait cycle. Positive/negative work indicates work added/removed from the bio- mechanical system. Solid bars, running simulations; striped bars, walking simulations. rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 12 of both (i) a flexible tendon and (ii) the ability to account for energy storage and return may be important when inferring whether a muscle acts as a strut or spring. Relative to the hip, the knee undergoes greater joint excursions during walking and running in birds. As a result, studies of avian gait have historically characterized muscles crossing the knee as critical to driving movement [52]. On the other hand, models of human walking and run- ning have found muscles crossing the knee to primarily act as brakes, absorbing energy during stance [53,54]. Ostriches are uniquely situated—as birds they likely use similar mechanics to smaller cursorial birds but are larger in size and thus may require similar mechanics to larger bipedal animals such as humans. An examination of muscular roles provides evi- dence that ostrich gait is at least partly hip-driven, with the major biarticular hip-to-knee muscles acting as motors and generating much of the positive work in both gaits (table 4 and figure 12: ILFB, FCLP, FCM). Bi-articular muscles are thought to act primarily to transfer energy across joints (i.e. as a strut) and the function of the ostrich bi-articular hip extensors as a motor may be greater than previously inferred. In contrast, despite generating large forces, the uni- articular hip extensors (PIFML, CFP) had mechanical work values near zero and acted as struts. This result is consistent with previous inverse dynamics analyses (i.e. joint-level ana- lyses) that predict little hip joint work [2]. However the muscle-level analysis performed here, which includes work done by multi-joint muscles, shows that total hip muscle work may be disproportionate to joint-level estimates and suggests that ostriches may use more complex hip–knee interactions than humans to drive their limbs. The major knee extensors (FMTL, FMTIM) acted as brakes during stance, suggesting that ostriches, like humans, employ these muscles to assist in maintaining whole-body stability. Of the muscles active in swing, only IC and AMB2 acted as motors, indicating that these muscles are the key drivers of swing phase mechanics, especially limb protraction. Avian distal limb muscles are remarkably specialized, con- sisting of extremely long tendons that have high energy storage and return potential [2,15,17,18]. In this study, regardless of the type of simulation, the lateral gastrocnemius (GL) and digital flexors generated large but nearly equal amounts of negative and positive work, resulting in near zero net mechanical work in both gaits (figures 7 and 11–13). These muscles acted as springs, first absorbing energy during early stance and then returning this energy during late stance. The magnitudes of 0 –35 –70 0 35 70 negative work (J) positive work (J) 0 35 70 positive work (J) 0 –35 –70 negative work (J) stance hip rotators biarticular hip/knee knee extensors gastrocnemius digital flexors ankle flexors other muscles swing hip rotators biarticular hip/knee knee extensors gastrocnemius digital flexors ankle flexors other muscles RSO RCMCR RCMCC WSO WCMCR WCMCC Figure 13. Positive and negative musculotendon work generated by different muscle groups over the stance and swing phases of a gait cycle. Positive/negative work indicates work added/removed from limb and were calculated from the corresponding portion of the power curves. Muscles were grouped as either hip rotators (ITCa, ITCp, ITCR, ITM), biarticular hip/knee (ILa, ILp, ILFBa, ILFBp, FCLP, FCM), knee extensors (FMTL, FMTIM, FMTM), gastrocnemius (GL, GIM, GM), digital flexors (FDL, FHL, FL, FPPD3, FPD3, FPD4), ankle flexors (EDL, TCf, TCt) or other muscles (OM, IFE, IFI, ISF, PIFML, CFP, AMB1, AMB2, IC). Solid bars, running simulations; striped bars, walking simulations. rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 13 positive and negative work generated by these MTUs were also greater during running than walking (e.g. 266.5 versus 237.8 J and 49.6 versus 44.0 J; RCMCC versus WCMCC), congruent with the notion that these MTUs are acting as springs that use tendon energy storage and return (figure 13). The increased distal limb muscle activity and work observed in the running simulations is consistent with the widely held notion that ostriches increase their reliance on these specialized elastic structures during higher speed movements to improve running economy [2,17]. In both gaits, individual muscle excitation timing and integrated muscle activity occurred primarily during either stance or swing, suggesting that primary muscle functional roles may be associated with gait phases (figures 3–7). These data allowed for general muscle classification, which was found to be insensitive to simulation type and generally Table 4. Muscle functional roles based on muscle–tendon unit excitation, force and mechanical work. Differences in roles between walking and running are shown in italics. Muscles performing roles in both swing and stance have roles that are separated by a colon (:) with their role in swing first (e.g. AMB2 acts as a motor during swing, then acts as a brake during stance). muscle abbreviation classification primary role running walking running walking M. iliotibialis cranialis IC swing swing motor motor M. iliotibialis lateralis (cranial part) ILa swing stance strut brake M. iliotibialis lateralis (caudal part) ILp stance stance strut brake M. ambiens, ventral (pubic) head AMB1 both swing strut : strut strut M. ambiens, dorsal (iliac) head AMB2 both swing motor : brake motor M. femorotibialis lateralis FMTL stance stance brake brake M. femorotibialis intermedius FMTIM stance stance brake brake M. femorotibialis medialis FMTM swing swing strut strut M. iliofibularis (cranial part) ILFBa both stance brake : motor motor M. iliofibularis (caudal part) ILFBp stance stance motor motor M. iliotrochantericus caudalis (cranial part) ITCa stance stance strut strut M. iliotrochantericus caudalis (caudal part) ITCp both stance strut : strut strut M. iliofemoralis externus IFE stance off strut M. iliotrochantericus medius ITM stance stance strut strut M. iliotrochantericus cranialis ITCR stance stance strut strut M. iliofemoralis internus IFI swing off strut M. flexor cruris medialis FCM stance stance motor motor M. flexor cruris lateralis pars pelvica FCLP stance stance motor motor M. ischiofemoralis ISF off off M. puboischiofemorales medialis þ lateralis PIFML stance stance strut strut M. obturatorius medialis OM swing swing strut strut M. caudofemoralis pars pelvica (et caudalis) CFP stance stance strut strut M. gastrocnemius pars lateralis GL stance stance motor motor M. gastrocnemius pars intermedius GIM stance stance strut strut M. gastrocnemius pars medialis GM stance stance strut brake M. fibularis longus FL stance stance brake brake M. flexor digitorum longus FDL stance stance spring spring M. flexor perforans et perforatus digitorum 3 FPPD3 stance stance spring spring M. flexor perforans digitorum 3 FPD3 stance stance spring spring M. flexor perforans digitorum 4 FPD4 stance stance spring spring M. flexor hallucis longus FHL stance off spring M. extensor digitorum longus EDL swing swing strut strut M. tibialis cranialis (femoral head) TCf swing swing strut strut M. tibialis cranialis (tibial head) TCt swing swing strut strut rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 14 consistent between the two gaits; only seven of the 34 muscle actuators had gait-specific classifications (table 4). In all seven muscles with gait-specific classifications, the running gaits had additional excitation periods that were not observed in the walking simulations. For example, AMB1 was excited during swing in both gaits. In running, AMB1 also had an additional excitation period during stance (figure 5). These findings may be due to the higher mechanical demands associated with running and muscles may take on additional roles to assist with meeting these demands. Although a broad division based on gait phases could be identified for individual muscles, this division did not scale to anatomical groups. For example, within the femorotibialis muscle group, FMTM and FMTL were classified as stance phase muscles but FMTIM was classified as a swing phase muscle based on excitation timing. Similarly, the cranial por- tions of M. iliofibularis (ILFBa) and M. iliotibialis lateralis (ILa) had different classifications from the caudal portions (ILFBp, ILp) during running (table 4). Previous EMG studies have also suggested that muscles within anatomical groups are differentially excited. Marsh et al. [22] showed that the Mm. femorotibialis and M. iliofibularis usually had two exci- tation periods during running—one during stance and a second during swing. Gatesy [23] also found the cranial and caudal compartments of the M. iliotibialis lateralis to have distinct activity patterns. Our study, combined with the previous EMG work, highlights the need to exercise cau- tion when assuming that anatomically similar muscles also have similar functions during movement. In addition, the present results further suggest that even general classification of muscles based solely on excitation relative to stance or swing phase mechanics may be too simplistic. For example, despite their primary activity being clearly associated with either stance or swing, many limb muscles in this study also had small amounts of excitation over transition regions (e.g. late stance to early swing) [22,43]. The reasons for this low level excitation are less clear: activity may be associated with a secondary minor functional role or may be a result of time delays between muscle activity and force gener- ation—future work directed at resolving this uncertainty (e.g. combining simulations with induced acceleration and/ or segment power analyses [55–57]) is warranted. When constructing optimizations designed to reproduce experimental data, OpenSim allows the user to apply ‘reserve actuators’ to each joint in the model to compensate for any mechanical forces that could not be satisfied by the muscles alone. Because the optimization framework only uses these actuators when muscle forces are insufficient, the actuator values can provide a rough estimate of how experimental data and musculoskeletal model inaccuracies influence a simu- lation. During human movements, a threshold value of 5% of the net joint moments for reserve actuator values (average and peak) has been suggested as one indicator of a high-quality simulation [58]. In this study, average reserve actuators fell below 10% of net joint moments in 37 of the 48 cases (table 3, average). The most notable exception occurred in the average hip ad–abduction moment, which exceeded 50% in all six simulations. Peak values were more variable but hip ad–abduction, ankle and MTP flexion–extension reserve actuators were high in most of the simulations (table 3). One plausible reason for the high average and peak reserve actua- tor values is that they are compensating for unmodelled passive tissues and structures. Functionally, passive tissues act primarily as struts or springs, generating high forces but little mechanical work. To further assess whether the reserve actuators represent unmodelled passive structures, the posi- tive, negative and net mechanical work generated by each actuator was calculated. Except for ankle and MTP flexion– extension, net mechanical work was generally low (i.e. less than 5% of the 134.7–224 J in total muscle work; table 3 and figures 2 and 12), suggesting that most reserve actuators likely represented passive structures. Ostriches, like most birds, have remarkably few hip adduc- tor muscles [9,59]. This is not surprising, because inverse dynamics analyses have shown that the intersegmental hip abduction moment is less than half the hip extension moment during stance in running [2]. However, many of the biarticular hip extensors and knee flexors, which are the main drivers during gait (table 4 and figure 13), also have large hip abduction moment arms. Therefore, these muscles generate a very large hip abduction moment during stance that cannot be counteracted by adductor muscles alone. Instead, passive mechanisms, such as bony contact between the femur and antitrochanter and strong ligaments [12–14] likely oppose this abduction moment. In our study, the hip ad–abduction reserve actuator was used to represent these passive mechanisms that are not explicitly modelled. Both the net mechanical work generated and the pattern of work gener- ation exhibited by this reserve actuator were consistent with it representing passive tissues. During stance, this actuator gener- ates an equal amount of negative and positive work, resulting in little net mechanical work during the modelled motions. In addition, the negative work associated with the hip ab– adduction actuator was generated during early stance and the positive work was generated during mid-to-late stance, consist- ent with the expected energetics of a passive structure that can stretch to absorb and then return energy (table 3 and figure 2). To further test if the hip ad–abduction reserve actuator represented unmodelled passive tissues and better under- stand how these tissues may influence muscle coordination, a series of post hoc simulations using CMCC were generated in which the hip adduction reserve actuator was systemati- cally reduced (i.e. reducing passive tissue contributions). As passive force contributions decreased, the muscle IC, despite acting as a hip flexor, was increasingly recruited during stance owing to its small hip adduction moment. After IC was maximally recruited, hip extension muscle activity was decreased to reduce the induced hip abduction moment by these muscles, replaced by increasing the torque generated by the hip extension reserve actuator. Both the recruitment of IC during stance, which has been found to be active exclusively during swing in other birds [22,23], and the increased reliance on the hip extension reserve actuator to power the motion suggest that passive hip structures are important during ostrich gait. The avian hip is an excellent example of a joint where non-muscular soft tissues and bony stops deserve careful consideration in dynamic analyses of locomotion. However, rigorously implementing sufficiently accurate passive structures introduces additional challenges when building models and simulations. Rigid body contact models exist that could be implemented to model bony stops [60–63]. However, implementing these contact models is difficult as detailed information of both the under- lying contact geometry and detailed joint motion data are necessary (i.e. subject-specific models), which are rarely rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 15 available. In addition, contact models can be computationally expensive, especially when implemented at multiple joints, further increasing the time required to generate an optimal simulation. Similar constraints and limitations are associated with modelling other non-muscular passive tissues, where detailed knowledge of joint and tissue geometry is necessary. One alternative approach that has been used successfully in numerous human studies is to quantify the total passive behaviour of a joint using regression equations [64–66]. These equations are usually generated in the form of a net passive torque as a function of a single joint angle. However, creating these characteristic regression functions requires extensive cadaver-based work, especially when trying to characterize how the tissues interact between multiple DOF at a joint. On the other hand, the ankle and MTP reserve actuators generated a substantial amount of positive work, suggesting that they did not represent passive structures but were compensating for muscle deficiencies. Peak MTP reserve actuator values occurred during mid-stance to assist the digi- tal flexors, whereas peak ankle reserve values occurred during mid-swing to assist the ankle flexors. To confirm that muscle weakness was responsible for the simulations requiring these reserve actuators, an additional RCMCC run- ning simulation was performed in which the maximum isometric force of the digital flexors and ankle flexors was doubled. Doubling the strength of the digital flexors elimi- nated the need for the MTP reserve actuator, confirming that these muscles appear to be weak relative to the motion requirements. This result is consistent with findings in pre- vious human running studies, where models of the plantar flexor muscles were incapable of generating sufficient torque to overcome the mechanical demands at the ankle joint [5,67]. Surprisingly, doubling the maximum isometric force of the ankle muscles did not reduce the required ankle flexion reserve torque—in fact, the required reserve torque was higher in this simulation. Further inspection revealed that the antagonistic digital flexors were passively generating force during mid-swing owing to muscle fibres operating at fibre lengths greater than the optimal fibre length. In the model, the ankle flexors cannot counteract these passive muscle forces using the current force ratio between the two muscle groups. In general, muscle fibre excursions tended to be larger than might be expected empirically, especially over regions where the joints also underwent large angle changes such as those found in swing (electronic supplementary material, figure S1). Lumped-parameter muscle models, like the Hill-type muscle model used here, tend to overestimate fibre excur- sions, which may explain why the digital flexors produce passive force during swing [45,68]. Despite these model inconsistences, all six simulations predicted overall muscle coordination patterns consistent with previously collected guinea fowl EMG data (figure 3, [20,21]). In addition, the percentage of muscle activity that occurs during swing (13.9–38.6%; see electronic sup- plementary material, table S2) compares favourably with previous muscle blood flow data suggesting that one quarter of the energetic cost of running occurs during swing in guinea fowl [22]. Combined with the good excitation timing comparisons in the vast majority of the muscles, these data indicate that the excitation patterns predicted by the simulations in this study are, in general, biologically reasonable and realistic. The high level of similarity bet- ween the predicted ostrich muscle coordination patterns and those of smaller cursorial birds also suggests that, despite experiencing a large change in size, ostriches appear to have conserved a gait coordination pattern inher- ited from a common avian ancestor, which is unsurprising given the apparent conservatism in avian pelvic limb muscle activity [23,50]. Although muscle functional roles were found to be insen- sitive to the three different optimization frameworks, there were some subtle differences in muscle quantities. During both walking and running, total muscle activity was consist- ently lower in the SO simulations than in both CMC simulations. This is most likely a direct result of the CMC simulations including excitation–activation dynamics, which can increase muscle co-contraction. The CMCC simulations also generated greater muscle forces despite having similar iAct values to the other simulations (e.g. figures 5 and 9; TCf, TCt), with differences likely due to the incorporation of fibre–tendon dynamics that create substantial changes in the force generation properties of muscle. Caution should be taken when eliminating muscle–tendon dynamics from bio- mechanical analyses, especially when investigating specific muscle quantities, motions that require large changes in joint motion, or muscles with relatively long tendons. Further tests against a gold standard (i.e. muscle fibre length measure- ments obtained via sonomicrometry or tendon force measurements via tendon buckles) should provide additional insight into how sensitive specific muscle quantities may be to muscle–tendon dynamics and optimization approach. Our study shows how combining detailed musculoskeletal models with optimization techniques can provide a rich and varied dataset that complements and enhances existing empirical methods used in comparative biomechanics research. Similar to reductionist models [69,70], these models are well suited to theoretical studies that can elucidate under- lying principles and constraints governing motion. For example, this study has generated estimates of muscle exci- tation, force and musculotendon work during walking and running in an ostrich, which were used to identify muscle functional roles. Muscle roles were found to be insensitive to optimization approach, with the bi-articular hip and knee muscles acting as motors and digital flexors acting as springs during stance. The IC and AMB2 were the main drivers of the swing motion. Passive tissues at the hip also appear to play an important role in ostrich running, acting as a strut to prevent excessive hip abduction. Future models should incorporate non-muscular soft tissues and bony stops, which also deserve careful consideration when modelling or performing dynamic analyses of locomotion of fossil taxa. Data accessibility. The musculoskeletal model and motion data used in this study are available via Dryad at http://dx.doi.org/10.5061/ dryad.fh3h6. Author contributions. J.W.R. modified the original musculoskeletal model, ran computer simulations, performed the analyses and drafted the manuscript. J.R. and J.R.H. assisted in study conception and design, helped interpret the study findings, and provided comments on manuscript drafts. All authors gave final approval for publication. Competing interests. We have no competing interests. Funding. This project was partially supported by BBSRC and NERC grants (grant no. BB/I02204X/1 and NE/K004751/1 to J.R.H.) and fellowships from the NSF (to J.R.H.) and the Vice Principal of Research at the Royal Veterinary College (to J.W.R.). rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 16 Acknowledgements. We thank Thor Besier, David Lloyd, Paul Fornier and Denham Heliams for their contribution to ostrich gait data collection and the two anonymous reviewers who provided very helpful suggestions during the review process. References 1. Watson RR, Rubenson J, Coder L, Hoyt DF, Propert MWG, Marsh RL. 2011 Gait-specific energetics contributes to economical walking and running in emus and ostriches. Proc. R. Soc. B 278, 2040–2046. (doi:10.1098/rspb.2010.2022) 2. Rubenson J, Lloyd DG, Heliams DB, Besier TF, Fournier PA. 2011 Adaptations for economical bipedal running: the effect of limb structure on three-dimensional joint mechanics. J. R. Soc. Interface 8, 740–755. (doi:10.1098/rsif.2010.0466) 3. Rubenson J, Heliams DB, Lloyd DG, Fournier PA. 2004 Gait selection in the ostrich: mechanical and metabolic characteristics of walking and running with and without an aerial phase. Proc. R. Soc. Lond. B 271, 1091–1099. (doi:10.1098/rspb. 2004.2702) 4. Alexander RM, Maloiy GMO, Njau R, Jayes AS. 1979 Mechanics of running of the ostrich (Struthio camelus). J. Zool. 187, 169–178. (doi:10.1111/j. 1469-7998.1979.tb03941.x) 5. Hutchinson JR. 2004 Biomechanical modeling and sensitivity analysis of bipedal running ability. I. Extant taxa. J. Morphol. 262, 421–440. (doi:10.1002/jmor.10241) 6. Jindrich DL, Smith NC, Jespers K, Wilson AM. 2007 Mechanics of cutting maneuvers by ostriches (Struthio camelus). J. Exp. Biol. 210, 1378–1390. (doi:10.1242/jeb.001545) 7. Kambic RE, Roberts TJ, Gatesy SM. 2014 Long-axis rotation: a missing degree of freedom in avian bipedal locomotion. J. Exp. Biol. 217, 2770–2782. (doi:10.1242/jeb.101428) 8. Rubenson J, Lloyd DG, Besier TF, Heliams DB, Fournier PA. 2007 Running in ostriches (Struthio camelus): three-dimensional joint axes alignment and joint kinematics. J. Exp. Biol. 210, 2548–2562. (doi:10.1242/jeb.02792) 9. Hutchinson JR, Gatesy SM. 2000 Adductors, abductors, and the evolution of archosaur locomotion. Paleobiology 26, 734–751. (doi:10. 1666/0094-8373(2000)026,0734:AAATEO.2.0. CO;2) 10. Hutchinson JR, Allen V. 2009 The evolutionary continuum of limb function from early theropods to birds. Naturwissenschaften 96, 423–448. (doi:10. 1007/s00114-008-0488-3) 11. Souter T, Cornette R, Pedraza J, Hutchinson J, Baylac M. 2010 Two applications of 3D semi-landmark morphometrics implying different template designs: the theropod pelvis and the shrew skull. C.R. Palevol. 9, 411–422. (doi:10.1016/j.crpv.2010. 09.002) 12. Hertel F, Campbell KE. 2007 The antitrochanter of birds: form and function in balance. Auk 124, 789. (doi:10.1642/0004-8038(2007)124[789:TAOBFA]2.0. CO;2) 13. Firbas W, Zweymu¨ller K. 1971 Uber des Hu¨ftgelenk der Ratiten. Geg. Morph. Jrb. 116, 91–103. 14. Tsai HP, Holliday CM. 2015 Articular soft tissue anatomy of the archosaur hip joint: structural homology and functional implications. J. Morphol. 276, 601–630. (doi:10.1002/jmor.20360) 15. Biewener AA. 1998 Muscle function in vivo: a comparison of muscles used for elastic energy savings versus muscles used to generate mechanical power. Am. Zool. 38, 703–717. (doi:10.1093/icb/ 38.4.703) 16. Wilson AM. 2000 Optimization of the muscle– tendon unit for economical locomotion in cursorial animals. In Skeletal muscle mechanics (ed. W Herzog), pp. 517–547. Chichester, UK: Wiley & Sons. 17. Smith NC, Wilson AM. 2013 Mechanical and energetic scaling relationships of running gait through ontogeny in the ostrich (Struthio camelus). J. Exp. Biol. 216, 841–849. (doi:10.1242/jeb. 064691) 18. Roberts TJ, Marsh RL, Weyand PG, Taylor CR. 1997 Muscular force in running turkeys: the economy of minimizing work. Science 275, 1113–1115. (doi:10. 1126/science.275.5303.1113) 19. Dickinson MH, Farley CT, Full RJ, Koehl MA, Kram R, Lehman S. 2000 How animals move: an integrative view. Science 288, 100–106. (doi:10.1126/science. 288.5463.100) 20. Higham TE, Biewener AA, Delp SL. 2011 Mechanics, modulation and modelling: how muscles actuate and control movement. Phil. Trans. R. Soc. B 366, 1463–1465. (doi:10.1098/rstb.2010.0354) 21. Roberts TJ, Azizi E. 2011 Flexible mechanisms: the diverse roles of biological springs in vertebrate movement. J. Exp. Biol. 214, 353–361. (doi:10. 1242/jeb.038588) 22. Marsh RL, Ellerby DJ, Carr JA, Henry HT, Buchanan CI. 2004 Partitioning the energetics of walking and running: swinging the limbs is expensive. Science 303, 80–83. (doi:10.1126/science.1090704) 23. Gatesy SM. 1999 Guineafowl hind limb function. II: Electromyographic analysis and motor pattern evolution. J. Morphol. 240, 127–142. (doi:10.1002/ (SICI)1097-4687(199905)240:2,127::AID-JMOR 4.3.0.CO;2-Q) 24. Gatesy SM, Ba¨ker M, Hutchinson JR. 2009 Constraint-based exclusion of limb poses for reconstructing theropod dinosaur locomotion. J. Vertebr. Paleontol. 29, 535–544. (doi:10.1671/ 039.029.0213) 25. Sellers WI, Manning PL. 2007 Estimating dinosaur maximum running speeds using evolutionary robotics. Proc. R. Soc. B 274, 2711–2716. (doi:10. 1098/rspb.2007.0846) 26. Bates KT, Schachner ER. 2012 Disparity and convergence in bipedal archosaur locomotion. J. R. Soc. Interface 9, 1339–1353. (doi:10.1098/rsif. 2011.0687) 27. Liu MQ, Anderson FC, Pandy MG, Delp SL. 2006 Muscles that support the body also modulate forward progression during walking. J. Biomech. 39, 2623–2630. (doi:10.1016/j. jbiomech.2005.08.017) 28. McGowan CP, Kram R, Neptune RR. 2009 Modulation of leg muscle function in response to altered demand for body support and forward propulsion during walking. J. Biomech. 42, 850–856. (doi:10.1016/j.jbiomech.2009.01.025) 29. Zajac FE, Neptune RR, Kautz SA. 2003 Biomechanics and muscle coordination of human walking: part II: lessons from dynamical simulations and clinical implications. Gait Posture 17, 1–17. (doi:10.1016/ S0966-6362(02)00069-3) 30. Lee SSM, Piazza SJ. 2009 Built for speed: musculoskeletal structure and sprinting ability. J. Exp. Biol. 212, 3700–3707. (doi:10.1242/jeb.031096) 31. Miller RH, Umberger BR, Caldwell GE. 2012 Limitations to maximum sprinting speed imposed by muscle mechanical properties. J. Biomech. 45, 1092–1097. (doi:10.1016/j.jbiomech.2011.04.040) 32. An KN, Kwak BM, Chao EY, Morrey BF. 1984 Determination of muscle and joint forces: a new technique to solve the indeterminate problem. J. Biomech. Eng. 106, 364–367. (doi:10.1115/1. 3138507) 33. Kaufman KR, An KW, Litchy WJ, Chao EY. 1991 Physiological prediction of muscle forces I. Theoretical formulation. Neuroscience 40, 781–792. (doi:10.1016/0306-4522(91)90012-D) 34. Erdemir A, McLean S, Herzog W, van den Bogert AJ. 2007 Model-based estimation of muscle forces exerted during movements. Clin. Biomech. (Bristol, Avon) 22, 131–154. (doi:10.1016/j.clinbiomech. 2006.09.005) 35. Anderson FC, Pandy MG. 2001 Static and dynamic optimization solutions for gait are practically equivalent. J. Biomech. 34, 153–161. (doi:10.1016/ S0021-9290(00)00155-X) 36. Miller RH, Umberger BR, Caldwell GE. 2009 Muscle forces in the lower limb predicted by static and dynamic optimization. In American Society of Biomechanics 33rd Annual Meeting, State College, Pennsylvania, PA, 26–29 August. See http:// www.asbweb.org/conferences/2009/937.pdf. 37. Lin Y-C, Dorn TW, Schache AG, Pandy MG. 2012 Comparison of different methods for estimating muscle forces in human movement. Proc. Inst. Mech. Eng. Part H J. Eng. Med. 226, 103–112. (doi:10.1177/0954411911429401) 38. Morrow MM, Rankin JW, Neptune RR, Kaufman KR. 2014 A comparison of static and dynamic optimization muscle force predictions during rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 17 wheelchair propulsion. J. Biomech. 47, 3459–3465. (doi:10.1016/j.jbiomech.2014.09.013) 39. Hutchinson JR, Rankin JW, Rubenson J, Rosenbluth KH, Siston RA, Delp SL. 2015 Musculoskeletal modelling of an ostrich (Struthio camelus) pelvic limb: influence of limb orientation on muscular capacity during locomotion. PeerJ 3, e1001. (doi:10. 7717/peerj.1001) 40. Delp SL, Anderson FC, Arnold AS, Loan P, Habib A, John CT, Guendelman E, Thelen DG. 2007 OpenSim: open-source software to create and analyze dynamic simulations of movement. IEEE Trans. Biomed. Eng. 54, 1940–1950. (doi:10.1109/TBME. 2007.901024) 41. DeMers MS, Pal S, Delp SL. 2014 Changes in tibiofemoral forces due to variations in muscle activity during walking. J. Orthop. Res. 32, 769–776. (doi:10.1002/jor.22601) 42. Thelen DG, Anderson FC. 2006 Using computed muscle control to generate forward dynamic simulations of human walking from experimental data. J. Biomech. 39, 1107–1115. (doi:10.1016/j. jbiomech.2005.02.010) 43. Gatesy SM. 1999 Guineafowl hind limb function. I: Cineradiographic analysis and speed effects. J. Morphol. 240, 115–125. (doi:10.1002/(SICI)1097- 4687(199905)240:2,115::AID-JMOR3.3.0.CO;2-Y) 44. Nyakatura JA, Andrada E, Grimm N, Weise H, Fischer MS. 2012 Kinematics and center of mass mechanics during terrestrial locomotion in northern lapwings (Vanellus vanellus, Charadriiformes). J. Exp. Zool. Part A Ecol. Genet. Physiol. 317, 580–594. (doi:10. 1002/jez.1750) 45. Millard M, Uchida T, Seth A, Delp SL. 2013 Flexing computational muscle: modeling and simulation of musculotendon dynamics. J. Biomech. Eng. 135, 021005. (doi:10.1115/1.4023390) 46. Burkholder TJ, Lieber RL. 2001 Sarcomere length operating range of vertebrate muscles during movement. J. Exp. Biol. 204, 1529–1536. 47. Lichtwark GA, Wilson AM. 2007 Is Achilles tendon compliance optimised for maximum muscle efficiency during locomotion? J. Biomech. 40, 1768–1775. (doi:10.1016/j.jbiomech.2006.07.025) 48. Manal K, Buchanan TS. 2004 Subject-specific estimates of tendon slack length: a numerical method. J. Appl. Biomech. 20, 195–203. 49. Spector SA, Gardiner PF, Zernicke RF, Roy RR, Edgerton VR. 1980 Muscle architecture and force– velocity characteristics of cat soleus and medial gastrocnemius: implications for motor control. J. Neurophysiol. 44, 951–960. 50. Gatesy SM. 1994 Neuromuscular diversity in archosaur deep dorsal thigh muscles. Brain Behav. Evol. 43, 1–14. (doi:10.1159/000113619) 51. Rankin JW, Richter WM, Neptune RR. 2011 Individual muscle contributions to push and recovery subtasks during wheelchair propulsion. J. Biomech. 44, 1246–1252. (doi:10.1016/j. jbiomech.2011.02.073) 52. Gatesy S. 1990 Caudefemoral musculature and the evolution of theropod locomotion. Paleobiology 16, 170–186. 53. Roberts TJ, Belliveau RA. 2005 Sources of mechanical power for uphill running in humans. J. Exp. Biol. 208, 1963–1970. (doi:10.1242/jeb.01555) 54. Pires NJ, Lay BS, Rubenson J. 2014 Joint-level mechanics of the walk-to-run transition in humans. J. Exp. Biol. 217, 3519–3527. (doi:10.1242/jeb.107599) 55. Kimmel SA, Schwartz MH. 2006 A baseline of dynamic muscle function during gait. Gait Posture 23, 211–221. (doi:10.1016/j.gaitpost.2005.02.004) 56. Zajac FE, Neptune RR, Kautz SA. 2002 Biomechanics and muscle coordination of human walking. Part I: introduction to concepts, power transfer, dynamics and simulations. Gait Posture 16, 215–232. (doi:10. 1016/S0966-6362(02)00068-1) 57. Neptune RR, Kautz SA, Zajac FE. 2001 Contributions of the individual ankle plantar flexors to support, forward progression and swing initiation during walking. J. Biomech. 34, 1387–1398. (doi:10.1016/ S0021-9290(01)00105-1) 58. Hicks JL, Uchida TK, Seth A, Delp SL. 2015 Is my model good enough? Best practices for verification and validation of musculoskeletal models and simulations of movement. J. Biomech. Eng. 137, 020905. (doi:10.1115/1.4029304) 59. Gangl D, Weissengruber GE, Egerbacher M, Forstenpointner G. 2004 Anatomical description of the muscles of the pelvic limb in the ostrich (Struthio camelus). Anat. Histol. Embryol. 33, 100–114. (doi:10.1111/j.1439-0264.2003.00522.x) 60. Guess TM, Liu H, Bhashyam S, Thiagarajan G. 2013 A multibody knee model with discrete cartilage prediction of tibio-femoral contact mechanics. Comput. Methods Biomech. Biomed. Engin. 16, 256–270. (doi:10.1080/10255842.2011.617004) 61. Bei Y, Fregly BJ. 2004 Multibody dynamic simulation of knee contact mechanics. Med. Eng. Phys. 26, 777–789. (doi:10.1016/j.medengphy. 2004.07.004) 62. Lerner ZF, DeMers MS, Delp SL, Browning RC. 2015 How tibiofemoral alignment and contact locations affect predictions of medial and lateral tibiofemoral contact forces. J. Biomech. 48, 644–650. (doi:10. 1016/j.jbiomech.2014.12.049) 63. Delp SL. 2001 Three-dimensional dynamic simulation of total knee replacement motion during a step-up task. J. Biomech. Eng. 123, 599. (doi:10. 1115/1.1406950) 64. Silder A, Whittington B, Heiderscheit B, Thelen DG. 2007 Identification of passive elastic joint moment–angle relationships in the lower extremity. J. Biomech. 40, 2628–2635. (doi:10. 1016/j.jbiomech.2006.12.017) 65. Davy DT, Audu ML. 1987 A dynamic optimization technique for predicting muscle forces in the swing phase of gait. J. Biomech. 20, 187–201. (doi:10. 1016/0021-9290(87)90310-1) 66. Domalain M, Vigouroux L, Berton E. 2010 Determination of passive moment–angle relationships at the trapeziometacarpal joint. J. Biomech. Eng. 132, 071009. (doi:10.1115/1.4001397) 67. Hof AL, Van Zandwijk JP, Bobbert MF. 2002 Mechanics of human triceps surae muscle in walking, running and jumping. Acta Physiol. Scand. 174, 17–30. (doi:10.1046/j.1365-201x.2002. 00917.x) 68. Blemker SS, Delp SL. 2006 Rectus femoris and vastus intermedius fiber excursions predicted by three-dimensional muscle models. J. Biomech. 39, 1383–1391. (doi:10.1016/j.jbiomech.2005.04. 012) 69. Srinivasan M, Ruina A. 2006 Computer optimization of a minimal biped model discovers walking and running. Nature 439, 72–75. (doi:10.1038/ nature04113) 70. Usherwood JR, Channon AJ, Myatt JP, Rankin JW, Hubel TY. 2012 The human foot and heel-sole-toe walking strategy: a mechanism enabling an inverted pendular gait with low isometric muscle force? J. R. Soc. Interface 9, 2396–2402. (doi:10.1098/rsif. 2012.0179) rsif.royalsocietypublishing.org J. R. Soc. Interface 13: 20160035 18

Inferring muscle functional roles of the ostrich pelvic limb during walking and running using computer optimization.

[]

Rankin, Jeffery W,Rubenson, Jonas,Hutchinson, John R

eng

PMC3503441

Hindawi Publishing Corporation Pulmonary Medicine Volume 2012, Article ID 542402, 10 pages doi:10.1155/2012/542402 Research Article Determination of Best Criteria to Determine Final and Initial Speeds within Ramp Exercise Testing Protocols Sidney C. da Silva,1 Walace D. Monteiro,2, 3 Felipe A. Cunha,2, 4 Jonathan Myers,5 and Paulo T. V. Farinatti2, 3 1Department of Sports Science, Brazilian Olympic Committee, Avenida das Am´ericas 899, 22631-000 Rio de Janeiro, RJ, Brazil 2Laboratory of Physical Activity and Health Promotion, Rio de Janeiro State University, Rua S˜ao Francisco Xavier 524, Sala 8121F, 20550-900 Rio de Janeiro, RJ, Brazil 3Graduate Program in Sciences of Physical Activity, Salgado de Oliveira University, Rua Marechal Deodoro 217, No. 2 Andar, 24030-060 Niteroi, RJ, Brazil 4Graduate Program in Medical Sciences, Rio de Janeiro State University, Avenida Professor Manoel de Abreu, 444/No. 2 Andar, Vila Isabel, 20550-170 Rio de Janeiro, RJ, Brazil 5Cardiology Division, Palo Alto VA Health Care System, Cardiology 111C, 3801 Miranda Avenue, Palo Alto, CA 94304, USA Correspondence should be addressed to Paulo T. V. Farinatti, pfarinatti@gmail.com Received 20 May 2012; Revised 25 September 2012; Accepted 25 September 2012 Academic Editor: Darcy D. Marciniuk Copyright © 2012 Sidney C. da Silva et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. This study compared strategies to define final and initial speeds for designing ramp protocols. VO2 max was directly assessed in 117 subjects (29 ± 8 yrs) and estimated by three nonexercise models: (1) Veterans Specific Activity Questionnaire (VSAQ); (2) Rating of Perceived Capacity (RPC); (3) Questionnaire of Cardiorespiratory Fitness (CRF). Thirty seven subjects (30 ± 9 yrs) performed three additional tests with initial speeds corresponding to 50% of estimated VO2 max and 50% and 60% of measured VO2 max. Significant differences (P < 0.001) were found between VO2 max measured (41.5 ± 6.6 mL·kg−1·min−1) and estimated by VSAQ (36.6±6.6 mL·kg−1·min−1) and CRF (45.0±5.3 mL·kg−1·min−1), but not RPC (41.3±6.2 mL·kg−1·min−1). The CRF had the highest ICC, the lowest SEE, and better limits of agreement with VO2 max compared to the other instruments. Initial speeds from 50%–60% VO2 max estimated by CRF or measured produced similar VO2 max (40.7±5.9; 40.0±5.6; 40.3±5.5 mL·kg−1·min−1 resp., P = 0.14). The closest relationship to identity line was found in tests beginning at 50% VO2 max estimated by CRF. In conclusion, CRF was the best option to estimate VO2 max and therefore to define the final speed for ramp protocols. The measured VO2 max was independent of initial speeds, but speeds higher than 50% VO2 max produced poorer submaximal relationships between workload and VO2. 1. Introduction Exercise capacity is an independent predictor of risk for car- diovascular disease and mortality among asymptomatic and symptomatic individuals [1–3]. Hence the determination of maximal oxygen uptake (VO2 max) is considered to be one of the most important health-related parameters and has been widely used to evaluate cardiorespiratory fitness in health and illness [4–7]. However, the determination of exercise capacity is closely related to the test protocol employed [8]. An extensive body of evidence has shown that ramp exercise protocols offer advantages over traditional protocols, because the increase in external work occurs in a constant and continuous fashion, and when designing the protocol the rate of increase in workload can be individualized by a previous estimate of maximal exercise capacity [7, 9–12]. This is associated with greater linearity between VO2 and work rate compared to 2 Pulmonary Medicine traditional protocols with large and disproportionate work rate increments [9, 11, 13]. Moreover, ramp protocols induce more uniform hemodynamic and respiratory responses, facilitating the acquisition of information at submaximal intensities, such as the ventilatory threshold [9, 13]. Despite the apparent advantages over traditional exercise testing, standardized criteria to guide the application of ramp protocols remain sparse. For instance, a limitation of ramp protocols is the requirement to estimate maximal exercise capacity from an activity scale and then adjust the ramp rate accordingly [14]. In practical terms, an underestimation of maximal exercise capacity will result in a prolonged total test duration, while an overestimation will result in premature test termination and, therefore, inappropriate test protocol for eliciting a true VO2 max [15]. However, there is no consensus in the literature concerning this issue. Available recommendations are generally vague and largely limited to the premise that tests should last between 8 and 12 min [4, 7, 14–17]. The same occurs with regard to the initial work rate of the test—actually we could not find recommendations of standard procedures for its determination [18]. Thus, the first objective of the present study was to compare three nonexercise models to predict maximal exercise capacity as criteria to determine the final speed of maximal treadmill ramp protocols. A second purpose was to investigate how different initial speeds calculated from %VO2 max influenced the VO2 max measured in the tests. 2. Material and Methods 2.1. Subjects. A group of 117 subjects (47 women) aged between 18 and 51 years (mean: 29.1 ± 7.6 yrs), with no previous experience in high performance physical training, volunteered for the study. Exclusion criteria included a clinical diagnosis of any clinical condition that could limit exercise performance and the use of any medication with potential cardiovascular influence. All participants were fully informed about the procedures and potential risks before giving written consent to take part in the study, which was approved by the local Institutional Research Ethics Committee. 2.2. Procedures. A flowchart of the 1st and 2nd studies is presented in Figure 1, detailing the procedures adopted to determine the workload increments using the nonexercise models (1st study—final speed) and different percent VO2 max intensities (2nd study—initial speed). All 117 subjects enrolled in the first study. After signing the informed consent, the subjects performed the following procedures in a single visit to the laboratory: (a) anthropometric measurements; (b) application of three nonexercise models to estimate VO2 max (Veterans Specific Activity Questionnaire (VSAQ), [19, 20]; Rating of Perceived Capacity (RPC) [21]; Questionnaire of Cardio-respiratory Fitness (CRF) [22]); (c) cardiopulmonary exercise testing. The VSAQ was originally developed by Myers et al. [19, 20] with the specific purpose of individualizing ramp protocols. The VSAQ includes a list of physical activities with scores ranging from 1 to 13. The responder indicates which of the listed activities would cause fatigue or shortness of breath. Subjects evaluated in the initial studies with the VSAQ had low cardiorespiratory fitness and a high prevalence of overweight/obesity, hypertension, or coronary disease. Even though further studies have demonstrated that the instrument also provided adequate estimation of VO2 max in healthy active populations [5, 8], there is a lack of research specifically designed to assess its validity within the application of ramp protocols in healthy subjects. The RPC may be considered a variation of the VSAQ [21], presenting different maximal MET levels (ranging from 1 to 20), which are linked to physical activities of several intensities. Subjects rate their perceived capacity by choosing the most strenuous activity they could sustain for 30 min. However, the RPC has been not validated through direct comparison with exercise capacity using cardiopulmonary exercise testing. The CRF was not specifically developed to design ramp protocols, but it has been extensively applied as a nonexercise model to estimate the maximal cardiorespiratory capacity [22]. It is a progressive scale with scores for the intensity of the activities ranging from 0 to 7. The subjects must select the most appropriate score according to the physical activities performed in the last 30 days. The CRF was selected because of the unusual methodological meticulousness applied to its development. A large sample (N = 799) of men and women aged 19 to 79 years was tested. The estimated VO2 max was compared to directly measured data, and the questionnaire was cross-validated with another population, which is uncommon in studies assessing such instruments [23, 24]. In the first study, the increase in work rate within the cardiopulmonary exercise test (CPET1) was individualized to elicit each subject’s limit of tolerance in 10 min, and treadmill grade was set at 0%. Final and initial speeds were determined using ACSM equations for treadmill running [7], considering the intensities corresponding to the highest VO2 max estimated by the non-exercise models (final speed) and 50% of this value (initial speed). The choice of 50% of the estimated VO2 max to determine the initial speed was based on a previous pilot study involving 35 subjects. In this pilot study, the initial speed was set at 1/3 of the estimated VO2 max, which corresponded to a mean speed of 4.3 km·h−1 and a work rate increase of 0.88 km·h−1 each minute. The protocols lasted approximately 12 min (11.3 ± 2.2 min) and subjects remained walking, for about 4 min. Thus, an intensity of 50% VO2 max would probably shorten the test and increase the time in which the subjects would be actually running. A subgroup of 37 subjects (17 women; age: 29.1±7.6 yrs) was randomly selected to participate in the second study. These subjects performed three additional cardiopulmonary exercise tests, separated by 72 to 120 h intervals. The increase in work rate and treadmill grade were the same applied in CPET1. In the first test (CPET1bis), the final speed was determined using the best non-exercise model as defined in the first study, and the initial speed set at 50% of this value. The other tests (CPET2 and CPET3) were then per- formed using the results of CPET1bis as reference. In brief, Pulmonary Medicine 3 CPET1 CPET1bis CPET2 CPET3 • Final speed calculated from the the three questionnaires (VSAQ, RPC ,and CRF). • Initial speed calculated from 50% of the estimated value. • Final speed calculated from CRF. • Initial speed calculated from 50% of the estimated value. • Final speed calculated from CPET1bis. • Initial speed calculated from 50% of the measured value. • Final speed calculated from the CPET1bis. • Initial speed calculated from 60% of the measured value. 1st study 2nd study • Anthropometric measurements. • is reported as mL·kg−1·min−1 • Speed in m·min−1 (converted to km·h−1) • G = grade expressed in decimal form (ii) Speedinitial = speedfinal × percentage (iii) Increment ratio = (speedfinal − speedinitial)/10 minutes Stages for determining the final and initial test speeds (i) Speedfinal = ( − 3.5) × 100%/(0.2 + 0.9 × G) (N = 117) (N = 37) highest estimated by the estimated by the measured during VO2 max VO2 max VO2 max VO2 max measured during VO2 max VO2 Figure 1: Flowchart of the 1st and 2nd studies including the procedures adopted to determine the workload increments, using nonexercise models to estimate VO2 max and ACSM running equation to calculate the treadmill speeds. VO2 max: maximal oxygen uptake; CPET: cardiopulmonary exercise test; VSAQ: Veterans Specific Activity Questionnaire; RPC: Rating of Perceived Capacity; CRF: Questionnaire of Cardiorespiratory Fitness. the final speed in CPET1bis was estimated from the maximal exercise capacity provided by CRF, whereas in both CPET2 and CPET3 it corresponded to the speed associated with the VO2 max assessed in CPET1bis. The initial speeds corre- sponded to 50% VO2 max estimated (CPET1bis), 50% VO2 max measured (CPET2), and 60% VO2 max measured (CPET3). This approach allowed to observe whether initial speeds ranging from 50 to 60% VO2 max (estimated or measured) influenced the results of the tests. In the first study the CPET1 was applied by a researcher blinded for the results of the non-exercise models. In the second study, the sequence of tests was defined by a counterbalanced crossover design. The participants were blinded for the %VO2 used to establish the initial speeds, and the evaluator was blinded for the purposes of the study. The cardiopulmonary exercise test protocols were per- formed using a super-ATL treadmill (Inbramed, Florianop- olis, SC, Brazil), and VO2 was averaged and recorded every 30 s. The 30 s time average provided a good compromise between removing noise from VO2 data while maintain- ing the underlying trend [25]. Data was assessed using a mouthpiece and noseclip. Gas exchange was assessed using a VO2000 analyzer (Medical Graphics, Saint Louis, MO, USA), which was calibrated with a certified standard mixture of oxygen (17.01%) and carbon dioxide (5.00%), balanced with nitrogen. The flows and volumes for the pneumotachograph were calibrated with a 3 L syringe (Hans Rudolph, Kansas, MO, USA). Heart rate was monitored using a Polar S-810 device (Polar, Kempele, Finland). Mean ambient temperature and relative humidity during testing were 22.4 ± 1.8◦C (range 18–23) and 62.5 ± 4.1% (range 50– 75%), respectively. The criteria for test interruption followed the recom- mendations of the American College of Sports Medicine [7]. The test was considered to achieve peak capacity when at least three of the following criteria were observed [26]: (a) maximum voluntary exhaustion as reflected by a score of 10 on the Borg CR-10 scale; (b) ≥95% predicted HR max (220—age) or presence of an HR plateau (ΔHR between two consecutive work rates ≤4 beats·min−1); (c) presence of a VO2 plateau (ΔVO2 between two consecutive work rates <2.1 mL·kg−1·min−1); (d) respiratory exchange ratio > 1.15. Participants were verbally encouraged to achieve maximal effort. Holding onto the side or front rails of the treadmill was not permitted. 2.3. Statistical Analyses. Data normality was confirmed by univariate analysis. Therefore the intraclass correlation coef- ficient (ICC) was used to verify the concordance between the VO2 max assessed in CPET1 and the VO2 max estimated by the non-exercise models. Limits of agreement and bias for measured and estimated VO2 max were determined according to the Bland and Altman method [27]. Intraclass correlation (ICC), R-square coefficients (r2), and standard errors of estimate (SEE) between actual and estimated VO2 max were also calculated. The VO2 max values obtained in CPET1bis, CPET2, and CPET3 were compared by repeated measures ANOVA. 4 Pulmonary Medicine Additionally, linear regression was performed for each sub- ject on each protocol in order to compare the relationships between workload and VO2, considering data in every 30 s of exercise. Mean ± SD values of intercepts and slopes were determined for each linear regression model. Student t-tests for paired samples were used to test whether the intercepts and slopes were significantly different from 0 and 1, respectively [12], and to test possible differences between the regression lines, as described in detail elsewhere [28]. The r2 and SEE for the regression models obtained in all tests were calculated as supplementary criteria to define the best initial speed. Two-tailed statistical significance for all tests was accepted as P ≤ 0.05. All statistical analyses were performed using Statistica 7.0 (Statsoft, Tulsa, OK, USA) and SPSS 8.0 (IBM, Chicago, IL, USA) statistical analysis software. 3. Results An achieved statistical power of 0.96 for an effect size of 0.25 was obtained by performing a post hoc power analysis (GPower version 3.0.10, Kiel, University of Kiel, Germany) based on the sample size, P value, number of repeated measures, and groups. Table 1 presents the characteristics of the samples comparing strategies to define final and initial speeds. Table 2 presents values for the assessed VO2 max (mL·kg−1·min−1) by age and sex groups. In the first study, mean duration of CPET1 was 13.3 ± 2.1 min for initial and final speeds of 5.9 ± 0.9 km·h−1 and 14.7 ± 2.1 km·h−1, respectively. Significant differences were detected between VO2 max assessed in CPET1 (41.5 ± 6.6 mL·kg−1·min−1) and VO2 max estimated from VSAQ and CRF (VO2 max VSAQ = 36.6±6.6 mL·kg−1·min−1, P < 0.0001; VO2 max CRF = 45.0 ± 5.3 mL·kg−1·min−1; P < 0.0001), but not from RPC (VO2 max RPC = 41.3 ± 6.2 mL·kg−1·min−1, P = 0.99). Figure 2 shows the Bland-Altman analysis, including the limits of agreement for estimated and measured VO2 max. Table 3 presents values for R-square, SEE, and ICC between VO2 max measured and estimated by the questionnaires. The RPC provided the lowest mean difference between VO2 max directly assessed in CPET1 and estimated from the questionnaires (RPC = 0.24 mL·kg−1·min−1; CRF = −3.54 mL·kg−1·min−1; VSAQ = 4.94 mL·kg−1·min−1; P = 0.05). However, the CRF exhibited better limits of agreement compared to the other instruments. The higher values obtained for CRF with regard to R-square and ICC were consistent with the results of the Bland- Altman analysis. The SEE between assessed and estimated VO2 max was also lower in CRF compared to VSAQ and RPC. Table 4 shows the distribution of VO2 max assessed in CPET1 according to tertiles, as well the percent agreement between estimated and measured VO2 max in each tertile. The nonparametric Kendall’s tau-b correlation between tertiles was similar across the three questionnaires and measured VO2 max. However the correlation using the CRF was higher over RPC and VSAQ—the proportion of subjects assigned in the same tertile category was superior for CRF compared to the other questionnaires, and the distribution was more homogeneous. With regard to the second study, mean durations of CPET1bis, CPET2, and CPET3 were 13.7 ± 1.8 min, 10.7 ± 1.9 min, and 10.6 ± 0.9 min, respectively. No differences were detected between VO2 max assessed in CPET1bis (used as reference to define final and initial speeds in CPET2 and CPET3), CPET2, and CPET3 (CPET1bis = 40.7 ± 5.9 mL·kg−1·min−1; CPET2 = 39.8 ± 5.6 mL·kg−1·min−1; CPET3 = 40.3 ± 5.5 mL·kg−1·min−1; P = 0.142). Mean initial speeds applied in CPET1bis, CPET2, and CPET3 were 5.7 ± 0.8 km·h−1, 8.1 ± 0.9 km·h−1, and 9.1 ± 1.1 km·h−1, respectively. Table 5 shows the relationships between work- load and VO2 in the ramp test protocols initiating with speeds corresponding to 50% and 60% VO2 max either measured or estimated (slopes, intercepts, R-square, and SEE). CPET1bis showed the closest relationship with the theoretical identity line (slope = 1 and intercept = 0), with the highest R-square and lowest SEE in comparison with CPET2 and CPET3. 4. Discussion The present study aimed to compare different strategies to define final and initial speeds when designing ramp exercise testing protocols for healthy young populations. Three nonexercise models were employed to estimate maximal cardiorespiratory capacity and therefore the final speed. The choice of VSAQ, RPC, and CRF to estimate the VO2 max was due to the fact that these instruments have been frequently applied in previous studies and have been shown to have good potential to estimate the maximal cardiorespiratory capacity in different populations [23, 24]. Two relative intensities (%VO2 max) using different initial treadmill speeds were tested. The values obtained for the VO2 max assessed in CPET1 are consistent with reference values reported by previous research [4, 7, 14, 16]. Our findings on the ICC, R-square, SEE, and dispersion in the Bland-Altman plot (see Figure 2) suggest that there are advantages in using the CRF to determine the final speed, in comparison with the other instruments. In contrast, the VSAQ had the poorest precision and highest variability with respect to VO2 max estimation. In their original study, Myers et al. [19] reported a stronger association between estimated and achieved cardiorespira- tory capacity over the present data (r = 0.79; SEE = 4.97 mL·kg−1·min−1; P = 0.001 versus r = 0.40; SEE = 7.63 mL·kg−1·min−1; P = 0.0001, resp.). However, subjects in the two studies differed considerably in terms of clinical and fitness status, which may have contributed to such discrepancy, since poor conditioned individuals are more likely to interrupt earlier the test due to peripheral fatigue. Moreover, Myers et al. [19] did not directly assess the VO2 max in their original research. In a later study, these investigators [20] validated the VSAQ measuring VO2 max directly in a larger sample (n = 337). Subjects had similar characteristics as those in the original study, but the results were more Pulmonary Medicine 5 Table 1: Characteristics of the subjects participating in the comparisons regarding the final (N = 117) and initial (N = 37) speeds. Age (years) Body mass (kg) Height (cm) Body fat (%) VO2 max (mL·kg−1·min−1) G M F G2 G M F G2 G M F G2 G M F G2 G M F G2 Mean 29.1 29.8 28.2 29.7 71.7 79.7 59.7 72.4 171.2 176.7 163.1 170.4 15.2 11.7 20.4 16.8 41.5 43.9 37.8 40.7 SD 7.6 7.9 7.0 8.6 14.9 12.6 8.8 17.9 9.1 5.9 6.5 10.0 7.0 5.8 5.3 6.5 6.6 5.8 6.1 5.9 Minimum 18 18 19 18 46.7 57.4 46.7 46.5 150.0 163.3 150.0 150.5 2.9 2.9 11.2 6.0 25.6 32.8 25.6 28.5 Maximum 51 47 51 51 92.9 92.9 91.8 90.3 190.0 190.0 176.0 190.0 32.8 27.1 32.8 28.3 61.6 61.6 54.8 52.4 G: total sample (n = 117); M: males (n = 70); F: females (n = 47); G2: subgroup for initial speed comparison (n = 37). 6 Pulmonary Medicine Table 2: Descriptive values for VO2 max (mL·kg−1·min−1) by age and sex groups. Age (years) Males (N = 70) Females (N = 47) 18–29 (N = 39) 30–39 (N = 20) >40 (N = 11) 18–29 (N = 32) 30–39 (N = 10) >40 (N = 5) Mean 46.2 41.1 39.4 39.0 34.3 37.0 SD 5.8 4.4 3.9 5.9 6.3 4.6 Minimum 36.5 32.8 34.0 26.2 25.6 29.5 Maximum 61.5 47.9 44.2 54.8 45.3 40.4 Table 3: Mean difference (mL·kg−1·min−1), R-square coefficient, standard error of estimate, and intraclass correlation between VO2 max assessed and estimated by three non-exercise models (N = 117). Total (N = 117) Males (N = 70) Females (N = 47) VO2 max VO2 max VO2 max Mean difference r2 SEE ICC P Mean difference r2 SEE ICC P Mean difference r2 SEE ICC P VSAQ 4.94 (11.9%) 0.16 7.63 0.57 <0.0001 −1.81 (−4.1%) 0.05 7.92 0.36 <0.0317 −1.24 (−3.3%) 0.07 7.17 0.42 <0.040 RPC 0.24 (1.0%) 0.09 7.60 0.46 <0.001 3.22 (7.3%) 0.17 1.70 0.58 <0.0001 −3.49 (−9.2%) 0.07 8.35 0.42 <0.035 CRF −3.54 (−8.5%) 0.53 5.75 0.83 <0.0001 −3.89 (−8.9%) 0.37 6.01 0.76 <0.0001 −2.90 (−7.7%) 0.47 5.36 0.81 <0.0001 VSAQ: Veteran Specific Activity Questionnaire using the following equation: VO2 (mL·kg−1·min−1) = (4.7 + 0.97 (VSAQ) − 0.06 (age) × 3.5); for women this value was multiplied by 0.85 [8]; RPC: Rating of Perceived Capacity; CRF: Cardiorespiratory Fitness. similar to our findings (r = 0.42; SEE = 9.1 mL·kg−1·min−1; P = 0.001). Maeder et al. [5] compared the VO2 max obtained in tests using cycle ergometer and treadmill with the exercise capacity estimated by the VSAQ in healthy subjects. The cor- relations were similar to our data (cycle ergometer: r = 0.46 and treadmill: r = 0.50; P < 0.0001). More recently, Maeder et al. [8] used the VSAQ to select the optimal treadmill ramp protocol in highly trained individuals and reported a similar correlation between estimated and measured VO2 max (r = 0.47), even when using the VSAQ modified nomogram (r = 0.56). Although the VSAQ was developed to facilitate the individualization of ramp protocols, previous research has not ratified this purpose in all populations. Actually, the available evidence does not support its use in determining the final speed within ramp protocols in healthy and well- conditioned populations. Actually the VSAQ has been shown to be more appropriate to estimate the VO2 max in unfit individuals [20, 29]. The present results confirm this idea. Precision using the VSAQ was lower compared to the other instruments, and the same categorization was obtained in less than 40% of cases. Furthermore, the Bland-Altman plots suggested that in our sample the VO2 max was systematically overestimated by the VSAQ. The RPC closely paralleled VO2 max assessed in CPET1 (mean difference of 0.24 mL·kg−1·min−1 or 1%), but exhib- ited high variability, as evidenced by the Bland-Altman method and SEE (7.60 mL·kg−1·min−1). This variation accounted for the relatively low ICC and R-square values. It is noteworthy that RPC was developed in a sample of 87 young, healthy women (age = 48.4 ± 17.4 years) [21]. However, our experience with this method suggests that strong agreement between estimated and actual VO2 max can be also obtained in men. Interestingly, although our sample consisted of young women (age = 28.2 ± 7.0 years), the comparison between VO2 max directly measured and estimated by RPC showed greater concordance (ICC) and lower variation (SEE) among men versus women (ICC = 0.58 versus 0.42 and SEE = 1.70 mL·kg−1·min−1 versus 8.35 mL·kg−1·min−1, resp.). A possible explanation for this is that in the original RPC study the VO2 max was estimated from the work performed on cycle ergometer, and not directly measured. The VO2 max was estimated using maximal work and body mass, assuming as constants the amount of oxygen required for each Watt of power during ramp cycling (10.93 mL·min−1·W−1) and VO2 at rest when sitting on the cycle (4.3 mL·min−1). However these unpublished data have been previously determined in a group of healthy men [21], and no information was provided with regard to their possible application in females. The CRF has been widely used to estimate maximal cardiorespiratory capacity [12, 30–35]. Although it was not originally developed to help designing ramp protocols, our results indicate that it works well for this purpose. The original study by Matthews et al. [22] showed a higher correlation between VO2 max measured and estimated from CRF than the present study, in a sample of 390 men (r = 0.82 versus r = 0.61, resp.) and 409 women (r = 0.83 versus r = 0.69, resp.). However, the SEEs in the total sample (5.7 mL·kg−1·min−1 versus 5.8 mL·kg−1·min−1) and in gender subgroups (men: 6.3 mL·kg−1·min−1 ver- sus 6.0 mL·kg−1·min−1; women: 5.0 mL·kg−1·min−1 ver- sus 5.4 mL·kg−1·min−1) were similar in the two studies. The Bland-Altman analysis showed limits of agreement higher over VSAQ and comparable to RPC, but the CRF had the greatest ICC. In addition, the tertile classifications obtained from CRF were more accurate compared to the other nonexercise models. Overall, CRF showed higher concordance with measured VO2 max, lower dispersion, and better capacity to discriminate Pulmonary Medicine 7 20 30 40 50 60 70 −35 −30 −25 −20 −15 −10 −5 0 5 10 15 20 25 30 35 −12.1 22 Bias 4.9 difference (mL·kg−1·min−1) Mean (mL·kg−1·min−1) Sd Sd VO2 VO2 (a) −14.7 15.2 Bias 0.2 20 30 40 50 60 70 −35 −30 −25 −20 −15 −10 −5 0 5 10 15 20 25 30 35 difference (mL·kg−1·min−1) (mL·kg−1·min−1) Sd Sd MeanVO2 VO2 (b) −12.5 5.4 Bias 20 30 40 50 60 70 −35 −30 −25 −20 −15 −10 −5 0 5 10 15 20 25 30 35 difference (mL·kg−1·min−1) Mean (mL·kg−1·min−1) Sd Sd −3.5 VO2 VO2 (c) Figure 2: Bland-Altman plot for the individual differences between VO2 max assessed in CPET1 and VO2 max estimated by VSAQ (a), RPC (b), and CRF (c). The first and third horizontal dashed lines in each graph represent the 95% limits of agreement for VSAQ, RPC, CRF, and VSAQ, corresponding, respectively, to −12.1 to 22.0 (−29.1 to 53.0%); −14.7 to 15.2 (−35.5 to 36.6%); −12.5 to 5.4 (−30,0% to 13,0%). Sd: standard deviation of the differences. Table 4: Percentage of participants ranked in the same tertile, percentage of total agreement, tau-b correlation coefficients between VO2 max measured and estimated by three non-exercise models (VSAQ, RPC, and CRF) (N = 117). 1st Tertile (n = 39) 2nd Tertile (n = 39) 3rd Tertile (n = 39) Total (N = 117) R (tau-b) VO2 max versus VSAQ 66.66% (26) 5.12% (2) 38.46% (15) 36.75% (43) 0.833 VO2 max versus RPC 43.58% (17) 25.64% (10) 43.58% (17) 37.60% (44) 0.992 VO2 max versus CRF 69.23% (27) 41.02% (16) 58.97% (23) 56.41% (66) 0.983 VSAQ: Veteran Specific Activity Questionnaire; RPC: Rating of Perceived Capacity; CRF: Questionnaire of Cardiorespiratory Fitness. subjects with high and low cardio-respiratory capacity in comparison to VSAQ and RPC. Notably, the CRF may be limited when assessing cardiorespiratory capacity in subjects with VO2 max > 55.0 mL·kg−1·min−1 [29], which could be a problem when designing ramp protocols in highly fit individuals. However, fewer than 20% of ordinary healthy individuals achieve this level [7]. It therefore seems unlikely that the final speed would be wrongly determined from inaccurate estimation of VO2 max estimation, at least in most healthy nonathletic subjects. In what concerns the second study, the literature is mixed regarding criteria to determine the initial speed for ramp testing [9, 11]. Recommendations from different expert panels are also ambiguous with regard to this issue 8 Pulmonary Medicine Table 5: Intercept, slope, R-square (r2), and standard error of estimate (SEE) for the regression models obtained in ramp protocols initiating with speeds corresponding to 50% of the estimated VO2 max (CPET1bis), 50% of the measured VO2 max (CPET2), and 60% of the measure VO2 max (CPET3). Y intercept Slope r-Square SEE (mL·kg−1·min−1) VO2versus speed in CPET1bis −4.882 ± 2.696∗ 0.96 ± 0.027§ 0.93 ± 0.050 2.14 ± 0.67 VO2versus speed in CPET2 −8.270 ± 6.312∗ 0.94 ± 0.029§ 0.89 ± 0.054 2.19 ± 0.55 VO2versus speed in CPET3 −14.666 ± 8.958∗ 0.92 ± 0.036§ 0.86 ± 0.065 2.48 ± 0.67 ∗Intercept significantly different from zero (P < 0.0001). §Slope significantly different from 1.0 (P < 0.0001). [4, 7, 14, 15], and no formal criteria are available on this important aspect of ramp protocols. Our findings suggested that initial speeds within the range corresponding to 50% to 60% VO2 max influenced the duration of the test (CPET1bis = 13.7 ± 1.8 min > CPET2 = 10.7 ± 0.9 min ∼= CPET3 = 10.6 ± 0.9 min, P < 0.0001), but not the achieved VO2 max (CPET1bis = 40.7 ± 5.9 mL·kg−1·min−1 ∼= CPET2 = 40.0 ± 5.6 mL·kg−1·min−1 ∼= CPET3 = 40.3 ± 5.5 mL·kg−1·min−1, P = 0.14). From these results, any initial speed within this range would be appropriate for performing ramp tests. In contrast, the relationship between workload and VO2 among the tests was affected by the initial speed. Considering the identity line as a reference for the ideal regression between workload and VO2, the current results suggest that higher initial speed produced the lowest R-squares (e.g., poorest adjustment to the identity line) (CPET3— 60% VO2 max < CPET2—50% VO2 max < CPET1bis—50% VO2 max). Early research confirms the concept that the initial speed applied does not influence measured VO2 max. Kang et al. compared three incremental treadmill protocols ( ˚Astrand, Bruce, and Costill/Fox) in 25 sedentary sub- jects (10 women) [36]. The protocols began with speeds of 9.7 km·h−1, 2.5 km·h−1, and 14.4 km·h−1, respectively, and no differences in VO2 max were detected. The rela- tionship between workload and VO2 was not specifically addressed, but the authors considered that this could have been good, at least in the Costill/Fox protocol. The high initial speed significantly shortened the tests (to about 5 min) and precluded the identification of the ventilatory threshold. In 1991, Myers et al. compared VO2 max obtained during ramp and conventional staged protocols (Bruce and Balke modified), which were very different with regard to the combination of initial speed, treadmill grade, and workload increment. The duration of tests was significantly different (Bruce: 6.6 ±1.5 min versus Balke: 10.4 ±3.4 min and Ramp: 9.1 ± 1.4 min, P < 0.05), with little impact on VO2 max (Bruce: 22.3 ± 8.0 mL·kg−1·min−1 versus Balke: 21.1 ± 8.0 mL·kg−1·min−1 and Ramp: 21.0 ± 8.0 mL·kg−1·min−1, P < 0.05). However, slopes and SEE for the regres- sion curves between workload and VO2 showed more linear relationships in the ramp protocol (Bruce: slope = 0.62 and SEE = 4.0 mL·kg−1·min−1; Balke: Slope = 0.79 and SEE = 3.4 mL·kg−1·min−1; Ramp: Slope = 0.80 and SEE = 2.5 mL·kg−1·min−1). In other words, differences in the protocol design may reflect on physiological rela- tionships in submaximal workloads, but not necessarily on the assessed VO2 max. Our findings seem to ratify this idea. In conclusion, CRF was superior in comparison with RPC and VSAQ to estimate maximal cardio-respiratory capacity and should be preferred when attempting to deter- mine an appropriate speed for ramp testing. Initial speeds within the range corresponding to 50–60% VO2 max estimated or measured did not affect assessed VO2 max. Nevertheless, speeds higher than 50% VO2 max may influence the quality of submaximal relationships between work rate and VO2. Moreover, higher speeds applied at the beginning of ramp protocols may hinder the performance of subjects with poor fitness levels and compromise test results. This information should be considered when data from exercise testing is used to establish relative exercise intensities for exercise prescription. Acknowledgments This paper was supported by grants from FAPERJ (Carlos Chagas Foundation for the Research Support in the State of Rio de Janeiro, Rio de Janeiro, Brazil) and CNPq (Brazilian Council for the Technological and Scientifical Development, Bras´ılia, Brazil). References [1] S. Mora, R. F. Redberg, Y. Cui et al., “Ability of exercise testing to predict cardiovascular and all-cause death in asymptomatic women: a 20-year follow-up of the lipid research clinics prevalence study,” JAMA, vol. 290, no. 12, pp. 1600–1607, 2003. [2] M. Gulati, H. R. Black, L. J. Shaw et al., “The prognostic value of a nomogram for exercise capacity in women,” The New England Journal of Medicine, vol. 353, no. 5, pp. 468–475, 2005. [3] E. S. H. Kim, H. Ishwaran, E. Blackstone, and M. S. Lauer, “External prognostic validations and comparisons of age- and gender-adjusted exercise capacity predictions,” Journal of the American College of Cardiology, vol. 50, no. 19, pp. 1867–1875, 2007. [4] R. J. Gibbons, G. J. Balady, J. T. Bricker et al., “ACC/AHA 2002 guideline update for exercise testing: summary article. A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to update the 1997 exercise testing guidelines),” Journal of the Pulmonary Medicine 9 American College of Cardiology, vol. 40, no. 8, pp. 1531–1540, 2002. [5] M. Maeder, T. Wolber, R. Atefy et al., “Impact of the exercise mode on exercise capacity: bicycle testing revisited,” Chest, vol. 128, no. 4, pp. 2804–2811, 2005. [6] P. G. Snell, J. Stray-Gundersen, B. D. Levine, M. N. Hawkins, and P. B. Raven, “Maximal oxygen uptake as a parametric measure of cardiorespiratory capacity,” Medicine and Science in Sports and Exercise, vol. 39, no. 1, pp. 103–107, 2007. [7] American College of Sports Medicine (ACSM), Guidelines of Exercise Testing and Exercise Prescription, Lea & Febiger, Philadelphia, Pa, USA, 8nd edition, 2009. [8] M. Maeder, T. Wolber, R. Atefy et al., “A nomogram to select the optimal treadmill ramp protocol in subjects with high exercise capacity: validation and comparison with the Bruce protocol,” Journal of Cardiopulmonary Rehabilitation, vol. 26, no. 1, pp. 16–23, 2006. [9] J. Myers, N. Buchanan, D. Walsh et al., “Comparison of the ramp versus standard exercise protocols,” Journal of the American College of Cardiology, vol. 17, no. 6, pp. 1334–1342, 1991. [10] B. J. Whipp, J. A. Davis, F. Torres, and K. Wasserman, “A test to determine parameters of aerobic function during exercise,” Journal of Applied Physiology Respiratory Environmental and Exercise Physiology, vol. 50, no. 1, pp. 217–221, 1981. [11] J. Myers, N. Buchanan, D. Smith et al., “Individualized ramp treadmill; Observations on a new protocol,” Chest, vol. 101, no. 5, pp. 236–241, 1992. [12] F. A. Cunha, A. W. Midgley, W. D. Monteiro, and P. T. V. Farinatti, “Influence of cardiopulmonary exercise testing protocol and resting VO2 Assessment on %HR2max, %HRR, %VO2max and %VO2R Relationships,” International Journal of Sports Medicine, vol. 31, no. 5, pp. 319–326, 2010. [13] J. Myers and D. Bellin, “Ramp exercise protocols for clinical and cardiopulmonary exercise testing,” Sports Medicine, vol. 30, no. 1, pp. 23–29, 2000. [14] G. F. Fletcher, G. J. Balady, E. A. Amsterdam et al., “Exercise standards for testing and training: a statement for healthcare professionals from the American Heart Association,” Circula- tion, vol. 104, no. 14, pp. 1694–1740, 2001. [15] A. W. Midgley, D. J. Bentley, H. Luttikholt, L. R. McNaughton, and G. P. Millet, “Challenging a dogma of exercise physiology: does an incremental exercise test for valid V ·O2max determi- nation really need to last between 8 and 12 minutes?” Sports Medicine, vol. 38, no. 6, pp. 441–447, 2008. [16] American Thoracic Society/American College of Chest Physi- cians, “ATS/ACCP Statement on cardiopulmonary exercise testing,” American Journal of Respiratory and Critical Care Medicine, vol. 167, no. 2, pp. 211–277, 2003. [17] A. Mezzani, P. Agostoni, A. Cohen-Solal et al., “Standards for the use of cardiopulmonary exercise testing for the functional evaluation of cardiac patients: a report from the exercise physiology section of the European association for cardio- vascular prevention and rehabilitation,” European Journal of Cardiovascular Prevention and Rehabilitation, vol. 16, no. 3, pp. 249–267, 2009. [18] S. C. da Silva, W. D. Monteiro, and P. T. V. Farinatti, “Exercise maximum capacity assessment: a review on the traditional protocols and the evolution to individualized models,” Revista Brasileira de Medicina do Esporte, vol. 17, no. 5, pp. 363–369, 2011. [19] J. Myers, W. Herbert, P. Ribisl, and V. F. Froelicher, “A nomogram to predict exercise capacity from a specific activity questionnaire and clinical data,” American Journal of Cardiol- ogy, vol. 73, no. 8, pp. 591–596, 1994. [20] J. Myers, D. Bader, R. Madhavan, and V. Froelicher, “Val- idation of a specific activity questionnaire to estimate exercise tolerance in patients referred for exercise testing,” American Heart Journal, vol. 142, no. 6, pp. 1041–1046, 2001. [21] A. G. M. Wis´en, R. G. Farazdaghi, and B. Wohlfart, “A novel rating scale to predict maximal exercise capacity,” European Journal of Applied Physiology, vol. 87, no. 4-5, pp. 350–357, 2002. [22] C. E. Matthews, D. P. Heil, P. S. Freedson, and H. Pastides, “Classification of cardiorespiratory fitness without exercise testing,” Medicine and Science in Sports and Exercise, vol. 31, no. 3, pp. 486–493, 1999. [23] G. A. Maranh˜ao Neto, P. M. Lourenc¸o, and P. T. Farinatti, “Prediction of aerobic fitness without stress testing and applicability to epidemiological studies: a systematic review,” Cadernos de Sa´ude P´ublica, vol. 20, no. 1, pp. 48–56, 2004. [24] G. De Albuquerque Maranh˜ao Neto, A. C. M. P. De Leon, and P. De Tarso Veras Farinatti, “Cross-cultural equivalence of three scales used to estimate cardiorespiratory fitness in the elderly,” Cadernos de Sa´ude P´ublica, vol. 24, no. 11, pp. 2499– 2510, 2008. [25] A. W. Midgley, L. R. Mcnaughton, and S. Carroll, “Effect of the ˙Vo2 time-averaging interval on the reproducibility of ˙Vo2max in healthy athletic subjects,” Clinical Physiology and Functional Imaging, vol. 27, no. 2, pp. 122–125, 2007. [26] E. T. Howley, D. R. Bassett, and H. G. Welch, “Criteria for maximal oxygen uptake: review and commentary,” Medicine and Science in Sports and Exercise, vol. 27, no. 9, pp. 1292– 1301, 1995. [27] J. M. Bland and D. G. Altman, “Statistical methods for assess- ing agreement between two methods of clinical measurement,” The Lancet, vol. 1, no. 8476, pp. 307–310, 1986. [28] J. H. Zar, Comparing Simple Linear Regression Equations. Bio- statistical Analysis, Prentice-Hall, Englewoods Cliff, NJ,USA, 1984. [29] P. McAuley, J. Myers, J. Abella, and V. Froelicher, “Evaluation of a specific activity questionnaire to predict mortality in men referred for exercise testing,” American Heart Journal, vol. 151, no. 4, pp. 890.e1–890.e7, 2006. [30] A. S. Jackson, S. N. Blair, M. T. Mahar, L. T. Wier, R. M. Ross, and J. E. Stuteville, “Prediction of functional aerobic capacity without exercise testing,” Medicine and Science in Sports and Exercise, vol. 22, no. 6, pp. 863–870, 1990. [31] D. P. Heil, P. S. Freedson, L. E. Ahlquist, J. Price, and J. M. Rippe, “Nonexercise regression models to estimate peak oxygen consumption,” Medicine and Science in Sports and Exercise, vol. 27, no. 4, pp. 599–606, 1995. [32] J. D. George, W. J. Stone, and L. N. Burkett, “Non-exercise ·VO2max estimation for physically active college students,” Medicine and Science in Sports and Exercise, vol. 29, no. 3, pp. 415–423, 1997. [33] R. Jurca, A. S. Jackson, M. J. LaMonte et al., “Assessing car- diorespiratory fitness without performing exercise testing,” American Journal of Preventive Medicine, vol. 29, no. 3, pp. 185–193, 2005. [34] D. I. Bradshaw, J. D. George, A. Hyde et al., “An accu- rate VO2max nonexercise regression model for 18-65-year-old adults,” Research Quarterly for Exercise and Sport, vol. 76, no. 4, pp. 426–432, 2005. [35] L. T. Wier, A. S. Jackson, G. W. Ayers, and B. Arenare, “Nonexercise models for estimating V·O2max with waist 10 Pulmonary Medicine girth, percent fat, or BMI,” Medicine and Science in Sports and Exercise, vol. 38, no. 3, pp. 555–561, 2006. [36] J. Kang, E. C. Chaloupka, M. A. Mastrangelo, G. B. Biren, and R. J. Robertson, “Physiological comparisons among three maximal treadmill exercise protocols in trained and untrained individuals,” European Journal of Applied Physiology, vol. 84, no. 4, pp. 291–295, 2001.

Determination of Best Criteria to Determine Final and Initial Speeds within Ramp Exercise Testing Protocols.

11-01-2012

da Silva, Sidney C,Monteiro, Walace D,Cunha, Felipe A,Myers, Jonathan,Farinatti, Paulo T V

eng

PMC7432299

International Journal of Environmental Research and Public Health Review Decreased Blood Glucose and Lactate: Is a Useful Indicator of Recovery Ability in Athletes? Woo-Hwi Yang 1,* , Hyuntae Park 2,* , Marijke Grau 3 and Oliver Heine 4 1 Graduate School of Sports Medicine, CHA University, Seongnam-si, Gyeonggi-do 13503, Korea 2 Department of Health Care and Science, College of Health Science, Dong-A University, Busan 49315, Korea 3 Department of Molecular and Cellular Sports Medicine, Institute of Cardiovascular Research and Sports Medicine, German Sport University Cologne, 50933 Cologne, Germany; m.grau@dshs-koeln.de 4 Olympic Training Centre Rhineland, 50933 Cologne, Germany; heine@osp-rheinland.de * Correspondence: ywh1235@cha.ac.kr (W.-H.Y.); htpark@dau.ac.kr (H.P.) Received: 9 July 2020; Accepted: 28 July 2020; Published: 29 July 2020



 Abstract: During low-intensity exercise stages of the lactate threshold test, blood lactate concentrations gradually diminish due to the predominant utilization of total fat oxidation. However, it is unclear why blood glucose is also reduced in well-trained athletes who also exhibit decreased lactate concentrations. This review focuses on decreased glucose and lactate concentrations at low-exercise intensity performed in well-trained athletes. During low-intensity exercise, the accrued resting lactate may predominantly be transported via blood from the muscle cell to the liver/kidney. Accordingly, there is increased hepatic blood flow with relatively more hepatic glucose output than skeletal muscle glucose output. Hepatic lactate uptake and lactate output of skeletal muscle during recovery time remained similar which may support a predominant Cori cycle (re-synthesis). However, this pathway may be insufficient to produce the necessary glucose level because of the low concentration of lactate and the large energy source from fat. Furthermore, fatty acid oxidation activates key enzymes and hormonal responses of gluconeogenesis while glycolysis-related enzymes such as pyruvate dehydrogenase are allosterically inhibited. Decreased blood lactate and glucose in low-intensity exercise stages may be an indicator of recovery ability in well-trained athletes. Athletes of intermittent sports may need this recovery ability to successfully perform during competition. Keywords: aspartate transaminase; Cori cycle; hepatic blood flow; oxaloacetate; phosphoenolpyruvate carboxykinase; pyruvate dehydrogenase 1. Introduction Clinical physicians and sports scientists have used lactate threshold (LT) tests for over fifty years because their application is considered extremely useful for recommendations on individual exercise intensity in cardiac patients and trained athletes [1,2]. Endurance athletes regularly undergo these tests in order to control individual exercise intensity during endurance training [1,3–6]. Both respiratory and metabolic parameters are commonly utilized to identify the anaerobic threshold [1] and oxygen uptake (VO2) during exercise performance influenced by the percentage of maximal oxygen uptake (VO2max) at LT. The workout test is performed either on a bicycle ergometer or on a treadmill applying different steps [1,7]. The ramp test is applied to determine VO2max and lactate values at each step in order to analyze the metabolic system and physiological performance [8]. The number of scientific studies on LT has increased enormously and among the diagnostics of endurance performance in sports, submaximal exercise is probably one of the most relevant [1,2,5,9–13]. For instance, increased exercise intensity at four millimoles per liter lactate was commonly observed as the lactate threshold in Int. J. Environ. Res. Public Health 2020, 17, 5470; doi:10.3390/ijerph17155470 www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2020, 17, 5470 2 of 16 endurance-trained athletes, and this value is highly associated with the potential maximal lactate steady state level (MLSS) [4,14,15]. A rightward shift of the exponential lactate curve can generally be interpreted as improved endurance capacity [2,16–18]. Furthermore, validated LT concepts such as aerobic-anaerobic transition using lactate and gas exchange parameters were applied and refined by several scientists [2,4,10,12,18–24]. To measure the exercise capacity, numerous studies have been focused on altered blood glucose concentrations following moderate-to-high-intensity exercise in LT tests. The metabolic changes in blood glucose concentration during low-intensity exercise in LT test are not analyzed [2,4,14,15,18–23,25–33]. Glucose 6-phosphate, supplied through breakdown of muscle glycogen and blood glucose, is metabolized to lactate and re-synthesized to adenosine triphosphate (ATP) by substrate-level phosphorylation reactions [34]. The blood glucose of endurance-trained athletes is decreased during the early stages of LT testing while blood lactate concentration (below lactate baseline concentration; LTAer or < two millimoles per liter) is also reduced. This exercise area is commonly referred to as regenerative endurance training [2]. In these low exercise stages, it seems likely that blood lactate concentrations gradually decrease as a result of the predominance of total fat oxidation [2,14,25–27]. In terms of energy metabolism, fat is also used as an energy source and represents the main energy source in moderate exercise under aerobic conditions. However, fat oxidation cannot predominantly be used to meet the energy demand during high-intensity exercise. Under this condition, carbohydrate oxidation represents the primary source of energy [25,35]. In turn, at low-intensity, triglycerides in adipocytes are hydrolyzed into glycerol and free fatty acids (lipolysis) which are then converted into acetyl-CoA by ß-oxidation in the mitochondria. At low-intensity exercise levels of 25% VO2max, plasma fatty acids are delivered for energy production [25,36]. In light of this, it is understandable why lactate values in blood begin to decrease at this exercise intensity as more pyruvate and lactate are used aerobically than are generated via anaerobic glycolysis [14]. However, the reduction in blood glucose during low-intensity exercise is difficult to explain. Blood glucose concentrations are usually increased incrementally with exercise from low to high intensity because carbohydrate metabolism partly contributes to aerobic glycolysis during low-intensity exercise [14,25]. The aim of this literature review was to describe possible relationships between exercise intensity, glucose and lactate at the low-intensity exercise stages of the LT test. To date, it is unclear why blood glucose is reduced while lactate values are also decreased during low-intensity exercise. Therefore, comprehensive aspects of the underlying physiological and molecular biologic background are considered. We suggest that decreased blood glucose and lactate at low-intensity exercise (LT test) are relevant signals for the recovery ability of well-trained athletes in intermittent and endurance sports. 2. Materials and Methods Literature studies were performed using online data bases including Scopus, PubMed (Medline) and Web of Science and published articles were retrieved (1929–2019). Major keywords regarding lactate threshold test (“LT”, “lactate threshold”, “MLSS”, “endurance”, “aerobic”, “anaerobic” and “recovery”) and physiological and biochemical reactions occurring during low-intensity exercise (“glycolysis”, “gluconeogenesis”, “glycogenesis”, “lactate metabolism”, “glucose metabolism”, “MCT”, “fat oxidation”, “oxaloacetate”, “pyruvate”, “AMPK”, “hepatic blood flow”, “skeletal muscle blood flow”, “skeletal muscle lactate output” and “hepatic lactate uptake”) were used in diverse combinations. Original full-text articles and reviews in English language published in scientific journals were included. Articles describing human and animal species were included. Conference articles, posters and studies with information overlapping with another publication were excluded. Based on a review of overlapping articles, the most recent or the most comprehensive articles were selected. After the initial searches identified articles, of which 167 were screened from the aforementioned databases. 30 articles were excluded because of unavailable full-text articles (11) and absence of specific data related to blood glucose and lactate without exercise (19). Of these, 115 articles were screened for eligibility, while 22 were excluded due to lack of useful data related to exercise physiology and Int. J. Environ. Res. Public Health 2020, 17, 5470 3 of 16 clinical features (Figure 1). One author (W.-H.Y.) reviewed the titles and abstracts of studies and the remaining 167 articles using the foregoing search strategy. Another author (H.P.) reviewed the article inclusion/exclusion criteria. Eligible articles were retrieved and independently assessed by two authors (W.-H.Y. and H.P.). The disagreement between authors over the eligibility of remaining articles was resolved through discussion with other collaborating authors (M.G. and O.H.). Furthermore, two authors (W.-H.Y. and H.P.) independently extracted data from articles based on study features and populations, type of intervention, measurement procedure and outcomes. Figure 1. Flow chart outlining the literature search strategy. 3. Utilization of Fat Oxidation During Low-Intensity Exercise The entire energy system, including phosphagens, glycolysis and oxidative phosphorylation, is simultaneously used during all levels of exercise intensity. In general, it seems important which energy system is predominantly used during different exercise intensities and exercise volumes. The energy storage of human fat is effectively unlimited during exercise [37]. Accordingly, one gram of fat provides about 40.79 kJ of energy. Very lean individuals of 70 kg and 10% body fat approximately have 285.56 kJ of endogenous fat energy [38]. With regard to low-intensity exercise, the oxidative metabolism from carbohydrate and fat is predominant. Adipocytes store large amounts of energy in the form of Int. J. Environ. Res. Public Health 2020, 17, 5470 4 of 16 triglycerides which amount to 200–625 Megajoule (MJ) in humans with normal body compositions of 10–30% body fat [25,36]. The energy expenditure derived from fat comes from various sources including plasma fatty acids from lipolysis in adipose tissue, fatty acids liberated from hydrolysis of circulating very low density lipoprotein (VLDL)-triacylglycerol and fatty acids from lipolysis of triacylglycerol located in lipid droplets in the skeletal muscle [39]. Plasma triglycerides are used as a crucial energy source in the muscle. However, when triglyceride in muscle cells are catalyzed by lipoprotein lipase, their contribution to energy demands during high-intensity exercise is limited [40]. During low-intensity exercise (25% VO2max), overall energy is obtained from plasma fatty acids with an additional small contribution from blood glucose. The rate of plasma fatty acid oxidation is similar to the rate of fatty acid oxidation (26 µmol·kg−1·min−1) in endurance-trained athletes. Furthermore, an increase in exercise intensity from 25% to 85% VO2max resulted in a progressive decline of fatty acid oxidation along with a proportional reduction of its concentration in blood [25]. This was due to insufficient transport of outflowing blood and albumin from adipose tissue into the systemic circulation [36,41]. 4. Lactate, Glucose, Enzymatic Responses and Cori Cycle During Exercise Lactate is produced during glycolysis, which is one of the metabolic pathways through which glucose can be utilized to provide energy. Lactate production from glycolysis occurs in muscle when exercise intensity increased [27]. Anaerobic conditions were not essential for the production of lactate in animal experiments (tail shaker muscle; western diamondback rattlesnake) [42] thus indicating that energy systems (phosphagen, glycolytic and oxidative) started to work simultaneously while the dissociation between lactate and hypoxic or anoxic conditions was orderly conformed [27]. Another study using the same model in ischemic and normoxic situations showed that increased rates of glycolysis could occur independently of O2 [43]. Such muscle conditions indicated the capability for exercise without fatigue [27] because of high blood flow rates that allowed the rapid turnover of H+ and lactate within the cell (and also other metabolites that may be involved in the fatigue process) [27,44]. These results indicated that, in addition to lactate production during anoxic or hypoxic situations, lactate was also produced as a metabolite due to adequate oxygenation [27]. Formerly, the understanding of lactate physiology was that lactate transport took place through simple diffusion (e.g., in the bloodstream) from cellular compartments to the blood. Increased lactate concentrations were believed to be a consequence of glycolytic flux rates [45–48]. In addition, previous studies had shown that three pathways were involved in lactate transport in red blood cells (RBC)—(i) H+ coupled transporter, (ii) band 3 protein Cl−/HCO3- mediated exchange with inorganic anions and (iii) passive diffusion of lactate across the lipid bilayer [49–51]. Nowadays, monocarboxylate transport (MCT) proteins (14 isoforms in total) are known to play critical roles in lactate transport. Cluster of differentiation 147 (CD147) functions as an ancillary protein that chaperones MCT1 and MCT4 to the cell membrane (muscle, red blood cell and liver). Human, rat and horse muscles express MCTl and MCT4. Both MCT1 and MCT4 need of an ancillary protein CD147 for their activity [52–54]. MCT1 and 4 are the predominant MCT transporters in human skeletal muscle while MCT2 is prominently expressed in the liver and brain [55,56]. MCT1 is coordinately expressed with isoforms of lactate dehydrogenase (LDH). High levels of MCT1 and LDH are found in oxidative muscle fibers [57]. In addition, MCT1 is the most important protein for lactate transport into or out of RBC [58,59]. In contrast, the low affinity transporter MCT4 was shown to be relevant for the net export of lactate from the cell which was predominantly expressed in glycolytic type IIA fibers [60]. MCTs transfer lactate into and out of cells and other organs such as liver, kidney, heart and brain [61,62]. These are now known as lactate shuttle mechanisms. The intracellular lactate shuttle mechanism is based on mitochondria-localized LDH (mLDH) for the re-synthesis between lactate and pyruvate [63]. During lactate production at rest and during submaximal exercise, pyruvate is converted to lactate by lactate dehydrogenase (LDH and mLDH) reaction. In addition, lactate can be reversibly converted to pyruvate by the intracellular lactate shuttle mechanisms [27,61,64]. Int. J. Environ. Res. Public Health 2020, 17, 5470 5 of 16 The liver is capable of eliminating lactate during exercise [65,66]. The Cori cycle, refers to the metabolic pathway of lactate-produced by anaerobic glycolysis in the muscle cells-moved to the liver and converted to glucose in order to ultimately return to the muscles [67]. Intensive exercise may impair the Cori cycle resulting in increased blood lactate concentrations which can be affected by decreased hepatosplanchnic blood flow [68]. Nielsen et al. [69] reported that arterial lactate was decreased because of reductions in lactate release from the working muscles during prolonged exercise (2 h and ~70% of VO2max, respectively). In contrast, liver clearance of lactate was maintained during a 2 h exercise phase. Lactate release by legs was significantly increased with increased work rate (~90% of VO2max during 20 min). However, the uptake of hepatic lactate constituted only one-tenth of the leg lactate production compared with 25% during prolonged exercise, while hepatic blood flow was markedly decreased, and leg blood flow increased. This reduction in hepatic extraction ratio may influence the rise in arterial lactate concentrations when exercise intensity is increased. On the other hand, leg lactate output and hepatic lactate uptake were similar (0.5 ± 0.3 and 0.55 ± 0.25 mmol·min−1, respectively) and the hepatic blood flow was accordingly increased during a recovery period (20 min) between exercises [69]. This study result showed that a two-third reduction in hepatic blood flow was among the most distinct changes during high-intensity exercise. With more intensive sympathetic activation and a cardiac output of more than 30 L·min−1, indocyanine green dye (ICG) eliminations may even approximate zero [70]. Therefore, a reciprocal relationship existed between liver and leg blood flow. During resting condition, hepatic blood flow was 19% of cardiac output which decreased to 2% during high-intensity exercise. This indicates that splanchnic organs contribute as a “blood donor” to the systemic circulation [69,71]. Glucose utilization and total glucose production are balanced by the concentration of glucose in arterial blood. As described above, the Cori cycle is responsible for lactate to glucose conversion in the liver [67]. However, if the hepatosplanchnic blood flow reaches a minimum, resulting in a reduction in hepatic venous O2 saturation to 6%, the contribution of the Cori cycle to glucose production appears to decrease during exercise [68]. During prolonged exercise, relative hypoglycemia may emerge although the rate of glucose appearance is significantly increased [72–75]. Therefore, muscle glucose uptake can be increased with time during prolonged exercise [65,69]. During high-intensity exercise, leg glucose uptake was increased while hepatic glucose output was significantly decreased (6.2 ± 1.3 and 1.9 ± 0.41 mmol·min−1, respectively). Furthermore, another study outcome showed that when exercise intensity was higher than 50% of VO2max the rate of gluconeogenesis was decreased because of the reduced hepatic blood flow [45]. In comparison to these levels, leg glucose uptake was markedly lower than hepatic glucose output during rest and recovery times (0.3 ± 0.1, 1.9 ± 0.5 and 1.55 ± 0.23, 2.34 ± 0.75 mmol·min−1, respectively) [69]. During recovery, despite hepatic blood flow being relatively increased, the Cori cycle (gluconeogenesis) may be insufficient to provide the needed glucose for maintaining blood glucose concentrations. The mechanism of attenuation of gluconeogenesis by sympathetic nervous system and upregulation of glycogenolysis still remains unclear [76,77]. The hepatic artery is sustained with α- and β-receptors [78,79]. A high level of epinephrine could cause an increase in hepatic glucose production, partly owing to an increased supply of gluconeogenic substrates such as alanine—and partly associated with a direct action on the liver cells [80]. In contrast, exercise with β-receptor blockade led to decreased hepatic uptake of gluconeogenic precursors, decreased lactate uptake and increased glucose output [76]. Furthermore, interleukins were released from active muscle during exercise and these are relevant for hepatic glucose production [77]. Decreased hepatosplanchnic blood flow may reduce the available number of hepatic sinusoids. Norepinephrine decreases the hepatic blood volume—even the plasma volume in hepatic sinusoids may be influenced [81]. Blood flow reductions of 30–40% during hemorrhage in the pig resulted in a reduction of hepatic norepinephrine uptake which induced a partial sinusoidal collapse [82]. In addition, Nielsen et al. [68] showed that a decreased intrinsic hepatic elimination of ICG during exercise caused a reduction of active sinusoidal area in human. Int. J. Environ. Res. Public Health 2020, 17, 5470 6 of 16 According to the aspects described above, lactate and glucose concentrations in well-trained endurance athletes gradually decreases during low-intensity exercise. As a large part of gluconeogenesis, accumulating lactate may be predominantly oxidized during rest or low-intensity exercise in the liver. However, this may be insufficient to produce appropriate glucose concentrations because of the low concentration of lactate and in addition the large energy source derived from fat between rest and low-intensity exercise. Further assumptions and available evidence are discussed in upcoming sections. 5. Allosteric Regulation between Glycolysis and Gluconeogenesis Allosteric regulation between glycolysis and gluconeogenesis can depend on the release of insulin, glucagon and cortisol. The role of glucagon and cortisol is to increase the concentration of blood glucose, while in opposition, insulin decreases blood glucose [83–85]. Hormonal regulation of metabolic reactions in the liver occurs by two major mechanisms. First, glucagon and β-adrenergic agonists interact with plasma membrane receptors which are associated with adenylate cyclase. The activity of these membrane-bound receptors increases intracellular cyclic adenosine monophosphate (cAMP) which drives the activation of cAMP-dependent protein kinase and catalyzes the phosphorylation of many protein substrates. Finally, these cascading events induce the activation of gluconeogenesis and inhibit glycolysis [86,87]. Second, those hormones act via alterations in intracellular calcium ion (Ca2+) concentration levels. Alpha-adrenergic agonists, vasopressin and angiotensin interact with their specific plasma membrane receptors to induce two intracellular messengers, myoinositol-1,4,5-trisphosphate and 1,2-diacylglycerol [88]. These increase intracellular Ca2+, which in combination with calmodulin or other effectors, stimulates a number of Ca2+-associated protein kinases including Ca2+/calmodulin-dependent protein kinase, phosphorylase kinase and protein kinase C. Furthermore, protein kinase catalyzes phosphorylation of many protein substrates which lead to alterations in gluconeogenic and glycolytic flux [86]. Phosphoenolpyruvate is partly recycled to pyruvate (for short-term hormonal regulation), during gluconeogenesis in perfused liver and isolated hepatocytes [89–92]. This flux is strongly inhibited by glucagon and cAMP. Liver-pyruvate kinase (PK), an allosteric enzyme, inhibits sigmoidal kinetics with regard to phosphoenolpyruvate (PEP). This enzyme is allosterically activated by fructose 1, 6-bisphosphate (Fru-1, 6-P2) and repressed by alanine and ATP. The in vitro studies of physiological concentrations of alanine, ATP and PEP, showed that these enzymes would be inhibited if they are not activated by Fru-1, 6-P2 [93,94]. From rest through moderate intensity exercise, ATP is primarily generated from fat oxidation [27]. The increased plasma/blood glucose concentration inhibits non-esterified fatty acid (NEFA) released by adipose tissue, by secreting insulin. In turn, elevated NEFA can decrease insulin and glucose concentrations and fatty acids are predominantly released and oxidized [95]. Moreover, Khani et al. [84] suggested that the infusion of cortisol increased NEFA. Therefore, it is important to recognize that cortisol increases lipolysis and NEFA concentrations. They also found moderate correlations between NEFA and gluconeogenesis in different observed groups (r = 0.599–0.665). These results do not indicate cause and influence. However, associations between NEFA and the rate of gluconeogenesis suggested existence of a relationship [96–98]. Mitochondrial acetyl-CoA acts as a key allosteric activator of pyruvate carboxylase (PC) which is activated from increased fatty acid. This allosteric activator leads to increased production of oxaloacetate for gluconeogenesis oxidation [99]. 6. Regulation of AMPK in Energy Metabolism AMP-activated protein kinase (AMPK) is a key regulator of physiological energy dynamics and functions by limiting anabolic, while facilitating catabolic pathways. AMPK is a heterotrimer and possesses an α (α1 and α2)-catalytic subunit and β (β1 and β2) and γ (γ1 and γ2 and γ3)-catalytic subunits. Three subunit combinations exist in human skeletal muscle. These are α1/β2/γ1, α2/β2/γ1 and α2/β2/γ3 [100]. γ3 is expressed predominantly in glycolytic skeletal muscle while there is very low Int. J. Environ. Res. Public Health 2020, 17, 5470 7 of 16 level of γ3 expression in oxidative muscles. The role of α1, α2 and γ3, during contraction of skeletal muscle or while exercising, in glucose metabolism has been broadly investigated [100,101]. The function of AMPK is to be a sensor in most tissues and organs including liver, skeletal muscle, heart, hypothalamus and adipose tissue. AMPK works by influencing enzymatic activities directly and is involved in biosynthesis of carbohydrates, lipids and proteins. AMPK regulates glucose and lipid metabolism and activates hepatic AMPK causing increased fatty acid oxidation [102]. AMPK is activated by a variety of exercise stresses. These typically alter cellular AMP: ATP ratio, either by increasing ATP or decreasing ATP production due to hypoxia, glucose deprivation or inhibition of mitochondrial oxidative phosphorylation [103]. There are conflicting results related to fatty acid oxidation induced by exercise. The concentration of AMPK induces fatty acid oxidation by utilizing an AMPK activator, 5-aminoimidazole-4-carboxamide-1-β-D-ribofuranoside (AICAR). In common with exercise, in skeletal muscles, AICAR affects the phosphorylation of acetyl-CoA carboxylase 2 (ACC2) which is an isoform of squamous cell carcinoma (SCC). AICAR increases the rate of uptake of long-chain fatty acids in cardiac myocytes via translocation of fatty acid translocase (FAT)/CD36 to the sarcolemma [104]. This mechanism reduces the level of malonyl-CoA and releases the inhibition of fatty acids uptake into mitochondria via carnitine palmitoyl transferase 1, resulting in fatty acid oxidation [101,105]. However, α2-AMPK activation is unnecessary for increasing fatty acid oxidation during exercise of low-intensity [101]. Miura et al. [101] suggested that in skeletal muscle, α2-AMPK may not have a major role in the shift to fatty acid oxidation from glucose oxidation while fasting, because the respiratory quotient ratio and utilization of oxygen during the fasting state remained constant between α1-AMPK -dominant-negative transgenic mice and wild-type littermates. Peripheral lipolysis can be maximally stimulated at the lowest exercise intensity in humans as well as at 25% of VO2max, whereas uptake of plasma glucose and oxidation of muscle glycogen increases with exercise intensity [106]. At 30% of VO2max, such as during prolonged low-intensity exercise, free fatty acid oxidation increases progressively while glucose oxidation is decreased [107]. Nevertheless, the increased activity of α2-AMPK is not necessary for increases in oxidation of fatty acid in skeletal muscle during endurance performance [101]. 7. Fat Oxidation Stimulates Gluconeogenesis and Can Decrease Glucose in Blood Low-intensity exercise causes fat oxidation that increases gluconeogenesis which occurs mostly in the liver and kidney [25,36,108–113]. In both, the liver and kidney, glycerol can be converted directly to glycerol 3-phosphate by glycerol kinase when glycerol is plentiful. Glycerol 3-phosphate is further converted to dihydroxyacetone phosphate by glycerol 3-phosphate dehydrogenase for gluconeogenesis. The direct conversion to glycerol 3-phosphate from free glycerol is believed to be trivial in skeletal muscle as well as in adipose tissue due to lower activity of glycerol kinase [114–117]. However, Guo et al. [118] suggested that the capacity for using blood glycerol for intracellular triacylglycerol (TG) synthesis in skeletal muscle is greater than was seen in previous studies [114–117]. Indeed, glucose was a constitutive substrate for muscle TG glycerol synthesis which may have provided TG derived glycerol with carbons [118]. The almost complete loss of 3H label in relation to 14C from blood glucose in muscle TG glycerol suggests (calculation of 3H label from glucose) that glucose passed through PC catabolized reactions and thus gluconeogenic precursors may also have paved their way to triglyceride glycerol [118,119]. Consequently, a pattern of preference to blood glycerol via blood glucose for TG glycerol synthesis in type 1 fiber-rich muscle was observed and its glycerol kinase activity was higher than in other types of muscle. Blood glycerol for intramuscular synthesis of TG glycerol is associated with the capacity of muscle to oxidize as well as store fatty acids [115,118]. Fatty acid availability is increased by mobilization of triglycerides in liver and adipose tissue. Increased fatty acid oxidation induces an increased rate of ketone-body formation and increased tissue concentrations of acetyl-Coenzyme A (CoA), fatty acyl-CoA and reduced NAD+ [108]. The oxidation of pyruvate is decreased due to inhibition of pyruvate oxidase by acetyl-CoA or competitive CoA Int. J. Environ. Res. Public Health 2020, 17, 5470 8 of 16 between pyruvate oxidase and the fatty acid oxidation system [120,121]. Accordingly, the complex mechanisms of pyruvate dehydrogenase include inhibition of end products by increases in concentration of mitochondrial acetyl-CoA, NADH and ATP. These can originate from fatty acid oxidation as well [122,123]. Pyruvate is converted to acetyl-CoA by pyruvate dehydrogenase (PDH). However, increased mitochondrial acetyl-CoA from fatty acid ß-oxidation activates pyruvate dehydrogenase kinases (PDHKs 1–4) and PC, which results in inhibition of PDH [108,124]. This enables the entrance of acetyl-CoA, derived from fatty acid oxidation in the first span of the tricarboxylic acid (TCA) cycle, to generate citrate. However, employment of the second span of the TCA cycle becomes reliant on NADH and FADH2 re-oxidation originating from fatty acid ß-oxidation [125]. Fatty acid oxidation can involve increased acetyl-CoA and inhibition of PDH. The TCA cycle must be substituted to allow continued function (anaplerosis) if its anions are removed. Due to inhibited PDH, PC is the major anaplerotic enzyme which immediately synthesizes oxaloacetate from pyruvate in the mitochondria [126]. During stimulated gluconeogenesis, the oxaloacetate concentration can fall due to increased PC activity. This can be explained by conversion of pyruvate into oxaloacetate that is still formed to malate. Thus, the metabolic production of the pyruvate carboxylase reaction may be related to the sum of malate and oxaloacetate [108,127,128]. In many tissues, the activity of PC is high (e.g., 10 to 12 units per gram in liver) and acetyl-CoA plays a role as a positive allosteric regulator of the enzyme, respectively [126]. This anaplerotic mechanism is mandatory during gluconeogenesis and lipogenesis when the reversible reaction from oxaloacetate to malate in the cytosol takes place with the aid of the malate–aspartate shuttle for gluconeogenesis or citrate for lipogenesis (oxaloacetate to cytosol, acetyl-CoA, malonyl-CoA) which exists in mitochondria and is still metabolized from glucose or fatty acids. Therefore, malate, but not oxaloacetate can traverse the inner membrane of mitochondria. Additionally, formation of malate is promoted by increased delivery of NADH from fatty acid oxidation [129]. Aspartate is converted to oxaloacetate to recruit cytosolic oxaloacetate by cytosolic aspartate aminotransferase. Hence, the effect of net redox-reaction of malate–aspartate shuttle is the oxidation of NADH to NAD+ in cytosol and reduction to NADH in the matrix. Accordingly, malate, aspartate and citrate are transferred precursors for oxaloacetate to gluconeogenesis [125,126]. It is equally relevant to remove TCA cycle intermediates and to avoid accumulated anions in the mitochondrial matrix. Cataplerotic reactions relate to disposal of TCA cycle intermediates. Phosphoenolpyruvate carboxykinase (PEPCK), which highly important in cataplerosis, generates PEP from oxaloacetate to be utilized for gluconeogenesis in the liver and kidney. Pyruvate is transported into the mitochondria where it is then converted into oxaloacetate or acetyl-CoA, respectively by PC and PDH. Mitochondrial oxaloacetate depends largely upon the distribution of PEPCK between cytosol and mitochondria [130]. The increase in phosphoenolpyruvate concentration is associated with decreased oxaloacetate concentration which may indicate activation of PEPCK [108,131]. The PEP from glycolysis otherwise can be converted to pyruvate that is decarboxylated to acetyl-CoA for ensuing oxidation to carbon dioxide (CO2) in the TCA cycle of muscle [126,132]. From muscle, glutamine can be transported to the kidney where ammonia is formed by utilization of the amino and amide groups. For generation of ammonia, glutamine goes through anaplerotic reactions to build α-ketoglutarate which joins the TCA cycle and is consequently metabolized to malate. Malate is further oxidized in the cytosol to oxaloacetate and to PEP and then to glucose [126]. The gluconeogenic pathway in liver and kidney is as follows: PEP, 2 phosphoglycerate, 3 phosphoglycerate, 1.3-bisphosphoglycerate, glyceraldehyde 3-phosphate (G3P) ↔ dihydroxyacetone phosphate (DHAP), fructose 1.6-bisphosphate, fructose 6-phosphate, glucose 6-phosphate and glucose [126]. Subsequently, reduced blood glucose concentration at the initial stages of low-intensity exercise (LT test) may occur because the glucose, through gluconeogenesis, seems to be transported to muscle cells via blood as an ongoing process and may be used as substrate for muscle TG glycerol synthesis. However, it is unclear whether this transferred blood glucose enters glycolysis or glycogenesis in Int. J. Environ. Res. Public Health 2020, 17, 5470 9 of 16 muscles. The enzyme activity in human is similar to animals like the rat. Earlier studies were conducted by investigations using isotope tracers and arterial-venous difference. Although outstanding studies of gluconeogenesis were developed, phenomena in humans were never investigated because of technical limitations such as gluconeogenesis from lactate [64]. The factors mentioned above are summarized in Figure 2. Figure 2. Summarized illustration of all described factors for decreased blood glucose and lactate values during the initial stages of lactate threshold test. ADP—adenosine diphosphate; ATP—adenosine triphosphate; ASAT—aspartate transaminase; DHAP—dihydroxyacetone phosphate; ETC;—electron transport chain; FAD—flavin adenine dinucleotide; F 1;6-B—fructose 1;6-bisphosphate; F 6-P—fructose 6-phosphate; G3P—glyceraldehyde 3-phasphate; G 6-P—glucose 6-phosphate; LDH—lactate dehydrogenase; PDH—pyruvate dehydrogenase; PEP—phosphoenolpyruvate; Pi—inorganic phosphate; MCT—monocarboxylate transporter; NAD—nicotinamide adenine dinucleotide; 1;3-BG—1;3-bisphosphoglycerate; 2-PG—2 Phosphoglycerate 3-PG—3 phosphoglycerate. 8. Conclusions It is already known that fat oxidation is predominantly utilized to perform low-intensity exercise. This exercise area is crucial for estimating the recovery ability of athletes. During the low-intensity exercise, the accrued resting lactate may predominantly be transported via blood from muscle cells to the liver/kidney (ongoing moment) while lactate from muscle cells is less oxidized by the intracellular lactate shuttle mechanism [45,64]. Furthermore, increased hepatic blood flow according to relatively more hepatic glucose output than glucose output of skeletal muscle and similar remained hepatic lactate uptake and lactate output of skeletal muscle during recovery time may support aspects of the predominant activation of gluconeogenesis (Cori cycle). However, it may be insufficient to induce the production of needed glucose because of the low concentration of lactate and the large energy source from fat between rest and low-intensity exercise. Insufficient sympathetic drive also may influence blood glucose and lactate concentrations [68,69,133,134]. Fatty acid oxidation activates key enzymes and hormonal responses of gluconeogenesis such as PK, PC, PEPCK, glucagon, cortisol and other associated regulators such as cAMP and intracellular Ca+, while glycolysis-related enzyme such as PDH are allosterically inhibited [93,94,99,108,126–128,130–132]. The efficient use of fat oxidation during low-intensity exercise and its effect during LT test exhibited a rightward shift of the exponential lactate curve. This can be interpreted as improved Int. J. Environ. Res. Public Health 2020, 17, 5470 10 of 16 regenerative ability, lactate threshold and endurance capacity [2,16–18]. Hence, decreased blood lactate and glucose may be a signal of efficient utilization of fat oxidation and improved recovery during low-intensity exercise. In particular, athletes of intermittent sports may need this recovery ability to improve performance in competition. The efficiency of fat oxidation during low-intensity exercise in athletes may be important to improve the regeneration of exercise performance between and during competition after highly intensive exercise load. In addition, strength athletes such as weight lifter may need this recovery ability to optimize the repeated high intensity training session because of the need for ATP re-synthesis. Athletes with a relatively poor endurance capability and the general population may show increased blood glucose concentrations during low-intensity exercise. It indicates that they also use significant amounts of glucose to perform low-intensity exercise. Athletes and the general population need to low-intensity exercises which activate the corresponding enzymes via fat oxidation resulting in enhanced endurance and recovery. Studies and findings of above-mentioned review were actively investigated and this review can be a first step toward identifying the associations between exercise intensity, blood glucose and lactate at the low-intensity exercise stages of LT test. Further studies are expected to investigate how these key enzymes and hormonal responses during low-intensity exercise are actually activated in humans with regard to gluconeogenesis. Author Contributions: Conceptualization, W.-H.Y., M.G. and O.H.; methodology, W.-H.Y., H.P., M.G. and O.H.; writing—original draft preparation, W.-H.Y.; writing—review and editing, W.-H.Y., H.P. and M.G.; visualization, W.-H.Y. and H.P.; project administration, W.-H.Y. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Acknowledgments: This study was supported by Dong-A University research grant. Conflicts of Interest: The authors declare no conflicts of interest. Abbreviations AICAR. Aminoimidazole-4-carboxamide-1-β-D-ribofuranoside ATP Adenosine triphosphate AMPK Adenosine monophosphate-activated protein kinase Ca2+ Calcium ions cAMP Cyclic adenosine monophosphate FADH2 Flavin adenine dinucleotide Fru-1,6-P2 Fructose 1,6-bisphosphate ICG Indocyanine green dye LT Lactate threshold mLDH Mitochondria-localized lactate dehydrogenase MCT Monocarboxylate transport MLSS Maximal lactate steady state NADH Nicotinamide adenine dinucleotide NEFA Non-esterified fatty acid PDH Pyruvate dehydrogenase PC Pyruvate carboxylase PEP Phosphoenolpyruvate PEPCK Phosphoenolpyruvate carboxykinase PK Pyruvate kinase SCC Squamous cell carcinoma TCA Tricarboxylic acid TG Triacylglycerol VO2 Oxygen uptake VO2max Maximal oxygen uptake Int. J. Environ. Res. Public Health 2020, 17, 5470 11 of 16 References 1. Messias, L.H.D.; Polisel, E.E.C.; Manchado-Gobatto, F.B. Advances of the reverse lactate threshold test: Non-invasive proposal based on heart rate and effect of previous cycling experience. PLoS ONE 2018, 13, e0194313. [CrossRef] 2. Faude, O.; Kindermann, W.; Meyer, T. Lactate threshold concepts. Sports Med. 2009, 39, 469–490. [CrossRef] 3. Wasserman, K.; McIlroy, M.B. Detecting the threshold of anaerobic metabolism in cardiac patients during exercise. Am. J. Cardiol. 1964, 14, 844–852. [CrossRef] 4. Heck, H.; Mader, A.; Hess, G.; Mucke, S.; Muller, R.; Hollmann, W. Justification of the 4-mmol/l lactate threshold. Int. J. Sports Med. 1985, 6, 117–130. [CrossRef] 5. Albesa-Albiol, L.; Serra-Payá, N.; Garnacho-Castaño, M.A.; Guirao Cano, L.; Pleguezuelos Cobo, E.; Maté-Muñoz, J.L.; Garnacho-Castaño, M.V. Ventilatory efficiency during constant-load test at lactate threshold intensity: Endurance versus resistance exercises. PLoS ONE 2019, 14, e0216824. [CrossRef] 6. Jotta, B.; Coutinho, A.B.B.; Pino, A.V.; Souza, M.N. Lactate threshold by muscle electrical impedance in professional rowers. Rev. Sci. Instrum. 2017, 88, 045105. [CrossRef] 7. Allen, W.K.; Seals, D.R.; Hurley, B.F.; Ehsani, A.A.; Hagberg, J.M. Lactate threshold and distance-running performance in young and older endurance athletes. J. Appl. Physiol. 1985, 58, 1281–1284. [CrossRef] 8. Coyle, E.F. Integration of the physiological factors determining endurance performance ability. Exerc. Sport Sci. Rev. 1995, 23, 25–63. [CrossRef] 9. Sjodin, B.; Svedenhag, J. Applied physiology of marathon running. Sports Med. 1985, 2, 83–99. [CrossRef] 10. Faria, E.W.; Parker, D.L.; Faria, I.E. The science of cycling. Sports Med. 2005, 35, 285–312. [CrossRef] 11. Atkinson, G.; Davison, R.; Jeukendrup, A.; Passfield, L. Science and cycling: Current knowledge and future directions for research. J. Sports Sci. 2003, 21, 767–787. [CrossRef] 12. Jones, A.M. The physiology of the world record holder for the women’s marathon. Int. J. Sports Sci. Coach. 2006, 1, 101–116. [CrossRef] 13. Wahl, P.; Manunzio, C.; Vogt, F.; Strütt, S.; Volmary, P.; Bloch, W.; Mester, J. Accuracy of a Modified Lactate Minimum Test and Reverse Lactate Threshold Test to Determine Maximal Lactate Steady State. J. Strength Cond. Res. 2017, 31, 3489–3496. [CrossRef] 14. Beneke, R.; Leithäuser, R.M.; Ochentel, O. Blood lactate diagnostics in exercise testing and training. Int. J. Sports Physiol. Perform. 2011, 6, 8–24. [CrossRef] 15. Mader, A.; Heck, H. A theory of the metabolic origin of “anaerobic threshold”. Int. J. Sports Med. 1986, 7, 45–65. [CrossRef] 16. Yoshida, T.; Udo, M.; Chida, M.; Ichioka, M.; Makiguchi, K.; Yamaguchi, T. Specificity of physiological adaptation to endurance training in distance runners and competitive walkers. Eur. J. Appl. Physiol. Occup. Physiol. 1990, 61, 197–201. [CrossRef] 17. Acevedo, E.O.; Goldfarb, A.H. Increased training intensity effects on plasma lactate, ventilatory threshold, and endurance. Med. Sci. Sports Exerc. 1989, 21, 563–568. [CrossRef] 18. Bosquet, L.; Léger, L.; Legros, P. Methods to determine aerobic endurance. Sports Med. 2002, 32, 675–700. [CrossRef] 19. Svedahl, K.; MacIntosh, B.R. Anaerobic threshold: The concept and methods of measurement. Can. J. Appl. Physiol. 2003, 28, 299–323. [CrossRef] 20. Yeh, M.P.; Gardner, R.M.; Adams, T.; Yanowitz, F.; Crapo, R. “Anaerobic threshold”: Problems of determination and validation. J. Appl. Physiol. 1983, 55, 1178–1186. [CrossRef] 21. Dotan, R. Reverse lactate threshold: A novel single-session approach to reliable high-resolution estimation of the anaerobic threshold. Int. J. Sports Physiol. Perform. 2012, 7, 141–151. [CrossRef] 22. Meyer, T.; Lucia, A.; Earnest, C. A conceptual framework for performance diagnosis and training prescription from submaximal gas exchange parameters-theory and application. Int. J. Sports Med. 2005, 26, 1–11. [CrossRef] 23. Skinner, J.S.; Mclellan, T.H. The transition from aerobic to anaerobic metabolism. Res. Q. Exerc. Sport 1980, 51, 234–248. [CrossRef] 24. Midgley, A.W.; McNaughton, L.R.; Jones, A.M. Training to enhance the physiological determinants of long-distance running performance. Sports Med. 2007, 37, 857–880. [CrossRef] Int. J. Environ. Res. Public Health 2020, 17, 5470 12 of 16 25. Coyle, E.F. Substrate utilization during exercise in active people. Am. J. Clin. Nutr. 1995, 61, 968–979. [CrossRef] 26. Billat, V.L.; Sirvent, P.; Py, G.; Koralsztein, J.-P.; Mercier, J. The concept of maximal lactate steady state. Sports Med. 2003, 33, 407–426. [CrossRef] 27. Philp, A.; Macdonald, A.L.; Watt, P.W. Lactate–a signal coordinating cell and systemic function. J. Exp. Biol. 2005, 208, 4561–4575. [CrossRef] 28. Simões, H.G.; Grubert Campbell, C.S.; Kokubun, E.; Denadai, B.S.; Baldissera, V. Blood glucose responses in humans mirror lactate responses for individual anaerobic threshold and for lactate minimum in track tests. Eur. J. Appl. Physiol. Occup. Physiol. 1999, 80, 34–40. [CrossRef] 29. Simões, H.G.; Campbell, C.S.; Kushnick, M.R.; Nakamura, A.; Katsanos, C.S.; Baldissera, V.; Moffatt, R.J. Blood glucose threshold and the metabolic responses to incremental exercise tests with and without prior lactic acidosis induction. Eur. J. Appl. Physiol. 2003, 89, 603–611. [CrossRef] 30. Simões, H.G.; Hiyane, W.C.; Benford, R.E.; Madrid, B.; Prada, F.A.; Moreira, S.R.; de Oliveira, R.J.; Nakamura, F.Y.; Campbell, C.S. Lactate threshold prediction by blood glucose and rating of perceived exertion in people with type 2 diabetes. Percept. Mot. Skills 2010, 111, 365–378. [CrossRef] 31. Restan, A.Z.; Zacche, E.; da Silva, S.B.; Cerqueira, J.A.; Carfiofi, A.C.; Queiroz-Neto, A.; Camacho, A.A.; Ferraz, G.C. Lactate and glucose thresholds and heart rate deflection points for Beagles during intense exercise. Am. J. Vet. Res. 2019, 80, 284–293. [CrossRef] 32. Ferraz, G.; D Angelis, F.; Teixeira-Neto, A.R.; Freitas, E.; Lacerda-Neto, J.; Queiroz-Neto, A. Blood lactate threshold reflects glucose responses in horses submitted to incremental exercise test. Arg. Bras. Med. Vet. Zootec. 2008, 60, 256–259. [CrossRef] 33. Junior, P.B.; de Andrade, V.L.; Campos, E.Z.; Kalva-Filho, C.A.; Zagatto, A.M.; de Araujo, G.G.; Papoti, M. Effect of Endurance Training on The Lactate and Glucose Minimum Intensities. J. Sports Sci. Med. 2018, 17, 117–123. 34. Rodríguez, F.A.; Mader, A. Energy systems in swimming. In World Book of Swimming. From Science to Performance; Nova: New York, NY, USA, 2011; pp. 225–240. 35. Horowitz, J.F.; Klein, S. Lipid metabolism during endurance exercise. Am. J. Clin. Nutr. 2000, 72, 558–563. [CrossRef] 36. Bülow, J.; Madsen, J. Influence of blood flow on fatty acid mobilization from lipolytically active adipose tissue. Pflugers Arch. 1981, 390, 169–174. [CrossRef] 37. Gonzalez, J.T.; Fuchs, C.J.; Betts, J.A.; Van Loon, L.J. Liver glycogen metabolism during and after prolonged endurance-type exercise. Am. J. Physiol. Endocrinol. Metab. 2016, 311, 543–553. [CrossRef] 38. Jeukendrup, A.; Wallis, G.A. Measurement of substrate oxidation during exercise by means of gas exchange measurements. Int. J. Sports Med. 2005, 26, 28–37. [CrossRef] 39. Kiens, B.; Alsted, T.J.; Jeppesen, J. Factors regulating fat oxidation in human skeletal muscle. Obes. Rev. 2011, 12, 852–858. [CrossRef] 40. Oscai, L.; Essig, D.; Palmer, W. Lipase regulation of muscle triglyceride hydrolysis. J. Appl. Physiol. 1990, 69, 1571–1577. [CrossRef] 41. Hodgetts, V.; Coppack, S.W.; Frayn, K.N.; Hockaday, T. Factors controlling fat mobilization from human subcutaneous adipose tissue during exercise. J. Appl. Physiol. 1991, 71, 445–451. [CrossRef] 42. Conley, K.E.; Lindstedt, S.L. Energy-saving mechanisms in muscle: The minimization strategy. J. Exp. Biol. 2002, 205, 2175–2181. [PubMed] 43. Kemper, W.F.; Lindstedt, S.L.; Hartzler, L.K.; Hicks, J.W.; Conley, K.E. Shaking up glycolysis: Sustained, high lactate flux during aerobic rattling. Proc. Natl. Acad. Sci. USA 2001, 98, 723–728. [CrossRef] [PubMed] 44. Moon, B.R.; Hopp, J.J.; Conley, K.E. Mechanical trade-offs explain how performance increases without increasing cost in rattlesnake tailshaker muscle. J. Exp. Biol. 2002, 205, 667–675. [PubMed] 45. MacRae, H.; Dennis, S.C.; Bosch, A.N.; Noakes, T.D. Effects of training on lactate production and removal during progressive exercise in humans. J. Appl. Physiol. 1992, 72, 1649–1656. [CrossRef] [PubMed] 46. Juel, C.; Honig, A.; Pilegaard, H. Muscle lactate transport studied in sarcolemmal giant vesicles: Dependence on fibre type and age. Acta Physiol. Scand. 1991, 143, 361–366. [CrossRef] 47. Watt, P.W.; MacLennan, P.A.; Hundal, H.S.; Kuret, C.M.; Rennie, M.J. l (+)-Lactate transport perfused rat skeletal muscle: Kinetic characteristics and sensitivity to pH and transport inhibitors. Biochim. Biophys. Acta Biomembr. 1988, 944, 213–222. [CrossRef] Int. J. Environ. Res. Public Health 2020, 17, 5470 13 of 16 48. Gladden, L.B.; Crawford, R.E.; Webster, M.J.; Watt, P.W. Rapid tracer lactate influx into canine skeletal muscle. J. Appl. Physiol. 1995, 78, 205–211. [CrossRef] 49. Deuticke, B. Monocarboxylate transport in erythrocytes. J. Membr. Biol. 1982, 70, 89–103. [CrossRef] 50. Kim, C.; Goldstein, J.; Brown, M. cDNA cloning of MEV, a mutant protein that facilitates cellular uptake of mevalonate, and identification of the point mutation responsible for its gain of function. J. Biol. Chem. 1992, 267, 23113–23121. 51. Garcia, C.K.; Goldstein, J.L.; Pathak, R.K.; Anderson, R.G.; Brown, M.S. Molecular characterization of a membrane transporter for lactate, pyruvate, and other monocarboxylates: Implications for the Cori cycle. Cell 1994, 76, 865–873. [CrossRef] 52. Koho, N.M.; Hyyppä, S.; Pösö, A.R. Monocarboxylate transporters (MCT) as lactate carriers in equine muscle and red blood cells. Equine Vet. J. Suppl. 2006, 38, 354–358. [CrossRef] [PubMed] 53. Júnior, W.H.F.; Garcia de Carvalho, J.R.; Mendes de Almeida, M.L.; Macedo Lemos, E.G.; Brioschi Soares, O.A.; Ribeiro, G.; de Queiroz-Neto, A.; de Camargo Ferraz, G. Differential Expression of Monocarboxylate Transporter 1 and Ancillary Protein CD147 in Red Blood Cells of Show Jumping Horses. J. Equine Vet. Sci. 2019, 81, 102791. [CrossRef] [PubMed] 54. Ng, M.; Louie, J.; Cao, J.; Felmlee, M.A. Developmental Expression of Monocarboxylate Transporter 1 and 4 in Rat Liver. J. Pharm. Pharm. Sci. 2019, 22, 376–387. [CrossRef] [PubMed] 55. McClelland, G.B.; Khanna, S.; González, G.F.; Butz, C.E.; Brooks, G.A. Peroxisomal membrane monocarboxylate transporters: Evidence for a redox shuttle system? Biochem. Biophys. Res. Commun. 2003, 304, 130–135. [CrossRef] 56. Magistretti, P.J.; Allaman, I. Lactate in the brain: From metabolic end-product to signalling molecule. Nat. Rev. Neurosci. 2018, 19, 235–249. [CrossRef] 57. McCullagh, K.J.; Juel, C.; O’Brien, M.; Bonen, A. Chronic muscle stimulation increases lactate transport in rat skeletal muscle. Mol. Cell Biochem. 1996, 156, 51–57. [CrossRef] 58. Skelton, M.S.; Kremer, D.E.; Smith, E.W.; Gladden, L.B. Lactate influx into red blood cells of athletic and nonathletic species. Am. J. Physiol. 1995, 268, 1121–1128. [CrossRef] 59. Opitz, D.; Lenzen, E.; Opiolka, A.; Redmann, M.; Hellmich, M.; Bloch, W.; Brixius, K.; Brinkmann, C. Endurance training alters basal erythrocyte MCT-1 contents and affects the lactate distribution between plasma and red blood cells in T2DM men following maximal exercise. Can. J. Physiol. Pharmacol. 2015, 93, 413–419. [CrossRef] 60. Wilson, M.C.; Jackson, V.N.; Heddle, C.; Price, N.T.; Pilegaard, H.; Juel, C.; Bonen, A.; Montgomery, I.; Hutter, O.F.; Halestrap, A.P. Lactic acid efflux from white skeletal muscle is catalyzed by the monocarboxylate transporter isoform MCT3. J. Biol. Chem. 1998, 273, 15920–15926. [CrossRef] 61. Brooks, G.A. The lactate shuttle during exercise and recovery. Med. Sci. Sports Exerc. 1986, 18, 360–368. [CrossRef] 62. van Hall, G.; Stromstad, M.; Rasmussen, P.; Jans, O.; Zaar, M.; Gam, C.; Quistorff, B.; Secher, N.H.; Nielsen, H.B. Blood lactate is an important energy source for the human brain. J. Cereb. Blood Flow Metab. 2009, 29, 1121–1129. [CrossRef] [PubMed] 63. Brooks, G.A.; Dubouchaud, H.; Brown, M.; Sicurello, J.P.; Eric Butz, C. Role of mitochondrial lactate dehydrogenase and lactate oxidation in the intracellular lactate shuttle. Proc. Natl. Acad. Sci. USA 1999, 96, 1129–1134. [CrossRef] [PubMed] 64. Brooks, G.A. The Science and Translation of Lactate Shuttle Theory. Cell Metab. 2018, 27, 757–785. [CrossRef] [PubMed] 65. Ahlborg, G.; Wahren, J.; Felig, P. Splanchnic and peripheral glucose and lactate metabolism during and after prolonged arm exercise. J. Clin. Investig. 1986, 77, 690–699. [CrossRef] [PubMed] 66. Wahren, J.; Felig, P.; Ahlborg, G.; Jorfeldt, L. Glucose metabolism during leg exercise in man. J. Clin. Investig. 1971, 50, 2715–2725. [CrossRef] [PubMed] 67. Cori, C.F.; Cori, G.T. Glycogen formation in the liver from d-and l-lactic acid. J. Biol. Chem. 1929, 81, 389–403. 68. Nielsen, H.B.; Clemmesen, J.O.; Skak, C.; Ott, P.; Secher, N.H. Attenuated hepatosplanchnic uptake of lactate during intense exercise in humans. J. Appl. Physiol. 2002, 92, 1677–1683. [CrossRef] 69. Nielsen, H.B.; Febbraio, M.A.; Ott, P.; Krustrup, P.; Secher, N.H. Hepatic lactate uptake versus leg lactate output during exercise in humans. J. Appl. Physiol. 2007, 103, 1227–1233. [CrossRef] Int. J. Environ. Res. Public Health 2020, 17, 5470 14 of 16 70. Nielsen, H.B.; Boushel, R.; Madsen, P.; Secher, N.H. Cerebral desaturation during exercise reversed by O2 supplementation. Am. J. Physiol. 1999, 277, 1045–1052. [CrossRef] [PubMed] 71. Katz, L.; Rodbard, S. The integration of the vasomotor responses in the liver with those in other systemic vessels. J. Pharmacol. Exp. Ther. 1939, 67, 407–422. 72. Coyle, E.F.; Hagberg, J.M.; Hurley, B.F.; Martin, W.H.; Ehsani, A.A.; Holloszy, J.O. Carbohydrate feeding during prolonged strenuous exercise can delay fatigue. J. Appl. Physiol. Respir. Environ. Exerc. Physiol. 1983, 55, 230–235. [CrossRef] [PubMed] 73. Bergman, B.C.; Butterfield, G.E.; Wolfel, E.E.; Lopaschuk, G.D.; Casazza, G.A.; Horning, M.A.; Brooks, G.A. Muscle net glucose uptake and glucose kinetics after endurance training in men. Am. J. Physiol. 1999, 277, E81–E92. [CrossRef] [PubMed] 74. Bergman, B.C.; Horning, M.A.; Casazza, G.A.; Wolfel, E.E.; Butterfield, G.E.; Brooks, G.A. Endurance training increases gluconeogenesis during rest and exercise in men. Am. J. Physiol. Endocrinol. Metab. 2000, 278, E244–E251. [CrossRef] [PubMed] 75. Kjaer, M.; Engfred, K.; Fernandes, A.; Secher, N.H.; Galbo, H. Regulation of hepatic glucose production during exercise in humans: Role of sympathoadrenergic activity. Am. J. Physiol. 1993, 265, E275–E283. [CrossRef] 76. Ahlborg, G.; Juhlin-Dannfelt, A. Effect of beta-receptor blockade on splanchnic and muscle metabolism during prolonged exercise in men. J. Appl. Physiol. 1994, 76, 1037–1042. [CrossRef] 77. Gleeson, M. Interleukins and exercise. J. Physiol. 2000, 529, 1. [CrossRef] 78. Greenway, C.V.; Lawson, A.E. Beta-adrenergic receptors in the hepatic arterial bed of the anesthetized cat. Can. J. Physiol. Pharmacol. 1969, 47, 415–419. [CrossRef] 79. Greenway, C.V.; Lawson, A.E.; Mellander, S. The effects of stimulation of the hepatic nerves, infusions of noradrenaline and occlusion of the carotid arteries on liver blood flow in the anaesthetized cat. J. Physiol. 1967, 192, 21–41. [CrossRef] 80. Stevenson, R.W.; Steiner, K.E.; Connolly, C.C.; Fuchs, H.; Alberti, K.G.; Williams, P.E.; Cherrington, A.D. Dose-related effects of epinephrine on glucose production in conscious dogs. Am. J. Physiol. 1991, 260, E363–E370. [CrossRef] 81. Greenway, C.V.; Bass, L. Derecruitment in cat liver: Extension of undistributed parallel tube model to effects of low hepatic blood flow on ethanol uptake. Can. J. Physiol. Pharmacol. 1989, 67, 1225–1231. [CrossRef] 82. Rasmussen, A.; Skak, C.; Kristensen, M.; Ott, P.; Kirkegaard, P.; Secher, N.H. Preserved arterial flow secures hepatic oxygenation during haemorrhage in the pig. J. Physiol. 1999, 516, 539–548. [CrossRef] [PubMed] 83. Litwack, G. Chapter 8-Glycolysis and Gluconeogenesis. In Human Biochemistry; Litwack, G., Ed.; Academic Press: Boston, MA, USA, 2018; pp. 183–198. [CrossRef] 84. Khani, S.; Tayek, J.A. Cortisol increases gluconeogenesis in humans: Its role in the metabolic syndrome. Clin. Sci. 2001, 101, 739–747. [CrossRef] [PubMed] 85. Levitt, N.S.; Lambert, E.V.; Woods, D.; Hales, C.N.; Andrew, R.; Seckl, J.R. Impaired glucose tolerance and elevated blood pressure in low birth weight, nonobese, young South African adults: Early programming of cortisol axis. J. Clin. Endocrinol. Metab. 2000, 85, 4611–4618. [PubMed] 86. Pilkis, S.J.; El-Maghrabi, M.R.; Claus, T.H. Hormonal regulation of hepatic gluconeogenesis and glycolysis. Annu. Rev. Biochem. 1988, 57, 755–783. [CrossRef] 87. Pilkis, S.; Claus, T. Hepatic gluconeogenesis/glycolysis: Regulation and structure/function relationships of substrate cycle enzymes. Annu. Rev. Nutr. 1991, 11, 465–515. [CrossRef] 88. Exton, J.H. Mechanisms involved in alpha-adrenergic phenomena. Am. J. Physiol. Endocrinol. Metab. 1985, 248, 633–647. [CrossRef] 89. Freidmann, B.; Goodman, E.H., Jr.; Saunders, H.L.; Kostos, V.; Weinhouse, S. An estimation of pyruvate recycling during gluconeogenesis in the perfused rat liver. Arch. Biochem. Biophys. 1971, 143, 566–578. [CrossRef] 90. Rognstad, R. Cyclic AMP induced inhibition of pyruvate kinase flux in the intact liver cell. Biochem. Biophys. Res. Commun. 1975, 63, 900–905. [CrossRef] 91. Rognstad, R.; Katz, J. Effects of hormones and of ethanol on the fructose 6-P-fructose 1,6-P2 futile cycle during gluconeogenesis in the liver. Arch. Biochem. Biophys. 1976, 177, 337–345. [CrossRef] 92. Rognstad, R.; Katz, J. Role of pyruvate kinase in the regulation of gluconeogenesis from L-lactate. J. Biol. Chem. 1977, 252, 1831–1833. Int. J. Environ. Res. Public Health 2020, 17, 5470 15 of 16 93. Flory, W.; Peczon, B.D.; Koeppe, R.E.; Spivey, H.O. Kinetic properties of rat liver pyruvate kinase at cellular concentrations of enzyme, substrates and modifiers. Biochem. J. 1974, 141, 127–131. [CrossRef] [PubMed] 94. van Berkel, T.J.; de Jonge, H.R.; Koster, J.F.; Hulsmann, W.C. Kinetic evidence for the presence of two forms of M2-type pyruvate kinase in rat small intestine. Biochem. Biophys. Res. Commun. 1974, 60, 398–405. [CrossRef] 95. Randle, P.; Garland, P.; Hales, C.; Newsholme, E. The glucose fatty-acid cycle its role in insulin sensitivity and the metabolic disturbances of diabetes mellitus. Lancet 1963, 281, 785–789. [CrossRef] 96. Tayek, J.A.; Katz, J. Glucose production, recycling, Cori cycle, and gluconeogenesis in humans: Relationship to serum cortisol. Am. J. Physiol. Endocrinol. Metab. 1997, 272, E476–E484. [CrossRef] 97. Katz, J.; Tayek, J.A. Gluconeogenesis and the Cori cycle in 12-, 20-, and 40-h-fasted humans. Am. J. Physiol. Endocrinol. Metab. 1998, 275, 537–542. [CrossRef] 98. Tayek, J.A.; Katz, J. Glucose production, recycling, and gluconeogenesis in normals and diabetics: A mass isotopomer [U-13C] glucose study. Am. J. Physiol. Endocrinol. Metab. 1996, 270, 709–717. [CrossRef] 99. Oh, K.-J.; Han, H.-S.; Kim, M.-J.; Koo, S.-H. CREB and FoxO1: Two transcription factors for the regulation of hepatic gluconeogenesis. BMB Rep. 2013, 46, 567. [CrossRef] 100. Birk, J.B.; Wojtaszewski, J.F. Predominant α2/β2/γ3 AMPK activation during exercise in human skeletal muscle. J. Physiol. 2006, 577, 1021–1032. [CrossRef] 101. Miura, S.; Kai, Y.; Kamei, Y.; Bruce, C.R.; Kubota, N.; Febbraio, M.A.; Kadowaki, T.; Ezaki, O. α2-AMPK activity is not essential for an increase in fatty acid oxidation during low-intensity exercise. Am. J. Physiol. Endocrinol. Metab. 2009, 296, E47–E55. [CrossRef] 102. Viollet, B.; Guigas, B.; Leclerc, J.; Hébrard, S.; Lantier, L.; Mounier, R.; Andreelli, F.; Foretz, M. AMP-activated protein kinase in the regulation of hepatic energy metabolism: From physiology to therapeutic perspectives. Acta Physiol. 2009, 196, 81–98. [CrossRef] 103. Towler, M.C.; Hardie, D.G. AMP-activated protein kinase in metabolic control and insulin signaling. Circ. Res. 2007, 100, 328–341. [CrossRef] 104. Luiken, J.J.; Coort, S.L.; Willems, J.; Coumans, W.A.; Bonen, A.; van der Vusse, G.J.; Glatz, J.F. Contraction-induced fatty acid translocase/CD36 translocation in rat cardiac myocytes is mediated through AMP-activated protein kinase signaling. Diabetes 2003, 52, 1627–1634. [CrossRef] [PubMed] 105. Merrill, G.F.; Kurth, E.J.; Hardie, D.G.; Winder, W.W. AICA riboside increases AMP-activated protein kinase, fatty acid oxidation, and glucose uptake in rat muscle. Am. J. Physiol. Endocrinol. Metab. 1997, 273, E1107–E1112. [CrossRef] [PubMed] 106. Romijn, J.A.; Coyle, E.F.; Sidossis, L.S.; Gastaldelli, A.; Horowitz, J.F.; Endert, E.; Wolfe, R.R. Regulation of endogenous fat and carbohydrate metabolism in relation to exercise intensity and duration. Am. J. Physiol. 1993, 265, 380–391. [CrossRef] [PubMed] 107. Ahlborg, G.; Felig, P.; Hagenfeldt, L.; Hendler, R.; Wahren, J. Substrate turnover during prolonged exercise in man. Splanchnic and leg metabolism of glucose, free fatty acids, and amino acids. J. Clin. Investig. 1974, 53, 1080–1090. [CrossRef] 108. Williamson, J.R. Mechanism for the stimulation in vivo of hepatic gluconeogenesis by glucagon. Biochem. J. 1966, 101, 11C. [CrossRef] 109. Clore, J.N.; Glickman, P.S.; Helm, S.T.; Nestler, J.E.; Blackard, W.G. Evidence for dual control mechanism regulating hepatic glucose output in nondiabetic men. Diabetes Care 1991, 40, 1033–1040. [CrossRef] 110. Puhakainen, I.; Yki-Järvinen, H. Inhibition of lipolysis decreases lipid oxidation and gluconeogenesis from lactate but not fasting hyperglycemia or total hepatic glucose production in NIDDM. Diabetes 1993, 42, 1694–1699. [CrossRef] 111. Chen, X.; Iqbal, N.; Boden, G. The effects of free fatty acids on gluconeogenesis and glycogenolysis in normal subjects. J. Clin. Investig. 1999, 103, 365–372. [CrossRef] 112. Alsahli, M.; Gerich, J.E.; practice, C. Renal glucose metabolism in normal physiological conditions and in diabetes. Diabetes Res. Clin. Pract. 2017, 133, 1–9. [CrossRef] 113. van Loon, L.J.; Greenhaff, P.L.; Constantin-Teodosiu, D.; Saris, W.H.; Wagenmakers, A.J. The effects of increasing exercise intensity on muscle fuel utilisation in humans. J. Physiol. 2001, 536, 295–304. [CrossRef] 114. Seltzer, W.K.; Angelini, C.; Dhariwal, G.; Ringel, S.P.; McCabe, E.R. Muscle glycerol kinase in Duchenne dystrophy and glycerol kinase deficiency. Muscle Nerve 1989, 12, 307–313. [CrossRef] 115. Newsholme, E.; Taylor, K. Glycerol kinase activities in muscles from vertebrates and invertebrates. Biochem. J. 1969, 112, 465–474. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2020, 17, 5470 16 of 16 116. Robinson, J.; Newsholme, E. Glycerol kinase activities in rat heart and adipose tissue. Biochem. J. 1967, 104, 2C. [CrossRef] [PubMed] 117. Ryall, R.L.; Goldrick, R. Glycerokinase in human adipose tissue. Lipids 1977, 12, 272–277. [CrossRef] [PubMed] 118. Guo, Z.; Jensen, M.D. Blood glycerol is an important precursor for intramuscular triacylglycerol synthesis. J. Biol. Chem. 1999, 274, 23702–23706. [CrossRef] 119. Guo, Z.; Lee, W.P.; Katz, J.; Bergner, A.E. Quantitation of positional isomers of deuterium-labeled glucose by gas chromatography/mass spectrometry. Anal. Biochem. 1992, 204, 273–282. [CrossRef] 120. Walter, P.; Paetkau, V.; Lardy, H.A. Paths of carbon in gluconeogenesis and lipogenesis III. The role and regulation of mitochondrial processes involved in supplying precursors of phosphoenolpyruvate. J. Biol. Chem. 1966, 241, 2523–2532. 121. Garland, P.; Randle, P.J. Control of pyruvate dehydrogenase in the perfused rat heart by the intracellular concentration of acetyl-coenzyme A. Biochem. J. 1964, 91, 6C. 122. Sugden, M.C.; Holness, M.J. Mechanisms underlying regulation of the expression and activities of the mammalian pyruvate dehydrogenase kinases. Arch. Physiol. Biochem. 2006, 112, 139–149. [CrossRef] 123. Sugden, M.C.; Holness, M.J. Recent advances in mechanisms regulating glucose oxidation at the level of the pyruvate dehydrogenase complex by PDKs. Am. J. Physiol. Endocrinol. Metab. 2003, 284, E855–E862. [CrossRef] 124. Utter, M.F.; Keech, D.B. Pyruvate carboxylase. J. Biol. Chem. 1963, 238, 2603. 125. Sugden, M.C.; Holness, M.J. The pyruvate carboxylase-pyruvate dehydrogenase axis in islet pyruvate metabolism: Going round in circles? Islets 2011, 3, 302–319. [CrossRef] [PubMed] 126. Owen, O.E.; Kalhan, S.C.; Hanson, R.W. The key role of anaplerosis and cataplerosis for citric acid cycle function. J. Biol. Chem. 2002, 277, 30409–30412. [CrossRef] [PubMed] 127. Klingenberg, M.; Bucher, T. Biological oxidations. Annu. Rev. Biochem. 1960, 29, 669–708. [CrossRef] [PubMed] 128. Henning, H.; Stumpf, B.; Ohly, B.; Seubert, W. On the mechanism of gluconeogenesis and its regulation. 3. The glucogenic capacity and the activities of pyruvate carboxylase and PEP-carboxylase of rat kidney and rat liver after cortisol treatment and starvation. Biochem. Z. 1966, 344, 274. 129. Shrago, E.; Lardy, H.A. Paths of carbon in gluconeogenesis and lipogenesis II. Conversion of precursors to phosphoenolpyruvate in liver cytosol. J. Biol. Chem. 1966, 241, 663–668. 130. Hanson, R.W.; Mehlman, M.A.; Lardy, H.A. Gluconeogenesis, Its Regulation in Mammalian Species; John Wiley and Sons: New York, NY, USA, 1976; p. 592. 131. Weber, G.; Singhal, R.; Stamm, N.; Srivastava, S. Hormonal induction and suppression of liver enzyme biosynthesis. In Proceedings of the Federation Proceedings, Bethesda, MD, USA; 1946; p. 745. 132. Jungas, R.L.; Halperin, M.L.; Brosnan, J.T. Quantitative analysis of amino acid oxidation and related gluconeogenesis in humans. Physiol. Rev. 1992, 72, 419–448. [CrossRef] 133. Pellerin, L.; Pellegri, G.; Bittar, P.G.; Charnay, Y.; Bouras, C.; Martin, J.L.; Stella, N.; Magistretti, P.J. Evidence supporting the existence of an activity-dependent astrocyte-neuron lactate shuttle. Dev. Neurosci. 1998, 20, 291–299. [CrossRef] 134. Noakes, T.D.; St Clair Gibson, A.; Lambert, E.V. From catastrophe to complexity: A novel model of integrative central neural regulation of effort and fatigue during exercise in humans: Summary and conclusions. Br. J. Sports Med. 2005, 39, 120–124. [CrossRef] © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Decreased Blood Glucose and Lactate: Is a Useful Indicator of Recovery Ability in Athletes?

07-29-2020

Yang, Woo-Hwi,Park, Hyuntae,Grau, Marijke,Heine, Oliver

eng

PMC9781885

Citation: Skoki, A.; Rossi, A.; Cintia, P.; Pappalardo, L.; Štajduhar, I. Extended Energy-Expenditure Model in Soccer: Evaluating Player Performance in the Context of the Game. Sensors 2022, 22, 9842. https://doi.org/10.3390/s22249842 Academic Editor: Michael E. Hahn Received: 18 October 2022 Accepted: 12 December 2022 Published: 14 December 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). sensors Article Extended Energy-Expenditure Model in Soccer: Evaluating Player Performance in the Context of the Game Arian Skoki 1 , Alessio Rossi 2,3,* , Paolo Cintia 2 , Luca Pappalardo 2,3 and Ivan Štajduhar 1,4 1 Department of Computer Engineering, Faculty of Engineering, University of Rijeka, Vukovarska 58, 51000 Rijeka, Croatia 2 Department of Computer Science, University of Pisa, Largo B. Pontecorvo 3, 56127 Pisa, Italy 3 Institute of Information Science and Technologies (ISTI), National Research Council of Italy (CNR), Giuseppe Moruzzi 1, 56124 Pisa, Italy 4 Center for Artificial Intelligence and Cybersecurity, University of Rijeka, R. Matejcic 2, 51000 Rijeka, Croatia * Correspondence: alessio.rossi@di.unipi.it Abstract: Every soccer game influences each player’s performance differently. Many studies have tried to explain the influence of different parameters on the game; however, none went deeper into the core and examined it minute-by-minute. The goal of this study is to use data derived from GPS wearable devices to present a new framework for performance analysis. A player’s energy expenditure is analyzed using data analytics and K-means clustering of low-, middle-, and high- intensity periods distributed in 1 min segments. Our framework exhibits a higher explanatory power compared to usual game metrics (e.g., high-speed running and sprinting), explaining 45.91% of the coefficient of variation vs. 21.32% for high-, 30.66% vs. 16.82% for middle-, and 24.41% vs. 19.12% for low-intensity periods. The proposed methods enable deeper game analysis, which can help strength and conditioning coaches and managers in gaining better insights into the players’ responses to various game situations. Keywords: game intensity; clustering; machine learning; fatigue; fitness tracking 1. Introduction The intensity of a soccer game is dependent on a wide range of factors: quality of the opposition, period of the season, weather, team form, game status, etc. There is a lot of work dealing with the influence of various parameters on said factors—such as location, opponent quality, and game outcome—by using the most important metrics derived from wearable sensors (e.g., total distance, accelerations, and decelerations). However, one of the most dominant factors in soccer is the quality of the opposition. Higher-quality opponents usually require higher physical demands during the game, which, in turn, results in increased values of the total distance (TD), maximal speed, average speed, frequency of high-intensity actions (HIAs) [1], and events related to changes in velocity (accelerations and decelerations) [2]. This is not always true, because it heavily depends on the context and the play styles of each team. According to Garcıa-Unanue et al. [3], away games accumulate significantly more TD (+230.65 m, 95% IC: 21.94 to 438.19, ES: 0.46, p = 0.031), but there were no differences found that depended on the opponent quality. However, an analysis between the first and second half revealed a significant reduction in TD covered by the players against lower-level teams (−290.42 m, 95% CI: −557.82 to −23.01, ES: 0.72, p = 0.033) and medium-level teams (−374.56 m, 95% CI: −549.21 to −199.70, ES: 0.71, p < 0.001). Congested periods can also have a great impact on player performance and reduce the number of HIAs that a player can sustain [4]. Differences in intensities are present across playing positions [5]. Therefore, central midfielders (CM) and wide midfielders (WM) cover greater TD and have higher average speed, whereas WMs and fullbacks (FB) cover higher distances at high-speed running (HSR) and sprinting [6]. Center Sensors 2022, 22, 9842. https://doi.org/10.3390/s22249842 https://www.mdpi.com/journal/sensors Sensors 2022, 22, 9842 2 of 16 backs (CB) and forwards (FW) usually cover the smallest distances [7], but FWs cover more HSR distance than CMs [6]. Concerning the game outcome, recent work reported higher workloads in won games compared to losses and draws, higher TD and HSR distance in the second half of defeats, and lower average speed in wins compared to draws and losses [8]. When looking at score change influence on the game demands, there have been reports that CMs, WMs, and FWs have increased TD while winning, but CBs and WBs have higher TD and HSR while losing [9,10]. Interpretation of scoreline influence needs to be done carefully, because there is a need for a clear scoreline definition for which to consider scoreline effects [11]. Several studies [3,6,7,11,12] have described various context situations and their effects on the wearable-derived metrics. Nevertheless, there were no attempts to increase the level of data sampling and examine physical demand change in a game on a minute-by- minute basis. Another problem is the usage of different GPS providers, which have various thresholds for determining sprint distance, HSR distance, number of sprints, number of detected acceleration and deceleration phases, etc.; hence, the concept cannot be easily transferred. Nowadays, most providers allow users to set their own preferred thresholds. However, this invokes other problems, such as many different thresholds used for the same parameters (no consensus). In addition, accumulated values (which are the most-used ones) provide an overall image, but often that is not particularly useful as it represents an average of the game. A team can play an average-intensity game overall but actually perform a very intensive first half and a less-intensive second half of the game [8]. This information is very important, along with individual analysis of a particular player’s effort within the game. Intensity inspection within the context of scored and conceded goals can provide information about which player could be a better substitute depending on the game scoreline. Not every game is the same—each one provides unique demands [13]. Therefore, the training load in the preceding week should be adjusted according to expenditure in the previous game, and player fitness needs to be tracked regularly to evaluate season meso- and macro- cycles [14]. For this purpose, better methods for game analysis are needed than the ones that are currently used. Hence, the aim of this study is to minimize the effect of various GPS provider thresh- olds by taking an energy expenditure approach. The study shows how to develop a framework of data analytics for evaluating workload intensity as the game goes on. In particular, this enables a detailed examination of the workload throughout the game and provides a baseline for better understanding of the game demands depending on the context. Moreover, methods proposed in this study can be used for objective tracking of the players through meso- and macro-cycles of the season. 2. Materials and Methods 2.1. Study Design The data were acquired during official and preseason games of a professional soccer club. The process of data collection was executed by using GPS wearable sensors, GPexe pro2 (Exelio Srl, Udine, Italy), with a sampling rate of 18 Hz. According to position, the players were divided into five categories: center back (CB), wing back (WB), midfielder (MF), wide forward (WF), and forward (FW). The sensor provider enabled the usage of two types of data: (1) GM-5MIN (GPS metrics of expenditure through 5 min intervals) and (2) metabolic power events (MPEs), (HIAs, which occur throughout the whole course of a game). 2.2. Subjects In total, there were 38 male soccer players (age 25 ± 3 years; height 1.81 ± 0.06 m; weight 76 ± 5 kg) that played at least one game during the acquisition period. The goal- keeper (GK) position was not recorded and therefore it was not used for analysis. All the players that were wearing the sensor during the games were included in the dataset. This applies to the substitute players, too. Playing positions counted 11 CBs, 7 WBs, 8 MFs, 3 Sensors 2022, 22, 9842 3 of 16 WFs, and 9 FWs. The club allowed the research team to access players’ data, and informed consent was provided. 2.3. Data Acquisition The data collection process started in January 2021 and ended in March 2022, which rounded a period of two half-seasons (including preseasons) and included 80 games. The game data were acquired using GPexe pro2 devices that collected all of the standard parameters for analysis [15] with the addition of metabolic expenditure features [16]. Furthermore, GPS metrics derived from the whole game (90 min data) were used as a baseline for explaining and validating the descriptive power of the proposed clustering methods. In the rest of the paper, we refer to this data type as GM-GAME. 2.4. Procedures and Variables Before each game, players would put on the GPexe pro2 device, which was located in a wearable vest. The sensor notified the wearer with a red blinking light when there was a need for re-calibration. This was done very easily by spinning the device for a couple of seconds. After the match, collected data were downloaded from the device and uploaded to the manufacturer’s platform using the GPexe bridge application (version 8.3.6). The online web application (version 7.4.46) computed all the metrics in a 5 min sampling window. In order to minimize problems due to different thresholds and varying results, only energy-based metrics and TD were used in the analysis. This included: total time played (min), distance (m), average metabolic power (W/kg), energy (J/kg), anaerobic energy (J/kg), MPE count, MPE average recovery time (s), MPE average recovery power (W/kg), walk distance (m), running distance (m), walk energy (J/kg), and running energy (J/kg). Average metabolic power was calculated by multiplying speed and energy cost (the description of parameter calculation was taken from the GPexe dictionary for athletic performance monitoring, which is an internal document accessible only to their users), which has been thoroughly described by di Prampero et al. [16]. The energy variable is an estimation of both the energy required to cover a given distance at a constant speed and the energy needed to perform speed variations. The same is true for calculating anaerobic energy, with a difference of taking into account the player’s maximal VO2 as a measurement threshold [17]. The MPEs are defined as phases during the exercise (or a game) based on a difference between the estimated metabolic power and oxygen consumption (the description of parameter calculation was taken from the GPexe dictionary for athletic performance monitoring, which is an internal document accessible only to their users). Since the maximal VO2 of each athlete can be directly or indirectly assessed, this value allows individual analysis and overcomes limitations of other models that are based on specific speed or acceleration thresholds. The MPE recovery (power and time) is detected by the power decrease that happens in order to repay previously contracted oxygen debt (the description of parameter calculation was taken from the GPexe dictionary for athletic performance monitoring, which is an internal document accessible only to their users). The features regarding walking and running are not defined by a fixed speed threshold but depend rather on different combinations of speed and acceleration [16]. All the presented features are shown in Table 1. Many of them were related to metabolic power for the reason of avoiding difficulties involving speed, acceleration, and deceleration threshold values [17]. The accounted features were aggregated and derived from GM-5MIN and MPE data, which are described thoroughly in the following sections. 2.4.1. MPE Data A special feature of GPexe wearables is the focus on metabolic expenditure using MPE. The main difference and benefit of the approach using MPE is that it does not take into account only acceleration and deceleration for detecting HIA [16], but instead it is focused on energy expenditure; thus, the problem of setting the threshold is avoided. An issue Sensors 2022, 22, 9842 4 of 16 regarding the acceleration approach lies in setting up a threshold for detecting these events, which heavily influences the resulting values. Moreover, there exists a lack of agreement and information regarding the choice of methods for acceleration filtering [18]. Instead, the alternative approach proposed by di Prampero et al. in 2005 is used [19], which sets the standard for calculating MPE. This approach assumes that accelerated running on flat terrain is equivalent to constant running uphill at a constant speed at a certain angle. Power events happen often in the game, and they are the most important factor in energy expenditure. In the full game, there are more than 100 MPEs per player, which differ in duration and power. An example of the data for one player and a single game can be seen in Figure 1. The MPE dataset contained information about the start and end timestamp of an event, duration in seconds, maximal speed, and average power spent. The energy expenditure of an MPE was calculated by multiplying the average power and duration of the event, which is shown in Equation (1). EMPE = Pavg ∗ t (1) Figure 1. MPE in-game distribution for a single player. This could be done because most of the events are short in duration, up to 20 s, with a median value of 5.8, a mean value of 6.5, and a standard deviation of 3.35 s. The distribution of event duration can be seen in Figure 2. The described data provide information about the peak energy expenditure. However, this is not enough for a complete understanding of expenditure within a game because there are no data about the period in which a player was recovering (see Figure 3). Data regarding what is happening in the recovery period are lacking; therefore, if one would use only MPE data, then only the information about the HIA would be considered. The reality is that recovery can be passive or active: a player can stand, walk, jog or run. This information is crucial for understanding in-game player recovery. Sensors 2022, 22, 9842 5 of 16 Figure 2. MPE duration histogram. The distribution median value is 5.8, the mean is 6.5, and the standard deviation is 3.35. Figure 3. Recovery time within the power events (blue) in a 60 s time frame. 2.4.2. GM-5MIN Data MPE data occurrence is discrete across the whole game. That means that these events (short in duration) are always separated by periods of recovery. The energy that is spent during MPEs is equivalent to 30% of the total energy consumption (the description of parameter calculation was taken from the GPexe dictionary for athletic performance moni- toring, which is an internal document accessible only to their users), which can be seen in Figure 4. The remaining 70% of energy expenditure is ignored by this type of data. The lack of information about a player’s expenditure during the recovery periods is addressed with the introduction of GM-5MIN data, which contains an average of values in a given 5 min period. All the events within the 5 min interval were hence taken into consideration, including walking, jogging, running, sprinting, TD, energy, etc. This enabled the acqui- sition of the full 100% of energy expenditure within the observed period. An example of the GM-5MIN and MPE energy expenditure for one player and a game can be seen in Figure 4. Ideally, one would like to have these values in the 1 min interval and thus have a Sensors 2022, 22, 9842 6 of 16 detailed image of expenditure. Unfortunately, due to the intrinsic limits of the GPS tracking device, the processing time for shorter intervals exponentially increases with regard to the duration of the interval and could not be extracted in a reasonable time. This is the reason for settling on a 5 min period: it is short enough to capture game details but also long enough for calculating accumulation metrics such as average metabolic power, average recovery power, etc. Figure 4. GM-5MIN vs. MPE expenditure across 90 min. Blue bar charts represent GM-5MIN across the 90 min of the game. Orange bars represent the energy expenditure of MPEs. 2.5. Data Preprocessing To fully describe the intensity of a game, the MPE and GM-5MIN parameters needed to be combined together. This was done by iterating through each game for all the players. As the exact times and durations of both MPEs and GM-5MIN were known, they could be combined. The process consisted of merging three main data sources, which included: (1) GM-5MIN features in the preceding 5 min period (GM-5MIN-PRIOR), (2) MPE features in the preceding 3 and 5 min periods (MPE-PRIOR), and (3) MPE features in the observed minute (MPE-CURRENT). The entire processing workflow is shown in Figure 5. The resulting dataset counted 25 parameters, which are described in Table 1. The process to create GM-5MIN-PRIOR consisted of using the GM-5MIN data of a 5 min period that preceded the current processing minute. The aim of this was to measure the overall expenditure just before the observed minute. The second step was adding MPE- PRIOR features, which gave information about the peak intensity in the period preceding the observed minute. Finally, information about the observed minute was added by the introduction of MPE-CURRENT features. Sensors 2022, 22, 9842 7 of 16 Figure 5. Preprocessing and clustering flowchart is divided into 3 parts. The first step shows how GM-5MIN-PRIOR, MPE-PRIOR, and MPE-CURRENT are combined. Next, dimensionality reduction using PCA is performed to prepare the data for K-means clustering. The final step consists of using the clustering algorithm to obtain low-, middle-, and high-intensity events throughout the game and to perform MFit analysis on the players. Sensors 2022, 22, 9842 8 of 16 Table 1. A list and description of features used for clustering. Features are divided into 3 categories depending on the data source: GM-5MIN-PRIOR, MPE-PRIOR, and MPE-CURRENT. Feature Name Description GM-5MIN-PRIOR Distance (m) Distance covered in the last 5 min. MPE count Number of MPEs in the last 5 min Anaerobic energy (J/kg) Anaerobic energy spent in the last 5 min Average metabolic power (W/kg) Average metabolic power spent in the last 5 min Average MPE time (s) Average MPE duration in the last 5 min Average MPE recovery time (s) Average recovery time in the last 5 min Average MPE recovery power (W/kg) Average recovery power in the last 5 min Walk energy (J/kg) Energy spent walking in the last 5 min Running energy (J/kg) Energy spent running in the last 5 min General energy (J/kg) Energy spent on all activities in the last 5 min Total number of MPEs Number of MPEs up to that moment in the game Total energy spent (J/kg) Energy spent up to that moment in the game MPE-PRIOR MPE energy (3 min) Energy spent on MPEs in the last 3 min MPE energy (5 min) Energy spent on MPEs in the last 5 min MPE count (3 min) Number of MPEs in the last 3 min MPE count (5 min) Number of MPEs in the last 5 min Recovery time (s) (3 min) Recovery time (s) in the last 3 min Recovery time (s) (5 min) Recovery time (s) in the last 5 min Average recovery time (s) (3 min) Average recovery time (s) in the last 3 min Average recovery time (s) (5 min) Average recovery time (s) in the last 5 min Total recovery time (s) Recovery time up to that moment in the game MPE-CURRENT MPE energy spent (J/kg) Energy spent on MPEs in the observed minute Event count Number of MPEs in the observed minute Average recovery time (s) Average recovery time in the observed minute Recovery time (s) Recovery time in the observed minute On top of described variables in Section 2.4, additional ones were derived from these data and incorporated into the feature set. In order to keep information about the duration of the game and the influence of fatigue, cumulative (total) features were created. This group of features comprised: the total number of MPEs, total energy spent (J/kg), and total recovery time (s). Information about MPE average recovery time was only available for a 5 min period preceding the observed minute. To get information about the average recovery time in the 3 min preceding and the observed minute, a new metric was derived from MPE data. The metric represents absolute recovery time in seconds (within the observed period) divided by the number of events in that period. An example is shown in Figure 3, where there are 4 MPEs and 5 recovery periods, which gives 18 s of work and 42 s of recovery. An average recovery time is thus equal to Equation (2), which gives an average of 8.4 s for the observed example. tavg_recovery = trecovery NMPE + 1 (2) The MPE features were used to explain the work done in the current processing minute and also the last 3 and 5 min periods. Combined with GM-5MIN, the dataset had information about (1) overall energy consumption and recovery time in the preceding 5 min, (2) MPE energy consumption and recovery time in the preceding period of 3 and 5 min, and (3) MPE energy consumption and recovery time in the current minute. After all the players and games were processed, extreme values for every feature were limited to fit the ceiling value in order to denoise the data. This value was determined by calculating the threshold at which 99.5% of the data would fit in that range. All the values that were higher than the ceiling threshold were limited to that threshold. Sensors 2022, 22, 9842 9 of 16 2.6. Clustering Analysis To account for the scarcity of data instances (limited number of games and players) and to suppress overfitting, the number of features was reduced to successfully apply the clustering algorithm. We used principal component analysis (PCA) for dimensionality reduction. Before that, each feature was normalized by applying a min–max normalization. The next step was to divide the resulting data points (after min–max normalization) into different categories. As this was a new approach and the labels were unknown beforehand, an unsupervised clustering algorithm had to be used. For this purpose, K-means was selected to evaluate how many intensity zones were present in the dataset. Before fitting centroids on the data, halftime and the start of the game needed to be excluded from training. By analyzing the data, a threshold of 200 m was set as the minimum distance; thus, all the instances that were lower than that value were ignored. Inspection showed that these outlier distance values were part of the halftime recovery and the first 5 min of the game (lacking information about the energy expenditure in the 5 min that preceded). The chosen threshold of 200 m enabled the exclusion of all the outlier values. Failure to do so would have heavily affected the results for the low-intensity zone, which is described later in Section 3.2. The K-means algorithm was tested for k ∈ [2, 3, 4, . . . , 15] clusters using the within-cluster sum of squares (WCSS). The best number of clusters was defined through the elbow method on WCSS values. Clustering analysis is a prerequisite for better understanding the physical demands of the game and for future research about the influence of various game context variables on players’ physical behavior. 2.7. Clustering Application After the dataset had been created (see Figure 5), we clustered intensity profiles (groups) that enabled assessment of the effort performed by the players as the game goes on. Moreover, a Markov chain analysis was conducted in order to estimate players’ capabilities of intensity shifts during a game. This part of the paper may be more interesting for specific readers; therefore, a detailed description of the methods used and the newly created MFit index can be found in the Supplementary Materials. Section S1 explains the process of creating an MFit index, while Section S2 shows the results of such analysis by tracking the players’ fitness through meso- and macro- cycles of the season. 3. Results In this section, we explain how the proposed clustering method can be used for detailed minute-by-minute game analysis that can be performed both individually or on a team basis. It also provides an example of game load comparison in Section 3.3, with the premise that each game has unique physical demands. 3.1. Clustering Analysis Results PCA analysis showed that the optimal number of components for dimensionality reduction is seven, preserving 92.8% of data variance (see Figure 6). Table 2 presents the most important features of every component, with the minimal importance of a single feature being 0.3 (explaining 30% of the data). Based on these seven principal components, the elbow of the score was obtained by analyzing a different number of classes by the K- means algorithm. We determined that the optimal number of clusters was three (Figure 6), corresponding to low, middle, and high intensities in the game. The most important features and their distributions across clusters are shown in Table 3. Sensors 2022, 22, 9842 10 of 16 Figure 6. Analysis of K-means clustering with WCSS method and PCA component inspection. The optimal number of clusters is chosen using an elbow method. The resulting cluster number is 3, with the number of PCA components being 7, which corresponds to 92.8% of the original data variance. Table 2. The most important features (above 0.3) for every PCA component. Feature PC1 PC2 PC3 PC4 PC5 PC6 PC7 Distance (m) ✓ MPE count ✓ Anaerobic energy (J/kg) ✓ Average metabolic power (W/kg) ✓ ✓ Average MPE time (s) ✓ Average MPE recovery time (s) ✓ Walk energy (J/kg) ✓ ✓ Running energy (J/kg) ✓ General energy (J/kg) ✓ Total event count ✓ Total energy spent (J/kg) ✓ ✓ MPE energy (3 min) ✓ ✓ ✓ MPE energy (5 min) ✓ Average recovery time (s) (3 min) ✓ Average recovery time (s) (5 min) ✓ ✓ Total recovery time (s) ✓ MPE energy spent (J/kg) ✓ Event count ✓ ✓ Average recovery time (s) ✓ ✓ The K-means algorithm assigns each particular minute in the game to one of the three possible groups. It is expected that higher intensity causes more energy consumption and an increased number of MPEs but less time spent in recovery. To distinguish between low, middle, and high intensity, the groups needed to be examined in more detail. For this purpose, each feature distribution across classes was inspected by calculating the mean Sensors 2022, 22, 9842 11 of 16 and standard deviation. As described in Section 2.5, clustering features were divided into three groups: (1) MPE-CURRENT, (2) MPE-PRIOR, and (3) GM-5MIN-PRIOR. The first one gives information about the current intensity, while the latter two explain what actually caused the current state. It can be seen that the low-intensity group does not contain any MPEs, but at the same time, it has the highest energy consumption in the last 5 min (GM-5MIN features—Energy (J)). The low-intensity group is a result of higher energy consumption in the preceding period and can be characterized as an active recovery period for the player. Table 3. Clustering group results. Darker background color represents higher intensity. The low- intensity group shows no MPE activity in the observed minute, but this is likely a result of very high expenditure in the 5 min period that preceded it. The middle-intensity group exhibits higher expenditure according to MPE features but shows lower values of GM-5MIN in 5 min prior. The high-intensity group produces the highest expenditure in all types of features. Parameter Name Low Group Middle Group High Group MPE features (1 min) µ σ µ σ µ σ Energy (J) 0 0 180 150 230 280 Event count 0 0 1.8 0.9 2 1 Average recovery time (s) 60 0 20 7 18 8 MPE features (3 min before) µ σ µ σ µ σ Energy (J/kg) 3 min 400 350 400 330 700 350 MPE count (3 min) 3.8 2.2 4.0 2.2 6 2 Recovery time (s) (3 min) 155 16 154 15 140 16 GM-5MIN features (5 min before) µ σ µ σ µ σ Energy (J/kg) 2900 1000 1300 1000 2800 500 MPE count 6 3.5 4 3.5 9 2.5 Anaerobic energy (J/kg) 750 400 500 400 1000 150 Avg. MPE recovery time (s) 60 80 50 60 23 8 Running energy (J/kg) 1400 800 1000 800 2250 450 3.2. In-Depth Game Visualization Every game is unique to each player. Therefore, a visual representation of intensity zones throughout the game for a single player in a particular game is shown in Figure 7. It can be clearly seen that a player needs to make a stop after a certain number of high- intensity actions. The period following (middle and low intensity) could be a recovery period; however, the reader should note that it is dependent on the game context (e.g., penalty, set-piece, video-assisted referee decision). Scoring minutes are shown in order to better visualize the potential effect of scored and conceded goals on the game tempo. The distribution of the intensity zones across the span of the game is shown in Figure 8. It can be clearly seen that the high-intensity periods decrease as the game goes toward the end. At the same time, middle intensity periods start high and then slowly increase from the 10th min. Low intensity periods stay stable and slightly increase towards the end. This is mostly due to the effect of fatigue. The same inspection can be made for the whole team, but this requires additional analysis. Average minute-by-minute intensity can be aggregated by grouping playing positions and showing their average minute-by-minute intensity values for each position or by taking an average intensity for each minute using all the players in the team. It should be noted that this kind of analysis is inferior to individual inspection, and a lot of information can be lost in the aggregation. Sensors 2022, 22, 9842 12 of 16 Figure 7. Intensity clusters across a 90-min period with the score-change timestamps for a single player in a particular game. High intensity is represented with brown–red, middle with yellow– orange, and low intensity with blue. Vertical discontinued red lines are goals conceded, and green ones are goals scored. Figure 8. High-, middle-, and low-intensity cluster distribution across a period of a game in 5 min intervals. 3.3. Evaluation through Game Load To show the effect and benefits of the proposed clustering approach, the results are compared with the GM-GAME data (each GPS feature separately) that is regularly used for athlete load monitoring. The premise that each game is unique was tested by looking at the distribution of high-, middle-, and low-intensity minutes for the whole team. The clustering algorithm provides information about the three groups (i.e., high, middle, and low), and the same thing needs to be done for GM-GAME data to enable comparison. Currently, by using GM-GAME data, there exists no consensus on how one could measure intensity using particular parameters. In order to compare the approach presented in this study with GM-GAME data, we need to draw parallels between the intensity clusters and a single GM-GAME feature. Therefore, the assumption was that GM-GAME data equivalents for Sensors 2022, 22, 9842 13 of 16 the high-intensity group would be the distance covered above 20 km/h; for low intensity, it would be equal to walking distance (m); and for middle intensity, it would be equal to running distance (<20 km/h, walking excluded). Table 4 provides the percentage of intensity groups in both the clustering algorithm and GM-GAME features. To compare the variability of each group, the coefficient of variation (CV) was calculated for each column. The results (row CV) show the variance in % for the particular column. The aim of this was to assess between-game variability and the potential for describing each game’s unique demands. Clearly, the clustering approach has superior CV results with the additional capability of detailed in-game inspection. Further, the distribution of values is more natural–mostly middle- and high-intensity clusters and a smaller numer of low-intensity clusters, which is a reasonably competitive game-demand distribution. On the other hand, GM-GAME puts focus on middle- and low-, with very little influence due to high-intensity. According to this, soccer players are not giving their maximal effort very often in competitive games, which is hard to believe. This confirms that tracking the intensity of the game is not a trivial problem, and that it is very hard to explain game intensity by using a single GPS parameter. Hence, based on these results and using only GM-GAME, we can speculate that there is no distinction between scoreline change, quality of the opponent, or some other context of the game. This shows the limitation of GM-GAME data and a need for better methods for intensity evaluation. A comparison of the MFit index approach (mentioned in Section 2.7) and GM-GAME parameters can be seen in Figure S5 of the Supplementary Materials. Table 4. Comparison of clustering algorithm and GM-GAME data for describing game load (63 games) based on CV. Bolded values mark higher explanatory power for a particular cluster. Game Id Clustering with K-Means GM-GAME Data High Middle Low High Middle Low 1 0.3572 0.4042 0.2538 0.0743 0.5845 0.4497 2 0.2456 0.5178 0.2507 0.0881 0.5116 0.4089 3 0.3300 0.5092 0.1737 0.0511 0.5717 0.4257 ... ... ... ... ... ... ... 79 0.4338 0.3946 0.1863 0.0679 0.6124 0.4352 80 0.2611 0.3997 0.3487 0.0695 0.5011 0.3963 µ ± σ 0.31 ± 0.14 0.51 ± 0.16 0.20 ± 0.05 0.07 ± 0.01 0.57 ± 0.1 0.42 ± 0.08 CV 45.91% 30.66% 24.41% 21.32% 16.82% 19.12% 4. Discussion This study provides a deep dive into game intensity in soccer by using data ac- quired from wearable sensors. Every game is unique and, therefore, should be treated independently. In the literature, a lot of work has analyzed the intensity concerning the scoreline [9,11,12,20]. This was done by comparing relevant metrics (e.g., HSR, distance, and sprint distance) depending on the score and also the quality of the opponent. However, all of the research was only focused on comparing the GM-GAME data from wearable sensors, which give overall information about the game. There have been no attempts to understand how players perform on a minute-to-minute basis and how fatigue influences the final minutes of the game. To go deep into the game intensity, the proposed analytical approach takes into consideration two data sources (GM-5MIN and MPE data) as the game goes by. Surely, the possibility of including 1 min periods instead of 5 min periods would yield a better representation of the game. On the other hand, by using smaller periods of the game, accumulation error would go up, and this amount would vary between the providers [21]. Moreover, processing smaller intervals would be very long or even impossible with state- of-the-art tracking devices. To account for this, the framework of data analytics provided in this study enabled a detailed inspection of the game. In particular, the intensity of the Sensors 2022, 22, 9842 14 of 16 game could be grouped into three main zones, i.e., low-, middle-, and high-intensity. As shown in Table 3, in low-intensity actions, players do not engage in any activity (recovery period), but this was a result of the preceding 5 min intense period. Differently, the middle- and high-intensity groups show medium and strong MPE activity, respectively, and shorter periods of recovery. However, the middle-intensity group shows low-intensity actions 5 min prior, while the high-intensity group is preceded by strong GM-5MIN expenditure. These groups enable us to analyze player status as the game goes on. As a matter of fact, Figures 7 and 8 show intensity groups of a single player for the entire game. The figures suggest that there are several high-intensity periods in the starting part of the game, which recedes as the game goes by, indicating the fatigue status of the player or a change in the game intensity. Moreover, based on Figure 7, the game score seems to have no correlation with the intensity period. However, more research is needed to assess the effect of different intensity patterns on game scoring status, the tactical approach of a team, the roles of the players, and other game events. The proposed framework provides a basis for additional analysis of the game con- text. This includes the changes in intensity cluster distribution depending on: the team’s possession style, the quality of the opponent, or the change of scoreline. By using more seasons and acquiring more data about the substitute players, a better analysis of their energy expenditure can be made. Currently, there are not enough substitute players with similar playing time and the same playing position. Further, with the addition of multiple teams with different playing styles, the model can be adapted according to e.g., possession or counter-attacking football. This, however, needs further investigation and is out of the scope of this study. The applicability of the model was shown using in-game visualization of the player’s low-, middle-, and high- intensity clusters. The ability to provide better information about the game intensity compared to GM-GAME parameters was shown by calculating the CV of each cluster. The results proved that tracking the game intensity is not a trivial task, i.e., it cannot be observed from the aggregation level. However, the approach proposed in this study gives promising results because it takes into account minute-by-minute changes in energy expenditure. However, the reader should note that all of these analyses can be affected by the game characteristics, which can influence intensity and, consequently, physical demands. The main limitation of this study is that only one team was used in the analysis. Therefore, we should inspect this on more teams to provide information about the accuracy and validity of this approach. However, in this study, we present a new approach for analyzing real-time intensity during a game that could be applied to each team. Hence, every team could create its own personalized model in accordance with individual players’ characteristics by applying our analytical approach. 5. Conclusions This study is focused on diving deeper into players’ energy expenditure throughout the full length of a soccer game. It provides procedures for processing and clustering the data for the purpose of inspecting the physical game performance of the players on a minute-by-minute basis. This approach can help practitioners to better understand the specific context of the game (i.e., when the team decreases physical performance) but also provide ground for further inspection of fatigue, match context, and how the energy of the players is spent through the course of a game. Sensors 2022, 22, 9842 15 of 16 Supplementary Materials: The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/s22249842/s1, Section S1. MFit Methodology; Section S2. MFit Analysis: Tracking Players’ Fitness Status. Figure S1: Maximum high-intensity repetitions per game for a particular player; Figure S2: Transition probability matrix used in MFit analysis for the example shown in Figure 7 of the main paper. The player has a very high possibility (83.33%) of remaining in the high-intensity zone given that he is already performing in the high-intensity zone; Figure S3: MFit-5 through the season, made with a 5-min high-intensity repetition probability. An example is provided for each role, based on a single player. The red dotted line refers to the probability repetition average of all the players in the dataset; Figure S4: MFit-10 through the season, made with a 10-min high-intensity repetition probability. An example is provided for each role, based on a single player. The red dotted line refers to the probability repetition average of all the players in the dataset; Figure S5: Analytical comparison between widely used GM-GAME and a MFit-10 for a player in CB position. GM-GAME features show very low variability and, therefore, an inability to express player physical effort differences on a game-to-game basis. Reference [22] is cited in the Supplementary Materials. Author Contributions: A.S. carried out data collection, analysis, and interpretation. A.R. and P.C. made substantial contributions to the concept and design. A.S. and A.R. wrote the first manuscript draft, and all the authors were involved in revising it critically. L.P. and I.Š. did supervision of the work done as well as in-depth revision and editing of the manuscript. All authors have read and agreed to the published version of the manuscript. Funding: This work was supported by the Horizon 2020 project EuroCC 951732 National Competence Centres in the Framework of EuroHPC and by the University of Rijeka, Croatia, grant number uniri-tehnic-18-15. Moreover, this work was also supported by the European Community’s H2020 Program under the funding scheme INFRAIA-2019-1: Research Infrastructures grant agreement 871042, www.sobigdata.eu, SoBigData. The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: Not applicable. Acknowledgments: We would like to thank the Transnational Access program of SoBigData++ for their support, which permitted meeting between the researchers authoring this study. Conflicts of Interest: The authors declare that they have no conflicts of interest relevant to the content of this article. Abbreviations The following abbreviations are used in this manuscript: GPS Global Positioning System TD Total Distance HSR High-Speed Running MPE Metabolic Power Event HIA High Intensity Action CB Center Back FB Full Back WB Wing Back MF Midfielder WM Wide Midfielder WF Wide Forward FW Forward PCA Principal Component Analysis WCSS Within-Cluster Sum of Squares CV Coefficient of Variation MFit Markov Fitness GM GPS Metrics Sensors 2022, 22, 9842 16 of 16 References 1. Aquino, R.; Munhoz Martins, G.H.; Palucci Vieira, L.H.; Menezes, R.P. Influence of Match Location, Quality of Opponents, and Match Status on Movement Patterns in Brazilian Professional Football Players. J. Strength Cond. Res. 2017, 31, 2155–2161. [CrossRef] 2. Rago, V.; Silva, J.; Mohr, M.; Randers, M.; Barreira, D.; Krustrup, P.; Rebelo, A. Influence of opponent standard on activity profile and fatigue development during preseasonal friendly soccer matches: A team study. Res. Sport. Med. 2018, 26, 413–424. 3. García-Unanue, J.; Pérez-Gómez, J.; Giménez, J.V.; Felipe, J.L.; Gómez-Pomares, S.; Gallardo, L.; Sánchez-Sánchez, J. Influence of contextual variables and the pressure to keep category on physical match performance in soccer players. PLoS ONE 2018, 13, e0204256. 4. Palucci Vieira, L.H.; Aquino, R.; Lago-Peñas, C.; Munhoz Martins, G.H.; Puggina, E.F.; Barbieri, F.A. Running Performance in Brazilian Professional Football Players During a Congested Match Schedule. J. Strength Cond. Res. 2018, 32, 313–325. [CrossRef] 5. Oliva-Lozano, J.M.; Fortes, V.; Krustrup, P.; Muyor, J.M. Acceleration and sprint profiles of professional male football players in relation to playing position. PLoS ONE 2020, 15, e0236959. 6. Teixeira, J.E.; Leal, M.; Ferraz, R.; Ribeiro, J.; Cachada, J.M.; Barbosa, T.M.; Monteiro, A.M.; Forte, P. Effects of Match Location, Quality of Opposition and Match Outcome on Match Running Performance in a Portuguese Professional Football Team. Entropy 2021, 23, 973. 7. Aquino, R.; Carling, C.; Palucci Vieira, L.H.; Martins, G.; Jabor, G.; Machado, J.; Santiago, P.; Garganta, J.; Puggina, E. Influence of Situational Variables, Team Formation, and Playing Position on Match Running Performance and Social Network Analysis in Brazilian Professional Soccer Players. J. Strength Cond. Res. 2020, 34, 808–817. [CrossRef] 8. Nobari, H.; Oliveira, R.; Brito, J.P.; Pérez-Gómez, J.; Clemente, F.M.; Ardigò, L.P. Comparison of Running Distance Variables and Body Load in Competitions Based on Their Results: A Full-Season Study of Professional Soccer Players. Int. J. Environ. Res. Public Health 2021, 18, 2077. 9. Ponce-Bordón, J.C.; Díaz-García, J.; López-Gajardo, M.A.; Lobo-Triviño, D.; López del Campo, R.; Resta, R.; García-Calvo, T. The Influence of Time Winning and Time Losing on Position-Specific Match Physical Demands in the Top One Spanish Soccer League. Sensors 2021, 21, 6843. 10. Lorenzo-Martinez, M.; Kalén, A.; Rey, E.; López-Del Campo, R.; Resta, R.; Lago-Peñas, C. Do elite soccer players cover less distance when their team spent more time in possession of the ball? Sci. Med. Footb. 2021, 5, 310–316. . 24733938.2020.1853211. [CrossRef] 11. Redwood-Brown, A.J.; O’Donoghue, P.G.; Nevill, A.M.; Saward, C.; Dyer, N.; Sunderland, C. Effects of situational variables on the physical activity profiles of elite soccer players in different score line states. Scand. J. Med. Sci. Sports 2018, 28, 2515–2526. [CrossRef] [PubMed] 12. Trewin, J.; Meylan, C.; Varley, M.C.; Cronin, J. The influence of situational and environmental factors on match-running in soccer: A systematic review. Sci. Med. Footb. 2017, 1, 183–194. 13. Liu, H.; Gómez, M.A.; Gonçalves, B.; Sampaio, J. Technical performance and match-to-match variation in elite football teams. J. Sports Sci. 2016, 34, 509–518. 14. Oliveira, R.; Brito, J.P.; Martins, A.; Mendes, B.; Marinho, D.A.; Ferraz, R.; Marques, M.C. In-season internal and external training load quantification of an elite European soccer team. PLoS ONE 2019, 14, e0209393. 15. Tan, J.H.; Polglaze, T.; Peeling, P. Validity and reliability of a player-tracking device to identify movement orientation in team sports. Int. J. Perform. Anal. Sport 2021, 21, 790–803. 16. Prampero, P.E.d.; Osgnach, C. Metabolic Power in Team Sports—Part 1: An Update. Int. J. Sports Med. 2018, 39, 581–587. 17. Osgnach, C.; Prampero, P.E.d. Metabolic Power in Team Sports—Part 2: Aerobic and Anaerobic Energy Yields. Int. J. Sport. Med. 2018, 39, 588–595. 18. Delves, R.I.M.; Aughey, R.J.; Ball, K.; Duthie, G.M. The Quantification of Acceleration Events in Elite Team Sport: A Systematic Review. Sports Med. 2021, 7, 45. [CrossRef] 19. Prampero, P.; Fusi, S.; Sepulcri, L.; Morin, J.B.; Belli, A.; Antonutto, G. Sprint running: A new energetic approach. J. Exp. Biol. 2005, 208, 2809–2816. [CrossRef] 20. Klemp, M.; Memmert, D.; Rein, R. The influence of running performance on scoring the first goal in a soccer match. Int. J. Sport. Sci. Coach. 2021, 17, 17479541211035382. 21. Hennessy, L.; Jeffreys, I. The Current Use of GPS, Its Potential, and Limitations in Soccer. Strength Cond. J. 2018, 40, 83–94. [CrossRef] 22. Cian, C. Macrocycle Overview—Does Player Tracking aid Periodized Peak Performance? Available online: https://pro.statsports. com/macrocycle-overview-does-player-tracking-aid-periodized-peak-performance/ (accessed on 1 December 2022).

Extended Energy-Expenditure Model in Soccer: Evaluating Player Performance in the Context of the Game.

12-14-2022

Skoki, Arian,Rossi, Alessio,Cintia, Paolo,Pappalardo, Luca,Štajduhar, Ivan

eng

PMC10651037

Table 1S. Number of people in each category by age group. Significant trend toward younger individuals reporting higher running volume, with more than 75% of the elite group falling between the ages of 18 and 35. S1 Table. Table 2S. Full running volume vs. blood biomarker results Biomarker ANOVA p-value Trend p-value lowest mean highest mean Alb <1e-16 <0.001 MVAM PRO ALT <1e-16 <1e-16 SED PRO AST <1e-16 <0.001 SED PRO B12 <0.001 <0.001 SED PRO BASOS 0.001 0.004 LVAM PRO BASOS_PCT <0.001 0.156 SED PRO Ca 0.007 0.030 MVAM PRO Chol <0.001 0.005 PRO SED CK <1e-16 <1e-16 SED PRO Cor <0.001 0.675 SED PRO D <1e-16 0.424 SED PRO DHEAS <0.001 <0.001 SED PRO EOS <0.001 0.371 HVAM SED EOS_PCT <0.001 0.137 HVAM MVAM FE <0.001 0.119 SED PRO Fer <1e-16 <1e-16 MVAM SED Fol <1e-16 <0.001 SED PRO FT <0.001 0.013 SED PRO GGT <1e-16 <0.001 PRO SED Glu 0.087 0.184 PRO SED Hb 0.002 <0.001 MVAM PRO HCT 0.053 0.055 MVAM PRO HDL <1e-16 <0.001 SED PRO HbA1c <0.001 0.010 PRO SED hsCRP <0.001 0.176 PRO SED K <1e-16 <0.001 SED LVAM LDL <0.001 0.006 PRO SED LYMPHS <0.001 0.008 PRO SED LYMPHS_PCT <1e-16 0.417 SED PRO Biomarker ANOVA p-value Trend p-value lowest mean highest mean MCH 0.197 0.077 SED PRO MCHC <1e-16 0.276 SED PRO MCV <0.001 <0.001 SED PRO Mg <0.001 0.276 PRO SED MONOS <0.001 0.175 PRO SED MONOS_PCT <0.001 0.137 SED LVAM MPV 0.058 0.089 SED HVAM Na <1e-16 0.622 HVAM SED NEUT <0.001 0.007 PRO SED NEUT_PCT <0.001 0.764 PRO SED PLT <0.001 0.058 LVAM SED RBC 0.016 0.880 MVAM SED RBC_Mg <0.001 0.773 PRO SED RDW <1e-16 0.002 PRO SED SHBG <1e-16 0.004 SED PRO Tes <1e-16 0.675 MVAM LVAM Tg <1e-16 <1e-16 PRO SED TIBC <0.001 0.417 LVAM MVAM TS <1e-16 0.298 SED PRO WBC <1e-16 <1e-16 PRO SED S2 Table. Table 3S. 2S-MR results with BMI as the exposure and select biomarkers as outcomes S3. S3 Table. Table 4S. 2S-MR results with BMI with biomarkers as exposures and BMI as outcome to assess reverse causality id.exposure id.outcome outcome exposure method nsnp b se pval ebi-a- GCST004631 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Basophil percentage of white cells || id:ebi-a-GCST004631 MR Egger 55 0.02850274 0.03058163 0.35555179 ebi-a- GCST004631 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Basophil percentage of white cells || id:ebi-a-GCST004631 Weighted median 55 0.01115946 0.01802141 0.53576264 ebi-a- GCST004631 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Basophil percentage of white cells || id:ebi-a-GCST004631 Inverse variance weighted 55 -0.0133255 0.01548404 0.38946137 ebi-a- GCST004631 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Basophil percentage of white cells || id:ebi-a-GCST004631 Simple mode 55 0.00553647 0.03733532 0.88266604 ebi-a- GCST004631 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Basophil percentage of white cells || id:ebi-a-GCST004631 Weighted mode 55 -0.0005482 0.02054445 0.97881084 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-a-1012 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Plasma cortisol || id:ieu-a-1012 Wald ratio 1 -0.0162841 0.0285215 0.56803913 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-a-1050 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Ferritin || id:ieu-a-1050 MR Egger 4 -0.0692185 0.02954547 0.14388806 ieu-a-1050 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Ferritin || id:ieu-a-1050 Weighted median 4 -0.0458501 0.01819113 0.01171994 ieu-a-1050 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Ferritin || id:ieu-a-1050 Inverse variance weighted 4 -0.0401901 0.01571805 0.01055975 ieu-a-1050 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Ferritin || id:ieu-a-1050 Simple mode 4 -0.040832 0.02570683 0.21040991 ieu-a-1050 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Ferritin || id:ieu-a-1050 Weighted mode 4 -0.0508659 0.01904314 0.07562034 S4. S4 Table. id.exposure id.outcome outcome exposure method nsnp b se pval ukb-b-11349 ieu-b-40 body mass index || id:ieu-b- 40 Folate || id:ukb-b- 11349 Wald ratio 1 0.04546597 0.058838 32 0.439683 8 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-b-114 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Fasting glucose || id:ieu-b-114 MR Egger 30 -0.0453952 0.113428 53 0.692038 95 ieu-b-114 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Fasting glucose || id:ieu-b-114 Weighted median 30 0.00266784 0.031883 98 0.933316 16 ieu-b-114 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Fasting glucose || id:ieu-b-114 Inverse variance weighted 30 -0.0341858 0.052964 48 0.518637 21 ieu-b-114 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Fasting glucose || id:ieu-b-114 Simple mode 30 -0.01322 0.067068 73 0.845115 55 ieu-b-114 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Fasting glucose || id:ieu-b-114 Weighted mode 30 0.00016512 0.029158 44 0.995520 48 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-a-270 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Haemoglobin concentration || id:ieu-a-270 MR Egger 15 0.00144021 0.07587 258 0.98514 372 ieu-a-270 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Haemoglobin concentration || id:ieu-a-270 Weighted median 15 0.01307681 0.02023 195 0.51805 628 ieu-a-270 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Haemoglobin concentration || id:ieu-a-270 Inverse variance weighted 15 -0.0334432 0.02660 806 0.20879 683 ieu-a-270 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Haemoglobin concentration || id:ieu-a-270 Simple mode 15 -0.1129022 0.05448 673 0.05720 502 ieu-a-270 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Haemoglobin concentration || id:ieu-a-270 Weighted mode 15 0.01648048 0.02138 074 0.45363 255 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-b-103 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 HbA1C || id:ieu-b- 103 MR Egger 11 0.01268313 0.08396 0.88325 885 ieu-b-103 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 HbA1C || id:ieu-b- 103 Weighted median 11 -0.0069654 0.03248 061 0.83019 839 ieu-b-103 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 HbA1C || id:ieu-b- 103 Inverse variance weighted 11 0.0283815 0.03446 064 0.41017 145 ieu-b-103 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 HbA1C || id:ieu-b- 103 Simple mode 11 -0.0257779 0.06293 863 0.69075 483 ieu-b-103 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 HbA1C || id:ieu-b- 103 Weighted mode 11 -0.0208929 0.03914 067 0.60514 756 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 MR Egger 26 0.26135491 0.14952 437 0.09326 378 ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 Weighted median 26 0.08455861 0.04638 846 0.06832 803 ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 Inverse variance weighted 26 -0.0072778 0.05705 404 0.89849 796 ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 Simple mode 26 0.09054068 0.07897 524 0.26246 646 ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 Weighted mode 26 0.09962869 0.04750 758 0.04626 112 id.exposure id.outcome outcome exposure method nsnp b se pval ebi-a- GCST005068 ieu-b-40 body mass index || id:ieu-b- 40 LDL cholesterol || id:ebi-a-GCST005068 MR Egger 4 0.01267769 0.05816 618 0.84767 998 ebi-a- GCST005068 ieu-b-40 body mass index || id:ieu-b- 40 LDL cholesterol || id:ebi-a-GCST005068 Weighted median 4 -0.0332107 0.01134 854 0.00342 875 ebi-a- GCST005068 ieu-b-40 body mass index || id:ieu-b- 40 LDL cholesterol || id:ebi-a-GCST005068 Inverse variance weighted 4 -0.0320938 0.00863 712 0.00020 257 ebi-a- GCST005068 ieu-b-40 body mass index || id:ieu-b- 40 LDL cholesterol || id:ebi-a-GCST005068 Simple mode 4 -0.036639 0.01673 254 0.11628 833 ebi-a- GCST005068 ieu-b-40 body mass index || id:ieu-b- 40 LDL cholesterol || id:ebi-a-GCST005068 Weighted mode 4 -0.034971 0.01482 963 0.09956 398 id.exposure id.outcome outcome exposure method nsnp b se pval ebi-a- GCST90002336 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Mean corpuscular volume || id:ebi-a- GCST90002336 MR Egger 9 0.01627788 0.03256 415 0.63249 338 ebi-a- GCST90002336 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Mean corpuscular volume || id:ebi-a- GCST90002336 Weighted median 9 -0.0169769 0.01096 864 0.12167 813 ebi-a- GCST90002336 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Mean corpuscular volume || id:ebi-a- GCST90002336 Inverse variance weighted 9 -0.0212777 0.01246 612 0.08785 201 ebi-a- GCST90002336 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Mean corpuscular volume || id:ebi-a- GCST90002336 Simple mode 9 -0.0220168 0.01637 922 0.21575 632 ebi-a- GCST90002336 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Mean corpuscular volume || id:ebi-a- GCST90002336 Weighted mode 9 -0.0231087 0.01392 193 0.13552 035 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-a-1008 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Platelet count || id:ieu-a-1008 MR Egger 32 -0.0006727 0.00070 143 0.34517 459 ieu-a-1008 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Platelet count || id:ieu-a-1008 Weighted median 32 0.00013472 0.00021 428 0.52954 228 ieu-a-1008 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Platelet count || id:ieu-a-1008 Inverse variance weighted 32 -0.0003268 0.00024 878 0.18895 678 ieu-a-1008 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Platelet count || id:ieu-a-1008 Simple mode 32 9.63E-05 0.00034 761 0.78352 525 ieu-a-1008 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Platelet count || id:ieu-a-1008 Weighted mode 32 0.00011659 0.00024 456 0.63688 891 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 MR Egger 26 0.26135491 0.14952 437 0.09326 378 ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 Weighted median 26 0.08455861 0.04591 029 0.06550 111 ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 Inverse variance weighted 26 -0.0072778 0.05705 404 0.89849 796 ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 Simple mode 26 0.09054068 0.07653 474 0.24793 721 ieu-a-275 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red blood cell count || id:ieu-a-275 Weighted mode 26 0.09962869 0.04605 998 0.04030 183 id.exposure id.outcome outcome exposure method nsnp b se pval ebi-a- GCST006804 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red cell distribution width || id:ebi-a- GCST006804 MR Egger 122 -0.0341912 0.03423 066 0.31987 855 ebi-a- GCST006804 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red cell distribution width || id:ebi-a- GCST006804 Weighted median 122 0.01563428 0.00970 492 0.10718 764 ebi-a- GCST006804 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red cell distribution width || id:ebi-a- GCST006804 Inverse variance weighted 122 0.0300081 0.01723 486 0.08166 104 ebi-a- GCST006804 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red cell distribution width || id:ebi-a- GCST006804 Simple mode 122 0.01857783 0.02026 892 0.36119 26 ebi-a- GCST006804 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Red cell distribution width || id:ebi-a- GCST006804 Weighted mode 122 0.01128028 0.01217 303 0.35594 689 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-a-302 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Triglycerides || id:ieu-a-302 MR Egger 55 -0.0445751 0.03601 334 0.22126 815 ieu-a-302 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Triglycerides || id:ieu-a-302 Weighted median 55 -0.0315603 0.01590 831 0.04726 817 ieu-a-302 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Triglycerides || id:ieu-a-302 Inverse variance weighted 55 -0.0214287 0.02233 545 0.33735 629 ieu-a-302 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Triglycerides || id:ieu-a-302 Simple mode 55 -0.0456892 0.02952 211 0.12755 31 ieu-a-302 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 Triglycerides || id:ieu-a-302 Weighted mode 55 -0.0297723 0.01209 893 0.01709 375 id.exposure id.outcome outcome exposure method nsnp b se pval ieu-b-30 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 white blood cell count || id:ieu-b-30 MR Egger 475 -0.0347487 0.02445 633 0.15602 056 ieu-b-30 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 white blood cell count || id:ieu-b-30 Weighted median 475 -0.0307482 0.01204 555 0.01069 025 ieu-b-30 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 white blood cell count || id:ieu-b-30 Inverse variance weighted 475 -0.040535 0.01168 163 0.00052 05 ieu-b-30 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 white blood cell count || id:ieu-b-30 Simple mode 475 0.00133145 0.03782 675 0.97193 623 ieu-b-30 ukb-a-248 Body mass index (BMI) || id:ukb-a-248 white blood cell count || id:ieu-b-30 Weighted mode 475 -0.0184754 0.02085 621 0.37614 876 id.exposure id.outcome outcome exposure method nsnp b se pval ukb-d- 30830_raw ieu-b-40 body mass index || id:ieu-b- 40 SHBG || id:ukb-d- 30830_raw MR Egger 135 0.00121356 0.00180 979 0.50366 715 ukb-d- 30830_raw ieu-b-40 body mass index || id:ieu-b- 40 SHBG || id:ukb-d- 30830_raw Weighted median 135 0.00102542 0.00037 386 0.00609 279 ukb-d- 30830_raw ieu-b-40 body mass index || id:ieu-b- 40 SHBG || id:ukb-d- 30830_raw Inverse variance weighted 135 -0.002129 0.00115 3 0.06482 013 ukb-d- 30830_raw ieu-b-40 body mass index || id:ieu-b- 40 SHBG || id:ukb-d- 30830_raw Simple mode 135 -0.0011256 0.00124 723 0.36841 147 id.exposure id.outcome outcome exposure method nsnp b se pval ebi-a-GCST006098 ebi-a- GCST005068 LDL cholesterol || id:ebi-a- GCST005068 Vigorous physical activity || id:ebi-a-GCST006098 MR Egger 7 4.55735741 5.397582 61 0.437005 89 ebi-a-GCST006098 ebi-a- GCST005068 LDL cholesterol || id:ebi-a- GCST005068 Vigorous physical activity || id:ebi-a-GCST006098 Weighted median 7 -0.6208293 0.838434 47 0.459019 39 ebi-a-GCST006098 ebi-a- GCST005068 LDL cholesterol || id:ebi-a- GCST005068 Vigorous physical activity || id:ebi-a-GCST006098 Inverse variance weighted 7 -0.2426177 0.660161 46 0.713236 61 ebi-a-GCST006098 ebi-a- GCST005068 LDL cholesterol || id:ebi-a- GCST005068 Vigorous physical activity || id:ebi-a-GCST006098 Simple mode 7 -1.231375 1.532513 13 0.452333 86 ebi-a-GCST006098 ebi-a- GCST005068 LDL cholesterol || id:ebi-a- GCST005068 Vigorous physical activity || id:ebi-a-GCST006098 Weighted mode 7 -1.0821029 1.017395 34 0.328430 07 ebi-a-GCST006098 ieu-a-1050 Ferritin || id:ieu-a-1050 Vigorous physical activity || id:ebi-a-GCST006098 MR Egger 5 -5.2248204 5.495151 03 0.411848 23 ebi-a-GCST006098 ieu-a-1050 Ferritin || id:ieu-a-1050 Vigorous physical activity || id:ebi-a-GCST006098 Weighted median 5 -0.7040971 0.647521 49 0.276872 22 ebi-a-GCST006098 ieu-a-1050 Ferritin || id:ieu-a-1050 Vigorous physical activity || id:ebi-a-GCST006098 Inverse variance weighted 5 -0.1408587 0.767230 61 0.854332 25 ebi-a-GCST006098 ieu-a-1050 Ferritin || id:ieu-a-1050 Vigorous physical activity || id:ebi-a-GCST006098 Simple mode 5 -0.8515134 0.843286 04 0.369732 73 ebi-a-GCST006098 ieu-a-1050 Ferritin || id:ieu-a-1050 Vigorous physical activity || id:ebi-a-GCST006098 Weighted mode 5 -0.9327437 0.769158 52 0.291972 96 ebi-a-GCST006098 ieu-a-302 Triglycerides || id:ieu-a-302 Vigorous physical activity || id:ebi-a-GCST006098 MR Egger 5 -1.4775783 1.334354 04 0.348955 32 ebi-a-GCST006098 ieu-a-302 Triglycerides || id:ieu-a-302 Vigorous physical activity || id:ebi-a-GCST006098 Weighted median 5 -0.1485007 0.262002 58 0.570856 3 ebi-a-GCST006098 ieu-a-302 Triglycerides || id:ieu-a-302 Vigorous physical activity || id:ebi-a-GCST006098 Inverse variance weighted 5 -0.0745059 0.221142 62 0.736182 24 ebi-a-GCST006098 ieu-a-302 Triglycerides || id:ieu-a-302 Vigorous physical activity || id:ebi-a-GCST006098 Simple mode 5 -0.1338663 0.368514 73 0.734794 89 ebi-a-GCST006098 ieu-a-302 Triglycerides || id:ieu-a-302 Vigorous physical activity || id:ebi-a-GCST006098 Weighted mode 5 -0.1563444 0.328858 88 0.659289 19 ebi-a-GCST006098 ukb-b-11349 Folate || id:ukb-b-11349 Vigorous physical activity || id:ebi-a-GCST006098 MR Egger 7 0.44992252 2.058173 78 0.835602 06 ebi-a-GCST006098 ukb-b-11349 Folate || id:ukb-b-11349 Vigorous physical activity || id:ebi-a-GCST006098 Weighted median 7 -0.2016639 0.320053 64 0.528633 09 ebi-a-GCST006098 ukb-b-11349 Folate || id:ukb-b-11349 Vigorous physical activity || id:ebi-a-GCST006098 Inverse variance weighted 7 -0.2577298 0.244615 28 0.292060 17 ebi-a-GCST006098 ukb-b-11349 Folate || id:ukb-b-11349 Vigorous physical activity || id:ebi-a-GCST006098 Simple mode 7 0.0568742 0.467024 35 0.907049 56 ebi-a-GCST006098 ukb-b-11349 Folate || id:ukb-b-11349 Vigorous physical activity || id:ebi-a-GCST006098 Weighted mode 7 0.07601041 0.465339 68 0.875610 65 Table 5S. 2S-MR results with vigorous physical activity as exposure and blood biomarkers as outcomes S5. S5 Table. ebi-a-GCST006098 ukb-d- 30070_irnt Red blood cell (erythrocyte) distribution width || id:ukb-d- 30070_irnt Vigorous physical activity || id:ebi-a-GCST006098 MR Egger 7 1.57820562 1.795700 92 0.419691 32 ebi-a-GCST006098 ukb-d- 30070_irnt Red blood cell (erythrocyte) distribution width || id:ukb-d- 30070_irnt Vigorous physical activity || id:ebi-a-GCST006098 Weighted median 7 -0.0549663 0.143906 26 0.702491 75 ebi-a-GCST006098 ukb-d- 30070_irnt Red blood cell (erythrocyte) distribution width || id:ukb-d- 30070_irnt Vigorous physical activity || id:ebi-a-GCST006098 Inverse variance weighted 7 -0.2321729 0.213298 82 0.276380 26 ebi-a-GCST006098 ukb-d- 30070_irnt Red blood cell (erythrocyte) distribution width || id:ukb-d- 30070_irnt Vigorous physical activity || id:ebi-a-GCST006098 Simple mode 7 -0.0408618 0.177334 68 0.825419 27 ebi-a-GCST006098 ukb-d- 30070_irnt Red blood cell (erythrocyte) distribution width || id:ukb-d- 30070_irnt Vigorous physical activity || id:ebi-a-GCST006098 Weighted mode 7 -0.0920344 0.182147 56 0.631385 89 ebi-a-GCST006098 ukb-d- 30830_irnt SHBG || id:ukb-d-30830_irnt Vigorous physical activity || id:ebi-a-GCST006098 MR Egger 7 0.46503747 1.973712 9 0.823077 52 ebi-a-GCST006098 ukb-d- 30830_irnt SHBG || id:ukb-d-30830_irnt Vigorous physical activity || id:ebi-a-GCST006098 Weighted median 7 0.11870029 0.151914 89 0.434590 52 ebi-a-GCST006098 ukb-d- 30830_irnt SHBG || id:ukb-d-30830_irnt Vigorous physical activity || id:ebi-a-GCST006098 Inverse variance weighted 7 0.33282495 0.213695 39 0.119358 06 ebi-a-GCST006098 ukb-d- 30830_irnt SHBG || id:ukb-d-30830_irnt Vigorous physical activity || id:ebi-a-GCST006098 Simple mode 7 0.07719013 0.190096 22 0.698784 8 ebi-a-GCST006098 ukb-d- 30830_irnt SHBG || id:ukb-d-30830_irnt Vigorous physical activity || id:ebi-a-GCST006098 Weighted mode 7 0.08827653 0.182289 32 0.645365 99 id.exposure id.outcome outcome exposure Egger intercept se pval ebi-a-GCST006098 ebi-a- GCST005068 LDL cholesterol || id:ebi-a- GCST005068 Vigorous physical activity || id:ebi-a-GCST006098 -0.0463331 0.051710 55 0.411305 82 ebi-a-GCST006098 ieu-a-1050 Ferritin || id:ieu-a-1050 Vigorous physical activity || id:ebi-a-GCST006098 0.04824757 0.051622 26 0.418926 33 ebi-a-GCST006098 ieu-a-302 Triglycerides || id:ieu-a-302 Vigorous physical activity || id:ebi-a-GCST006098 0.01391277 0.013048 39 0.364507 21 ebi-a-GCST006098 ukb-b-11349 Folate || id:ukb-b-11349 Vigorous physical activity || id:ebi-a-GCST006098 -0.0065813 0.019005 84 0.743228 24 ebi-a-GCST006098 ukb-d- 30070_irnt Red blood cell (erythrocyte) distribution width || id:ukb-d- 30070_irnt Vigorous physical activity || id:ebi-a-GCST006098 -0.0168346 0.016580 53 0.356535 74 ebi-a-GCST006098 ukb-d- 30830_irnt SHBG || id:ukb-d-30830_irnt Vigorous physical activity || id:ebi-a-GCST006098 -0.0012295 0.018225 48 0.948828 18 Tests for horizontal pleiotropy id.exposure id.outcome outcome exposure method nsn p b se pval ebi-a- GCST006098 ukb-b-10217 Sweets intake || id:ukb-b- 10217 Vigorous physical activity || id:ebi- a-GCST006098 MR Egger 7 - 0.0903539 1.5078935 0.95454 011 ebi-a- GCST006098 ukb-b-10217 Sweets intake || id:ukb-b- 10217 Vigorous physical activity || id:ebi- a-GCST006098 Weighted median 7 - 0.2367422 0.21224606 0.26467 304 ebi-a- GCST006098 ukb-b-10217 Sweets intake || id:ukb-b- 10217 Vigorous physical activity || id:ebi- a-GCST006098 Inverse variance weighted 7 - 0.1760505 0.17921106 0.32592 043 ebi-a- GCST006098 ukb-b-10217 Sweets intake || id:ukb-b- 10217 Vigorous physical activity || id:ebi- a-GCST006098 Simple mode 7 - 0.2770321 0.31802813 0.41719 033 ebi-a- GCST006098 ukb-b-10217 Sweets intake || id:ukb-b- 10217 Vigorous physical activity || id:ebi- a-GCST006098 Weighted mode 7 - 0.2596458 0.31161898 0.43662 814 ebi-a- GCST006098 ukb-b-11679 Type of special diet followed: Vegetarian || id:ukb-b-11679 Vigorous physical activity || id:ebi- a-GCST006098 MR Egger 6 0.3098933 2 0.88272515 0.74325 054 ebi-a- GCST006098 ukb-b-11679 Type of special diet followed: Vegetarian || id:ukb-b-11679 Vigorous physical activity || id:ebi- a-GCST006098 Weighted median 6 0.0220304 6 0.0642637 0.73173 884 ebi-a- GCST006098 ukb-b-11679 Type of special diet followed: Vegetarian || id:ukb-b-11679 Vigorous physical activity || id:ebi- a-GCST006098 Inverse variance weighted 6 0.0230808 9 0.04874316 0.63584 178 ebi-a- GCST006098 ukb-b-11679 Type of special diet followed: Vegetarian || id:ukb-b-11679 Vigorous physical activity || id:ebi- a-GCST006098 Simple mode 6 0.0214745 2 0.10369298 0.84410 404 ebi-a- GCST006098 ukb-b-11679 Type of special diet followed: Vegetarian || id:ukb-b-11679 Vigorous physical activity || id:ebi- a-GCST006098 Weighted mode 6 0.0245167 9 0.09506696 0.80677 012 ebi-a- GCST006098 ukb-b-1996 Salad / raw vegetable intake || id:ukb-b-1996 Vigorous physical activity || id:ebi- a-GCST006098 MR Egger 7 0.6661970 2 1.2339482 0.61244 007 ebi-a- GCST006098 ukb-b-1996 Salad / raw vegetable intake || id:ukb-b-1996 Vigorous physical activity || id:ebi- a-GCST006098 Weighted median 7 0.5139604 0.10299379 6.03E- 07 ebi-a- GCST006098 ukb-b-1996 Salad / raw vegetable intake || id:ukb-b-1996 Vigorous physical activity || id:ebi- a-GCST006098 Inverse variance weighted 7 0.5044851 7 0.13388668 0.00016 456 ebi-a- GCST006098 ukb-b-1996 Salad / raw vegetable intake || id:ukb-b-1996 Vigorous physical activity || id:ebi- a-GCST006098 Simple mode 7 0.6270295 4 0.14130406 0.00438 784 Table 6S. 2S-MR results with vigorous physical activity as exposure and blood biomarkers as outcomes t lifestyle habits S6. S6 Table. ebi-a- GCST006098 ukb-b-1996 Salad / raw vegetable intake || id:ukb-b-1996 Vigorous physical activity || id:ebi-a- GCST006098 Weighted mode 7 0.58940834 0.14675 62 0.00698 862 ebi-a- GCST006098 ukb-b-2209 Oily fish intake || id:ukb-b- 2209 Vigorous physical activity || id:ebi-a- GCST006098 MR Egger 7 0.95472369 2.00996 591 0.65481 403 ebi-a- GCST006098 ukb-b-2209 Oily fish intake || id:ukb-b- 2209 Vigorous physical activity || id:ebi-a- GCST006098 Weighted median 7 0.53975424 0.15310 124 0.00042 273 ebi-a- GCST006098 ukb-b-2209 Oily fish intake || id:ukb-b- 2209 Vigorous physical activity || id:ebi-a- GCST006098 Inverse variance weighted 7 0.48170818 0.21898 245 0.02782 414 ebi-a- GCST006098 ukb-b-2209 Oily fish intake || id:ukb-b- 2209 Vigorous physical activity || id:ebi-a- GCST006098 Simple mode 7 0.82403974 0.20581 323 0.00708 805 ebi-a- GCST006098 ukb-b-2209 Oily fish intake || id:ukb-b- 2209 Vigorous physical activity || id:ebi-a- GCST006098 Weighted mode 7 0.78663663 0.29210 032 0.03590 724 ebi-a- GCST006098 ukb-b-3881 Fresh fruit intake || id:ukb- b-3881 Vigorous physical activity || id:ebi-a- GCST006098 MR Egger 7 1.30879467 1.11224 869 0.29227 451 ebi-a- GCST006098 ukb-b-3881 Fresh fruit intake || id:ukb- b-3881 Vigorous physical activity || id:ebi-a- GCST006098 Weighted median 7 0.46001672 0.08901 784 2.37E- 07 ebi-a- GCST006098 ukb-b-3881 Fresh fruit intake || id:ukb- b-3881 Vigorous physical activity || id:ebi-a- GCST006098 Inverse variance weighted 7 0.38633487 0.12863 774 0.00267 088 ebi-a- GCST006098 ukb-b-3881 Fresh fruit intake || id:ukb- b-3881 Vigorous physical activity || id:ebi-a- GCST006098 Simple mode 7 0.59733916 0.11693 25 0.00220 313 ebi-a- GCST006098 ukb-b-3881 Fresh fruit intake || id:ukb- b-3881 Vigorous physical activity || id:ebi-a- GCST006098 Weighted mode 7 0.58037912 0.12022 717 0.00291 812 ebi-a- GCST006098 ukb-b-4616 Nap during day || id:ukb-b- 4616 Vigorous physical activity || id:ebi-a- GCST006098 MR Egger 7 -1.8168346 0.85758 58 0.08766 702 ebi-a- GCST006098 ukb-b-4616 Nap during day || id:ukb-b- 4616 Vigorous physical activity || id:ebi-a- GCST006098 Weighted median 7 -0.3203349 0.09563 435 0.00080 934 ebi-a- GCST006098 ukb-b-4616 Nap during day || id:ukb-b- 4616 Vigorous physical activity || id:ebi-a- GCST006098 Inverse variance weighted 7 -0.1836272 0.12240 112 0.13356 03 ebi-a- GCST006098 ukb-b-4616 Nap during day || id:ukb-b- 4616 Vigorous physical activity || id:ebi-a- GCST006098 Simple mode 7 -0.3872459 0.12727 48 0.02272 693 ebi-a- GCST006098 ukb-b-4616 Nap during day || id:ukb-b- 4616 Vigorous physical activity || id:ebi-a- GCST006098 Weighted mode 7 -0.3872459 0.11249 117 0.01375 948 ebi-a- GCST006098 ukb-b-6324 Processed meat intake || id:ukb-b-6324 Vigorous physical activity || id:ebi-a- GCST006098 MR Egger 7 0.06859678 1.00946 468 0.94845 629 ebi-a- GCST006098 ukb-b-6324 Processed meat intake || id:ukb-b-6324 Vigorous physical activity || id:ebi-a- GCST006098 Weighted median 7 -0.6292381 0.13853 861 5.57E- 06 ebi-a- GCST006098 ukb-b-6324 Processed meat intake || id:ukb-b-6324 Vigorous physical activity || id:ebi-a- GCST006098 Inverse variance weighted 7 -0.5496216 0.11346 205 1.27E- 06 ebi-a- GCST006098 ukb-b-6324 Processed meat intake || id:ukb-b-6324 Vigorous physical activity || id:ebi-a- GCST006098 Simple mode 7 -0.6879646 0.21816 478 0.01972 901 ebi-a- GCST006098 ukb-b-6324 Processed meat intake || id:ukb-b-6324 Vigorous physical activity || id:ebi-a- GCST006098 Weighted mode 7 -0.6939066 0.21655 974 0.01850 116 id.exposure id.outcome outcome exposure Egger intercept se pval ebi-a- GCST006098 ukb-b-10217 Sweets intake || id:ukb-b- 10217 Vigorous physical activity || id:ebi-a- GCST006098 -0.000797 0.01392 445 0.95657 282 ebi-a- GCST006098 ukb-b-11679 Type of special diet followed: Vegetarian || id:ukb-b-11679 Vigorous physical activity || id:ebi-a- GCST006098 -0.0025594 0.00786 299 0.76112 246 ebi-a- GCST006098 ukb-b-1996 Salad / raw vegetable intake || id:ukb-b-1996 Vigorous physical activity || id:ebi-a- GCST006098 -0.0015039 0.01139 436 0.90014 308 ebi-a- GCST006098 ukb-b-2209 Oily fish intake || id:ukb-b- 2209 Vigorous physical activity || id:ebi-a- GCST006098 -0.0043991 0.01856 093 0.82205 339 ebi-a- GCST006098 ukb-b-3881 Fresh fruit intake || id:ukb- b-3881 Vigorous physical activity || id:ebi-a- GCST006098 -0.0085789 0.01027 085 0.44163 822 ebi-a- GCST006098 ukb-b-4616 Nap during day || id:ukb-b- 4616 Vigorous physical activity || id:ebi-a- GCST006098 0.01518897 0.00791 926 0.11322 178 ebi-a- GCST006098 ukb-b-6324 Processed meat intake || id:ukb-b-6324 Vigorous physical activity || id:ebi-a- GCST006098 -0.0057496 0.00932 2 0.56437 631 Tests for horizontal pleiotropy id.exposure id.outcome outcome exposure method nsn p b se pval ukb-b-1996 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Salad / raw vegetable intake || id:ukb-b-1996 MR Egger 2 0 0.3848011 0.33578299 0.26681 018 ukb-b-1996 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Salad / raw vegetable intake || id:ukb-b-1996 Weighted median 2 0 0.2725700 2 0.06073408 7.19E- 06 ukb-b-1996 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Salad / raw vegetable intake || id:ukb-b-1996 Inverse variance weighted 2 0 0.3160797 5 0.0657002 1.50E- 06 ukb-b-1996 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Salad / raw vegetable intake || id:ukb-b-1996 Simple mode 2 0 0.3937231 5 0.14841231 0.01570 282 ukb-b-1996 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Salad / raw vegetable intake || id:ukb-b-1996 Weighted mode 2 0 0.3979305 4 0.16695029 0.02773 826 ukb-b-6324 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Processed meat intake || id:ukb-b- 6324 MR Egger 2 3 0.1350953 3 0.17451244 0.44748 219 ukb-b-6324 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Processed meat intake || id:ukb-b- 6324 Weighted median 2 3 - 0.0754533 0.03693307 0.04105 493 ukb-b-6324 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Processed meat intake || id:ukb-b- 6324 Inverse variance weighted 2 3 - 0.1085666 0.03736846 0.00366 899 ukb-b-6324 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Processed meat intake || id:ukb-b- 6324 Simple mode 2 3 - 0.0569462 0.08563101 0.51295 016 ukb-b-6324 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Processed meat intake || id:ukb-b- 6324 Weighted mode 2 3 - 0.0357803 0.07544894 0.64000 771 ukb-b-4616 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Nap during day || id:ukb-b-4616 MR Egger 8 9 - 0.1310942 0.09139345 0.15504 488 ukb-b-4616 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Nap during day || id:ukb-b-4616 Weighted median 8 9 - 0.0537114 0.0279262 0.05443 828 ukb-b-4616 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Nap during day || id:ukb-b-4616 Inverse variance weighted 8 9 - 0.0652623 0.02490192 0.00877 309 ukb-b-4616 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Nap during day || id:ukb-b-4616 Simple mode 8 9 - 0.0625663 0.0649349 0.33792 525 ukb-b-4616 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Nap during day || id:ukb-b-4616 Weighted mode 8 9 - 0.0482764 0.05502095 0.38264 819 ukb-b-3881 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Fresh fruit intake || id:ukb-b-3881 MR Egger 5 3 - 0.0277245 0.12616196 0.82694 037 Table 7S. 2S-MR with healthy/unhealthy dietary habits as exposures and vigorous physical activity as outcome to assess reverse causality S7 Table. ukb-b-3881 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Fresh fruit intake || id:ukb-b-3881 Weighted median 53 0.16136177 0.03846 63 2.73E- 05 ukb-b-3881 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Fresh fruit intake || id:ukb-b-3881 Inverse variance weighted 53 0.2103584 0.03697 867 1.28E- 08 ukb-b-3881 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Fresh fruit intake || id:ukb-b-3881 Simple mode 53 0.16760638 0.09363 947 0.07929 148 ukb-b-3881 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Fresh fruit intake || id:ukb-b-3881 Weighted mode 53 0.14738639 0.08743 063 0.09783 589 id.exposure id.outcome outcome exposure Egger intercept se pval ukb-b-1996 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Salad / raw vegetable intake || id:ukb-b- 1996 -0.0007564 0.00362 056 0.83686 113 ukb-b-6324 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Processed meat intake || id:ukb-b- 6324 -0.0037415 0.00262 036 0.16803 755 ukb-b-4616 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Nap during day || id:ukb-b-4616 0.00064846 0.00086 601 0.45600 425 ukb-b-3881 ebi-a- GCST006098 Vigorous physical activity || id:ebi-a-GCST006098 Fresh fruit intake || id:ukb-b-3881 0.00227091 0.00115 335 0.05440 223 Tests for horizontal pleiotropy Figure 1S. Assumptions of Mendelian randomization: (1) the genetic instrument is associated with the exposure. (2) the genetic instrument should not associate with a confounder. (3) the genetic instrument should affect the outcome only via the exposure. S1 Fig. Figure 2S. Blood biomarker levels with respect to self-reported running volume and professional athletes in males (m) and females (f). Fig. S2. S2 Fig. Figure 3S. 2S-MR scatter plot showing effects of vigorous physical activity as the exposure on (A) SHBG (B) Red blood cell count (C) Folate (D) Triglycerides (E) Ferritin (F) LDL as outcomes (see Table 5S for statistical significance). A B C D E F Fig. S3. S3 Fig. Figure 4S. 2S-MR scatter plot showing effects of vigorous physical activity as the exposure on (A) oily fish consumption (B) salad intake(C) fresh fruit intake (D) processed meat intake (E) daytime napping (F) sweets intake (G) vegetarian diet (see Table 6S for statistical significance). E F G Fig. S4. S4 Fig. Figure 5S. 2S-MR scatter plot showing effects of dietary behaviors as the exposures on vigorous physical activity: (A) salad intake (B) processed meat intake (C) daytime napping (D) fresh fruit intake (see Table 7S for statistical significance). A B C D S5 Fig.

Dose response of running on blood biomarkers of wellness in generally healthy individuals.

11-15-2023

Nogal, Bartek,Vinogradova, Svetlana,Jorge, Milena,Torkamani, Ali,Fabian, Paul,Blander, Gil

eng

PMC4887549

REVIEW ARTICLE Is There an Economical Running Technique? A Review of Modifiable Biomechanical Factors Affecting Running Economy Isabel S. Moore1 Published online: 27 January 2016  The Author(s) 2016. This article is published with open access at Springerlink.com Abstract Running economy (RE) has a strong relation- ship with running performance, and modifiable running biomechanics are a determining factor of RE. The purposes of this review were to (1) examine the intrinsic and extrinsic modifiable biomechanical factors affecting RE; (2) assess training-induced changes in RE and running biomechanics; (3) evaluate whether an economical running technique can be recommended and; (4) discuss potential areas for future research. Based on current evidence, the intrinsic factors that appeared beneficial for RE were using a preferred stride length range, which allows for stride length deviations up to 3 % shorter than preferred stride length; lower vertical oscillation; greater leg stiffness; low lower limb moment of inertia; less leg extension at toe-off; larger stride angles; alignment of the ground reaction force and leg axis during propulsion; maintaining arm swing; low thigh antagonist–agonist muscular coactivation; and low activation of lower limb muscles during propulsion. Extrinsic factors associated with a better RE were a firm, compliant shoe–surface interaction and being barefoot or wearing lightweight shoes. Several other modifiable biomechanical factors presented inconsistent relationships with RE. Running biomechanics during ground contact appeared to play an important role, specifically those dur- ing propulsion. Therefore, this phase has the strongest direct links with RE. Recurring methodological problems exist within the literature, such as cross-comparisons, assessing variables in isolation, and acute to short-term interventions. Therefore, recommending a general economical running technique should be approached with caution. Future work should focus on interdisciplinary longitudinal investigations combining RE, kinematics, kinetics, and neuromuscular and anatomical aspects, as well as applying a synergistic approach to understanding the role of kinetics. Key Points Running biomechanics during ground contact, particularly those related to propulsion, such as less leg extension at toe-off, larger stride angles, alignment of the ground reaction force and leg axis, and low activation of the lower limb muscles, appear to have the strongest direct links with running economy. Inconsistent findings and limited understanding still exist for several spatiotemporal, kinematic, kinetic, and neuromuscular factors and how they relate to running economy. 1 Introduction For competitive runners, decreasing the time needed to complete a race distance is crucial. Consequently, there is a need to understand the determinants of running perfor- mance. Several physiological determinants have been identified, which include a high maximal oxygen uptake ( _VO2max) [1, 2], lactate threshold [3, 4], and running economy (RE) [5, 6]. & Isabel S. Moore imoore@cardiffmet.ac.uk 1 Cardiff School of Sport, Cardiff Metropolitan University, Cardiff CF23 6XD, Wales, UK 123 Sports Med (2016) 46:793–807 DOI 10.1007/s40279-016-0474-4 In a heterogeneous group of runners, _VO2max is strongly related to running performance [7]. However, in a group of runners with a similar _VO2max, _VO2max cannot be used to discern between those who out-perform others [6]. A measure that can distinguish between good and poor run- ning performers is the rate of oxygen consumed at a given submaximal running velocity, termed RE [5, 8, 9], with lower oxygen consumption ( _VO2) indicating better RE during steady-state running. For a group of runners with a similar _VO2max, RE can differ by as much as 30 % and is a better predictor of running performance than _VO2max [6, 8, 10]. Several researchers have reported strong associations between RE and running performance [5, 7, 11, 12]. Additionally, RE differs substantially between elite, trained (recreational), and untrained runners and also between males and females [13–17]. Saunders et al. [18] proposed the following determinants of RE: training, environment, physiology, anthropometry, and running biomechanics. Studies utilizing interventions show RE can be improved [19], meaning it is a ‘trainable’ parameter [20]. Improvements in RE have ranged from 2 to 8 % using various short-term training modes, such as plyometric [21– 23], strength and resistance [24–27], whole-body vibration [28], interval [29–31], altitude [32, 33], and endurance running [34, 35]. In comparison, long-term physiological training can improve RE by 15 % [12]. Jones [12] reported that such an improvement over 9 years was probably a crucial factor in the elite marathon runner’s continued improvement in running performance. For intervention studies concerned with improving RE, the initial fitness level of participants is particularly important [18], with a high initial fitness level perhaps explaining why not all interventions have successfully improved RE [36–39]. Nevertheless, the trainability of RE suggests certain factors affecting RE can be modified. One such factor that can influence RE is an individual’s running biomechanics. Understanding what constitutes an economical running technique has been the focus of much research. Specific factors include spatiotemporal factors [40, 41], lower limb kinematics [34, 42], kinetics [9, 43, 44], neuromuscular factors [45–48], the shoe–surface interaction [49–54], and trunk and upper limb biomechanics [55–57]. Synthesizing the literature within this field of research has received limited attention, with some still drawing upon descriptors provided up to 20 years ago [18, 58]. Much research has been conducted since, in an attempt to answer the question: is there an economical running technique? Therefore, the purposes of this review are to (1) examine the intrinsic and extrinsic modifiable biomechanical factors affecting RE; (2) assess training-induced changes in RE and running biomechanics; (3) evaluate whether an economical running technique can be recommended; and (4) discuss potential areas for future research directions. 2 Modifiable Biomechanical Factors Affecting Running Economy Several modifiable biomechanical factors may affect RE. Each factor can be considered either intrinsic (internal) or extrinsic (external). Intrinsic factors refer to an individual’s running biomechanics. These factors can be further cate- gorised as spatiotemporal (parameters relating to changes in and/or phases of the gait cycle, such as ground contact time and stride length); kinematics (the movement patterns, such as lower limb joint angles); kinetics (the forces that cause motion, such as ground reaction force [GRF]); and neuromuscular (the nerves and muscles, such as the acti- vation and coactivation of muscles). The extrinsic factors covered in this review relate to the shoe–surface interaction and focus on footwear, orthotics, and running surface. Evidence for how each factor affects RE is reviewed and discussed. 3 Spatiotemporal Factors Stride frequency and stride length are mutually dependent and define running speed. If running speed is kept constant, increasing either stride frequency or stride length will result in a decrease of the other. Runners appear to natu- rally choose a stride frequency or stride length that is economically optimal, or at least very near to being eco- nomically optimal. This innate, subconscious fine-tuning of running biomechanics is referred to as self-optimization [34, 42]. Studies supporting this self-optimizing theory generally use acute manipulations of stride frequency or stride length and mathematical curve-fitting procedures to derive the most economical stride frequency and length [40, 59–61]. Interestingly, a trained runner’s mathematical optimal stride frequency or stride length is, on average, 3 % faster or 3 % shorter than their preferred frequency or length [40, 59, 61]. Acute and short-term manipulations whereby stride length has been shortened by 3 % show RE to be unaf- fected [50, 62], whereas stride length deviations greater than 6 % are detrimental to RE [59]. Collectively, these results suggest there is an optimal stride length ‘range’ that trained runners can acutely adopt without compromising their RE. This range appears to be the preferred stride length minus 3 % to the preferred stride length. Impor- tantly, even in a fatigued state, trained runners reduce their stride frequency compared with a non-fatigued state and 794 I. S. Moore 123 produce a preferred stride frequency that is similar to their optimal stride frequency achieved in a fatigued state [60]. These results imply that trained runners can dynamically self-optimize their running biomechanics in response to their physiological state. For novice runners, the difference between preferred and mathematically optimal stride fre- quencies is greater than for trained runners (8 vs. 3 %) [59] (Fig. 1). Therefore, generalizing the principle of an optimal stride length range to all runners should be done with caution, as self-optimization appears to be a physiological adaptation resulting from greater running experience. Similar to stride frequency and stride length, vertical oscillation can be altered. Acute interventions have shown that increasing vertical oscillation leads to increases in _VO2 [41, 63]. Additionally, vertical oscillation increases when running to exhaustion. However, vertical oscillation chan- ges are minimal and increases in _VO2 are large [64, 65], meaning several other physiological and biomechanical factors contribute to increases in _VO2 during fatigue [66, 67]. Furthermore, decreases in vertical oscillation have been shown when individuals run barefoot and their RE improves [50], probably due to a smaller vertical dis- placement during stance [52]. Yet, it must be noted that shoe mass and other biomechanical changes associated with barefoot running also influence such RE improve- ments (see Sect. 3.4). Another study has shown that decreasing vertical oscillation can slightly improve RE, but only if the absolute height of the body’s center of mass (CoM) is not changed [68]. Collectively, these results imply that reducing the magnitude of vertical displacement should be encouraged. It is possible that reducing vertical displacement improves RE by reducing the metabolic cost associated with supporting body weight, as a smaller ver- tical impulse would be produced [69]. Additionally, it could make a runner more mechanically efficient, as a low displacement of the body’s CoM produces a low mechan- ical energy cost, since the body is performing less work against gravity [70]. Notwithstanding these encouraging results, findings show that female runners have a lower vertical oscillation than their male counterparts, but findings are conflicting regarding whether females are more or less economical than males [13, 16, 71]. Eriksson et al. [72] demonstrated that vertical oscillation could be successfully lowered using visual and auditory feedback, and that runners found it more natural to change vertical oscillation than step fre- quency. However, to date, only one study has assessed the effect of specifically decreasing a runner’s vertical oscil- lation. This means research has not tried to manipulate vertical oscillation, in a similar manner to stride frequency and stride length, to determine whether runners have an optimal magnitude of vertical oscillation or whether run- ners would simply benefit from lowering their vertical oscillation to improve RE. The time the foot spends in contact with the ground has equivocal results regarding its association with RE. Several studies have failed to find any relationship between ground contact time and RE [9, 42, 73, 74], whilst some have observed a better RE to be associated with longer contact times [75, 76] and others have found the opposite to be true [11, 77]. It is suggested that short ground-contact times incur a high metabolic cost because faster force production is required, meaning metabolically expensive fast twitch muscle fibers are recruited [78, 79]. Conversely, long ground-contact times may incur a high metabolic cost because force is produced slowly, meaning longer braking phases when runners undergo deceleration [77]. Whilst both arguments appear plausible, it has been argued that being able to reduce the amount of speed lost during ground contact is the most important aspect rather than the time in contact with it [77, 80–82]. Combining this with evidence that individuals can produce shorter ground- contact times, but similar deceleration times and RE when forefoot striking compared with rearfoot striking [83], suggests that the time spent decelerating may influence RE. Another factor that may affect the body’s deceleration is how far ahead of the body the foot strikes the ground. Fig. 1 Individual differences (selected-optimal) in stride frequency (a) and running cost (b) for novice (left) and trained runners (right) on day 1 (black bars) and day 2 (grey bars). 2 test days were used to assess the reliability of measures and were separated by at least 48 h. RCopt running cost of optimal stride frequency, RCsel running cost of self-selected stride frequency, SFopt optimal stride frequency based on minimal running cost, SFsel self-selected stride frequency. X denotes that optimal stride frequency and, consequently, optimal running cost could not be established in these five trials. Reproduced from de Ruiter et al. [59] by permission of Taylor & Francis Ltd, http://www.tandfonline.com Modifiable Biomechanical Factors Affecting Running Economy 795 123 Evidence from step rate manipulation investigations and global gait re-training studies instructing runners to adopt a Pose running method, suggest that significantly decreasing the horizontal distance between the body’s CoM and foot at initial ground contact reduces peak braking and propulsive forces [84, 85] and braking impulses (less speed lost) applied by the runner [86, 87]. Yet, both performance and RE were unaffected during the gait re-training [85], potentially because too many running biomechanics were modified at once. Others have suggested that a runner’s optimal stride frequency is a trade-off between the meta- bolic cost associated with braking impulses and those associated with swinging the leg [87]. Further work into this braking strategy is required to understand the impli- cations for RE. Increasing the absolute time spent in the swing phase has been associated with better RE by several researchers [11, 42, 43]. However, others have failed to find any relationship between the two [43, 71]. Findings from Barnes et al. [43] suggest that sex also affects this rela- tionship; however, this has not been corroborated by others [11, 71]. It is conceivable that a longer absolute swing time means runners spend a smaller proportion of the gait cycle in contact with the ground, which is believed to be the metabolically expensive phase of the cycle. It is important to note that the swing and ground contact times will impact the stride frequency and stride length of a runner, and it is perhaps the relationship between all these aspects that should be considered. 3.1 Lower Limb Kinematic Factors Various kinematic parameters have been identified as being associated with better RE in cross-comparison studies; greater plantarflexion velocity [75], greater horizontal heel velocity at initial contact [75], greater maximal thigh extension angle with the vertical [75], greater knee flexion during stance [42], reduced knee range of motion during stance [88], reduced peak hip flexion during braking [88], slower knee flexion velocity during swing [42, 71], greater dorsiflexion and faster dorsiflexion velocity during stance [71], slower dorsiflexion velocity during stance [88], and greater shank angle at initial contact [42]. Intra-individual comparisons have identified later occurrence of peak dor- siflexion, slower eversion velocity at initial contact, and less knee flexion at push-off as being associated with improved RE [34]. One of the few kinematic variables to have strong support from both cross- and intra-individual comparisons as being beneficial for RE is a less extended leg at toe-off [34, 42, 50, 71, 75, 89]. Evidence has shown that this can be achieved through less plantarflexion and/or less knee extension as the runner pushes off the ground (Fig. 2). Hip extension is also likely to contribute, but studies have typically focused on the knee and ankle angles. Less leg extension could produce greater propulsive force, as identified by Moore et al. [34], by potentially allowing the leg extensor muscles to operate at a more favorable position on the force–length curve and higher gear ratios (GRF moment arm to muscle–tendon moment arm) being obtained. Both strategies could maximize force production [90, 91]. Additionally, less leg extension would reduce the amount of flexion needed during swing by already being partially flexed and potentially reduce the leg’s moment of inertia, lowering the energy required to flex the leg during the swing phase. Previous research has shown that reduced leg moment of inertia lowers the leg’s mechanical demand during the swing phase, as well as the metabolic demand, of walking [92]. Therefore, it is conceivable that a similar relationship exists when running, but this needs investigating. Another kinematic during the push-off phase that has been associated with better RE is stride angle, which is defined as the angle of the parable tangent of the CoM at toe-off [11, 93, 94]. Larger stride angles appear to be beneficial for lowering _VO2 and can be achieved by either increasing swing time or decreasing stride length. How- ever, the system (Optojump Next) used by each study [11, 93, 94] only tracks the foot during ground contact and not the CoM. Therefore, only inferences can be made 165º 157º -25º -19º Pre Post Fig. 2 Differences in knee angle (top) and ankle angle (bottom) at toe-off between pre and post measurements. Pre refers to baseline running biomechanics and post refers to running biomechanics after 10 weeks of running whereby beginner runners improved their running economy and altered their running technique. Reproduced from Moore et al. [34], with permission 796 I. S. Moore 123 regarding the trajectory angle of the CoM and other pos- sible kinematic changes. Future work focusing on the push- off phase should assess CoM trajectory in relation to kinematics and kinetics, as increasing swing time would also increase the vertical displacement of the CoM based on previous calculations [11, 93, 94] and observations [95]. Crucially, research suggests that increases to these spa- tiotemporal parameters appear to have contradictory rela- tionships with RE [11, 41–43, 63]. Foot strike patterns have been implicated as a modifiable factor affecting RE [96], with some researchers arguing that the most economical strike pattern is forefoot striking, even when RE is not assessed [97–99]. However, empirical evidence refutes this claim. Findings shows no difference in RE between rearfoot and forefoot striking at slow (B3 ms-1) [51, 83, 100, 101], medium (3.1–3.9 ms-1) [83, 100, 101], and fast speeds (C4.0 ms-1) [83, 100] or rearfoot and midfoot striking at medium speeds [76]. However, others have shown rearfoot striking to be more economical than midfoot striking at slow running speeds [102]. Interestingly, habitual forefoot strikers can change to a rearfoot strike without detrimental consequences to RE, while an imposed forefoot strike in habitual rearfoot strikers produces worse RE at slow and medium speeds [100]. Based on the current literature, foot strike appears to have a negligible effect upon RE, with only habitual rearfoot strikers likely to experience a worsening of RE by switching foot strike patterns. 3.2 Kinetic Factors Early research reported that RE was proportional to the vertical component of GRF (e.g., force required to support body weight) and was termed the ‘cost of generating force’ hypothesis [79, 103, 104]. However, later investigations have used a task-by-task approach to partition RE into individual biomechanical tasks [105]. Such work has demonstrated that braking (decelerating the body) and propulsive (accelerating the body) forces also incur meta- bolic costs [105]. Typically, the three components of GRF (anterior-posterior, medial–lateral, and vertical) have been independently assessed, with evidence suggesting lower vertical impact force [42], lower peak medial–lateral force [42, 75], lower anterior–posterior braking force [73], and higher anterior–posterior propulsive force [34] are eco- nomical. However, numerous studies have also failed to identify similar associations between RE and individual GRF components [26, 73, 74]. To understand the metabolic costs incurred during running Arellano and Kram [106] advocate using a syn- ergistic approach, rather than the ‘cost of generating force’ hypothesis or task-by-task approach. Using this approach, the vertical force (supporting body weight) and forward propulsive force (accelerating the body) incur the greatest metabolic cost (Fig. 3). However, very few biomechanical studies have utilized such an approach. Storen et al. [74] demonstrated that it could be usefully applied as they found significant relationships between the summation of peak vertical and anterior–posterior forces and 3-km performance (r = -0.71) and RE (r = - 0.66). Their findings show that lower forces were asso- ciated with a better running performance and RE. Addi- tionally, Moore et al. [107] reported near perfect alignment of the angle of the resultant GRF vector (all three components) with the angle of the longitudinal leg axis vector during propulsion when novice runners improved their RE. This change in alignment was asso- ciated with a change in RE (rs = 0.88), suggesting that minimizing the muscular effort of generating force during propulsion is beneficial to RE [107]. Associations have also been found between GRF impulses and RE, with lower braking [87], total, and net vertical impulses related to a better RE [9]. However, this finding is not consistent in the literature [77]. Through collectively considering the deceleration and acceleration (anterior–posterior) impulses, a runner’s change in momentum can be determined. One pilot study has utilized this technique, but reported similar changes in momentum pre and post a 10-week running program that improved RE [107]. The authors suggested that such a short-term training program might not have been long enough to induce modifications in momentum [107]. It is also conceivable that a synergistic approach should be applied to momentum and speed lost during braking. The magnitude of the GRF during running has a linear relationship with the body’s vertical displacement [108], suggesting the leg acts like a spring during ground contact [44]. Therefore, use of the spring-mass model to describe the body’s bounce during the support phase of running has been widespread. The springs’ stiffness is the ratio of deformation (vertical displacement) to the force applied to it (vertical GRF) and therefore represents the stiffness of the whole body’s musculoskeletal system [109]. Leg stiffness represents the ratio of maximal vertical force to maximal vertical leg spring compression [110]. Greater leg stiffness has been associated with a better RE [44], whilst fatiguing runs to volitional exhaustion have led to reduc- tions in leg stiffness [64, 65]. Furthermore, alterations to extrinsic factors, such as increasing surface compliance, can lead to decreases in leg stiffness, resulting in a worse RE [111]. Running in minimalist footwear can increase leg stiffness and improve RE compared with traditional and cushioned footwear [112, 113]. Interestingly, leg stiffness is predominately associated with ground-contact time rather than step frequency [114]. Thus, to try and increase leg stiffness, runners are advised to focus on shortening Modifiable Biomechanical Factors Affecting Running Economy 797 123 ground-contact time rather than increasing step frequency. Such an approach may be beneficial for RE improvements. As leg stiffness represents the stiffness of the whole musculoskeletal system, several factors relating to stiffness are unmodifiable, such as muscle crossbridges and tendon stiffness. However, neuromuscular activation is a modifi- able characteristic that can modulate stiffness. 3.3 Neuromuscular Factors The preactivation of muscles prior to ground contact, ter- med muscle tuning, is believed to increase muscle–tendon stiffness [77], potentially enhance muscular force genera- tion via the stretch–shortening cycle (SSC) [115], and affect leg geometry at initial ground contact [116–118]. Nigg et al. [119] studied the effect of shoe midsole char- acteristics on RE and preactivation, and, whilst no overall shoe-dependent changes were found in either variable, systematic individual changes in vastus medialis preacti- vation were evident. Runners who produced higher vastus medialis preactivation independent of shoe condition also had a higher _VO2 [119]. However, given the small changes in RE (\2 %) the differences may be due to test–retest measurement error and are unlikely to represent a mean- ingful change in RE [120]. Greater muscular activity of the lower limbs has been reported as a potential mechanism behind increasing _VO2 and thus is seen as detrimental to RE [73]. The intuitive link between muscle activity and RE stems from muscles needing to utilize oxygen to activate, and thereby, control movement patterns and stabilize joints. Therefore, greater muscle activation, as typically measured using surface electromyography (EMG), is thought to require a higher _VO2 and lead to a worsening of RE. In line with this, findings have shown a higher activation of the gastrocne- mius during propulsion and of the biceps femoris during braking and propulsion to be associated with higher _VO2 [73]. Additionally, Abe et al. [45] found an increase in _VO2 during a prolonged run was associated with a decrease in the ratio of eccentric–concentric vastus lateralis activity. This change in eccentric–concentric ratio was due to an increase in activity during propulsion (concentric phase). Collectively, these findings suggest that needing to utilize greater muscle activation to propel the runner forwards, possibly due to a reduced efficiency of the SSC, is detri- mental to RE. Bourdin et al. [121] support this notion, as they found lower eccentric–concentric ratios of vastus lateralis activity were associated with a higher energetic cost of running. Importantly, however, this relationship was more promi- nent when inter-individual differences were being assessed and was weaker when intra-individual differences were considered. Sinclair et al. [88] also found a higher activity of the vastus medialis to be related to a worse RE when comparing different runners. Conversely, Pinnington and colleagues [122, 123] have suggested that intra-individual increases in _VO2 associated with running on sand com- pared with on a firm surface are partially due to increased activation of the quadriceps and hamstrings muscles involved in greater hip and knee range of motion. Fig. 3 The a cost of generating force, b individual task-by-task, and c synergistic task-by-task approach partition the net metabolic cost of human running into its biomechanical constituents. The cost of generating force approach and the individual task-by-task approach both illustrate that body weight support is the primary determinant of the net metabolic cost of human running. In the individual task-by-task approach, forward propulsion represents the second largest determi- nant. The individual task-by-task approach leads to an overestimation, while the synergistic task-by-task approach suggests that the synergistic tasks of body weight support and forward propulsion are the primary determinants of the net metabolic cost of human running. Note that leg swing and lateral balance exact a relatively small net metabolic cost. If we sum all the biomechanical tasks, the synergistic task-by-task approach accounts for 89 % of the net metabolic cost of human running, leaving 11 % of unexplained metabolic cost, and the cost of generating force accounts for 80 %, leaving 20 % of unexplained metabolic cost. Reproduced from Arellano and Kram [106], with permission from Oxford University Press 798 I. S. Moore 123 However, as _VO2 and EMG data were collected in separate studies, causal interpretations should be made with caution. Larger intra- and inter-individual variations in lower limb muscle activity duration and timing of peak activation have been reported in novice compared with experienced run- ners [124], suggesting that greater running exposure may alter neuromuscular control. However, longitudinal inves- tigations are needed to confirm this. Conflicting results have also been reported for the role of muscular coactivation in relation to RE [46–48], whereby muscular coactivation is defined as the simulta- neous activation of two muscles. Heise et al. [47] found a negative relationship between RE and the coactivation of the rectus femoris and gastrocnemius, suggesting coacti- vation of biarticular muscles is economical, whereas Moore et al. [48] reported a positive relationship. Furthermore, muscular coactivation of the proximal agonist–antagonist leg muscles, rectus femoris and biceps femoris, has also been shown to have a positive association with RE, meaning such coactivation is detrimental to RE [46, 48]. Coactivation of the proximal thigh antagonist–agonist muscles occurs during the loading phase of stance as the knee flexes. Without such coactivation, it is likely that the leg would collapse [125], but essentially the muscles are performing opposing movements. Using two muscles to control such a movement would therefore incur a greater metabolic cost than using one muscle, potentially decreasing the efficiency of the SSC. Investigations into the effect of orthotics on muscular activation during ground contact and RE have provided inconsistent findings. Kelly et al. [126] reported that alter- ations to muscular activity when wearing orthotics during a 1-h run were not accompanied by changes in RE. Con- trastingly, Burke and Papuga [127] observed improvements in RE when runners ran in custom-made orthotics rather than shoe-fitted insoles, yet there were no changes in lower limb muscular activity. However, the mass of the different orthotics used by Burke and Papuga [127], and the potential effect the orthotics had on running biomechanics, were not assessed and may have influenced their findings. 3.4 Shoe–Surface Interaction Factors There is a general consensus that running in traditional running trainers is detrimental to RE compared with run- ning barefoot or in lightweight, minimalist trainers, due to the added shoe mass [49–52, 128, 129]. A recent meta- analysis suggested that a shoe mass (per pair) of less than 440 g does not affect RE, but a shoe mass greater than 440 g negatively affects RE [129]. However, when shoe mass is taken into account, evidence regarding footwear effects on RE is equivocal due to different methodologies used. Mathematically correcting for different footwear mass when expressing _VO2 in relative terms supports the above statement that running in traditional trainers is detrimental to RE compared with barefoot or minimalist footwear running [50]. However, strapping weights equal to the mass of a shoe to participants’ feet results in either similar RE [52] or worse RE when barefoot compared with shod [49]. One reason for this discrepancy is that mathe- matically adjusting _VO2 technically adjusts the whole body’s mass rather than the foot’s mass and does not take into account the decrease in lower limb moment of inertia. When the foot’s CoM is altered (weights strapped to the top of foot) _VO2 is worse when barefoot [49], but when the foot’s CoM is unchanged (weights evenly distributed on the foot), _VO2 is similar between barefoot and shod con- ditions [52]. Therefore, changes to lower limb moment of inertia, and not just shoe mass, appear to affect RE. Findings from Scholz et al. [130] support this notion by showing greater lower limb moment of inertia was asso- ciated with higher _VO2. Other shoe characteristics, such as stiffness [131], comfort [132], and cushioning [133], are likely to effect RE and thus, may have also contributed to the equivocal findings regarding footwear effects on RE when shoe mass is taken into account. However, if shoe mass is not adjusted for, running barefoot or in lightweight, minimalist trainers improves RE compared with traditional running trainers (shoe mass [440 g). Changing footwear can also change the level of cush- ioning underfoot. Frederick et al. [134] proposed the ‘cost of cushioning’ hypothesis, stating that actively cushioning the body whilst running may incur a metabolic cost. Therefore, shoes with limited cushioning or no cushioning (such as being barefoot) would result in an individual having to actively cushion the body using the lower limb muscles [117] and lead to an increase in _VO2. Some evi- dence to support this claim is provided by Franz et al. [49], who found that running in shoes with increasing mass had a lower metabolic power demand than running barefoot with increasing mass strapped to their feet. These results therefore show that running without cushioning has a higher metabolic demand than running with cushioning, even when added shoe mass is similar. However, results from Divert et al. [52] suggest it may be mechanical energy that is increased rather than _VO2 when barefoot. This means that barefoot running leads to mechanical efficiency improvements due to greater work being done for the same _VO2 compared with shod running. Further, it appears there is an ‘optimal’ level of surface cushioning for good RE. When running barefoot on a treadmill, 10 mm of surface cushioning was more benefi- cial for RE than no surface cushioning and 20 mm of surface cushioning [53]. When considering natural running Modifiable Biomechanical Factors Affecting Running Economy 799 123 terrain, Pinnington and Dawson [122] found running on grass elicited a lower _VO2 than running on sand. This is likely due to the damping effects of sand, leading to an increase in mechanical work done during stance [135]. Therefore, a firmer surface that returns the energy it absorbs will benefit a runner’s RE. Moreover, a firm sur- face with reduced stiffness, and thus greater compliance, will return more energy due to the surface’s elastic rebound and improve RE [111]. This theory can also be applied to running shoes, as Worobets et al. [54] showed that a softer shoe, which was more compliant and lost less energy during impact than a control shoe, improved RE. Additionally, shoes with a high forefoot bending elasticity can increase propulsive force and reduce contact time and gastrocnemius muscle activation during slow (\3 ms-1), but not medium (3.1–3.9 ms-1), running speeds compared with a flexible forefoot region [136]. Such shoes may therefore improve RE due to enhancing propulsion; however, no _VO2 data were gathered during the study, so direct associations cannot be made. Consequently, it is likely that a medium level of cushioning, that returns energy, is beneficial for RE compared with the shoe–surface cushioning being too compliant or too hard. Footwear (or lack of) can also affect running biome- chanics. Several modifications to running biomechanics may potentially benefit RE, whilst others may not. For example, in comparison with shod running, barefoot running can shorten ground contact time and stride length [49–52, 128, 137–140], increase knee flexion at initial contact [139], increase leg stiffness [52, 139, 141, 142], decrease vertical oscillation [50, 138], increase propulsive force [143], and reduce plantarflexion at toe-off [50, 139]. The most com- monly cited change when running barefoot is a more ante- rior foot strike pattern brought about by a flatter foot, such as switching from a rearfoot to a forefoot strike pattern [50, 98, 137, 139, 140, 142, 144]. However, evidence shows many confounding variables affect foot strike, including speed [97, 145], surface stiffness [146], stride length [50], and famil- iarization with barefoot running [147]. Therefore, footwear (or lack of) alone cannot explain changes in foot strike. Based on the several findings above, it can be suggested that acute exposure to running barefoot may be beneficial for RE, especially if performed on a surface with a medium level of cushioning. Aside from acute exposure, the effect of individual adaptations due to short- and long-term exposure to barefoot running on RE and running biomechanics is currently unknown. 3.5 Trunk and Upper Limb Biomechanical Factors The relationship between RE and trunk and upper body biomechanics has received limited research attention compared with lower limb biomechanics. Swinging the arms during running plays an important role as it con- tributes to vertical oscillation [55, 56]; counters vertical angular momentum of the lower limbs [148]; and mini- mizes head, shoulder, and torso rotation [149, 150]. Eliminating arm swing by placing the hands on top of the head can be detrimental to RE [41, 149], whilst placing the hands behind the back or across the chest has provided inconsistent findings [41, 56, 63, 149, 150]. However, there is no evidence to suggest that individuals can alter arm kinematics to improve RE and thus, running performance. Therefore, based on current evidence, individuals are encouraged to maintain their natural arm swing whilst running. Suppressing arm swing can alter several lower limb biomechanics and kinetics. For example, restraining the arms behind the back and across the chest decreases peak vertical force, increases peak hip and knee flexion angles during stance, and reduces knee adduction during stance [151]. These biomechanical changes appear to be due to the loss of arm motion rather than the body’s CoM moving position [151], suggesting that arm motion plays an integral role in an individual’s running technique. Further, the greater knee flexion and reduced peak vertical force observed when arm swing is suppressed suggests that leg stiffness decreases, which may explain the change in RE found in some studies [41, 56, 149]. However, currently, the relationship between leg stiffness and arm motion during running is unknown. It has been suggested that a forward trunk lean during running improves RE [58], based on findings from Wil- liams and Cavanagh [42]. Yet, a forward lean has also been implicated as detrimental to RE. Hausswirth et al. [57] compared the _VO2 during a marathon run (2 h, 15 min) with that during a 45-minute run and found the marathon run had a higher _VO2 and greater forward trunk lean. However, this finding should be interpreted in light of the other modifications to running biomechanics when com- paring the marathon run with the 45-min run, such as the 13 % shorter stride lengths. It is possible that shortening the stride lengths by this amount incurred the highest _VO2 rather than the forward lean. Additionally, the biome- chanical changes could be due to muscular fatigue resulting from the difference in running time between the two con- ditions (1 h, 30 min), meaning muscular fatigue could have led to increases in _VO2. For women runners, breast kinematics also have the potential to affect RE and running biomechanics. Evidence shows that breast kinematics can affect running kinetics [152], trunk lean via changes in breast support [153], and lower limb biomechanics, in particular knee angle and step length [154]. These findings imply there may be alterations 800 I. S. Moore 123 to RE, particularly if the changes in step length are greater than 3 % of the preferred step length. Further work that simultaneously assesses RE, breast kinematics, breast support, and lower limb biomechanics is warranted to assess whether there is a direct association between the measures. 4 Simultaneously Modifying Running Biomechanics and Running Economy Through Training Short- and mid-term training interventions (3–12 weeks) have been conducted to assess relationships between run- ning biomechanics and RE. But to date, no long-term training interventions have been performed. Early inter- ventions primarily focused on spatiotemporal factors, with Morgan et al. [155] showing that trained runners with uneconomical stride lengths could be retrained using audio-feedback over 3 weeks to produce mathematically derived optimal stride lengths and improved RE. In con- trast, Messier and Cirillo [95] failed to find improvements in RE when using verbal and visual feedback for 5 weeks to change specific running biomechanics, such as longer stride lengths, shorter ground-contact time, and reduced vertical oscillation. However, optimal stride length was not mathematically determined prior to the intervention, meaning suitable procedures were not used and several running biomechanics either were not modified or, in the case of vertical oscillation, actually increased after the intervention. Bailey and Messier [156] also found that if runners were able to freely choose their stride length over 7 weeks, there was no change in RE. Similarly, if runners were restricted to their initial freely chosen stride length over 7 weeks, RE was unaffected [156]. Interventions concerned with instructing runners to retrain their running biomechanics towards a specific glo- bal running technique, such as Pose, Chi and midstance to midstance running, has generally resulted in either no improvement in RE [62, 85] or a worsening of RE [157]. Whilst these techniques are often advocated as efficient forms of running [157, 158], and all the interventions led to modified running biomechanics, currently there appears to be no evidence to substantiate the claims that they benefit RE. It is conceivable that the failure of global running techniques to improve RE is because they are not targeting the right running biomechanics or because they are trying to change too many at the same time. Running gait retraining has also focused on reducing injury risk [159–162], but only one study has assessed the effect of such retraining on RE as well [163]. Clansey et al. [163] provided trained runners with gait re-training using real-time visual feedback over 3 weeks to modify impact- loading variables associated with tibial stress fracture risk. Runners reduced peak tibial acceleration and loading rates without changing RE. Thus, gait re-training to reduce injury risk can be performed without necessarily affecting running performance. This is possibly because the gait alterations were predominantly during the impact phase and have minimal effect on RE, as individuals increased plantarflexion at initial contact and exhibited a more anterior foot strike. Moore et al. [34] reported that novice runners could self- optimize their running gait over 10 weeks of running training, with 94 % of the variance of change in RE explained by less knee extension at toe-off, a later occur- rence of peak dorsiflexion, and slower eversion velocity at initial contact. Furthermore, trained, habitually shod run- ners can improve their RE when running in minimalist footwear after a 4-week intervention exposing them to running in minimalist footwear [96]. Although very few running gait parameters were assessed by Warne and Warrington [96], runners did exhibit a more anterior foot strike when more economical. Whilst collectively these results support short-term biomechanical self-optimization to running training, a previous investigation failed to find RE improvements and biomechanical changes in trained runners after 6 weeks of running [36]. Consequently, novice runners may be more responsive to self-optimiza- tion in the short-term than trained runners; however pro- viding trained runners with a novel stimulus, such as different footwear, can lead to short-term self-optimization. Thus, self-optimization is a physiological adaptation to running acquired through greater experience of the stimu- lus. For trained runners, the majority of this physiological adaptation may have already occurred. A summary of how training interventions have affected RE is presented in Fig. 4. 5 Is there an Economical Running Technique? Based on the literature, several modifiable factors that can potentially improve RE have been identified, as well as factors that have conflicting or limited findings regarding their relationship with RE (Table 1). From this summary, it is clear that biomechanics during ground contact play an important role. Furthermore, evidence shows that many of the running biomechanics identified occur during propul- sion, suggesting that this phase has the strongest direct links with RE. However, theoretical deceleration strategies, such as short braking times and minimizing the speed lost during braking, may translate to more economical strate- gies in the propulsive phase and mediate the relationship between propulsion and RE. Therefore, utilizing the prin- ciples of the SSC is encouraged. Modifiable Biomechanical Factors Affecting Running Economy 801 123 Considering the empirical evidence, one economical running strategy could be aiming to shorten ground-contact times whilst maintaining stride frequency, which may facilitate greater leg stiffness, larger stride angles, and longer swing times. However, such a strategy may increase vertical oscillation and encourage greater muscular activity during propulsion. Another strategy could involve aligning the resultant GRF more closely with the leg axis during propulsion. This may help minimize muscular activity and agonist–antagonist coactivation and could be produced as a result of reducing leg extension at toe-off. An experienced runner’s naturally chosen stride length is self-optimized to within 3 % of the mathematically derived optimal. Deviating between naturally chosen and mathematically optimal will only have a negligible effect on RE. However, novice runners have not acquired the running experience necessary to self-optimize as effec- tively. Therefore, a short-term running training program for novice runners can lead to running biomechanics being modified to benefit RE. However, long-term running training has seldom been investigated. Consequently, longitudinal investigations assessing the development of running biomechanics in both novice runners and experi- enced runners are required to better understand self-opti- mization for RE improvements. Notwithstanding the identified modifiable factors affecting RE, prescribing an economical way of running has its limitations based on the current empirical evidence. The majority of studies have used cross-comparison methodologies or are restricted to one running population. Additionally, it is evident from the numerous studies ana- lyzing intra-individual changes that group differences, which statistically hold more power, provide limited con- clusions of modifications to running biomechanics [88, 119, 164]. Also, very few studies have assessed running biomechanics during the swing phase, even though current findings indicate the position of the CoM and leg during this phase may be crucial to conserving energy and reducing _VO2. Exploring running biomechanics during swing and the interaction with stance-phase biomechanics is recommended in future work. Furthermore, the role of unmodifiable factors and how they may interact with Was the training programme < 13 weeks? Did the training programme focus on changing specific running biomechanics? Was stride length or stride frequency manipulated? Were participants exposed to a novel stimulus? RE unchanged Optimal stride length/ frequency not mathematically determined [95] NO No studies NO RE improved Optimal stride length/ frequency mathematically determined [155] Was the training programme focused on achieving a global running technique? NO RE unchanged Tibial acceleration reduced [163] YES NO RE improved Novice runners increased running volume [34] RE improved Recreational runners exposed to novel footwear [96] RE worsened Pose running technique [157] RE unchanged Pose and midstance to midstance running technique [62, 85] RE worsened Recreational runners increased running volume [36] YES YES YES NO YES Fig. 4 Summary of the training programs that have simultaneously measured running economy and running biomechanics. The effect on running economy is denoted in bold. RE running economy 802 I. S. Moore 123 modifiable factors is an area requiring investigation. For example, Cavanagh and Williams [40] reported that indi- viduals with long legs had a larger increase in _VO2 when shortening their strides compared with lengthening them. In contrast, individuals with shorter legs had a larger increase in _VO2 when lengthening their stride than when shortening it. Biomechanical case studies of economical runners have not been published, but could provide interesting findings if an in-depth runner profile was provided. Such a profile would need to encompass factors such as running biome- chanics, anatomical structures, functional capacity (e.g., flexibility, muscular strength, and stiffness), shoe degra- dation, injury history, and training protocols [165]. Whilst only the former have been discussed here, the interaction between an individual’s anatomical structures—such as foot morphology, leg length, and tendon stiffness—and their running biomechanics is likely to be influential upon RE. This is certainly a direction for future research to pursue, as it could identify novel relationships and inter- actions that inform larger, cohort studies. 6 Conclusion One of the determining factors of running performance is RE. Modifiable running biomechanical factors that affect RE include spatiotemporal factors, lower limb kinematics, kinetics, neuromuscular factors, shoe–surface interac- tions, and trunk and upper limb biomechanics. Several intrinsic factors that appear to benefit RE are a self- selected stride length with a 3 % shorter stride length range, lower vertical oscillation, greater leg stiffness, low lower limb moment of inertia, alignment of the GRF and leg axis vectors, less leg extension at toe-off, larger stride angles, maintaining arm swing, low muscle activation during propulsion, and low antagonist–agonist thigh coactivation. In regards to extrinsic factors, better RE was found to be associated with a firm, compliant shoe-surface interaction and being barefoot or wearing lightweight shoes. Other modifiable biomechanical factors, such as ground contact time, impact force, anterior–posterior forces, trunk lean, lower limb biarticular muscle coacti- vation, and orthotics, presented inconsistent relationships with RE. Collectively, the evidence shows that many of the running biomechanics identified occur during propulsion, suggesting that this phase has the strongest direct links with RE. However, recurring methodological problems exist within the literature, such as cross-com- parisons, assessing variables in isolation, and acute to short-term interventions. Further, intra-individual differ- ences due to unmodifiable factors limit the findings of cross-comparisons, and future research should look to investigate longitudinal interventions and assess runners on an individual basis. Consequently, recommending an economical running technique should be approached with caution. Directions for further work within the field should focus on a synergistic approach to assessing kinetics as well as integrated approaches combining _VO2, kinematics, kinetics, and neuromuscular and anatomical aspects to increase our understanding of economical running technique. Table 1 Modifiable intrinsic and extrinsic running biomechanics and their effect on running economy Evidenced effect on RE Intrinsic Extrinsic Spatiotemporal Kinetics Kinematics Neuromuscular Beneficial Self-selected stride length (minus 3 %) Greater leg stiffness Less leg extension at toe-off Low muscle activation during propulsion Firm, compliant shoe-surface interaction Low vertical oscillation Alignment of GRF and leg axis during propulsion Large stride angle Low agonist– antagonist coactivation Barefoot or lightweight shoes (\440 g) Low lower limb moment of inertia Maintain arm swing Conflicting Ground contact time Impact force Trunk lean Biarticular coactivation Orthotics Swing time Anterior–posterior forces Limited or unknown Horizontal distance between the foot and CoM at initial contact Impulses Swing phase Vastus medialis preactivation Braking/deceleration time Foot-strike pattern Speed lost during ground contact Breast kinematics CoM centre of mass, GRF ground reaction force, RE running economy Modifiable Biomechanical Factors Affecting Running Economy 803 123 Acknowledgments The author would like to thank Professor Andrew Jones and Dr. Victoria Stiles for their critical comments on earlier versions of the manuscript. Compliance with Ethical Standards Funding No sources of funding were used to assist in the prepa- ration of this article. Conflicts of interest Isabel Moore declares she has no conflicts of interest relevant to the content of this review. Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http:// creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. References 1. Billat VL, Demarle A, Slawinski J, et al. Physical and training characteristics of top-class marathon runners. Med Sci Sports Exerc. 2001;33:2089–97. 2. Foster C. VO2 max and training indices as determinants of competitive running performance. J Sports Sci. 1983;1:13–22. 3. Farrell PA, Wilmore JH, Coyle EF, et al. Plasma lactate accu- mulation and distance running performance. Med Sci Sports. 1979;11:338–44. 4. Tanaka K, Matsuura Y. Marathon performance, anaerobic threshold, and onset of blood lactate accumulation. J Appl Physiol Respir Environ Exerc Physiol. 1984;57:640–3. 5. Conley DL, Krahenbuhl GS. Running economy and distance running performance of highly trained athletes. Med Sci Sports Exerc. 1980;12:357–60. 6. Morgan DW, Baldini FD, Martin PE, et al. Ten kilometer per- formance and predicted velocity at VO2max among well-trained male runners. Med Sci Sports Exerc. 1989;21:78–83. 7. Pollock ML. Submaximal and maximal working capacity of elite distance runners. Part I: cardiorespiratory aspects. Ann N Y Acad Sci. 1977;301:310–22. 8. Daniels JT. A physiologist’s view of running economy. Med Sci Sports Exerc. 1985;17:332–8. 9. Heise GD, Martin PE. Are variations in running economy in humans associated with ground reaction force characteristics? Eur J Appl Physiol. 2001;84:438–42. 10. Costill DL, Thomason H, Roberts E. Fractional utilization of the aerobic capacity during distance running. Med Sci Sports. 1973;5:248–52. 11. Santos-Concejero J, Tam N, Granados C, et al. Stride angle as a novel indicator of running economy in well-trained runners. J Strength Cond Res. 2014;28:1889–95. 12. Jones AM. The physiology of the world record holder for the women’s marathon. Int J Sports Sci Coach. 2006;1:101–16. 13. Bransford DR, Howley ET. Oxygen cost of running in trained and untrained men and women. Med Sci Sports Exerc. 1977;9:41–4. 14. Morgan DW, Bransford DR, Costill DL, et al. Variation in the aerobic demand of running among trained and untrained sub- jects. Med Sci Sports Exerc. 1995;27:404–9. 15. Daniels J, Daniels N. Running economy of elite male and elite female runners. Med Sci Sports Exerc. 1992;24:483–9. 16. Helgerud J, Storen O, Hoff J. Are there differences in running economy at different velocities for well-trained distance run- ners? Eur J Appl Physiol. 2010;108:1099–105. 17. Barnes KR, Kilding AE. Running economy: measurement, norms, and determining factors. Sports Med Open. 2015;1:8. 18. Saunders PU, Pyne DB, Telford RD, et al. Factors affecting running economy in trained distance runners. Sports Med. 2004;34:465–85. 19. Barnes KR, Kilding AE. Strategies to improve running econ- omy. Sports Med. 2015;45:37–56. 20. Jones AM, Carter H. The effect of endurance training on parameters of aerobic fitness. Sports Med. 2000;29:373–86. 21. Saunders PU, Telford RD, Pyne DB, et al. Short-term plyo- metric training improves running economy in highly trained middle and long distance runners. J Strength Cond Res. 2006;20:947–54. 22. Spurrs RW, Murphy AJ, Watsford ML. The effect of plyometric training on distance running performance. Eur J Appl Physiol. 2003;89:1–7. 23. Turner AM, Owings M, Schwane JA. Improvement in running economy after 6 weeks of plyometric training. J Strength Cond Res. 2003;17:60–7. 24. Barnes KR, Hopkins WG, McGuigan MR, et al. Effects of resistance training on running economy and cross-country per- formance. Med Sci Sports Exerc. 2013;45:2322–31. 25. Guglielmo LGA, Greco CC, Denadai BS. Effects of strength training on running economy. Int J Sports Med. 2009;30:27–32. 26. Paavolainen L, Hakkinen K, Hamalainen I, et al. Explosive- strength training improves 5-km running time by improving running economy and muscle power. J Appl Physiol. 1999;86:1527–33. 27. Støren Ø, Helgerud J, Støa EM, et al. Maximal strength training improves running economy in distance runners. Med Sci Sports Exerc. 2008;40:1087–92. 28. Cheng CF, Cheng KH, Lee YM, et al. Improvement in running economy after 8 weeks of whole-body vibration training. J Strength Cond Res. 2012;26:3349–57. 29. Barnes KR, Hopkins WG, McGuigan MR, et al. Effects of different uphill interval-training programs on running economy and performance. Int J Sports Physiol Perf. 2013;8:639–47. 30. Denadai BS, Ortiz MJ, Greco CC, et al. Interval training at 95% and 100% of the velocity at VO2 max: effects on aerobic physiological indexes and running performance. Appl Physiol Nutr Metab. 2006;31:737–43. 31. Franch J, Madsen K, Djurhuus MS, et al. Improved running economy following intensified training correlates with reduced ventilatory demands. Med Sci Sports Exerc. 1998;30:1250–6. 32. Saunders PU, Pyne DB, Gore CJ. Endurance training at altitude. High Alt Med Biol. 2009;10:135–48. 33. Saunders PU, Telford RD, Pyne DB, et al. Improved running economy in elite runners after 20 days of simulated moderate- altitude exposure. J Appl Physiol. 2004;96:931–7. 34. Moore IS, Jones AM, Dixon SJ. Mechanisms for improved running economy in beginner runners. Med Sci Sports Exerc. 2012;44:1756–63. 35. Beneke R, Hutler M. The effect of training on running economy and performance in recreational athletes. Med Sci Sports Exerc. 2005;37:1794–9. 36. Lake MJ, Cavanagh PR. Six weeks of training does not change running mechanics or improve running economy. Med Sci Sports Exerc. 1996;28:860–9. 37. Ramsbottom R, Williams C, Fleming N, et al. Training induced physiological and metabolic changes associated with improve- ments in running performance. Br J Sports Med. 1989;23:171–6. 38. Ferrauti A, Bergermann M, Fernandez-Fernandez J. Effects of a concurrent strength and endurance training on running 804 I. S. Moore 123 performance and running economy in recreational marathon runners. J Strength Cond Res. 2010;24:2770–8. 39. Roschel H, Barroso R, Tricoli V, et al. Effects of strength training associated with whole body vibration training on run- ning economy and vertical stiffness. J Strength Cond Res. 2015;29:2215–20. 40. Cavanagh PR, Williams KR. The effect of stride length variation on oxygen uptake during distance running. Med Sci Sports Exerc. 1982;14:30–5. 41. Tseh W, Caputo JL, Morgan DW. Influence of gait manipulation on running economy in female distance runners. J Sports Sci Med. 2008;7:91–5. 42. Williams KR, Cavanagh PR. Relationship between distance running mechanics, running economy, and performance. J Appl Physiol. 1987;63:1236–45. 43. Barnes KR, McGuigan MR, Kilding AE. Lower-body determi- nants of running economy in male and female distance runners. J Strength Cond Res. 2014;28:1289–97. 44. Dalleau G, Belli A, Bourdin M, et al. The spring-mass model and the energy cost of treadmill running. Eur J Appl Physiol. 1998;77:257–63. 45. Abe D, Muraki S, Yanagawa K, et al. Changes in EMG char- acteristics and metabolic energy cost during 90-min prolonged running. Gait Posture. 2007;26:607–10. 46. Frost G, Dowling J, Dyson K, et al. Cocontraction in three age groups of children during treadmill locomotion. J Electromyogr Kinesiol. 1997;7:179–86. 47. Heise G, Shinohara M, Binks L. Biarticular leg muscles and links to running economy. Int J Sports Med. 2008;29:688–91. 48. Moore IS, Jones AM, Dixon SJ. Relationship between metabolic cost and muscular coactivation across running speeds. J Sci Med Sport. 2013;17:671–6. 49. Franz JR, Wierzbinski CM, Kram R. Metabolic cost of running barefoot versus shod: is lighter better? Med Sci Sports Exerc. 2012;44:1519–25. 50. Moore IS, Jones AM, Dixon SJ. The pursuit of improved run- ning performance: can changes in cushioning and somatosen- sory feedback influence running economy and injury risk? Footwear Sci. 2014;6:1–11. 51. Perl DP, Daoud AI, Lieberman DE. Effects of footwear and strike type on running economy. Med Sci Sports Exerc. 2012;44:1335–43. 52. Divert C, Mornieux G, Freychat P, et al. Barefoot-shod running differences: shoe or mass effect? Int J Sports Med. 2008;29:512–8. 53. Tung KD, Franz JR, Kram R. A test of the metabolic cost of cushioning hypothesis during unshod and shod running. Med Sci Sports Exerc. 2014;46:324–9. 54. Worobets J, Wannop JW, Tomaras E, et al. Softer and more resilient running shoe cushioning properties enhance running economy. Footwear Sci. 2014;6:147–53. 55. Arellano CJ, Kram R. The energetic cost of maintaining lateral balance during human running. J Appl Physiol. 2012;112:427–34. 56. Arellano CJ, Kram R. The effects of step width and arm swing on energetic cost and lateral balance during running. J Biomech. 2011;44:1291–5. 57. Hausswirth C, Bigard AX, Guezennec CY. Relationships between running mechanics and energy cost of running at the end of a triathlon and a marathon. Int J Sports Med. 1997;18:330–9. 58. Anderson T. Biomechanics and running economy. Sports Med. 1996;22:76–89. 59. de Ruiter CJ, Verdijk PW, Werker W, et al. Stride frequency in relation to oxygen consumption in experienced and novice runners. Eur J Sport Sci. 2013;14:251–8. 60. Hunter I, Smith GA. Preferred and optimal stride frequency, stiffness and economy: changes with fatigue during a 1-h high- intensity run. Eur J Appl Physiol. 2007;100:653–61. 61. Connick MJ, Li FX. Changes in timing of muscle contractions and running economy with altered stride pattern during running. Gait Posture. 2014;39:634–7. 62. Craighead DH, Lehecka N, King DL. A novel running mechanic’s class changes kinematics but not running economy. J Strength Cond Res. 2014;28:3137–45. 63. Egbuonu ME, Cavanagh PR, Miller TA. Degradation of running economy through changes in running mechanics. Med Sci Sports Exerc. 1990;22:S17. 64. Fourchet F, Girard O, Kelly L, et al. Changes in leg spring behaviour, plantar loading and foot mobility magnitude induced by an exhaustive treadmill run in adolescent middle-distance runners. J Sci Med Sport. 2014;18:199–203. 65. Hayes PR, Caplan N. Leg stiffness decreases during a run to exhaustion at the speed at VO2max. Eur J Sport Sci. 2014;14:556–62. 66. McKenna MJ, Hargreaves M. Resolving fatigue mechanisms determining exercise performance: integrative physiology at its finest! J Appl Physiol. 2008;104:286–7. 67. Levine BD. _V(O(2), max): what do we know, and what do we still need to know? J Physiol. 2008;586:25–34. 68. Halvorsen K, Eriksson M, Gullstrand L. Acute effects of reducing vertical displacement and step frequency on running economy. J Strength Cond Res. 2012;26:2065–70. 69. Teunissen LP, Grabowski A, Kram R. Effects of independently altering body weight and body mass on the metabolic cost of running. J Exp Biol. 2007;210:4418–27. 70. Slawinski JS, Billat VL. Difference in mechanical and energy cost between highly, well, and nontrained runners. Med Sci Sports Exerc. 2004;36:1440–6. 71. Williams KR, Cavanagh PR, Ziff JL. Biomechanical studies of elite female distance runners. Int J Sports Med. 1987;8(Suppl 2):107–18. 72. Eriksson M, Halvorsen KA, Gullstrand L. Immediate effect of visual and auditory feedback to control the running mechanics of well-trained athletes. J Sports Sci. 2011;29:253–62. 73. Kyrolainen H, Belli A, Komi PV. Biomechanical factors affecting running economy. Med Sci Sports Exerc. 2001;33:1330–7. 74. Storen O, Helgerud J, Hoff J. Running stride peak forces inversely determine running economy in elite runners. J Strength Cond Res. 2011;25:117–23. 75. Williams KR, Cavanagh PR. Biomechanical correlates with running economy in elite distance runners. Proceedings of the North American Congress on Biomechanics. Montreal; 1986. p. 287–8. 76. Di Michele R, Merni F. The concurrent effects of strike pattern and ground-contact time on running economy. J Sci Med Sport. 2013;17:414–8. 77. Nummela AT, Keranen T, Mikkelsson LO. Factors related to top running speed and economy. Int J Sports Med. 2007;28:655–61. 78. Roberts TJ, Kram R, Weyand PG, et al. Energetics of bipedal running. I. Metabolic cost of generating force. J Exp Biol. 1998;201:2745–51. 79. Kram R, Taylor CR. Energetics of running: a new perspective. Nature. 1990;346:265–7. 80. Kaneko M, Ito A, Fuchimoto T, et al. Influence of running speed on the mechanical efficiency of sprinters and distance runners. In: Winter DA, Norman RW, Wells RP, Heyes KC, Patla AE, editors. Biomechanics IX-B. Champaign: Human Kinetics; 1985. p. 307–12. 81. Nummela AT, Paavolainen L, Sharwood KA, et al. Neuromus- cular factors determining 5 km running performance and Modifiable Biomechanical Factors Affecting Running Economy 805 123 running economy in well-trained athletes. Eur J Appl Physiol. 2006;97:1–8. 82. Kong PW, De Heer H. Anthropometric, gait and strength characteristics of Kenyan distance runners. J Sports Sci Med. 2008;7:499–504. 83. Ardigo LP, Lafortuna C, Minetti AE, et al. Metabolic and mechanical aspects of foot landing type, forefoot and rearfoot strike, in human running. Acta Physiol Scand. 1995;155:17–22. 84. Arendse RE, Noakes TD, Azevedo LB, et al. Reduced eccentric loading of the knee with the pose running method. Med Sci Sports Exerc. 2004;36:272–7. 85. Fletcher G, Bartlett R, Romanov N, et al. Pose method tech- nique improves running performance without economy changes. Int J Sports Sci Coach. 2008;3:365–80. 86. Heiderscheit BC, Chumanov ES, Michalski MP, et al. Effects of step rate manipulation on joint mechanics during running. Med Sci Sports Exerc. 2011;43:296–302. 87. Lieberman DE, Warrener AG, Wang J, et al. Effects of stride frequency and foot position at landing on braking force, hip torque, impact peak force and the metabolic cost of running in humans. J Exp Biol. 2015;218:3406–14. 88. Sinclair J, Taylor PJ, Edmundson CJ, et al. The influence of footwear kinetic, kinematic and electromyographical parameters on the energy requirements of steady state running. Mov Sport Sci. 2013;80:39–49. 89. Cavanagh PR, Pollock ML, Landa J. A biomechanical com- parison of elite and good distance runners. Ann N Y Acad Sci. 1977;301:328–45. 90. Rassier DE, MacIntosh BR, Herzog W. Length dependence of active force production in skeletal muscle. J Appl Physiol. 1999;86:1445–57. 91. Carrier D, Heglund N, Earls K. Variable gearing during loco- motion in the human musculoskeletal system. Science. 1994;265:651–3. 92. Royer TD, Martin PE. Manipulations of leg mass and moment of inertia: effects on energy cost of walking. Med Sci Sports Exerc. 2005;37:649–56. 93. Santos-Concejero J, Tam N, Granados C, et al. Interaction effects of stride angle and strike pattern on running economy. Int J Sports Med. 2014;35:1118–23. 94. Santos-Concejero J, Granados C, Irazusta J, et al. Differences in ground contact time explain the less efficient running economy in North African runners. Biol Sport. 2013;30:181–7. 95. Messier SP, Cirillo KJ. Effects of a verbal and visual feedback system on running technique, perceived exertion and running economy in female novice runners. J Sports Sci. 1989;7:113–26. 96. Warne JP, Warrington GD. Four-week habituation to simulated barefoot running improves running economy when compared with shod running. Scand J Med Sci Sports. 2014;24:563–8. 97. Hasegawa H, Yamauchi T, Kraemer WJ. Foot strike patterns of runners at the 15-km point during an elite-level half marathon. J Strength Cond Res. 2007;21:888–93. 98. Lieberman DE, Venkadesan M, Werbel WA, et al. Foot strike patterns and collision forces in habitually barefoot versus shod runners. Nature. 2010;463:531–5. 99. Jenkins DW, Cauthon DJ. Barefoot running claims and contro- versies: a review of the literature. J Am Podiatr Med Assoc. 2011;101:231–46. 100. Gruber AH, Umberger BR, Braun B, et al. Economy and rate of carbohydrate oxidation during running with rearfoot and fore- foot strike patterns. J Appl Physiol. 2013;115:194–201. 101. Cunningham CB, Schilling N, Anders C, et al. The influence of foot posture on the cost of transport in humans. J Exp Biol. 2010;213:790–7. 102. Ogueta-Alday A, Rodriguez-Marroyo JA, Garcia-Lopez J. Rearfoot striking runners are more economical than midfoot strikers. Med Sci Sports Exerc. 2014;46:580–5. 103. Farley CT, McMahon TA. Energetics of walking and running: insights from simulated reduced-gravity experiments. J Appl Physiol. 1992;73:2709–12. 104. Taylor CR, Heglund NC, McMahon TA, et al. Energetic cost of generating muscular force during running: a comparison of large and small animals. J Exp Biol. 1980;86:9–18. 105. Chang YH, Kram R. Metabolic cost of generating horizontal forces during human running. J Appl Physiol. 1999;86:1657–62. 106. Arellano CJ, Kram R. Partitioning the metabolic cost of human running: a task-by-task approach. Integr Comp Biol. 2014;54:1084–98. 107. Moore IS, Jones AM, Dixon SJ. Reduced oxygen cost of running is related to alignment of the resultant GRF and leg axis vector: a pilot study. Scand J Med Sci Sports. 2015. doi:10.1111/sms.12514. 108. Cavagna GA, Franzetti P, Heglund NC, et al. The determinants of the step frequency in running, trotting and hopping in man and other vertebrates. J Physiol. 1988;399:81–92. 109. Butler RJ, Crowell HP 3rd, Davis IM. Lower extremity stiffness: implications for performance and injury. Clin Biomech. 2003;18:511–7. 110. Divert C, Baur H, Mornieux G, et al. Stiffness adaptations in shod running. J Appl Biomech. 2005;21:311–21. 111. Kerdok AE, Biewener AA, McMahon TA, et al. Energetics and mechanics of human running on surfaces of different stiffnesses. J Appl Physiol. 2002;92:469–78. 112. Lussiana T, Fabre N, Hebert-Losier K, et al. Effect of slope and footwear on running economy and kinematics. Scand J Med Sci Sports. 2013;23:246–53. 113. Lussiana T, He´bert-Losier K, Mourot L. Effect of minimal shoes and slope on vertical and leg stiffness during running. J Sport Health Sci. 2015;4:195–202. 114. Morin JB, Samozino P, Zameziati K, et al. Effects of altered stride frequency and contact time on leg-spring behavior in human running. J Biomech. 2007;40:3341–8. 115. Ruan M, Li L. Approach run increases preactivation and eccentric phases muscle activity during drop jumps from dif- ferent drop heights. J Electromyogr Kinesiol. 2010;20:932–8. 116. Muller R, Grimmer S, Blickhan R. Running on uneven ground: leg adjustments by muscle pre-activation control. Hum Mov Sci. 2010;29:299–310. 117. Boyer KA, Nigg BM. Muscle activity in the leg is tuned in response to impact force characteristics. J Biomech. 2004;37:1583–8. 118. Boyer KA, Nigg BM. Changes in muscle activity in response to different impact forces affect soft tissue compartment mechan- ical properties. J Biomech Eng. 2007;129:594–602. 119. Nigg BM, Stefanyshyn DJ, Cole G, et al. The effect of material characteristics of shoe soles on muscle activiation and energy aspects during running. J Biomech. 2003;36:569–75. 120. Saunders PU, Pyne DB, Telford RD, et al. Reliability and variability of running economy in elite distance runners. Med Sci Sports Exerc. 2004;36:1972–6. 121. Bourdin M, Belli A, Arsac LM, et al. Effect of vertical loading on energy cost and kinematics of running in trained male sub- jects. J Appl Physiol. 1995;79:2078–85. 122. Pinnington HC, Dawson B. The energy cost of running on grass compared to soft dry beach sand. J Sci Med Sport. 2001;4:416–30. 123. Pinnington HC, Lloyd DG, Besier TF, et al. Kinematic and electromyography analysis of submaximal differences running on a firm surface compared with soft, dry sand. Eur J Appl Physiol. 2005;94:242–53. 806 I. S. Moore 123 124. Chapman AR, Vicenzino B, Blanch P, et al. Is running less skilled in triathletes than runners matched for running training history? Med Sci Sports Exerc. 2008;40:557–65. 125. Montgomery WH 3rd, Pink M, Perry J. Electromyographic analysis of hip and knee musculature during running. Am J Sports Med. 1994;22:272–8. 126. Kelly LA, Girard O, Racinais S. Effect of orthoses on changes in neuromuscular control and eerobic cost of a 1-h run. Med Sci Sports Exerc. 2011;43:2335–43. 127. Burke JR, Papuga MO. Effects of foot orthotics on running economy: methodological considerations. J Manip Physiol Ther. 2012;35:327–36. 128. Burkett LN, Kohrt WM, Buchbinder R. Effects of shoes and foot orthotics on VO2 and selected frontal plane knee kinematics. Med Sci Sports Exerc. 1985;17:158–63. 129. Fuller JT, Bellenger CR, Thewlis D, et al. The effect of footwear on running performance and running economy in distance run- ners. Sports Med. 2014;45:411–22. 130. Scholz MN, Bobbert MF, Van Soest AJ, et al. Running biomechanics: shorter heels, better economy. J Exp Biol. 2008;211:3266–71. 131. Roy J-PR, Stefanyshyn DJ. Shoe midsole longitudinal bending stiffness and running economy, joint energy, and EMG. Med Sci Sports Exerc. 2006;38:562–9. 132. Luo G, Stergiou P, Worobets J, et al. Improved footwear com- fort reduces oxygen consumption during running. Footwear Sci. 2009;1:25–9. 133. Frederick EC, Howley ET, Powers S. Lower oxygen demands of running in soft-soled shoes. Res Q Exerc Sport. 1986;57:174–7. 134. Frederick EC, Clarke TE, Larsen JL, et al. The effect of shoe cushioning on the oxygen demands on running. In: Nigg BM, Kerr BA, editors. Biomechanical aspects of sports shoes and playing surfaces. Calgary: University of Calgary; 1983. p. 107–14. 135. Lejeune TM, Willems PA, Heglund NC. Mechanics and ener- getics of human locomotion on sand. J Exp Biol. 1998;201:2071–80. 136. Chen C-H, Tu K-H, Liu C, et al. Effects of forefoot bending elasticity of running shoes on gait and running performance. Hum Mov Sci. 2014;38:163–72. 137. McCallion C, Donne B, Fleming N, et al. Acute differences in foot strike and spatiotemporal variables for shod, barefoot or minimalist male runners. J Sports Sci Med. 2014;13:280–6. 138. Vincent HK, Montero C, Conrad BP, et al. Metabolic responses of running shod and barefoot in mid-forefoot runners. J Sports Med Phys Fit. 2014;54:447–55. 139. De Wit B, De Clercq D, Aerts P. Biomechanical analysis of the stance phase during barefoot and shod running. J Biomech. 2000;33:269–78. 140. Moore IS, Pitt W, Nunns M, et al. Effects of a seven-week minimalist footwear transition programme on footstrike modality, pressure variables and loading rates. Footwear Sci. 2014;7:17–29. 141. Chambon N, Delattre N, Gueguen N, et al. Is midsole thickness a key parameter for the running pattern? Gait Posture. 2014;40:58–63. 142. Squadrone R, Gallozzi C. Biomechanical and physiological comparison of barefoot and two shod conditions in experienced barefoot runners. J Sports Med Phys Fitness. 2009;49:6–13. 143. Paquette MR, Zhang S, Baumgartner LD. Acute effects of barefoot, minimal shoes and running shoes on lower limb mechanics in rear and forefoot strike runners. Footwear Sci. 2013;5:9–18. 144. Hamill J, Russell E, Gruber A, et al. Impact characteristics in shod and barefoot running. Footwear Sci. 2011;3:33–40. 145. Breine B, Malcolm P, Frederick EC, et al. Relationship between running apeed and initial foot contact patterns. Med Sci Sports Exerc. 2014;46:1595–603. doi:10.249/MSS. 0000000000000267. 146. Allison HG, JuliaFreedman S, Peter B, et al. Footfall patterns during barefoot running on harder and softer surfaces. Footwear Sci. 2013;5:39–44. 147. Moore IS, Dixon SJ. Changes in sagittal plane kinematics with treadmill familiarization to barefoot running. J Appl Biomech. 2014;30:626–31. 148. Hinrichs RN. Upper extremity function in running II: angular momentum considerations. J Appl Biomech. 1987;3:242–63. 149. Arellano CJ, Kram R. The metabolic cost of human running: is swinging the arms worth it? J Exp Biol. 2014;217:2456–61. 150. Pontzer H, Holloway JH 4th, Raichlen DA, et al. Control and function of arm swing in human walking and running. J Exp Biol. 2009;212:523–34. 151. Miller RH, Caldwell GE, Van Emmerik RE, et al. Ground reaction forces and lower extremity kinematics when running with suppressed arm swing. J Biomech Eng. 2009;131:124502. 152. White JL, Scurr JC, Smith NA. The effect of breast support on kinetics during overground running performance. Ergonomics. 2009;52:492–8. 153. Milligan A, Mills C, Corbett J, et al. The influence of breast support on torso, pelvis and arm kinematics during a five kilo- meter treadmill run. Hum Mov Sci. 2015;42:246–60. 154. Milligan A. The effect of breast support on running biome- chanics. PhD thesis. University of Portsmouth; 2013. http:// eprints.port.ac.uk/14846/. Accessed 10 Dec 2015. 155. Morgan DW, Martin P, Craib M, et al. Effect of step length optimization on the aerobic demand of running. J Appl Physiol. 1994;77:245–51. 156. Bailey SP, Messier SP. Variations in stride length and running economy in male novice runners subsequent to a seven-week training program. Int J Sports Med. 1991;12:299–304. 157. Dallam GM, Wilber RL, Jadelis K, et al. Effect of a global alteration of running technique on kinematics and economy. J Sports Sci. 2005;23:757–64. 158. Romanov N, Fletcher G. Runners do not push off the ground but fall forwards via a gravitational torque. Sports Biomech. 2007;6:434–52. 159. Crowell HP, Davis IS. Gait retraining to reduce lower extremity loading in runners. Clin Biomech. 2011;26:78–83. 160. Davis IS, Crowell HP, Fellin RE, et al. Reduced impact loading following gait retraining over a 6-month period. Gait Posture. 2009;30:S4–5. 161. Diebal AR, Gregory R, Alitz C, et al. Forefoot running improves pain and disability associated with chronic exertional compart- ment syndrome. Am J Physiol. 2012;40:1060–7. 162. Willy RW, Scholz JP, Davis IS. Mirror gait retraining for the treatment of patellofemoral pain in female runners. Clin Bio- mech. 2012;27:1045–51. 163. Clansey AC, Hanlon M, Wallace ES, et al. Influence of tibial shock feedback training on impact loading and running econ- omy. Med Sci Sports Exerc. 2014;46:973–81. 164. Willwacher S, Ko¨nig M, Braunstein B, et al. The gearing function of running shoe longitudinal bending stiffness. Gait Posture. 2014;40:386–90. 165. Williams KR. Biomechanical factors contributing to marathon race success. Sports Med. 2007;37:420–3. Modifiable Biomechanical Factors Affecting Running Economy 807 123

Is There an Economical Running Technique? A Review of Modifiable Biomechanical Factors Affecting Running Economy.

[]

Moore, Isabel S

eng

PMC3826314

Hindawi Publishing Corporation The Scientific World Journal Volume 2013, Article ID 680326, 4 pages http://dx.doi.org/10.1155/2013/680326 Research Article Oxygen Uptake in Maximal Effort Constant Rate and Interval Running Daniel Pratt, Brendan J. O’Brien, and Bradley Clark School of Health Sciences, University of Ballarat, Mt Helen, Ballarat, VIC 3353, Australia Correspondence should be addressed to Brendan J. O’Brien; b.obrien@ballarat.edu.au Received 30 June 2013; Accepted 30 July 2013 Academic Editors: N. Berretta and R. Inoue Copyright © 2013 Daniel Pratt et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. This study investigated differences in average ̇VO2 of maximal effort interval running to maximal effort constant rate running at lactate threshold matched for time. The average ̇VO2 and distance covered of 10 recreational male runners ( ̇VO2 max: 4158 ± 390 mL⋅min−1) were compared between a maximal effort constant-rate run at lactate threshold (CRLT), a maximal effort interval run (INT) consisting of 2 min at ̇VO2 max speed with 2 minutes at 50% of ̇VO2 repeated 5 times, and a run at the average speed sustained during the interval run (CR submax). Data are presented as mean and 95% confidence intervals. The average ̇VO2 for INT, 3451 (3269–3633) mL⋅min−1, 83% ̇VO2 max, was not significantly different to CRLT, 3464 (3285–3643) mL⋅min−1, 84% ̇VO2 max, but both were significantly higher than CR sub-max, 3464 (3285–3643) mL⋅min−1, 76% ̇VO2 max. The distance covered was significantly greater in CLRT, 4431 (4202–3731) metres, compared to INT and CR sub-max, 4070 (3831–4309) metres. The novel finding was that a 20-minute maximal effort constant rate run uses similar amounts of oxygen as a 20-minute maximal effort interval run despite the greater distance covered in the maximal effort constant-rate run. 1. Introduction The principal objective of endurance training is to evoke supracompensation in the physiological systems restraining the maximal sustainable competition speed. The physio- logical systems most noted for regulating the speed of an endurance runner are the convective supply of oxygen to the muscles and the rate at which oxygen can be metabolized in the muscles to resynthesize adenosine triphosphate (ATP) [1]. It is proposed that the training strategy that sustains the highest oxygen use ( ̇VO2) for the longest is the most effective strategy to improve running performance [2]. ̇VO2 is typically assessed by the minute rate of pulmonary oxygen uptake during running [3]. The training strategies used by athletes can be broadly classed into constant rate running or interval running, where interval involves higher speeds of running interspersed with slower “recovery” speeds. Interval running evokes a greater total ̇VO2 than constant rate running [4] when the average speed of the treatments is controlled. Additionally, Daussin et al. [5] revealed that interval training over several weeks’ results in greater gains in cycling performance, metabolic, and cardiorespiratory adaptation. However, any assumption interval training is superior to constant rate training may be erroneous and an artefact of the research design. Generally, the work (run speed or cycle wattage) completed in a specific time frame has been controlled in the experimental treatments to ensure that comparisons in training adaptation are not biased by differences in work of the training strategies. However, by controlling work in the constant rate to interval training, the sustainable constant rate speed/wattage is less than the maximal sustainable speed (i.e., the constant rate training is still “submaximal”). For example, in O’Brien et al.’s [4] investigation, it was reported that interval running used more oxygen than constant rate running; however, the participants performed the constant rate run at a speed equivalent to the interval running mean speed, estimated to be only 75% ̇VO2 max, which was most likely below the lactate thresh- old or the fastest speed able to be sustained continuously by the runner. Consequently, equalising speed or work of the interval or constant rate runs may mask the optimal training strategy for athletes, and for all practical purposes, 2 The Scientific World Journal matching maximal effort over a duration recommended to improve cardiorespiratory fitness is more appropriate. The ACSM currently recommends that 20 minutes of exercise is required to improve cardiorespiratory fitness [6]. Therefore, we aim to compare the total ̇VO2 of a maximal effort interval run to a maximal effort constant rate run, matched for time, 20 minutes. 2. Methods 2.1. General Design of Study. This study is a quantitative study with a crossover design where participants in ran- dom sequence completed constant rate and interval training treadmill running at their individual-perceived maximal effort speed to investigate which strategy results in greater pulmonary oxygen uptake per minute ( ̇VO2). 2.2. Participants. Ten “fit” males ( ̇VO2 max 4158 ± 390 mL⋅min−1) were tested through recruitment via the university and personal contacts. The participants were aged from 18 to 40 years old. 2.3. Experimental Protocol. Each participant included com- peted two preliminary running tests and three experimental runs which were compared. Preliminary test 1: an initial maximal treadmill test to establish ̇VO2 max and the speed at which it is achieved. Preliminary test 2: a 5 km run time trial to estimate the maximal constant-rate speed approximating lactate threshold. Experimental test 1: a maximal effort interval tread- mill run consisting of 5 × 2 minute intervals at the speed corresponding to ̇VO2 max (s ̇VO2 max) during the high periods and 5 × 2 minute intervals at 0.5 s ̇VO2 max. Experimental test 2: a maximal effort constant rate treadmill run at the highest velocity that could be sustainable speed over 20 minutes (constant rate approximating lactate threshold run). This was deter- mined from the speed calculated from a 5 km time trial performed on a public park. Experimental test 3: a constant rate treadmill run at a speed determined from the average speed of the interval protocol used in Experimental test 1. 2.4. Experimental Procedure. The initial preliminary test of ̇VO2 max and its corresponding speed was conducted in an exercise physiology laboratory. Prior to the ̇VO2 max test, participants were fitted with a two-way breathing valve (Hans Rudolph, USA), and expired air was collected into an online metabolic system (Moxus, USA) to analyse ̇VO2. The metabolic system was calibrated before each test using ambient air and gas of known composition. The ̇VO2 max test commenced at 9 km⋅h−1at a gradient of 1%, and treadmill speed was increased by 1 km⋅h−1 every 2 minutes until volitional exhaustion. ̇VO2 max was determined as the highest 60-second ̇VO2 value recorded during the test. Within a week of ̇VO2 max determination, the 5 km time trial test was performed on flat terrain at a public park. After the two preliminary tests, the participants com- pleted the interval and the constant rate runs on the exercise physiology laboratory treadmill on separate days in random sequence. The experimental runs were preceded by a stan- dardized 5-minute warm-up run on the treadmill at 60% of ̇VO2 max followed by 2-minute rest. To control the confound- ing variables of diet, hydration, and fatigue, the participants were asked to consume 8–10 g of carbohydrate per kg of body weight, drink adequate fluid to maintain hydration, and sleep a minimum of 7 hours the night prior to testing. During all experimental treadmill runs, expired air was collected for metabolic analysis as per the initial maximal test. The ̇VO2 was recorded continuously in 30-second segments during each 20-minute run to determine the average ̇VO2. To confirm if the runs were the highest sustainable perceived effort for 20 minutes, each participant initially ran at the speed determined from the preliminary tests. The constant- rate run at lactate threshold was initially attempted by all participants at the speed determined from the 5 km time trial performed at the public park. The interval run on the treadmill was initially attempted at the final treadmill speed from the ̇VO2 max test, with the recovery periods set at 50% of the final treadmill speed. If the participant completed the 20 minutes in either the interval or constant rate run at lactate threshold, they undertook the run on another day at a higher speed. If the participant could not complete the 20- minute run, they ran on another day at a lower speed. The increase or decrease in speed was subjectively determined by the participant to their projected perception of what they felt could be a maximal effort. Originally, we planned to alter increments or decrements in speed by 0.2 km⋅h−1, although it quickly became apparent that some individuals felt 0.2 km⋅h−1 changes would be too “easy” or “not enough,” so we decided it was more appropriate for the individual to determine their own speed adjustments to establish a maximal perceived effort. The number of runs to determine a maximal effort was capped at three attempts for ethical and time constraints. The fastest speed able to be sustained for 20 minutes by the participant was used in the statistical analysis. The mean final treadmill speed from the initial ̇VO2 max test was 16.1 km⋅h−1, and the mean final effort sustainable interval speed was 16.3/8.15 km⋅h−1. The mean time of the 5 km time was 14 km⋅h−1 although this was not tolerated well on the laboratory treadmill by the majority of participants, with the mean maximal effort speed being 13.4 km⋅h−1. 2.5. Statistical Analyses. Differences in average ̇VO2 and mean distance covered between the three run protocols were analysed using linear mixed models (LMMs), with “type” as a fixed effect. Two error covariance structures were tested—independence (zero covariance) and repeated mea- sures structures (compound symmetry—constant covari- ance between each pair of types). Models were compared The Scientific World Journal 3 Table 1: Mean average ̇VO2 (mL⋅min−1), ̇VO2/ ̇VO2 max (%), and distance covered (metres) for the three treatments with 95% confidence intervals. Interval Submaximal constant rate Constant rate at lactate threshold Mean average ̇VO2 3451 (3269, 3633)† 3141 (2969, 3314)∗∧ 3464 (3285, 3643)† ̇VO2/ ̇VO2max (%) 83 (79, 88)† 76 (72, 80)∗∧ 84 (80, 89)† Distance covered (metres) 4070 (3831, 4309) 4070 (3831, 4309) 4470 (4202, 4737)∗† ∗P < 0.05 versus respective value in the interval run. †𝑃 < 0.05 versus respective value in submaximal constant rate run. ∧P < 0.05 versus respective value in constant rate at lactate threshold run. using likelihood ratio tests, which confirmed the compound symmetry structure. Paired 𝑡-tests with Bonferroni correc- tion were conducted to determine the significance of pairwise differences. Assumptions of normality and homogeneous variance of errors were tested by graphical display and analysis of residuals and found to be normally distributed. Significance was assumed at the 5% level. All statistical analyses were carried out using SPSS Version 19. 3. Results The mean ̇VO2 of the three running protocols is presented in Table 1. The mean ̇VO2 and ̇VO2/ ̇VO2 max (%) were similar between the interval and constant rate at lactate threshold runs but were significantly greater in both maximal effort runs compared to submaximal constant rate run. The dis- tance covered during the constant rate at lactate threshold run was significantly greater (𝑃 < 0.05) than the distance covered during the maximal Interval and submaximal constant rate runs. 4. Discussion The purpose of this study is to elucidate whether constant- rate running has the potential to equal or exceed the oxygen uptake of maximal effort interval training by comparing the ̇VO2 between maximal interval and constant rate run efforts, matched for duration of running, 20 minutes. The major finding of this study is that interval running and constant- rate running use similar amounts of oxygen when performed at the maximal sustainable speed for an individual. Both maximal interval and the constant rate at lactate threshold run resulted in a significantly greater (𝑃 < 0.05) mean ̇VO2 consumption compared to the submaximal con- stant rate run (3451 and 3434 versus 3141 mL⋅min−1). This difference can be explained by the higher average relative intensity of the exercise of the maximal interval and the constant rate at lactate threshold runs compared to the submaximal constant rate run (83% and 84% versus 76% ̇VO2/ ̇VO2 max (%)). The similar oxygen requirement of both maximal running strategies challenges the assumption that interval training is a superior form of training to maximal effort constant rate training. Previous studies report interval training results in greater total ̇VO2 of a workout compared to constant-rate training [2, 4, 7, 8] and Daussin et al. [5] clearly showed physiological adaptations were superior after interval training. However, Billat et al. [2] and Demarie et al. [7] used a very high intensity for the constant rate run (approximately 92% of v ̇VO2 max) that did not allow exercise to be sustained for a duration from (eight to ten minutes) normally sustained in typical endurance athlete training (at least 20 minutes). On the other hand, the studies by O’Brien et al. [4] and Daussin et al. [5] performed the constant rate run at a submaximal intensity (72% ̇VO2 max and approximately at 60% ̇VO2 max, resp.) that does not drive ̇VO2 near ̇VO2 max. The significance of our finding is that when matched for duration, constant rate approximating lactate threshold training places similar aerobic “load” as maximal interval training and therefore may be equally effective in enhancing running performance. Future research is required to compare a constant rate at lactate threshold training versus maximal effort interval training performed over several weeks to determine if any has a superior outcome on time trial performance. Interestingly, the constant rate at lactate threshold running resulted in a significantly greater distance being covered than interval running (4470 versus 4070 m), despite using similar amounts of oxygen. Consequently, maximal effort constant-rate running is a more effective and more economic strategy to cover a set distance in 20 minutes. The most likely explanation of the greater oxygen use in interval running is the excess postoxygen consumption that accumulates after each of the 2 min high intensity efforts. The excess post oxygen consumption is attributable to a number of factors but most likely is consequential to greater need for phosphate creatine restoration [9] and sodium/potassium regulation associated with repeated high intensity efforts that have a high anaerobic reliance [10]. 4.1. Limitations. A limitation of this study was the determi- nation of maximal effort that was capped at three attempts for each of the interval and constant-rate at lactate threshold runs. In the ideal experimental model, we would have requested participants to report more frequently to the laboratory to pinpoint maximal effort more precisely (i.e., any further increase in treadmill speed would lead to failure to complete the 20-minute run). Our treadmills minimum increment capability is 0.1 km⋅h−1. However for logistical and ethical reasons, volunteers subjectively nominated the treadmill running speed they perceived approximated their personal maximal tolerable effort, with the knowledge the third and final effort was the last opportunity to determine a “maximal” effort. The initial speeds were based on the initial speeds they ran at, which were based on the 5 km time trial 4 The Scientific World Journal and final speed of the ̇VO2 max test. Unfortunately due to technical malfunction, blood lactate concentration changes during the incremental test to determine lactate threshold could not be analysed, although we believe the best gauge of maximal constant-rate effort is ultimately determined from actual time trial performance. Hence, 5 km was chosen as the time trial distance as it was estimated to be completed in approximately 20 minutes. The mean time of the 5 km time trial completed was 21 min and 24 seconds. 5. Conclusion The primary aim of this paper is to contribute to the knowledge of the most effective training regimens athletes should embrace to optimise improvements in 5 km run per- formance. It is acknowledged to address this question further research needs to compare the effects of training strategies over time. Our data indicates that constant-rate running at lactate threshold should be considered worthy of inclusion in investigations as it imposes an identical aerobic metabolic load as interval running over the duration of a time-matched training bout. Another interesting finding is that constant- rate running at lactate threshold allows more distance to be covered and is therefore a more economic training strategy if covering distance is the goal. 5.1. Practical Applications (i) The similar mean ̇VO2 between constant rate at lactate threshold and interval runs indicates that both train- ing strategies may be equally effective in stimulating physiological adaptation and enhancing run perfor- mance. (ii) Constant rate at lactate threshold running will allow athletes to cover 10% further distance in 20 minutes compared to interval running. Conflict of Interests The authors declare that they have no conflict of interests. References [1] D. R. Bassett Jr. and E. T. Howley, “Limiting factors for max- imum oxygen uptake and determinants of endurance perfor- mance,” Medicine and Science in Sports and Exercise, vol. 32, no. 1, pp. 70–84, 2000. [2] V. L. Billat, J. Slawinski, V. Bocquet et al., “Intermittent runs at the velocity associated with maximal oxygen uptake enables subjects to remain at maximal oxygen uptake for a longer time than intense but submaximal runs,” European Journal of Applied Physiology and Occupational Physiology, vol. 81, no. 3, pp. 188– 196, 2000. [3] T. Hale, “History of developments in sport and exercise phys- iology: A. V. Hill, maximal oxygen uptake, and oxygen debt,” Journal of Sports Sciences, vol. 26, no. 4, pp. 365–400, 2008. [4] B. J. O’Brien, J. Wibskov, W. L. Knez, C. D. Paton, and J. T. Harvey, “The effects of interval-exercise duration and intensity on oxygen consumption during treadmill running,” Journal of Science and Medicine in Sport, vol. 11, no. 3, pp. 287– 290, 2008. [5] F. N. Daussin, J. Zoll, S. P. Dufour et al., “Effect of interval versus continuous training on cardiorespiratory and mitochondrial functions: relationship to aerobic performance improvements in sedentary subjects,” American Journal of Physiology, vol. 295, no. 1, pp. R264–R272, 2008. [6] S. S. Morey, “ACSM revises guidelines for exercise to maintain fitness,” American Family Physician, vol. 59, no. 2, p. 473, 1999. [7] S. Demarie, J. P. Koralsztein, and V. Billat, “Time limit and time at ̇VO2max, during a continuous and an intermittent run,” Jour- nal of Sports Medicine and Physical Fitness, vol. 40, no. 2, pp. 96–102, 2000. [8] A. Zafeiridis, H. Sarivasiliou, K. Dipla, and I. S. Vrabas, “The effects of heavy continuous versus long and short intermittent aerobic exercise protocols on oxygen consumption, heart rate, and lactate responses in adolescents,” European Journal of Applied Physiology, vol. 110, no. 1, pp. 17–26, 2010. [9] G. Dupont and S. Berthoin, “Time spent at a high percentage of ̇VO2 max for short intermittent runs: active versus passive recovery,” Canadian Journal of Applied Physiology, vol. 29, pp. S3–S16, 2004. [10] W. McGarvey, R. Jones, and S. Petersen, “Excess post-exercise oxygen consumption following continuous and interval cycling exercise,” International Journal of Sport Nutrition and Exercise Metabolism, vol. 15, no. 1, pp. 28–37, 2005.

Oxygen uptake in maximal effort constant rate and interval running.

09-01-2013

Pratt, Daniel,O'Brien, Brendan J,Clark, Bradley

eng

PMC3989295

The Positive Effects of Priming Exercise on Oxygen Uptake Kinetics and High-Intensity Exercise Performance Are Not Magnified by a Fast-Start Pacing Strategy in Trained Cyclists Renato Aparecido Correˆa Carita´, Camila Coelho Greco*, Benedito Se´rgio Denadai Human Performance Laboratory, IB – UNESP, Rio Claro, Sa˜o Paulo, Brazil Abstract The purpose of this study was to determine both the independent and additive effects of prior heavy-intensity exercise and pacing strategies on the VO2 kinetics and performance during high-intensity exercise. Fourteen endurance cyclists (VO2max = 62.868.5 mL.kg21.min21) volunteered to participate in the present study with the following protocols: 1) incremental test to determine lactate threshold and VO2max; 2) four maximal constant-load tests to estimate critical power; 3) six bouts of exercise, using a fast-start (FS), even-start (ES) or slow-start (SS) pacing strategy, with and without a preceding heavy- intensity exercise session (i.e., 90% critical power). In all conditions, the subjects completed an all-out sprint during the final 60 s of the test as a measure of the performance. For the control condition, the mean response time was significantly shorter (p,0.001) for FS (2764 s) than for ES (3265 s) and SS (3266 s). After the prior exercise, the mean response time was not significantly different among the paced conditions (FS = 2465 s; ES = 2565 s; SS = 2665 s). The end-sprint performance (i.e., mean power output) was only improved (,3.2%, p,0.01) by prior exercise. Thus, in trained endurance cyclists, an FS pacing strategy does not magnify the positive effects of priming exercise on the overall VO2 kinetics and short-term high-intensity performance. Citation: Carita´ RAC, Greco CC, Denadai BS (2014) The Positive Effects of Priming Exercise on Oxygen Uptake Kinetics and High-Intensity Exercise Performance Are Not Magnified by a Fast-Start Pacing Strategy in Trained Cyclists. PLoS ONE 9(4): e95202. doi:10.1371/journal.pone.0095202 Editor: Maria` Alemany, University of Barcelona, Faculty of Biology, Spain Received February 11, 2014; Accepted March 24, 2014; Published April 16, 2014 Copyright:  2014 Carita´ et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This research was supported by grants from Conselho Nacional de Desenvolvimento Cientı´fico e Tecnolo´gico and Fundac¸a˜o de Amparo a` Pesquisa do Estado de Sa˜o Paulo. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: grecocc@rc.unesp.br Introduction Exercise intensity domains (i.e., moderate, heavy and severe) are defined according to the blood lactate and oxygen uptake (VO2) responses obtained during constant-work-rate exercise [1]. Critical power (CP – the asymptote of the power-time relationship) is considered the lower boundary of the severe-intensity domain [2]. Indeed, during constant-work-rate exercise performed within the severe domain, the VO2 rises inexorably (as the slow component of the VO2 kinetics increases) to the maximal oxygen uptake (VO2max). Exercise tolerance within the severe domain can be predicted and is defined by the curvature constant of the power– time relationship (W9) [3]. Several lines of evidence indicate that the interaction between VO2 kinetics, W’ and the attainment of VO2max can contribute to exercise intolerance during exercise performed in the severe-intensity domain [4]. Some interventions (e.g., pacing, priming exercise and nitrate supplementation) that are used to improve VO2 kinetics (i.e., t – the time taken to reach 63% of the increase in VO2 above baseline and/or the slow component of VO2 kinetics) can reduce the W’ utilization during the initial phase of exercise, improving performance [5] and exercise tolerance [6] during severe-intensity exercise. Pacing strategy (i.e., the pattern of the rate of energy expenditure) has important effects on exercise tolerance [7] and performance [5]. The self-selected pacing strategy adopted during a time trial is controlled by a complex regulatory system, in which integrated neural control regulates exercise intensity to prevent homeostatic disturbances that might cause injury [8]. Factors such as exercise modality, event duration and performance level can influence the self-selected pacing strategy [9,10]. Some studies have demonstrated that a fast-start pacing strategy has a positive effect on performance during sports events of up to approximately 2–3 min in duration [11,12]; in these events, energy is provided by both aerobic and anaerobic pathways [13]. In these conditions, the VO2 kinetics is significantly faster, sparing W’ utilization during the initial phase of exercise [5,7]. Interestingly, Jones et al. [7] found that the percentage reduction in the mean response time of VO2 was significantly correlated (r = 0.85, p,0.05) with the percentage improvement in exercise tolerance when a fast start was compared with an even-paced exercise. Warm-up exercise has been extensively performed by athletes before their participation in subsequent vigorous exercise. Indeed, priming exercise performed at heavy or severe intensities domain can improve exercise tolerance during severe-intensity exercise (submaximal and perimaximal) [6,14]. These positive alterations have been attributed, at least in part, to enhancement of the overall VO2 kinetics [6,15]. Gerbino et al. [16] and MacDonald et al. [17] demonstrated that prior heavy exercise accelerated the PLOS ONE | www.plosone.org 1 April 2014 | Volume 9 | Issue 4 | e95202 monoexponential kinetics (i.e., mean response time) during a second bout of heavy exercise performed 6 min after the first bout. Later, studies using a more comprehensive model (two or three components) to analyze VO2 kinetics [6,18] demonstrated that this overall acceleration could be attributed to the increased amplitude of the primary component and the reduced amplitude of the slow component, with the time constant of the primary component (i.e., t) remaining unaffected. A similar response (i.e., increased amplitude of the primary component and unchanged t) was found for exercise that was performed at perimaximal intensities (100%, 110% and 120% of VO2max) after prior heavy exercise [14]. Thus, the interventions discussed above (i.e., priming exercise and pacing) seem to have different effects on the VO2 kinetics during severe-intensity exercise. The mechanism that underpins these effects is unclear. However, it is possible that different factors contribute to the VO2 response profile under these conditions (pacing vs. priming exercise). Priming exercise seems to increase blood flow, oxygenation, oxidative enzyme activity and electro- myographic activity, thus accelerating the overall VO2 response to severe exercise [15]. A positive pacing strategy, which causes a higher initial rate of muscle ATP hydrolysis, can magnify the VO2 ‘‘error signal’’, i.e., the difference between the instantaneous supply and the required rates of oxidative phosphorylation [5]. The absolute rate at which VO2 increases after the onset of exercise is a positive function of the ‘‘error signal’’ [19]; therefore, a fast-start pacing strategy results in faster VO2 kinetics [5,7]. Given this scenario, it is possible that priming exercise can amplify the positive effect of a fast-start pacing strategy on VO2 kinetics and exercise tolerance/performance during high-intensity exer- cise. However, the possible additive effects of priming exercise and pacing strategy on these variables are unknown. The purpose of this study was to determine the independent and additive effects of prior heavy-intensity exercise and pacing strategies on VO2 kinetics and performance during high-intensity exercise. The following hypotheses were proposed: 1) A fast-start pacing strategy would shorten the mean response time, and increase peak power output and mean power output during short- term high-intensity exercise; and 2) Priming exercise would shorten the mean response time, and increase peak power output and mean power output during short-term high-intensity exercise irrespectively of the utilized pacing strategy. Materials and Methods Ethics statement The present study was approved by the Ethics Committee of the Biosciences Institute – Rio Claro of Sa˜o Paulo State University, and all subjects provided written informed consent prior to participation. The study was performed in accordance with the declaration of Helsinki. Subjects Fourteen endurance cyclists (2665 years; 7169 kg; 17568 cm) with at least 5 years of experience in the modality volunteered to participate in the present study; these athletes were competing in regional- to national-level meets. The subjects were familiar with the laboratory testing procedures, as they were previously involved in similar evaluations. The subjects were instructed to be fully rested and hydrated at least 3 h postprandially when reporting to the laboratory and to refrain from using caffeine-containing food or beverages, drugs, alcohol, cigarette, or any form of nicotine 24 h before testing. Each subject was tested in a climate-controlled (21–22uC) laboratory at the same time of day (62 h) to minimize the effects of diurnal biological variation. Experimental design The subjects were required to visit the laboratory on 11 different occasions, separated by at least 24 h, within a period of three weeks. The first visit to the laboratory was to undergo an incremental test to determine the lactate threshold, VO2max and the power output at VO2max (PVO2max). On the following four visits, the subjects underwent four constant-load tests (75%, 80%, 85% and 100% of PVO2max) to exhaustion, in random order, to determine the parameters of the power-duration relationship (i.e., CP and W9). The CP model was used to estimate the workload that would be expected to lead to exhaustion in 3 min (P3-min). From the 6th to the 11th visit, the subjects performed three different pacing strategies (fast start, even start, and slow start) with and without a preceding heavy-intensity exercise session. Incremental protocol The incremental protocol was performed on a cycle ergometer (Lode Excalibur Sport, Lode BC, Groningen, Netherlands) with the subjects pedaling at a constant self-selected pedal rate (between 70 and 90 rpm). The chosen pedal rate along with saddle and handle bar height and configuration was recorded and reproduced in subsequent tests. The initial power output was 120 W for 3 min and was then increased by 20 W every 3 min. Capillary blood samples were collected within the final 20 s of each stage for the determination of the blood lactate concentration ([La]). The [La] were determined (YSI 2300, Yellow Springs, Ohio, USA) immediately and the test was stopped when the [La] rose above 4 mM. Plots of [La] against the power output were provided by two independent reviewers, who determined the lactate threshold as the first sudden and sustained increase in blood lactate above resting concentrations [20]. After a rest period of 30 min, the participants performed a fast ramp test. The test began with an initial 5 min of cycling at 25 W below their previously determined lactate threshold, and the power was subsequently increased by 5 W every 12 s until voluntary exhaustion. The protocol was terminated when a drop of more than 5 rpm of their self-selected cadence occurred for more than 5 seconds despite strong verbal encouragement. VO2max was defined as the highest average 15-s VO2 value recorded during the incremental test. Pulmonary gas exchange was measured continuously using a breath-by-breath analyzer (Cosmed Quark PFTergo, Rome, Italy). Before each test, the O2 and CO2 analysis systems were calibrated using ambient air and a gas of known O2 and CO2 concentration according to the manufacturer’s instructions, while the gas analyzer turbine flowmeter was calibrated using a 3-L syringe. The heart rate was also monitored throughout the tests (Polar, Kempele, Finland). The PVO2max was defined as the power output at which VO2max occurred. The work rate that would require 50%D (work rate at the lactate threshold plus 50% of the difference between the work rate at the lactate threshold and VO2max) was subsequently calculated. Determination of the power–duration relationship The exercise protocol began with a 10 min warm-up at lactate threshold, followed by 5 min of rest prior to the commencement of the exhaustive trial [21]. Thereafter, the subjects exercised for 3 min at 20 W followed by a constant-workload test (75%, 80%, 85% and 100% of PVO2max) to voluntary exhaustion or until the subject could not maintain the required cadence (i.e., a cadence ,5 rpm of the preferred cadence) despite verbal encouragement [21]. These tests were conducted at the same cadence as the Priming Exercise, Pacing Strategy and Short-Term Performance PLOS ONE | www.plosone.org 2 April 2014 | Volume 9 | Issue 4 | e95202 incremental test. During these testing sessions, the participants were not informed of the imposed work rate, their performance times or their heart rate. The exercise tolerance (tlim) was measured to the nearest second. The three equivalents of the 2- parameter model [P = (W9/tlim)+CP; tlim = W9/(P-CP); W = CP?tlim+W9] were used to fit the data and estimate CP and W9 [22] using an iterative nonlinear regression procedure (Microcal Origin 7.5; Northampton, MA, USA) for each subject. The CP and W9 estimates from the 3 equations were compared to select the best fit using the model associated with the lowest standard error for CP (SEE) [23,24]. Experimental sessions The exercise protocol began with a 5 min warm-up at lactate threshold, followed by 7 min of rest. Thereafter, the subjects performed 3 min at 20 W before the experimental conditions. In the even-start (ES) condition, the athletes performed 2 min of constant-load exercise at P3-min, followed by a 1-min all-out exercise period. In the fast-start (FS) condition, the first 90 s of exercise was performed as the work rate was reduced linearly from 110% to 90% of P3-min, followed by a 1-min all-out exercise period. In the slow-start (SS) condition, the first 90 s of exercise was performed as the work rate was increased linearly from 90% to 110% of P3-min, followed by a 1-min all-out exercise period. The last 30 s of the FS and SS conditions was performed at P3- min [5]. During the first 2 min of exercise, a hyperbolic mode was used (fixed power), which was immediately changed to a linear mode (power dependent on the cadence) during the all-out exercise. These experimental conditions were performed in random order with and without previous exercise (Figure 1). 1-min all-out exercise Following the 2-min pacing exercises, the athletes performed 1 min of all-out exercise. They were required to reach the peak power as quickly as possible and to exert maximal effort during the whole test. Throughout the 1-min test, the athletes were given verbal encouragement but were not informed of time elapsed. The VO2 was measured breath-by-breath during the exercise, and the data were reduced to 15-s stationary averages. The resistance to pedaling was calculated using the preferred cadence obtained during the incremental test and the workload corresponding to 50%D: 1 min resis tan ce~50%D=preferred cadence2 ð1Þ The following performance data were obtained from the all-out test: peak power output, time to peak power output and mean power output. Prior exercise The prior exercise conditions involved participants performing 3 min of baseline cycling at 20 W, followed by a square-wave transition to a work rate requiring 90% CP (i.e., heavy-intensity exercise). At 6 min, the subjects were allowed to ‘‘spin down’’ against zero resistance for 1 min and then rested passively for 6 min before remounting the ergometer and pedaling for 3 min at 20 W. After this 3-min period, one of the three pacing conditions was immediately imposed as described above. One minute before and immediately after these exercise bouts, a fingertip capillary blood sample was taken to determine the blood [La]. The subjects repeated this process on separate days and in a randomized order until all experimental trials were completed. VO2 kinetics The breath-by-breath data from each exercise test were filtered manually to remove outlying breaths, which were defined as breaths deviating by more than four standard deviations from the preceding five breaths. The breath-by-breath data were subse- quently linearly interpolated to provide second-by-second values and aligned by time to the start of the exercise, and a nonlinear least squares algorithm was used to fit the data thereafter. A single- exponential model without a time delay and with a fitting window commencing at t = 0 s (equivalent to the mean response time) was used to characterize the kinetics of the overall VO2 response during initial phase (i.e., 90 s) of the different pacing strategies for all subjects. The following equation describes this model: VO2(t)~VO2baselinezA½1{e{(t=t) ð2Þ where VO2(t) represents the absolute VO2 at a given time t, VO2baseline represents the mean VO2 measured over the final 60 s of baseline pedaling, and A and t represent the amplitude and time constant, respectively, which describe the overall increase in VO2 above the baseline. The oxygen deficit was also calculated for the same time period (i.e., 90 s) by multiplying the mean response time and the DVO2. Statistical analysis The data are reported as the means 6SD. The normality of data was checked by the Shapiro-Wilk test. The data were analyzed using two-way ANOVA (prior exercise vs. pacing strategy), with Fisher’s LSD test where appropriate. For all statistics, the significance level was set at p#0.05. Results During the ramped incremental test, the subjects attained a peak work rate (i.e., PVO2max) of 411645 W, a VO2max of 4.4360.47 L.min21, a peak [La] of 8.661.6 mM and a maximal heart rate of 19368 bpm. The CP and the W9 were 283635 W and 2266 kJ, respectively. The P3-min was calculated to be 407647 W. The goodness-of-fit of the power-time relationship was R2 = 0.98. The SEE of the CP estimation was 7.066.6 W. The parameters of the VO2 kinetics during the paced exercise trials (FS, ES and SS) with and without prior exercise are presented in table 1. The measurements of VO2 amplitude (i.e., A) revealed a main effect of prior exercise (F = 37.95; p,0.001), but no interaction was detected (F = 1.62; p = 0.212). Similarly, the absolute VO2 (i.e., A+ VO2baseline) of the pacing exercises revealed a main effect of prior exercise, but no interaction was detected (F = 0.81; p = 0.452). The analysis of the O2 deficit values revealed a significant interaction (F = 3.95; p = 0.028), indicating that the effect from previous exercise occurred only for the ES and SS conditions. Post hoc analyses revealed a significant reduction in the O2 deficit only for the SS (p = 0.003) and ES (p,0.001) conditions after prior exercise. The effect of the pacing strategy was only significant when comparing SS with FS (p = 0.002) and ES with FS (p,0.001) during the control condition. The analysis of the mean response time values revealed a significant interaction (F = 3.59; p = 0.037), indicating that the effect of previous exercise occurred only for the ES and SS conditions. Post hoc analyses revealed a significant reduction in mean response time for the SS (p,0.001) and ES (p,0.001) conditions after prior exercise. The effect of the pacing strategy was only significant when comparing SS with FS (p,0.001) and ES with FS (p,0.001) in the control condition. Figure 2 shows the VO2 responses during the paced Priming Exercise, Pacing Strategy and Short-Term Performance PLOS ONE | www.plosone.org 3 April 2014 | Volume 9 | Issue 4 | e95202 exercise trials (FS, ES and SS) with and without prior exercise in a representative subject. The VO2peak attained during the sprint for the FS, ES and SS conditions was significantly lower than VO2max, both with (FS = 4.1960.35, ES = 4.1460.33 and SS = 4.0360.32 L.min21; F = 5.678, p = 0.002) and without (FS = 3.9060.32, ES = 4.0160.30 and SS = 3.9960.73 L.min21; F = 5.678, p = 0.002) a prior exercise session. The VO2peak attained during the sprint for the FS, ES and SS conditions was unaffected by the prior exercise and pacing strategies (p.0.05). The parameters of exercise performance during the paced exercise trials (FS, ES and SS) with and without prior exercise are presented in table 2. The measurements of peak power output revealed a main effect of prior exercise (F = 61.72; p,0.001), but no interaction was detected (F = 2.28; p = 0.116). Similarly, the measurements of mean power output revealed a main effect of prior exercise (F = 6.54; p = 0.015), but no interaction was detected (F = 0.14; p = 0.873). The analysis of the time to peak power output values revealed no significant interaction (F = 0.52; p = 0.598), and no significant main effect of prior exercise (F = 0.59; p = 0.447) and pacing strategy (F = 2.38; p = 0.106). Figure 1. Study design for the six separate exercises conditions for a representative individual. Panels A, B and C - Paced exercises in the control condition, using slow start, even start and fast start, respectively. Panels D, E and F - Paced exercise preceded by previous heavy exercise (PHE), using slow start, even start and fast start, respectively. doi:10.1371/journal.pone.0095202.g001 Priming Exercise, Pacing Strategy and Short-Term Performance PLOS ONE | www.plosone.org 4 April 2014 | Volume 9 | Issue 4 | e95202 Figure 3 shows the power output during the paced exercise trials (FS, ES and SS) with and without prior exercise in a representative subject. The measurements of the [La] values before the paced exercises revealed a main effect of prior exercise (F = 58.77; p,0.001), but no interaction was detected (F = 0.01; p = 0.992). The mean [La] values after prior exercise for the FS, ES and SS conditions were 1.8760.50, 1.7860.61 and 1.7360.74 mM, respectively. The mean control [La] values for the FS, ES and SS conditions were 1.1460.41, 1.0160.18 and 0.9860.24 mM, respectively. The analysis of the [La] values after the sprint revealed no significant interaction (F = 2.00; p = 0.149), with no significant main effect of prior exercise (F = 0.79; p = 0.381) and pacing strategy (F = 1.15; p = 0.329). The mean [La] values after prior exercise for the FS, ES and SS conditions were 11.061.98, 11.162.39 and 10.162.32 mM, respectively. The mean control [La] values for the FS, ES and SS conditions were 11.363.11, 9.4662.15 and 10.262.28 mM, respectively. Discussion The purpose of this study was to determine the independent and additive effects of prior heavy-intensity exercise and pacing strategies on VO2 kinetics and performance during high-intensity exercise. Similar to previous studies, we have demonstrated that both priming exercise [6] and pacing strategies (i.e., FS) [5,7] accelerated the overall VO2 kinetics (i.e., mean response time). However, our study reveals, for the first time, that an FS pacing strategy does not magnify the positive effects of prior heavy- intensity exercise on the overall VO2 kinetics. Moreover, the performance during high-intensity exercise (i.e., peak power output and mean power output) was enhanced only by prior heavy-intensity exercise. These data confirm and extend the proposal that the changes (i.e., speeding/slowing) in the VO2 kinetics during the initial phase of different pacing strategies (FS, ES and SS) are not necessarily associated with the changes in performance during short-term high-intensity exercise [5]. Some studies have found that the overall VO2 kinetics is accelerated by an FS pacing strategy when compared with ES and SS strategies [5,7]. Factors such as exercise modality [5,11] and aerobic performance level [6] do not appear to influence the effects of the FS pacing strategy on the overall VO2 kinetics. Thus, our data confirm that an FS pacing strategy can improve the overall VO2 kinetics during high-intensity exercise in trained endurance cyclists. Studies have shown a direct proportionality between the products of PCr splitting and muscle or pulmonary VO2 [19]. An FS pacing strategy requires a greater initial rate of muscle ATP hydrolysis, resulting in a greater initial D [PCr]/D time. Thus, a more rapid accumulation of the metabolites (ADP, Pi and Ca2+) that stimulate oxidative phosphorylation would be Figure 2. Oxygen uptake (VO2) responses during the pacing exercise conditions in a representative subject. The horizontal line superimposed on each panel indicates the subject’s VO2max. Panels A, B and C - slow start, even start and fast start pacing conditions, respectively. Grey circles and black circles - paced exercise trials, with and without prior exercise, respectively. Notice that the VO2 response is speeded using the fast start pacing strategy only in the control condition. Thus, the fast start pacing strategy does not magnify the positive effects of prior heavy-intensity exercise on overall VO2 response. doi:10.1371/journal.pone.0095202.g002 Table 1. Parameters of the oxygen uptake (VO2) kinetics during paced exercise trials (FS, ES and SS), with and without prior exercise. Control After heavy exercise Significance FS ES SS FS ES SS VO2b (L.min21) 1.27 1.26 1.20 1.13 1.26 1.13 NS 0.20 0.21 0.22 0.22 0.25 0.23 A (L.min21) 2.67 2.73 2.63 2.89 2.86 2.91 *F = 37.95 0.31 0.40 0.36 0.42 0.42 0.40 p,0.001 Absolute VO2 (L.min21) 3.94 3.99 3.82 4.03 4.12 4.04 *F = 12.86 0.31 0.36 0.33 0.39 0.31 0.34 p = 0.001 MRT (s) 27 32 32 24 25 26 {F = 3.59 4 5N 6N 5 5` 5` P = 0.037 CI 95 (s) 2.2 2.2 2.5 1.5 1.9 2.1 - 0.7 0.5 0.8 0.6 0.7 0.7 O2 deficit (L) 1.22 1.48 1.45 1.16 1.15 1.22 {F = 3.95 0.25 0.27N 0.38N 0.40 0.25` 0.30` P = 0.028 Data are the mean +SD. N = 14. VO2b, baseline oxygen uptake; A, amplitude; Absolute VO2, VO2b+A; MRT, mean response time; CI 95, 95% confidence interval for MRT estimation. FS, fast start; ES, even start; SS, slow start. *Main effect of previous exercise; {Prior vs. pacing interaction; `p,0.05 relative to the control condition; Np,0.05 relative to the FS condition. doi:10.1371/journal.pone.0095202.t001 Priming Exercise, Pacing Strategy and Short-Term Performance PLOS ONE | www.plosone.org 5 April 2014 | Volume 9 | Issue 4 | e95202 observed during an FS pacing strategy. Accordingly, Bailey et al. [5] used near-infrared spectroscopy (NIRS) to verify that FS strategies might be linked to increased muscle O2 extraction. The classic experiments of Gerbino et al. [16] demonstrated that prior heavy exercise accelerated the monoexponential kinetics (i.e., mean response time) during a second bout of heavy exercise performed 6 min after the first. Later, studies using different experimental designs (e.g., intensities, durations of recovery time and age group) [6,25] confirmed the seminal results obtained by Gerbino et al. [16]. Our experimental results revealed that both the mean response time and VO2 amplitude (the overall increase in VO2 above the baseline) were modified by previous heavy exercise. The increased VO2 amplitude during a second bout of severe exercise has been considered important for exercise tolerance/performance, because the slow component of the VO2 kinetics, the change in blood lactate concentration and the aerobic contribution are positively modified during the second bout of exercise. Central (increases in bulk O2 delivery) and peripheral (convective O2 delivery and increased activity of mitochondrial enzymes) factors are possible explanations for this altered VO2 response profile during the second bout of exercise [15]. To the best of our knowledge, this study is the first to determine the possible additive effects of priming exercise and pacing strategy on VO2 kinetics during severe-intensity exercise. We have demonstrated that previous heavy-intensity exercise accelerated the overall VO2 kinetics only during SS and ES pacing strategies. Moreover, there was no significant difference among the FS, ES and SS preceded by previous heavy-intensity exercise. Thus, the effects of priming exercise on VO2 kinetics during severe-intensity exercise are dependent on pacing strategy. Moreover, these effects do not appear to be magnified by an FS pacing strategy. Together, these results suggest that previous exercise has great potential to enhance the overall VO2 kinetics and that an FS pacing strategy does not amplify its effects. In the present study, we did not use a biexponential model to characterize the VO2 kinetics because we were unable to repeat each trial to enhance the signal-to-noise ratio of the VO2 responses (see below). Thus, the parameters of the VO2 kinetics were not characterized. However, the exercise intensity used during the different pacing strategies was similar to PVO2max; therefore, the VO2 slow component, that elevates the VO2 above the steady- state value predicted from the sub-lactate threshold VO2-work rate relationship [26], cannot be detected under these circumstances. Thus, in line with other studies, the previous exercise may have only enhanced the VO2 amplitude [15], while the pacing strategy enhanced the time constant of the primary component of the VO2 response [5,7]. Nevertheless, previous heavy-intensity exercise blunted the effects of the FS pacing strategy on the overall VO2 kinetics. Therefore, the alterations caused by previous exercise (available O2, convective O2 delivery, activity of mitochondrial enzymes and motor unit recruitment) appear to prevent the effects of an FS pacing strategy on the VO2 response. In line with this statement, Rossiter et al. [27] have found that prior high-intensity exercise reduced the amplitude of the [PCr] response, with the initial rate of [PCr] change (d[PCr]/dt) remaining unaffected during a second bout of heavy exercise. These alterations are suggestive of a reduced t[PCr] during primed exercise [27], although the difference between conditions (i.e., control, 34 s vs. primed exercise, 32 s) did not reach statistical significance. Moreover, it has been demonstrated that the intramuscular enzyme activity status (i.e., pyruvate dehydrogenase complex - PDC), can allow a greater flux of acetyl groups into the mitochondria for oxidation [28]. An increased activation of pyruvate dehydrogenase complex might reduce both substrate- level phosphorylation (i.e., glycolysis and the creatine kinase and adenylate kinase reactions) [28] and the primary component time Figure 3. Power output during the pacing exercise conditions in a representative subject. Panels A, B and C - slow start, even start and fast start pacing conditions, respectively. Grey lozenges and black squares - paced exercise trials, with and without prior exercise, respectively. Notice the effect of prior exercise on performance, irrespectively of the pacing strategy used. doi:10.1371/journal.pone.0095202.g003 Table 2. Parameters of exercise performance during paced exercise trials (FS, ES and SS), with and without prior exercise. Control After heavy exercise Significance FS ES SS FS ES SS PP (W) 606 558 553 674 617 586 *F = 61.72 101 93 94 110 110 106 p,0.001 TPP (s) 6 6 7 6 6 7 NS 2 2 3 1 1 1 MPO (W) 400 396 386 415 408 396 *F = 6.54 44 43 69 60 47 68 p = 0.015 Data are the mean +SD. N = 14. PP, peak power output; TPP, time to peak power output; MPO, mean power output. FS, fast start; ES, even start; SS, slow start; *Main effect of previous exercise. doi:10.1371/journal.pone.0095202.t002 Priming Exercise, Pacing Strategy and Short-Term Performance PLOS ONE | www.plosone.org 6 April 2014 | Volume 9 | Issue 4 | e95202 constant [29]. Thus, these mechanisms (altered [PCr] kinetics and/or increased PDC activity) might have blunted the effects of an FS pacing strategy on the VO2 response (i.e., VO2 ‘‘error signal’’). However, future studies, with appropriated experimental design, should be conducted to confirm (or not) these hypotheses. Previous heavy- or severe-intensity exercises have been shown to improve exercise tolerance during both submaximal [6] and perimaximal exercise [14]. This positive effect seems to be influenced by an optimal interaction between prior exercise intensity and recovery duration [6]. We have provided the first demonstration that prior heavy-intensity exercise enhances per- formance (peak power output and mean power output), and this increase occurs independently of the chosen pacing strategy. Improved exercise tolerance during submaximal intensity has been observed when the amplitude of the slow component of the VO2 kinetics decreased and the overall VO2 kinetics were faster [6]. Indeed, we have observed that the overall VO2 kinetics was faster and the VO2 amplitude increased (and thus the magnitude of the O2 deficit was reduced) after heavy-intensity exercise. Thus, previous heavy-intensity exercise can reduce the W9 utilization during the initial phase of exercise, improving performance during short-term high-intensity exercise. It has been proposed that the mild lactic acidosis caused by prior heavy exercise might increase oxygen delivery by stimulating vasodilatation and a rightward shift in the oxyhaemoglobin dissociation curve (i.e. the Bohr effect) [16]. Based on data obtained in active subjects (VO2max,50 mL.kg21.min21), some studies have suggested that a baseline blood lactate concentration of ,3 mM results in an increased time to exhaustion during subsequent high-intensity exercise [6,14]. Moderate-intensity prior exercise, which did not alter the baseline blood lactate concen- tration, does not enhance VO2 kinetics or exercise tolerance during subsequent high-intensity exercise performed by active subjects [16]. Similarly, Bailey et al. [6] have shown that the effect of prior heavy exercise on VO2 kinetics is prevented when baseline blood [lactate] recovers to ,2 mM. However, we have verified that a baseline blood lactate concentration of ,1.8 mM has enhanced both overall VO2 kinetics and short-term high-intensity performance in trained endurance cyclists. The low blood lactate concentration found 9 min after heavy intensity exercise can be explained, at least in part, by increased rate of blood lactate removal found in aerobic trained athletes [30]. Interestingly, Burnley et al. [31] have found that moderate-intensity prior exercise enhanced both primary VO2 amplitude and exercise performance in well-trained cyclists (VO2max ,58 mL.kg21.min21). Thus, in aerobic trained cyclists, it seems that the presence of an elevated blood lactate concentration is not a sine qua non condition for improving VO2 kinetics and short-term high-intensity performance after prior exercise. Some studies found that an FS pacing strategy can improve exercise tolerance [7] and performance [5] during short-term high-intensity exercise. In the present study, the pacing strategy did not significantly influence the exercise performance, although the overall VO2 kinetics was improved by the FS pacing strategy. Some interventions (priming exercise and pacing) have shown similar results [5,6], indicating that changes in the overall VO2 kinetics will not necessarily enhance exercise tolerance/perfor- mance during subsequent high-intensity exercise. Interestingly, Bailey et al. [5] reported that utilizing an FS pacing strategy with active individuals (VO2max ,52 mL.kg21.min21) improved both the overall VO2 kinetics and exercise performance during subsequent high-intensity exercise. In the present study, we analyzed trained endurance cyclists (VO2max = 62 mL.kg21.- min21). Thus, differences in aerobic fitness might explain, at least in part, these different results. Bailey et al. [5] proposed that the attainment of VO2max during high-intensity exercise bouts, when this is ordinarily not possible, is essential for improving exercise performance. Given the finite speed of the VO2 response, the exercise durations at the extreme domain [32] would be too short to permit attainment of VO2max [21]. Thus, the attainment of VO2max would allow a more complete depletion of W9 and consequently allow better exercise performance [5]. Indeed, we have verified that VO2max was not attained during the FS pacing strategy. However, future studies using different experimental designs should be conducted to test this relationship. Because of the nature of the present experiments, certain limitations of the study should be considered when interpreting its findings. The determination of the VO2 response parameters in the heavy- and severe-intensity domain using only one transition can have potential limitations (i.e., low confidence in the response parameters). Repeated bouts have traditionally been averaged to improve the signal-to-noise ratio of data [33]. However, due to the extremely demanding nature of the exercise testing and the frequent laboratory visits (11), only one trial was conducted for each experimental condition. Although we only measured one transition, the signal-to-noise ratio of the data can be improved by using higher VO2 amplitudes [33]. Therefore, higher VO2 amplitudes, as utilized in the present study, correspond to smaller confidence intervals. Indeed, the 95% confidence interval for the estimation of mean response time was ,3 s for all conditions (Table 1). In summary, we have demonstrated in trained endurance cyclists that priming heavy-intensity exercise has a positive effect on both overall VO2 kinetics and short-term high-intensity performance. However, the FS pacing strategy only modified the overall VO2 kinetics. This finding suggests that faster overall VO2 kinetics does not, per se, determine the performance (i.e., peak power output and mean power output) during high-intensity exercise. The FS pacing strategy does not magnify the positive effects of prior heavy-intensity exercise on the overall VO2 kinetics. Thus, the modifications caused by priming exercise preclude the effects of the FS pacing strategy on the overall VO2 kinetics. Finally, priming exercise seems to have greater potential than FS pacing strategies to enhance both overall VO2 kinetics and short-term high-intensity performance in trained endurance cyclists. Author Contributions Conceived and designed the experiments: CCG BSD. Performed the experiments: RACC CCG. Analyzed the data: RACC CCG BSD. Contributed reagents/materials/analysis tools: RACC CCG BSD. Wrote the paper: RACC CCG BSD. References 1. Gaesser GA, Poole DC (1996) The slow component of oxygen uptake kinetics in humans. Exerc Sport Sci Rev 24: 35–71. 2. Poole DC, Ward SA, Gardner GW, Whipp BJ (1988) Metabolic and respiratory profile of the upper limit for prolonged exercise in man. Ergonomics 31: 1265– 1279. 3. Jones AM, Vanhatalo A, Burnley M, Morton RH, Poole DC (2010) Critical power: implications for determination of VO2max and exercise tolerance. Med Sci Sports Exerc 42: 1876–1890. 4. Jones AM, Grassi B, Christensen PM, Krustrup P, Bangsbo J, et al. (2011) Slow component of VO2 kinetics: mechanistic bases and practical applications. Med Sci Sports Exerc 43: 2046–2062. Priming Exercise, Pacing Strategy and Short-Term Performance PLOS ONE | www.plosone.org 7 April 2014 | Volume 9 | Issue 4 | e95202 5. Bailey SJ, Vanhatalo A, DiMenna FJ, Wilkerson DP, Jones AM (2011) Fast-start strategy improves VO2 kinetics and high-intensity exercise performance. Med Sci Sports Exerc 43: 457–467. 6. Bailey SJ, Vanhatalo A, Wilerson DP, Dimenna FJ, Jones AM (2009) Optimizing the ‘‘priming’’ effect: influence of prior exercise intensity and recovery duration on O2 uptake kinetics and severe-intensity exercise tolerance. J Appl Physiol 107: 1743–1756. 7. Jones AM, Wilkerson DP, Vanhatalo A, Burnley M (2008) Influence of pacing strategy on O2 uptake and exercise tolerance. Scand J Med Sci Sports 18: 615– 626. 8. Noakes TD, St Clair Gibson A, Lambert EV (2005) From catastrophe to complexity: a novel model of integrative central neural regulation of effort and fatigue during exercise in humans: summary and conclusions. Br J Sports Med 39: 120–124. 9. Brown MR, Delau S, Desgorces FD (2010) Effort regulation in rowing races depends on performance level and exercise mode. J Sci Med Sport 13: 613–617. 10. Foster C, De Koning JJ, Hettinga F, Lampen J, La Clair KL, et al. (2003) Pattern of energy expenditure during simulated competition. Med Sci Sports Exerc 35: 826–831. 11. Bishop D, Bonetti D, Dawson B (2002) The influence of pacing strategy on VO2 and supramaximal kayak performance. Med Sci Sports Exerc 34: 1041–1047. 12. Foster C, Snyder AC, Thompson NN, Green MA, Foley M, et al. (1993) Effect of pacing strategy on cycle time trial performance. Med Sci Sports Exerc 25: 383–388. 13. Spencer MR, Gastin PB (2001) Energy system contribution during 200- to 1500- m running in highly trained athletes. Med Sci Sports Exerc 33: 157–162. 14. Jones AM, Wilkerson DP, Burnley M, Koppo K (2003) Prior heavy exercise enhances performance during subsequent perimaximal exercise. Med Sci Sports Exerc 35: 2085–2092. 15. Jones AM, Koppo K, Burnley M (2003) Effects of prior exercise on metabolic and gas exchange responses to exercise. Sports Med 33: 949–971. 16. Gerbino A, Ward SA, Whipp BJ (1996) Effects of prior exercise on pulmonary gas-exchange kinetics during high-intensity exercise in humans. J Appl Physiol 80: 99–107. 17. MacDonald M, Pedersen PK, Hughson RL (1997) Acceleration of VO2 kinetics in heavy submaximal exercise by hyperoxia and prior high intensity exercise. J Appl Physiol 83: 1318–1325. 18. Burnley M, Davison G, Baker JR (2011) Effects of priming exercise on VO2 kinetics and the power-duration relationship. Med Sci Sports Exerc 43: 2171– 2179. 19. Rossiter HB, Ward SA, Kowalchuk JM, Howe FA, Griffiths JR, et al. (2002) Dynamic asymmetry of phosphocreatine concentration and O2 uptake between the on- and off-transients of moderate- and high-intensity exercise in humans. J Physiol 15: 991–1002. 20. Carter H, Jones AM, Barstow TJ, Burnley M, Williams CA, et al. (2000) Oxygen uptake kinetics in treadmill running and cycle ergometry: a comparison. J Appl Physiol 89: 899–907. 21. Caputo F, Denadai BS (2008) The highest intensity and the shortest duration permitting attainment of maximal oxygen uptake during cycling: effects of different methods and aerobic fitness level. Eur J Appl Physiol 103: 47–57. 22. Bull AJ, Housh TJ, Johnson GO, Perry SR (2000) Effect of mathematical modeling on the estimation of critical power. Med Sci Sports Exerc 32: 526–530. 23. Hill DW, Smith JC (1994) A method to ensure the accuracy of estimates of anaerobic capacity derived using the critical power concept. J Sports Med Phys Fitness 34: 23–37. 24. Vanhatalo A, McNaughton LR, Siegler J, Jones AM (2010) Effect of induced alkalosis on the power-duration relationship of ‘‘all-out’’ exercise. Med Sci Sports Exerc 42: 563–570. 25. Barker AR, Jones AM, Armstrong N (2010) The influence of priming exercise on oxygen uptake, cardiac output, and muscle oxygenation kinetics during very heavy-intensity exercise in 9- to 13-yr-old boys. J Appl Physiol 109: 491–500. 26. Vanhatalo A, Poole DC, DiMenna FJ, Bailey SJ, Jones AM (2011) Muscle fiber recruitment and the slow component of O2 uptake: constant work rate vs. all-out sprint exercise. Am J Physiol Regul Integr Comp Physiol 300: R700–R707. 27. Rossiter HB, Ward SA, Kowalchuk JM, Howe FA, Griffiths JR, et al. (2001) Effects of prior exercise on oxygen uptake and phosphocreatine kinetics during high-intensity knee-extension exercise in humans. J Physiol 537: 291–303. 28. Timmons JA, Poucher SM, Constantin-Teodosiu D, Macdonald IA, Greenhaff PL (1997) Metabolic responses from rest to steady state determine contractile function in ischemic skeletal muscle. Am J Physiol 273: E233–238. 29. Campbell-O’Sullivan SP, Constantin-Teodosiu D, Peirce N, Greenhaff PL (2002) Low intensity exercise in humans accelerates mitochondrial ATP production and pulmonary oxygen kinetics during subsequent more intense exercise. J Physiol 538: 931–939. 30. Messonnier L, Freund H, Denis C, Fe´asson L, Lacour JR (2006) Effects of training on lactate kinetics parameters and their influence on short high-intensity exercise performance. Int J Sports Med 27: 60–66. 31. Burnley M, Doust JH, Jones AM (2005) Effects of prior warm-up regime on severe-intensity cycling performance. Med Sci Sports Exerc 37: 838–845. 32. Hill DW, Poole DC, Smith JC (2002) The relationship between power and the time to achieve VO2max. Med Sci Sports Exerc 34: 709–714. 33. Lamarra N, Whipp BJ, Ward SA, Wasserman K (1987) Effect of interbreath fluctuations on characterizing exercise gas exchange kinetics. J Appl Physiol 62: 2003–2012. Priming Exercise, Pacing Strategy and Short-Term Performance PLOS ONE | www.plosone.org 8 April 2014 | Volume 9 | Issue 4 | e95202

The positive effects of priming exercise on oxygen uptake kinetics and high-intensity exercise performance are not magnified by a fast-start pacing strategy in trained cyclists.

04-16-2014

Caritá, Renato Aparecido Corrêa,Greco, Camila Coelho,Denadai, Benedito Sérgio

eng

PMC8863837

1 Vol.:(0123456789) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports Explaining the differences of gait patterns between high and low‑mileage runners with machine learning Datao Xu1, Wenjing Quan1,2,3, Huiyu Zhou1,4, Dong Sun1, Julien S. Baker5* & Yaodong Gu1* Running gait patterns have implications for revealing the causes of injuries between higher‑mileage runners and low‑mileage runners. However, there is limited research on the possible relationships between running gait patterns and weekly running mileages. In recent years, machine learning algorithms have been used for pattern recognition and classification of gait features to emphasize the uniqueness of gait patterns. However, they all have a representative problem of being a black box that often lacks the interpretability of the predicted results of the classifier. Therefore, this study was conducted using a Deep Neural Network (DNN) model and Layer‑wise Relevance Propagation (LRP) technology to investigate the differences in running gait patterns between higher‑mileage runners and low‑mileage runners. It was found that the ankle and knee provide considerable information to recognize gait features, especially in the sagittal and transverse planes. This may be the reason why high‑mileage and low‑mileage runners have different injury patterns due to their different gait patterns. The early stages of stance are very important in gait pattern recognition because the pattern contains effective information related to gait. The findings of the study noted that LRP completes a feasible interpretation of the predicted results of the model, thus providing more interesting insights and more effective information for analyzing gait patterns. With an increase of the number of recreational runners, the injuries caused by overuse running are increasing1,2. The etiology of excessive use of running injuries is multifactorial, which may result from the interaction of many factors of external uncertainties (e.g., weekly running days, weekly running mileages, running environment, footwear) and internal risk (e.g., biomechanics factors, foot strike pattern, anatomic factors, age, gender)3. The injury rate among recreational runners has been recorded as high as 29.4%, with overuse knee injuries (e.g., knee anterior pain and iliotibial band syndrome) being the most reported4. Previous studies have shown that weekly running mileage is a major risk factor related to running injuries1,5, and there are significant differences in injuries between higher-mileage runners (self-reported running more than 32 km per week) and low-mileage runners (self-reported running less than 25 km per week)6. The higher-mileage weekly runners show higher rates of hip and hamstring injuries7, while the low-mileage weekly runners show higher rates of knee injuries8. Gait patterns are an important factor in decoding gait characteristics, which is related to revealing motor injuries and gait recognition9,10. Therefore, running gait patterns have implications for understanding the causes of injuries between higher-mileage runners and low-mileage runners. However, there is limited research on the possible relationship between running gait patterns and weekly running mileages. Biomechanical analysis of higher-mileage and low-mileage runners may be useful in order to better under- stand the potential relationship between running mileage and specific types of injuries. However, current research on the biomechanical performance of running gait of high-mileage and low-mileage runners mainly focuses on kinematics. Boyer et al. used the principal component analysis found that there were recognizable differ- ences in the kinematics of the sagittal and frontal planes of the ankle, the frontal plane of the knee, the frontal and transverse plane of the hip in the stance phase between high-mileage and low-mileage runners11. Clermont et al. then combined the methods of principal component analysis with support vector machines with kinematic OPEN 1Faculty of Sports Science, Ningbo University, Ningbo 315211, China. 2Faculty of Engineering, University of Pannonia, Veszprém, Hungary. 3Savaria Institute of Technology, Eötvös Loránd University, Budapest, Hungary. 4School of Health and Life Sciences, University of the West of Scotland, Glasgow G72 0LH, Scotland, UK. 5Department of Sport, Physical Education and Health, Hong Kong Baptist University, Hong Kong 999077, China. *email: jsbaker@hkbu.edu.hk; guyaodong@nbu.edu.cn 2 Vol:.(1234567890) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ data to classify runners based on mileage, and found that the classification performance of gait kinematics of high-mileage and low-mileage runners had high accuracy, which means there was high identifiability in the gait kinematics between high-mileage and low-mileage runners12. However, the kinetics (joint moments) of biomechanical parameters also play an important role in identifying damage patterns, especially in revealing the stresses on the major joints13,14. Therefore, both kinematics and kinetics should be considered to improve the recognition of gait patterns and reveal the pattern characteristics in a more detailed way when recognizing the running gait patterns of high-mileage and low-mileage runners. When analyzing variables related to gait patterns, the previous method mainly examines the influence of single-time discrete gait variables. Previous methods have successfully addressed many important clinical and scientific questions related to human gait, but there are some inherent limitations. For example, when discrete variables are extracted from time-series variables, a large amount of data is lost 10. In addition, a single pre- selected gait variable may miss potentially meaningful information represented by other unselected variables and correlated variables. Therefore, given the shortcomings of traditional methods, machine learning techniques (such as hierarchical clustering analysis, support vector machines, artificial neural networks, etc.) and multivari- able statistical analysis have been used to examine and analyze human motion based on time-series gait patterns in recent years9,12,15–17. The progressive development of advanced motion capture equipment makes it possible to collect a large amount of clinical biomechanics data, which results in the increasing application of machine learning in clinical biomechanics16,18,19. For example, artificial neural networks and support vector machines are used for pattern recognition and classification of gait features to emphasize the uniqueness of gait patterns9,10,12. Machine learning approaches can be very successful in solving many clinical biomechanical problems related to classification systems and providing new insights from complex model systems. However, they all have the same problem of being a black box that doesn’t provide any information about what makes the decisions20,21. In other words, these models often lack the interpretability of the predicted results of the classifier22. The main reason for this lack of interpretability is the nonlinearity of various mappings that process the original data set (such as gait patterns) to their characteristic representation and then to the classifier function. In gait pattern recognition, this prevents experts in the relevant fields from carefully verifying classification decisions, because simple answers of "yes" or "no" sometimes have little or limited value. Therefore, Layer-wise Relevance Propaga- tion (LRP) technology is proposed to solve the problem of lack of interpretability22. LRP is a technology used to identify important relevance (that is, by measuring the contribution of each input variable to the overall predict outcomes) through backward propagation in neural networks22,23. LRP has been successfully applied to classifica- tion recognition tasks in many scenarios, such as text, image, and even gait pattern recognition9,10,24. Therefore, the application of LPR in running gait pattern recognition can improve the overall transparency of the classifier and make the classification results interpretable, thus providing reliable clinical biomechanical diagnostic results. Therefore, the purpose of this study was to investigate the differences in running gait patterns between higher-mileage runners and low-mileage runners. Specifically, the aim of this study was: (1) To train a deep neural network (DNN) model by using the kinematics and kinetics data of runners with different weekly running mileages as input variables to classify and recognize the gait characteristics of runners with higher-mileage and low-mileage runners. (2) To evaluate the classifier performance of DNN classification models based on different input variables (separate kinematic inputs; separate kinetic inputs; kinematic and kinetic inputs together). (3) To identify the relevance of relevant variables and time points between higher-mileage and low-mileage runners by using LRP technology. (4) To explore LRP as a method for data reduction and explain the classification decision of the DNN classifier model based on the high relevant variables. Results Performance of deep neural network classification models. For the matrices M , 75 TP, 5 FN, 77 TN and 3 FP were obtained by DNN classifier. For the matrices Mkinematics , 75 TP, 5 FN, 69 TN and 11 FP were obtained by DNN classifier. For the matrices Mkinetics , 70 TP, 10 FN, 77 TN and 3 FP were obtained by DNN classifier. All classification performance parameters are presented in Fig. 1. For the classifier of the DNN models based on the matrices M (Fig. 1A), the model showed the higher accuracy rate (accuracy rate: 95%) than the matrices Mkinematics (accuracy rate: 90.00%) and matrices Mkinetics (accuracy rate: 91.88%). In general, the clas- sifier of the DNN models based on the matrices M presented a perfect accuracy rate, specificity rate, as well as precision rate compared to separate matrices Mkinematics and Mkinetics . At the same time, the classifier of the DNN models based on the matrices M showed the higher F1 − score (0.9494) and MCC (0.9003) than the matrices Mkinematics and matrices Mkinetics (Fig. 1C). Overall, the classifier performance based on the matrices M achieved an F1 − score and MCC score of very strong relationships. The ROC curves are showed in Fig. 1, the ROC curves of the classifier of the DNN models based on the matrices M (Fig. 1A) presented a good classification performance during the overall area. However, the ROC curves based on the matrices Mkinematics (Fig. 1B) show the worse classification performance during the about (0FPR−0.1FPR) ∗ (0.4FPR−1FPR) area, and the matrices Mkinetics (Fig. 1C) show the worse classification perfor- mance during the about (0FPR−0.7FPR) ∗ (0.9FPR−1FPR) area. The classifier of the DNN models based on the matrices M show the higher AUC (0.9427) than the matrices Mkinematics (AUC: 0.8981) and matrices Mkinetics (AUC: 0.9097). Overall, the classifier of the DNN models based on the matrices M has a good performance from the perspective of overall indicators. Results of LPR. The relative contribution of variables during the overall stance phase are showed in Fig. 2A, the variables recorded at every 1% of the stance interval are related to successfully matching the stride pat- tern between the higher-mileage runners and lower-mileage runners. The contribution of variables during the 3 Vol.:(0123456789) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ 1%-47% stance phase (contribution: 52.54%) was higher than the contribution of variables during the 48%-100% stance phase (contribution: 47.46%) to the successful classification. The summed contribution of the relevance score of each joint (ankle, knee, hip) of each plane (sagittal, frontal, transverse) of kinematics (joint angle) and kinetics (joint moment) trajectories are showed in Fig. 2C. The summed contribution rate of the relevance score of the ankle, knee, hip was 43.16%, 35.98%, 20.86%, respectively. The summed contribution rate of the relevance score of the sagittal, frontal, transverse was 39.90%, 32.24%, 27.86%, respectively. The most relevant trajectory variables were the ankle dorsiflexion-plantarflexion angle (9.69%), the knee internal–external rotation angle (9.59%), the ankle dorsiflexion-plantarflexion moment (9.37%), and the knee flexion–extension moment (9.39%). Secondly, the relevant trajectory variables were the knee flexion–extension angle (7.19%), the hip abduction–adduction angle (8.64%), and the ankle inversion- eversion moment (7.93%). However, there was little relevance score in the variables of knee abduction–adduc- tion angle (1.93%), hip flexion–extension angle (1.90%), hip internal–external rotation angle (1.85%), ankle internal–external rotation moment (2.99%), knee abduction–adduction moment (1.70%), hip flexion–extension moment (2.36%), hip internal–external rotation moment (1.18%). The detailed distribution of relevance score during each joint (ankle, knee, hip) of each plane (sagittal, frontal, transverse) of kinematics (joint angle) and kinetics (joint moment) are showed in Fig. 2B. There were revealing Figure 1. The classifier of the DNN models based on the matrices M , Mkinematics , Mkinetics . (A) The classifier of the DNN models based on the matrices M . (B) The classifier of the DNN models based on the matrices Mkinematics . (C) The classifier of the DNN models based on the matrices Mkinetics . ROC: Receiver Operating Characteristic; AUC: Area Under the ROC Curve; MCC: Matthews Correlation Coefficient; TPR: True Positive Rate; FPR: False Positive Rate. 4 Vol:.(1234567890) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ findings contributing to distribution of the variables on time points between the higher-mileage runners and lower-mileage runners during the overground running movement patterns. Notable highly relevant variables (the top 200 variables with the highest correlation relevance, all of them had a relevance score of over 0.7) during the stance are showed in Fig. 3. For the kinematics of the ankle, there was high relevance score in dorsiflexion-plantarflexion angle during the 1%–18%, 47%–51%, 88%–95% stance phase; in inversion-eversion angle during the 69%–72%, 98%–99% stance phase; in internal–external rotation Figure 2. The LPR results in the average absolute relevance score of every variable in a stride pattern. (A) The relative contribution of variables during the overall stance phase (0%–100%). (B) The detailed distribution of relevance score during each joint (ankle, knee, hip) of each plane (sagittal, frontal, transverse) of kinematics (joint angle) and kinetics (joint moment). The darker colors mean high relevance variables, the lighter colors mean low relevance variables. The model relied more on darker color variables; the lighter colors variables had less relevance with correctly classified gait patterns. (C) The summed contribution of the relevance score of each joint (ankle, knee, hip) of each plane (sagittal, frontal, transverse) of kinematics (joint angle) and kinetics (joint moment) trajectories. Figure 3. Notable highly relevant variable during each joint (ankle, knee, hip) of each plane (sagittal, frontal, transverse) of kinematics (joint angle) and kinetics (joint moment). The top 200 variables with the highest correlation relevance, all of them had a relevance score of over 0.7. 5 Vol.:(0123456789) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ angle during the 19%–34% stance phase. For the kinematics of the knee, there was high relevance score in flexion–extension angle during the 3%-21% stance phase; in internal–external rotation angle during the 6%, 11%–34%, 37%–41%, 81%–88% stance phase. For the kinematics of the hip, there was high relevance score in abduction–adduction angle during the 10%–14%, 68%, 77%–83% stance phase. For the kinetics of the ankle, there was high relevance score in the dorsiflexion-plantarflexion moment during the 2%–4%, 9%–11%, 13%–21%, 28%–34%, 95%–97% stance phase; in the inversion-eversion moment during the 32%–35% stance phase. For the kinetics of the knee, there was high relevance score in the flexion–extension moment during the 3%–11%, 14%–33%, 69%–70% stance phase; in internal–external rotation moment during the 26%–34% stance phase. For the kinetics of the hip, there was high relevance score in the abduction–adduc- tion moment during the 37%–44% stance phase. Discussion This study aimed to investigate the differences in running gait patterns between higher-mileage runners and low- mileage runners. The objectives were to firstly train the DNN model by using the running gait kinematics (joint angle) and kinetics (joint moment) dataset as input variables to classify and recognize the gait characteristics of runners with higher-mileage and low-mileage runners. Secondly, to evaluate the classifier performance of DNN classification models based on different input variables (separate kinematic inputs; separate kinetic inputs; kinematic and kinetic inputs together). Finally, to use LRP to identify the relevance of relevant variables and time points between higher-mileage and low-mileage runners, and explain the classification decision of DNN classifier model based on those high relevant variables. According to our research results, higher-mileage and low-mileage runners have discernable differences in gait characteristics, independently in relation to the perspective of kinematics or kinetics variables. When the classifier of the DNN models is only based on the kinematics as the input variables, the model shows good clas- sification performance (Fig. 1A: accuracy rate is 90.00%). This supports previous findings of Clermont et al., who successfully classified higher- and low-mileage runners with 92.59% accuracy, showing that there are discernible differences in running gait kinematics between higher-mileage and low-mileage runners12. At the same time, when the classifier of the DNN models is only based on the kinetics as the input variables, the model accuracy rate is 91.88%, but when combining kinematics and kinetics as input variables, the model accuracy rate reaches 95%. In our study, the F1−score and MCC were used to evaluate the performance of the classifier, which can provide a good evaluation of the performance of the classifier34,35. In our results, the classifier of combining kinematics and kinetics as input variables obtained a higher F1−score (0.9494) and MCC (0.9003), as well as a higher AUC (0.9427). These results show that running gait kinetics data can increase the pattern recognition rate of gait characteristics between higher-mileage and low-mileage runners, at least in terms of classifier model performance. Therefore, the relevant research should consider the combination of kinematics and kinetics data sets rather than only simply kinematics when analyzing gait characteristics, if it is possible. It can provide more effective gait pattern information for the field of medical biomechanics. Of course, compared to only collecting kinematics data, both collecting kinematics and kinetics increase the difficulty of collection, especially in the absence of relevant equipment. In the research of gait pattern recognition, it is often necessary to record a large amount of data in order to better recognize gait patterns36, which makes it difficult to complete an accurate interpretation of gait pat- tern recognition results with few variables as possible. In this study, the variables were imported into the DNN model for training, and then the relevance score of each variable’s contribution to the gait pattern recognition results was obtained through LRP. The results of gait pattern recognition can be accurately interpreted by using highly correlated variables, which undoubtedly provides more important and effective information for gait pat- tern recognition. As shown in Fig. 2, not all variables contribute significantly to identifying the gait patterns of higher-mileage and low-mileage runners. The contribution of variables during the 1%–47% stance phase was higher than the contribution of variables during the 48%–100% stance phase to the successful recognize gait pattern (as shown in Fig. 2A). In other words, the early stage of the stance phase covers the interpretability of higher-mileage and low-mileage runners in gait pattern recognition. Horst et al. found that the most significant individual gait characteristics appeared in the early stage of the stance phase when they analyzed individual gait patterns in barefoot walking using LRP10. At the same time, Hoitz et al. found that the early stage of stance phase (1%–30%) has a more significant contribution to gait pattern recognition than the late stage of the stance phase9. The differences in foot strike patterns (from rearfoot strikes to forefoot strikes) are more readily observed in the early stages of stance37. These results seem to suggest that the early stages of stance may play a more important and meaningful role in identifying gait patterns. It also provides insights for other researchers who should focus on the early stages of stance when investigating gait patterns, at least for now the evidence suggests that early stages of stance contain more meaningful information about gait patterns. In addition to showing a more significant contribution during the early stages of stance, the summed contribu- tion of the relevance score of each joint of each plane of kinematics and kinetics trajectories are also inconsistent. As shown in Fig. 2C, our results show that the most relevant trajectory variables were the ankle dorsiflexion- plantarflexion angle, the knee internal–external rotation angle, the ankle dorsiflexion-plantarflexion moment, and the knee flexion–extension moment. The sagittal plane of the ankle and knee plays an important role in recognition gait patterns between high-milage and low-milage runners, which also confirms previous findings that the sagittal plane should be considered11. The hip appears to play a small role in identifying the gait patterns of higher-mileage and low-mileage runners, no matter from the perspective of kinematics or kinetics. However, when the top 200 variables with the highest correlation relevance score (as shown in Fig. 3, all of them had a relevance score of over 0.7) were extracted9, the high relevance score was shown in the abduction–adduction angle (moment) during the 10%–14%, 68%, 77%–83% (37%–44%) stance phase. Previous studies have shown 6 Vol:.(1234567890) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ that high-mileage runners exhibit larger hip adduction and have a higher risk of hip injury compared to low- mileage runners7,11. Therefore, it is permissible to use the gait characteristic of the hip frontal plane to identify gait patterns in higher-mileage and low-mileage runners, which can provide more information about injuries and individual characteristics. At the same time, the ankle and knee provide considerable information to recognize gait features, especially in the sagittal and transverse planes. It also suggests that runners adjust their gait pat- terns during the running gait stance phase, leading to more flexion of the knee and more valgus of the foot12,38. Therefore, the high-mileage runners show higher rates of hip and hamstring injuries and low-mileage runners show higher rates of knee injuries may be due to their different gait patterns. In general, LRP completes a feasible interpretation of the predicted results of the model, thus providing more interesting insights and more effective information for analyzing gait patterns. The relevance score results of LRP output enable machine learning algorithms (such as artificial neural networks) to predict and analyze multiple variables of the gait cycle from different time points. Compared with traditional gait analysis methods (based on a single pre-selected variable), machine learning algorithms in the field of medical biomechanics seem to be better able to correlate human movement with related injuries and diseases in multiple dimensions16,39. At the same time, the explainable relevance score results of gait pattern recognition show that the variables related to a particular gait pattern recognition are not confined to a single gait feature, nor are they evenly distributed across all gait features. In summary, the results of LRP demonstrate its applicability to the understanding and interpretation of machine learning prediction results in clinical (biomechanical) gait analysis. In other words, the application of machine learning in gait analysis combined with LRP is well worth considering by research- ers, which also provides some references for future clinical (biomechanical) analysis and diagnostic research. The current study has some limitations. First of all, only male runners were included in this study, so the results of this study apply only to male runners. In the future, female runners can be combined to explore the differences in gait patterns among different mileage runners. Secondly, the current study used uniform runners’ running speeds (3.3 m/s ± 10%) to minimize the biomechanical differences due to different running speeds40. Because of the differences in training levels and running habits between high-mileage and low-mileage runners, there may be a small number of runners not showing the most realistic gait pattern. In general, however, the subjects were given enough time to familiarize themselves to the uniform speed prior to formal experimental data collection, which compensated for any possible errors outlined. Conclusion Considering the combination of kinematics and kinetics data sets rather than only simply kinematics when analyzing gait characteristics can increase the pattern recognition rate of gait characteristics between higher- mileage and low-mileage runners, as well as providing more effective and efficient gait pattern information. The ankle and knee provide considerable information that can help recognize gait features, especially in the sagittal and transverse planes. This may be the reason why high-mileage and low-mileage runners have different injury patterns due to their different gait patterns. The early stages of the stance are also very important in the term of gait pattern recognition because it contains more effective information about gait patterns. LRP completes a feasible interpretation of the predicted results of the model, thus providing more interesting insights and more effective information for analyzing gait patterns. Thus, researchers should consider combining LRP when they apply machine learning in gait analysis. Methods Participants. This study recruited 80 male healthy runners: 40 higher-mileage runners (age: 35.51 ± 10.32 y, height: 172.30 ± 8.13  cm, body mass: 65.33 ± 7.46  kg, running experience: 8.56 ± 7.74, weekly mile- age: 44.31 ± 13.67  km), 40 lower-mileage runners (age: 33.90 ± 9.74 y, height: 173.40 ± 6.96  cm, body mass: 68.58 ± 8.20 kg, running experience: 4.71 ± 3.19, weekly mileage: 15.28 ± 5.30 km). The criteria for inclusion were no serious lower extremity musculoskeletal injury, no history of major lower extremity surgery, or any other injury factors that might interfere with the study in the previous 6 months. According to previous studies11,12, “lower-mileage” runners were defined as those who self-reported running less than 25 km per week, while “higher-mileage” runners were defined as those who ran more than 32 km per week. Participants were informed of the purpose, requirements, and procedures of the experiment. This study was performed in accordance with the Declaration of Helsinki, the study protocol was approved (Approval Number: RAGH20210326) by the Eth- ics Committee of Ningbo University, and the written informed consent was provided and signed by all subjects. Experimental protocol and procedures. The experiment was conducted in the biomechanics labora- tory at the Research Academy of Grand Health, Ningbo University. Three-dimensional lower limb joint kinemat- ics data were collected at 200 Hz using a Vicon (Vicon Metrics Ltd., United Kingdom) motion capture system (eight Infrared cameras). In an identical time frame, the ground reaction force (GRF) data were synchronously collected using a 1000 Hz in-ground AMTI force plate (AMTI, Watertown, United States). Vicon motion capture system and AMTI force plate are connected through Vicon Nexus 1.8.6 software to achieve the synchronous collection. This study selected the right leg as the analytical limb, so the 12.5 mm diameter standard reflective marker was attached to the pelvis and right lower limb25: right anterior superior iliac spine, left anterior superior iliac spine, right posterior superior iliac spine, left posterior superior iliac spine, right medial condyle, right lateral condyle, right medial malleolus, right lateral malleolus, right first metatarsal head, right fifth metatarsal head, right distal interphalangeal joint of the second toe. At the same time, three tracking clusters were labeled on the right middle and lateral thigh, right middle and lateral shank, right heel. A stadiometer and a calibrated scale were used to measure the subject’s body mass and height respectively. 7 Vol.:(0123456789) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ All subjects were asked to wear leggings and tights and uniform standard running shoes (Anta Flashedge, China). All runners were heel strikers. Prior to the formal experiment, subjects warmed up by jogging for 10 min in the laboratory environment at a self-selected speed. Following warm up, they then familiarized themselves with the experiment process and conducted preliminary experimental data collection. The infrared timers were placed on either side of the 20-m track to measure the participants’ running speed (specific location: 4-m behind/ in front of the force plate). The subjects were asked to run naturally across the track at a speed of 3.3 ± 10% meters per second and land with their right foot on the force plate in a natural unconsciousness way26. The test was considered valid when the subject was observed and measured to run at the correct speed and in a natural manner. A total of 10 recordings of valid data were collected for each subject. Data collection and processing Based on the study of Xu et al.27, the initial contact force point was determined as the vertical GRF greater than 10 N. The stance phase was defined as the force plate from the initial contact force point to the right lower limb leaving the force plate (force value to zero). The whole data set was preprocessed using Vicon Nexus 1.8.6 soft- ware. Firstly, the data of the reflective marker trajectory coordinates and the GRF data are exported from Vicon Nexus into C3D format file, and then the C3D format file is imported into Visual 3-D software (version 6.7.3, C-Motion Inc., Germantown, United States) for modeling and further processing. According to Winter’s study in relation to the filter selected frequency, the most appropriate signal-to-noise ratio was selected by carrying out residual analysis of the data of subsets28. Finally, fourth-order zero-phase lag Butterworth low-pass filters were selected to filter the data (Filter frequency, kinematics data: 10 Hz, kinetic data: 20 Hz). The pelvis model was developed according to the CODA model, and the hip joint center location was defined by regression Eqs. 29. The right hip joint center (RHJC) according to Eq. (1) and left hip joint center (LHJC) according to Eq. (2) was identified by the anterior superior iliac spine (ASIS): The center position of each segment was determined by the coordinates of the reflective markers, and then the joint angles of each segment were calculated. Finally, the joint kinetics (joint moment) was calculated by the inverse kinetics algorithm in Visual 3-D software. All joint kinematics and joint kinetics data were then imported into MATLAB R2019a (Visual R2019a, MathWorks, United States) to process further. For each joint (ankle, knee, hip) of each plane (sagittal, frontal, transverse) of kinematics (joint angle) and kinetics (joint moment) data, all were extracted to expand into 100 data point curves by custom MATLAB script. Finally, two matrices can be obtained: Data analysis Neural networks are widely parallel networks of adaptive simple units whose organization can simulate the interactions of biological nervous systems to real-world objects30. Neural networks with more than two hid- den layers are defined as deep neural networks, and deep neural network (DNN) is generally considered to improve the accuracy of the whole model31. The application of the DNN model in this study was mainly biased to improve the accuracy of the model, so a DNN model with ten hidden layers was designed under the condition of repeated model training and adjustment according to the actual data. The matrices Mkinematics , Mkinetics , and M = Mkinematics + Mkinetics was conducted using Layer-wise Relevance Propagation (LRP) respectively. Firstly, a deep neural network (DNN) was established that included one input layer, ten hidden layers, and one output layers, and the per layer nodes were determined by the input data shape32. Therefore, for the dataset Mkinematics and Mkinetics , the nodes of the input layer, hidden layers, and output layer were 900, 1800, and 2. For the dataset M , the nodes of the input layer, hidden layers, and output layer were 1800, 3600, and 2. As shown in Fig. 4A, the layers of the neural network are fully connected, which means the neuron of the n-th layers must be connected to the neuron of the (n + 1) -t h layer. A linear relation function and an activation function were used to calculate the new values between layers, and the linear relationship function of the model constructed in this study was The wi is the connection weight of the i-th neuron, and the xi is the input from the i-th neuron. The hidden layer activation function was used the hyperbolic tangent function The batch size was set 25, and the epoch limit was set 3000. At the same time, the data of the higher-mileage runner was set at positive class, and the data of the lower-mileage runner was set to negative class. Before the (1) RHJC = (0.36 ∗ ASISDistance, −0.19 ∗ ASISDistance, −0.3 ∗ ASIS_Distance) (2) LHJC = (−0.36 ∗ ASISDistance, −0.19 ∗ ASISDistance, −0.3 ∗ ASIS_Distance) Mkinematics = 800  80 subjects ∗ 10 trials  ∗ 900  3 joint ∗ 3 plane ∗ 100 data points  Mkinetics = 800  80 subjects ∗ 10 trials  ∗ 900  3 joint ∗ 3plane ∗ 100 data points  (3) z = m  i=1 wixi + b (4) gx = ex − e−x ex + e−x 8 Vol:.(1234567890) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ data training, the 10 data sets of successful trials for each subject were taken as a whole, and then randomly extracted the data sets of 32 higher-mileage and 32 lower-mileage runners as training sets (a total of 640 sample data sets), the remaining data sets of 8 higher-mileage and 8 lower-mileage runners as test sets (a total of 160 sample data sets). Following DNN training, the relevance score was calculated by the LRP, and the performance of the classifier was evaluated by the accuracy achieved and other parameters. Layer‑wise relevance propagation. Layer-wise Relevance Propagation (LRP) is technology used to identify important relevance through backward propagation in neural networks. Backward propagation is a conservative relevance redistribution process in which the neurons that contribute the most to the upper layer receive the most relevance from the upper layer. In general, LRP aims to narrow the gap between the classifica- tion and interpretability of multi-layer neural networks on nonlinear cores22,23. The overall idea is to understand the contribution of a single feature of dataset x to the prediction f (x) made by the classifier f in pattern recognition and classification tasks. That is, the positive or negative contribution of each feature to the classification result for dataset x can be calculated, and the degree of such contribution can be accurately measured to a certain extent (The contribution of each input feature x(d) to a particular predic- tion f (x) . In the setting of the classifier is a mapping f : Rv → R1 , f (x) > 0 indicates the existence of a learning structure. The constraint of classification is to find the differential contribution relative to the most uncertain state of the classification, which is then represented by the root point f (x0) = 0 . By factoring the prediction f (x) into the sum of the individual input feature x(d): In the classifier, whether for nonlinear support vector machines or neural networks, the first layer is the input features, and the last layer is the predicted output of the classifier. Meanwhile, each layer is part of the features extracted from the dataset x after running the classification algorithm. The l-th layer is modeled as a vector z =  zl d V(l) d=1 with dimensionality V(l) . LRP has a relevance score R(l+1) d for each dimension z(l+1) d of vector z (5) f (x) = V  d=1 Rd Figure 4. (A) A description of the neurons and weight connections of the DNN by the interpretation of the different variables and indices from multilayers. Left is the process of establishing f (x) by forward pass of DNN. Right is the process of calculating relevance score R(1) d by LRP back pass. On the upper right side is the algorithm summary about the complete LRP procedure for DNN. (B) A description of the confusion matrix of binary classifier. 9 Vol.:(0123456789) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ at layer l + 1 . A relevance score R(l) d is found in each dimension zl d of vector z near the next layer l of the input layer, as shown in the following formula: The inter-hierarchical relevance is represented by the message Rl,l+1 i←j between neuron i and j , and these mes- sages can be sent along with each connection. As shown in Fig. 4A, the output f (x) is then passed from one neuron to the next by backward propagation. The relevance of neurons is defined as the sum of incoming mes- sages, then the sum runs over the sinks at layer l + 1 for a fixed neuron i at a layer l. The Input of the next neuron in the direction defined during classification, then the sum runs over the sources at layer l for a fixed neuron k at layer l + 1 . In general, this can be expressed as: The relevance of each layer is calculated by backward propagation: the relevance R(l) i is expressed as a func- tion of the upper relevance R(l+1) j , and back propagates the relevance until the input feature is reached. As shown in Fig. 4A, through the relevance of the neuron R(l+1) j to the classification decision f (x) , the relevance is then decomposed according to the message Ri←j sent to the upper layer of neurons. Holding the conservation property: For the linear network f (x) =  i zij , the relevance is Rj = f (x) , and the decomposition directly by Ri←j = zij . Through hyperbolic tangent function and rectification function two monotone increasing functions, the pre- activation function zij provides a reasonable way to measure the relative contribution of xi to Rj for each neuron. Based on the proportion of local pre-activation and global pre-activation, the selection of association decomposi- tion is obtained: The relevance Ri←j are shown in: Multiplier accounts represent the relevance absorbed by the bias term, and the residual bias correlations can be reassigned to each neuron xi . According to the determined rule (Eq. 10), through adding up the correlations of all neurons in the upper layer i (combined Eqs. (7) and (8)), the overall relevance of all neurons in the next layer j can be obtained: The relevance propagates from one layer to another until it reaches the input feature x(d) , where the relevance R(1) d provides the hierarchical eigen-decomposition required for the decision f (x) . The upper right side of Fig. 4A summarized the algorithm of the complete LRP procedure for DNN. More details can be found by referring to Lapuschkin et al22. All algorithms were run in MATLAB R2019a (Natick, Massachusetts: The MathWorks Inc.), through self-written scripts according to the layer-wise relevance propagation toolbox33. The relevance of correctly classified gait patterns was extracted by defining logical variables, and then a rel- evance score was assigned to each input variable. LRP determines the correlation between each variable and the predicted results of the model, and normalizes the LRP-derived association patterns to their respective maximum values for comparison. After then, the average of all relevant patterns was determined and the error was rectified. The rectified average was smoothed, whereby the present point was weighted with 50%, and the previous and following points were weighted with 25%. For the smoothing process, the weighted values were set such that their total equaled 1 and a repetition of the procedure would approximate a Gaussian filter. Each of these steps was performed three times to get the desired result. Finally, the smoothed correlation pattern was rescaled from 0 (no correlation) to 1 (the highest correlation)9. Since the input variables are collected in the time domain, and the adjacent values are interdependent, the fluctuation of the relevance score can be reduced by smoothing. To explore the influence of different variables on the accuracy of model classification, all variables were sorted according to the correlation between variables, and then the top 200 variables with the highest relevance scores were selected to explain and analyze the gait pattern. (6) f (x) = · · · =  d∈l+1 Rl+1 d =  d∈l Rl d = · · · =  d R1 d (7) R(l) j =  k:i is input for neuron k R(l,l+1) i←k (7) (8) R(l+1) k =  i:i is input for neuron k R(l,l+1) i←k (9)  i R(l,l+1) i←j = R(l+1) j (10) Rl,l+1 i←j = zij zj ∗ Rl+1 j (11)  i R(l,l+1) i←j = R(l+1) j ∗  1 − bj zj  (12) R(l) i =  j R(l,l+1) i←j 10 Vol:.(1234567890) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ Evaluate the performance of the classifier. Combine the results of the classification model into a 2 ∗ 2 table called confusion matrix m =  TP FN FP TN  (more details are shown in Fig. 4B) which fully describes the results of the classification task34. Then, the following indicators were calculated to evaluate the performance of the classifier. 1. The accuracy of a classifier on a given set of tests is the percentage of tuples that are correctly classified by the classifier: 2. The sensitivity (also called recall) is the true positive cases recognition rate, which means the percentage of positive tuples correctly identified: 3. The specificity is the true positive cases recognition rate, which means the percentage of negative tuples correctly identified: 4. The precision is a measure of accuracy, which means the percentage of tuples marked as positive that are actually positive: 5. F1 − score is the harmonic average of accuracy and recall rate, which means the recall rate is weighted once as much as the precision: 6. Receiver Operating Characteristic (ROC) curves is a useful visual tool for comparing classifier models, which can provide objective and neutral advice regardless of cost/benefit when making decisions. The ROC curve shows the tradeoff between the true positive rate (TPR) and the false positive rate (FPR) for the classifier model. The increase in TPR comes at the expense of the increase in FPR: The Y-axis of the ROC curve represents TPR and the X-axis represents FPR, and the area under the ROC curve ( AUC ) is a measure of model accuracy: 7. Matthew’s correlation coefficient (MCC) is a contingency matrix method34. MCC can be used to calculate the Pearson product-moment correlation coefficient 35 between the actual value and the predicted value: Received: 19 September 2021; Accepted: 8 February 2022 References 1. Van Gent, R. et al. Incidence and determinants of lower extremity running injuries in long distance runners: a systematic review. Br. J. Sports Med. 41, 469–480 (2007). 2. Saragiotto, B. T., Yamato, T. P. & Lopes, A. D. What do recreational runners think about risk factors for running injuries? A descrip- tive study of their beliefs and opinions. J. Orthop. Sports Phys. 44, 733–738 (2014). 3. Van der Worp, M. P. et al. Injuries in runners; a systematic review on risk factors and sex differences. PLoS ONE 10, e0114937 (2015). 4. Taunton, J. et al. A prospective study of running injuries: the Vancouver Sun Run “In Training” clinics. Br. J. Sports Med. 37, 239–244 (2003). 5. Hootman, J. M. et al. Predictors of lower extremity injury among recreationally active adults. Clin. J. Sports Med. 12, 99–106 (2002). accuracy = TP + TN P + N sensitivity/recall = TP TP + FN specificity = TN FP + TN precision = TP TP + FP F1 − score = 2 ∗ precision ∗ recall precision + recall TPR = TP TP + FN FPR = FP FP + TN AUC = (TPR − FPR + 1) 2 MCC = TP ∗ TN − FP ∗ FN √(TP + FP) ∗ (TP + FN) ∗ (TN + FP) ∗ (TN + FN) 11 Vol.:(0123456789) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ 6. Van Middelkoop, M., Kolkman, J., Van Ochten, J., Bierma Zeinstra, S. & Koes, B. W. Risk factors for lower extremity injuries among male marathon runners. Scand. J. Med Sci. Sports 18, 691–697 (2008). 7. Wen, D. Y., Puffer, J. C. & Schmalzried, T. P. Lower extremity alignment and risk of overuse injuries in runners. Med. Sci. Sports Exerc. 29, 1291–1298 (1997). 8. Messier, S. P., Davis, S. E., Curl, W. W., Lowery, R. B. & Pack, R. J. Etiologic factors associated with patellofemoral pain in runners. Med. Sci. Sports Exerc. 23, 1008–1015 (1991). 9. Hoitz, F., von Tscharner, V., Baltich, J. & Nigg, B. M. Individuality decoded by running patterns: movement characteristics that determine the uniqueness of human running. PLoS ONE 16, e0249657 (2021). 10. Horst, F., Lapuschkin, S., Samek, W., Müller, K.-R. & Schöllhorn, W. I. Explaining the unique nature of individual gait patterns with deep learning. Sci. Rep. 9, 1–13 (2019). 11. Boyer, K. A., Silvernail, J. F. & Hamill, J. The role of running mileage on coordination patterns in running. J. Appl. Biomech. 30, 649–654 (2014). 12. Clermont, C. A., Phinyomark, A., Osis, S. T. & Ferber, R. Classification of higher-and lower-mileage runners based on running kinematics. J. Sports Health Sci. 8, 249–257 (2019). 13. Xu, D., Lu, J., Baker, J. S., Fekete, G. & Gu, Y. Temporal kinematic and kinetics differences throughout different landing ways fol- lowing volleyball spike shots. Proc. Inst. Mech. Eng. P: J. Sports. Eng. Technol. https:// doi. org/ 10. 1177/ 17543 37121 10094 85 (2021). 14. Xu, D., Jiang, X., Cen, X., Baker, J. S. & Gu, Y. Single-leg landings following a volleyball spike may increase the risk of anterior cruciate ligament injury more than landing on both-legs. Appl. Sci. 11, 130 (2021). 15. Chau, T. A review of analytical techniques for gait data. Part 2: neural network and wavelet methods. Gait Posture 13, 102–120 (2001). 16. Schöllhorn, W. I. Applications of artificial neural nets in clinical biomechanics. Clin. Biomech. 19, 876–898 (2004). 17. Jauhiainen, S., Pohl, A. J., Äyrämö, S., Kauppi, J. P. & Ferber, R. A hierarchical cluster analysis to determine whether injured run- ners exhibit similar kinematic gait patterns. Scand. J. Med. Sci. Sports 30, 732–740 (2020). 18. Phinyomark, A., Petri, G., Ibáñez-Marcelo, E., Osis, S. T. & Ferber, R. Analysis of big data in gait biomechanics: current trends and future directions. J. Med. Biol. Eng. 38, 244–260 (2018). 19. Figueiredo, J., Santos, C. P. & Moreno, J. C. Automatic recognition of gait patterns in human motor disorders using machine learn- ing: a review. Med. Eng. Phy. 53, 1–12 (2018). 20. Baehrens, D. et al. How to explain individual classification decisions. J. Mach. Learn. Res. 11, 1803–1831 (2010). 21. Montavon, G., Samek, W. & Müller, K.-R. Methods for interpreting and understanding deep neural networks. Digit. Signal Process. 73, 1–15 (2018). 22. Bach, S. et al. On pixel-wise explanations for non-linear classifier decisions by layer-wise relevance propagation. PLoS ONE 10, e0130140 (2015). 23. Montavon, G., Binder, A., Lapuschkin, S., Samek, W. & Müller, K.-R. Layer-wise relevance propagation: an overview. Explainable AI: interpreting, explaining and visualizing deep learning, 193–209 (2019). 24. Samek, W., Binder, A., Montavon, G., Lapuschkin, S. & Müller, K.-R. Evaluating the visualization of what a deep neural network has learned. IEEE T. Neural Netw. Learn. 28, 2660–2673 (2016). 25. Xu, D., Zhou, H., Baker, J., István, B. & Gu, Y. An investigation of differences in lower extremity biomechanics during single-leg landing from height using bionic shoes and normal shoes. Front. Bioeng. Biotechnol. 9, 679123. https:// doi. org/ 10. 3389/ fbioe. 2021. 679123 (2021). 26. Quan, W., Ren, F., Sun, D., Fekete, G. & He, Y. Do novice runners show greater changes in biomechanical parameters?. Appl. Bionics Biomech. 8, 1–8 (2021). 27. Xu, D. et al. The Differences in lower extremity joints energy dissipation strategy during landing between athletes with symptomatic patellar tendinopathy (PT) and without patellar tendinopathy (UPT). Mol. Cell. Biomech. 18, 107–118 (2021). 28. Winter, D. A. Biomechanics and motor control of human movement (John Wiley & Sons, Hoboken, 2009). 29. Bell, A. L., Pedersen, D. R. & Brand, R. A. A comparison of the accuracy of several hip center location prediction methods. J. Biomech. 23, 617–621 (1990). 30. Kohonen, T. An introduction to neural computing. Neural Netw. 1, 3–16 (1988). 31. Liu, W. et al. A survey of deep neural network architectures and their applications. Neurocomputing 234, 11–26 (2017). 32. Binder, A., Bach, S., Montavon, G., Müller, K.-R. & Samek, W. Layer-Wise Relevance Propagation for Deep Neural Network Architectures. In Information science and applications (ICISA) 2016 (eds Kim, K. J. & Joukov, N.) 913–922 (Springer, Singapore, 2016). 33. Lapuschkin, S., Binder, A., Montavon, G., Müller, K.-R. & Samek, W. The LRP toolbox for artificial neural networks. J. Mach. Learn. Res. 17, 3938–3942 (2016). 34. Chicco, D. & Jurman, G. The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. BMC Genomics 21, 1–13 (2020). 35. Powers, D. M. Evaluation: from precision, recall and F-measure to ROC, informedness, markedness and correlation. arXiv preprint http:// arxiv. org/ abs/ 2010. 16061 (2020). 36. Nigg, B. M., Baltich, J., Maurer, C. & Federolf, P. Shoe midsole hardness, sex and age effects on lower extremity kinematics during running. J. Biomech. 45, 1692–1697 (2012). 37. Almeida, M. O., Davis, I. S. & Lopes, A. D. Biomechanical differences of foot-strike patterns during running: a systematic review with meta-analysis. J. Orthop. Sports Phys. 45, 738–755 (2015). 38. Dempster, J., Dutheil, F., & Ugbolue, U. C. The Prevalence of Lower Extremity Injuries in Running and Associated Risk Factors: A Systematic Review. Phys. Act. Health 5, 133–145 (2021). 39. Slijepcevic, D. et al. On the Explanation of Machine Learning Predictions in Clinical Gait Analysis. arXiv preprint http:// arxiv. org/ abs/ 1912. 07737 (2019). 40. Padulo, J., Annino, G., Migliaccio, G. M., D’Ottavio, S. & Tihanyi, J. Kinematics of running at different slopes and speeds. J. Strength Cond. Res. 26, 1331–1339 (2012). Author contributions Conception: D.X., Y.G. Design and perform the experiment: D.X., W.Q., H.Z. Critical interpretation of the data: all authors. Manuscript drafting, critical revision prior to the submission: all authors. Funding This study was sponsored by the National Natural Science Foundation of China (No. 81772423), Key Project of the National Social Science Foundation of China (19ZDA352), Key R&D Program of Zhejiang Province China (2021C03130), and K. C. WongMagna Fund in Ningbo University. 12 Vol:.(1234567890) Scientific Reports | (2022) 12:2981 | https://doi.org/10.1038/s41598-022-07054-1 www.nature.com/scientificreports/ Competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to J.S.B. or Y.G. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2022

Explaining the differences of gait patterns between high and low-mileage runners with machine learning.

02-22-2022

Xu, Datao,Quan, Wenjing,Zhou, Huiyu,Sun, Dong,Baker, Julien S,Gu, Yaodong

eng

PMC6093601

RESEARCH ARTICLE Effects of short-term in-season break detraining on repeated-sprint ability and intermittent endurance according to initial performance of soccer player Alejandro Rodrı´guez-Ferna´ndez1,2,3, Javier Sa´nchez-Sa´nchez1,3, Rodrigo Ramirez- Campillo3,4, Jose´ Antonio Rodrı´guez-Marroyo1, Jose´ Gerardo Villa Vicente1,3, Fabio Yuzo Nakamura3,5,6* 1 Institute of Biomedicine (IBIOMED), Department of Physical Education and Sports, University of Leo´n, Leo´n, Spain, 2 Faculty of Physical Activity Sciences and Sports, University Isabel I, Burgos, Spain, 3 Unit Assessment and Planning of Sports Training, Faculty of Education, Pontifical University of Salamanca, Salamanca, Spain, 4 Department of Physical Activity Sciences, Universidad de Los Lagos, Osorno, Chile, 5 Department of Medicine and Aging Sciences “G. d´Annunzio” University of Chieti-Pescara, Chieti, Italy, 6 The College of Healthcare Sciences, James Cook University, Townsville, Queensland, Australia * fabioy_nakamura@yahoo.com.br Abstract To better understand the detraining effects in soccer, the purpose of the study was to ana- lyse if performance level of soccer players modulate repeated-sprint ability (RSA) and inter- mittent endurance changes during 2-weeks of detraining (i.e., in-season break). Seventeen professional and sixteen young elite soccer players of two different teams performed, before and after 2-weeks of detraining, the RSA test and the Yo-Yo Intermittent Recovery Test, level 1 (YYIR1). Before detraining, professional players perform better (p < 0.05) RSA best time (RSAbest) than young players. A decrease (p < 0.05) in RSAbest, RSA total time (RSAtotal) and mean time (RSAmean) performance was observed in both teams, without changes in RSA fatigue index (Sdec). No significant changes in distance covered during YYIR1 was observed in any team. Before detraining, faster players from both teams (FG) (following the median split technique, soccer players with RSAbest  3.95 s) performed bet- ter (p < 0.01) in RSAtotal, RSAmean and RSAbest, but worse (p < 0.01) in Sdec. Although FG and the slower players (SG, RSAbest > 3.95 s) showed a worse (p < 0.05) RSAtotal, RSAbest and RSAmean performance after detraining (ES = 1.5, 1.4 and 2.9; ES = 0.6, 1.2 and 0.6; for FG and SG, respectively), the deterioration was greater in the FG for RSAbest (p < 0.05) and RSAtotal (ES = 1.46). After detraining, FG improved (p < 0.05) Sdec performance. In conclu- sion, a 2-week in-season break (detraining) period induced a worse RSA, with no effect on intermittent endurance in professional and elite young soccer players, with greater detrimen- tal effects on RSAtotal and RSAbest in FG. In addition, Sdec does not seem to be sensitive to changes in RSA after a 2-week in-season break. PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 1 / 10 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Rodrı´guez-Ferna´ndez A, Sa´nchez-Sa´nchez J, Ramirez-Campillo R, Rodrı´guez-Marroyo JA, Villa Vicente JG, Nakamura FY (2018) Effects of short-term in-season break detraining on repeated- sprint ability and intermittent endurance according to initial performance of soccer player. PLoS ONE 13(8): e0201111. https://doi.org/10.1371/journal. pone.0201111 Editor: Johnny Padulo, National Center of Medicine and Science in Sport, TUNISIA Received: August 8, 2017 Accepted: June 11, 2018 Published: August 15, 2018 Copyright: © 2018 Rodrı´guez-Ferna´ndez et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: The authors received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. Introduction The ability to perform short-duration multiple sprints interspersed with short recovery times has been termed “repeated-sprint ability” (RSA) [1]. Although debate exists regarding the main factors determining soccer performance [2], the importance of RSA is recognized [3,4]. For instance, significant correlation exists between very-high intensity running distances cov- ered during matches and mean sprint time on a RSA test [5]. Besides this, single and repeated- sprint efforts are frequently involved in crucial moments of match-play [6], including creation of goal scoring opportunities. Therefore, constant evaluation of RSA throughout the season can provide valuable information to coaches and athletes. In addition to RSA, intermittent high-intensity endurance is also considered crucial to performance in soccer [7]. Although the importance of total running distance covered at high-intensity in soccer could be masked by the technical-tactical level of a team [8], players at a higher standard of competition tend to perform significantly more high-intensity run- ning than those at a lower standard [9], and this ability can be assessed by the Yo-Yo inter- mittent recovery test, level 1 (YYIR1) [10]. Previous studies have shown no effects of detraining after one week in the Yo-Yo Intermittent Recovery Test, level 2 [11]. However, significant (p < 0.01) detrimental effects of ~5 and ~22% after three days and 2-weeks of inactivity, respectively, was reported in a study [12]. Aside from the lack of agreement across studies, to our knowledge, no study has analyzed the effects of short-term detraining after an in-season break using the YYIR1 as a marker of performance. Despite YYIR1 and YYIR2 test performances correlate very largely [13], the lower level of speed effort required by the YYIR1 might better detect changes in aerobic fitness of players after a detraining period. The effects of different training programs on RSA have been assessed in soccer players [14– 17]. However, the impact of in-season unloading or detraining on RSA in soccer players remains unresolved. For instance, previous studies have analyzed the effects of 1- [11] or 2-week [12,18] off-season detraining periods on RSA (complete interruption of training), showing detrimental effects of ~2% after two weeks for RSA total time and ~3% for sprinting speed in the last five repetitions during a 20-m RSA test with a total of 10 repetitions [18]. However, RSA best time and RSA fatigue index were not affected after 2-weeks of inactivity [12,18]. Possibly, the loading pattern (e.g., deliberate overload leading to overreaching) and the athlete’s performance level previous to detraining might modulate such changes. A better understanding of this phenomenon is relevant since most soccer leagues encompass a period without competitive matches [19] or in-season break detraining period. Similar to off-season breaks, in-season breaks can also lead to short-term detraining [12,18,20]. This may induce cardiovascular and neuromuscular deconditioning [21], which can potentially impair RSA. Notably, RSA is related to chronological age [1], competitive level [4] and intermittent high-intensity endurance [22]. However, the interaction between detrain- ing, RSA, age, competitive level, and intermittent high-intensity endurance is unknown and needs to be clarified. RSA is usually assessed by the total time (RSAtotal) [23], mean time (RSAmean) [4,24], best time (RSAbest) [14], and fatigue index or the percentage decrement score (Sdec) [25]. Although Sdec is considered a reliable RSA marker [26], recent research has suggested that “absolute” performance values (i.e., total, mean and best times) can be more reliable and sensitive to training effects [4,27]. In response to detraining Sdec showed contradictory results, with signif- icant (p = 0.04) detrimental changes (5.8 ± 2.8% to 7.8 ± 3.2%) [11] or no significant changes (5.9 ± 2.3% to 7.6 ± 2.8%) [12] after 1 or 2-week of detraining, respectively. The knowledge of the extent of RSA responses to a short-term detraining period can help physical trainers to Effects of detraining on repeated-sprint ability and intermittent endurance PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 2 / 10 foresee eventual changes during the in-season breaks, and to implement adequate training strategies to optimize fitness levels and training time after returning from the break. Therefore, the main aim of the study was to analyse if initial performance level (sprinting speed) of soccer players modulates repeated-sprint ability (RSA) and intermittent endurance changes during 2-weeks of detraining (i.e., in-season break). A substantial detraining effect for both professional senior and young elite soccer players and the influence of “baseline” fitness level (pre-detraining) on this effect were assumed as the working hypotheses of this study. Material and methods Study design The 2-week in-season break period took place during the mid-phase of the season (i.e., Christ- mas holidays), with players not involved in any competitive game or training session (team or individual session) during this period. Specifically, players were asked to refrain for any type of physical activity training (other than daily life physical activity) during the in-season break. After the break, players were individually interviewed to assess the level of accomplishment of the aforementioned requirement. In addition, players were asked to reduce meaningful changes in their diet, although this was not controlled. The same training week was repro- duced before and after the 2-week Christmas break (Table 1). The tests were imbedded into the training sessions, so that there was no disturbance to the training plan. During day one, anthropometry (ISAK procedure) and RSA test were performed (16:30 to 18:30 pm) and, dur- ing day two, the YYIR1 was completed (16:30 pm) and used to estimate VO2max of soccer players [13]. Prior to the two exercise tests, soccer players performed the same warm-up (i.e., low-intensity running, dynamic flexibility, 20-m run-ups). Participants were fully familiarized with testing protocols, which were routinely performed in the respective investigated clubs. Athletes refrained from vigorous high-intensity exercise 24 hours before testing sessions and were instructed to maintain normal daily food and water intake and to avoid any leisure sport activity or self-administered exercise throughout the study period. Players were required to wear their usual training uniforms and football boots during the tests, performed in their respective habitual training venues. Participants Male professional senior (n = 17, PT; age: 24.0 ± 2.8 years; height: 179.6 ± 1.8 cm; body mass: 74.5 ± 4.6 kg; VO2max: 58.29 ± 3.0 ml
kg-1
min-1) and young soccer players (n = 16, YT; age: 18.3 ± 0.8 years; height: 173.5 ± 9.9 cm; body mass: 65.4 ± 1.3 kg; VO2max: 54.65 ± 2.1 ml
kg-1
min-1) volunteered for the investigation. Players from both teams had a minimum soc- cer experience of 7 years at the commencement of the study. Bout groups performed three (YT) to four (PT) training sessions and one national-level match per week, in the three months preceding this study. Written informed consent was signed by all players (and their parents or Table 1. Schematic representation of a training week before the intervention period in young (YT) and professional (PT) soccer players. Monday Tuesday Wednesday Thursday Friday Saturday Sunday YT Strength/power and injury prevention RSA test, small sided games (4 vs 4 to 6 vs 6), aerobic power and tactical drills Yo-Yo test, speed/reaction soccer and tactical work game Official match PT Strength/power and injury prevention RSA test, small sided games (4 vs 4 to 6 vs 6), aerobic power and tactical drills Strength/power and tactical work game (match simulation) Yo-Yo test, speed/reaction soccer and strategy drills Activation Official match RSA: repeated-sprint ability https://doi.org/10.1371/journal.pone.0201111.t001 Effects of detraining on repeated-sprint ability and intermittent endurance PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 3 / 10 guardians for under 18 years of age athletes) after a brief but detailed explanation about the aims, benefits, and risks involved with this investigation. The study was conducted according to the Declaration of Helsinki and the Institutional Research Ethics Committee (Universidad Isabel I) granted approval for the study. Yo-Yo intermittent recovery test Athletes completed the YYIR1 test as previously described [26]. Running speed was set through an acoustic signal amplified by speakers (SONY-ENG2031, Japan), connected to a notebook (Acer TravelMater 57201, Taiwan). Maximal heart rate was measured with a heart rate moni- tor (Polar Team Sport System1, Polar Electro Oy, Finland) and total distance was recorded by the number of runs. The test was finished when i) subjects were unable to complete two conse- cutive 20-m runs at the pace dictated by the acoustic signal, or ii) when athletes achieved voli- tional exhaustion [28]. Subjects were instructed and motivated to achieve maximal effort during testing (validated by the achievement of ±10% of predicted maximal heart rate). Repeated-sprint ability test The RSA test involved eight maximal 30-m sprints, separated by 25 seconds of active recovery between sprints. Approximately two seconds before each sprint subjects assumed the start position [14] with the front foot placed 0.5 m behind the first photocell (DSD Laser System1, Leon, Spain), as previously described [23]. Immediately after the warm-up, each player com- pleted a single criterion-reference sprint and this trial was used as the criterion score during the subsequent sprints [4]. Then, athletes rested for 5 minutes before commencement of the RSA test. If the first sprint-time of the RSA test was 2.5% greater (i.e., worse) than the criterion-refer- ence sprint time, subjects were requested to rest for further 5 minutes and then restart the test. The RSAbest, RSAmean, RSAtotal and Sdec [4,5,14] were determined. The Sdec was calculated as (RSAtotal/(RSAbest × total number of sprints) × 100)– 100 [26]. Statistical analyses The results are expressed as mean ± standard deviation (SD). The change in tests performance is presented as percentage (Δ = [after value—before value] / before value). In addition to the comparison between PT and YT, the median split technique was used to divide the pooled par- ticipants into fast (FG) and slow (SG) performers, according to the median value calculated to RSAbest [5,27]. Normal distribution of data was confirmed by using Kolmogorov-Smirnov test and normal probability plot. A two-way ANOVA with repeated measures on time [before vs. after intervention] × team [professional vs. young players]) and time [before vs. after interven- tion] × group [fast vs. slow players] was used to analyse results. When a significant F value was found, Bonferroni’s post hoc test was applied to establish differences between means. Cohen’s d effect size (ES) was calculated and qualitatively assessed as trivial (0–0.19), small (0.20–0.49), medium (0.50–0.79) and large (0.80 and greater) [28]. The p<0.05 criterion was used to estab- lish statistical significance. Analyses were performed using standard statistical software (SPSS, v.17.0, Chicago, Illinois, USA). Results Yo-Yo intermittent recovery test The YYIR1 performance was not significantly affected (p > 0.05) by the in-season break (PT: before 2,368 ± 265 m, after 2,256 ± 283 m ES = 0.42; YT: before 2,054 ± 289 m, after 1,986 ± 321 m ES = 0.23). Effects of detraining on repeated-sprint ability and intermittent endurance PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 4 / 10 Repeated-sprint ability test Regarding the effects of in-season break according to playing level, before detraining, PT showed significantly (p < 0.05) better RSAbest than YT (~ 2.8%) (Table 2). The RSAbest, RSAmean and RSAtotal performance was worsened after the in-season break in both YT (p < 0.05) and PT (p < 0.01) (Table 2), with no difference between PT and YT. Regarding the effects of in-season break according to baseline RSA performance, the median of the RSAbest was 3.95 s (27.3 km
h-1), categorizing players into FG (< 3.95 s) and SG ( 3.95 s) (Table 3). The FG group (n = 14) was composed of 8 professional and 6 young soc- cer players. The SG (n = 19) was composed of 9 professional and 10 young soccer players. Before detraining, the FG showed better performances in RSAbest (~ 6.8%), RSAmean (~ 4.1%), and RSAtotal (~ 4.1%), (p < 0.01; Table 3), confirming the appropriateness of the median split technique. The players from the FG showed a greater impairment in the RSAbest (ES = 2.09), RSAmean (ES = 1.04) and RSAtotal (ES = 1.46) than the players from the SG (ES = 0.56, 1.02, 0.58, respectively) (Table 3). Table 2. RSA in professional (PT; n = 17) and young elite (YT; n = 16) soccer players. Before After Δ ES RSAbest (s) YT 4.03 ± 0.15 4.11 ± 0.14  1.9 ± 3.0 1.03 (large) PT 3.92 ± 0.11 † 4.04 ± 0.13  3.0 ± 2.7 1.03 (large) RSAmean (s) YT 4.19 ± 0.12 4.26 ± 0.17  1.7 ± 2.6 0.65 (medium) PT 4.12 ± 0.12 4.22 ± 0.12  2.3 ± 2.6 1.03 (large) RSAtotal (s) YT 33.52 ± 0.97 34.12 ± 1.40  1.7 ± 2.6 0.51 (medium) PT 32.91 ± 0.91 33.80 ± 0.94  2.3 ± 2.6 1.03 (large) Sdec (%) YT 3.90 ± 1.65 3.69 ± 1.61 -0.21 ± 2.2 0.13 (trivial) PT 5.21 ± 1.91 4.48 ± 2.14 -0.73 ± 2.4 0.36 (small) RSA: repeated sprint ability; RSAbest, RSAmean and RSAtotal: best, mean and total time in the RSA test, respectively; Sdec: percentage decrement score; Δ: percentage change; ES: effect size. , : denote difference compared with before values (p < 0.05 and p < 0.01, respectively); †: denote difference between teams (p < 0.05). https://doi.org/10.1371/journal.pone.0201111.t002 Table 3. RSA in slow (SG, n = 19) and fast (FG, n = 14) performers€. Before After Δ ES RSAbest (s) SG 4.08 ± 0.02 4.14 ± 0.03  1.5 ± 2.6 0.56 (medium) FG 3.82 ± 0.02 †† 3.98 ± 0.04  4.0 ± 2.5 † 2.09 (large) RSAmean (s) SG 4.23 ± 0.02 4.31 ± 0.03  2.0 ± 2.6 1.02 (large) FG 4.06 ± 0.03 †† 4.16 ± 0.04  2.2 ± 2.5 1.04 (large) RSAtotal (s) SG 33.81 ± 0.19 34.49 ± 0.26  2.0 ± 2.6 0.58 (medium) FG 32.45 ± 0.7 †† 33.21 ± 0.29  2.2 ± 2.5 1.46 (large) Sdec (%) SG 3.64 ± 0.52 4.25 ± 2.01 0.5 ± 1.9 0.29 (small) FG 5.91 ± 1.92 †† 3.93 ± 1.84  -2.0 ± 2.2 † 1.13 (large) €: the median split technique was used to divide subjects into FG and SG performers, according to median RSAbest of 3.95 s; RSA: repeated sprint ability; RSAbest, RSAmean and RSAtotal: best, mean and total time in the RSA test, respectively; Sdec: percentage decrement score; Δ: percentage change; ES: effect size. , : denote difference compared with before values (p < 0.05 and p < 0.01, respectively); †, ††: denote difference between SG and FG (p < 0.05 and p < 0.01, respectively). https://doi.org/10.1371/journal.pone.0201111.t003 Effects of detraining on repeated-sprint ability and intermittent endurance PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 5 / 10 Discussion The main aim of the study was to analyse if initial performance level (sprinting speed) of soc- cer players can modulate repeated-sprint ability (RSA) and intermittent endurance changes during 2-weeks of detraining (i.e., in-season break). The major findings showed detrimental changes in RSA, but not intermittent endurance performance, in both PT and YT. Also, RSAbest was greatly impaired in FG compared to SG players (p<0.05, with a large 2.09 ES ver- sus medium 0.56 ES), as well as RSAtotal (large 1.46 ES versus medium 0.58 ES). To our knowledge, this is the first controlled study showing the response of YYIR1 and RSA performance after an in-season break. However, a previous study [29] analysed the effects of a 2-weeks in-season break on male professional (age, 24.3 ± 4.2 years) Australian Football player’s fitness, including skinfolds, and heart rate plus rating of perceived exertion while plyers performed submaximal running velocity (12 km
h-1), high-intensity intermittent run- ning exercise, and a standardized handball game. The aforementioned study showed increased levels of strength and cardiorespiratory fitness, despite a small increase in skinfold thickness. However, the athletes did not fully stop training, contrary to our study where soccer players interrupted training completely. In addition, in the aforementioned study, the authors did not control the training that athletes completed during the break. It is possible that the break allowed players to come back in January well recovered with preserved or even increased levels of strength and cardiorespiratory fitness. According to our results, a complete reduction of training during 2-weeks of in-season break did not affect YYIR1 performance in PG or YG. Accordingly, high-intensity intermittent endurance performance might be more resilient to detraining [30] than some of its physiologi- cal correlates, such as maximal oxygen uptake (VO2max) [31], and other factors affecting high-intensity intermittent endurance performance [13]. Of note, FG players showed greater impairment in RSAbest performance after detraining compared to SG players. Due to their greater initial performance level, FG players may have experienced greater detraining effects [31], with increased negative effects on fast-twitch mus- cle fibers [32], the ability to use ATP and phosphocreatine [33], accompanied with greater pro- duction of metabolic by-products [34], which may negatively affect motor units recruitment and synchronization [35], and thus RSA [36]. In a similar study, faster futsal players assessed at the beginning of the pre-season also lost more of their sprinting speed than their slower peers [37]. In summary, independent from age, compared to slower players, faster players seem to be more negatively affected in their RSA by detraining. Therefore, fast team sports players may need more attention from technical staffs due to their tendency to lose their sprint quality in several phases of the preparation. The FG players showed an enhanced Sdec performance after detraining. Although tapering effects may help explain this result [38], no other RSA value was improved after detraining. Previous studies have found worsening (5.8 ± 2.8% vs 7.8 ± 3.2%, p = 0.04) or maintenance (5.9 ± 2.3% vs 7.6 ± 2.8%) of Sdec after one [11] or two [12] weeks of detraining, respectively. Rapid RSA impairment with detraining might be related to reductions in resting phosphoryla- tion status of the Na+-K+ pump [18]. The different RSA testing protocols used between studies may partially help to explain the relatively different results. However, the relationship between Sdec and the performance achieved in the first sprint may also contribute to the different results. In this sense, a slower first sprint (i.e., impaired sprint performance after detraining) will induce a reduction in Sdec values[39], which is translated into a false-positive improvement of Sdec. Thus, a more probable explanation stems on the poor validity and sensitivity of Sdec to detect actual performance impairment during RSA test and negative changes expected in response to short-term detraining. Although Sdec have been presented as a reliable marker of Effects of detraining on repeated-sprint ability and intermittent endurance PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 6 / 10 RSA performance [26], its use have been questioned, given that Sdec is the least reliable param- eter calculated from RSA tests [4]. In fact, Sdec have showed poor sensitivity to training [3]. Hence, it appears that Sdec is a poor indicator of RSA performance changes in response to an in-season break period in soccer. It is possible that, for a better assessment of Sdec, the ideal sprint performance at the moment of RSA measurement should not be considered, but to take into account the better sprint of the athlete for the given sprint distance. One potential limitation of this study was the estimation of VO2max through the YYIR1 test, since it could not be determined during laboratory-based maximal graded test. The direct measurement of VO2max (and metabolic thresholds) could offer more physiological informa- tion regarding the effects of detraining in soccer. Conclusion A 2-week in-season break (detraining) period impaired RSA, with no effect on intermittent endurance in professional and elite young soccer players, with greater impairment of RSAtotal and RSAbest in faster players, independent of their age category. In addition, Sdec does not seem to be sensitive to changes in RSA after a 2-week in-season break. Coaches should take these findings into consideration for appropriate training schedule after in-season break in order to regain the performance indices lost during the break. Practical applications According to our results (poorer RSA after a 2-weeks in-season break), coaches and practition- ers should considered an individualization of training loads during such periods, considering principles such as the minimal-effective dose, especially for players with greater fitness level, as these may be negatively affected to a greater magnitude by short-detraining periods. Supporting information S1 File. DatosGlobal. (XLSX) S2 File. Datos. (SAV) S1 Dataset. DataSet. (XLSX) Author Contributions Conceptualization: Alejandro Rodrı´guez-Ferna´ndez, Javier Sa´nchez-Sa´nchez, Rodrigo Ramirez-Campillo, Jose´ Antonio Rodrı´guez-Marroyo, Jose´ Gerardo Villa Vicente, Fabio Yuzo Nakamura. Formal analysis: Alejandro Rodrı´guez-Ferna´ndez, Javier Sa´nchez-Sa´nchez, Jose´ Antonio Rodrı´guez-Marroyo. Investigation: Alejandro Rodrı´guez-Ferna´ndez, Javier Sa´nchez-Sa´nchez. Methodology: Alejandro Rodrı´guez-Ferna´ndez, Javier Sa´nchez-Sa´nchez, Rodrigo Ramirez- Campillo, Jose´ Antonio Rodrı´guez-Marroyo, Jose´ Gerardo Villa Vicente, Fabio Yuzo Nakamura. Project administration: Alejandro Rodrı´guez-Ferna´ndez, Javier Sa´nchez-Sa´nchez. Effects of detraining on repeated-sprint ability and intermittent endurance PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 7 / 10 Resources: Alejandro Rodrı´guez-Ferna´ndez. Supervision: Javier Sa´nchez-Sa´nchez. Writing – original draft: Alejandro Rodrı´guez-Ferna´ndez, Javier Sa´nchez-Sa´nchez, Rodrigo Ramirez-Campillo, Fabio Yuzo Nakamura. Writing – review & editing: Alejandro Rodrı´guez-Ferna´ndez, Javier Sa´nchez-Sa´nchez, Rodrigo Ramirez-Campillo, Jose´ Antonio Rodrı´guez-Marroyo, Jose´ Gerardo Villa Vicente, Fabio Yuzo Nakamura. References 1. Mujika I, Spencer M, Santisteban J, Goiriena JJ, Bishop D. Age-related differences in repeated-sprint ability in highly trained youth football players. J Sport Sci. 2009/12/08. 2009; 27: 1581–1590. https://doi. org/10.1080/02640410903350281 PMID: 19967589 2. Reilly T, Gilbourne D. Science and football: a review of applied research in the football codes. J Sport Sci. 2003/10/29. 2003; 21: 693–705. https://doi.org/10.1080/0264041031000102105 PMID: 14579867 3. Iaia F, Fiorenza M, Perri E, Alberti G, Millet GP, Bangsbo J. The Effect of Two Speed Endurance Train- ing Regimes on Performance of Soccer Players. PLoS One. 2015/09/24. 2015; 10: e0138096. https:// doi.org/10.1371/journal.pone.0138096 PMID: 26394225 4. Impellizzeri FM, Rampinini E, Castagna C, Bishop D, Ferrari Bravo D, Tibaudi A, et al. Validity of a repeated-sprint test for football. Int J Sport Med. 2008/04/17. 2008; 29: 899–905. https://doi.org/10. 1055/s-2008-1038491 PMID: 18415931 5. Rampinini E, Bishop D, Marcora S, Ferrari Bravo D, Sassi R, Impellizzeri F. Validity of simple field tests as indicators of match-related physical performance in top-level professional soccer players. Int J Sport Med. 2006/10/07. 2007; 28: 228–235. https://doi.org/10.1055/s-2006-924340 PMID: 17024621 6. Oliver JL, Armstrong N, Williams CA. Relationship between brief and prolonged repeated sprint ability. J Sci Med Sport. 2007/12/14. 2009; 12: 238–243. https://doi.org/10.1016/j.jsams.2007.09.006 PMID: 18077213 7. Iaia F, Rampinini E, Bangsbo J. High-intensity training in football. Int J Sport Physiol Perform. 2009/12/ 04. 2009; 4: 291–306. Available: http://www.ncbi.nlm.nih.gov/pubmed/19953818 8. Bradley PS, Carling C, Gomez Diaz A, Hood P, Barnes C, Ade J, et al. Match performance and physi- cal capacity of players in the top three competitive standards of English professional soccer. Hum Mov Sci. 2013/08/28. 2013; 32: 808–821. https://doi.org/10.1016/j.humov.2013.06.002 PMID: 23978417 9. Bangsbo J, Norregaard L, Thorso F. Activity profile of competition soccer. Can J Sport Sci. 1991/06/01. 1991; 16: 110–116. Available: http://www.ncbi.nlm.nih.gov/pubmed/1647856 PMID: 1647856 10. Mohr M, Krustrup P, Bangsbo J. Match performance of high-standard soccer players with special refer- ence to development of fatigue. J Sport Sci. 2003/07/10. 2003; 21: 519–528. https://doi.org/10.1080/ 0264041031000071182 PMID: 12848386 11. Joo CH. The effects of short-term detraining on exercise performance in soccer players. J Exerc Reha- bil. 2016; 12: 54–59. https://doi.org/10.12965/jer.160280 PMID: 26933661 12. Christensen PM, Krustrup P, Gunnarsson TP, Kiilerich K, Nybo L, Bangsbo J. VO2 kinetics and perfor- mance in soccer players after intense training and inactivity. Med Sci Sport Exerc. 2011; 43: 1716– 1724. https://doi.org/10.1249/MSS.0b013e318211c01a PMID: 21311360 13. Bangsbo J, Iaia FM, Krustrup P. The Yo-Yo intermittent recovery test: a useful tool for evaluation of physical performance in intermittent sports. Sport Med. 2007/12/18. 2008; 38: 37–51. 14. Buchheit M, Mendez-Villanueva A, Delhomel G, Brughelli M, Ahmaidi S. Improving repeated sprint abil- ity in young elite soccer players: repeated shuttle sprints vs. explosive strength training. J Strength Cond Res. 2010/03/13. 2010; 24: 2715–2722. https://doi.org/10.1519/JSC.0b013e3181bf0223 PMID: 20224449 15. Campos-Vazquez MA, Romero-Boza S, Toscano-Bendala FJ, Leon-Prados JA, Suarez-Arrones LJ, Gonzalez-Jurado JA. Comparison of the effect of repeated-sprint training combined with two different methods of strength training on young soccer players. J Strength Cond Res. 2014/09/17. 2015; 29: 744–751. https://doi.org/10.1519/JSC.0000000000000700 PMID: 25226307 16. Iaia F, Fiorenza M, Larghi L, Alberti G, Millet G, Girard O. Short-or long-rest intervals during repeated- sprint training in soccer? Csernoch L, editor. PLoS One. 2017; 12: e0171462. https://doi.org/10.1371/ journal.pone.0171462 PMID: 28199402 Effects of detraining on repeated-sprint ability and intermittent endurance PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 8 / 10 17. Owen AL, Wong del P, Paul D, Dellal A. Effects of a periodized small-sided game training intervention on physical performance in elite professional soccer. J Strength Cond Res. 2012/09/25. 2012; 26: 2748–2754. https://doi.org/10.1519/JSC.0b013e318242d2d1 PMID: 23001394 18. Thomassen M, Christensen PM, Gunnarsson TP, Nybo L, Bangsbo J. Effect of 2-wk intensified training and inactivity on muscle Na+-K+ pump expression, phospholemman (FXYD1) phosphorylation, and performance in soccer players. J Appl Physiol. 2010; 108: 898–905. https://doi.org/10.1152/ japplphysiol.01015.2009 PMID: 20133439 19. Bangsbo J. Energy demands in competitive soccer. J Sports Sci. 1994; 12 Spec No: S5–12. Available: http://www.ncbi.nlm.nih.gov/pubmed/8072065 20. Melchiorri G, Ronconi M, Triossi T, Viero V, De Sanctis D, Tancredi V, et al. Detraining in young soccer players. J Sport Med Phys Fit. 2014; 54: 27–33. Available: http://www.ncbi.nlm.nih.gov/pubmed/ 24445542 21. Mujika I, Padilla S. Muscular characteristics of detraining in humans. Med Sci Sport Exerc. 2001/07/28. 2001; 33: 1297–1303. 22. Chaouachi A, Manzi V, Wong del P, Chaalali A, Laurencelle L, Chamari K, et al. Intermittent endurance and repeated sprint ability in soccer players. J Strength Cond Res. 2010/09/18. 2010; 24: 2663–2669. https://doi.org/10.1519/JSC.0b013e3181e347f4 PMID: 20847706 23. Rodriguez-Fernandez A, Sanchez Sanchez J, Rodriguez-Marroyo JA, Casamichana D, Villa JG. Effects of 5-week pre-season small-sided-game-based training on repeat sprint ability. J Sports Med Phys Fitness. Italy; 2017; 57: 529–536. https://doi.org/10.23736/S0022-4707.16.06263-0 24. Buchheit M. Repeated-sprint performance in team sport players: associations with measures of aerobic fitness, metabolic control and locomotor function. Int J Sport Med. 2012/02/01. 2012; 33: 230–239. https://doi.org/10.1055/s-0031-1291364 PMID: 22290323 25. Glaister M. Multiple-sprint work: methodological, physiological, and experimental issues. Int J Sport Physiol Perform. 2009/02/06. 2008; 3: 107–112. Available: http://www.ncbi.nlm.nih.gov/pubmed/ 19193958 26. Glaister M, Howatson G, Pattison JR, McInnes G. The reliability and validity of fatigue measures during multiple-sprint work: an issue revisited. J Strength Cond Res. 2008/08/21. 2008; 22: 1597–1601. https://doi.org/10.1519/JSC.0b013e318181ab80 PMID: 18714226 27. Buchheit M. Fatigue during Repeated Sprints: precision needed. Sport Med. 2012/01/12. 2012; 42: 165–167. https://doi.org/10.2165/11598220-000000000-00000 PMID: 22233536 28. Krustrup P, Mohr M, Amstrup T, Rysgaard T, Johansen J, Steensberg A, et al. The yo-yo intermittent recovery test: physiological response, reliability, and validity. Med Sci Sport Exerc. 2003/04/04. 2003; 35: 697–705. https://doi.org/10.1249/01.MSS.0000058441.94520.32 PMID: 12673156 29. Buchheit M, Morgan W, Wallace J, Bode M, Poulos N. Physiological, psychometric, and performance effects of the Christmas break in Australian Football. Int J Sports Physiol Perform. 2015; 10: 120–123. https://doi.org/10.1123/ijspp.2014-0082 PMID: 24806508 30. Nakamura FY, Suzuki T, Yasumatsu M, Akimoto T. Moderate running and plyometric training during off-season did not show a significant difference on soccer-related high-intensity performances com- pared with no-training controls. J Strength Cond Res. 2012; 26: 3392–3397. https://doi.org/10.1519/ JSC.0b013e3182474356 PMID: 22207263 31. Mujika I, Padilla S. Detraining: loss of training-induced physiological and performance adaptations. Part I: short term insufficient training stimulus. Sport Med. 2000/08/31. 2000; 30: 79–87. Available: http:// www.ncbi.nlm.nih.gov/pubmed/10966148 32. Kuzon WM Jr., Rosenblatt JD, Huebel SC, Leatt P, Plyley MJ, McKee NH, et al. Skeletal muscle fiber type, fiber size, and capillary supply in elite soccer players. Int J Sport Med. 1990/04/01. 1990; 11: 99– 102. https://doi.org/10.1055/s-2007-1024770 PMID: 2338382 33. Greenhaff PL, Nevill ME, Soderlund K, Bodin K, Boobis LH, Williams C, et al. The metabolic responses of human type I and II muscle fibres during maximal treadmill sprinting. J Physiol. 1994/07/01. 1994; 478 (Pt 1: 149–155. Available: http://www.ncbi.nlm.nih.gov/pubmed/7965830 34. Ratel S, Williams CA, Oliver J, Armstrong N. Effects of age and recovery duration on performance dur- ing multiple treadmill sprints. Int J Sport Med. 2006/01/03. 2006; 27: 1–8. https://doi.org/10.1055/s- 2005-837501 PMID: 16388435 35. Esbjornsson-Liljedahl M, Sundberg CJ, Norman B, Jansson E. Metabolic response in type I and type II muscle fibers during a 30-s cycle sprint in men and women. J Appl Physiol. 1999/10/12. 1999; 87: 1326–1332. Available: http://www.ncbi.nlm.nih.gov/pubmed/10517759 PMID: 10517759 36. Girard O, Mendez-Villanueva A, Bishop D. Repeated-sprint ability—part I: factors contributing to fatigue. Sports Med. 2011; 41: 673–94. https://doi.org/10.2165/11590550-000000000-00000 PMID: 21780851 Effects of detraining on repeated-sprint ability and intermittent endurance PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 9 / 10 37. Nakamura FY, Pereira LA, Rabelo FN, Ramirez-Campillo R, Loturco I. Faster Futsal Players Perceive Higher Training Loads and Present Greater Decreases in Sprinting Speed During the Preseason. J strength Cond Res. 2016; 30: 1553–62. https://doi.org/10.1519/JSC.0000000000001257 PMID: 26562717 38. Mujika I. The influence of training characteristics and tapering on the adaptation in highly trained individ- uals: a review. Int J Sport Med. 1998/12/05. 1998; 19: 439–446. https://doi.org/10.1055/s-2007-971942 PMID: 9839839 39. Mendez-Villanueva A, Hamer P, Bishop D. Physical fitness and performance. Fatigue responses during repeated sprints matched for initial mechanical output. Med Sci Sport Exerc. 2007/11/30. 2007; 39: 2219–2225. https://doi.org/10.1249/mss.0b013e31815669dc PMID: 18046194 Effects of detraining on repeated-sprint ability and intermittent endurance PLOS ONE | https://doi.org/10.1371/journal.pone.0201111 August 15, 2018 10 / 10

Effects of short-term in-season break detraining on repeated-sprint ability and intermittent endurance according to initial performance of soccer player.

08-15-2018

Rodríguez-Fernández, Alejandro,Sánchez-Sánchez, Javier,Ramirez-Campillo, Rodrigo,Rodríguez-Marroyo, José Antonio,Villa Vicente, José Gerardo,Nakamura, Fabio Yuzo

eng

PMC6235347

RESEARCH ARTICLE Repeatability and predictive value of lactate threshold concepts in endurance sports Jules A. A. C. HeubergerID1*, Pim Gal1, Frederik E. StuurmanID1, Wouter A. S. de Muinck KeizerID1,2, Yuri Mejia Miranda1, Adam F. Cohen1,3 1 Centre for Human Drug Research, Leiden, the Netherlands, 2 Free University of Amsterdam, Amsterdam, the Netherlands, 3 Department of Internal Medicine, Leiden University Medical Centre, Leiden, the Netherlands * jheuberger@chdr.nl Abstract Introduction Blood lactate concentration rises exponentially during graded exercise when muscles pro- duce more lactate than the body can remove, and the blood lactate-related thresholds are parameters based on this curve used to evaluate performance level and help athletes opti- mize training. Many different concepts of describing such a threshold have been published. This study aims to compare concepts for their repeatability and predictive properties of endurance performance. Methods Forty-eight well-trained male cyclists aged 18–50 performed 5 maximal graded exercise tests each separated by two weeks. Blood lactate-related thresholds were calculated using eight different representative concepts. Repeatability of each concept was assessed using Cronbach’s alpha and intra-subject CV and predictive value with 45 minute time trial tests and a road race to the top of Mont Ventoux was evaluated using Pearson correlations. Results Repeatability of all concepts was good to excellent (Cronbach’s alpha of 0.89–0.96), intra- subject CVs were low with 3.4–8.1%. Predictive value for performance in the time trial tests and road race showed significant correlations ranging from 0.65–0.94 and 0.53–0.76, respectively. Conclusion All evaluated concepts performed adequate, but there were differences between concepts. One concept had both the highest repeatability and the highest predictability of cycling per- formance, and is therefore recommended to be used: the Dmax modified method. As an easier to apply alternative, the lactate threshold with a fixed value of 4 mmol/L could be used as it performed almost as well. PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 1 / 16 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Heuberger JAAC, Gal P, Stuurman FE, de Muinck Keizer WAS, Mejia Miranda Y, Cohen AF (2018) Repeatability and predictive value of lactate threshold concepts in endurance sports. PLoS ONE 13(11): e0206846. https://doi.org/10.1371/journal. pone.0206846 Editor: Laurent Mourot, University of Bourgogne France Comte´, FRANCE Received: August 18, 2018 Accepted: October 19, 2018 Published: November 14, 2018 Copyright: © 2018 Heuberger et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All data files are available from the figshare database (DOI: 10. 6084/m9.figshare.7240571). Funding: The authors received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. Trial registration Dutch Trial Registry NTR5643 Introduction The measurement of blood lactate is extensively used in sports medicine, although there is debate on how lactate affects fatigue in endurance athletes. [1] Nevertheless, the concentration of lactate in the blood relative to the exercise intensity is a relevant marker of endurance per- formance. [2–5] This can be visualized in a blood lactate curve (BLC) using a maximal graded exercise test (GXT): as the workload on the athlete increases over time, blood lactate concen- trations (bLa) are measured at defined intervals. During high intensity contractions lactate is formed along with H+ in the muscles, [6] followed by an increased elimination of lactate from plasma. [7, 8] When elimination becomes saturated, bLa will start to rise when production exceeds clearance. This (exponential) rise in bLa in the BLC is of importance, as the corre- sponding exercise intensity is associated with endurance performance since it correlates with the transition from aerobic to anaerobic workout. [9] Since the 1960’s BLCs have been ana- lysed trying to accurately determine a point in this curve that predicts endurance performance. Although many terms have been used for this point, in this work they will be termed lactate threshold (LT) concepts. BLCs and LT concepts can be used to assess ‘endurance fitness’ in athletes, [10] and to evaluate the effects of and to prescribe training exercises for individual athletes. [4, 5] Therefore these measures are relevant in sports medicine, both in amateur and professional sports. But as LT is based on a maximal exercise test protocol that does not directly mimic endurance exercise, finding a single point in the resulting BLC that has a strong relation to endurance performance is challenging. Moreover, determining where this single point lies in the relatively smooth curve, that is the result of a complex system of factors, can prove difficult as well. On the other hand, the more accurate method of determining maxi- mum lactate steady state (MLSS), using several sessions with different workloads takes more time, which is the reason why an approximation of MLSS using lactate threshold concepts was developed. [11] A previous literature review showed that there are many methods used to analyse the BLCs, with approximately 25 different concepts identified in literature to describe some form of LT. [9] These different concepts are used interchangeably throughout scientific studies and in sports and show variable repeatability and predictive value. Moreover, populations that were included in different studies often differed in training status, age and category of sport. For these reasons there is debate about these LT concepts. [9] The aim of this study is to evaluate the repeatability and predictive value of representative concepts using a large dataset of BLCs from a group of well-trained cyclists who performed multiple GXTs, time trials and an uphill road race in the setting of a clinical study. Materials and methods Study design and participants Blood lactate curve data in this paper were generated in a previously published study. [12] Briefly, the study was a double-blind, randomized, placebo controlled, parallel, single centre trial to evaluate the effects of recombinant human erythropoietin (rHuEPO) in forty-eight healthy male cyclists aged 18 to 50. Informed consent was obtained from all individual Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 2 / 16 participants included in the study. The study was approved by the Independent Ethics Com- mittee of the Foundation Evaluation of Ethics in Biomedical Research (Stichting Beoordeling Ethiek Biomedisch Onderzoek, Assen, Netherlands). The study is registered in the Dutch Trial Registry (Nederlands Trial Register), number NTR5643. For inclusion, participants had to be well-trained, as evaluated by a maximum power-to-weight ratio during the GXT at screening that should exceed 4 W/kg. During the eleven week study duration, twenty-four participants received weekly rHuEPO injections and twenty-four received placebo injections for eight weeks. Participants had to maintain their regular training schedule during the study. Procedures Maximal exercise tests. Five GXTs were performed on a Monark LC4r ergometer (COSMED, Rome, Italy) with approximately 2-week intervals between each test, see Fig 1. After a two-minute warm-up at 75 Watts, the GXT dictated an increase in pedalling resistance to 175 Watts, which increased by an additional 25 Watts every five minutes. Between 4:15 and 4:45 into each step and immediately after termination of the exercise test, blood was drawn to measure bLa. Gas exchange was measured using a Quark CPET system (COSMED, Rome, Italy) and breath-by-breath sampling technology. During the test cadence had to be main- tained between 70 and 90 rpm. The test terminated when cadence could not be maintained above 70 rpm or when a participant stopped the test. Lactate determination. During the GXTs blood for lactate determination was drawn via an IV cannula (Venflon 7 Pro Safety, BD, Switzerland) with a 30 cm extension set between the cannula and a three way stopcock for blood sampling in the antecubital vein. Before the first and after every sampling the stopcock and extension set were flushed with 2 mL saline. Before blood sampling 0.5 mL was withdrawn from the stopcock to remove any remaining saline. Next, 1 mL of blood was taken from the stopcock. Within ten seconds from withdrawal the blood was placed on the Lactate Pro 2 (Arkray, Kyoto, Japan) strip which was then inserted in the Lactate Pro 2 device. The same device was used throughout the whole study and was given at least 20 minutes to adjust to the room temperature before sampling. Time trial tests. The time trial tests were performed twice on the same ergometer used for the GXT, with the first (TT1) 3–8 days after the first GXT and the second (TT2) one week after GXT four. Participants were instructed to produce the highest mean power output during a 45-minute period at a cadence of 70–90 rpm, attempting to mimic competitive cycling time trials. At the start of the test pedalling resistance was set at 80% of the maximal power reached during GXT1. Participants could adjust the power by indicating to increase or decrease in power by steps of 10 Watts. They were informed of the remaining time on a regular basis dur- ing the test. Mont Ventoux race. Approximately one week after the last GXT participants competi- tively climbed the Mont Ventoux (Vaucluse de´partement, France) via Be´doin, a climb of approximately 21.5 km with an average gradient of 7.5%. The race was preceded by a stage of 110 km in the French Provence (total elevation gain 1524 m) that was completed collectively. Fig 1. Study design. Study design showing timing of different tests. Time point 0 weeks indicates start of treatment (rHuEPO or placebo) for all participants. GXT, graded exercise test; TT, time trial test; RR, road race. https://doi.org/10.1371/journal.pone.0206846.g001 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 3 / 16 Racing bikes of participants were equipped with a Single Leg Power Meter SGY-PM910H2 (Pioneer Europe, Antwerp, Belgium) with Shimano Ultegra 6800 crank (Shimano, Osaka, Japan) to log power data on the bicycle during the race. Data were uploaded to the dedicated database Cyclo-Sphere. Lactate threshold concepts. The BLCs from the GXTs were then used to calculate several representative LT concepts. Concepts were selected as follows: First, published concepts were retrieved from a review by Faude et al. [9] and by a literature research within the PubMed database. The database was searched for the search terms ‘lactate threshold’, ‘aerobic thresh- old’, ‘anaerobic threshold’, ‘endurance performance’ or ‘maximal lactate steady state’ or similar terms in different combinations. The references of the selected articles were searched for fur- ther relevant articles. Secondly, retrieved concepts were divided into seven different categories, see S1 Table. A few retrieved concepts could not be implemented, reasons being lacking lactate concentrations in the recovery phase after exercise and no availability of the full text article describing the method of the concept despite various efforts obtaining it. (S1 Table, listed under “not selected categories”). From each remaining category, concepts that were represen- tative and were used frequently in other research were selected. If there were multiple concepts in one category that were commonly used and fundamentally different in methodology, more than one concept of that category was included in the analysis. Selecting multiple commonly used, but very similar concepts from one category was not deemed useful for the purpose of this study. This resulted in a final selection of eight concepts from the five implementable cate- gories for analysis in our study. Implementation of lactate threshold concepts. All selected concepts were implemented according to the articles that described the concept (S1 Table). When exact reproduction of the method was not feasible due to the use of different parameters (e.g. running velocity was used), we approximated the description as close as possible (e.g. we used power output). For concepts that required data fitting of the blood lactate curve a third-order polynomial was cho- sen, based on the shape of the blood lactate curve data and given that it is a proven method, although there is no generally accepted method for data fitting. [9] An example of a blood lac- tate curve with a depiction of all lactate threshold concepts is shown in Fig 2. LT1. Similar to what Tanaka described [13] we plotted bLa (mmol/L) versus power (W). Three authors (JH, WdMK and PG) were asked to independently select the first point in the BLC that marks a substantial increase above resting level. LT1 was defined as the power value corresponding to the point selected by at least two researchers, or in cases without consensus, the three researchers discussed until consensus was reached. LT2. Coyle et al. [14] determined LT as 1 mmol/L above a visually determined baseline in the BLC. We took the lactate measurement chosen as LT1 and calculated the mean of the mea- surements preceding this point to create an average baseline value. The power value belonging to the first measured lactate value after baseline that supersedes the baseline value plus 1 mmol/L was considered LT2. LT3. As Dickhuth et al., [15] we determined the minimum lactate equivalent (the lowest value when bLa is divided by work intensity) using third-order polynomial fitting and added 1.5 mmol/L to the corresponding bLa, termed individual anaerobic threshold in the paper, to find the power value on the fitted polynomial of the BLC and termed it LT3. LT4. As described by Amann et al., [16] we calculated the first rise of 1 mmol/L or more between two bLa measurements where the next rise was similar or larger than 1 mmol/L. The measurement that preceded this first increase was considered LT4. LT5. Based on the method described by Dickhuth et al., [17] we divided bLa (mmol/L) by the 30 second average VO2 (mL/min/kg) and plotted it against power. These values were Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 4 / 16 interpolated with a third-order polynomial and the power value at the lowest point in this curve was considered LT5. LT-4mmol. A widely used concept is the LT-4mmol method, as described for example by Sjodin et al. [18] The power in the interpolated third-order polynomial BLC that corresponds to a bLa of 4 mmol/L was considered LT-4mmol. Dmax and Dmax modified. Similar to the method proposed by Cheng et al., [19] we plot- ted bLa versus power, interpolated with a third-order polynomial and plotted a line from the first measurement to the last measurement. The point in the interpolated BLC that has the maximum perpendicular distance with that line was considered Dmax. A modified version as Fig 2. Graphical representation of lactate threshold concepts. Example of a blood lactate curve with the location of the different lactate threshold concepts for this particular curve. Open circles: observed blood lactate values at each exercise intensity; Black curve: third-order polynomial; Grey dashed line: baseline; Green circle and arrow: LT1, observer-determined first rise in blood lactate; Yellow circle and arrow: LT2, first observed blood lactate value more than 1 mmol/L above baseline; Pink circle and arrow: LT3, minimum lactate equivalent (blood lactate divided by power) plus 1.5 mmol/L; Purple circle and arrow: LT4, first blood lactate value that shows an increase of at least 1 mmol/L; Orange circle and arrow: LT5, minimum lactate equivalent (blood lactate divided by VO2); Brown circle and arrow and dashed line: LT-4mmol, value at 4 mmol/L; Red circle and arrow and dashed line: Dmax, value with the maximum perpendicular distance to the polynomial from the dashed line; Blue circle and arrow and dashed line: Dmax-mod, value with the maximum perpendicular distance to the polynomial from the dashed line. https://doi.org/10.1371/journal.pone.0206846.g002 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 5 / 16 described by Bishop et al., [20] uses the measurement that precedes an increase of at least 0.4 mmol/L instead of the first bLa measurement to draw the line to the last measurement, which is termed Dmax modified (Dmax-mod). Data management Data was stored in a validated database system (Promasys, Omnicomm Inc., Fort Lauderdale, USA) and checked for accuracy and completeness. Blinded data review before code-breaking and analysis was performed according to a standard procedure at our unit. This included eval- uating whether the GXT was performed to maximal ability, which was based on power, VO2 and bLa values and report by the subject. Statistical analysis We used statistical software R version 3.4.0 [21] to plot measurements, calculate the third- order polynomial that best fits the data using polynomial regression with the R-function lm (y~poly(3)), implement the LT concepts and perform the statistical testing. R was used with the following packages: dplyr 0.5.0, [22] psych 1.7.5, [23] tidyr 0.6.3. [24] Data of all subjects enrolled in the study were used in the analysis. Repeatability. To measure repeatability we determined the weighted intra-subject coeffi- cient of variation (CV) and the Cronbach’s alpha based on the five GXT results for each LT concept. Weighted intra-subject CV was calculated correcting for missing values (CV based on the sum of the variance per subject multiplied by the amount of measurements, divided by the total amount of measurements). Both the weighted intra-subject CV and Cronbach’s alpha were calculated only using data from participants receiving placebo, as there might have been longitudinal effects of rHuEPO treatment on the GXTs. Predictive properties. For the predictive properties we calculated the Pearson correlation between each LT concept and the mean power of the corresponding relevant endurance parameter. The LT concept from the GXT closest in time to the endurance tests TT1 and TT2 and road race (see Fig 1), namely GXT 1, 4 and 5 respectively, were used for correlations between the LT concept and corresponding average power output. In addition, the difference between each measurement pair was calculated and averaged to create the mean difference between the LT concept and endurance test power. This value indicates how the power at the LT concept translates to average endurance power in a time trial or race. For these Pearson correlation and mean difference analyses both subjects receiving rHuEPO and placebo were included. This was done as LT concepts are designed to be a predictive parameter for endur- ance exercise, which should be irrespective of a subject being treated with rHuEPO or not. In addition, given that the measurements of each pair are at most a week apart, no changes in the LT concept or endurance performance are expected due to rHuEPO. Moreover, GXT1 and TT1 were performed before starting the treatment period, and no rHuEPO administrations took place between GXT5 and the race. For these analyses therefore no treatment effect was expected and pooling was considered appropriate. Results In total 49 subjects entered the study, of which 47 were completers (Fig 3); one subject dropped out after having performed the first GXT and time trial test and was replaced. One other sub- ject dropped out after completing two GXTs and one time trial test and was not replaced. Sub- ject characteristics can be found in Table 1. Of the remaining 238 planned GXTs, five were not performed due to illness or injury. An additional 22 were excluded from analysis, five due to having less than four bLa samples for the GXT, most others due to the fact that subjects Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 6 / 16 Fig 3. CONSORT flowchart. https://doi.org/10.1371/journal.pone.0206846.g003 Table 1. Subject disposition. All subjects Placebo subjects N 48 24 Age (years) 33.6 (20.0–50.0) 33.8 (20.0–50.0) Weight (kg) 76.9 (9.0; 59.2–95.6) 76.9 (8.9; 59.2–95.6) Height (cm) 186 (7.3; 172–203) 186 (6.7; 174–203) Maximal Power output per kg (W/kg) 4.36 (4.03–5.18) 4.36 (4.03–4.94) VO2,max (mL/min/kg) 55.7 (4.6; 45.3–67.5) 56.0 (4.1; 47.0–62.8) Values are presented as mean (standard deviation (SD) where appropriate; range where). VO2,max: maximal oxygen consumption. https://doi.org/10.1371/journal.pone.0206846.t001 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 7 / 16 indicated having physical problems (e.g. illness/injury, sore legs from recent exercise) poten- tially affecting test results, leaving 211 GXTs (of which 109 from placebo subjects) with analy- sable lactate threshold data. A total of 96 time trial tests were performed and used in the analysis, and power data of 37 subjects was available for the road race. Out of the 47 subjects that completed the study, three could not participate in the road race, four did not reach the finish line due to exhaustion, and three did not have a power meter on their bike, therefore lacking power data for the road race. Lactate threshold concepts and endurance All eight LT concepts were successfully implemented on the GXT data; for LT1 which was determined visually by three researchers, a unanimous decision about the lactate threshold was reached in 56.8% of the tests, in 40.0% of the cases two out of three researchers agreed and there was originally no consensus in 3.2% of the tests. Several concepts were based on the third-order polynomial data fitting, mean r-squared values of all individual curves were 0.978 (SD = 0.032, range 0.716–1.000). Mean values for each LT concept of the placebo group can be found in Table 2. Mean (SD) power output for TT1 was 268 W (28 W) in the placebo group and 271 W (29 W) in the rHuEPO group, and estimated mean for TT2 was 277 W and 283 W for the placebo and rHuEPO groups respectively. Estimated mean power during RR were 266 W and 257 W for the two groups, during a mean race time of 1 h 37 min 45 s (SD = 12 min 40 s) and 1 h 38 min 23 s (SD = 14 min 9 s), respectively. Repeatability The overall intra-subject CV of each LT concept is indicated in Table 2, and shows some minor differences between concepts, with LT3, LT-4mmol, Dmax and Dmax-mod having CVs < 5% and LT5 having the highest intra-subject CV with 8.1%. The Cronbach’s alpha val- ues for all LT concepts in the placebo group are between 0.89 and 0.97 and although 95% CIs largely overlap, the same four concepts as observed for intra-subject CVs perform best with Cronbach’s alpha values >0.95 (Table 3). Predictive properties Pearson correlation coefficients and the mean difference between each correlation pair are listed in Table 4. All correlations are highly significant (p<0.0002), indicating the null Table 2. Mean lactate threshold concept power output. GXT number LT1 (W) LT2 (W) LT3 (W) LT4 (W) LT5 (W) LT-4mmol (W) Dmax (W) Dmax-mod (W) 1 283.3 (29.9; 225– 350) 292.9 (37.2; 250– 375) 286.1 (32.9; 219– 352) 275.0 (41.1; 200– 350) 225.0 (31.2; 175– 283) 301.8 (41.0; 222– 381) 275.7 (24.6; 222– 323) 299.5 (35.3; 225– 367) 2 283.0 (22.3; 225– 375) 293.2 (31.0; 250– 375) 288.7 (29.2; 231– 373) 276.1 (34.0; 175– 375) 231.8 (25.7; 175– 300) 305.0 (33.0; 234– 389) 280.0 (23.6; 233– 339) 301.2 (28.7; 237– 369) 3 281.0 (29.5; 225– 400) 290.5 (27.9; 250– 400) 285.7 (26.3; 240– 390) 272.6 (33.5; 225– 375) 224.5 (29.5; 175– 318) 300.8 (28.9; 253– 411) 278.7 (20.9; 250– 343) 297.5 (29.6; 257– 413) 4 283.7 (35.8; 225– 400) 292.4 (38.0; 225– 425) 291.6 (29.1; 240– 392) 272.8 (36.9; 200– 400) 229.8 (29.5; 175– 323) 307.0 (34.2; 249– 415) 284.0 (22.7; 232– 338) 308.7 (35.6; 251– 396) 5 278.3 (28.5; 225– 325) 290.2 (37.5; 225– 350) 285.0 (31.6; 216– 339) 271.7 (37.9; 200– 350) 230.4 (30.5; 175– 274) 297.2 (38.9; 204– 364) 280.3 (20.3; 245– 325) 298.9 (29.2; 253– 365) Overall 282.1 (5.7%) 292.2 (5.0%) 287.7 (3.6%) 274.1 (5.6%) 228.4 (8.1%) 302.7 (3.8%) 280.0 (3.4%) 301.5 (4.3%) Weighted mean power output (SD; range) for the placebo group at every exercise test. Overall combined (based on 109 GXTs) for each lactate threshold concept (CV). CV is weighted intra-subject CV. https://doi.org/10.1371/journal.pone.0206846.t002 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 8 / 16 hypothesis that the correlation is equal to zero can be rejected. The strength of the relationship differs for different concepts. Correlation with TT1 was very strong for Dmax-mod and strong for all other concepts except LT5, which showed a moderate correlation. Correlation with TT2 was strong for all concepts except LT5, which showed a moderate correlation. Correlation with RR was strong for Dmax and Dmax-mod, and moderate for all other concepts. Dmax- mod has the highest correlation with time trial test 1 (r = 0.94), LT-4mmol with time trial test 2 (r = 0.85) and Dmax-mod with road race power (r = 0.76). The mean difference with the endurance parameters differs substantially between concepts, ranging from the lactate thresh- old on average being up to 45.3 W lower than the related endurance parameter for LT5 to 36.6 W higher for LT-4mmol. Linear regression between each LT concept and average race power, including accompanying r2 values, is plotted in Fig 4. Discussion All LT concepts that were included in this analysis performed good on repeatability and rea- sonable to good on predicting a lab-based time trial and a real-life road race. Nevertheless, this Table 3. Cronbach’s alpha for each lactate threshold concept. Lactate threshold concept Cronbach’s alpha Lower 95% CI Upper 95% CI LT1 0.91 0.85 0.96 LT2 0.95 0.92 0.98 LT3 0.97 0.94 0.99 LT4 0.94 0.91 0.98 LT5 0.89 0.82 0.96 LT 4_mmol 0.96 0.94 0.99 Dmax 0.96 0.93 0.98 Dmax-mod 0.96 0.94 0.98 Cronbach’s alpha for the placebo group for each lactate threshold concept with 95% confidence interval (CI). https://doi.org/10.1371/journal.pone.0206846.t003 Table 4. Predictive value of lactate threshold concepts. Lactate threshold concept Pearson correlation Mean difference (SD) TT1 TT2 RR TT1 TT2 RR n = 42 n = 46a n = 34b n = 42 n = 46a n = 34b LT1 0.78 0.74 0.54 -11.3 (18.3) -8.7 (22.1) -9.8 (37.2) LT2 0.87 0.80 0.53 -23.2 (16.0) -18.5 (19.3) -27.4 (39.3) LT3 0.88 0.84 0.64 -16.2 (14.3) -16.1 (18.1) -21.9 (33.8) LT4 0.78 0.82 0.61 -3.5 (23.7) -0.6 (26.2) -13.5 (36.1) LT5 0.67 0.65 0.58 43.7 (21.7) 45.3 (23.5) 39.1 (35.9) LT-4mmol 0.88 0.85 0.61 -31.7 (19.4) -32.3 (23.0) -36.6 (36.1) Dmax 0.89 0.82 0.73 -4.4 (12.1) -3.8 (15.8) -13.4 (32.4) Dmax-mod 0.94 0.84 0.76 -27.3 (11.8) -29.9 (16.6) -33.7 (29.1) Pearson correlation between each lactate threshold concept in GXT 1 and time trial test 1 (TT1), GXT 4 and time trial test 2 (TT2) and GXT 5 and average road race (RR) power for all subjects combined. All correlations are significant (p<0.0002). To determine potential differences in power output between the LT concept and time trial power or race power, mean difference (SD) between each measurement pair is calculated. Negative values indicate lactate threshold power is higher than exercise test average power. a For LT5 n = 44; b for LT5 n = 32. https://doi.org/10.1371/journal.pone.0206846.t004 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 9 / 16 study identified several LT concepts that outperformed the others in the setting of this trial. The best method being Dmax-mod, but Dmax, LT-4mmol and LT3 performed well too. Methodology The design of the exercise protocol, for example stage duration, is known to impact blood lactate curves. [25, 26] We selected an exercise protocol with five minute stages and 25 W increments because it takes 3–4 minutes for the body to reach steady state and lactate accom- panying that effort level can be measured accurately. [27] In addition, longer protocols may be more sensitive to performance changes. [25] As described in more detail elsewhere, [12] GXT results show our subjects were well-trained with maximal power output and VO2 max values comparable to elite cyclists and triathletes when using longer exercise protocols. [28, 29] All evaluated concepts were applied to data from the same exercise tests, with the same sampling and assay method, and the same fitting procedure was used for those applicable concepts. As a result such factors could not have affected the comparison between concepts within this study. The current study was designed in that way to give the most accurate estimate of performance parameters and its controlled set-up seems to be the most robust and valuable way to deter- mine differences between concepts. Nevertheless, when any of these factors are changed (e.g. using a different exercise test protocol) it is possible the outcomes might not translate perfectly. With regards to data fitting, the third-order polynomial in the applicable concepts performed well given the high mean r-squared values observed. Selection of concepts After inspection of all identified lactate concepts, it became clear that there were similarities between quite some of them. For this reason, the concepts were grouped into categories, and a selection was made of concepts to be analysed to have at least one representative per category Fig 4. Linear regression lactate threshold concept power and average race power. Linear regression of lactate threshold power and average race power per LT concept for all subjects depicting linear regression line (solid line) and 95% confidence interval (dotted lines). r2: R-squared or coefficient of determination is the proportion of the variance in the dependent variable that is predictable from the independent variable. https://doi.org/10.1371/journal.pone.0206846.g004 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 10 / 16 and thereby ensuring that results from this study would be informative for all regularly used lactate concepts. The selection includes concepts such as a fixed lactate value (usually at 4 mmol/L) and the visually determined LT concepts that have been used since the conception of the LT, and more recent concepts such as LT3, LT4 and Dmax and Dmax-mod. [9] Mean threshold The mean power output (Table 2) is relatively constant over time for each concept. These results confirm there was no placebo-effect on any of the LT concepts, although such an impact would already theoretically be improbable. What can also be seen is that not all con- cepts seem to identify the same point in the blood lactate curve: LT5 gives the lowest estimate of LT (228.4 W), much lower than other concepts (274.1–302.7 W). LT-4mmol and Dmax- mod have the highest estimates (302.7 and 301.5 W), indicating these concepts identify differ- ent intensities of performance and have different physiological meanings. Applying the termi- nology as described in Faude et al, [9] based on mean threshold and mean difference with TT and RR (Table 4), some concepts seem to be more related to the aerobic threshold (LT5), oth- ers to the aerobic/anaerobic transition (e.g. LT1, LT4, Dmax) or the anaerobic threshold (LT- 4mmol, Dmax-mod). Repeatability Intra-subject CV’s over all five measurements were low (3.4–8.1%) and Cronbach’s alphas high (0.89–0.97), indicating repeatability of all concepts over the study period of approximately 8 weeks was good. This corresponds well to previous findings of repeatability for power or speed at different lactate concepts, both in terms of CV, determined at 1.3–5.9% in a meta-ana- lytic review, [30] and in terms of Pearson correlations 0.88–0.96 [31, 32] or ICC 0.98–0.99. [33] One study applied different LT concepts to the same dataset from two exercise tests and showed that intra-subject CV’s and correlation was good for LT2, LT-4mmol and a concept similar to LT4 (CV 3–4% and r  0.85), but not for Dmax (10.3% and 0.57). [34] Our data, based on more subjects (24 versus 14) and more measurements per subject (5 versus 2), dis- putes this relatively poor repeatability for Dmax. However, our study does show differences between concepts, with LT3, LT-4mmol, Dmax and Dmax-mod having the lowest intra-sub- ject CV (<5%) and the highest Cronbach’s alpha (>0.95). Correlation with performance As we have established that CV and repeatability for all LT concepts was good, the most rele- vant question is whether these concepts correlate to actual endurance performance. As previ- ously indicated, it is highly unlikely that the rHuEPO treatment impacted this particular correlation analysis. When analysing the groups separately, some differences in correlation coefficients could be observed between the two groups (data not shown), but these differences were already present for the correlation between GXT1 and TT1 when treatment had not yet started, indicating that this was not due to rHuEPO treatment. Because combining all subjects generates more informative and robust results being based on a bigger population, pooling the groups was considered justified. Data in Table 4 show that for all concepts correlations with time trial tests were higher com- pared to the road race (based on all subjects median of all concepts r = 0.875 for TT1 and 0.82 for TT2, versus 0.61 for RR). This is most likely partly due to additional variability in the road race due to the circumstances (e.g. weather, uphill racing with changes in steepness over the course, and differences in race duration (range 72–126 minutes)). Possibly there was also a minor impact of using different equipment for power measurement during the RR, as it was Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 11 / 16 not measured on the ergometer but on the subjects’ bike. What can also be seen is that correla- tion of the LT concepts with TT1 in general is slightly higher than with TT2. More importantly however, correlation for both time trials show that the ranking among different concepts is very similar, confirming the results are robust. It seems that in general, using a technique of interpolation for the BLC has superior performance, as LT concepts that were based on the third-order polynomial derived from the individual lactate concentration measurements (LT3, LT5, LT-4mmol, Dmax, Dmax-mod) performed better than the ones that used actual mea- sured bLa values without interpolation (LT1, LT2 and LT4), with the exception of LT5. This poor performance of LT5 is most likely due to the fact that it is conceptually different from the other concepts; it is the power at the minimum lactate equivalent, in this case the lowest value for the lactate-VO2 ratio. In contrast, LT3 also uses a form of the minimum lactate equivalent, but it adds 1.5 mmol/L to this value. As can be seen in Table 2 and Fig 2, this leads to LT5 on average determining a point even before the first rise in lactate concentration as determined by LT1. This concept therefore relates to much lower (aerobic) work intensities than the other concepts. Additionally it is less repeatable (see Table 3). From all tested concepts LT5 corre- lates least with 45 min TT performance, but for the longer RR performance relative to the other concepts it performs somewhat better than for TT. This could mean that is this concept is more related to long-term exercise efforts. Many studies previously evaluated correlations of LT concepts with endurance perfor- mance, of which most used running performance. An overview of these studies by Faude et al shows a median r = 0.84–0.92 for several different LT concepts for endurance distances (>5km), [9] comparable to our results. There are fewer studies that have compared LT con- cepts and their correlation with different types of cycling endurance performance, [16, 20, 26, 35–38] but correlation with endurance performances (30–90 minutes) for each concept seem to vary between these studies, see Table 5. In addition, the comparison between concepts within these studies shows varying conclusions about which is the best concept. This could partially be due to differences between studies, for example study populations differ (mean VO2max ranges from 48 to 68 mL/kg/min, and some studying female, others male cyclists and/or triathletes). However, they are more or less as heterogeneous as our population with an SD of 4–8 mL/kg/min on VO2max. The applied exercise protocols all used long stages similar to ours (3–5 minutes), although the increases in workload differ (20-50W). Finally, correlation to endurance exercise was based on time trials that lasted between approximately 30 to 90 min- utes (our TT of 45 min at the lower end and RR of on average 98 min at the higher end), a dif- ference that might impact the correlation to different LT concepts. Nevertheless, taking these differences into account, comparison is possible, albeit with some caution. Moreover, a robust Table 5. Reported correlations between LT concepts and endurance performance. Correlation reported in publication Lactate threshold concept LT1 LT2 LT4 LT-4mmol Dmax Dmax-mod LTlog Amann [16] - 0.72 0.59 0.60 - - - Bentley [37] - - - 0.54 0.77 - 0.91 Bishop [20] 0.81 - 0.61 0.81 0.84 0.83 0.69 Borszcz [35] - 0.31 - 0.56 0.75 - - McNaughton [38] - - - 0.90 0.91 - 0.86 Nichols [36] - - 0.88 0.67 - - - Literature data for LT concepts and correlation with 30–90 minute-during performances. LTlog: the power output at which bLa starts to increase when log(bLa) is plotted against log(power output). https://doi.org/10.1371/journal.pone.0206846.t005 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 12 / 16 and valid LT concept should perform well in any of these datasets. What can be observed is that all these concepts except LT1, Dmax and Dmax-mod have shown correlations below 0.75, and that in all four direct comparisons that evaluated both Dmax and LT-4mmol, Dmax showed a higher correlation. This latter finding could be due to the fact that LT-4mmol is less robust to changes in settings such as exercise protocol duration, sampling site and lactate analyser because of its fixed nature. Our study expands on this information, and compared to previous studies as reviewed in Table 5, is based on approximately 2–4 times more subjects, therefore allowing for more robust conclusions. This is especially true since our population is a heterogeneous well-trained, and therefore relevant, group (range maximal power output at baseline 256–425 W). Similar to what can be extracted from the literature, our study too shows that Dmax and Dmax-mod have highest correlations with time trial performance, although LT-4mmol and LT3 show a similarly high correlation in our study. For the correlation with RR, there are slightly larger differences between concepts. Correlation is highest for Dmax and Dmax-mod, mainly because for the other concepts correlation for a few subjects is very poor, as visualized in Fig 4 (e.g. for LT-4mmol). These findings combined, we conclude that Dmax, and even more so Dmax-mod, have the best correlation with endurance performance. One recent study evaluated correlation between MLSS, which could be considered to be the gold standard for the physiological endurance threshold, and different LT concepts generated from GXTs with different protocol durations. [26] This study concluded that for a GXT with 4-min- ute steps (most similar to our GXT), correlation was high for many of the concepts, but validity was highest for LT-2.5mmol, Dmax-mod, and two modified versions of Dmax-mod. In con- trast, LT2, LT-4mmol and Dmax showed much higher mean differences with MLSS and there- fore were designated as invalid estimates of MLSS. Combining these findings with our own results, Dmax-mod determined in a GXT with approximately 5-minute stages is both a valid estimate of MLSS and has a high correlation with actual endurance performance. Absolute power difference The mean difference of each concept with the endurance parameter gives an indication of how the absolute power of the LT concept corresponds to the average power produced during TT and RR. On average, power is higher compared to the endurance test for each concept (except the poorest performing concept LT5). This difference in power between LT concepts and endurance test is possibly due to having to sustain the power for a much longer time during the endurance tests, needing a systematic lower power in order to cope with the effort. Inter- estingly, Dmax-mod and LT-4mmol, concepts that show among the highest correlations, have the largest difference in absolute power (approximately 30 W). Given the high correlation with performance this should not disqualify these concepts, but one should take into account that there is a systematic difference with endurance performance of approximately 30 W. Conclusions LT concepts are correlated with endurance performance, but a review showed that many dif- ferent concepts are used in literature, which is undesirable. [9] Also for cycling performance, there is no consensus on which LT concept should be applied and results vary highly. [16, 20, 35–38] In this study we compared eight different representative LT concepts on the same large cycling performance dataset to evaluate repeatability and predictive properties. All concepts showed high repeatability, and correlated with endurance performance. However, LT3, LT- 4mmol, Dmax and Dmax-mod showed the best repeatability, and had the highest correlation with time trial performance. As correlation with performance was consistently high for Dmax and Dmax-mod, also with the uphill road race, the latter performing slightly better on each Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 13 / 16 criterion, and because Dmax-mod was previously shown to be a valid estimate of MLSS, we would recommend using Dmax-mod when analyzing the blood lactate curve. Supporting information S1 Table. Lactate threshold concept categories. (DOCX) S1 Protocol. Study protocol CHDR1514. (PDF) Author Contributions Conceptualization: Jules A. A. C. Heuberger, Adam F. Cohen. Data curation: Jules A. A. C. Heuberger, Pim Gal, Frederik E. Stuurman. Formal analysis: Jules A. A. C. Heuberger, Wouter A. S. de Muinck Keizer, Yuri Mejia Miranda. Investigation: Jules A. A. C. Heuberger, Pim Gal, Frederik E. Stuurman. Methodology: Jules A. A. C. Heuberger. Supervision: Jules A. A. C. Heuberger, Adam F. Cohen. Writing – original draft: Jules A. A. C. Heuberger. Writing – review & editing: Pim Gal, Frederik E. Stuurman, Wouter A. S. de Muinck Keizer, Yuri Mejia Miranda, Adam F. Cohen. References 1. Cairns SP. Lactic acid and exercise performance: culprit or friend? Sports Med. 2006; 36(4):279–91. https://doi.org/10.2165/00007256-200636040-00001 PMID: 16573355 2. Atkinson G, Davison R, Jeukendrup A, Passfield L. Science and cycling: current knowledge and future directions for research. J Sports Sci. 2003; 21(9):767–87. https://doi.org/10.1080/ 0264041031000102097 PMID: 14579871 3. Kindermann W, Simon G, Keul J. The significance of the aerobic-anaerobic transition for the determina- tion of work load intensities during endurance training. Eur J Appl Physiol Occup Physiol. 1979; 42 (1):25–34. PMID: 499194 4. Londeree BR. Effect of training on lactate/ventilatory thresholds: a meta-analysis. Med Sci Sports Exerc. 1997; 29(6):837–43. PMID: 9219214 5. Antonutto G, Di Prampero PE. The concept of lactate threshold. A short review. J Sports Med Phys Fit- ness. 1995; 35(1):6–12. PMID: 7474995 6. Robergs RA, Ghiasvand F, Parker D. Biochemistry of exercise-induced metabolic acidosis. Am J Phy- siol Regul Integr Comp Physiol. 2004; 287(3):R502–16. https://doi.org/10.1152/ajpregu.00114.2004 PMID: 15308499 7. MacRae HS, Dennis SC, Bosch AN, Noakes TD. Effects of training on lactate production and removal during progressive exercise in humans. J Appl Physiol (1985). 1992; 72(5):1649–56. 8. Stanley WC, Gertz EW, Wisneski JA, Neese RA, Morris DL, Brooks GA. Lactate extraction during net lactate release in legs of humans during exercise. J Appl Physiol (1985). 1986; 60(4):1116–20. 9. Faude O, Kindermann W, Meyer T. Lactate threshold concepts: how valid are they? Sports Med. 2009; 39(6):469–90. https://doi.org/10.2165/00007256-200939060-00003 PMID: 19453206 10. Coyle EF, Coggan AR, Hopper MK, Walters TJ. Determinants of endurance in well-trained cyclists. J Appl Physiol (1985). 1988; 64(6):2622–30. 11. Billat VL, Sirvent P, Py G, Koralsztein JP, Mercier J. The concept of maximal lactate steady state: a bridge between biochemistry, physiology and sport science. Sports Med. 2003; 33(6):407–26. https:// doi.org/10.2165/00007256-200333060-00003 PMID: 12744715 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 14 / 16 12. Heuberger J, Rotmans JI, Gal P, Stuurman FE, van’t Westende J, Post TE, et al. Effects of erythropoie- tin on cycling performance of well trained cyclists: a double-blind, randomised, placebo-controlled trial. Lancet Haematol. 2017; 4(8):e374–e86. https://doi.org/10.1016/S2352-3026(17)30105-9 PMID: 28669689 13. Tanaka H. Predicting running velocity at blood lactate threshold from running performance tests in ado- lescent boys. Eur J Appl Physiol Occup Physiol. 1986; 55(4):344–8. PMID: 3758032 14. Coyle EF, Martin WH, Ehsani AA, Hagberg JM, Bloomfield SA, Sinacore DR, et al. Blood lactate thresh- old in some well-trained ischemic heart disease patients. J Appl Physiol Respir Environ Exerc Physiol. 1983; 54(1):18–23. https://doi.org/10.1152/jappl.1983.54.1.18 PMID: 6826403 15. Dickhuth H-H, Yin L, Niess A, Rocker K, Mayer F, Heitkamp HC, et al. Ventilatory, lactate-derived and catecholamine thresholds during incremental treadmill running: relationship and reproducibility. Int J Sports Med. 1999; 20(2):122–7. https://doi.org/10.1055/s-2007-971105 PMID: 10190774 16. Amann M, Subudhi AW, Foster C. Predictive validity of ventilatory and lactate thresholds for cycling time trial performance. Scand J Med Sci Sports. 2006; 16(1):27–34. https://doi.org/10.1111/j.1600- 0838.2004.00424.x PMID: 16430678 17. Dickhuth H.-H.; Huonker M. MT, Drexler H., Berg A., Keul J. Individual anaerobic threshold for evalua- tion of competitive athletes and patients with left ventricular dysfunctions. Advances in ergometry. 1991. 18. Sjodin B, Jacobs I. Onset of blood lactate accumulation and marathon running performance. Int J Sports Med. 1981; 2(1):23–6. https://doi.org/10.1055/s-2008-1034579 PMID: 7333732 19. Cheng B, Kuipers H, Snyder AC, Keizer HA, Jeukendrup A, Hesselink M. A new approach for the deter- mination of ventilatory and lactate thresholds. Int J Sports Med. 1992; 13(7):518–22. https://doi.org/10. 1055/s-2007-1021309 PMID: 1459746 20. Bishop D, Jenkins DG, Mackinnon LT. The relationship between plasma lactate parameters, Wpeak and 1-h cycling performance in women. Med Sci Sports Exerc. 1998; 30(8):1270–5. PMID: 9710868 21. Chambers J. Project R The R Project for Statistical Computing [3.4.0:[https://www.r-project.org/. 22. Hadley Wickham RF, Lionel Henry, Kirill Mu¨ller. dplyr: A Grammar of Data Manipulation dplyr: A Gram- mar of Data Manipulation2017 [https://cran.r-project.org/web/packages/dplyr/. 23. Revelle W. psych: Procedures for Psychological, Psychometric, and Personality Research psych: Pro- cedures for Psychological, Psychometric, and Personality Research2017 [https://cran.r-project.org/ web/packages/psych/. 24. Hadley Wickham LH. tidyr: Easily Tidy Data with ’spread()’ and ’gather()’ Functions tidyr: Easily Tidy Data with ’spread()’ and ’gather()’ Functions2017 [https://cran.r-project.org/web/packages/tidyr/. 25. Bentley DJ, Newell J, Bishop D. Incremental exercise test design and analysis: implications for perfor- mance diagnostics in endurance athletes. Sports Med. 2007; 37(7):575–86. https://doi.org/10.2165/ 00007256-200737070-00002 PMID: 17595153 26. Jamnick NA, Botella J, Pyne DB, Bishop DJ. Manipulating graded exercise test variables affects the validity of the lactate threshold and [Formula: see text]. PLoS One. 2018; 13(7):e0199794. https://doi. org/10.1371/journal.pone.0199794 PMID: 30059543 27. Thoden JS. Testing aerobic power. In: MacDougall JD, Wenger HA, Green HJ, editors. Physiological testing of the high-performance athlete. Champaign: Human Kinetics; 1991. p. 107–74. 28. San Millan I, Bing K, Brill C, Hill JC, Miller LE. Randomized controlled trial of Micro-Mobile Compression (R) on lactate clearance and subsequent exercise performance in elite male cyclists. Open Access J Sports Med. 2013; 4:221–7. https://doi.org/10.2147/OAJSM.S51956 PMID: 24379728 29. Bentley DJ, McNaughton LR. Comparison of W(peak), VO2(peak) and the ventilation threshold from two different incremental exercise tests: relationship to endurance performance. J Sci Med Sport. 2003; 6(4):422–35. PMID: 14723392 30. Hopkins WG, Schabort EJ, Hawley JA. Reliability of power in physical performance tests. Sports Med. 2001; 31(3):211–34. https://doi.org/10.2165/00007256-200131030-00005 PMID: 11286357 31. Weltman A, Snead D, Stein P, Seip R, Schurrer R, Rutt R, et al. Reliability and validity of a continuous incremental treadmill protocol for the determination of lactate threshold, fixed blood lactate concentra- tions, and VO2max. Int J Sports Med. 1990; 11(1):26–32. https://doi.org/10.1055/s-2007-1024757 PMID: 2318561 32. Grant S, McMillan K, Newell J, Wood L, Keatley S, Simpson D, et al. Reproducibility of the blood lactate threshold, 4 mmol.l(-1) marker, heart rate and ratings of perceived exertion during incremental treadmill exercise in humans. Eur J Appl Physiol. 2002; 87(2):159–66. https://doi.org/10.1007/s00421-002-0608- 2 PMID: 12070627 33. Pfitzinger P, Freedson PS. The reliability of lactate measurements during exercise. Int J Sports Med. 1998; 19(5):349–57. https://doi.org/10.1055/s-2007-971929 PMID: 9721059 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 15 / 16 34. Pallares JG, Moran-Navarro R, Ortega JF, Fernandez-Elias VE, Mora-Rodriguez R. Validity and Reli- ability of Ventilatory and Blood Lactate Thresholds in Well-Trained Cyclists. PLoS One. 2016; 11(9): e0163389. https://doi.org/10.1371/journal.pone.0163389 PMID: 27657502 35. Borszcz FK, Tramontin AF, de Souza KM, Carminatti LJ, Costa VP. Physiological Correlations With Short, Medium, and Long Cycling Time-Trial Performance. Research quarterly for exercise and sport. 2018; 89(1):120–5. https://doi.org/10.1080/02701367.2017.1411578 PMID: 29334005 36. Nichols JF, Phares SL, Buono MJ. Relationship between blood lactate response to exercise and endur- ance performance in competitive female master cyclists. Int J Sports Med. 1997; 18(6):458–63. https:// doi.org/10.1055/s-2007-972664 PMID: 9351693 37. Bentley DJ, McNaughton LR, Thompson D, Vleck VE, Batterham AM. Peak power output, the lactate threshold, and time trial performance in cyclists. Med Sci Sports Exerc. 2001; 33(12):2077–81. PMID: 11740302 38. McNaughton LR, Roberts S, Bentley DJ. The relationship among peak power output, lactate threshold, and short-distance cycling performance: effects of incremental exercise test design. J Strength Cond Res. 2006; 20(1):157–61. https://doi.org/10.1519/R-15914.1 PMID: 16506862 Lactate threshold concepts in endurance exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0206846 November 14, 2018 16 / 16

Repeatability and predictive value of lactate threshold concepts in endurance sports.

11-14-2018

Heuberger, Jules A A C,Gal, Pim,Stuurman, Frederik E,de Muinck Keizer, Wouter A S,Mejia Miranda, Yuri,Cohen, Adam F

eng

PMC9268959

Citation: Jaén-Carrillo, D.; Roche-Seruendo, L.E.; Molina-Molina, A.; Cardiel-Sánchez, S.; Cartón-Llorente, A.; García-Pinillos, F. Influence of the Shod Condition on Running Power Output: An Analysis in Recreationally Active Endurance Runners. Sensors 2022, 22, 4828. https://doi.org/10.3390/s22134828 Academic Editors: Robert Crowther and Carlo Ricciardi Received: 2 June 2022 Accepted: 23 June 2022 Published: 26 June 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). sensors Article Influence of the Shod Condition on Running Power Output: An Analysis in Recreationally Active Endurance Runners Diego Jaén-Carrillo 1 , Luis E. Roche-Seruendo 1 , Alejandro Molina-Molina 1,2,* , Silvia Cardiel-Sánchez 1, Antonio Cartón-Llorente 1 and Felipe García-Pinillos 2,3 1 Campus Universitario, Universidad San Jorge, Autov A23 km 299, Villanueva de Gállego, 50830 Zaragoza, Spain; djaen@usj.es (D.J.-C.); leroche@usj.es (L.E.R.-S.); scardiels@usj.es (S.C.-S.); acarton@usj.es (A.C.-L.) 2 Department of Physical Education and Sport, University of Granada, 18016 Granada, Spain; fegarpi@gmail.com 3 Departamento de Educación Física, Deporte y Recreación, Universidad de La Frontera, Temuco 4780000, Chile * Correspondence: amolinam@usj.es; Tel.: +34-625538520 Abstract: Several studies have already analysed power output in running or the relation between VO2max and power production as factors related to running economy; however, there are no studies assessing the difference in power output between shod and barefoot running. This study aims to identify the effect of footwear on the power output endurance runner. Forty-one endurance runners (16 female) were evaluated at shod and barefoot running over a one-session running protocol at their preferred comfortable velocity (11.71 ± 1.07 km·h−1). The mean power output (MPO) and normalized MPO (MPOnorm), form power, vertical oscillation, leg stiffness, running effectiveness and spatiotemporal parameters were obtained using the Stryd™ foot pod system. Additionally, footstrike patterns were measured using high-speed video at 240 Hz. No differences were noted in MPO (p = 0.582) and MPOnorm (p = 0.568), whereas significant differences were found in form power, in both absolute (p = 0.001) and relative values (p < 0.001), running effectiveness (p = 0.006), stiffness (p = 0.002) and vertical oscillation (p < 0.001). By running barefoot, lower values for contact time (p < 0.001) and step length (p = 0.003) were obtained with greater step frequency (p < 0.001), compared to shod running. The prevalence of footstrike pattern significantly differs between conditions, with 19.5% of runners showing a rearfoot strike, whereas no runners showed a rearfoot strike during barefoot running. Running barefoot showed greater running effectiveness in comparison with shod running, and was consistent with lower values in form power and lower vertical oscillation. From a practical perspective, the long-term effect of barefoot running drills might lead to increased running efficiency and leg stiffness in endurance runners, affecting running economy. Keywords: barefoot; footstrike; stiffness; sensor; wearable 1. Introduction Endurance running events range from 3000 m to over 160 km in ultra-marathons. Nowadays, both the number of runners in endurance races and the number of organised races have increased. For example, 19,076 runners (19.51% women) finished the half marathon in Valencia in 2020 (Spain). The lower limb muscles execute three distinctive functions during such events: (i) force and power generation; (ii) shock absorption; and (iii) store and release elastic energy [1], thus compromising running economy. Endurance runners experienced improvements in muscle strength and power, among others, after an 8-week training intervention directly affecting running economy and, thus, performance [2]. The novel appearance of wearable devices capable of obtaining kinetic and kinematic data during running offers sports practitioners a new way to quantify workload by acquiring valuable metrics such as spatiotemporal parameters, power, and leg stiffness. Sensors 2022, 22, 4828. https://doi.org/10.3390/s22134828 https://www.mdpi.com/journal/sensors Sensors 2022, 22, 4828 2 of 9 Although a wide range of endurance runners tend to collide with the ground first with the heel when shod [3], the switch from shod to barefoot running implies a tendency toward a midfoot (MFS) or forefoot strike pattern (FFS), influencing factors such as contact (CT) and flight time (FT), step frequency (SF), step length (SL), loading rate and leg compliance [4–7]. While leg stiffness increases with barefoot running in comparison to shod running [8], a significant reduction in running dynamic stability was found when changing from shod to barefoot conditions [9]. Moreover, greater activation levels in the intrinsic muscles of the foot have been found with the stance phase in barefoot running, in comparison with shod running, producing adjustment during the compression of the longitudinal arch. This results in a greater ability to recoil elastic energy when running barefoot [10,11]. The recent appearance of wearable power meters on the running scene may change training and competition by providing power output values for endurance runners. These sensors may allow us to monitor and quantify workload from a fair and objective perspec- tive with accurate replication, as they already do in cycling [12]. Velocity and both the body height and weight of a runner, as well as external conditions such as slope and wind, may influence power output in running [13,14]. Although the level of agreement between power meter systems in running and two theoretical models for power output analysis has been assessed [15], the lack of scientific evidence for the use and interpretation of such metrics in endurance runners may prevent sport practitioners from adopting them as a means to monitor and assess running performance. A recent wearable system (i.e., Stryd™) calculates power production while running, separating this metric into two parts: power and form power. Apparently, power reflects the power output associated with changes in the athlete’s horizontal movement. Form power, however, represents the power output production caused by the combination of the oscillatory up and down movements of the centre of mass and lateral power when the athlete moves forward. This system employs mathematical calculations to estimate these two parameters from kinematic data collected from the described movements executed by the runner’s foot [16]. In addition, in a recent review [17], the Stryd foot pod was noted for its reliability and compatibility with metabolic power, compared to other commercially available portable running power devices. While several studies have already analysed power output in running [18,19] and others have investigated the relation between VO2max and power production [16,20], to the best of the authors’ knowledge, there are no studies assessing the difference in power output between shod and barefoot running. In order to bridge this gap, this study aims to identify the effect of footwear on power output in endurance runners. It is hypothesised that increased effectiveness, leg stiffness and power production would be identified in barefoot running. 2. Materials and Methods 2.1. Subjects Forty-one recreationally active endurance runners (25 males; age = 28.5 ± 6.9 years; height = 1.73 ± 0.08 m; body mass = 68.2 ± 11.6 kg), recruited by convenience, volunteered to take part in this study. All the participants were 18 years of age or older, capable of running 10,000 m in under 50 min (44.02 ± 4.22 min), injury-free for the last 6 months and were completing no fewer than 2 running sessions per week, therefore meeting the inclusion criteria. Every participant signed a formal consent form, aligned with the bioethics of the World Medical Association’s Declaration of Helsinki (2013). Once the objectives and procedures of the study were explained, participants were assured that they were free to leave the study at any time. The study was approved by the Ethics Committee of San Jorge University (009-18/19), from which sport sciences students were recruited. 2.2. Procedures Participants completed two testing trials over a one-session running protocol at their preferred comfortable velocity (11.71 ± 1.07 km·h−1) for data collection at the San Jorge University Biomechanics Laboratory (Zaragoza, Spain) in April 2019. Both trials were Sensors 2022, 22, 4828 3 of 9 completed on a motorised treadmill with a slope maintained at 0% (HP cosmos Pulsar 4P; HP cosmos Sports & Medical, Gmbh, Nußdorf, Germany). Participants warmed up for 5 min on the treadmill where the velocity was increased and decreased several times until a comfortable velocity was achieved [21]. For each trial, participants completed two successive 3 min running bouts (i.e., shod for the first and barefoot for the latter), separated by a 2 min period to change from shod to barefoot condition. Since power output [19] and spatiotemporal parameters [22] reach a steady state in less than 2 min, data were recorded during both running trials and 6–8 strides were analysed [23]. 2.3. Materials and Testing Both body weight and height were measured for each participant, utilising a weigh- ing scale (Tanita BC-601; TANITA Corp., Maeno-Cho, Itabashi-ku, Tokyo, Japan) and a stadiometer (SECA 222; SECA Corp., Hamburg, Germany), respectively. For this study, a commercially available wearable power meter, Stryd™ (Stryd Summit Powermeter; Stryd, Inc., Boulder, CO, USA), was clipped on the laces of the runner’s shoe when running shod and placed and secured with tape on the runner’s instep during barefoot running (Figure 1). This lightweight, reinforced carbon-fibre foot pod (weight: 9.1 g) is based on a 6-axis inertial motion sensor (3-axis gyroscope, 3-axis accelerometer) and provides kinetic and kinematic data. During barefoot running, participants ran with socks to avoid friction injuries to the soles of their feet caused by the treadmill belt. When running shod, participants wore their traditional training shoes. The power meter was linked to the manufacturer’s mobile application (StrydApp, version 5.13), downloaded on a smartphone (iPhone 8, Apple Inc., Cupertino, CA, USA), for recording data. Sensors 2022, 22, x FOR PEER REVIEW 3 of 9 University Biomechanics Laboratory (Zaragoza, Spain) in April 2019. Both trials were completed on a motorised treadmill with a slope maintained at 0% (HP cosmos Pulsar 4P; HP cosmos Sports & Medical, Gmbh, Nußdorf, Germany). Participants warmed up for 5 min on the treadmill where the velocity was increased and decreased several times until a comfortable velocity was achieved [21]. For each trial, participants completed two suc- cessive 3 min running bouts (i.e., shod for the first and barefoot for the latter), separated by a 2 min period to change from shod to barefoot condition. Since power output [19] and spatiotemporal parameters [22] reach a steady state in less than 2 min, data were recorded during both running trials and 6–8 strides were analysed [23]. 2.3. Materials and Testing Both body weight and height were measured for each participant, utilising a weigh- ing scale (Tanita BC-601; TANITA Corp., Maeno-Cho, Itabashi-ku, Tokyo, Japan) and a stadiometer (SECA 222; SECA Corp., Hamburg, Germany), respectively. For this study, a commercially available wearable power meter, Stryd™ (Stryd Sum- mit Powermeter; Stryd, Inc., Boulder, CO, USA), was clipped on the laces of the runner’s shoe when running shod and placed and secured with tape on the runner’s instep during barefoot running (Figure 1). This lightweight, reinforced carbon-fibre foot pod (weight: 9.1 g) is based on a 6-axis inertial motion sensor (3-axis gyroscope, 3-axis accelerometer) and provides kinetic and kinematic data. During barefoot running, participants ran with socks to avoid friction injuries to the soles of their feet caused by the treadmill belt. When running shod, participants wore their traditional training shoes. The power meter was linked to the manufacturer’s mobile application (StrydApp, version 5.13), downloaded on a smartphone (iPhone 8, Apple Inc., Cupertino, CA, USA), for recording data. Figure 1. Representation for the placement of the Stryd™ power meter clipped on the laces of the runner’s shoe (left picture) and placed and secured with tape on the runner’s instep during barefoot running (central and right picture). Average power output (w; ratio of total of watts generated to total run time), form power (w; previously described), mean power output (w (MPO)), normalised MPO (w/kg (MPOnorm)), vertical oscillation (cm; quantity of up and down movement generated dur- ing running), leg stiffness (kN/m; ratio of the maximal force at the initial touchdown to the maximum leg compression at the middle of the stance phase) and running effective- ness (kg/N; ratio of speed to power) were obtained using the Stryd™ power meter. Additionally, the running spatiotemporal parameters of contact time (time the foot spends in contact with the ground (CT)), flight time (time from toes-off to initial contact of the same foot (FT)), step length (distance covered between initial contact of one foot and the initial contact of the other foot (SL)) and step frequency (number of ground con- tacts that occurred in a minute (SF)) were also measured utilising the Stryd™ system, which has been previously validated for such purposes [18]. The foot strike pattern (FSP) exhibited by the participants was recorded using high- speed video at 240 Hz (Imaging Source DFK 33UX174, The Imaging Source Europe Figure 1. Representation for the placement of the Stryd™ power meter clipped on the laces of the runner’s shoe (left picture) and placed and secured with tape on the runner’s instep during barefoot running (central and right picture). Average power output (w; ratio of total of watts generated to total run time), form power (w; previously described), mean power output (w (MPO)), normalised MPO (w/kg (MPOnorm)), vertical oscillation (cm; quantity of up and down movement generated during running), leg stiffness (kN/m; ratio of the maximal force at the initial touchdown to the maximum leg compression at the middle of the stance phase) and running effectiveness (kg/N; ratio of speed to power) were obtained using the Stryd™ power meter. Additionally, the running spatiotemporal parameters of contact time (time the foot spends in contact with the ground (CT)), flight time (time from toes-off to initial contact of the same foot (FT)), step length (distance covered between initial contact of one foot and the initial contact of the other foot (SL)) and step frequency (number of ground contacts that occurred in a minute (SF)) were also measured utilising the Stryd™ system, which has been previously validated for such purposes [18]. The foot strike pattern (FSP) exhibited by the participants was recorded using high- speed video at 240 Hz (Imaging Source DFK 33UX174, The Imaging Source Europe GmbH; Bremen, Germany). The camera was placed perpendicular to the treadmill from a sagittal Sensors 2022, 22, 4828 4 of 9 view at 2 m from the centre of the treadmill and at a height of 0.30 m, which has been previously validated for such a purpose [24]. Three different FSP were identified in the present study [4]: rearfoot strike pattern (RFS), where the heel contacts the ground first; MFS, in which the outside edge of the foot contacts the ground first; and FFS, where the forefoot touches down first. 2.4. Statistical Analysis Descriptive data are shown as mean (±SD), frequency and percentage. To determine the differences between nominal variables, McNemar’s test was used. The mean differences between values were analysed via pairwise mean comparisons (t-test) and the magnitude of the differences was expressed by means of the Cohen’s d effect size (ES) and interpreted as trivial (<0.19), small (0.2–0.49), medium (0.5–0.79) and large (≥0.8) [25]. All statistical analyses were performed using SPSS (version 25, SPSS Inc., Chicago, IL, USA) and statistical significance was accepted at α = 0.05. 3. Results Significant differences (p < 0.05) were found in spatiotemporal gait characteristics during running when comparing shod and barefoot conditions (Table 1). When running barefoot, lower values for CT (p < 0.001, ES = 0.46) and SL (p = 0.003, ES = 0.13) were obtained with greater SF (p < 0.001, ES = 0.59), compared to those reported during shod running at the same comfortable velocity. The prevalence of FSP significantly differs (p < 0.034) between conditions, with 19.5% of runners showing RF, 56.1% MF and 24.4% FF during the shod condition, whereas no runners showed RF during barefoot running, with 31.8% and 68.2% showing MF and FF, respectively. Table 1. Spatiotemporal gait characteristics during running shod and barefoot at comfortable velocity. Shod Condition Barefoot Condition p-Value (d) FSP (n, %) ˆ RF 8 (19.5) 0 (0) 0.019 MF 23 (56.1) 13 (31.8) 0.033 FF 10 (24.4) 28 (68.2) 0.012 CT (s) 0.261 (0.020) 0.252 (0.019) <0.001 (0.46) FT (s) 0.111 (0.018) 0.108 (0.017) 0.053 (0.17) SL (m) 1.11 (0.15) 1.09 (0.15) 0.003 (0.13) SF (spm) 162.06 (8.06) 166.99 (8.22) <0.001 (0.59) ˆ indicates that a McNemar test was conducted to compare frequencies; d: Cohen’s d effect size; FSP: foot strike pattern; RF: rearfoot; MF: midfoot; FF: forefoot; CT: ground contact time; FT: flight time; SL: step length; SF: step frequency. The comparisons between conditions (i.e., shod vs. barefoot) revealed no differences in MPO (p = 0.582, ES = 0.02) and MPOnorm (p = 0.568, ES = 0.03), whereas significant differences were found in form power, in both absolute (p = 0.001, ES = 0.14) and relative values (p < 0.001, ES = 0.33), running effectiveness (p = 0.006, ES = 0.36), stiffness (p = 0.002, ES = 0.20) and vertical oscillation (p < 0.001, ES = 0.48) (Table 2). Table 2. Power output and related parameters during running shod and barefoot at comfortable velocity. Shod Condition Barefoot Condition p-Value (d) MPO (W) 210.05 (44.16) 210.73 (44.24) 0.582 (0.02) MPOnorm (W/kg) 3.07 (0.32) 3.08 (0.32) 0.568 (0.03) Form power (W) 69.95 (12.51) 68.28 (12.19) 0.001 (0.14) Form power (%) 33.6 (2.8) 32.7 (2.7) <0.001 (0.33) Running effectiveness 0.95 (0.05) 0.97 (0.06) 0.006 (0.36) Leg Stiffness (kN/m) 10.26 (1.86) 10.65 (1.93) 0.002 (0.20) Vertical oscillation (cm) 7.93 (0.98) 7.48 (0.90) <0.001 (0.48) d: Cohen’s d effect size; MPO: mean power output; MPOnorm: normalised mean power output. Sensors 2022, 22, 4828 5 of 9 4. Discussion This study sought to determine the effect of footwear on power output in long distance runners, comparing data collected by the Stryd™ system during both shod and barefoot running. The main finding of this study was that endurance runners showed greater running effectiveness when running barefoot in comparison with shod running, being consistent with lower values in form power and lower vertical oscillation. Additionally, our findings support those by previous studies reporting biomechanical alterations in runners who changed from traditional shoes to barefoot while running, such as the adoption of MFS or FFS, higher SF, and both shorter CT and SL [4–7]. Given the novelty of power output in endurance running, there is a lack of scientific evidence regarding this metric and, in particular, with power output in barefoot running, making this discussion challenging. From a biomechanical standpoint, the tendency towards the runner’s adoption of MF or FF when switching from traditional running shoes to barefoot is supported by Lieberman and colleagues, as they stated that habitually shod runners adopted flatter foot positions at the initial contact when barefoot running [4]. In the same line, the runner’s showed significantly higher SF (3%) when running barefoot (166.99 ± 8.22 spm, p < 0.001) during the present study; this reinforces previous works [5,6], which found that habitually shod runners significantly increased their SF when barefoot running [5] and as velocity increased [6]. Likewise, Cochrum and colleagues reported that barefoot running at 50% VO2max resulted in a 2.4% shorter SL in comparison with shod running conditions [6], endorsing the findings reported here, where SL is significantly shorter in barefoot running (1.83%) in comparison with running in shod conditions (1.09 ± 0.15 m and 1.11 ± 0.15 m, respectively; p < 0.05) at a comfortable velocity. Significantly smaller CT was also reported in barefoot running (0.252 ± 0.019 s, p < 0.001), supporting previous studies [5,7]. Divert and colleagues [5] reported that CT significantly increased in shod running, and Lussiana and colleagues [7] found shorter CT when comparing minimalist to traditional shoes. Although all these studies share shod– barefoot comparison in their procedures, the slight differences may be due to their different methodologies. While our participants completed one single, steady-state, comfortable velocity testing session on a motorised treadmill at a maintained slope of 0%, the studies discussed above are based on an incremental gradient protocol [7], a two-session protocol made up of six running bouts of 4 min [5], and four treadmill running testing sessions [6]. It should also be noted that none of those studies utilised the Stryd™ power meter to analyse the spatiotemporal parameters, in either the shod or barefoot condition. Considering that it has been proposed that the addition of 100 g to the foot reduces running economy by 1% [5,26], barefoot running would optimise the stretch-shortening cycle behaviour, buffering and releasing elastic energy [27] and increasing leg stiffness by the adoption of a plyometric movement pattern [28]. Moreover, as the foot’s core muscle system produces an adaptation during compression of the longitudinal arch, which results in increased ability to recoil elastic energy over the stance phase [10,11], one might expect greater power output in barefoot running compared to running in the shod condition. It should be noted that the footwear condition does not influence MPO or MPOnorm, but it does significantly influence form power (watts and percent), running effectiveness, leg stiffness and vertical oscillation in endurance runners. Given that power can be defined as the product of force and velocity [29], and that in the present study both running bouts (i.e., shod and barefoot) were executed at the same comfortable velocity for every participant, it seems reasonable that MPO and MPOnorm remained stable under both footwear conditions. The power output values reported in the present study during shod running (210.05 ± 44.16 W) are supported by those found by previous works using the Stryd™ power meter at the same running velocity [15,19]. From a practical application of MPO in endurance runners, the Functional Threshold Power is a performance index referring to the highest MPO maintained for around 60 min running without the onset of fatigue [30], commonly used to determine training intensities (i.e., training zones) and quantify athletes’ responses to training stimuli [30,31]. In a recent Sensors 2022, 22, 4828 6 of 9 study, MPO and MPOnorm have shown a strong relationship with the Functional Threshold Power at submaximal running from 10 min to 30 min [32]. Our results have shown no significant differences for MPO and MPOnorm between the shod and barefoot running conditions, suggesting no changes in load intensities between the two footwear conditions. Thus, a runner could maintain their training loads based on watts. However, the authors recommend being cautious with this information as intra-articular loads could be different due to biomechanical changes caused by barefoot running, such as the change in FSP from RFS to FFS, greater ankle stiffness, lower impact load and brake load, greater knee flexion at ground contact, and reduced tibialis anterior muscle activity, among other factors [33]. Regarding form power, and according to the manufacturer’s manual (https://www. stryd.com/guide (accessed on 1 June 2022)), most athletes would exhibit form power ranging from 30 to 100 W. The findings reported in our study show significant differences between both running conditions (69.95 ± 12.51 W when shod, 68.28 ± 12.19 W when barefoot) and seem to align with the manufacturer’s statement. It can be argued that form power relates to leg stiffness and vertical oscillation in endurance running in different ways. It is known that lower-limb stiffness varies across footwear conditions, resulting in increased leg stiffness when barefoot in comparison to shod running [5,7,8,10,27]. This statement is supported in the present study as the leg stiffness values are significantly greater in barefoot running (10.65 ± 1.93 kN/m). Since increased leg stiffness optimizes elastic energy recoil and enhances running economy [34], it is reasonable to find lower values of form power and increased leg stiffness in barefoot running, making the reverse equally valid in shod running. The values for vertical oscillation exhibit significant differences between shod (7.93 ± 0.98 cm) and barefoot running (7.48 ± 0.9 cm). The significantly lower values found in barefoot running demonstrate that runners show less vertical oscillation as they run, positively affecting running economy [35,36], which is also associated with the increased leg stiffness found under this footwear condition [7], consequently producing lower form power under this footwear condition. The Stryd™ system also offers a running effectiveness metric. This novel metric is referred to as the ratio of running velocity to power (https://www.trainingpeaks. com/blog/wko4-new-metrics-for-running-with-power/ (accessed on 1 June 2022)). The findings reported here align with the proposed values for this value (i.e., ~1 kg/N), showing significantly higher effectiveness (p < 0.005) in barefoot running (0.97 ± 0.06 kg/N). Although it has not been reported before, this metric might be useful for coaches and practitioners as white papers have stated that the closer the running effectiveness value to 1 kg/N, the more effective runners are at transforming external power into ve- locity (https://docs.google.com/document/u/2/d/e/2PACX-1vTzjH-Ns_GInUm4lAxi3 cVOQpzzKcWNF6VEX271s-QGYFHjwMgyLhhmu5i21-1_CaC3eL0B817rQo8k/pub (ac- cessed on 1 June 2022)). This metric must not be used interchangeably with running economy as they are completely different parameters. However, following the statements of the aforementioned white papers, running effectiveness might represent running econ- omy from a mechanical standpoint. Regarding running economy, referred to as the energy required to maintain submax- imal velocity efforts [37], several studies have not found significant differences between shod and barefoot running [38,39]. These studies did not control FSP, which may influ- ence these comparisons [5]. Of note, after controlling for maximal oxygen consumption (VO2max) and footwear conditions (i.e., barefoot, minimal, and traditional running shoes), Cochrum et al. [6] stated that barefoot running provides less metabolic benefit over cush- ioned shoes. This finding is supported by previous work, whose authors suggested that the design of cushioned shoes offers metabolic savings compared to barefoot running [28]. Our results, in contrast, provide evidence that barefoot running can be more metabolically beneficial than running in shoes. In addition, we believe that the transition from running in traditional shoes to less cushioned shoes, minimalist or barefoot running should be carried out gradually, as has already been recently proposed in a 10-week pain and injury free retraining program [40]. Sensors 2022, 22, 4828 7 of 9 The findings described here are based on entirely mechanical parameters; therefore, they should not be transferred to physiological terms. These findings should also not be extrapolated to injury management or competition, as we detailed the changes that occur in shod and barefoot running regarding kinetic and kinematics parameters. Finally, there are some limitations to consider. Firstly, the protocol was completed on a motorised treadmill at a comfortable velocity, preventing the readers from extrapolating these findings to other velocities. The participants wore their own running shoes in shod running, therefore increasing the ecological validity of the study. It would be of interest for the research community to assess running power output and related metrics considering different types of footwear, given the current revolution in the design of running shoes. The participants were habitually shod runners; therefore, the novelty of the task might influence the outcomes of barefoot running. Ultimately, the lack of scientific evidence on this topic made the discussion section especially complex. However, notwithstanding the aforementioned limitations, the present study offers new insights into the power production in endurance running, as well as the use of power meters and the interpretation of the metrics they provide, which might be of high value for clinicians, coaches and athletes who aim to introduce this metric into training and competition. 5. Conclusions The results obtained show that, besides the already known spatiotemporal gait char- acteristic adaptations, barefoot running reported greater values in running effectiveness in comparison with shod running, being consistent with lower values in form power and lower vertical oscillation related to running economics. Future studies are needed to exam- ine whether the long-term effect of short periods of barefoot running might contribute to increased running efficiency and leg stiffness in endurance runners, which would affect running economy. Author Contributions: D.J.-C., L.E.R.-S. and F.G.-P. defined the experimental design and conceptual- ized the approach. D.J.-C., A.M.-M., S.C.-S. and A.C.-L. collected the data. D.J.-C. and F.G.-P. carried out the statistical analysis. D.J.-C. wrote the paper. All authors reviewed the manuscript for scientific content. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki, and approved by the Ethics Committee of San Jorge University (009-18/19). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Acknowledgments: The authors would like to acknowledge the study participants involved in recruitment. Conflicts of Interest: The authors declare no conflict of interest. References 1. Lai, A.K.; Biewener, A.A.; Wakeling, J.M. Muscle-specific indices to characterise the functional behaviour of human lower-limb muscles during locomotion. J. Biomech. 2019, 89, 134–138. [CrossRef] [PubMed] 2. Li, F.; Wang, R.; Newton, R.U.; Sutton, D.; Shi, Y.; Ding, H. Effects of complex training versus heavy resistance training on neuromuscular adaptation, running economy and 5-km performance in well-trained distance runners. PeerJ 2019, 7, e6787. [CrossRef] [PubMed] 3. Hanley, B.; Bissas, A.; Merlino, S.; Gruber, A.H. Most marathon runners at the 2017 IAAF World Championships were rearfoot strikers, and most did not change footstrike pattern. J. Biomech. 2019, 92, 54–60. [CrossRef] [PubMed] 4. Lieberman, D.E.; Venkadesan, M.; Werbel, W.A.; Daoud, A.I.; D’Andrea, S.; Davis, I.S.; Mang’Eni, R.O.; Pitsiladis, Y. Foot strike patterns and collision forces in habitually barefoot versus shod runners. Nature 2010, 463, 531–535. [CrossRef] 5. Divert, C.; Mornieux, G.; Freychat, P.; Baly, L.; Mayer, F.; Belli, A. Barefoot-Shod Running Differences: Shoe or Mass Effect? Int. J. Sports Med. 2008, 29, 512–518. [CrossRef] 6. Cochrum, R.G.; Connors, R.T.; Coons, J.M.; Fuller, D.K.; Morgan, D.W.; Caputo, J.L. Comparison of Running Economy Values While Wearing No Shoes, Minimal Shoes, and Normal Running Shoes. J. Strength Cond. Res. 2017, 31, 595–601. [CrossRef] Sensors 2022, 22, 4828 8 of 9 7. Lussiana, T.; Hébert-Losier, K.; Mourot, L. Effect of minimal shoes and slope on vertical and leg stiffness during running. J. Sport Health Sci. 2015, 4, 195–202. [CrossRef] 8. Sinclair, J.; Atkins, S.; Taylor, P.J. The Effects of Barefoot and Shod Running on Limb and Joint Stiffness Characteristics in Recreational Runners. J. Mot. Behav. 2016, 48, 79–85. [CrossRef] 9. Ekizos, A.; Santuz, A.; Arampatzis, A. Transition from shod to barefoot alters dynamic stability during running. Gait Posture 2017, 56, 31–36. [CrossRef] 10. Shih, Y.; Lin, K.-L.; Shiang, T.-Y. Is the foot striking pattern more important than barefoot or shod conditions in running? Gait Posture 2013, 38, 490–494. [CrossRef] 11. Perl, D.P.; Daoud, A.I.; Lieberman, D.E. Effects of footwear and strike type on running economy. Med. Sci. Sports Exerc. 2012, 44, 1335–1343. [CrossRef] [PubMed] 12. Passfield, L.; Hopker, J.; Jobson, S.; Friel, D.; Zabala, M. Knowledge is power: Issues of measuring training and performance in cycling. J. Sports Sci. 2017, 35, 1426–1434. [CrossRef] [PubMed] 13. van Dijk, H.; van Megen, R. The Secret of Running: Maximum Performance Gains through Effective Power Metering and Training Analysis; Meyer & Meyer Sport: London, UK, 2017. 14. García-Pinillos, F.; Latorre-Roman, P.A.; Roche-Seruendo, L.E.; García-Ramos, A. Prediction of power output at different running velocities through the two-point method with the Stryd™ power meter. Gait Posture 2019, 68, 238–243. [CrossRef] 15. Cerezuela-Espejo, V.; Hernández-Belmonte, A.; Courel-Ibáñez, J.; Conesa-Ros, E.; Martínez-Cava, A.; Pallarés, J.G. Running power meters and theoretical models based on laws of physics: Effects of environments and running conditions. Physiol. Behav. 2020, 223, 112972. [CrossRef] [PubMed] 16. Austin, C.; Hokanson, J.; McGinnis, P.; Patrick, S. The Relationship between Running Power and Running Economy in Well- Trained Distance Runners. Sports 2018, 6, 142. [CrossRef] 17. Jaén-Carrillo, D.; Roche-Seruendo, L.E.; Cartón-Llorente, A.; Ramírez-Campillo, R.; García-Pinillos, F. Mechanical Power in Endurance Running: A Scoping Review on Sensors for Power Output Estimation during Running. Sensors 2020, 20, 6482. [CrossRef] [PubMed] 18. García-Pinillos, F.; Roche-Seruendo, L.E.; Marcén-Cinca, N.; Marco-Contreras, L.A.; Latorre-Román, P.A. Absolute Reliability and Concurrent Validity of the Stryd System for the Assessment of Running Stride Kinematics at Different Velocities. J. Strength Cond. Res. 2021, 35, 78–84. [CrossRef] 19. García-Pinillos, F.; Soto-Hermoso, V.M.; Latorre-Román, P.; Párraga-Montilla, J.A.; Roche-Seruendo, L.E. How Does Power During Running Change when Measured at Different Time Intervals? Laryngo-Rhino-Otol. 2019, 40, 609–613. [CrossRef] 20. Cerezuela-Espejo, V.; Hernández-Belmonte, A.; Courel-Ibáñez, J.; Conesa-Ros, E.; Mora-Rodríguez, R.; Pallarés, J.G. Are we ready to measure running power? Repeatability and concurrent validity of five commercial technologies. Eur. J. Sport Sci. 2020, 21, 341–350. [CrossRef] 21. Weir, G.; Willwacher, S.; Trudeau, M.B.; Wyatt, H.; Hamill, J. The Influence of Prolonged Running and Footwear on Lower Extremity Joint Stiffness. Med. Sci. Sports Exerc. 2020, 52, 2608–2614. [CrossRef] 22. García-Pinillos, F.; Latorre-Román, P.A.; Ramírez-Campillo, R.; Párraga-Montilla, J.A.; Roche-Seruendo, L.E. Minimum time required for assessing step variability during running at submaximal velocities. J. Biomech. 2018, 80, 186–195. [CrossRef] [PubMed] 23. Besser, M.P.; Kmieczak, K.; Schwartz, L.; Snyderman, M.; Wasko, J.; Selby-Silverstein, L. Representation of temporal spatial gait parameters using means in adults without impairment. Gait Posture 1999, 9, 113. 24. Esculier, J.-F.; Silvini, T.; Bouyer, L.J.; Roy, J.-S. Video-based assessment of foot strike pattern and step rate is valid and reliable in runners with patellofemoral pain. Phys. Ther. Sport 2018, 29, 108–112. [CrossRef] [PubMed] 25. Cohen, J. Statistical Power Analysis for the Behavioral Sciences; Routledge: London, UK, 2013. 26. Franz, J.R.; Wierzbinski, C.M.; Kram, R. Metabolic cost of running barefoot versus shod: Is lighter better? Med. Sci. Sports Exerc. 2012, 44, 1519–1525. [CrossRef] [PubMed] 27. Divert, C.; Mornieux, G.; Baur, H.; Mayer, F.; Belli, A. Mechanical Comparison of Barefoot and Shod Running. Laryngo-Rhino-Otol. 2005, 26, 593–598. [CrossRef] 28. Spurrs, R.W.; Murphy, A.J.; Watsford, M.L. The effect of plyometric training on distance running performance. Eur. J. Appl. Physiol. 2003, 89, 1–7. [CrossRef] 29. Halliday, D.R.; Resnick, R. Fundamentals of Physics; Wiley: New York, NY, USA, 2007. 30. Allen, H.; Coggan, A.R.; McGregor, S. Training and Racing with a Power Meter; VeloPress: Boulder, CO, USA, 2019. 31. Lajoie, C.; Laurencelle, L.; Trudeau, F. Physiological Responses to Cycling for 60 Minutes at Maximal Lactate Steady State. Can. J. Appl. Physiol. 2000, 25, 250–261. [CrossRef] 32. Cartón-Llorente, A.; García-Pinillos, F.; Royo-Borruel, J.; Rubio-Peirotén, A.; Jaén-Carrillo, D.; Roche-Seruendo, L.E. Estimating Functional Threshold Power in Endurance Running from Shorter Time Trials Using a 6-Axis Inertial Measurement Sensor. Sensors 2021, 21, 582. [CrossRef] 33. Hall, J.P.L.; Barton, C.; Jones, P.R.; Morrissey, D. The Biomechanical Differences Between Barefoot and Shod Distance Running: A Systematic Review and Preliminary Meta-Analysis. Sports Med. 2013, 43, 1335–1353. [CrossRef] 34. Albracht, K.; Arampatzis, A. Exercise-induced changes in triceps surae tendon stiffness and muscle strength affect running economy in humans. Eur. J. Appl. Physiol. 2013, 113, 1605–1615. [CrossRef] Sensors 2022, 22, 4828 9 of 9 35. Halvorsen, K.; Eriksson, M.; Gullstrand, L. Acute Effects of Reducing Vertical Displacement and Step Frequency on Running Economy. J. Strength Cond. Res. 2012, 26, 2065–2070. [CrossRef] [PubMed] 36. Cavagna, G.A.; Heglund, N.C.; Willems, P.A. Effect of an increase in gravity on the power output and the rebound of the body in human running. J. Exp. Biol. 2005, 208, 2333–2346. [CrossRef] [PubMed] 37. Daniels, J.T. A physiologist’s view of running economy. Med. Sci. Sports Exerc. 1985, 17, 332–338. [CrossRef] [PubMed] 38. Hanson, N.J.; Berg, K.; Deka, P.; Meendering, J.R.; Ryan, C. Oxygen cost of running barefoot vs. running shod. Int. J. Sports Med. 2011, 32, 401–406. [CrossRef] [PubMed] 39. Squadrone, R.; Gallozzi, C. Biomechanical and physiological comparison of barefoot and two shod conditions in experienced barefoot runners. J. Sports Med. Phys. Fit. 2009, 49, 6. 40. Molina-Molina, A.; Latorre-Román, P. Mercado-Palomino, E.; Delgado-García, G.; Richards, J.; Soto-Hermoso, V.M. The effect of two retraining programs, barefoot running vs increasing cadence, on kinematic parameters: A randomized controlled trial. Scand. J. Med. Sci. Sports 2022, 32, 533–542. [CrossRef]

Influence of the Shod Condition on Running Power Output: An Analysis in Recreationally Active Endurance Runners.

06-26-2022

Jaén-Carrillo, Diego,Roche-Seruendo, Luis E,Molina-Molina, Alejandro,Cardiel-Sánchez, Silvia,Cartón-Llorente, Antonio,García-Pinillos, Felipe

eng

PMC4552464

Additional file 1: Online Survey Dam tot Damloop File format: pdf Online Survey Dam tot Damloop 2014 The aim of this study is to gain insight in the positive and negative effects of training for the Dam tot Damloop on participants. This study is conducted by the Amsterdam University of Applied Sciences. You are invited to participate in this study, because you are registered for the Dam tot Damloop 2014. Your participation to this study is voluntarily. You can decide if you want to participate and you are allowed to quit at any time. For this study, we ask you to fill in an online survey, this will take approximately 15 minutes. After 6 months we will send you another survey. By then, you can decide if you want to participate in this follow-up survey. Your answers will be kept strictly confidential. If you have questions about this study, please contact Dr. Marije Baart de la Faille – Deutekom (m.baart.de.la.faille@hva.nl). Please indicate if you would like to participate in this research. If you do not want to participate, click on “no”. If you mark “yes” this means you: - have read the information described above - participate voluntarily - are 18 years or older  Yes  No For which distance did you subscribe?  16 KM (10 EM)  6.4 KM (4 EM)  Other (please give further information) … How often did you participate in the Dam tot Damloop?  This was the first time  This was the second time  This was the third time  This was the fourth time  This was the fifth time  This was the sixth time or more often  Do not know / no answer Did you actually participate in the Dam tot Damloop?  Yes  No What was the reason for not starting?  Being sick  Injury  Overtraining  I did not want to start  Weather conditions  Family or personal circumstances  Other, namely … Did you train for the Dam tot Damloop?  Yes  No Did you finish the Dam tot Damloop?  Yes  No What was your time at the Dam tot Damloop? … hours … minutes Do you usually take your phone with you during running?  Yes  No Did you use an app for exercising?  Yes  No Which app did you use during training for the Dam tot Damloop?  Dam tot Damloop 2014 app  Myasics  Adidas miCoach  RunKeeper  Get Runningapp  Nike + iPod / I Phone app  Runtastic  Strava  Endomundo  App with Renate Wennemars: Running Coach powered by the athletics union Could you indicate what the duration of the period of training was for the Dam tot Damloop?  Did not train or barely  Trained 1-5 weeks  Trained 6-11 weeks  Trained 12 weeks or more  No separate training period (I exercise throughout the whole year)  Do not know / no answer We are interested in the consequence of your participation in the Dam tot Damloop on amount of physical exercise. How much kilometres did you train per week prior to your training period for the Dam tot Damloop? Less than 5 km per week 5-10 km per week 10-20 km per week 20-30 km per week More than 30 km per week Do not know / no answer How much kilometres did you train per week during your training period for the Dam tot Damloop? Less than 5 km per week 5-10 km per week 10-20 km per week 20-30 km per week More than 30 km per week Do not know / no answer Do you think that training for the Dam tot Damloop had an effect your health?  No effect  Yes, I feel much healthier  Yes, I feel healthier  Yes, I feel less healthy  Yes, I feel much less healthy In total, how many times did you perform sports during the last 12 months? If you do not know the exact number, please give an estimation that is as accurate as possible. … times We are interested in the effect of your participation in the Dam tot Damloop on the behaviour that influences your health. Previously, questions about sports and exercise have been asked. That is why we, in the next section, ask for other aspects of behaviour that may have been influenced by the Dam tot Damloop. Alcohol consumption On average, how many glasses of alcohol did you drink per week prior to your training period for the Dam tot Damloop? None 1-3 glasses per week 4-7 glasses per week 8-14 glasses per week More than 14 glasses per week Do not know / no answer On average, how many glasses of alcohol did you drink per week during your training period for the Dam tot Damloop? None 1-3 glasses per week 4-7 glasses per week 8-14 glasses per week More than 14 glasses per week Do not know / no answer Smoking behaviour How often did you smoke prior to your training period for the Dam tot Damloop? Never Occasionally 1-3 pieces a day 4-10 pieces a day More than 10 pieces a day Do not know / no answer How often did you smoke during your training period for the Dam tot Damloop? Never Occasionally 1-3 pieces a day 4-10 pieces a day More than 10 pieces a day Do not know / no answer To what extent do you agree with the following theses in relation to the training for the Dam tot Damloop? I eat healthier. Totally agree Agree Neutral Do not agree Totally do not agree I feel more energetic. Totally agree Agree Neutral Do not agree Totally do not agree I know that performing sports is not my thing. Totally agree Agree Neutral Do not agree Totally do not agree The chance is high that I will keep on performing sports on the long-term. Totally agree Agree Neutral Do not agree Totally do not agree I feel better about myself. Totally agree Agree Neutral Do not agree Totally do not agree I see myself more as an athlete. Totally agree Agree Neutral Do not agree Totally do not agree I did not change anything in my lifestyle. Totally agree Agree Neutral Do not agree Totally do not agree I encouraged others in my surrounding to perform sports. Totally agree Agree Neutral Do not agree Totally do not agree I lost weight. Totally agree Agree Neutral Do not agree Totally do not agree I feel tired more often. Totally agree Agree Neutral Do not agree Totally do not agree Are you a man or a woman?  Man  Woman Wat Is your body height in centimetres at this moment? … cm What is your body weight in kilogrammes at this moment? … kg What is your year of birth? ….

App use, physical activity and healthy lifestyle: a cross sectional study.

08-28-2015

Dallinga, Joan Martine,Mennes, Matthijs,Alpay, Laurence,Bijwaard, Harmen,Baart de la Faille-Deutekom, Marije

eng

PMC9012714

Vol.:(0123456789) 1 3 European Journal of Applied Physiology (2022) 122:1179–1187 https://doi.org/10.1007/s00421-022-04903-9 ORIGINAL ARTICLE Acute intense fatigue does not modify the effect of EVA and TPU custom foot orthoses on running mechanics, running economy and perceived comfort Ken Van Alsenoy1,2 · Joong Hyun Ryu3 · Olivier Girard1,4 Received: 23 August 2021 / Accepted: 29 January 2022 / Published online: 24 February 2022 © The Author(s) 2022 Abstract We determined whether fatigue modifies the effect of custom foot orthoses manufactured from ethyl-vinyl acetate (EVA) and expanded thermoplastic polyurethane (TPU) materials, both compared to standardized footwear (CON), on running mechanics, running economy, and perceived comfort. Eighteen well-trained, males ran on an instrumented treadmill for 6 min at the speed corresponding to their first ventilatory threshold (13.8 ± 1.1 km/h) in three footwear conditions (CON, EVA, and TPU). Immediately after completion of a repeated-sprints exercise (8 × 5 s treadmill sprints, rest = 25 s), these run tests were replicated. Running mechanics, running economy and perceived comfort were determined. Two-way repeated measures ANOVA [condition (CON, EVA, and TPU) × fatigue (fresh and fatigued)] were conducted. Flight time shortened (P = 0.026), peak braking (P = 0.016) and push-off (P = 0.032) forces decreased and vertical stiffness increased (P = 0.014) from before to after the repeated-sprint exercise, independent of footwear condition. There was a global fatigue-induced deterioration in running economy (− 1.6 ± 0.4%; P < 0.001). There was no significant condition × fatigue [except mean loading rate (P = 0.046)] for the large majority of biomechanical, cardio-respiratory [except minute ventilation (P = 0.020) and breathing frequency (P = 0.019)] and perceived comfort variables. Acute intense fatigue does not modify the effect of custom foot orthoses with different resilience characteristics on running mechanics, running economy and perceived comfort. Keywords Fatigue · Orthotics · Material resilience · Stride pattern · Economy of locomotion · Footwear comfort Abbreviations ANOVA Repeated-measures analysis of variance CFO Custom foot orthotics CON Control condition EVA Ethyl-vinyl acetate GRF Ground reaction force RE Running economy RPE Ratings of perceived exertion TPU Thermoplastic polyurethane Introduction Custom foot orthoses (CFOs), which refer to shoe inserts built from a three-dimensional representation of the ath- lete’s feet, have become a contemporary topic in footwear biomechanics literature. Wearing CFOs is used to provide foot support and shock absorption during ground contact through re-distribution of plantar loading and a better maintenance of foot stability (Crago et al. 2019). Prior studies investigating CFOs effects on key biomechani- cal indicators have reported mixed results with beneficial (Worobets et al. 2014; Wilkinson et al. 2018) or unchanged (Lewinson et al. 2013) adjustments in stride pattern. Dis- crepant findings may relate to intrinsic properties (i.e., energy return and longitudinal bending stiffness) of tested insoles provoking specific biomechanical modifications Communicated by Jean -Rene Lacour. * Ken Van Alsenoy Ken.VanAlsenoy@aspetar.com * Olivier Girard olivier.girard@uwa.edu.au 1 Aspetar, Orthopaedic and Sports Medicine Hospital, FIFA Medical Centre of Excellence, Doha, Qatar 2 Centre for Health, Activity and Rehabilitation Research (CHEARR), Queen Margaret University, Edinburgh, UK 3 Sports Science Department, Aspire Academy, Doha, Qatar 4 School of Human Sciences (Exercise and Sport Science), The University of Western Australia, Perth, WA, Australia 1180 European Journal of Applied Physiology (2022) 122:1179–1187 1 3 in running gait (Sinclair et al. 2016). Quantifying stride mechanical adjustments in response to CFOs is crucial to better understand how footwear features may eventually influence the energetic cost of running (Moore 2016). Running economy (RE), the steady-state oxygen uptake at a constant submaximal speed, is considered a key physi- ological measure for distance runners (Barnes and Kilding 2015). The effect of inserts (foot orthoses and sock absorb- ing insoles) on RE in distance runners has produced incon- sistent results (Crago et al. 2019). In a study by Burke and Papuga (2012), six recreational athletes consumed at least 3% less oxygen while running with CFOs compared to their shoe-fitted insoles. In contrast, the effect of wear- ing CFOs manufactured from ethyl-vinyl acetate (EVA) and expanded thermoplastic polyurethane (TPU) materi- als, both compared to standardized (shoe only), on RE was considered negligible and marginally improved (albeit not significantly), respectively (Van Alsenoy et al. 2019). Additionally, it is known that the amount of cushioning material in the shoe can influence RE (Tung et al. 2014), and that these effects might be mediated by comfort per- ception (Mundermann et al. 2003). Footwear comfort is paramount since it largely influ- ences the adherence to ongoing use of CFOs. The percep- tion of load attenuation (i.e., perceived comfort), result- ing from the level of somatosensory feedback experienced (Robbins and Hanna 1987), has been related to RE (Luo et  al. 2009; Lindorfer et  al. 2020). Comfort-induced changes, as a result of load attenuation on certain anatomi- cal foot structures, potentially contribute to reduction in metabolic demands via more economical stride character- istics (Moore et al. 2014). Analyzing the runner’s percep- tion of the CFOs’ cushioning properties and the associated biomechanical adjustments induced by different inserts features (yet with identical geometry) may also help to better elucidate their effectiveness at protecting against fatigue effects (Hintzy et al. 2015). Tolerance to ground impact is often compromised as fatigue appears, which in turn may limit performance and/or increase injury risk, notably by increasing the magnitude and rate of loading (Li et al. 2020). To date, little attention has been paid to the effectiveness of CFOs at reducing impact loading in situations of intense fatigue (i.e., repeated ‘all out’ efforts; Girard et al. 2020), with participants typically tested in ‘fresh’ conditions only. Reportedly, CFOs reduced plantar loading under the hallux, medial midfoot, and lateral midfoot compared to prefabricated insoles by ~ 30–35% post-fatigue (12 min at treadmill speed of ~ 14.4 km/h) (Lucas-Ceuvas et al. 2014). It is therefore plausible that the use of CFOs may become a more important protective mechanism for excessive mechanical constraints in the lower extremities once runners become fatigued. Because increased fatigue differently affects the biomechanical pattern of running (Brocherie et al. 2016), assessing the effects of CFOs for spatio-temporal, spring-mass model and antero-posterior variables is relevant. This study determined whether acute intense fatigue modifies the effect of CFOs manufactured from ethyl-vinyl acetate (EVA) and expanded thermoplastic polyurethane (TPU) materials, both compared to standardized footwear (CON), on running mechanics, RE, and perceived comfort. Methods Participants Eighteen male well-trained athletes (mean ± SD age, 38.9 ± 5.1 years; body height, 175.3 ± 5.8 cm; body mass 74.9 ± 7.7 kg; maximal oxygen uptake, 49.1 ± 6.6 mL/min/ kg; maximal aerobic speed, 18.4 ± 1.6 km/h) were recruited for this study. They trained on average 8.8 ± 3.7 h per week in the 3 months leading up to the data collection with an average weekly running distance of 37.6 ± 26.7 km. Thirteen were rear-foot strikers, one was a midfoot striker and four were forefoot strikers at 10 km/h. Written informed consent was obtained from participants, and the study was approved by Anti-Doping Laboratory Ethics Committee in Qatar (IRB Application Number 2017000201) and conducted according to the Declaration of Helsinki. Study design About 1 week before testing, participants undertook a pre- liminary session. They completed a continuous, maximal incremental running test where the individual ventilatory threshold, and corresponding running speed that was used for the three following intervention sessions, were deter- mined. Briefly, participants started running at 9 km/h with speed increases of 0.5 km/h every 30 s. The test ended with voluntary exhaustion of the participants. Verbal encourage- ment was only given by the researcher guiding the runners throughout the session. Ventilatory threshold was deter- mined using the criteria of an increase in minute ventila- tion/oxygen uptake with no increase in minute ventilation/ carbon dioxide and the departure from linearity of minute ventilation (Davis 1985). On three occasions, participants performed (in a counter- balanced randomized crossover design), at the same time of day (± 1 h) and 4–5 days apart, an exercise protocol (see below) in different footwear conditions: a control session where participants ran with standardized (i.e., only shoe liner inserted) footwear, CFOs made of EVA and TPU. After arrival to the laboratory, CFOs were inserted bilaterally in participants’ shoes. The participants and the researcher who was directly involved in guiding the session were visually 1181 European Journal of Applied Physiology (2022) 122:1179–1187 1 3 blinded from the CFO materials. Participants were asked to avoid strenuous exercise in the 12 h, as well as refrain from food and caffeine for 4 h preceding their visits to the laboratory and were encouraged to replicate their diet and training pattern for all visits. Laboratory conditions were similar throughout all running sessions (mean ± SD tempera- ture 20.7 ± 0.2 °C, relative humidity 60.4 ± 0.6%). Time of day was standardized for each participant over all sessions. Exercise protocol After a 10 min warm-up at 10 km/h, followed by a 3 min break used to put on the mask to collect expired gases, par- ticipants ran for 6 min at the speed associated with their first ventilatory threshold (13.8 ± 1.1 km/h) whereby run- ning mechanics, RE and perceived comfort were evaluated. Participants were then allowed 5 min to rest in a standing position prior to undertaking a fatiguing task that consisted of performing eight, 5 s sprints separated by 25 s of rest (Girard et al. 2020). Lastly, 2 min after the termination of the fatiguing task, participants repeated the 6 min run trial. The complete timing sequence from warm-up to finish was strictly controlled and guided by visual and verbal cues. Footwear During all running, the participants used neutral-like run- ning shoes (Pearl Izumi N2v2, Colorado, US) with an average European shoe size of 43.6 ± 1.6, a stack height of 23–24 mm and a heel drop of 4 mm. The two pairs of CFOs used by participants were based on an individual non- weight-bearing 3D scan of the foot using a Delcam iCube scanner (Elinvision, Karmėlava, Lithuania). CFOs were designed by a sport podiatrist with nearly 20 years of expe- rience, using the Orthomodel Pro CAD software (Autodesk, California, USA). Briefly, scans were imported into the soft- ware, markers were placed over the heel, first- and fifth met- atarsal and medial arch. A base model surface was adjusted to match the contour of the foot using cross-sectional views from the heel to the forefoot. The thickness of the orthotic was arbitrary set to 8 mm in an attempt to maximize the potential of the TPU beats inside the Infinergy® material (BASF, Ludwigshafen, Germany). All CFOs were direct- milled out of EVA and TPU materials and manually fin- ished to fit inside the shoes. Wear-in time between the first and second intervention session was 4.5 ± 2.5 days and 4.6 ± 2.8 days between the second and last intervention ses- sion. The mass of the three footwear conditions was on aver- age 600.3 ± 32.0 g, 647.3 ± 36.0 g and 681.1 ± 35.7 g for the shoes with its original liners (CON), with the custom EVA orthoses (EVA) and with the custom TPU orthoses (TPU), respectively. Running mechanics An instrumented treadmill [ADAL3D-WR, Medical Devel- opment—HEF Tecmachine, France; for details, see Belli et al. (2001)] was used for all running conditions. Briefly, it is mounted on a highly rigid metal frame, set at 0°grade incline, fixed to the ground through four piezoelectric force transducers (KI 9077b; Kistler, Winterthur, Switzerland) and installed on a specially engineered concrete slab to ensure maximal rigidity of the supporting ground (Girard et al 2017). In this study, the treadmill function was switched to either constant speed mode (i.e., to measure the constant speed running pattern with direct ground reaction force measurement) or constant motor torque mode (i.e., to allow participants to perform sprints; Morin et al. 2010). Over the last 2 minutes of each 6 min run, three-dimen- sional ground reaction force was continuously sampled at 1000 Hz. Ten consecutive steps recorded after running for ~ 4 min 15 s, ~ 4 min 45 s, ~ 5 min 15 s and ~ 5 min 45 s were subsequently averaged for final analysis. After appro- priate filtering (Butterworth-type 30 Hz low-pass filter), instantaneous data of vertical and antero-posterior ground reaction forces were averaged for each support phase when the vertical force was above 30 N. These data were deter- mined by measurement of the main spatio-temporal varia- bles: contact time (s), flight time (s) and step frequency (Hz) were reported. Peak braking and peak push-off forces (BW) along with duration of braking and push-off phases (s) were determined. Finally, average vertical loading rate (BW/s) was calculated as the mean value of the time-derivate of vertical force signal within the first 50 ms of the support phase (Li et al. 2020). A linear spring-mass model paradigm was used to inves- tigate the main mechanical integrative variables character- izing the lower limb behavior during running (McMahon and Cheng 1990). Vertical stiffness (kN/m) was calculated as the ratio of peak vertical forces (N) to the maximal verti- cal downward displacement of center of mass (m), which was determined by double integration of vertical accel- eration of center of mass over time during ground con- tact (Cavagna 1975). Leg stiffness (kN/m) was calculated as the ratio of peak vertical forces to the maximum leg spring compression [maximal vertical downward displace- ment + L0-√L0 2–(0.5 × running speed × contact time)2, in m], both occurring at mid-stance (Morin et al. 2005). Initial leg length (L0, great trochanter to ground distance in a stand- ing position) was determined from participant’s stature as L0 = 0.53 × stature (Morin et al. 2005). 1182 European Journal of Applied Physiology (2022) 122:1179–1187 1 3 Cardio‑respiratory variables Expired gases were collected by a metabolic cart (Jeager™ Oxycon Mobile, Carefusion, Hoechberg, Germany). Prior to each session, calibration of gas sensor was completed for ambient air and a known gas mixture (16% oxygen, 5% carbon dioxide). Turbine was calibrated using a 3 Liter (± 0.4%) syringe and automated high and low flow ventila- tion. Breath-by-breath gas samples were first averaged every 15 s and subsequently expressed as the average of the last 2 minutes of each 6 min run. Oxygen uptake expressed in both absolute (mL/min) and relative (mL/kg/min) terms, minute ventilation (L/min), breathing frequency (breaths/ min), tidal volume (L) were determined. Heart rate (beats/ min) was continuously measured by short-range telemetry (Polar, Kempele, Finland). RE was calculated as the oxygen uptake per body mass over speed, expressed in milliliters of oxygen consumed per kilogram per kilometer (mL/kg/km). The metabolic cart was suspended from the ceiling next to participants, so they did not have to support the additional weight of the system when running. Perceptual and comfort measures Within the first minute after finishing low-speed and high- speed runs, a global (6 min run) rating of perceived exertion value was collected using the 6–20 Borg scale. A modified version of the footwear comfort assessment tool, developed and tested on reliability by Mundermann (2002), was used to assess comfort associated with wearing each footwear condition using an iPad mini (Apple, California, US). This scale was used in previous studies to assess perceived com- fort (McPoil et al. 2011; Burke and Papuga 2012). For this study, only six of the nine items (‘overall comfort’, ‘heel cushioning’, ‘forefoot cushioning’, ‘medio-lateral control’, ‘arch height’ and ‘heel cup fit’) were scored on a digital, 150 mm visual analogic scale where 0 was defined as ‘not comfortable at all’ and 150 ‘most comfortable condition imaginable’. Statistical analysis Values are presented as mean ± SD. Two-way repeated measures analysis of variance (ANOVAs) [Condition (CON, EVA, TPU) × Fatigue (fresh and fatigued)] were used to compare investigated variables. To assess assumptions of variance, Mauchly’s test of sphericity was performed using all ANOVA results. A Greenhouse–Geisser correction was performed to adjust the degree of freedom if an assumption was violated, while post hoc pairwise-comparisons with Bonferroni-adjusted P values were performed if a signifi- cant main effect was observed. Partial eta-squared (ηp 2, with ηp 2 ≥ 0.06 representing a moderate effect and ηp 2 ≥ 0.14 a large effect) values were calculated. All statistical calcula- tions were performed using SPSS statistical software V.26.0 (IBM Corp., Armonk, USA). The significance level was set at P < 0.05. Results Running mechanics (Table 1) There was a significant main condition effect for mean loading rate, push-off duration and push-off peak force (all P ≤ 0.017; 0.27 ≤ ηp 2 ≤ 0.47). A significant main fatigue effect was noted for four out of nine variables studied: flight time, vertical stiffness, as well as braking and push- off durations (all P ≤ 0.032; 0.26 ≤ ηp 2 ≤ 0.32). There was no significant condition × fatigue [except mean loading rate (P = 0.046; ηp 2 = 0.19)] interactions for any stride mechanical variable. Cardio‑respiratory variables (Table 2) There was a significant main condition effect for heart rate (P = 0.027; ηp 2 = 0.19) only. All examined cardiorespira- tory variables changed significantly from fresh to fatigued state (all P ≤ 0.021; 0.28 ≤ ηp 2 ≤ 0.83), except tidal volume (P = 0.507; ηp 2 = 0.03). Only minute ventilation (P = 0.020; ηp 2 = 0.21) and breathing frequency (P = 0.019; ηp 2 = 0.21) displayed significant condition × fatigue interactions. Perceptual and Comfort measures (Table 3) There was a significant main condition effect for medio-lateral control and arch height (all P ≤ 0.027; 0.20 ≤ ηp 2 ≤ 0.30). Increased ratings of perceived exertion (P < 0.001; ηp 2 = 0.71) occurred under fatigue, while heel cushioning (P = 0.021; ηp 2 = 0.29) was also rated as more comfortable. Discussion Running mechanics One strength of our study is that running mechanics dur- ing constant submaximal runs were derived from direct ground reaction forces’ recording in both vertical and antero-posterior directions, as opposed to previous studies using tri-axial accelerometers to assess the effects of CFOs before and after an intense run (Lucas-Cuevas et al. 2014; Lucas-Cuevas et al. 2017). Lower mean loading rates were 1183 European Journal of Applied Physiology (2022) 122:1179–1187 1 3 recorded while wearing CFO made of TPU compared to EVA or no insert (CON), likely due to the higher resilience material properties (not measured). Contrary to the present study, previous studies observed lower loading rates in stiffer shoes (Henning et al. 1996; Milani et al. 1997). The often-used statement that, based upon their perception, runners adapt their technique with stiffer footwear to avoid high impact rates over the heel by cushioning the ground is not supported. Biomechanical manifestation of fatigue resulting from completion of a repeated-sprint treadmill exercise was gen- erally not modified by inserts. Out of nine biomechanical variables, only contact time, mean loading rate and brak- ing phase duration displayed statistically significant con- dition × fatigue interactions, with no systematic advantage of one or the other CFO condition. With fatigue, however, wearing CFOs (EVA or TPU) only generated subtle kinetic (mean loading rate: 1–2 BW/s) and kinematic (contact time and braking phase duration: 1–3 ms) adjustments compared to CON. It must therefore be questioned whether such small fatigue-related differences between footwear conditions are clinically relevant. Despite different resilience character- istics but similar configuration of tested CFOs, our novel findings indicate that participants who wore CFOs, either Table 1 Changes in biomechanical parameters for shoe only (CON), shoe with Ethyl-Vinyl Acetate orthotic (EVA), and shoe with Thermoplastic Poly-Urethane orthotic (TPU) conditions before (Fresh) and after (Fatigued) the completion of a repeated-sprint treadmill exercise Bold values indicate statistically significant ANOVA P values (P < 0.05) Values are mean ± SD. C and F, respectively, refer to ANOVA main effects of condition and fatigue and interaction between these two factors with P value and partial eta-squared (η2) in parentheses. Bold values indicate statistically significant findings † Significantly different from TPU, P < 0.05 *Significantly different from Fresh, P < 0.05 Variables CON EVA TPU ANOVA P value (η2) C F C × F Contact time (ms)  Fresh 223 ± 17 225 ± 17 226 ± 17 0.121 0.332 0.404  Fatigued 225 ± 18 225 ± 18 226 ± 17 (0.12) (0.06) (0.06) Flight time (ms)  Fresh 117 ± 18 116 ± 19 117 ± 17 0.354 0.026 0.571  Fatigued 112 ± 18* 113 ± 17* 115 ± 17* (0.06) (0.27) (0.03) Step frequency (Hz)  Fresh 2.97 ± 0.15 2.96 ± 0.14 2.94 ± 0.14 0.076 0.069 0.953  Fatigued 2.95 ± 0.15 2.94 ± 0.12 2.94 ± 0.12 (0.15) (0.19) (0.01) Mean loading rate (BW/s)  Fresh 62.3 ± 13.8† 62.4 ± 14.0† 55.0 ± 10.6 < 0.001 0.684 0.046  Fatigued 60.4 ± 14.1†* 62.2 ± 14.6† 56.1 ± 12.3 (0.46) (0.01) (0.19) Vertical stiffness (kN/m)  Fresh 34.8 ± 4.5 35.5 ± 4.3 35.2 ± 4.6 0.617 0.014 0.878  Fatigued 36.0 ± 5.3* 36.4 ± 4.5* 36.4 ± 5.0* (0.02) (0.32) (0.01) Leg stiffness (kN/m)  Fresh 15.7 ± 2.5 15.6 ± 2.2 15.5 ± 2.2 0.88 0.788 0.152  Fatigued 15.4 ± 2.1 15.5 ± 2.1 15.7 ± 1.9 (0.01) (0.01) (0.11) Braking phase duration (ms)  Fresh 109 ± 9 109 ± 8 110 ± 8 0.828 0.423 0.32  Fatigued 110 ± 9 110 ± 8 109 ± 7 (0.01) (0.040) (0.07) Push-off phase duration (ms)  Fresh 114 ± 11† 115 ± 12 116 ± 12 0.003 0.309 0.368  Fatigued 115 ± 12† 115 ± 13 117 ± 12 (0.31) (0.06) (0.06) Peak braking force (kN)  Fresh 0.59 ± 0.11 0.58 ± 0.13 0.60 ± 0.11 0.253 0.016 0.895  Fatigued 0.58 ± 0.11* 0.57 ± 0.13* 0.59 ± 0.11* (0.08) (0.31) (0.01) Peak push-off force (kN)  Fresh 0.41 ± 0.06† 0.39 ± 0.07 0.39 ± 0.06 0.017 0.032 0.867  Fatigued 0.40 ± 0.06†* 0.39 ± 0.06* 0.38 ± 0.06* (0.27) (0.26) (0.1) 1184 European Journal of Applied Physiology (2022) 122:1179–1187 1 3 made of EVA or TPU materials, produced essentially similar fatigue-induced adjustments in their stride pattern at con- stant treadmill speed. Inspection of changes in biomechanical variables induced by the repeated-sprint treadmill exercise indicates that glob- ally running technique is not profoundly modified under acute intense fatigue. Regardless of footwear condition, the runners mainly reduced their flight time and increased their vertical stiffness to maintain treadmill speed constant, while the magnitude of these adjustments also did not dif- fer between the two running speeds. Previously, both low (10 km/h) and high (20 km/h) constant speed running pat- terns were found unchanged from before to ~ 3 min after repeated-sprint exercises [four sets of five, 6 s sprints with 24 s recovery and 3 min between sets (Morin et al. 2012); three sets of five, 5 s sprints with 25 s recovery and 3 min between sets (Girard et al. 2017)] using the ADAL treadmill. Despite exacerbated cardio-respiratory and perceptual (rat- ing of perceived exertion) responses occurring after sprint- ing repeatedly, tested individuals may have not reached exertion levels where biomechanical manifestation of fatigue would cause more stressful ground impacts. Cardio‑respiratory variables As expected, physiological responses were elevated between before and after the fatiguing exercise. No significant inter- action was found between fatigue and footwear condi- tions for key cardio-respiratory variables (i.e., heart rate, oxygen uptake). This finding provides evidence that both types of CFOs (TPU and EVA materials) were not able to protect against fatigue-related deterioration in RE and/ or elevations in physiological strain compared to the shoe only (CON) condition. This may not be surprising since no fatigue-induced differences in mechanical variables could be detected across footwear conditions. Similarly, wearing CFOs that alters neuromuscular control during a submaxi- mal 1 h treadmill run was found ineffective to reduce the aerobic cost of running (Kelly et al. 2011). In our study, the biomechanical effect of wearing CFOs with or without fatigue was probably too small to induce meaningful differ- ences in physiological variables (expect minute ventilation that was lower in TPU compared to CON and to a lower extent EVA) across conditions. Another interesting observation was that RE at a speed corresponding to the first ventilatory threshold deterio- rated by ~ 1.7% with fatigue. Results from Day and Hahn Table 2 Changes in cardio- respiratory parameters for shoe only (CON), shoe with Ethyl- Vinyl Acetate orthotic (EVA), and shoe with Thermoplastic Poly-Urethane orthotic (TPU) conditions before (Fresh) and after (Fatigued) the completion of a repeated-sprint treadmill exercise Bold values indicate statistically significant ANOVA P values (P < 0.05) Values are mean ± SD. C and F, respectively, refer to ANOVA main effects of condition and fatigue and interaction between these two factors with P value and partial eta-squared (η2) in parentheses. Bold values indicate statistically significant findings † Significantly different from TPU, P < 0.05 *Significantly different from Fresh, P < 0.05 Variables CON EVA TPU ANOVA P value (η2) C F C × F Heart rate (bpm)  Fresh 164 ± 13† 163 ± 13 162 ± 13 0.027 < 0.001 0.904  Fatigued 172 ± 13†* 171 ± 12* 170 ± 13* (0.19) (0.60) (0.01) Running economy (mL/kg/km)  Fresh 190 ± 11 185 ± 14 188 ± 11 0.077 0.020 0.746  Fatigued 193 ± 12* 189 ± 14* 191 ± 13* (0.14) (0.28) (0.02) Oxygen uptake (mL/kg/min)  Fresh 43.7 ± 4.0 42.5 ± 4.5 43.4 ± 5.0 0.087 0.021 0.796  Fatigued 44.4 ± 4.8* 43.4 ± 4.5* 44.0 ± 5.5* (0.13) (0.28) (0.02) Minute ventilation (L/min)  Fresh 101 ± 16 101 ± 16 100 ± 15 0.058 < 0.001 0.020  Fatigued 119 ± 20†* 116 ± 21†* 112 ± 19* (0.15) (0.71) (0.21) Breathing frequency (breath/min)  Fresh 42.6 ± 6.8 44.9 ± 10.7 44.5 ± 8.6 0.496 < 0.001 0.019  Fatigued 50.9 ± 8.4†* 50.2 ± 8.9* 49.0 ± 8.3* (0.03) (0.83) (0.21) Tidal volume (L)  Fresh 2.39 ± 0.43 2.31 ± 0.57 2.30 ± 0.48 (0.204) 0.507 0.420  Fatigued 2.37 ± 0.50 2.35 ± 0.56 2.33 ± 0.51 (0.09) (0.03) (0.05) 1185 European Journal of Applied Physiology (2022) 122:1179–1187 1 3 (2019) suggest that optimal footwear longitudinal bend- ing stiffness to improve RE in fresh conditions is speed dependent. Whereas most participants running at 14 km/h elicited a minimum metabolic rate in the normal shoe, an increased number of participants were more economical in the stiff shoe (despite it weighing an extra 50 g compared to the normal shoe) at 17 km/h. In our study, EVA and TPU were ~ 50 g (+ 8%) and ~ 80 g (+ 14%) heavier compared to CON, respectively. The relationship between shoe mass and energy cost suggests that energy demands during running are greater as shoe mass is increased (i.e., + 1% for every added 100 g per shoe; Frederick 1984). Consequently, we cannot exclude that the additional weight of inserts may have confounded any protective effect of CFOs on RE when fatigue sets in. While our approach did not account for dif- ferent footwear mass, imposing running speeds relative to our participants’ physiological capacity rather than absolute speeds, as commonly done in the CFO-related literature on RE (Crago et al. 2019), was a strength. Comfort measures Significantly improved perceived comfort for medio-lateral control (∼20%) and arch height (∼25%) was reported for both EVA and TPU, yet with no difference between the two inserts, compared to CON. Despite statistical significance was not reached, other perceived comfort-related metrics (heel and forefoot cushioning, heel cup fit) displayed simi- lar trends. However, the presumably greater levels of rigid- ity of the sole of the EVA insert did not generate greater discomfort for the runners, also with similar or improved (e.g., lower loading rates) running biomechanics. Overall, in line with previous research in fresh conditions (Lindorfer et al. 2020), increase in comfort at the foot/shoe interface did not lead to improved RE and meaningful changes in biomechanical variables. Perhaps functional biomechanical variables that were not measured in this study (i.e., plantar pressure distribution, muscle activity, ankle and knee joint moments), known to be influenced by perceived comfort, Table 3 Changes in rating of perceived exertion (RPE) and comfort parameters for shoe only (CON), shoe with Ethyl- Vinyl Acetate orthotic (EVA), and shoe with Thermoplastic Poly-Urethane orthotic (TPU) conditions before (Fresh) and after (Fatigued) the completion of a repeated-sprint treadmill exercise Bold values indicate statistically significant ANOVA P values (P < 0.05) RPE was assessed using a 6–20 Borg scale and other comfort parameters were measures using a Visual Analog Scale (0–150 mm); Values are mean ± SD. Values are mean ± SD. C and F, respectively, refer to ANOVA main effects of condition and fatigue and interaction between these two factors with P value and partial eta-squared (η2) in parentheses. Bold values indicate statistically significant findings # Significantly different from EVA, P < 0.05 † Significantly different from TPU, P < 0.05 *Significantly different from Fresh, P < 0.05 Variables CON EVA TPU ANOVA P value (η2) C F C × F RPE  Fresh 12.7 ± 3.1 13.2 ± 3.1 13.0 ± 3.0 0.256 < 0.001 0.738  Fatigued 14.7 ± 4.2* 15.1 ± 3.1* 13.7 ± 3.1* (0.08) (0.71) (0.01) Overall comfort  Fresh 86 ± 32 93 ± 31 97 ± 25 0.089 0.163 0.121  Fatigued 81 ± 31 105 ± 21 101 ± 23 (0.15) (0.12) (0.14) Heel cushioning  Fresh 83 ± 33 96 ± 23 89 ± 25 0.090 0.021 0.813  Fatigued 85 ± 27* 102 ± 22* 93 ± 26* (0.15) (0.29) (0.01) Forefoot cushioning  Fresh 88 ± 34 96 ± 28 102 ± 23 0.079 0.326 0.086  Fatigued 84 ± 29 105 ± 24 104 ± 22 (0.16) (0.06) (0.16) Medio-lateral control  Fresh 83 ± 32#† 98 ± 26 101 ± 25 0.027 0.080 0.361  Fatigued 84 ± 30#† 107 ± 22 100 ± 24 (0.20) (0.18) (0.06) Arch height  Fresh 74 ± 36#† 92 ± 33 96 ± 24 0.009 0.346 0.060  Fatigued 72 ± 32#† 102 ± 28 95 ± 26 (0.30) (0.06) (0.18) Heel cup fit  Fresh 86 ± 29 95 ± 23 88 ± 29 0.135 0.206 0.416  Fatigued 85 ± 29 102 ± 22 93 ± 26 (0.12) (0.10) (0.05) 1186 European Journal of Applied Physiology (2022) 122:1179–1187 1 3 could explain observed differences between footwear con- ditions (Dinato et al. 2015). While ground reaction forces are commonly (similar to current approach) used as proxy measurements to reflect biomechanical loads imposed on the lower extremities as a whole, directly quantifying tissue and/ or structure-specific strain (i.e., longitudinal arch, Achilles’ tendon) remains a challenge (Verheul et al. 2020). Unexpectedly, perceived comfort ratings in different regions of the foot were in fact improved after completion of the repeated-sprint treadmill exercise. While the condi- tion × fatigue interaction was not significant, there was a trend for the two CFOs conditions to become more com- fortable in the fatigued state compared to CON. Contrast- ingly, during a 13 km run, decrement in perceived overall footwear comfort became significant only after 44 min of exercise (~ 7.8 km) (Hintzy et al. 2015). Discrepant findings between our results (i.e., assessment before and after a short and intense fatigue protocol) and previous studies (i.e., regu- lar assessments during prolonged running at lower inten- sity; Hintzy et al. 2015; Jimenez-Perez et al. 2021) could be explained by the methodology used for measuring biome- chanical manifestation of fatigue and the nature/degree of fatigue attained by participants. The clinical implication of our findings is that well-trained runners wearing CFOs made of either EVA or TPU materials should not fear deteriorated comfort ratings with acute intense fatigue. Limitations Several limitations must be considered. First, the inclusion of only male runners who were mainly habitual rear-foot strikers (~ 70%). Our findings may not be generalizable to runners with habitual midfoot/forefoot strike patterns and female population since biomechanical variables may dif- fer across various foot strikes and between genders (Moore 2016). Whereas the foot strike pattern of tested athletes was determined, our sample size of 18 participants (with only one and four forefoot and midfoot strikers) was too small to allow meaningful comparisons between groups. Addi- tionally, to reflect ecological situations, participants should undertake over-ground runs with their foot strikes recorded by a number of force plates laid in series. Because inher- ent characteristics of running shoes per se can alter running mechanics and/or RE (Hoogkamer et al. 2018), participants were not allowed to use their own running shoes. Standardi- zation of footwear conditions across participants (also with the use of personalized inserts) is a strength of our study from a methodological standpoint. Nonetheless, one could speculate that any protective effect of fatigue may have more apparent if individuals were wearing their habitual footwear or other types of foot orthoses. Conclusion Acute intense fatigue does not modify the effect of custom foot orthoses with different resilience characteristics (EVA or TPU materials both compared to standardized footwear) on running mechanics, running economy and perceived comfort. When facing acute intense fatigue, well-trained runners should not expect any protective effects from wear- ing CFOs. Acknowledgements The authors also thank Pr. Jean-Benoit Morin from the Université of Lyon, Saint Étienne, France, for his comment on our draft and help in providing the running mechanics data process- ing custom software. Author contributions OG and KVA conceived and designed research. OG and KVA conducted experiments. All authors analyzed data and interpreted results of experiments. OG and KVA drafted manuscript and prepared tables. All authors edited and revised manuscript. All authors approved final version of manuscript. Funding Open Access funding provided by the Qatar National Library. Data were collected using an instrumented treadmill funded by a QNRF grant (NPRP 4–760-3–217). Declarations Conflict of interest No potential conflict of interest was reported by the authors. Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. References Barnes KR, Kilding AE (2015) Running economy: measurement, norms, and determining factors. Sports Med-Open 1:8 Belli A, Bui P, Berger A, Geyssant A, Lacour JR (2001) A treadmill ergometer for three-dimensional ground reaction forces measure- ment during walking. J Biomech 34(1):105–112 Brocherie F, Millet GP, Morin J-B, Girard O (2016) Mechanical altera- tions to repeated treadmill sprints in normobaric hypoxia. Med Sci Sports Exerc 48(8):1570–1579 Burke JR, Papuga MO (2012) Effects of foot orthotics on running economy: methodological considerations. J Manipulative Physiol Ther 35(4):327–336 Cavagna GA (1975) Force platforms as ergometers. J Appl Physiol 39(1):174–179 1187 European Journal of Applied Physiology (2022) 122:1179–1187 1 3 Crago D, Bishop C, Arnold JB (2019) The effect of foot orthoses and insoles on running economy and performance in distance runners: a systematic review and meta-analysis. J Sports Sci 37(22):2613–2624 Davis JA (1985) Anaerobic threshold: review of the concept and direc- tions for future research. Med Sci Sports Exerc 17(1):6–21 Day E, Hahn M (2019) Optimal footwear longitudinal bending stiff- ness to improve running economy is speed dependent. Footwear Science 12(1):3–13 Flores N, Delattre N, Berton E, Rao G (2019) Does an increase in energy return and/or longitudinal bending stiffness shoe fea- tures reduce the energetic cost of running? Eur J Appl Physiol 119(2):429–439 Frederick EC (1984) Physiological and ergonomics factors in running shoe design. Appl Ergon 15:281–287 Girard O, Brocherie F, Tomazin K, Farooq A, Morin J-B (2016) Changes in running mechanics over 100-m, 200-m and 400-m treadmill sprints. J Biomech 9(4):1490–1497 Girard O, Brocherie F, Morin J-B, Millet GP (2017) Mechanical altera- tions during interval-training treadmill runs in high-level male team-sport players. J Sci Med Sport 20(1):87–91 Girard O, Morin J-B, Ryu JH, Van Alsenoy K (2020) Custom foot orthoses improve performance, but do not modify the biomechani- cal manifestation of fatigue, during repeated treadmill sprints. Eur J Appl Physiol 120(9):2037–2045 Hennig EM, Valiant GA, Liu Q (1996) Biomechanical variables and the perception of cushioning for running in various types of footwear. J Appl Biomech 12(2):143–150 Hintzy F, Cavagna J, Horvais N (2015) Evolution of perceived footwear comfort over a prolonged running session. Foot 25(4):220–223 Hoogkamer W, Kipp S, Frank JH, Farina EM, Luo G, Kram R (2018) A comparison of the energetic cost of running in marathon racing shoes. Sports Med 48:1009–1019 Jimenez-Perez I, Priego-Quesada JI, Camacho-García A, Ortiz C, de Anda RM, Pérez-Soriano P (2021) Impact accelerations during a prolonged run using a microwavable self-customised foot orthosis. Sports Biomech. https:// doi. org/ 10. 1080/ 14763 141. 2021. 19025 53 Kelly LA, Girard O, Racinais S (2011) Effect of orthoses on changes in neuromuscular control and aerobic cost of a 1-h run. Med Sci Sports Exerc 43(12):2335–2343 Lewinson RT, Worobets JT, Stefanyshyn DJ (2013) Knee abduction angular impulses during prolonged running with wedged insoles. Proc Inst Mech Eng [h] 227(7):811–814 Li SN, Hobbins L, Morin J-B, Ryu JH, Gaoua N, Hunter S, Girard O (2020) Running mechanics adjustments to perceptually-reg- ulated interval runs in hypoxia and normoxia. J Sci Med Sport 23(11):1111–1116 Lindorfer J, Kroll J, Schwameder H (2020) Does enhanced footwear comfort affect oxygen consumption and running biomechanics? Eur J Sport Sci 20(4):468–476 Lucas-Cuevas AG, Pérez-Soriano P, Llana-Belloch S, Macián-Romero C, Sánchez-Zuriaga D (2014) Effect of custom-made and prefab- ricated insoles on plantar loading parameters during running with and without fatigue. J Sports Sci 32(18):1712–1721 Luo SP, Worobets J, Nigg B, Stefanyshyn D (2009) Improved footwear comfort reduces oxygen consumption during running. Footwear Sci 1(1):25–29 McMahon TA, Cheng GC (1990) The mechanics of running: how does stiffness couple with speed? J Biomech 23(1):65–78 McPoil TG, Vicenzino B, Cornwall MW (2011) Effect of foot orthoses contour on pain perception in individuals with patellofemoral pain. J Am Podiatr Med Assoc 101(1):7–16 Milani TL, Hennig EM, Lafortune MA (1997) Perceptual and biome- chanical variables for running in identical shoe constructions with varying midsole hardness. Clin Biomech 12(5):294–300 ((bristol, Avon)) Moore IS (2016) Is there an economical running technique? a review of modifiable biomechanical factors affecting running economy. Sports Med 46(6):793–807 Moore IS, Jones A, Dixon S (2014) The pursuit of improved running performance: can changes in cushioning and somatosensory feed- back influence running economy and injury risk? Footwear Sci 6(1):1–11 Morin J-B, Dalleau G, Kyröläinen H, Jeannin T, Belli A (2005) A simple method for measuring stiffness during running. J Appl Biomech 21(2):167–180 Morin JB, Samozino P, Bonnefoy R, Edouard P, Belli A (2010) Direct measurement of power during one single sprint on treadmill. J Biomech 43(10):1970–1975 Mundermann A, Nigg BM, Stefanyshyn DJ, Humble RN (2002) Devel- opment of a reliable method to assess footwear comfort during running. Gait Posture 16(1):38–45 Mundermann A, Nigg BM, Humble RN, Stefanyshyn DJ (2003) Orthotic comfort is related to kinematics, kinetics, and EMG in recreational runners. Med Sci Sports Exerc 35(10):1710–1719 Robbins SE, Hanna AM (1987) Running-related injury preven- tion through barefoot adaptations. Med Sci Sports Exerc 19(2):148–156 Sinclair J, McGrath R, Brook O, Taylor PJ, Dillon S (2016) Influence of footwear designed to boost energy return on running econ- omy in comparison to a conventional running shoe. J Sports Sci 34(11):1094–1098 Tung KD, Franz JR, Kram R (2014) A test of the metabolic cost of cushioning hypothesis during unshod and shod running. Med Sci Sports Exerc 46(2):324–329 Van Alsenoy K, Ryu JH, Girard O (2019) The effect of EVA and TPU custom foot orthoses on running economy, running mechanics, and comfort. Front Sports Act Living 1:34 Verheul J, Nedergaard NJ, Vanrenterghem J, Robinson MA (2020) Measuring biomechanical loads in team sports–from lab to field. Sci Med Footb 4(3):246–252 Wilkinson M, Ewen A, Captan N, O’Learly D, Smith N, Stoneham R, Saxby L (2018) Textured insoles reduce vertical loading rate and increase subjective plantar sensation in overground running. Eur J Sport Sci 18(4):497–503 Worobets J, Wannop JW, Tomaras E, Stefanyshyn D (2014) Softer and more resilient running shoe cushioning properties enhance running economy. Footwear Sci 6(3):147–153 Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Acute intense fatigue does not modify the effect of EVA and TPU custom foot orthoses on running mechanics, running economy and perceived comfort.

02-24-2022

Van Alsenoy, Ken,Ryu, Joong Hyun,Girard, Olivier

eng

PMC8085187

1 Vol.:(0123456789) Scientific Reports | (2021) 11:9260 | https://doi.org/10.1038/s41598-021-88883-4 www.nature.com/scientificreports Track distance runners exhibit bilateral differences in the plantar fascia stiffness Hiroto Shiotani1,2, Ryo Yamashita3, Tomohiro Mizokuchi3, Natsuki Sado4, Munekazu Naito2,5 & Yasuo Kawakami1,2* Human steady-state locomotion modes are symmetrical, leading to symmetric mechanical function of human feet in general; however, track distance running in a counterclockwise direction exposes the runner’s feet to asymmetrical stress. This may induce asymmetrical adaptation in the runners’ foot arch functions, but this has not been experimentally tested. Here, we show that the plantar fascia (PF), a primary structure of the foot arch elasticity, is stiffer for the left than the right foot as a characteristic of runners, via a cross-sectional study on 10 track distance runners and 10 untrained individuals. Shear wave velocity (index of tissue stiffness: SWV) and thickness of PF and foot dimensions were compared between sides and groups. Runners showed higher PF SWV in their left (9.4 ± 1.0 m/s) than right (8.9 ± 0.9 m/s) feet, whereas untrained individuals showed no bilateral differences (8.5 ± 1.5 m/s and 8.6 ± 1.7 m/s, respectively). Additionally, runners showed higher left to right (L/R) ratio of PF SWV than untrained men (105.1% and 97.7%, respectively). PF thickness and foot dimensions were not significantly different between sides or groups. These results demonstrate stiffer PF in the left feet of runners, which may reflect adaptation to their running-specific training that involves asymmetrical mechanical loading. During human locomotion, the medial longitudinal arch of the foot is lowered while being stretched out in response to weight-bearing, and then recoils as the load is removed. Such a spring-like property of the foot arch helps to attenuate impact forces and store/release elastic strain energy leading to energy saving in running1,2. Previous studies indicate that the foot arch elasticity is attributed to the plantar fascia (PF)1,3,4. PF behaves vis- coelastically under load5,6, and its resistive tension helps to prevent the lengthening and lowering of the foot arch. During each foot contact of running, PF is repetitively loaded with the tension reaching as high as 0.6–3.7 times bodyweight with its longitudinal strain up to 6%7–10. Such sizable stress concentrates around the proximal site of PF11–13, which may be associated with the heterogeneity of mechanical and morphological properties (e.g., stiffness and thickness) of PF14–16 as well as the occurrence of plantar fasciitis17,18. The localized stiffness of PF can be quantitatively assessed as the shear wave velocity (SWV) in vivo14–16. PF has higher SWV (i.e., stiffer) at the proximal site than middle and distal sites14,16. Additionally, long-distance running induced a transient decrease of SWV at the proximal site of PF while long-distance runners showing smaller changes in SWV than untrained individuals15, suggesting that runners had built up a more resilient PF. These findings are evidence of PF adaptability to site-specific and chronic mechanical stress, which can be reflected in its stiffness and morphology. Human steady-state locomotion modes are symmetrical, leading to symmetric mechanical function of human feet in general; however, track distance running is performed always in a counterclockwise direction, i.e., the left leg being inside during curve running. In this phase, runners are required to generate greater forces with their left legs19,20 to exert centripetal force21. This is associated with the greater load on the left foot, resulting in the lowering of the left foot arch, and thus leading to an increase of mechanical stress to PF. Therefore, runners’ feet can be exposed to asymmetrical stress during running. This may induce asymmetrical adaptation in runners’ PF stiffness and morphology. Although PF SWV and thickness, and the foot dimensions were comparable between left and right sides in a healthy and untrained population14, this may not be true for track distance runners. If OPEN 1Faculty of Sport Sciences, Waseda University, Saitama, Japan. 2Human Performance Laboratory, Comprehensive Research Organization, Waseda University, Tokyo, Japan. 3School of Sport Sciences, Waseda University, Saitama, Japan. 4Faculty of Health and Sport Sciences, University of Tsukuba, Ibaraki, Japan. 5Department of Anatomy, Aichi Medical University, Aichi, Japan. *email: ykawa@waseda.jp 2 Vol:.(1234567890) Scientific Reports | (2021) 11:9260 | https://doi.org/10.1038/s41598-021-88883-4 www.nature.com/scientificreports/ the bilateral differences in runner’s feet can be confirmed, this provides an indication of a threshold of mechani- cal stress that causes adaptation of PF and foot arch functions. A profound understanding of PF adaptability is essential for improvements in their performance as well as prevention of plantar fasciitis. Therefore, the purpose of this study was to investigate the bilateral differences in mechanical and morpho- logical properties of PF and foot dimensions in track distance runners, as contrasted to untrained individuals. We hypothesized that track distance runners have bilateral differences in PF SWV, thickness, and the foot arch height, and that runners show sizable differences in the left to right (L/R) ratios of measured variables as com- pared to untrained individuals. Results In runners, SWV at the proximal site was significantly higher in left (9.4 ± 1.0 m/s) than right foot (8.9 ± 0.9 m/s) (p = 0.021, d = 0.813), but not at the middle (p = 0.782, d = 0.073) or distal sites (p = 0.554, d = 0.138) (Fig. 1). Even in a lefty runner (n = 1), PF SWV at the proximal site was higher for his left (10.0 m/s) than right foot (9.1 m/s). PF SWV at the proximal site was also higher for the left than the right feet both in rearfoot strike (n = 7, left: 9.0 ± 1.0 m/s and right: 8.6 ± 0.8 m/s, respectively) and forefoot strike runners (n = 3, left: 10.2 ± 0.3 m/s and right: 9.6 ± 0.5 m/s, respectively). In untrained men, SWV at each measurement site was not significantly different between left and right feet (p ≥ 0.222, d ≤ 0.264). PF thickness at each measurement site was not significantly dif- ferent between left and right feet in runners (p ≥ 0.327, d ≤ 0.141) or untrained men (p ≥ 0.411, d ≤ 0.305) (Fig. 1). Foot dimensions were not significantly different between left and right feet in either of runners or untrained men (Table 1). The L/R ratio of SWV at the proximal site was significantly higher in runners than untrained men (p = 0.027, d = 1.076), but not at the middle (p = 0.815, d = 0.107) or distal sites (p = 0.421, d = 0.369). The L/R ratios of thick- ness at any of the measurement sites or foot dimensions were not significantly different between groups (Table 2). Age, body height, body mass, BMI, and fractions of leg dominance and foot strike pattern were not signifi- cantly different between runners and untrained men (Table 3). All participants were healthy and free from injury of the lower extremity in the past 12 months and had no present or past history of plantar fasciitis. The runners had kept habitual running of at least 10 km/week for the past year, mainly on a running track, and their running experiences ranged between 9 and 16 years. Their personal best time of 5000 m ranged from 14′ 15 to 15′ 30. The untrained participants were either sedentary or lightly active, and none of them had been involved in any structured training program or continuous sports participation at least 12 months before the measurements. All participants used conventional running shoes rather than minimalist, high cushion, or high motion control shoes. Figure 1. Bilateral differences in SWV and thickness of runners and untrained men. Data are shown as means ± SD. 3 Vol.:(0123456789) Scientific Reports | (2021) 11:9260 | https://doi.org/10.1038/s41598-021-88883-4 www.nature.com/scientificreports/ Table 1. Bilateral differences in foot dimensions of runners and untrained men. Data are shown as means ± SD. Runners (n = 10) Untrained men (n = 10) Left Right p value Cohen’s d Left Right p value Cohen’s d Foot length (mm) 245.3 ± 7.3 245.9 ± 8.6 p = 0.520 247.4 ± 8.7 248.2 ± 8.1 p = 0.445 d = 0.075 d = 0.107 Dorsal height (mm) 61.1 ± 4.4 60.8 ± 4.2 p = 0.760 60.5 ± 5.5 61.1 ± 4.0 p = 0.671 d = 0.070 d = 0.125 Navicular height (mm) 43.3 ± 6.2 41.9 ± 6.8 p = 0.080 41.3 ± 4.8 40.9 ± 5.4 p = 0.507 d = 0.215 d = 0.078 Arch height ratio (%) 17.7 ± 2.7 17.1 ± 3.0 p = 0.064 16.7 ± 1.8 16.4 ± 1.9 p = 0.400 d = 0.210 d = 0.162 Table 2. Left/right ratios (%) of individual parameters in runners and untrained men. Data are shown as means ± SD. Bold fonts indicate significant difference between runners and untrained men (p < 0.05) with a “large” effect size (d ≥ 0.8). Variable Runners Untrained men p value Cohen’s d Shear wave velocity Proximal 105.1 ± 6.2 97.7 ± 7.6 p = 0.027 d = 1.076 Middle 101.8 ± 12.9 100.2 ± 17.2 p = 0.815 d = 0.107 Distal 104.6 ± 13.7 99.7 ± 12.8 p = 0.421 d = 0.360 Thickness Proximal 101.3 ± 6.4 99.6 ± 8.0 p = 0.610 d = 0.232 Middle 97.8 ± 5.9 98.8 ± 8.7 p = 0.756 d = 0.141 Distal 102.3 ± 14.1 97.6 ± 11.7 p = 0.429 d = 0.362 Foot length 99.8 ± 1.1 99.7 ± 1.3 p = 0.843 d = 0.083 Dorsal height 100.5 ± 3.9 99.1 ± 6.3 p = 0.583 d = 0.267 Navicular height 103.9 ± 5.8 101.2 ± 4.4 p = 0.270 d = 0.525 Arch height ratio 104.1 ± 6.0 101.6 ± 5.1 p = 0.329 d = 0.449 Table 3. Physical characteristics of participants. Data are shown as means ± SD. BMI body mass index, RFS rear foot strikers, FFS forefoot strikers. Age, body height, body mass, BMI, and fractions of leg dominance and foot strike pattern were not significantly different between runners and untrained men. Variable Runners Untrained men p value n 10 10 – Age (years) 22.0 ± 0.7 22.5 ± 1.4 0.309 Body height (m) 1.68 ± 0.04 1.70 ± 0.05 0.392 Body mass (kg) 55.5 ± 4.2 58.4 ± 5.6 0.062 BMI (kg/m2) 19.6 ± 1.2 20.3 ± 1.7 0.113 Dominant leg (Lefty:Righty) 1:9 1:9 1.000 Foot strike pattern (RFS:FFS) 7:3 10:0 0.060 Running experience (years) 11.0 ± 2.2 – – Running distance (km/week) 43.7 ± 35.4 – – 4 Vol:.(1234567890) Scientific Reports | (2021) 11:9260 | https://doi.org/10.1038/s41598-021-88883-4 www.nature.com/scientificreports/ Discussion To the best of our knowledge, this is the first study to investigate bilateral differences in the mechanical and morphological properties of PF in track distance runners. The most striking finding of the present study was that track distance runners showed stiffer PF at the proximal site in their left than the right feet, unlike the untrained participants. A number of previous studies addressing running mechanics22,23 and mechanical and morphological properties of the musculotendinous and fascial tissues15,24 in runners have focused on unilateral leg by assuming bilateral symmetry. Our findings however suggest that asymmetry in runners is the major issue that needs to be carefully considered. The greater load on the left foot during curve running19,20 can induce an increase of mechanical stress on PF. We previously revealed that running causes a decrease in PF SWV at its proximal site15, and this coincides with the simulation of stress distribution along PF11–13. The bilateral differences in PF stiffness at the proximal site in runners may reflect the adaptation to such stress accumulation in this region of the left foot during track running, regardless of the lateral dominance or foot strike patterns. The fact that PF stiffness in the right feet of runners was comparable to that of both feet in untrained individuals suggests a threshold of mechanical stress that causes adaptation of PF stiffness, which is side-specific for track runners. The finding that PF can be stiffened in response to a sufficient load is valuable for the general population toward injury prevention and rehabilitation as well as improvements in human locomotor performance. Since the runners in this study had no history of plantar fasciitis, they might have been successful examples who had been optimally adapted to their running-specific training. However, plantar fasciitis is one of the most common injuries in long-distance runners, regardless of their performance levels25,26. This injury frequently occurs around the proximal site of PF17,18 where the mechanical stress is concentrated11–13. Thus, there is a clinical implication that the left foot of track runners can suffer from a higher incidence of this injury if their training adaptation does not work well. The bilateral imbalances in strength, morphology, and running mechanics are considered to be risk factors for injury of runners27,28. Interactions between these factors for the occurrence of plantar fasciitis are worth examining in future studies. No bilateral differences in PF thickness of runners and untrained individuals are consistent with previous findings that PF thickness was not different between recreational runners and untrained individuals15, and that PF thickness was not influenced by physical activity29. These results, together with our findings, discard the pos- sibility of PF adaptability in terms of its thickness for reducing mechanical stress induced by distance running. No bilateral difference in the foot arch dimensions suggests that the arches of both feet of runners can fulfill their imposed roles through different mechanical properties with comparable morphology. We could not obtain the running mechanics and the foot arch deformation during running. This is one of the limitations of the present study. As our findings suggest a threshold of mechanical stress that causes adaptation of PF stiffness, quantifying the mechanical stress applied to bilateral feet during track running will lead to a better understanding of the nature of PF adaptability. Additionally, the runners who participated in the present study can be categorized as recreational level30. Runners of different performance levels (e.g., competitive and elite runners) show different running mechanics31,32 and fatigue responses33,34. Thus, competitive and elite runners have the possibility to exhibit different signs of adaptation in PF and foot morphology. Comparisons between runners in different performance levels should be incorporated in these future studies. Moreover, there is a vari- ation in the training volume of runners (Table 3). As we previously reported that long-distance running induced transient decreases of PF SWV15, the training volume can be a potential factor that affects PF properties and their adaptation. In addition, the material and mechanical properties of the running surface (e.g., rubber, asphalt, or grass) as well as shoe sole have the possibility to affect the magnitude of stress to the foot35–38. Future studies addressing the chronic effects of training volume and environment on PF adaptation as well as foot arch functions are needed. Lastly, it can be assumed that sprinters, participating in the event of 200 and 400 m in particular, and long/high jumpers may also apply asymmetrical stress to their feet with a greater magnitude of stress compared to distance runners. Further investigation of bilateral differences in PF characteristics and foot dimensions in other events and sports athletes can be an option of the future theme in understanding PF adaptability. In conclusion, this study showed bilateral differences in the mechanical but not in the morphological proper- ties of PF and foot arch dimensions in track distance runners as compared to untrained individuals. PF SWV at the proximal site was higher in the left feet of track distance runners while their right feet showing comparable values to that of untrained individuals. These results demonstrate stiffer proximal PF in the left feet of runners, which may reflect adaptation to their running-specific training that involves asymmetrical mechanical loading. Methods Study design and participants. A cross-sectional study was conducted at Waseda University (Tokoro- zawa campus) in Japan from August to November 2017. This study was approved by the Human Research Ethics Committee of Waseda University (reference number: 2016-310) and was carried out in accordance with the Declaration of Helsinki. Written informed consent was obtained from all participants before data collection. The necessary sample size was calculated from our preliminary results (n = 6 in each group; total = 12). A priori power analysis (G*Power v3.1, Heinrich Heine-Universität Dusseldorf, Germany) with an assumed type 1 error of 0.05 and a statistical power of 0.80 was conducted to find significant differences in PF SWV between left and right feet of runners and between groups, respectively. The critical sample sizes were estimated to be at least 7 runners and 9 in each group (total = 18), respectively. Thus, 10 track distance male runners and 10 untrained men were recruited in this study (Table 1). Twelve runners were eligible for participation in this study. Of these, 2 runners met the exclusion criteria of history of plantar fasciitis and operative treatment of the lower limb (Fig. 2). Finally, 10 runners and 10 untrained men who matched the baseline physical characteristics with those of runners were successfully recruited in this study. 5 Vol.:(0123456789) Scientific Reports | (2021) 11:9260 | https://doi.org/10.1038/s41598-021-88883-4 www.nature.com/scientificreports/ Before the main measurements, the profiles including age, body height, body mass, dominant leg, athletic experiences, exercise habits for the past year, foot strike pattern, history of injuries and operative treatment, and model of their running shoes were collected from all participants. The dominant leg was determined according to the participant’s favorite leg for kicking a ball. The foot strike pattern (rearfoot or forefoot strikers) of partici- pants was visually confirmed on another occasion15. Additionally, the training environment (e.g., affiliation and running surfaces), personal best time of 5000 m, and running volume/week for the year were asked for runners. To avoid any confounding factors, we recruited participants attending the same university, and attempted to match the baseline physical characteristics of untrained participants with those of runners. Participants were not allowed to perform any strenuous exercises for at least 24 h before the measurement. Ultrasound measurements. The supersonic shear imaging (SSI) and B-mode ultrasonography techniques with an Aixplorer ultrasound scanner (version 6.4, Supersonic Imagine, Aix-en-Provence, France) and a linear array probe (SL 15-4, Supersonic Imagine, Aix-en-Provence, France) were used to measure the mechanical and morphological properties of PF. SSI is a valid and reliable technique to evaluate the stiffness of skeletal muscles, tendons, and fasciae in vivo14,39–41. In principle, SSI uses multiple push pulses to generate the shear waves propa- gating within the soft tissues and measures their velocity (i.e., SWV). Since SWV is related to Young’s modulus and shear modulus of the soft tissues, it can be used as an index of stiffness42,43. Details of SSI measurement and data processing were based on our previous published work14,15. During ultrasound measurements, participants were requested to rest in a supine position on the examination bed with their knee fully extended. Additionally, their ankle and toe digits were secured to a custom-made fixture at the neutral position. PF was scanned at three different sites along the longitudinal line between the medial calcaneal tubercle and the second toe. The locations of measurement sites were that at the proximal (in the proximity to the calcaneus), middle (the level of navicular tuberosity), and distal (proximity to the second metatarsal head) (Fig. 3). The longitudinal line of the foot and the locations of the transducer were marked on the skin surface using a waterproof marker. The scanning head of the probe was coated with transmission gel. An acoustic standoff pad (Gelpad for StatUS, Enraf–Nonius, Rotterdam, Netherland) was used to avoid applying excessive compres- sion on the skin surface. Three images were obtained at each measurement site, and used for further analysis. After data collection, SWV at each measurement site was measured as the mean value within the region of interest (ROI) which was manually traced over the fascial boundaries of PF using a measurement tool included in the Aixplorer software (i.e., Q-box Trace). PF thickness at each measurement site was measured the distance between the superficial and deep fascial boundaries was measured to determine thickness using a measurement Figure 2. Flow diagram depicting participant selection. 12 runners were eligible for participation in this study. Two runners met the exclusion criteria of history of plantar fasciitis and operative treatment of the lower limb. Thus, 10 runners and 10 untrained men who matched the baseline physical characteristics with those of runners were included in this study. 6 Vol:.(1234567890) Scientific Reports | (2021) 11:9260 | https://doi.org/10.1038/s41598-021-88883-4 www.nature.com/scientificreports/ tool (i.e., Distance). For SWV and thickness at each measurement site, three images were analyzed at each meas- urement site, then the three values were averaged to obtain the representative value. Measurements of the foot dimensions. A foot scanner (JMS-2100CU, Dream GP, Osaka, Japan) was used to obtain three-dimensional foot shape data. Details of measurement and data processing were based on our previous study using the same system15. Participants were requested to stand in a relaxed position with their feet approximately shoulder-width apart. The longitudinal axis of their feet, which is the line connecting between the most posterior point of the heel and the head of the second toe, aligned parallel with the guidelines drawn on the footplate in the foot scanner. A laser scanner moved around the foot in an oval trajectory, measuring the foot dimensions and the anatomical marker positions based on laser line triangulation. After the scanning, foot length, dorsal height, and navicular height were measured. The foot length was defined as the length projected on the longitudinal axis between the most posterior point of the heel and the head of the first or second toe, whichever was longer. The dorsal height was defined as the height of the highest point from the floor at 55% of the length of the foot from the heel. The navicular height was defined as the height of the most medial point of the navicular bone from the floor. Additionally, the arch height ratio was calculated as the navicular height normalized to the foot length. Statistical analysis. The normality of the data was assessed using a Shapiro–Wilk test. After the normal- ity was confirmed, the difference in physical characteristics between groups were compared using an unpaired t-test. The fraction of dominant legs within each group was compared with a Pearson chi-squared test. Com- parisons of measured variables between left and right feet in each group were performed using a paired t-test. The L/R ratios were calculated for the measured variables, and were compared using an unpaired t-test between groups. Cohen’s d was calculated as a measure of effect size. For the within-subject factor, it was corrected for dependence between mean values using the following equation: d = Mdiff/SDpooled √2(1 − r) , where Mdiff is mean difference between conditions, SDpooled is pooled SD, and r is correlation between mean values44. Effect size is interpreted as trivial (d < 0.2), small (0.2 ≤ d < 0.5), medium (0.5 ≤ d < 0.8) and large effect (d ≥ 0.8)45. Sta- tistical significance was set at α = 0.05. Statistical analysis was performed using SPSS software (SPSS Statistics 25, IBM, Armonk, USA). Received: 17 November 2020; Accepted: 19 April 2021 References 1. Ker, R. F., Bennett, M. B., Bibby, S. R., Kester, R. C. & Alexander, R. M. The spring in the arch of the human foot. Nature 325, 147–149 (1987). 2. Stearne, S. M. et al. The foot’s arch and the energetics of human locomotion. Sci. Rep. 6, 19403 (2016). 3. Cifuentes-De la Portilla, C., Larrainzar-Garijo, R. & Bayod, J. Analysis of the main passive soft tissues associated with adult acquired flatfoot deformity development: A computational modeling approach. J. Biomech. 84, 183–190 (2019). 4. Huang, C. K., Kitaoka, H. B., An, K. N. & Chao, E. Y. Biomechanical evaluation of longitudinal arch stability. Foot Ankle 14, 353–357 (1993). 5. Pavan, P. G., Stecco, C., Darwish, S., Natali, A. N. & De Caro, R. Investigation of the mechanical properties of the plantar aponeu- rosis. Surg. Radiol. Anat. 33, 905–911 (2011). 6. Guo, J., Liu, X., Ding, X., Wang, L. & Fan, Y. Biomechanical and mechanical behavior of the plantar fascia in macro and micro structures. J. Biomech. 76, 160–166 (2018). 7. Chen, T. L., Wong, D. W., Wang, Y., Lin, J. & Zhang, M. Foot arch deformation and plantar fascia loading during running with rearfoot strike and forefoot strike: A dynamic finite element analysis. J. Biomech. 83, 260–272 (2019). 8. Giddings, V. L., Beaupr, G. S., Whalen, R. T. & Carter, D. R. Calcaneal loading during walking and running. Med. Sci. Sports Exerc. 32, 627–634 (2000). Figure 3. Representative shear wave and ultrasound B-mode images of the plantar fascia at the proximal, middle, and distal sites in a runner and an untrained participant. The region of interest (ROI) is bounded in red. 7 Vol.:(0123456789) Scientific Reports | (2021) 11:9260 | https://doi.org/10.1038/s41598-021-88883-4 www.nature.com/scientificreports/ 9. McDonald, K. A. et al. The role of arch compression and metatarsophalangeal joint dynamics in modulating plantar fascia strain in running. PLoS ONE 11, e0152602 (2016). 10. Wager, J. C. & Challis, J. H. Elastic energy within the human plantar aponeurosis contributes to arch shortening during the push- off phase of running. J. Biomech. 49, 704–709 (2016). 11. Chen, Y. N., Chang, C. W., Li, C. T., Chang, C. H. & Lin, C. F. Finite element analysis of plantar fascia during walking: A quasi-static simulation. Foot Ankle Int. 36, 90–97 (2015). 12. Cheng, H. Y., Lin, C. L., Wang, H. W. & Chou, S. W. Finite element analysis of plantar fascia under stretch-the relative contribution of windlass mechanism and achilles tendon force. J. Biomech. 41, 1937–1944 (2008). 13. Lin, S. C. et al. Stress distribution within the plantar aponeurosis during walking—A dynamic finite element analysis. J. Mech. Med. Biol. 14, 145005 (2014). 14. Shiotani, H., Yamashita, R., Mizokuchi, T., Naito, M. & Kawakami, Y. Site- and sex-differences in morphological and mechanical properties of the plantar fascia: A supersonic shear imaging study. J. Biomech. 85, 198–203 (2019). 15. Shiotani, H., Mizokuchi, T., Yamashita, R., Naito, M. & Kawakami, Y. Acute effects of long-distance running on mechanical and morphological properties of the human plantar fascia. Scand. J. Med. Sci. Sports 30, 1360–1368 (2020). 16. Shiotani, H., Maruyama, N., Kurumisawa, K., Yamagishi, T. & Kawakami, Y. Human plantar fascial dimensions and shear wave velocity change in vivo as a function of ankle and metatarsophalangeal joint positions. J. Appl. Physiol. 130, 390–399 (2021). 17. Wearing, S. C., Smeathers, J. E., Urry, S. R., Hennig, E. M. & Hills, A. P. The pathomechanics of plantar fasciitis. Sports Med. 36, 585–611 (2006). 18. League, A. C. Current concepts review: Plantar fasciitis. Foot Ankle Int. 29, 358–366 (2008). 19. Chang, Y. H. & Kram, R. Limitations to maximum running speed on flat curves. J. Exp. Biol. 210, 971–982 (2007). 20. Churchill, S. M., Trewartha, G., Bezodis, I. N. & Salo, A. I. Force production during maximal effort bend sprinting: Theory vs reality. Scand. J. Med. Sci. Sports 26, 1171–1179 (2016). 21. Greene, P. R. Running on flat turns: Experiments, theory, and applications. J. Biomech. Eng. 107, 96–103 (1985). 22. Bazuelo-Ruiz, B., Dura-Gil, J. V., Palomares, N., Medina, E. & Llana-Belloch, S. Effect of fatigue and gender on kinematics and ground reaction forces variables in recreational runners. PeerJ 6, e4489 (2018). 23. Sado, N., Yoshioka, S. & Fukashiro, S. A biomechanical study of the relationship between running velocity and three-dimensional lumbosacral kinetics. J. Biomech. 94, 158–164 (2019). 24. Fletcher, J. R. & MacIntosh, B. R. Changes in achilles tendon stiffness and energy cost following a prolonged run in trained distance runners. PLoS ONE 13, e0202026 (2018). 25. Taunton, J. E. et al. A retrospective case-control analysis of 2002 running injuries. Br. J. Sports Med. 36, 95–101 (2002). 26. van Gent, R. N. et al. Incidence and determinants of lower extremity running injuries in long distance runners: A systematic review. Br. J. Sports Med. 41, 469–480 (2007). 27. Zifchock, R. A., Davis, I. & Hamill, J. Kinetic asymmetry in female runners with and without retrospective tibial stress fractures. J. Biomech. 39, 2792–2797 (2006). 28. Rauh, M. J., Koepsell, T. D., Rivara, F. P., Rice, S. G. & Margherita, A. J. Quadriceps angle and risk of injury among high school cross-country runners. J. Orthop. Sports Phys. Ther. 37, 725–733 (2007). 29. Uzel, M., Cetinus, E., Ekerbicer, H. C. & Karaoguz, A. The influence of athletic activity on the plantar fascia in healthy young adults. J. Clin. Ultrasound 34, 17–21 (2006). 30. Clermont, C. A., Benson, L. C., Osis, S. T., Kobsar, D. & Ferber, R. Running patterns for male and female competitive and recrea- tional runners based on accelerometer data. J. Sports Sci. 37, 204–211 (2019). 31. Anderson, T. Biomechanics and running economy. Sports Med. 22, 76–89 (1996). 32. Williams, K. R. & Cavanagh, P. R. Relationship between distance running mechanics, running economy, and performance. J. Appl. Physiol. 63, 1236–1245 (1987). 33. Willwacher, S., Sanno, M. & Bruggemann, G. P. Fatigue matters: An intense 10 km run alters frontal and transverse plane joint kinematics in competitive and recreational adult runners. Gait Posture 76, 277–283 (2019). 34. Sanno, M., Willwacher, S., Epro, G. & Bruggemann, G. P. Positive work contribution shifts from distal to proximal joints during a prolonged run. Med. Sci. Sports Exerc. 50, 2507–2517 (2018). 35. Birch, J. V., Kelly, L. A., Cresswell, A. G., Dixon, S. J. & Farris, D. J. Neuromechanical adaptations of foot function to changes in surface stiffness during hopping. J. Appl. Physiol. 130, 1196 (2021). 36. Tessutti, V., Ribeiro, A. P., Trombini-Souza, F. & Sacco, I. C. Attenuation of foot pressure during running on four different surfaces: Asphalt, concrete, rubber, and natural grass. J. Sports Sci. 30, 1545–1550 (2012). 37. Kelly, L. A., Lichtwark, G. A., Farris, D. J. & Cresswell, A. G. Shoes alter the spring-like function of the human foot during running. J. R. Soc. Interface 13, 20160174 (2016). 38. Roy, J. P. & Stefanyshyn, D. J. Shoe midsole longitudinal bending stiffness and running economy, joint energy, and emg. Med. Sci. Sports Exerc. 38, 562–569 (2006). 39. Aubry, S. et al. Viscoelasticity in achilles tendonopathy: Quantitative assessment by using real-time shear-wave elastography. Radiology 274, 821–829 (2015). 40. Lacourpaille, L., Hug, F., Bouillard, K., Hogrel, J. Y. & Nordez, A. Supersonic shear imaging provides a reliable measurement of resting muscle shear elastic modulus. Physiol. Meas. 33, N19-28 (2012). 41. Otsuka, S., Shan, X. & Kawakami, Y. Dependence of muscle and deep fascia stiffness on the contraction levels of the quadriceps: An in vivo supersonic shear-imaging study. J. Electromyogr. Kinesiol. 45, 33–40 (2019). 42. Eby, S. F. et al. Validation of shear wave elastography in skeletal muscle. J. Biomech. 46, 2381–2387 (2013). 43. Helfenstein-Didier, C. et al. In vivo quantification of the shear modulus of the human achilles tendon during passive loading using shear wave dispersion analysis. Phys. Med. Biol. 61, 2485–2496 (2016). 44. Morris, S. B. & DeShon, R. P. Combining effect size estimates in meta-analysis with repeated measures and independent-groups designs. Psychol. Methods 7, 105–125 (2002). 45. Cohen, J. A power primer. Psychol. Bull. 112, 155–159 (1992). Acknowledgements This study was part of research activities of the Human Performance Laboratory, Comprehensive Research Organization, Waseda University. This study was supported by JSPS KAKENHI Grant Numbers 19J14912 and 16H01870. The authors express their gratitude to Dr. Pavlos Evangelidis, Dr. Takaki Yamagishi and Mr. Hidetaka Hayashi for grammatical corrections of the manuscript. Author contributions H.S., R.Y., T.M. and Y.K. designed the research. H.S., R.Y. and T.M. corrected the data. H.S. analyzed the data. H.S., R.Y., T.M., N.S., M.N. and Y.K. interpreted the results of experiments. H.S. wrote the main manuscript text and prepared Figures and Tables. H.S., N.S., M.N. and Y.K. revised the manuscript. All authors read and approved the final version of the manuscript. 8 Vol:.(1234567890) Scientific Reports | (2021) 11:9260 | https://doi.org/10.1038/s41598-021-88883-4 www.nature.com/scientificreports/ Competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to Y.K. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2021

Track distance runners exhibit bilateral differences in the plantar fascia stiffness.

04-29-2021

Shiotani, Hiroto,Yamashita, Ryo,Mizokuchi, Tomohiro,Sado, Natsuki,Naito, Munekazu,Kawakami, Yasuo

eng

PMC7769488

RESEARCH ARTICLE Effects of acute wearable resistance loading on overground running lower body kinematics Karl M. TrounsonID1,2☯*, Aglaja Busch3‡, Neil French Collier1‡, Sam RobertsonID1‡ 1 Institute for Health and Sport, Victoria University, Footscray, Victoria, Australia, 2 Western Bulldogs Football Club, Footscray, Victoria, Australia, 3 University Outpatient Clinic, Sports Medicine & Sports Orthopedics, University of Potsdam, Potsdam, Germany ☯ These authors contributed equally to this work. ‡ These authors also contributed equally to this work. * karl.trounson@live.vu.edu.au Abstract Field-based sports require athletes to run sub-maximally over significant distances, often while contending with dynamic perturbations to preferred coordination patterns. The ability to adapt movement to maintain performance under such perturbations appears to be train- able through exposure to task variability, which encourages movement variability. The aim of the present study was to investigate the extent to which various wearable resistance load- ing magnitudes alter coordination and induce movement variability during running. To inves- tigate this, 14 participants (three female and 11 male) performed 10 sub-maximal velocity shuttle runs with either no weight, 1%, 3%, or 5% of body weight attached to the lower limbs. Sagittal plane lower limb joint kinematics from one complete stride cycle in each run were assessed using functional data analysis techniques, both across the participant group and within-individuals. At the group-level, decreases in ankle plantarflexion following toe-off were evident in the 3% and 5% conditions, while increased knee flexion occurred during weight acceptance in the 5% condition compared with unloaded running. At the individual- level, between-run joint angle profiles varied, with six participants exhibiting increased joint angle variability in one or more loading conditions compared with unloaded running. Loading of 5% decreased between-run ankle joint variability among two individuals, likely in accor- dance with the need to manage increased system load or the novelty of the task. In terms of joint coordination, the most considerable alterations to coordination occurred in the 5% load- ing condition at the hip-knee joint pair, however, only a minority of participants exhibited this tendency. Coaches should prescribe wearable resistance individually to perturb preferred coordination patterns and encourage movement variability without loading to the extent that movement options become limited. PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 1 / 19 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Trounson KM, Busch A, French Collier N, Robertson S (2020) Effects of acute wearable resistance loading on overground running lower body kinematics. PLoS ONE 15(12): e0244361. https://doi.org/10.1371/journal.pone.0244361 Editor: Elena Bergamini, University of Rome, ITALY Received: July 19, 2020 Accepted: December 8, 2020 Published: December 28, 2020 Copyright: © 2020 Trounson et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: The data underlying this study are available on Dryad (doi:10.5061/ dryad.1c59zw3sk). Funding: The author(s) received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. Introduction Across many field-based sports, athletes must be capable of running long distances throughout a match [1–3]. Depending on the sport, total running distance can range from an average of 6 km in rugby league to 12 km in Australian Rules football [1]. In Australian Rules football, soc- cer, rugby league, and rugby sevens, most of the distance covered during match play can be classified as “low-intensity activity”, i.e., occurring at velocities <5.4 m.s-1 [1, 3]. While high- intensity efforts are often associated with significant match events, adequate sub-maximal run- ning capabilities are also important for effective opponent tracking and retention of team for- mations during different phases of play throughout a match [4]. As such, training aimed at developing sub-maximal overground running performance is evidently worthwhile. Development of sub-maximal running performance for field-based athletes is a multifacto- rial proposition and requires training of aerobic capacity, biomechanical factors for superior economy, and muscular strength [5–8]. Coaches should address these factors in training pre- scription and, in addition, athletes’ ability to adapt their running coordination patterns in accordance with the dynamic constraints of the sport [9, 10]. The capacity to exhibit “adapt- ability” in this sense allows for greater maintenance of performance in varied contexts and is a hallmark of higher performing athletes in many sports [10–13]. In field-based sport, organis- mic constraints in the form of local metabolite accumulation from intermittent anaerobic efforts [14, 15], muscle damage arising from high force eccentric contractions during decelera- tions [16], and muscular contusion from compressive force impacts [17], all present scenarios in which there is a challenge to an athlete’s preferred running coordinative structure, which must be adapted to. Critically, the implementation of a training intervention aimed at encouraging movement variability in diving [10] suggests that the capacity for athletes to harness movement system degeneracy to maintain a performance outcome is trainable. This notion is further supported by nonlinear pedagogical training interventions in youth tennis [18]. Individuals exposed to greater task variability during training displayed a greater number of unique movement clus- ters, indicating the presence of degeneracy, during performance tasks. Exposure to task vari- ability drives exploration of alternate movement strategies, or movement variability, as movement is adjusted to satisfy novel task demands [19]. Training in this way affords individ- uals the ability to adapt movement to maintain task performance under the varied constraints occurring in the dynamic sporting environment [10, 18, 20]. In the context of sub-maximal running kinematics, the effects of deliberately induced task variability through perturbation have been explored in research using elastic tubes attached from the hips to the ankles [21–23]. This intervention increases joint kinematic variability acutely, after which there is relatively rapid stabilisation around a slightly shifted coordinative structure [21, 23]. Although no post-training running test under novel conditions was under- taken, the performance benefits associated with exposure to constraints, which encourage movement variability in this way, are widely reported [24–28]. It is also worth noting that analyses of kinematic variability induced by constraint imple- mentation to date have typically focussed on group-level changes [21, 29, 30]. Increasingly, there is support for individual-level consideration given that intrinsic behavioural dynamics and baseline kinematic characteristics alter the extent to which a particular constraint is expe- rienced as a perturbation to the system [31–33]. Kinematic changes may vary markedly between individuals, which may not be clear when considering generalised responses, yet is important in a practical setting [34–36]. Lightweight wearable resistance (WR) may be a useful training tool for encouraging explo- ration of movement system degeneracy through movement variability. WR involves PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 2 / 19 attachment of small weights to particular body segments, such as the trunk, arms, thighs, and shanks [37]. To date, research has considered WR in its capacity as a movement specific over- load stimulus [38, 39], however, WR also presents a perturbation to coordination, which may induce movement variability. WR application alters segment inertial properties and as such can be considered an organismic constraint [40, 41]. Exposure to WR may ultimately be a use- ful stimulus for developing adaptable movement behaviours among athletes in preparation for changing organismic constraints faced during match play. This study aimed to describe the extent to which different acute lower limb WR loadings (1%, 3%, and 5% of body weight) alter coordination and induce movement variability during sub-maximal overground running. By considering both group- and individual-level responses, findings will provide context for coaches seeking to promote movement variability without imposing an excessive perturbation that limits movement options. Materials and methods Participants Fourteen participants (three female and 11 male; mean ± SD: age 28.3 ± 4.4 years; height: 179.9 ± 7.6 cm; body mass: 76.8 ± 6.1 kg) volunteered to participate in this study. Participants were included on the basis that they were currently undertaking, or had recent previous expe- rience (past year), in structured field-based sport competition. Participants in the study had no prior experience with WR. All participants provided written informed consent and were free from injury at the time of testing. All procedures used in this study complied with the criteria of the declaration of Helsinki and the ethical approval granted by the Victoria University Human Research Ethics Committee. Procedure Data collection apparatus. A 10-camera VICON motion analysis system (T-40 series, Vicon Nexus v2, Oxford, UK) sampling at 250 Hz was used for collection of kinematic data. A total of thirty-six reflective markers with 14mm diameter were attached to lower body land- marks on the pelvis, thighs, shanks, and feet according to the Plug-In-Gait model (Plug-In- Gait Marker Set, Vicon Peak, Oxford, UK) (Fig 1). Wearable resistance. Throughout testing, participants wore LilaTM ExogenTM (Sportbo- leh Sdh Bhd, Kuala Lumpur, Malaysia) compression shorts and calf sleeves. During WR expo- sure trials, a combination of 50, 100, and 200g fusiform shaped loads (with Velcro backing) totalling the required proportion of participants’ body weights were attached to the compres- sion garments (Fig 2). Loads were distributed in a 2:1 thigh:shank ratio about the centre of mass of each segment [42]. The required loads were added in an alternating fashion between the anterior and posterior surfaces, and between a proximal-dominant and distal-dominant orientation, in order to avoid a large shift in the centre of mass of each segment. Experimental setup. Testing was undertaken on a 20 m section of the Biomechanics Lab- oratory at Victoria University. Motion analysis cameras were arranged around the 10 m mark of the 20 m section and the approximate capture volume was 6.0 m long, 2.5 m high, and 3.0 m, wide. Data collection. Following application of compression garments and attachment of reflective markers, participants undertook an initial warm-up in which they ran back and forth along the 20 m section in a “shuttle” fashion for 2 min. Running velocity was dictated through the use of an audible metronome, which counted each second from 1–9, before repeating for every subsequent shuttle. Participants underwent a 2 min rest period following the first warm-up run before performing a second warm-up run for 1 min at an increased PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 3 / 19 Fig 1. Lower body Plug-In-Gait model. Blue markers define the required anatomical landmarks, red markers are used for tracking segments. https://doi.org/10.1371/journal.pone.0244361.g001 Fig 2. Lila™ Exogen™ compression shorts and calf sleeves with thigh and shank loading. https://doi.org/10.1371/journal.pone.0244361.g002 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 4 / 19 velocity defined by 6 s shuttle efforts. Owing to the requirement of 180˚ changes of direction after each shuttle, running velocities achieved through the capture area were greater than the theoretical straight-line velocity of 3.3 m.s-1. Analysis of pilot data showed mean ± SD veloci- ties of 4.16 ± 0.36 m.s-1 through the capture area. Such velocities are commonly described as “striding” or “running” in field based sports, but fall below the “high-intensity” classification, often defined as >5.4 m.s-1 [43, 44]. The first trial was performed with body weight only (BW), and participants completed 2 min worth of 20 m shuttles with the 6 s pacing speed per shuttle. Captures were taken each time participants passed through the 10 m mark capture area during runs from the start point to the 20 m mark only. Captures were not performed on the return shuttles. This process yielded capture of 10 complete strides across the 2 min trial. Participants performed three subsequent 2 min running trials in which they were allocated WR loading of 1%, 3%, and 5% of body weight in a randomised order. Each trial was inter- spersed with a 3 min rest period. The result of this protocol was 10 complete overground run- ning strides per condition, per participant. Data processing Visual 3D software (C-motion, Rockville, MD, USA) was used to construct a four segment model (pelvis, thigh, shank, and foot) for each participant. Within each participant, the leg on which most complete strides were successfully captured was used for analysis. This approach maximised available data given that individual stride characteristics tended to allow one side to be captured more consistently within the bounds of the 6 m capture area (see S1 Table for the leg used for each participant). Runs in which several marker trajectories were lost or accu- rate model construction could not be satisfied were excluded from analysis. Out of a possible 560 runs per-joint, 510 were successfully reconstructed for the hip, 530 for the knee, and 521 for the ankle. For a record of excluded runs and the participants and runs to which these per- tained, see S1 Table. For successfully reconstructed runs, marker trajectories were smoothed via a fourth order low-pass Butterworth filter with 10 Hz cut-off frequency, based on mean residual amplitudes [45]. Each run was trimmed to one complete stride cycle, which was defined as the period between two consecutive toe-off events on the same limb. Toe-off was defined by the initial rise in vertical displacement of the toe marker proceeding its lowest point at the end of the support phase [46, 47]. Time-continuous sagittal plane joint angles for the hip, knee, and ankle (o) were normalised to 100% of the stride cycle for further analysis. Posi- tive and negative joint angles were defined relative to the positions of joints in upright stand- ing. Positive joint angles indicate positions of hip flexion, knee flexion, and ankle dorsiflexion relative to standing, while negative joint angles indicate positions of hip extension, knee exten- sion, and ankle plantarflexion relative to standing. Data analysis Running velocity was compared across different loading conditions using a one-way repeated measures ANOVA with Bonferroni correction applied to post-hoc pairwise comparisons. A significance level of α = 0.05 was used. Statistical parametric mapping t-test. Comparisons between continuous joint angle kinematic data in the BW condition and each loading condition were performed across the group, including participants of both sexes, to identify global effects of loading. Statistical parametric mapping (SPM) t-tests were used in each instance with α = 0.05, as previously described [48, 49]. Kinematic data were estimated as functions using B-splines. A smoothing parameter of 0.01 was used in the fitting procedure. A t-statistic trajectory was created across the gait cycle and assessed in relation to a critical t-statistic, which was determined using a PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 5 / 19 permutation test by randomly shuffling the labels of the curves and recalculating the maximum t-statistic using these new labels. The analysis was done using R (version 3.6.0) and code used can be accessed at https://github.com/ktrounson/WR-running/blob/master/FDA%20t-test. Generalised additive model. Generalised additive models (GAMs) were fit to continuous joint angle data, with separate GAMs for each joint. In each case, data was modelled as a func- tion of the percentage of the stride cycle. Cyclic cubic regression splines were used to generate basis functions for each condition and smoothing was achieved using the restricted maximum likelihood method. Cubic regression splines are more appropriate for functional data that rep- resent repeated cycles of the same event [50]. The number of knots was increased until the maximum deviance explained by the model was reached. For visualisation of joint kinematic trends and between-run variability on an individual basis, runs from each participant were treated as random effects. The random effects estimates were plotted as a function of condition within each participant. Female participants are labelled F1-F3 and male participants are labelled M1-M11. All GAM code is provided at https://github.com/ktrounson/WR-running/blob/master/GAMs. Bivariate functional principal component analysis. Bivariate functional principal com- ponent analysis (bfPCA) applied to angle-angle kinematic data allows for the dominant modes of variation to be estimated. bfPCA was used to analyse concurrent hip-knee and knee-ankle kinematics using B-spline basis functions [51–53]. The smoothing parameter was selected using a generalised cross validation procedure and was set at 0.1 and 0.18 for the hip-knee and knee-ankle data, respectively. bfPCs were derived from the smoothed curves. Each bfPC was varimax rotated to assist with interpretation of results. The occurrence and magnitude of angle-angle variability was graphically represented by the first two bfPCs on individual plots containing the ensemble mean of curves along with two additional curves representing +/- 2SD of the bfPC scores for each bfPC. bfPCA was performed in R with code available at https://github.com/ktrounson/WR-running/blob/master/bfPCA. Individual-based 2D plots were generated in which mean bfPC scores for each condition were mapped along the first two bfPCs for each joint pairing. Positive scores along a dimension indicate that, on average, runs within this condition resembled more closely the characteristics of the ‘+’ curve, while negative scores indicate a closer resemblance to the ‘-’ curve. Results Mean running velocities across participants in each condition are included in Table 1. A signif- icant main effect of condition was evident (F = 4.77, p = 0.003). Post-hoc analysis showed slower running velocities in the 5% loading condition compared with all other conditions. SPM t-test Continuous ensemble means per-joint and per-condition with associated standard devia- tions are presented in Fig 3. Sections of significant difference between the BW condition and Table 1. Mean ± SD running velocities in each condition with post-hoc pairwise comparisons. Condition Running velocity (m.s-1) p-value vs. 1% p-value vs. 3% p-value vs. 5% BW 4.25 ± 0.43 1 1 0.017 1% 4.25 ± 0.47 1 0.005 3% 4.25 ± 0.48 0.022 5% 4.18 ± 0.44 BW, body weight; 1%, 1% of body weight WR loading; 3%, 3% of body weight WR loading; 5%, 5% of body weight WR loading. https://doi.org/10.1371/journal.pone.0244361.t001 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 6 / 19 each loading condition according to SPM t-tests are indicated. WR loading of 1% of body weight led to greater hip extension at 97–99% of the gait cycle (just prior to toe-off) com- pared with BW (P = 0.045). Loading of 3% of body weight resulted in less ankle plantarflex- ion from 12–18% of the gait cycle (during heel recovery) compared with BW (P = 0.035). Loading of 5% of body weight resulted in greater knee flexion from 66–81% of the gait cycle (during weight acceptance) (P < 0.001) and less ankle plantarflexion from 9–30% of the gait cycle (during heel recovery) compared with BW (P < 0.001). Pointwise t-statistics and the maximum critical value for a significance level of 0.05 for each set of curves are provided in S2 Table. Fig 3. SPM t-test per-joint and per-condition versus BW. (A) Hip joint BW versus 1%. (B) Hip joint BW versus 3%. (C) Hip joint BW versus 5%. (D) Knee joint BW versus 1%. (E) Knee joint BW versus 3%. (F) Knee joint BW versus 5%. (G) Ankle joint BW versus 1%. (H) Ankle joint BW versus 3%. (I) Ankle joint BW versus 5%. Solid lines represent ensemble means and accompanying shaded regions represent ± 1 SD. Grey shaded regions indicate regions of significant difference between curve sets. https://doi.org/10.1371/journal.pone.0244361.g003 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 7 / 19 GAMs The summary statistics of each joint GAM are shown in Table 2. For BW, the estimate indi- cates the mean joint angle across the stride cycle. For loading conditions, estimates indicate the difference in mean joint angle across the stride cycle versus BW. Estimated degrees of free- dom reflect the number of basis functions used to generate the smooths and therefore a higher number of estimated degrees of freedom suggests more variable data. For loading conditions, estimated degrees of freedom are in addition to those listed for BW. The GAM random effects estimates per-run are shown in Fig 4. Random effects estimates reflect the prevailing flexion-extension bias throughout the stride cycle relative to the group mean. The distribution of random effects estimates appeared to be more strongly driven by the participant in question than within-participant responses to WR loading. However, indi- vidual-level responses of note include instances in which loading increased between-run vari- ability, such as at the ankle in the 5% condition for F1, the knee in the 5% condition for M1, the knee in the 1% condition for M2, the knee in all loading conditions for M3, the hip and ankle in the 3% condition for F2, and the hip in the 5% condition for M7. Conversely, decreased between-run variability was evident at the ankle in the 3% condition for M1, the ankle in the 3% and 5% conditions for M2, the hip in the 5% condition for M5, the ankle in the 5% condition for F2 and M6, and the ankle in the 1% condition for F3. Shifts in the prevailing random effects estimates on the basis of loading appeared evident at the ankle in the 1% and 3% conditions for M2, the hip in the 3% condition for M3 and F2, the ankle in the 5% condi- tion for M6, the hip in the 3% and 5% conditions for M7, the hip in the 1% condition for F3, and the hip in the 1% and 3% conditions for M11. bfPCA For hip-knee joint coupling, bfPC1 explained 41.1% of the variability in the group data (Fig 5). Positive scorers on bfPC1 exhibited less knee flexion during the swing phase, while negative scorers exhibited greater knee flexion. bfPC2 explained 23.3% of the variability in the group data. Positive scorers on bfPC2 exhibited greater hip flexion during the swing phase, while neg- ative scorers exhibited less hip flexion and hip flexion was delayed compared with positive scorers. For knee-ankle joint coupling, bfPC1 explained 45.6% of the variability in the group Table 2. Generalised additive model summary statistics per-joint. Parametric coefficients Smooth terms Joint Condition Estimate Standard error t-value Pr(>|t|) EDF F p-value Hip BW (intercept) 14.16 1.35 10.5 > 0.001 12.96 22666.3 > 0.001 1% -0.65 0.06 -10.99 > 0.001 3.8 4.94 > 0.001 3% -0.12 0.06 -2.05 0.04 4.51 3.64 > 0.001 5% -0.71 0.06 -11.76 > 0.001 7.62 7.12 > 0.001 Knee BW (intercept) 43.21 1.15 37.65 > 0.001 8 32137.8 > 0.001 1% 0.55 0.1 5.25 > 0.001 4.05 1.47 0.009 3% 0.71 0.1 6.86 > 0.001 3.54 7.42 > 0.001 5% 0.15 0.1 1.48 0.14 7.29 52.18 > 0.001 Ankle BW (intercept) -18.5 1.06 -17.48 > 0.001 20.85 6141 > 0.001 1% 0.33 0.07 4.95 > 0.001 5.85 2.6 > 0.001 3% -0.15 0.07 -2.18 0.03 3.91 1.77 > 0.001 5% 1.31 0.07 19.49 > 0.001 5.55 5.06 > 0.001 EDF, estimated degrees of freedom; BW, body weight; 1%, 1% of body weight WR loading; 3%, 3% of body weight WR loading; 5%, 5% of body weight WR loading. https://doi.org/10.1371/journal.pone.0244361.t002 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 8 / 19 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 9 / 19 data. Positive scorers exhibited less knee flexion during the swing phase and negative scorers exhibited greater knee flexion. bfPC2 explained 16.8% of the variability in the group data. Posi- tive scorers on bfPC2 exhibited less ankle plantarflexion, particularly at touchdown, while neg- ative scorers exhibited greater ankle plantarflexion during late swing and touchdown. Mean individual bfPC scores along both bfPCs for hip-knee and knee-ankle joint pairs across runs in each condition are shown in Fig 6. Participants appeared to have mostly distinct joint coupling profiles and there was some impact of WR loading within-individuals. There was generally agreement between observations of condition-based shifts in random effects esti- mates from GAM analysis and differences in mean bfPC scores, including in the knee-ankle joint couple in the 3% condition for M2, the hip-knee joint couple in the 3% condition for M3 and F2, the knee-ankle joint couple in the 5% condition for M6, the hip-knee joint couple in the 3% and 5% conditions for M7, and the hip-knee joint couple in the 1% condition for F3 and M11. Additional condition-based shifts apparent from bfPCs included the hip-knee joint couple in the 1% condition in M2, the hip-knee joint couple in the 5% condition for M3, M6, M8, and M10, and the knee-ankle joint couple in the 1% condition for M11. Shifts that were identified from GAM analysis but that appeared to be minimal based on bfPC plots included the knee-ankle joint couple in the 1% condition for M3 and the hip-knee joint couple in the 3% condition for M11. Fig 4. GAM random effects estimates per-run, per-individual. Each major panel relates to a given participant, as denoted by labels. Joints are separated by the three minor panels within each participant plot. Conditions are expressed as categories within each joint and associated colours have been included for clarity. Positive estimates indicate greater hip flexion, knee flexion, and ankle dorsiflexion relative to the group mean. Negative estimates indicate greater hip extension, knee extension, and ankle plantarflexion. Thicker regions of coloured portions reflect a greater concentration of runs with similar random effects estimates. https://doi.org/10.1371/journal.pone.0244361.g004 Fig 5. bfPCA of hip-knee and knee-ankle joint couples throughout stride cycle. First two bfPCs from hip-knee and knee-ankle bfPCA with the percentage of group variability explained. Solid line represents the mean angle-angle curve. ‘+’ line represents positive scorers +2SD from the mean function. ‘-’ line represents negative scorers -2SD from the mean function. https://doi.org/10.1371/journal.pone.0244361.g005 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 10 / 19 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 11 / 19 Discussion This study examined the effects of lower limb WR loading on coordination tendencies during sub-maximal overground running. Specifically, the study sought to describe the effects of vari- ous WR magnitudes (1%, 3%, and 5% of body weight) on lower limb sagittal plane joint kine- matics at a group- and individual-level, both in terms of continuous gait cycle kinematics and between-run movement variability. The main findings at a group-level were that 3% and 5% loading decreased ankle plantarflexion during heel recovery, while 5% loading also increased knee flexion during weight acceptance, compared with BW running. In terms of joint cou- pling, 5% loading brought about the largest changes in coordination at the hip-knee joint pair. At an individual-level, six of the fourteen participants clearly exhibited increased between-run joint angle variability at one or more joints in one or more loading conditions compared with BW. Running velocity was slower in the 5% condition compared with all other conditions, how- ever, the magnitude of this difference was minimal at just 0.07 m.s-1. All participants main- tained speeds sufficient to successfully complete all shuttles within the allotted time frames in all conditions. In terms of kinematics at the group-level, slight decreases in ankle plantarflexion during heel recovery occurred with 3% loading compared with BW running. The 5% loading condi- tion led to more substantial decreases in ankle plantarflexion during heel recovery, as well as increased knee flexion during weight acceptance. The exhibition of greater knee flexion was likely a mechanism to mitigate increased peak ground reaction force arising from the greater system load [54, 55]. Explanations for less plantarflexion during heel recovery are more specu- lative. One possibility is that participants subconsciously attempted to offset the greater moment of inertia at the thigh by dorsiflexing the ankle to create a mechanical advantage dur- ing swing leg recovery [56, 57]. Alternatively, or perhaps in addition, heavier loading likely led to increased co-contraction of muscles around the ankle joint during stance for maintenance of stiffness and stability [55, 58]. Such alterations in motor unit recruitment and temporal sequencing of lower leg muscles may constrain the action of this joint during the subsequent propulsion and swing phase, with the joint returning to a relatively more neutral position more readily [59, 60]. The impact of coordination dynamics should also be considered. Indi- viduals performing novel motor tasks often exhibit freezing of distal biomechanical degrees of freedom to reduce coordinative complexity [61–63]. To the extent that running with an extra 5% of body weight on the lower limbs was perceived as a novel task, there may have been a ten- dency for participants to return to a more neutral ankle position following toe-off. Given these factors, 5% loading, and to a lesser extent 3% loading, may be excessive as a means of promot- ing movement variability for some individuals in the first instance. The group-level changes suggest a degree of convergence toward a common adaptation strategy and appear consistent with movement options being limited by task novelty and/or the need to manage high loads. Coaches should take this into consideration if prescribing WR for multiple athletes without individualisation [64]. Despite group-level trends, individual responses varied. The practical utility of WR for inducing movement variability is therefore likely to also be individual-dependent. Coaches should appreciate the range of individual responses and use the present findings as signposts to guide individual WR prescription in the field. Fig 6. Individual mean hip-knee and knee-ankle bfPC1 and bfPC2 scores per-condition. Each panel relates to a given participant, as denoted by labels. Mean hip- knee bfPC scores across runs within a condition are denoted by circle labels. Mean knee-ankle bfPC scores across runs within a condition are denoted by triangle labels. Separate conditions are indicated by distinct colours. https://doi.org/10.1371/journal.pone.0244361.g006 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 12 / 19 Participants F1, M1, M2, M3, F2, and M7 all exhibited increased between-run variability in mean angles at one or more joints in one or more loading conditions compared with BW. These high variability instances suggest that there was no readily accessible adaptive mode to satisfy the task goal in the presence of WR and instead a period of search and refinement of individuals’ preferred coordinative structures was required [65, 66]. WR in this context there- fore provides an opportunity to explore movement system degeneracy. For these individuals, exposure to WR over a training period may facilitate development of movement adaptability and allow running performance to be more readily maintained when perturbations arise in competition [10, 67]. The propensity for individuals to exhibit greater variability at one loading condition over another is largely a function of intrinsic behavioural dynamics, which dictate system tendencies such as attractor state stability and behavioural meta-stability [68, 69]. A definitive reduction in between-run variability was present in participants F2 and M6 at the ankle joint in the 5% loading condition. This may align with the proposed group-level hypothesis of distal joint freezing in this condition. While established literature tends to define freezing degrees of freedom as restricted movement of a joint within-trials, low between-trial variability is also indicative of constrained movement [20, 70]. Among these individuals, the perturbation of 5% loading may have been managed by increasing co-contractions and stiffen- ing muscles of the lower limbs, as occurs in the early stages of skill acquisition [61]. Interest- ingly, this can be considered an adaptive strategy in itself, particularly since performance of shuttle runs was successfully maintained. This therefore raises the need to clarify the benefit of perturbations that encourage movement variability during training versus those that limit movement variability. Findings from balance beam walking with different perturbation mag- nitudes demonstrate that learning under conditions in which sacral movement variability is maximised leads to superior learning and subsequent task performance post-training [71]. Substantially increasing the level of perturbation through an error augmenting device decreases movement variability in line with individuals attempting to maintain control of movement, and has suboptimal outcomes for post-training performance [71]. Separately, Chmielewski et al. [60] argue that increased co-contractions as a means of adapting to an ACL rupture reflect a suboptimal compensation pattern wherein the capacity to dynamically stabi- lise the injured knee without compromising knee motion has not yet been developed. Taken together, these findings highlight that large magnitude perturbations may be adapted to by reducing movement variability, however, skill acquisition is not facilitated under these condi- tions. Reduced movement variability affords fewer opportunities for internal models of limb dynamics to be updated, which may limit the extent to which adaptability is trained [72]. Among high performing field-based athletes, some individuals are likely to already have well developed functional movement adaptability [11, 12]. This is typified by an appropriate mix of movement pattern flexibility and stability, such that coordination can be readily adjusted in response to a perturbation and movement variability levels remain similar to those at baseline [10, 73]. Potential exemplars of this in the present study include participant M10 in all loading conditions and participant M7 in the 3% loading condition. Participants M4 and M9 exhibited no discernable joint kinematic changes at any loading magnitude. Between-run variability also appeared consistent across loading. For these individ- uals, loading even up to 5% of body weight may not have required additional exploitation of movement system degeneracy to satisfy the task goal [74]. As part of their intrinsic behavioural dynamics, these individuals likely defer to highly stable movement attractor states in the pres- ence of manageable perturbations [65, 75]. Practically, WR may not be appropriate to chal- lenge running coordination among such individuals, as loading beyond 5% of body weight on the lower limbs presents logistical difficulties due to load placement space limitations. PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 13 / 19 Lastly, it is interesting to note that participants M2 and M11 appeared to demonstrate multi- stability about the ankle joint. There appeared to be two dominant kinematic modes expressed with apparent condition dependence in M2 but not in M11. Coaches should appreciate that ath- letes may exhibit multi-stability, wherein two or more patterns of coordination are stable [68]. A limitation of the present study is that only sagittal plane kinematics were considered. Consequential alterations to kinematics may have also occurred in the transverse and frontal planes, or in trunk or upper body segments. In terms of the WR, an exactly equal distribution of load between the anterior and posterior segment surfaces could not always be guaranteed. In these instances, one surface of each segment experienced 50 g more loading, which although minimal, may have impacted ensuing running kinematics. In relation to data processing, it is important to acknowledge that although the initial rise in vertical displacement of the toe marker has previously been used to define toe-off during running actions [46, 47, 76], valida- tion of this detection method against force plate measures under the specific running velocities and floor surface conditions of the present study has not been performed. Lastly, despite the 3 min rest period allowed between running trials, residual after-effects following heavier loading conditions could have briefly impacted on the kinematics observed during lighter conditions [77]. When SPM t-tests were repeated with participants separated on the basis of having com- pleted the 1% condition immediately following the 5% condition as part of their randomisa- tion, it was evident that the group-level differences in hip extension between the 1% condition and BW were driven by these participants (S1 Fig). A larger sample size would provide clarity on this point by enabling direct statistical comparisons between participants who experienced the 1% condition immediately following the 5% condition, and those that did not. If this type of loading contrast was an effectual factor, fidelity could be improved by allowing participants to rest for longer or briefly run without loading in between trials to “re-establish” an unloaded baseline. Future research should specifically consider the effects of unloaded running immediately following a period of loading to clarify the propensity for acute coordinative changes to be retained following the removal of perturbation. Investigation into the impact of asymmetrical WR loading on coordination would also be worthwhile given the challenge to the movement system that such an intervention would pose. As understanding of the effects of WR loading develops, researchers and/or coaches should consider situating tasks such as loaded running in a representative, field-based environment. WR coupled with the inherent movement variabil- ity induced by dynamic constraints and affordances in this environment would present a fur- ther, more contextual, challenge to coordination [78]. Conclusions Exposure to WR of 5% of body weight increased knee flexion during weight acceptance and decreased ankle plantarflexion during heel recovery at the group-level. This appeared to be due to the high load and novelty of this condition. Among individuals that reflected group- level trends and exhibited decreased between-run variability at one or more joints, 5% loading may be an excessive perturbation, as exploration of alternate movement states is limited. Sev- eral participants exhibited increased between-run joint angle variability in one or more load- ing conditions compared with BW, suggesting exploration and refinement of coordinative structures under these conditions. The loading magnitudes at which these increases were elic- ited, however, varied between individuals. WR therefore appears to show utility for the pur- pose of perturbing coordination to encourage movement variability among certain individuals, though the loading magnitudes used should be determined on a case-by-case basis. PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 14 / 19 Supporting information S1 Table. List of legs analysed and joint angles unable to be reconstructed for analysis. Joint angle data that could not be reconstructed is highlighted in red. (DOCX) S2 Table. Pointwise t-statistics and maximum critical values for SPM t-tests. (DOCX) S1 Fig. Hip joint SPM t-test BW versus 1% separated based on condition order. (A) Hip joint BW versus 1% for participants in which 1% condition did not immediately proceed 5% condition. (B) Hip joint BW versus 1% for participants in which 1% condition immediately proceeded 5% condition. Solid lines represent ensemble means and accompanying shaded regions represent ± 1 SD. Grey shaded regions indicate regions of significant difference between curve sets. (TIF) Acknowledgments The authors would like to acknowledge the research participants for their involvement in this study. Author Contributions Conceptualization: Karl M. Trounson, Aglaja Busch, Sam Robertson. Data curation: Karl M. Trounson, Aglaja Busch, Neil French Collier. Formal analysis: Karl M. Trounson, Neil French Collier. Investigation: Karl M. Trounson, Aglaja Busch. Methodology: Karl M. Trounson, Aglaja Busch, Sam Robertson. Supervision: Sam Robertson. Visualization: Karl M. Trounson, Neil French Collier. Writing – original draft: Karl M. Trounson. Writing – review & editing: Karl M. Trounson, Neil French Collier, Sam Robertson. References 1. Varley MC, Gabbett T, Aughey RJ. Activity profiles of professional soccer, rugby league and Australian football match play. J Sports Sci. 2014; 32(20):1858–66. https://doi.org/10.1080/02640414.2013. 823227 PMID: 24016304 2. Dawson B, Hopkinson R, Appleby B, Stewart G, Roberts C. Player movement patterns and game activi- ties in the Australian Football League. J Sci Med Sport. 2004; 7(3):278–91. https://doi.org/10.1016/ s1440-2440(04)80023-9 PMID: 15518293 3. Suarez-Arrones LJ, Nunez FJ, Portillo J, Mendez-Villanueva A. Running demands and heart rate responses in men Rugby Sevens. J Strength Cond Res. 2012; 26(11):3155–9. https://doi.org/10.1519/ JSC.0b013e318243fff7 PMID: 22158098 4. Andrzejewski M, Konefał M, Chmura P, Kowalczuk E, Chmura J. Match outcome and distances cov- ered at various speeds in match play by elite German soccer players. Int J Perf Anal Spor. 2016; 16 (3):817–28. 5. Moore IS. Is there an economical running technique? A review of modifiable biomechanical factors affecting running economy. Sports Med. 2016; 46(6):793–807. https://doi.org/10.1007/s40279-016- 0474-4 PMID: 26816209 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 15 / 19 6. Morgan DW, Baldini FD, Martin PE, Kohrt WM. Ten kilometer performance and predicted velocity at VO2max among well-trained male runners. Med Sci Sports Exerc. 1989; 21(1):78–83. https://doi.org/ 10.1249/00005768-198902000-00014 PMID: 2927305 7. Conley DL, Krahenbuhl GS. Running economy and distance running performance of highly trained ath- letes. Med Sci Sports Exerc. 1980; 12(5):357–60. PMID: 7453514 8. Folland JP, Allen SJ, Black MI, Handsaker JC, Forrester SE. Running technique is an important compo- nent of running economy and performance. Med Sci Sports Exerc. 2017; 49(7):1412–23. https://doi. org/10.1249/MSS.0000000000001245 PMID: 28263283 9. Komar J, Seifert L, Thouvarecq R. What variability tells us about motor expertise: measurements and perspectives from a complex system approach. Mov Sport Sci/Sci Mot. 2015(89):65–77. 10. Barris S, Farrow D, Davids K. Increasing functional variability in the preparatory phase of the takeoff improves elite springboard diving performance. Res Q Exerc Sport. 2014; 85(1):97–106. https://doi.org/ 10.1080/02701367.2013.872220 PMID: 24749241 11. Schorer J, Baker J, Fath F, Jaitner T. Identification of interindividual and intraindividual movement pat- terns in handball players of varying expertise levels. J Mot Behav. 2007; 39(5):409–21. https://doi.org/ 10.3200/JMBR.39.5.409-422 PMID: 17827117 12. Chow JY, Davids K, Button C, Koh M. Organization of motor system degrees of freedom during the soc- cer chip: an analysis of skilled performance. Int J Sport Psychol. 2006; 37(2–3):207–29. 13. van Emmerik RE, Ducharme SW, Amado AC, Hamill J. Comparing dynamical systems concepts and techniques for biomechanical analysis. J Sport Health Sci. 2016; 5(1):3–13. https://doi.org/10.1016/j. jshs.2016.01.013 PMID: 30356938 14. Bishop DJ. Fatigue during intermittent-sprint exercise. Clin Exp Pharmacol Physiol. 2012; 39(9):836– 41. https://doi.org/10.1111/j.1440-1681.2012.05735.x PMID: 22765227 15. Morel B, Rouffet DM, Bishop DJ, Rota SJ, Hautier CA. Fatigue induced by repeated maximal efforts is specific to the rugby task performed. Int J Sports Sci Coa. 2015; 10(1):11–20. 16. Thompson D, Nicholas CW, Williams C. Muscular soreness following prolonged intermittent high-inten- sity shuttle running. J Sports Sci. 1999; 17(5):387–95. https://doi.org/10.1080/026404199365902 PMID: 10413266 17. Gabbe B, Finch C, Wajswelner H, Bennell K. Australian football: injury profile at the community level. J Sci Med Sport. 2002; 5(2):149–60. https://doi.org/10.1016/s1440-2440(02)80036-6 PMID: 12188087 18. Lee MC, Chow JY, Komar J, Tan CW, Button C. Nonlinear pedagogy: an effective approach to cater for individual differences in learning a sports skill. PLoS One. 2014; 9(8):e104744. https://doi.org/10.1371/ journal.pone.0104744 PMID: 25140822 19. Chow JY, Davids K, Hristovski R, Arau´jo D, Passos P. Nonlinear pedagogy: learning design for self- organizing neurobiological systems. New Ideas Psychol. 2011; 29(2):189–200. 20. Chow JY, Davids K, Button C, Koh M. Variation in coordination of a discrete multiarticular action as a function of skill level. J Mot Behav. 2007; 39(6):463–79. https://doi.org/10.3200/JMBR.39.6.463-480 PMID: 18055353 21. Haudum A, Birklbauer J, Muller E. The effect of external perturbations on variability in joint coupling and single joint variability. Hum Mov Sci. 2014; 36:246–57. https://doi.org/10.1016/j.humov.2014.02.004 PMID: 24636698 22. Haudum A, Birklbauer J, Muller E. The effect of an acute bout of rubber tube running constraint on kine- matics and muscle activity. J Sports Sci Med. 2012; 11(3):459–67. PMID: 24149354 23. Haudum A, Birklbauer J, Mu¨ller E. The influence of external perturbations on running kinematics and muscle activity before and after accommodation. J Sports Sci Med. 2012; 11(4):582. PMID: 24150066 24. Hernandez-Davo H, Urban T, Sarabia JM, Juan-Recio C, Moreno FJ. Variable training: effects on veloc- ity and accuracy in the tennis serve. J Sports Sci. 2014; 32(14):1383–8. https://doi.org/10.1080/ 02640414.2014.891290 PMID: 24702059 25. Garcı´a-Herrero JA, Sanchez-Sanchez J, Luis-Pereira JM, Menayo R. The effects of induced variability in the performance on shot in soccer. Int J Sports Sci Coach. 2016; 11(5):648–54. 26. Savelsbergh GJ, Kamper WJ, Rabius J, De Koning JJ, Scho¨llhorn W. A new method to learn to start in speed skating: A differencial learning approach. Int J Sport Psychol. 2010; 41(4):415. 27. Scho¨llhorn WI, Beckmann H, Janseen D, Drepper J. Stochastic perturbation in athletics field events enhance skill acquisition. In: Renshaw I, Davids K, Savelsbergh JP, editors. Motor learning in practice: A constraints-led approach. London: Routledge; 2010. 28. Harbourne RT, Stergiou N. Movement variability and the use of nonlinear tools: principles to guide phys- ical therapist practice. Phys Ther. 2009; 89(3):267–82. https://doi.org/10.2522/ptj.20080130 PMID: 19168711 PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 16 / 19 29. Apps C, Sterzing T, O’Brien T, Ding R, Lake M. Biomechanical locomotion adaptations on uneven sur- faces can be simulated with a randomly deforming shoe midsole. Footwear Sci. 2017; 9(2):65–77. 30. Frank NS, Callaghan JP, Prentice SD. Lower limb kinematic variability associated with minimal footwear during running. Footwear Sci. 2013; 5(3):171–7. 31. Kostrubiec V, Zanone PG, Fuchs A, Kelso JA. Beyond the blank slate: routes to learning new coordina- tion patterns depend on the intrinsic dynamics of the learner-experimental evidence and theoretical model. Front Hum Neurosci. 2012; 6:222. https://doi.org/10.3389/fnhum.2012.00222 PMID: 22876227 32. Scho¨llhorn W, Nigg B, Stefanyshyn D, Liu W. Identification of individual walking patterns using time dis- crete and time continuous data sets. Gait Posture. 2002; 15(2):180–6. https://doi.org/10.1016/s0966- 6362(01)00193-x PMID: 11869912 33. Scho¨llhorn WI, Bauer HU, editors. Identifying individual movement styles in high performance sports by means of self-organizing Kohonen maps. Proceedings of the XVI Annual Conference of the Interna- tional Society for Biomechanics in Sport; 1998 Jul 7–12; Konstanz, Germany. 34. Apps C, Ding R, Cheung JT-M, Sterzing T. Individual and generalized lower limb muscle activity and kinematic adaptations during walking on an unpredictable irregular surface. J Foot Ankle Res. 2014;7 (1). https://doi.org/10.1186/1757-1146-7-7 PMID: 24468080 35. Walker C, Warmenhoven J, Sinclair PJ, Cobley S. The application of inertial measurement units and functional principal component analysis to evaluate movement in the forward 3(1/2) pike somersault springboard dive. Sports Biomech. 2019; 18(2):146–62. https://doi.org/10.1080/14763141.2019. 1574887 PMID: 31042139 36. Alfonso M, Menayo R. Induced variability during the tennis service practice affect the performance of every tennis player individually and specifically. Eur J Hum Mov. 2019; 43:86–101. 37. Macadam P, Cronin JB, Uthoff AM, Feser EH. Effects of different wearable resistance placements on sprint-running performance: A review and practical applications. Strength Cond J. 2019; 41(3):79–96. 38. Simperingham K, Cronin J. Changes in sprint kinematics and kinetics with upper body loading and lower body loading using exogen exoskeletons: A pilot study. J Aust Strength Cond. 2014; 22(5):69–72. 39. Simperingham K, Cronin J, Pearson S, Ross A, editors. Changes in acceleration phase sprint biome- chanics with lower body wearable resistance. Proceedings of the 34th International Conference of Bio- mechanics in Sports; 2016 July 18–22, 2016; Tsukuba, Japan. 40. Field AP, Gill N, Macadam P, Plews D. Acute metabolic changes with thigh-positioned wearable resis- tances during submaximal running in endurance-trained runners. Sports. 2019; 7(8):187. https://doi. org/10.3390/sports7080187 PMID: 31375020 41. Martin PE, Cavanagh PR. Segment interactions within the swing leg during unloaded and loaded run- ning. J Biomech. 1990; 23(6):529–36. https://doi.org/10.1016/0021-9290(90)90046-6 PMID: 2341416 42. Macadam P, Simperingham KD, Cronin JB, Couture G, Evison C. Acute kinematic and kinetic adapta- tions to wearable resistance during vertical jumping. Eur J Sport Sci. 2017; 17(5):555–62. https://doi. org/10.1080/17461391.2017.1298672 PMID: 28316257 43. Sua´rez-Arrones LJ, Portillo LJ, Gonza´lez-Rave´ JM, Muñoz VE, Sanchez F. Match running performance in Spanish elite male rugby union using global positioning system. Isokinet Exerc Sci. 2012; 20(2):77– 83. 44. Carling C, Le Gall F, Dupont G. Analysis of repeated high-intensity running performance in professional soccer. J Sports Sci. 2012; 30(4):325–36. https://doi.org/10.1080/02640414.2011.652655 PMID: 22248291 45. Winter DA. Biomechanics and motor control of human movement. 4th ed. New York: John Wiley & Sons; 2009. 46. Schache AG, Blanch PD, Rath DA, Wrigley TV, Starr R, Bennell KL. A comparison of overground and treadmill running for measuring the three-dimensional kinematics of the lumbo–pelvic–hip complex. Clin Biomech. 2001; 16(8):667–80. https://doi.org/10.1016/s0268-0033(01)00061-4 PMID: 11535348 47. Nagahara R, Zushi K. Determination of foot strike and toe-off event timing during maximal sprint using kinematic data. J Sport Health Sci. 2013; 11:96–100. 48. Warmenhoven J, Harrison A, Robinson MA, Vanrenterghem J, Bargary N, Smith R, et al. A force profile analysis comparison between functional data analysis, statistical parametric mapping and statistical non-parametric mapping in on-water single sculling. J Sci Med Sport. 2018; 21(10):1100–5. https://doi. org/10.1016/j.jsams.2018.03.009 PMID: 29650339 49. Ramsay JO, Hooker G, Graves S. Functional Data Analysis with R and MATLAB: Springer Science & Business Media; 2009. 50. Wood SN, Goude Y, Shaw S. Generalized additive models for large data sets. J R Stat Soc C-Appl. 2015:139–55. PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 17 / 19 51. Warmenhoven J, Cobley S, Draper C, Harrison A, Bargary N, Smith R. Bivariate functional principal components analysis: considerations for use with multivariate movement signatures in sports biome- chanics. Sports Biomech. 2019; 18(1):10–27. https://doi.org/10.1080/14763141.2017.1384050 PMID: 29125036 52. Ramsay JO, Silverman BW. Functional Data Analysis: Wiley Online Library; 2005. 53. Harrison AJ, Ryan W, Hayes K. Functional data analysis of joint coordination in the development of ver- tical jump performance. Sports Biomech. 2007; 6(2):199–214. https://doi.org/10.1080/ 14763140701323042 PMID: 17892096 54. Liew B, Netto K, Morris S. Increase in leg stiffness reduces joint work during backpack carriage running at slow velocities. J Appl Biomech. 2017; 33(5):347–53. https://doi.org/10.1123/jab.2016-0226 PMID: 28530461 55. Silder A, Besier T, Delp SL. Running with a load increases leg stiffness. J Biomech. 2015; 48(6):1003– 8. https://doi.org/10.1016/j.jbiomech.2015.01.051 PMID: 25728581 56. Jeffreys I, Moody J. Strength and Conditioning for Sports Performance. Abingdon: Routledge; 2016. 57. Seagrave L, Mouchbahani R, O’Donnell K. Neuro-biomechanics of maximum velocity sprinting. New Stud Athl. 2009; 24:19–27. 58. Di Nardo F, Mengarelli A, Burattini L, Maranesi E, Agostini V, Nascimbeni A, et al. Normative EMG pat- terns of ankle muscle co-contractions in school-age children during gait. Gait Posture. 2016; 46:161–6. https://doi.org/10.1016/j.gaitpost.2016.03.002 PMID: 27131195 59. Marras WS, Rangaraajulu SL, Lavender SA. Trunk loading and expectation. Ergonomics. 1987; 30:551–62. https://doi.org/10.1080/00140138708969744 PMID: 3595552 60. Chmielewski TL, Hurd WJ, Rudolph KS, Axe MJ, Snyder-Mackler L. Perturbation training improves knee kinematics and reduces muscle co-contraction after complete unilateral anterior cruciate ligament rupture. Phys Ther. 2005; 85(8):740–9; discussion 50–4. PMID: 16048422 61. Vereijken B, Emmerik REA, Whiting HTA, Newell KM. Free(z)ing degrees of freedom in skill acquisition. J Mot Behav. 1992; 24(1):133–42. 62. Bernstein NA. The control and regulation of movements. London: Pergamon Press; 1967. 63. Southard D, Higgins T. Changing movement patterns: Effects of demonstration and practice. Res Q Exerc Sport. 1987; 58(1):77–80. 64. Bustos A, Metral G, Cronin J, Uthoff A, Dolcetti J. Effects of Warming Up With Lower-Body Wearable Resistance on Physical Performance Measures in Soccer Players Over an 8-Week Training Cycle. J Strength Cond Res. 2020; 34(5):1220–6. https://doi.org/10.1519/JSC.0000000000003498 PMID: 32149881 65. Davids K, Arau´jo D, Shuttleworth R, Button C. Acquiring skill in sport: A constraints-led perspective. Int J Comp Sci Sport. 2003; 2:31–9. 66. Davids K, Bennett SJ, Button C. Coordination and control of movement in sport: An ecological approach. Champaign, IL: Human Kinetics; 2008. 67. Newell KM, Kugler PN, van Emmerik REA, McDonald PV. Search strategies and the acquisition of coor- dination. In: Wallace SA, editor. Perspectives on the Coordination of Movement. Amsterdam, The Netherlands: North-Holland; 1989. p. 85–122. 68. Kelso JAS. Multistability and metastability: understanding dynamic coordination in the brain. Philos Trans R Soc Lond B Biol Sci. 2012; 367(1591):906–18. https://doi.org/10.1098/rstb.2011.0351 PMID: 22371613 69. Juarrero A. Dynamics in action: Intentional behavior as a complex system. Emergence. 1999; 2(2):24– 57. 70. Higuchi T, Imanaka K, Hatayama T. Freezing degrees of freedom under stress: kinematic evidence of constrained movement strategies. Hum Mov Sci. 2002; 21(5–6):831–46. https://doi.org/10.1016/s0167- 9457(02)00174-4 PMID: 12620722 71. Domingo A, Ferris DP. The effects of error augmentation on learning to walk on a narrow balance beam. Exp Brain Res. 2010; 206(4):359–70. https://doi.org/10.1007/s00221-010-2409-x PMID: 20853102 72. Wolpert DM, Ghahramani Z. Computational principles of movement neuroscience. Nat Neurosci. 2000; 3(11):1212–7. https://doi.org/10.1038/81497 PMID: 11127840 73. Newell KM, Corcos D. Variability and Motor Control. Champaign, IL: Human Kinetics; 1991. 74. Edelman GM, Gally JA. Degeneracy and complexity in biological systems. Proc Natl Acad Sci. 2001; 98 (24):13763–8. https://doi.org/10.1073/pnas.231499798 PMID: 11698650 75. Kelso JAS. Dynamic patterns: The self-organization of brain and behavior. Cambridge: MIT press; 1995. PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 18 / 19 76. Bezodis I, Thomson A, Gittoes M, Kerwin DG. Identification of instants of touchdown and take-off in sprint running using an automatic motion analysis system. XXV International Symposium on Biome- chanics in Sports; Ouro Preto, Brazil: International Society of Biomechanics in Sports; 2007. https:// doi.org/10.1080/14763140701324487 PMID: 17892094 77. Nakamoto H, Ishii Y, Ikudome S, Ohta Y. Kinesthetic aftereffects induced by a weighted tool on move- ment correction in baseball batting. Hum Mov Sci. 2012; 31(6):1529–40. https://doi.org/10.1016/j. humov.2012.04.005 PMID: 22698835 78. Arau´jo D, Davids K, Hristovski R. The ecological dynamics of decision making in sport. Psychol Sport Exerc. 2006; 7(6):653–76. PLOS ONE Effects of wearable resistance on running kinematics PLOS ONE | https://doi.org/10.1371/journal.pone.0244361 December 28, 2020 19 / 19

Effects of acute wearable resistance loading on overground running lower body kinematics.

12-28-2020

Trounson, Karl M,Busch, Aglaja,French Collier, Neil,Robertson, Sam

eng

PMC10403529

1 Vol.:(0123456789) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports Reliability of threshold determination using portable muscle oxygenation monitors during exercise testing: a systematic review and meta‑analysis Carlos Sendra‑Pérez 1, Jose Luis Sanchez‑Jimenez 1, Joaquín Martín Marzano‑Felisatti 1, Alberto Encarnación‑Martínez 1,2, Rosario Salvador‑Palmer 2,3 & Jose I. Priego‑Quesada 1,2,3* Over the last few years, portable Near‑Infrared Spectroscopy (NIRS) technology has been suggested for determining metabolic/ventilator thresholds. This systematic review and meta‑analysis aimed to assess the reliability of a portable muscle oxygenation monitor for determining thresholds during exercise testing. The proposed PICO question was: Is the exercise intensity of muscle oxygenation thresholds, using portable NIRS, reliable compared with lactate and ventilatory thresholds for exercise intensity determined in athletes? A search of Pubmed, Scopus and Web of Science was undertaken and the review was conducted following PRISMA guidelines. Fifteen articles were included. The domains which presented the highest biases were confounders (93% with moderate or high risk) and participant selection (100% with moderate or high risk). The intra‑class correlation coefficient between exercise intensity of the first ventilatory or lactate threshold and the first muscle oxygenation threshold was 0.53 (obtained with data from only 3 studies), whereas the second threshold was 0.80. The present work shows that although a portable muscle oxygenation monitor has moderate to good reliability for determining the second ventilatory and lactate thresholds, further research is necessary to investigate the mathematical methods of detection, the capacity to detect the first threshold, the detection in multiple regions, and the effect of sex, performance level and adipose tissue in determining thresholds. In many sports, various methods of exercise testing are performed for detecting metabolic/ventilatory thresh- olds. These zones or points are characterized by nonlinear increases of physiological outcomes (e.g., dot(V), oxygen volume (VO2), blood lactate, heart rate, etc.) so determining two physiological breakpoints that allow the three-phase model of intensities to be applied1–3. These data are important to trainers and athletes for assessing physical condition and programming intensities to optimize training and improving cardiovascular fitness and endurance4,5. Therefore, it is of great importance to have a reliable method for threshold detection6. The ventilatory or metabolic threshold is usually determined by gas exchange or blood lactate data respec- tively, obtained during incremental tests4,7. Gas exchange is one of the most commonly used methods for assess- ing the evolution of gas exchange measurements (dot(V), VO2, carbon dioxide volume (VCO2) and minute ventilation (VE)) that allow detection of the respiratory compensation point (also referred to as ventilatory OPEN 1Research Group in Sports Biomechanics (GIBD), Department of Physical Education and Sports, Faculty of Physical Activity and Sport Sciences, Universitat de València, C/Gascó Oliag, 3, 46010 Valencia, Spain. 2Red Española de Investigación del Rendimiento Deportivo en Ciclismo y Mujer (REDICYM), Consejo Superior de Deportes (CSD), Facultad de Ciencias de la Actividad Física y del Deporte, Campus d’Ontinyent, Laboratorio Biomecánica, Avda. Conde de Torrefiel n° 22, 46870 Ontinyent, Spain. 3Biophysics and Medical Physics Group, Department of Physiology, Universitat de València, Faculty of Medicine and Odontology, Avd. Blasco Ibañez 15, 46010 Valencia, Spain. *email: j.ignacio.priego@uv.es 2 Vol:.(1234567890) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ threshold (VT))8. For example, one method that is often used is the ventilatory method consists of determining the first and second ventilatory thresholds by detecting nonlinear increases in minute ventilation, the ventila- tory equivalent for oxygen, the ventilatory equivalent for carbon dioxide, oxygen uptake, and carbon dioxide production9. Another widely used method is the blood lactate measurement10. In contemporary physiology, lactate is considered a major metabolic intermediate that has a wide-ranging impact on energy substrate utiliza- tion, cell signaling, and adaptation11. It is also important for the mitochondria since lactate is the end product of glycolysis and plays a role in connecting oxygen-independent and oxygen-dependent energy production, as a major energy source for mitochondrial respiration4,11. Hence, lactate enters the mitochondrial reticulum to support cell energy homeostasis by oxidative phosphorylation, and this process helps lactate disposal11. Threshold determination using blood lactate concentration can be obtained from values fixed (e.g., 2 or 4 mmol  L−1)12 to mathematical models13,14. However, both methods have associated limitations such as the economic cost of gas exchange, and the neces- sity to extract a drop of blood or its incapacity to measure continuously for lactate15, all of which makes it inter- esting to explore new methodologies. Moreover, it has been suggested that determining thresholds using muscle oxygen saturation (SmO2) could be a valid alternative to pulmonary gas exchange or blood lactate methods16,17. Muscle oxygenation based on Near-Infrared Spectroscopy (NIRS) is a non-invasive technology that was described for the first time by Jöbsis in 1977, for monitoring in vivo cerebral oxygenation18. Nowadays, it is becoming very popular in the sports training field, thanks to the appearance of more affordable, easy to apply, and portable measuring devices19,20. Currently, NIRS technology is based on the modified Beer-Lambert’s law, which considers the dispersion of the nature of the tissues and their geometry21,22 (Eq. 1). NIRS technology detects the oxyhemoglobin ([O2Hb]) or deoxyhemoglobin ([HHb]) depending on light absorption, but in both cases, hemo- globin or myoglobin are referenced, since NIRS technology does not differentiate between chromophores (Eq. 2). Modified Beer-Lambert’s law Eq. (1), where “A” is the absorption, “I” is the luminous intensity (lm sr−1), “ ε ” is the extinction coefficient for the light absorbing compound of interest, “[C]” is the concentration of the compound of interest (e.g. [Hb], [Mb] and/or [cytox]), “L” is the source-detector distance (mm), “DPF” the dif- ferential path length factor and “G” is the factor reflecting non-absorption. Equation for calculating muscle oxygen saturation (SmO2) by the oxyhemoglobin (O2Hb) and deoxyhemo- globin (HHb) measured. NIRS technology in the sports field is being used to observe changes in the muscle metabolism of different muscles19. This has allowed us to measure local muscle performance during exercise, determining whether the muscles work optimally and if there is deoxygenation depending on exercise intensity20,23,24. Moreover, although several studies have suggested that portable NIRS technology can be used for determining muscle oxygenation thresholds17,25,26, and many studies have been published over the last few years, as far as the author knows, no systematic reviews and meta-analyses that validate the use of NIRS technology to detect thresholds have been undertaken. Therefore, the aim of this systematic review and meta-analysis was to evaluate the reliability of determining the exercise intensity of the muscle oxygenation threshold (using the portable NIRS) compared with detection, using a gold standard method during laboratory and field tests. Methods Literature search methodology. This systematic review and meta-analysis was carried out following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement27. The proposed PICO (Population, Intervention, Comparison and Outcomes of an article) question was: Is the exercise intensity of muscle oxygenation thresholds, using portable NIRS, reliable compared with lactate and ventilatory thresh- olds for exercise intensity determined in athletes? Three databases (PubMed, Scopus and Web of Science) were electronically searched on the 15th of June of 2023 using the following terms: “NIRS” OR “Near Infrared Spec- troscopy” OR “muscle oxygenation” OR “oximetry” AND with the terms and synonyms “threshold” OR “break- point” OR “inflection point”. Additionally, (AND) different terms such as “exercise” OR “sport” OR “physical activity” OR “running” OR “cycling” OR “swimming” were used. Every database employed its own term map- ping. The results were screened to identify relevant studies, first by abstract and finally by full text. Full texts underwent a thorough screening process to determine their eligibility for inclusion in the review. Only those texts that fulfilled all the predetermined criteria were considered for inclusion. The articles obtained were exported to Zotero (version 6.0.15, Corporation for Digital Scholarship, Vienna, USA) to eliminate duplicates, and the abstracts were uploaded to JBI SUMARI (The University of Adelaide, Adelaide, Australia) to carry out the first screening. Inclusion and exclusion criteria. The inclusion criteria established for the systematic review were as fol- lows: (1) Only studies written in English, Spanish or Portuguese, (2) studies using a portable and commercial NIRS for muscle oxygenation threshold detection, (3) studies using a gold standard (gas exchange or blood lac- tate methods) in addition to muscle oxygenation for thresholds detection, (4) studies with a healthy population between 18 and 65 years of age, and (5) experimental and quasi-experimental studies. (1) A = log I IO = ε[C]L ∗ DPF + G (2) SmO2 = O2Hb O2Hb + HHb × 100 3 Vol.:(0123456789) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ Study selection and data extraction. The first screening was performed by reviewing the abstracts of articles, after removing duplicates. Then, the selected articles were fully read to reach a decision. The entire process was carried out by two reviewers. When there was a disagreement on an abstract or article, it was sub- sequently discussed until a consensus was reached. For each study, the extracted data were: the authors and the year, the participants, a short description of the protocol, the thresholds calculated, the NIRS brand, the NIRS location, and the results. The data from each included article were extracted by two reviewers and confirmed by a third. Participants were categorized as elite, highly trained, trained and recreationally active following previous guidelines28,29. Risk of bias and quality of evidence assessment. The quality of the quasi-experimental studies included in the systematic review was assessed by two reviewers working independently using the ROBINS-I Scale. The ROBINS-I Scale evaluates risk of bias across 7 domains: confounding, selection of participants, clas- sification of interventions, deviations from intended interventions, missing data, measurement of outcomes and the selection of the reported results30. For each domain, the risk of bias assessment was categorized: no informa- tion, critical, serious, moderate or low30. When there was a disagreement between the reviewers a third reviewer was consulted. Meta‑analysis. A separate meta-analysis was performed to examine the reliability in determining intensity at each threshold using NIRS and the gold standard method (gas exchange and/or blood lactate). The intraclass correlation coefficient (ICC) and sample size were extracted for each study. For the studies that did not provide ICC values, the ICC value was calculated from obtaining the data from the datasets, tables and figures of the article, or on request from the authors. In the case of figures, data was extracted from scatter plots using the plot digitizer application31. If the data were not provided by the authors, the study was excluded from the analysis. ICC values were calculated based on a single rater-measurement, absolute-agreement, and 2-way random-effects model. For studies where it was possible to obtain more than one ICC value (e.g., because the intensity at the threshold was extracted using different automatic methods), these ICC values were averaged, using only one ICC value for each study to avoid statistical dependence31,32. ICC values were transformed to Fisher’s z scale and a random-effects model with Restricted Maximum Likelihood Estimation was used for the analysis33, assessing the type of gold standard compared (gas exchange or blood lactate) as a possible moderator. Q and I2 statistics were used for the homogeneity analysis. I2 values of around 25%, 50%, and 75% denoted low, moderate, and large heterogeneity, respectively. To assess the publication bias, funnel plot with Duval and Tweedie’s trim-and- fill method for imputing missing data and the Egger’s test were performed34,35. To facilitate the interpretation of the data, Fisher’s z values were then converted back to ICC values after completing the meta-analyses33. The ICC and associated 95% confidence intervals were interpreted as: poor (0.00–0.25), fair (0.26–0.50), moderate (0.51–0.75) and good (0.76–1.00)36. Statistical significance was established at p < 0.05. A meta-analysis was per- formed with the “metafor” package (version 4.2-0)37 in RSTUDIO (version 2023.06.0)38. Results Study selection. A total of 1,131 articles from databases of PubMed (237), Web of Science (507), and Sco- pus (387) were included, and 559 articles remained after removing duplicates. Finally, after selecting studies by their abstracts, 129 full articles were reviewed, of which 15 were included in the systematic review (Fig. 1). Participants characteristics. The systematic review included a sample of 344 participants (216 males and 128 females). Among these participants 33 were elite athletes, 208 highly trained athletes, 31 trained athletes and 72 recreationally active athletes. Moreover, athletes from various sports were included (soccer, cycling, running, triathlon and rowing) with laboratory protocols, since there are currently no studies carried out in field tests. The study characteristics and the main findings are summarized in Table 1. Methods used for determining muscle oxygenation threshold. The studies selected had deter- mined both muscle oxygenation threshold (MOT) (first and second) using different methods (Table 2). Most of the studies used the regression double linear representing 42% and wearable lactate threshold (WLT) was used in 25% of the studies included in the systematic review. Together, these two methods represented 67% of the studies included in the systematic review. However, visual identification was also used in two studies (17%). Risk of bias evaluation. The domains which presented the highest bias were due to confounding (7% with critical risk, 33% with serious risk and 53% with moderate risk), due to the selection of the participants (20% with serious risk and 80% with moderate risk), and due to the selection of the reported results (40% with moder- ate risk) (Figs. 2 and 3). For the other domains, most of the studies presented a low risk of bias (> 85%). Meta‑analyses. Of the 15 articles included in this review, the ICCs of 13 of them were obtained from the meta-analysis (Table 3). Of these 13 articles, the ICC was provided in the article itself in 3, was calculated from the data obtained in a dataset, table or figure in 8, and in 2 the ICC was provided directly by the authors (Table 3). A test of moderators was not performed for the first threshold due to the low number of studies (n = 3, Table 3). The Q test was not significant (Q(df = 2) = 1.01, p-val = 0.60) and the I2 was 0%, showing a low hetero- geneity. The Trim-and-fill method estimated 0 missing studies and Egger’s test was not significant (p = 0.46). The ICC of the first threshold was moderate (ICC = 0.53) but with a wide 95%CI[0.31, 0.69] (Fig. 4A). 4 Vol:.(1234567890) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ For the second threshold, no effect of moderators was observed (p = 0.94) at first. Therefore, a meta-analysis was performed without differentiating between the ICC obtained compared with lactate or gas exchange. The Q test was not significant (Q(df = 13) = 99.17, p < 0.001) and the I2 was of 86%, showing a large heterogeneity. The Trim-and-fill method estimated 0 missing studies and Egger’s test was not significant (p = 0.54). The ICC of the second threshold was good (ICC = 0.80, 95%CI[0.65, 0.89] (Fig. 4B). Discussion The aim of this systematic review and meta-analysis was to evaluate the reliability of determining exercise inten- sity using the muscle oxygenation threshold (with the portable NIRS) compared with a gold standard detection method during laboratory tests. The results of the review show that the methods mostly used to determine muscle oxygenation thresholds were regression double linear (46%), WLT (20%), and visual identification (20%). The meta-analysis revealed that of the 13 studies where ICC was obtained, only 3 studies assessed the first threshold, the mean ICC of 0.53 being observed between the exercise intensity obtained at the first muscle oxygenation threshold (MOT1) and first lactate threshold (LT1) or first ventilatory threshold (VT1). The mean ICC between second muscle oxygenation threshold (MOT2) and second lactate threshold (LT2) or second ventilatory threshold (VT2) was 0.80. Our meta-analyses were focused on showing whether the exercise intensity where the first and second thresh- olds were detected using the portable NIRS was more reliable than the gold standards methods (gas exchange and blood lactate). Table 1 shows how the relationship between MOT and VT was analyzed in 7 studies16,25,39–43 and in 9 studies for LT17,26,41,44–49. Figure 1. Study selection from the systematic review and meta-analysis (PRISMA). 5 Vol.:(0123456789) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ Study (year) Participantsa Protocol Thresholds NIRS device NIRS location Results/conclusions Batterson et al.44 N = 10 (M) Elite Soccer players GXT 3-min work + 30 s rest 9.0 km·h−1 W-UP ↑ 1.8 km·h−1 every 180 s *Treadmill LT2, LT1 Moxy VL, GC, BF MOT1 and MOT2 showed similar that LT1 and LT2 in all muscles analyzed. This show that SmO2 is useful for coaches Borges and Driller45 N = 7 (M); 7 (F) Highly trained athletes GXT 3-min W-UP (4.8 km·h−1) 3-min (9.3–11.7 km·h−1) ↑ 0.3–1.1 km·h−1 every 180 s *Treadmill LT2 BSX Insight GC MOT2 showed a high cor- relation. The wearable lactate threshold sensor could be implemented by coaches and athletes Cayot et al.46 N = 9 (M); 5 (F) Recreationally 2 × GXT (separated by 7–10 days) 5-min W-UP (25 W) ↑ 25 W every 180 s *Cycle ergometer LT2 Moxy VL MOT2 detection was moderately correlated with the LT2 and the heart rate. The results do not support the use of two different mathematical methods for MOT2 determination Contreras-Briceño et al.39 N = 8 (M); 7 (F) Highly trained triathletes GXT 2-min rest 3-min W-UP (100 W) ↑ 20 W every 80 s *Bike on a cycle ergometer VT2 Moxy 7th IC A good-to-excellent correlation was obtained between MOT2 and VT2 for each variable of all analyses in the 7th IC, muscle Driller et al.47 N = 10 (M); 5 (F) Highly Trained Cyclists GXT 3-min 80–120 W ↑ 20 W every 180 s *Bicycle on an ergometer LT2 BSX Insight GC LT2 determination through MOT2 showed an excellent correlation during cycling. These results were shown in all methods for LT2 detection Farzam et al.48 N = 15 (M); 3 (F) Recreationally GXT 4-min 30 W ↑30 W every 240 s *Cycle ergometer LT2 Humon Hex, MetaOX* RF MOT2 determination showed good agreements with LT2 NIRS portable and NIRS non-portable showed a good correlation during the exercise. A low-cost, wireless, wearable NIRS is a good predictor of the threshold Feldmann et al.16 N = 6 (M); 4 (F) Recreationally cyclists and runners GXT Run test: 5-min W-UP (3.0–3.5 km·h−1) ↑0.5 km·h−1 every 30 s Cycling test: 5-min W-UP (50–100W) ↑25W every 25 s *Treadmill *Cycle ergometer VT1, VT2 Moxy VL NIRS technology is suitable for determining VT1 and VT2. Additionally, SmO2min is a good indicator of cardiorespiratory fitness, as it correlated with VO2peak. Furthermore, no matter in which lateral vastus (right or left) the NIRS device was placed and the modality (cycling or running) it detected the MOT correctly McMorries et al.49 N = 7 (M); 14 (F) Trained Triathletes GXT ↑ 12–18 s per km every 180 s *Treadmill LT2 BSX Insight GC MOT2 showed similar values to LT2, when the thresholds were compareted using the heart rate Raleigh et al.41 N = 31 (M) Highly trained cyclist/triathletes GXT 15-min W-UP (120 W) 3-min (100 W) ↑ 25 W every 180 s *Cycle Ergometer LT2, VT2 Moxy VL MOT2, LT2 and VT2 were not different, but a poor correlation was obtained between them A good correlation was identi- fied between VT1 and LT1 Rodrigo-Carranza et al.25 N = 5 (M); 5 (F) Highly trained runners GXT 5-min W-Up (9 km·h−1) ↑ 1 km·h−1 every 60 s *Treadmill VT2 Humon Hex VL VT2 and MOT2 were positively correlated during running. Thus, the device presented a good predictor of the second threshold Osmani et al.40 N = 16 (M); 5 (F) Recreationally GXT 3-min work + 30 s rest 8.0 km  h−1 W-UP ↑ 1.2 km  h−1 every 180 s *Treadmill VT2 Humon Hex VL SmO2 data alone were not enough to determine the VT2 Also, SmO2 values of this device (Humon) do not correlate with other variables (blood lactate, RPE, HR and running power) Salas-Montoro et al.17 N = 32 (M); 58 (F) 23 Elite 67 Highly trained Cyclists GXT 5-min W-Up (15–20% of FTP) ↑ 25 W every 60 s *Cycle ergometer LT2 Humon Hex RF LT2 was excellently correlated with MOT2 when compared using power output, percentage of maximal aerobic power, heart rate and percentage of maxi- mum heart rate to MOT. The reliability of methods showed very good or excellent values in all cases (0.74–0.99) NIRS portable device can be an interesting tool for threshold detection for coaches without performing an on-site lactate test Continued 6 Vol:.(1234567890) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ The studies of Feldmann et al.16 and Van der Zwaard et al.42 compared the VT1 and LT1 with MOT1 in cycling and found ICC values (ICC = 0.56–0.65). These results are in line with other studies that determined thresh- olds with non-portable NIRS in cycling50. Moreover, a fair ICC in running was shown (ICC = 0.23–0.49)16,44. 28/07/2023 17:06:00 A lower number of studies assessed the first threshold compared with the second one (3 vs. 12 studies), maybe due to the difficulty of determining the MOT1, since the slope changes very slightly and the ICC value is not as good as the second threshold42. The second threshold was determined using the blood lactate concentration and muscle oxygenation in different sports such as cycling16,17,46–48, running44,45,49 and rowing26. ICC values showed a certain disparity and were fair, moderate or good (ICC = 0.29–0.90) in studies of running, although cycling studies showed a good ICC (ICC = 0.91–0.94). However, the ICC value of two studies were not obtained46,48. The remaining studies also Table 1. Summary of selected studies. W: watts; M: male; F: female; GXT: graded exercise test; W-UP: warm up; R: recovery; LT1: first lactate threshold; LT2: second lactate threshold; VT1: first ventilatory threshold; VT2: second ventilatory threshold; BF: biceps femoris; LD: lateral deltoid; IC: intercostal; RF: rectus femoris; VL: vastus lateralis; GC: gastrocnemius; RCP: respiratory compensation point; MOT1: first muscle oxygenation threshold; MOT2: second muscle oxygenation threshold. a Data are expressed as mean ± standard deviation. *Non-portable NIRS. Study (year) Participantsa Protocol Thresholds NIRS device NIRS location Results/conclusions Turnes et al.26 N = 13 (M) Highly trained rowers (1) GXT 3-min (130 W) ↑30 W every 180 s//R 30″ (2) 10-min W-Up + 5-rest 2000 m test *Rowing ergometer LT2 Portamon VL LT2 was moderately related to MOT2 during the rowing incre- mental test. However, the SmO2 in the VL presented a large vari- ability between participants Van der Zwaard et al.42 N = 30 (M); 10 (F) 9 Recreationally 10 Trained 21 Highly trained Cyclist and endurance trained GXT 3-min 1.5 W·kg−1 (85–145 W) ↑ 0.5 W·kg−1 (30–50 W) every 180 s *Cycle ergometer VT1, VT2 Portamon VL VT1 and VT2 were moderately related to MOT. The relation- ship increased in trained cyclists (0.68–0.84) compared with recreationally trained males (0.48–0.50) VT differed across sexes and training status, whereas MOT differed only across sexes Yogev et al.43 N = 17 (M); 5 (F) Highly trained Cyclist GXT 6-min W-Up (110–140 W) 4-min (70–100 W) ↑1 W every 2 s *Stationary bicycle trainer VT2 Moxy LD, VL VT2 and MOT2 showed a moderate relationship in both muscles The athletes and trainers could use portable NIRS to detect MOT Table 2. Methods for determining the muscle oxygenation threshold in the studies selected. MOT1: first muscle oxygenation threshold, MOT2: second muscle oxygenation threshold. a Also visually checked. b Inflection point at SmO2 values at the same point as the VT2. Methods for determining the MOT N (%) Threshold Studies Regression double linear 7 46 MOT1 & MOT2 16,26,39,41–44a Wearable lactate threshold (WLT) 3 20 MOT2 45,47,49 Visual identification (decrease of more than 15%) 3 20 MOT2 17,25,40b Application Humon Beta 2 7 MOT2 48 D-max or modified D-max 1 7 MOT2 46 Figure 2. Risk of bias summary. Created with ‘robvis’ application54. 7 Vol.:(0123456789) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ compared gas exchange with muscle oxygenation in the second threshold in cycling16,39,42,43 and running16,40. The results of the different studies suggest that the relationship between both methods in threshold determination is affected by the region assessed by the NIRS device, as good values (ICC = 0.92–0.97) were observed on assessing the intercostalis during cycling39. Moreover, the vastus lateralis presented moderate or good ICC in different investigations25,42, so the test or determination method chosen may also be critical. Different methods were developed to determine the thresholds in blood lactate concentration and gas exchange, which are commonly combined by users to find the most optimal inflection point51. Despite recent research into the application of NIRS technology for the purpose of obtaining thresholds, there is a lack of research on its methods of determination. The articles included in this systematic review use different methods for determining thresholds: BSX Insight (20%, N = 3)45,47,49, double linear regression (46%, N = 7)16,26,39,41–44, visual method17,25,40, Dmax or modified Dmax46 and applications of devices Humon Beta48. BSX Insight, which determines the threshold by making a comparison with blood lactate concentration, presented good values of ICC, although this used a patented method to determine MOT based on the inflection point of SmO2 during incremental testing45. However, as this system is commercial and patented, specific details of the algorithm used for said detection are unknown. Another important method is visual, which could be the most accurate for detecting the thresholds17 but with associated human error, or complementary to the previous one as was performed by Turnes et al.26 We recommend that future studies explore different methods to analyze thresholds using NIRS technology, to provide evidence on which are optimal, if several should be combined, or if some are more suitable for certain populations or sports. Figure 3. The risk of bias for each study. Created with ‘robvis’ application54. 8 Vol:.(1234567890) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ The muscles analyzed with NIRS portable had previously been studied by Perrey & Ferrari19, who showed that SmO2 was determined among different muscles (vastus lateralis, gastrocnemius medialis, intercostal, triceps brachii) and many sports (swimming, strength, skiing, speed skating, sailing, running, rugby, climbing, handball, cycling, kayak, judo, rowing, football, alpine skiing). Vastus lateralis was the muscle most assessed16,25,26,40–44,46, although other muscles such as gastrocnemius44,45,47,49, rectus femoris17, biceps femoris44, lateral deltoid43 or intercostal39 were also evaluated. Moreover, the muscles analyzed in each study depend on the sports performed in the testing, the main muscles involved in that activity being selected. For example, in cycling the muscle most assessed was the vastus lateralis as it is the main muscle contributing to power output production. However, some studies explored other regions during cycling which could affect the determination of the threshold17,47, although the rectus femoris is also a power output producer in this area where there could be a higher proportion of adipose tissue52 or because its neuromuscular activation is not affected by the increase in workload during the test (e.g., gastrocnemius)53. The systematic review also focused on exercise testing to determine whether the thresholds in the mus- cles (local thresholds) were analyzed or whether they are major exercise muscle. The articles included in this Table 3. The intraclass correlations (ICC) for the exercise intensity of muscle oxygenation threshold and the gold standard. ICC values used for the meta-analysis are in bold letters. MOT1: first muscle oxygenation threshold; MOT2: second muscle oxygenation threshold; LT1: first lactate threshold; LT2: second lactate threshold; VT1: first ventilatory threshold; VT2: second ventilatory threshold; mDmax: modified Dmax; VL: vastus lateralis; LD: lateral deltoid. Values used for the meta-analyses are in bold/italic. Study MOT method Gold standard method ICC source ICC Batterson et al.44 Segmented linear regression model LT1 and LT2 was determined using a mDmax Provided by the authors LT1 right VL: 0.38 LT1 left VL: 0.08 LT1 ICCmean: 0.23 LT2 right VL: 0.54 LT2 right VL: 0.60 LT2 ICCmean: 0.57 Borges and Driller45 Wearable lactate threshold sensor (WLT) LT2 was determined using the follow- ing methods: LSF, Dmax, mDmax, 4 mmol·L−1 and an increase greater than 1 mmol·L−1 Article LSF: 0.91 Dmax: 0.8 mDmax: 0.89 4mmoL: 0.98 1mmoL: 0.92 ICCmean: 0.90 Contreras-Briceño et al.39 Segmented linear regression model VT2 was determined with the visual method by two blinded researchers Calculated from data obtained from the Fig. 4 of the article ICC: 0.97 Cayot et al.46 Dmax and modified Dmax LT2 was determined using a Dmax and mDmax Authors did not provide the dataset after requestion – Driller et al.47 Wearable lactate threshold sensor (WLT) LT2 was determined using: TradLT, Dmax, mDmax and OBLA Calculated from data obtained from the Fig. 2 of the article TradLT: 0.96 Dmax: 0.88 mDmax: 0.97 OBLA: 0.96 ICCmean: 0.94 Farzam et al.48 Application Humon Beta LT2 was determined using the value of 4 mmol·L−1 lactate Authors did not provide the dataset after requestion – Feldmann et al.16 Segmented linear regression model VT1 and VT2 were detected with a seg- mented regression analysis Provided by the authors LT1 running: 0.49 LT1 cycling: 0.65 ICCmean: 0.57 LT2 running: 0.92 LT2 cycling: 0.92 ICCmean: 0.92 McMorries et al.49 Wearable lactate threshold sensor (WLT) LT2 was determined using the value of 4 mmol·L−1 lactate and an increase greater than 1 mmol·L−1 Calculated from data obtained from the Figure 6 of the article ICC: 0.29 Osmani et al.40 Visual identification VT2 was determined observing an inflec- tion point Calculated from data obtained from the Tables 1 and 2 of the article ICC: 0.23 Raleigh et al.41 Segmented linear regression model VT2 and LT2 were detected with a seg- mented regression analysis. The intersec- tion of two linear segments was defined as the threshold Article LT2: 0.54 VT2: 0.36 Rodrigo-Carranza et al.25 Visual identification VT2 was identified by the nonlinear increase Calculated from data obtained from the Table 1 of the article ICC: 0.84 Salas-Montoro et al.17 Visual identification LT2 was determined in an increase of at least 2 mmol·L−1 above baseline measure- ments Article ICC: 0.91 Turnes et al.26 Regression double linear and a visual identification LT was determined by linear interpola- tion given a fixed concentration of 3.5 mmol·L−1 Calculated from data obtained from the Table 2 of the article ICC: 0.65 Yogev et al.43 Regression double linear Regression double linear was used to detect the threshold with WKO5. This is similar to the V-slope method Calculated from dataset provided by the authors VT2 VL: 0.73 VT2 LD: 0.79 ICCmean: 0.76 Van der Zwaard et al.42 Intercept of two congregating regression lines VT detection method was the same as in MOT detection Calculated from the dataset (supporting files) of the study ICC VT1: 0.56 ICC VT2: 0.38 9 Vol.:(0123456789) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ systematic review analyzed 1 or 3 muscles at most at the same time. Moreover, most of these studies were focused on correlating the main muscles of exercise with blood lactate concentration or gas exchange, and it is impor- tant to take into account that lactate and gas exchange determine systemic changes, while NIRS technology can be used for determining a more local response. For this reason, further studies that analyze different muscles simultaneously would be interesting in order to understand what is happening in each muscle during exercise testing, and how some may be more related to systemic changes while others have more specific alterations. It is important to consider that the present meta-analysis is limited to only one measure of reliability (ICC), and more statistics are desirable (e.g., bias between methods) to improve the interpretation and application of the present results. Bias was not included due to the low number of studies that reported this data, and the dif- ferent units used (W, km·h−1or percentage) also posed a challenge. This point should be regarded as a limitation of the present work, and future meta-analysis with a higher number of studies should incorporate more reliable statistics. Some of the articles included in this review demonstrate mean bias between MOT2 and LT2 or VT2 ranging from 0.01 and 0.4 km·h−125,44,45,49, between 3.9 and 15.4 W39,41, 0.05 W·kg−117 and 10.7% of the power output26. However, Batterson et al.44 showed a higher mean bias for MOT and LT1 (1.1–1.2 km·h−1), and Driller et al.47 also demonstrated how the method of determination could affect the bias, with the lowest being for the Dmax method (17 W) and the highest for the OBLA method (37 W). Finally, the study of Feldmann et al.16 stated that in terms of power or speed, the bias represents one performance step (for this particular study, it was 25 W for cycling and 0.5 km·h−1 for running). Although the studies included present low risk of bias in most of the domains assessed, the analysis performed suggests that two domains presented a considerable risk of bias: confounders and the selection of the partici- pants. The main issues related to the confounding domain were the studies that did not consider the effect of Figure 4. Forest plots of the meta-analysis was performed for the intraclass correlation (ICC) of the exercise intensity obtained at the first (A) and second (B) threshold determination using NIRS and the gold standard (gas exchange or blood lactate). 10 Vol:.(1234567890) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ the training level of participants, prior activity or sex in their results. In some cases, only the value of correlation or intraclass correlation coefficient without the confidence interval appear in the reported results. However, the majority of studies had a missing data count bias and bias in measurement outcomes. Future studies should take into account these aspects, so as to control them as much as possible, to improve their quality and reduce their biases. Moreover, these aspects are possible sources of the high heterogeneity found in the meta-analysis. The main limitation of the present work is the small number of studies included in the meta-analysis (N = 13). In future, a higher number of studies incorporated into the current analysis could corroborate the results obtained. Moreover, there was a high heterogeneity between the different studies included. Regarding the methodology, the regions or the sample assessed, with participants ranging from national and international level competitors17 to recreational ones42, could affect the results of the metanalysis. Considering all the analyses carried out, we think that the following lines of research should be prioritized in this area: exploring which are the most appropriate mathematical detection methods depending on the sports or populations for NIRS, investigating whether it is possible to detect the first threshold, analyzing multiple regions at the same time to find out which ones are most related to systemic thresholds and which have a more specific behavior of the muscle itself, and understanding the differences in the detection of thresholds depending on sex, performance level, amount of adipose tissue or the changing of muscle length during exercise. Conclusion The present systematic review and meta-analysis shows that, although using a portable muscle oxygenation monitor has moderate to good reliability for determining the second threshold, further research is necessary to investigate the mathematical methods of detection, the capacity to detect the first threshold, detection in multiple regions, and the effect of sex, performance level and adipose tissue on threshold determination. Data availability The datasets used and/or analysed during the current study available from the corresponding author on reason- able request. Received: 5 May 2023; Accepted: 28 July 2023 References 1. Ribeiro, J. et al. Metabolic and ventilatory thresholds assessment in front crawl swimming. J. Sports Med. Phys. Fitness 55, 7 (2015). 2. Seiler, K. S. & Kjerland, G. Ø. Quantifying training intensity distribution in elite endurance athletes: Is there evidence for an “optimal” distribution?. Scand. J. Med. Sci. Sports 16, 49–56 (2006). 3. Stergiopoulos, D. C., Kounalakis, S. N., Miliotis, P. G. & Geladas, N. D. Second ventilatory threshold assessed by heart rate vari- ability in a multiple shuttle run test. Int. J. Sports Med. 42, 48–55 (2021). 4. Poole, D. C., Rossiter, H. B., Brooks, G. A. & Gladden, L. B. The anaerobic threshold: 50+ years of controversy. J. Physiol. 599, 737–767 (2021). 5. Skinner, J. S. & Mclellan, T. H. The transition from aerobic to anaerobic metabolism. Res. Q. Exerc. Sport 51, 234–248 (1980). 6. Halson, S. L. Monitoring training load to understand fatigue in athletes. Sports Med. 44, 139–147 (2014). 7. Caen, K. et al. Ramp vs. step tests: Valid alternatives to determine the maximal lactate steady-state intensity?. Eur. J. Appl. Physiol. 121, 1899–1907 (2021). 8. Caen, K., Bourgois, J. G., Stassijns, E. & Boone, J. A longitudinal study on the interchangeable use of whole-body and local exercise thresholds in cycling. Eur. J. Appl. Physiol. 122, 1657–1670 (2022). 9. Ferretti, G., Fagoni, N., Taboni, A., Vinetti, G. & di Prampero, P. E. A century of exercise physiology: Key concepts on coupling respiratory oxygen flow to muscle energy demand during exercise. Eur. J. Appl. Physiol. 122, 1317–1365 (2022). 10. Bentley, D. J., Newell, J. & Bishop, D. Incremental exercise test design and analysis. Sports Med. 37, 575–586 (2007). 11. Brooks, G. A. et al. Lactate in contemporary biology: A phoenix risen. J. Physiol. 600, 1229–1251 (2022). 12. Weltman, A. et al. Prediction of lactate threshold and fixed blood lactate concentrations from 3200-m running performance in male runners. Int. J. Sports Med. 08, 401–406 (1987). 13. Chalmers, S., Esterman, A., Eston, R. & Norton, K. Standardization of the Dmax method for calculating the second lactate thresh- old. Int. J. Sports Physiol. Perform. 10, 921–926 (2015). 14. Hofmann, P. & Tschakert, G. Intensity- and duration-based options to regulate endurance training. Front. Physiol. https:// doi. org/ 10. 3389/ fphys. 2017. 00337 (2017). 15. Iannetta, D., Qahtani, A., MattioniMaturana, F. & Murias, J. M. The near-infrared spectroscopy-derived deoxygenated haemoglobin breaking-point is a repeatable measure that demarcates exercise intensity domains. J. Sci. Med. Sport 20, 873–877 (2017). 16. Feldmann, A., Ammann, L., Gächter, F., Zibung, M. & Erlacher, D. Muscle oxygen saturation breakpoints reflect ventilatory thresholds in both cycling and running. J. Hum. Kinet. 83, 87–97 (2022). 17. Salas-Montoro, J.-A., Mateo-March, M., Sánchez-Muñoz, C. & Zabala, M. Determination of second lactate threshold using near- infrared spectroscopy in elite cyclists. Int. J. Sports Med. https:// doi. org/ 10. 1055/a- 1738- 0252 (2022). 18. Jöbsis, F. F. Noninvasive, infrared monitoring of cerebral and myocardial oxygen sufficiency and circulatory parameters. Science 198, 1264–1267 (1977). 19. Perrey, S. & Ferrari, M. Muscle oximetry in sports science: A systematic review. Sports Med. 48, 597–616 (2018). 20. Feldmann, A. M., Erlacher, D., Pfister, S. & Lehmann, R. Muscle oxygen dynamics in elite climbers during finger-hang tests at varying intensities. Sci. Rep. 10, 3040 (2020). 21. Barstow, T. J. Understanding near infrared spectroscopy and its application to skeletal muscle research. J. Appl. Physiol. 126, 1360–1376 (2019). 22. Rolfe, P. In vivo near-infrared spectroscopy. Annu. Rev. Biomed. Eng. 2, 715–754 (2000). 23. Seshadri, D. R. et al. Wearable sensors for monitoring the internal and external workload of the athlete. NPJ Digit. Med. 2, 71 (2019). 24. Marostegan, A. B. et al. Effects of different inspiratory muscle warm-up loads on mechanical, physiological and muscle oxygenation responses during high-intensity running and recovery. Sci. Rep. 12, 11223 (2022). 25. Rodrigo-Carranza, V., González-Mohíno, F., Turner, A. P., Rodriguez-Barbero, S. & González-Ravé, J. M. Using a portable near- infrared spectroscopy device to estimate the second ventilatory threshold. Int. J. Sports Med. 42, 905–910 (2021). 11 Vol.:(0123456789) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ 26. Turnes, T. et al. Association between deoxygenated hemoglobin breaking point, anaerobic threshold, and rowing performance. Int. J. Sports Physiol. Perform. 14, 1103–1109 (2019). 27. Page, M. J. et al. The PRISMA 2020 statement: An updated guideline for reporting systematic reviews. BMJ 372, n71 (2021). 28. McKay, A. K. A. et al. Defining training and performance caliber: A participant classification framework. Int. J. Sports Physiol. Perform. 17, 317–331 (2021). 29. Pauw, K. D. et al. Guidelines to classify subject groups in sport-science research. Int. J. Sports Physiol. Perform. 8, 111–122 (2013). 30. Thomson, H., Craig, P., Hilton-Boon, M., Campbell, M. & Katikireddi, S. V. Applying the ROBINS-I tool to natural experiments: An example from public health. Syst. Rev. 7, 15 (2018). 31. Drevon, D., Fursa, S. R. & Malcolm, A. L. Intercoder reliability and validity of webplotdigitizer in extracting graphed data. Behav. Modif. 41, 323–339 (2017). 32. Badenes-Ribera, L., Rubio-Aparicio, M., Sánchez-Meca, J., Fabris, M. A. & Longobardi, C. The association between muscle dys- morphia and eating disorder symptomatology: A systematic review and meta-analysis. J. Behav. Addict. 8, 351–371 (2019). 33. Botella, J., Suero, M. & Gambara, H. Psychometric inferences from a meta-analysis of reliability and internal consistency coef- ficients. Psychol. Methods 15, 386–397 (2010). 34. Duval, S. & Tweedie, R. Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta- analysis. Biometrics 56, 455–463 (2000). 35. Sterne, J. A. C. & Egger, M. Regression methods to detect publication and other bias in meta-analysis. In Publication Bias in Meta- Analysis 99–110 (Wiley, 2005). https:// doi. org/ 10. 1002/ 04708 70168. ch6. 36. Portney, L. G. & Watkins, M. P. Foundations of Clinical Research: Applications to Practice (Pearson/Prentice Hall, 2009). 37. Viechtbauer, W. Conducting meta-analyses in R with the metafor package. J. Stat. Softw. 36, 1–48 (2010). 38. R Core Team. R: A Language and Environment for Statistical Computing 2012. (R Foundation for Statistical Computing, 2022). 39. Contreras-Briceño, F. et al. Determination of the respiratory compensation point by detecting changes in intercostal muscles oxygenation by using near-infrared spectroscopy. Life 12, 444 (2022). 40. Osmani, F., Lago-Fuentes, C., Alemany-Iturriaga, J. & Barcala-Furelos, M. The relationship of muscle oxygen saturation analyzer with other monitoring and quantification tools in a maximal incremental treadmill test. Front. Physiol. https:// doi. org/ 10. 3389/ fphys. 2023. 11550 37 (2023). 41. Raleigh, C., Donne, B. & Fleming, N. Association between different non-invasively derived thresholds with lactate threshold during graded incremental exercise. Int. J. Exerc. Sci. 11, 391–403 (2018). 42. Van Der Zwaard, S. et al. Oxygenation threshold derived from near- Infrared spectroscopy: Reliability and its relationship with the first ventilatory threshold. PLoS ONE 11, e0162914 (2016). 43. Yogev, A. et al. Comparing the respiratory compensation point with muscle oxygen saturation in locomotor and non-locomotor muscles using wearable NIRS spectroscopy during whole-body exercise. Front. Physiol. 13, 818733 (2022). 44. Batterson, P. M., Kirby, B. S., Hasselmann, G. & Feldmann, A. Muscle oxygen saturation rates coincide with lactate-based exercise thresholds. Eur. J. Appl. Physiol. https:// doi. org/ 10. 1007/ s00421- 023- 05238-9 (2023). 45. Borges, N. R. & Driller, M. W. Wearable lactate threshold predicting device is valid and reliable in runners. J. Strength Cond. Res. 30, 2212–2218 (2016). 46. Cayot, T. E. et al. Estimating the lactate threshold using wireless near-infrared spectroscopy and threshold detection analyses. Int. J. Exerc. Sci. 14, 284–294 (2021). 47. Driller, M., Borges, N. & Plews, D. Evaluating a new wearable lactate threshold sensor in recreational to highly trained cyclists. Sports Eng. 19, 229–235 (2016). 48. Farzam, P., Starkweather, Z. & Franceschini, M. A. Validation of a novel wearable, wireless technology to estimate oxygen levels and lactate threshold power in the exercising muscle. Physiol. Rep. 6, e13664 (2018). 49. McMorries, R. M., Joubert, D. P., Jones, E. J. & Faries, M. D. A validation study of a noninvasive lactate threshold device. Int. J. Exerc. Sci. 12, 221–232 (2019). 50. Lin, C.-W., Huang, C.-F., Wang, J.-S., Fu, L.-L. & Mao, T.-Y. Detection of ventilatory thresholds using near-infrared spectroscopy with a polynomial regression model. Saudi J. Biol. Sci. 27, 1637–1642 (2020). 51. Jamnick, N. A., Pettitt, R. W., Granata, C., Pyne, D. B. & Bishop, D. J. An examination and critique of current methods to determine exercise intensity. Sports Med. 50, 1729–1756 (2020). 52. Niemeijer, V. M. et al. The influence of adipose tissue on spatially resolved near-infrared spectroscopy derived skeletal muscle oxygenation: The extent of the problem. Physiol. Meas. 38, 539–554 (2017). 53. Quesada, J. I. P., Bini, R. R., Diefenthaeler, F. & Carpes, F. P. Spectral properties of muscle activation during incremental cycling test. J. Sci. Cycl. 4, 7–13 (2015). 54. McGuinness, L. A. & Higgins, J. P. T. Risk-of-bias VISualization (robvis): An R package and Shiny web app for visualizing risk-of- bias assessments. Res. Synth. Methods 12, 55–61 (2021). Acknowledgements JM-F contribution was funded by a PhD fellowship (ref. FPU20/01060) from the Ministry of Universities of Spain. Author contributions C.S.P. and J.I.P.Q. had the conceptualization of the idea. All the authors contributed to the design of the study. C.S., J.S.J. and J.M.F. worked in the data curation. C.S.P. and J.I.P.Q. performed the statistical analysis and the data visualization. R.S.P., A.E.M. and J.I.P.Q. supervised the project. C.S.P. wrote the original draft of the manuscript, and all authors reviewed, edited, and agreed to the final version of the manuscript. Competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to J.I.P.-Q. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 12 Vol:.(1234567890) Scientific Reports | (2023) 13:12649 | https://doi.org/10.1038/s41598-023-39651-z www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2023

Reliability of threshold determination using portable muscle oxygenation monitors during exercise testing: a systematic review and meta-analysis.

08-04-2023

Sendra-Pérez, Carlos,Sanchez-Jimenez, Jose Luis,Marzano-Felisatti, Joaquín Martín,Encarnación-Martínez, Alberto,Salvador-Palmer, Rosario,Priego-Quesada, Jose I

eng

PMC6204994

SYMPOSIUM Understanding the Agility of Running Birds: Sensorimotor and Mechanical Factors in Avian Bipedal Locomotion Monica A. Daley1 Structure and Motion Lab, Royal Veterinary College, Hawkshead Lane, Hertfordshire, AL9 7TA, UK From the symposium “Sensory Feedback and Animal Locomotion: Perspectives from Biology and Biorobotics” presented at the annual meeting of the Society for Integrative and Comparative Biology, January 3–7, 2018 at San Francisco, California. 1E-mail: mdaley@rvc.ac.uk Synopsis Birds are a diverse and agile lineage of vertebrates that all use bipedal locomotion for at least part of their life. Thus birds provide a valuable opportunity to investigate how biomechanics and sensorimotor control are integrated for agile bipedal locomotion. This review summarizes recent work using terrain perturbations to reveal neuromechanical control strategies used by ground birds to achieve robust, stable, and agile running. Early experiments in running guinea fowl aimed to reveal the immediate intrinsic mechanical response to an unexpected drop (“pothole”) in terrain. When navigating the pothole, guinea fowl experience large changes in leg posture in the perturbed step, which correlates strongly with leg loading and perturbation recovery. Analysis of simple theoretical models of running has further confirmed the crucial role of swing-leg trajectory control for regulating foot contact timing and leg loading in uneven terrain. Coupling between body and leg dynamics results in an inherent trade-off in swing leg retraction rate for fall avoidance versus injury avoidance. Fast leg retraction minimizes injury risk, but slow leg retraction minimizes fall risk. Subsequent experiments have investigated how birds optimize their control strategies depending on the type of pertur- bation (pothole, step, obstacle), visibility of terrain, and with ample practice negotiating terrain features. Birds use several control strategies consistently across terrain contexts: (1) independent control of leg angular cycling and leg length actuation, which facilitates dynamic stability through simple control mechanisms, (2) feedforward regulation of leg cycling rate, which tunes foot-contact timing to maintain consistent leg loading in uneven terrain (minimizing fall and injury risks), (3) load-dependent muscle actuation, which rapidly adjusts stance push-off and stabilizes body me- chanical energy, and (4) multi-step recovery strategies that allow body dynamics to transiently vary while tightly reg- ulating leg loading to minimize risks of fall and injury. In future work, it will be interesting to investigate the learning and adaptation processes that allow animals to adjust neuromechanical control mechanisms over short and long timescales. Birds as an animal model for agile bipedal locomotion Birds are diverse and agile vertebrates capable of many combinations of aerial, terrestrial, and aquatic locomotion. Living birds vary in size from hum- mingbirds to ostriches, and exhibit diversity in the length and mass proportions of the wings and legs, reflecting adaptation for different locomotor ecolo- gies (Gatesy and Middleton 1997; Zeffer et al. 2003; Heers and Dial 2015). While wings and flight are a defining locomotor innovation of birds, many living bird species are impressive bipedal terrestrial athletes, and all birds use bipedal movement for at least some part of their lives (Abourachid and Ho¨fling 2012; Heers and Dial 2015). Birds inherited bipedalism and many hindlimb morphological features from theropod dinosaurs, an ancient lineage that first appeared around 230 million years ago (Gatesy and Middleton 1997). This diversity and bipedal legacy makes birds a valuable study system for investigating how morphology, biomechanics, and sensorimotor control are integrated for agile bipedal locomotion. Advance Access publication June 12, 2018  The Author(s) 2018. Published by Oxford University Press on behalf of the Society for Integrative and Comparative Biology. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/ by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Integrative and Comparative Biology Integrative and Comparative Biology, volume 58, number 5, pp. 884–893 doi:10.1093/icb/icy058 Society for Integrative and Comparative Biology What are the challenges in achieving agile bipedal locomotion? Legged locomotion is complex and dynamic, involv- ing abrupt foot-contact transitions and uncertainty due to variable terrain and sensorimotor errors. Animals must precisely control limb dynamics to move effectively over varied and uncertain terrain while avoiding falls, collisions, and injury (Daley 2016). It remains poorly understood how the sen- sory, neural, and mechanical components of control are integrated to achieve robust, stable, and agile locomotion. Here, robustness refers to how large a disturbance an animal can tolerate while still meeting the functional demands of the task, such as forward movement at an acceptable speed (Daley 2016). Disturbances can arise externally from the environ- ment, internally from sensorimotor noise (such as errors in motor commands), or from inaccurate sen- sory information (such as lack of visibility or conflict between sensory modalities). Stability quantifies how rapidly the system attenuates perturbations from steady-state locomotion, and agility refers to the ability to rapidly adjust locomotor dynamics to meet changing task demands (such as a rapid exten- sion of the leg to leap over an obstacle) (Daley 2016; Duperret et al. 2016). Locomotion must be robust, stable, and agile for effective locomotion in natural conditions. Avoiding slip, fall, and injury requires precise reg- ulation of foot-contact timing and leg-substrate in- teraction forces (Alexander 2002; Clark and Higham 2011; Birn-Jeffery and Daley 2012; Daley 2016). Yet, inherent uncertainty due to terrain variability, sen- sorimotor noise, and sensing errors mean that the system dynamics cannot be perfectly sensed or pre- dicted. Considering these challenges, the agility and robust stability of terrestrial animals is truly remark- able. Bipedal animals face the additional challenge that they have fewer legs to support the body com- pared to quadrupeds and other many-legged ani- mals. Quadrupeds can redistribute loads among the legs in response to perturbations—a strategy not available to a rapidly running biped. This likely makes the challenges for dynamic balance control, especially acute for bipedal animals. One inherent challenge of animal systems is sen- sorimotor delay that limits feedback response times (More et al. 2010; 2013). Sensorimotor delays in- clude delays from sensing, nerve transmission, syn- apses, muscle electromechanical coupling, and muscle force development (More et al. 2013). Delays necessitate the use of predictive feedforward control, because motor commands must be issued in advance of the required mechanical demands. Reactive feedback control is also crucial to modulate and update motor commands to correct for devia- tions between predicted and actual dynamics. Thus, sensorimotor delay necessitates that animals effec- tively integrate both predictive (feedforward) and re- active (feedback mediated) sensorimotor control mechanisms for effective locomotion (Rossignol et al. 2006; Ijspeert 2018). Nerve transmission delays increase with the ana- tomical distances of neural pathways. This physical constraint creates a direct link between neuroanat- omy and temporal scaling of control processes (Fig. 1). Considering this, delay has probably been Fig. 1. Schematic illustration of the hierarchical organization of vertebrate neuromechanical control. Transmission delays lead to a temporal scaling of sensorimotor processes that relate to the anatomical distances between sensors, neural networks and effectors. Consequently, central, peripheral and mechanical mechanisms must be integrated over short and long timescales. The fastest responses occur in the periphery, through intrinsic mechanics, intermediate responses occur through short-latency spinal reflexes, and slower responses involve processing and planning in higher brain centers. Agility of running birds 885 a selective factor in the evolution of a hierarchical organization of the nervous system. The fastest possible reactions occur locally, through intrinsic- mechanical responses to altered limb-substrate inter- actions (Brown and Loeb 2000; Daley and Biewener 2006; Daley et al. 2006; 2009). The shortest sensori- motor loops and fastest neural responses occur through monosynaptic spinal reflexes, and the lon- gest delays are associated with processing and pre- dictive planning in higher brain centers (Fig. 1) (Rossignol et al. 2006; Grillner et al. 2008; McCrea and Rybak 2008; McLean and Dougherty 2015; Kiehn 2016). This suggests the natural emergence of temporal scaling of sensorimotor control that relates to neuroanatomical organization. While the components of vertebrate sensorimotor systems are increasingly well understood, it remains unclear how these mechanisms are integrated over varying time- scales to achieve robust, stable, and agile locomotion in natural terrain contexts. The phrase “passive-dynamics” has often been used to refer to the intrinsic mechanical response of the locomotor system. However, this phrase can be somewhat misleading, because the intrinsic me- chanical response is actively tuned by the selection of a specific muscle activation pattern from the possible solutions that could meet the mechanical require- ments of the task (Brown and Loeb 2000). Each muscle activation solution will confer a unique set of characteristics in terms of muscle–tendon dynam- ics, impedance response, stability, robustness and sensitivity to perturbations, directional tuning, and energy cost (Brown and Loeb 2000; Inouye and Valero-Cuevas 2016; Valero-Cuevas 2016). Thus, in- trinsic mechanical responses are under some active control, because feedforward muscle activation pat- terns can be tuned through learning and experience to enable robust, stable, and agile performance. However, the processes and timescales of such tun- ing between intrinsic mechanics and muscle activa- tion patterns remain unclear. Terrain perturbation approaches help reveal neuromechanical control strategies Terrain perturbations are ubiquitous in nature and disrupt the predictability and timing of foot– substrate interactions, requiring transient locomotor responses to recover from disturbances. Understanding transient locomotor dynamics is important for reveal- ing natural locomotor behaviors, and for understand- ing the specific mechanical demands and constraints that have shaped animal locomotor control. Birds are particularly useful for such studies of transient locomotor dynamics, because it is possible to simultaneously measure in vivo muscle force– length dynamics, body dynamics, leg–substrate in- teraction forces, and joint mechanics during loco- motion (Daley and Biewener 2006; Daley et al. 2007, 2009). This facilitates integrated understand- ing of neuromechanical function. Considering the complex nature of neuromechanical control, it is useful to start by investigating the response to very simple terrain perturbation features, to mini- mize the number of confounding factors in the response. Initial terrain perturbation experiments in running guinea fowl aimed to reveal the immediate intrinsic mechanical response to an unexpected perturbation, in the time before a sensorimotor feedback response is possible (Daley and Biewener 2006; Daley et al. 2007, 2009). This work was inspired by earlier work in rapidly running cockroaches recovering from an impulsive perturbation (Jindrich and Full 2002) and studies of humans recovering from sud- den changes in terrain stiffness (Ferris et al. 1999; Moritz and Farley 2004). In the guinea fowl experi- ments, birds encountered a simple camouflaged pot- hole step, 8 cm deep (40% of leg length), covered by opaque tissue paper stretched across the gap (Daley et al. 2006). When navigating the unexpected drop, guinea fowl showed large changes in leg posture in the perturbed stance, which correlated strongly with leg loading and perturbation recovery. These findings highlighted the role of leg angular trajectory control for regulating foot contact timing and leg loading (Daley and Biewener 2006), which has also been found to be important in humans (Seyfarth et al. 2003; Daley and Biewener 2006). The dynamics fol- lowing a drop in terrain can be explained by the physics of a spring-loaded-inverted pendulum (SLIP) model with very simple swing leg control, in which the leg follows a sinusoidal, clock-like an- gular trajectory, retracting backwards toward the ground just before the swing-stance transition (Fig. 2; Seyfarth et al. 2003; Daley and Biewener 2006; Blum et al. 2014). In this model, contact angle depends on the duration of the ballistic flight time (Daley and Biewener 2006; Daley and Usherwood 2010; Blum et al. 2014). Although this is an ex- tremely simplistic model of running, it is sufficient to generate robustly stable gait dynamics (Seyfarth et al. 2003; Blum et al. 2014). Such model-based analyses of the coupling of body dynamics and leg angular cycling during run- ning over terrain perturbations have revealed an in- herent trade-off in leg control between terrain 886 M. A. Daley robustness and injury avoidance (Fig. 2; Daley and Usherwood 2010; Blum et al. 2014). Drops in terrain result in a delay of ground contact, longer fall time, and greater downward vertical velocity at contact. However, the specific dynamics of the response depends on how fast the leg is retracted during the falling phase, just before the swing-stance transition (Daley and Usherwood 2010). Fast leg retraction results in a large change in leg angle in response to a given change in terrain height, and earlier ground contact. This results in smaller fluctuations in verti- cal velocity and leg loading in response to terrain perturbations (Daley and Usherwood 2010; Blum et al. 2014). However, fast leg retraction also decreases the maximum terrain drop the animal can safely negotiate, a measure of robustness, and increases the risk that the leg will miss contact en- tirely, leading to a fall (Fig. 2B; Daley and Usherwood 2010; Blum et al. 2014). In contrast, slow leg retraction results in small changes in leg angle for a given terrain drop, ensuring foot contact even for large terrain perturbations, reducing risk of fall; however, this leads to larger increases in vertical velocity and leg loading in the stance following the drop, increasing risk of overload injury (Fig. 2C). Subsequent experiments have investigated how leg control strategies vary depending on the type of per- turbation (pothole, step, obstacle), and with ample practice negotiating visible terrain features. In com- paring locomotor control strategies between hidden and visible potholes, guinea fowl slow down in an- ticipation of visible potholes when they encounter them for the first time, and actually stumble more when negotiating the visible drop (Daley and Biewener 2006). Although the high-speed intrinsic- mechanical response to an unexpected drop is ro- bustly stable, birds may not always choose this strat- egy when they first encounter novel, visible terrain features, perhaps to minimize risk of injury. Blum and colleagues (2014) explored how animals manage the trade-off between terrain robustness and injury avoidance in leg angular control when given ample practice negotiating a visible drop in terrain. Under these conditions, guinea fowl converge upon a strategy similar to the hidden pothole strategy—they maintain high speeds and allow intrinsic leg mechan- ics to mediate the perturbation response (Blum et al. 2014). The authors compared the experimental measures to SLIP-model predictions with swing leg angular control optimized for disturbance rejection (robustness) versus load regulation (injury avoid- ance). The guinea fowl used a strategy that allowed body dynamics to transiently vary, with swing leg control optimized to maintain consistent leg loading in uneven terrain, which avoids both fall and injury conditions. Model analysis revealed that leg control optimized for disturbance rejection, to maintain Fig. 2. A trade-off in control of leg retraction rate for terrain robustness versus injury avoidance, illustrated by two boundary conditions. (A) Running dynamics modeled as a SLIP with the swing leg retracted toward the ground just before stance. Leg retraction rate influences the mechanical response in uneven terrain: (B) Fast leg retraction results in steeper leg contact angles and minimizes fluctuations in leg loading, but if leg loading angle (bTD) reaches 90-degrees, the leg will miss stance, risking a fall. Maximum terrain drop before a fall decreases with increasing rate of leg retraction. (C) Slow leg retraction ensures leg contact, minimizing fall risk, but incurs higher fluctuations in leg loading. Evidence suggests that birds optimize their leg retraction rate to minimize fluctuations in leg-loading (Blum et al. 2014), using in- termediate leg retraction rates that ensure contact while avoiding overload injury. Agility of running birds 887 steady body dynamics, demanded dramatic increases in leg loading, suggesting increased injury risk. This study also revealed that birds showed very little stride- to-stride variance in leg angular cycling rate in uneven terrain. In contrast, leg length actuation rapidly changed in response to altered leg posture and load- ing, resulting in rapid adjustment of stance push-off to stabilize body mechanical energy in the 1–2 steps following the perturbation (Blum et al. 2014). These studies have revealed optimization of leg angular cy- cling rate as an effective control strategy for locomo- tion in uneven terrain, allowing maintenance of consistent leg loading and high running speeds (Seyfarth et al. 2003; Daley and Biewener 2006; Daley and Usherwood 2010; Blum et al. 2011, 2014). In another series of experiments, Birn-Jeffery and colleagues investigated control strategies used by ground birds when negotiating visible obstacles, to investigate potential trade-offs in stance leg function (Birn-Jeffery and Daley 2012; Birn-Jeffery et al. 2014). Similar to the studies on terrain drops, the birds exhibited independent control of leg angular cycling and leg length trajectory, with higher stride- to-stride variance in leg length in uneven terrain (Fig. 3) When running over a visible obstacle, birds use a three-step negotiation strategy, with clear evi- dence of feedforward, predictive adjustments in the step preceding the obstacle (Fig. 3). Model-based analyses suggest that the strategy used by birds is most consistent with models optimized to regulate leg loading in uneven terrain, not to maintain steady body dynamics (Birn-Jeffery et al. 2014). Regulation of leg cycling rate can be viewed as a combined feedforward plus ‘preflexive’ control strat- egy that minimizes the need for reactive adjustments by exploiting the intrinsic mechanical coupling be- tween leg contact angle and leg loading. Experimental evidence from both humans and birds running over a range of terrain perturbations are con- sistent with leg angular trajectory as a key target of neural control (Seyfarth et al. 2003; Blum et al. 2011; Mu¨ller et al. 2016). Humans and birds allow body dynamics to transiently vary, but exhibit tight cou- pling between leg contact angle and leg loading across many different terrain contexts (Grimmer et al. 2008; Birn-Jeffery and Daley 2012; Birn-Jeffery et al. 2014; Blum et al. 2014; Mu¨ller et al. 2016). Empirical evi- dence from birds running over a range of terrain perturbations, including visible overground obstacles, treadmill obstacles, visible drops, visible and invisible potholes, all suggest that leg angular trajectory is: (1) relatively insensitive to perturbations and (2) adjusted subtly over longer time-scales. This suggests a context- dependent feedforward optimization of leg angular trajectory at higher levels in the control hierarchy to enable robust and stable locomotion with minimal control intervention (Birn-Jeffery and Daley 2012; Birn-Jeffery et al. 2014). Whereas leg angular trajectory appears insensitive to perturbations and adjusted over longer timescales, leg- length actuation shows high stride-to-stride variance, suggesting both predictive (feedforward) and reactive (feedback) adjustment in uneven terrain (Fig. 3; Birn- Jeffery and Daley 2012; Birn-Jeffery et al. 2014; Blum et al. 2014). Leg length actuation is sensitive to altered landing conditions, such that stance push-off is rapidly adjusted to stabilize the total mechanical energy of the body in uneven terrain. (Daley and Biewener 2006; Birn-Jeffery and Daley 2012; Birn-Jeffery et al. 2014; Fig. 3. Leg length and leg angular trajectories of pheasants ne- gotiating visible obstacles, illustrating a typical three-step strategy. At top, schematic illustration of the landing and take-off condi- tions of the bird during the step preceding (Step 1), the step on the obstacle (Step 0), and the obstacle dismount (Step þ1). Below, leg trajectory (length and angle) during running on level terrain (thin black lines, mean and 95% confidence intervals) and over an obstacle height of 30% leg length (thicker gray lines). Upward triangles indicate foot take-off at the end of stance. Leg length exhibits high stride-to-stride variance in uneven terrain, whereas leg angular trajectory follows a relatively consistent si- nusoidal trajectory, with only subtle changes in rate in anticipa- tion of terrain height changes. Data from Birn-Jeffery and Daley (2012). 888 M. A. Daley Blum et al. 2014). These findings suggest modular control of leg angular trajectory and leg-length actuation. While modular control of leg angular trajectory and leg-length actuation have emerged as consistent control strategies for robustly stable running, it remains less clear whether, and under what circumstances, leg stiff- ness serves as a direct target of control. Research on humans running over soft and hard surfaces suggests that humans regulate leg stiffness to maintain steady body trajectory (Ferris et al. 1999). However, the spe- cific terrain conditions used, soft and hard surfaces, did not allow a clear distinction between control priority for steady body trajectory versus consistent leg forces, because both were maintained. Humans running over visible downward steps exhibit anticipatory shifts in leg stiffness before a perturbation, but do not adjust leg stiffness within perturbed steps (Mu¨ller et al. 2012). In these human studies, subjects were specifically instructed to maintain constant running speed. In con- trast, birds negotiating terrain drops exhibit high var- iance in leg stiffness while allowing speed to transiently vary (Daley et al. 2007; Blum et al. 2011; Mu¨ller et al. 2016). Whether or not leg stiffness is directly regulated may depend on the context-dependent constraints on the locomotor task. Differences between birds and humans in stiffness regulation could also relate to leg morphology. Birds have a more crouched leg posture with four seg- ments, in contrast to the vertically oriented three- segment leg configuration of humans. The limb morphology of birds may allow more flexible adjust- ment of leg posture to accommodate terrain varia- tion, minimizing the need for active regulation of leg stiffness. This idea is supported by evidence from a study that directly compared control strategies in humans and birds from a model-based perspective (Blum et al. 2011). Birds exhibited a wider range of stable control solutions without adjusting leg stiff- ness, whereas humans are required to adjust leg stiff- ness to remain in the stable solution space (Blum et al. 2011). Additionally, this study showed that birds exhibited higher robustness to terrain height variation than humans, consistent with the more crouched posture enabling postural adjustments to minimize disturbances. In vivo muscle recordings reveal neuromuscular mechanisms underlying robust, stable, and agile locomotion While external measures of body and limb mechan- ics can help reveal task-level locomotor control strat- egies, these measures do not reveal the underlying neuromuscular mechanisms. In vivo recordings of muscle force, length, and activation dynamics during perturbed locomotion can help reveal the relative contributions of intrinsic mechanical, feedback, and feedforward control mechanisms. These studies also help reveal how neuromechanics of locomotion are integrated across levels of organization, from indi- vidual muscle–tendon dynamics to joint, whole limb, and body dynamics. The relationship between muscle activation and mechanical output is known to be nonlinear and dynamically variable, depending on instantaneous fascicle length, velocity and recent strain history (Askew and Marsh 1998; Josephson 1999; Edman 2012; Herzog 2014). In vivo measures of muscle function during steady-state locomotor behaviors have revealed muscle–tendon mechanisms for economic bipedal locomotion (Biewener and Baudinette 1995; Roberts et al. 1997; Daley and Biewener 2003), but do not reveal the mechanisms underlying robustness, stability, and agility in non- steady behaviors. In vivo recordings of distal hindlimb muscles of the guinea fowl during negotiation of uneven terrain has shown that these muscles exhibit rapid changes in force and work in response to altered foot– substrate interactions, contributing to the intrinsic stability of locomotion. During negotiation of unex- pected potholes, the peak force of the lateral gastroc- nemius muscle (LG) during stance decreases by 81% during perturbed steps compared to steady strides, despite maintaining the same electromyography (EMG) activation levels (Fig. 4; Daley et al. 2009). The muscle shortens rapidly during the initial per- turbation period, when the foot contacts and breaks through the false floor (tissue paper) and extends toward the true ground below (Fig. 4). In the sub- sequent stance period, peak muscle force is reduced, but peak ground reaction force is similar, and the muscle is stretched, resulting in energy absorption (Daley and Biewener 2006; Daley et al. 2009). This has a stabilizing effect on the body mechanical en- ergy, offsetting the increase in kinetic energy gained through exchange of gravitational potential energy during the fall (Daley and Biewener 2006; Daley et al. 2009). A similar but converse response is ob- served in upward steps and obstacles, in which in- creased stretch and longer length during force development during a step onto an obstacle results in higher force production and work output, increas- ing mechanical energy of the body (Daley and Biewener 2011; Fig. 5). These studies revealed that LG force–length dynamics rapidly adjust the degree of stance push-off in response to altered foot–sub- strate interactions. This load-dependent actuation Agility of running birds 889 response of distal hindlimb muscles provides rapid stabilization of body mechanical energy in uneven terrain, and is consistent with observed whole-body and leg dynamics. Subsequent modelling studies have also confirmed that load-dependent actuation increases robustness and stability of running dynam- ics (Schmitt and Clark 2009). Load- and posture-dependent shifts in muscle force and work occur without shifts in total muscle EMG activity during unexpected drop perturbations, revealing that intrinsic mechanisms play an impor- tant role in the response (Daley et al. 2009). However, increased EMG activity does contribute to the response during obstacle steps, likely mediated through short-latency proprioceptive reflexes (Daley and Biewener 2011). Interestingly, the qualitative pat- terns of muscle force–length dynamics remain similar in both unexpected and anticipated obstacle condi- tions (Daley et al. 2009; Daley and Biewener 2011). Nonetheless, while the overall force–length dynamics of the muscles remain similar across contexts, there is clear evidence of shifts in the relative contribution of intrinsic and neurally-mediated mechanisms of con- trol, depending on the sensory context. In a more recent study, Gordon and colleagues (2015) investigated context dependent shifts in sen- sorimotor control by comparing muscle activation patterns during obstacle negotiation at low and high speeds, and with low and high-contrast obstacles. In slower speed obstacle negotiation, an- ticipatory increases in muscle activity are apparent in Fig. 4. LG muscle length, force, and activation during the im- mediate response to a hidden pothole perturbation. Figure modified from Daley et al. (2009). At top, the guinea fowl is pictured at the time of ground contact after breaking through the false-floor of tissue paper. Below, thin lines indicate the mean and 95% confidence intervals for steady level running, and thick lines illustrate a perturbed drop step. Force and length are rapidly altered in response to the perturbation, although muscle activa- tion (EMG) remains similar to the level terrain condition. Fig. 5. Load- and posture-dependent actuation of the LG muscle during negotiation of uneven terrain. When leg posture is altered at the time of foot contact, altering the balance between muscle and external forces, muscle length during force development (Lt50) varies. Lt50 is the largest predictor of the force and total work output of the muscle (Wnet) ( Daley et al. 2009, Daley and Biewener 2011 ). This posture-dependent response is similar between unexpected perturbations and repeating obstacles. This suggests similar task-level control strategies across context, de- spite potential for differing contributions of intrinsic mechanical, feedforward, and feedback control mechanisms to the response. 890 M. A. Daley the steps preceding obstacles. At higher running speeds, the neuromuscular response is largely reac- tive, occurring after foot contact with the obstacle (Fig. 6; Gordon et al. 2015). Anticipatory increases in muscle activity are larger when the obstacles are more easily visible (higher contrast to surrounding terrain), but mainly in slower speed obstacle negoti- ation. In the higher speed condition, the response remains mainly reactive, despite increased obstacle visibility (Gordon et al. 2015). This likely relates to the sensorimotor delays involved in visual contribu- tions to path planning and navigation in higher brain centers. The results are consistent with a shift in sensorimotor control mechanisms with speed, with greater reliance on vision and anticipatory adjustments at slower speeds, and greater reliance on intrinsic mechanics and reactive feedback mech- anisms at high speeds. Thus, the regulation of mus- cle dynamics reflects a redundant system with coordinated contributions from intrinsic mechanical, feedback, and feedforward mechanisms. Conclusions While neuromechanical control of locomotion involves a complex interplay of mechanical and sen- sorimotor mechanisms, studies of running birds have revealed several strategies for robust, stable, and agile bipedal locomotion that are consistent across terrain contexts: (1) independent control of leg angular cy- cling and leg length actuation, which facilitates dy- namic stability through simple control mechanisms, (2) feedforward regulation of leg cycling rate to maintain consistent leg loading in uneven terrain, (3) load-dependent muscle actuation to stabilize body mechanical energy in response to disturbances, and (4) multi-step recovery strategies that allow body dynamics to transiently vary while tightly reg- ulating leg loading to minimize risks of fall and in- jury. Muscle proprioceptive feedback arising from non-steady force–length dynamics likely plays im- portant roles in effective tuning of perturbation responses over time, as well as maintaining accurate state estimates for internal models, path planning, and navigation in higher brain centers. However, it remains unclear how sensory feedback is integrated with spinal neural circuits and higher brain centers to adjust locomotor control over short and long time-scales. In future work, it will be interesting to investigate the learning and adaptation of neurome- chanical control mechanisms through repeated expo- sure to perturbations in controlled conditions. Acknowledgments Thanks to the past students, postdoctoral researchers and collaborators who have contributed to the work reviewed here, including Aleksandra Birn-Jeffery, Yvonne Blum, Christian Hubicki, Hamid Vejdani, Joanne Gordon, Jonathan Hurst and Andrew Biewener. Funding The work reviewed here was supported by grants from the Biotechnology and Biological Sciences Research Council of the UK (BB/H005838/1) and the Human Frontier Science Program (RGY0062/ 2010). Support for participation in this symposium was provided by Photron (https://photron.com), the Company of Biologists, the Society for Comparative and Integrative Biology (Divisions of Comparative Biomechanics, Vertebrate Morphology, Animal Behavior, and NNSB), the Air Force Office of Fig. 6. Context dependent-shifts in the contribution of predictive and reactive modulation of LG activity during obstacle negotia- tion. Guinea fowl running over obstacles on a treadmill en- countered a single footfall on an obstacle (black box) approximately once in 5–7 steps. Step ID corresponds to the sequence of steps with the obstacle encounter at step zero. LG exhibits predictive increases in muscle activity at slower walking speeds (0.7 m/s). Predictive shifts are larger when the obstacles are more visible (higher contrast) relative to the level terrain. At higher speeds (1.3 m/s) guinea fowl use a reactive strategy, with increases in LG activity occurring after foot contact with the obstacle. The influence of high versus low contrast terrain is greater at slower speeds, when the bird has a longer time to process visual information to modulate muscle activity. Agility of running birds 891 Scientific Research (FA9550-16-1-0165), and the National Science Foundation (IOS-1747859). References Abourachid A, Ho¨fling E. 2012. The legs: a key to bird evo- lutionary success. J Ornithol 153:193–8. Alexander RM. 2002. Stability and manoeuvrability of terres- trial vertebrates. Integr Comp Biol 42:158–64. Askew GN, Marsh RL. 1998. Optimal shortening velocity (V/Vmax) of skeletal muscle during cyclical contractions: length-force effects and velocity-dependent activation and deactivation. J Exp Biol 201:1527–40. Biewener AA, Baudinette RV. 1995. In vivo muscle force and elastic energy storage during steady-speed hopping of tam- mar wallabies (Macropus eugenii). J Exp Biol 198:1829–41. Birn-Jeffery AV, Daley MA. 2012. Birds achieve high robust- ness in uneven terrain through active control of landing conditions. J Exp Biol 215:2117–27. Birn-Jeffery AV, Hubicki CM, Blum Y, Renjewski D, Hurst JW, Daley MA. 2014. Don’t break a leg: running birds from quail to ostrich prioritise leg safety and economy on un- even terrain. J Exp Biol 217:3786–96. Blum Y, Birn-Jeffery A, Daley MA, Seyfarth A. 2011. Does a crouched leg posture enhance running stability and robust- ness?. J Theor Biol 281:97–106. Blum Y, Vejdani HR, Birn-Jeffery AV, Hubicki CM, Hurst JW, Daley MA. 2014. Swing-leg trajectory of running guinea fowl suggests task-level priority of force regulation rather than disturbance rejection. PLoS One 9:e100399. Brown IE, Loeb GE. 2000. A reductionist approach to creat- ing and using neuromechanical models. In: Winters JM, Crago PE, editors. Biomechanics and neural control of pos- ture and movement. New York: Springer-Verlag. p. 148–63. Clark AJ, Higham TE. 2011. Slipping, sliding and stability: locomotor strategies for overcoming low-friction surfaces. J Exp Biol 214:1369–78. Daley MA. 2016. Non-steady locomotion. In: Bertram JEA, editor. Understanding mammalian locomotion: concepts and applications. Hoboken (NJ): John Wiley & Sons, Inc. p. 277–306. Daley MA, Biewener AA. 2003. Muscle force-length dynamics during level versus incline locomotion: a comparison of in vivo performance of two guinea fowl ankle extensors. J Exp Biol 206:2941–58. Daley MA, Biewener AA. 2006. Running over rough terrain reveals limb control for intrinsic stability. Proc Natl Acad Sci U S A 103:15681–6. Daley MA, Biewener AA. 2011. Leg muscles that mediate stability: mechanics and control of two distal extensor muscles during obstacle negotiation in the guinea fowl. Philos Trans R Soc Lond B Biol Sci 366:1580–91. Daley MA, Felix G, Biewener AA. 2007. Running stability is enhanced by a proximo-distal gradient in joint neurome- chanical control. J Exp Biol 210:383–94. Daley MA, Usherwood JR. 2010. Two explanations for the compliant running paradox: reduced work of bouncing viscera and increased stability in uneven terrain. Biol Lett 6:418–21. Daley MA, Usherwood JR, Felix G, Biewener AA. 2006. Running over rough terrain: guinea fowl maintain dynamic stability despite a large unexpected change in substrate height. J Exp Biol 209:171–87. Daley MA, Voloshina A, Biewener AA. 2009. The role of intrinsic muscle mechanics in the neuromuscular control of stable running in the guinea fowl. J Physiol 587:2693–707. Duperret JM, Kenneally GD, Pusey JL, Koditschek DE. 2016. Towards a comparative measure of legged agility. Springer. p. 3–16. Edman KAP. 2012. Residual force enhancement after stretch in striated muscle. A consequence of increased myofilament overlap? J Physiol 590:1339–45. Ferris DP, Liang K, Farley CT. 1999. Runners adjust leg stiff- ness for their first step on a new running surface. J Biomech 32:787–94. Gatesy SM, Middleton KM. 1997. Bipedalism, flight, and the evolution of theropod locomotor diversity. J Vertebr Paleontol 17:308–29. Gordon JC, Rankin JW, Daley MA. 2015. How do treadmill speed and terrain visibility influence neuromuscular control of guinea fowl locomotion? J Exp Biol 218: 3010–22. Grillner S, Wallen P, Saitoh K, Kozlov A, Robertson B. 2008. Neural bases of goal-directed locomotion in vertebrates— an overview. Brain Res Rev 57:2–12. Grimmer S, Ernst M, Gu¨nther M, Blickhan R. 2008. Running on uneven ground: leg adjustment to vertical steps and self-stability. J Exp Biol 211:2989–3000. Heers AM, Dial KP. 2015. Wings versus legs in the avian bauplan: development and evolution of alternative locomo- tor strategies. Evolution 69:305–20. Herzog W. 2014. Mechanisms of enhanced force production in lengthening (eccentric) muscle contractions. J Appl Physiol 116:1407–17. Ijspeert AJ. 2018. Decoding the neural mechanisms underly- ing locomotion using mathematical models and bio- inspired robots: from lamprey to human locomotion. In: Bicchi A, Burgard W, editors. Robotics research. Springer Proceedings in Advanced Robotics. Vol. 2. Cham, Switzerland: Springer. p. 177–86. Inouye JM, Valero-Cuevas FJ. 2016. Muscle synergies heavily influence the neural control of arm endpoint stiffness and energy consumption. PLoS Comp Biol 12:e1004737. Jindrich DL, Full RJ. 2002. Dynamic stabilization of rapid hexapedal locomotion. J Exp Biol 205:2803–23. Josephson RK. 1999. Dissecting muscle power output. J Exp Biol 202:3369–75. Kiehn O. 2016. Decoding the organization of spinal circuits that control locomotion. Nat Rev Neurosci 17:224. McCrea DA, Rybak IA. 2008. Organization of mammalian locomotor rhythm and pattern generation. Brain Res Rev 57:134–46. McLean DL, Dougherty KJ. 2015. Peeling back the layers of locomotor control in the spinal cord. Curr Opin Neurobiol 33:63–70. More HL, Hutchinson JR, Collins DF, Weber DJ, Aung SK, Donelan JM. 2010. Scaling of sensorimotor control in ter- restrial mammals. Proc R Soc Lond Ser B Biol Sci (doi:10.1098rspb20100898). More HL, O’Connor SM, Brøndum E, Wang T, Bertelsen MF, Grøndahl C, Kastberg K, Hørlyck A, Funder J, 892 M. A. Daley Donelan JM. 2013. Sensorimotor responsiveness and reso- lution in the giraffe. J Exp Biol 216:1003–11. Moritz CT, Farley CT. 2004. Passive dynamics change leg mechanics for an unexpected surface during human hop- ping. J Appl Physiol 97:1313–22. Mu¨ller R, Birn-Jeffery AV, Blum Y. 2016. Human and avian running on uneven ground: a model-based comparison. J R Soc Interface 13:20160529. Mu¨ller R, Ernst M, Blickhan R. 2012. Leg adjustments during running across visible and camouflaged incidental changes in ground level. J Exp Biol 215:3072–9. Roberts TJ, Marsh RL, Weyand PG, Taylor CR. 1997. Muscular force in running turkeys: the economy of mini- mizing work. Science 275:1113–5. Rossignol S, Dubuc R, Gossard J-P. 2006. Dynamic senso- rimotor interactions in locomotion. Physiol Rev 86: 89–154. Schmitt J, Clark J. 2009. Modeling posture-dependent leg ac- tuation in sagittal plane locomotion. Bioinspiration Biomim 4:046005. Seyfarth A, Geyer H, Herr H. 2003. Swing-leg retraction: a simple control model for stable running. J Exp Biol 206:2547–55. Valero-Cuevas FJ. 2016. Fundamentals of neuromechanics. London: Springer-Verlag. Zeffer A, Johansson LC, Marmebro A˚ . 2003. Functional cor- relation between habitat use and leg morphology in birds (Aves). Biol J Linn Soc 79:461–84. Agility of running birds 893

Understanding the Agility of Running Birds: Sensorimotor and Mechanical Factors in Avian Bipedal Locomotion.

[]

Daley, Monica A

eng

PMC4076256

Swing-Leg Trajectory of Running Guinea Fowl Suggests Task-Level Priority of Force Regulation Rather than Disturbance Rejection Yvonne Blum1, Hamid R. Vejdani2, Aleksandra V. Birn-Jeffery1,3, Christian M. Hubicki2, Jonathan W. Hurst2, Monica A. Daley1* 1 Department of Comparative Biomedical Sciences, Royal Veterinary College, Hatfield, Hertfordshire, United Kingdom, 2 Mechanical, Industrial and Manufacturing Engineering, Oregon State University, Corvallis, Oregon, United States of America, 3 Department of Biology, University of California Riverside, Riverside, California, United States of America Abstract To achieve robust and stable legged locomotion in uneven terrain, animals must effectively coordinate limb swing and stance phases, which involve distinct yet coupled dynamics. Recent theoretical studies have highlighted the critical influence of swing-leg trajectory on stability, disturbance rejection, leg loading and economy of walking and running. Yet, simulations suggest that not all these factors can be simultaneously optimized. A potential trade-off arises between the optimal swing-leg trajectory for disturbance rejection (to maintain steady gait) versus regulation of leg loading (for injury avoidance and economy). Here we investigate how running guinea fowl manage this potential trade-off by comparing experimental data to predictions of hypothesis-based simulations of running over a terrain drop perturbation. We use a simple model to predict swing-leg trajectory and running dynamics. In simulations, we generate optimized swing-leg trajectories based upon specific hypotheses for task-level control priorities. We optimized swing trajectories to achieve i) constant peak force, ii) constant axial impulse, or iii) perfect disturbance rejection (steady gait) in the stance following a terrain drop. We compare simulation predictions to experimental data on guinea fowl running over a visible step down. Swing and stance dynamics of running guinea fowl closely match simulations optimized to regulate leg loading (priorities i and ii), and do not match the simulations optimized for disturbance rejection (priority iii). The simulations reinforce previous findings that swing-leg trajectory targeting disturbance rejection demands large increases in stance leg force following a terrain drop. Guinea fowl negotiate a downward step using unsteady dynamics with forward acceleration, and recover to steady gait in subsequent steps. Our results suggest that guinea fowl use swing-leg trajectory consistent with priority for load regulation, and not for steadiness of gait. Swing-leg trajectory optimized for load regulation may facilitate economy and injury avoidance in uneven terrain. Citation: Blum Y, Vejdani HR, Birn-Jeffery AV, Hubicki CM, Hurst JW, et al. (2014) Swing-Leg Trajectory of Running Guinea Fowl Suggests Task-Level Priority of Force Regulation Rather than Disturbance Rejection. PLoS ONE 9(6): e100399. doi:10.1371/journal.pone.0100399 Editor: Amir A. Zadpoor, Delft University of Technology (TUDelft), Netherlands Received January 24, 2014; Accepted May 27, 2014; Published June 30, 2014 Copyright:  2014 Blum et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This study was funded by grant RGY0062/2010 of the Human Frontier Science Program (HFSP). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * Email: mdaley@rvc.ac.uk Introduction Legged locomotion involves coordination of limb swing and stance phases with distinct yet tightly coupled dynamics. Studies of legged locomotion often focus primarily on the dynamics of the stance phase, during which an animal’s legs experience the greatest demands for force and power [1–8]. Yet, recent research highlights the critical role of swing-leg trajectory on locomotor dynamics—experimental evidence shows that leg loading is critically sensitive to the initial landing conditions (leg angle, leg length and body velocity) at the swing-stance transition [9–11], which are influenced by swing-leg trajectory. Running animals must effectively coordinate the interplay of swing-leg trajectory, landing conditions and stance leg loading [12–16]. For example, when running guinea fowl encounter an unexpected pothole, late- swing leg retraction leads to variation in leg contact angle, which explains 80% of the variance in stance leg impulse [17]. Thus, the swing-leg trajectory is a critical factor in the dynamics of legged locomotion, particularly during movement over uneven terrain. Recent theoretical studies have highlighted inherent trade-offs in swing-leg trajectory for walking and running in uneven terrain. Simple walking and running models have revealed that swing-leg velocity just before the stance transition influences numerous aspects of locomotor dynamics, including stability [14–16,18], robustness [19], leg work [19,20], disturbance rejection and collision impact energy losses [18]. Previous studies suggest these factors cannot be simultaneously optimized—resulting in a trade- off between two families of performance objectives: swing-leg velocity can be optimized to minimize peak forces, work and collision impacts [16,18–20], or to provide stability, disturbance rejection and robustness of body centre of mass (CoM) dynamics [15,16,18–20], but not all simultaneously. Thus, a potential trade- off has emerged between optimal swing-leg trajectory to regulate leg loading for injury avoidance, or alternatively, to facilitate steady PLOS ONE | www.plosone.org 1 June 2014 | Volume 9 | Issue 6 | e100399 gait through disturbance rejection. Yet, while theoretical studies suggest such a trade-off, there is no experimental data on how running animals optimize swing-leg trajectory for non-steady locomotion. Do running animals favor one end of this trade-off, or alternatively, find a compromise solution? Both disturbance rejection and injury avoidance have potential to be important task-level priorities for running animals. Disturbance rejection refers to minimizing the effect of perturbations on the body center of mass (CoM) trajectory [21]. Buffering the CoM motion against disturbances reduces the risk of fall, and may minimize need for active control intervention [22,23]. Furthermore, some experi- mental evidence has suggested steady CoM dynamics as an important task-level goal in legged locomotion [9,24]. However, minimizing leg impacts and peak forces may also be critical, because animal legs have relatively constant safety factors in musculoskeletal structures around 2–46 the peak forces of steady locomotion [25,26]. Perfect disturbance rejection can demand large leg forces [18–20], which could lead to musculoskeletal injury. Building legs to withstand very large forces would require carrying extra weight, so limited safety factors in animal legs may reflect a compromise between safety and economy. Specialized runners like cursorial ground birds appear to have a structure that is more optimized for economy, with relatively light legs and thin tendons, which could inherently limit safety factors [6,26,27], making them prone to injury [28,29]. Based on these consider- ations, we reason that both disturbance rejection and injury avoidance have potential to be important, yet sometimes conflicting, task-level priorities in animal locomotion. In this paper, we test the hypothesis that running guinea fowl use swing-leg trajectory optimized to regulate leg loading (reflecting priority for injury avoidance), against an alternative hypothesis that they use swing-leg trajectory optimized for disturbance rejection (reflecting priority for steady body dynam- ics). These hypotheses represent the two ends of the theoretical trade-off in swing-leg trajectory described above, providing useful points of comparison to animal behavior. In reality, animal swing- leg trajectory could reflect an intermediate compromise solution, which can be revealed by comparing experimentally observed swing trajectories to simulation predictions for the two hypothet- ical extremes. We experimentally measured swing-leg trajectory and stance dynamics of guinea fowl running over a visible step down in terrain, when given ample practice, distance and time to anticipate the drop. This contrasts with previous studies of the intrinsic-dynamic response to an unexpected terrain drop [10,17]. Here, we are focused on understanding the task-level priorities reflected in the ‘optimized’ locomotor behavior. To generate simulation predictions, we use a simple approach with swing-leg geometry that evolves as a function of time during the flight phase, according to a prescribed trajectory optimized to meet a specific performance objective [20,30]. The swing-leg trajectory determines the landing conditions at the swing-stance transition, and the landing conditions are used to predict stance dynamics based on a simple running model (see methods for model details). We generate simulations with swing-leg trajectory optimized for three specific performance objectives, the first two reflecting a priority to regulate leg loading, and the third reflecting a priority for disturbance rejection. Specifically, we optimize swing trajectory to achieve i) constant peak force, ii) constant axial impulse, or iii) perfect disturbance rejection (steady gait) in the step immediately following a downward step in terrain. Similar swing- leg control policies have been investigated previously in simula- tion: Ernst and colleagues investigated swing-leg trajectory optimized to target steady gait (constant speed and bounce height), to provide disturbance rejection in uneven terrain [30], and Vejdani and colleagues compared several possible priorities in simulation, including steady gait, constant leg work and constant leg loading [20]. Here we directly compare simulation predictions to new experimental data on guinea fowl running over a visible step down, to understand how task-level priorities influence swing- leg control in running birds. Optimization of swing-leg trajectory to achieve well-defined intrinsic-dynamic characteristics could be particularly important for animal locomotion because neuromuscular delays limit the rate of feedback-mediated responses to perturbations, and terrain conditions are not often perfectly known. Neuromuscular delays (synaptic, conduction, electromechanical and force development) can represent a large fraction of the step cycle in animals [10,31], and therefore limit the rate of feedback in both stance and swing. These neural delays are likely to be especially problematic at the swing-stance transition, when small changes in landing conditions have large influence on stance leg loading and body dynamics [9– 11]. If the animal’s knowledge of the terrain is imperfect, variation in terrain height leads to a disturbance, with the immediate response determined by feed-forward muscle activation and the system’s intrinsic dynamics [9,10,17]. Application of a prescribed swing-leg trajectory can provide well-defined intrinsic-dynamic response in terms of stability, disturbance rejection and leg loading characteristics, bridging neuromuscular delays and minimizing need for rapid neural feedback. The focus of this paper is to understand the task-level mechanical priorities reflected in the swing-leg trajectory used by running animals. Methods 1 Experiments Avian running trials were conducted on a 0:6|4:5 m runway. Five 0:6|0:9 m force plates (model 9287B, Kistler, Winterthur, Switzerland) were arranged in a row to record the ground reaction forces (sampling frequency 500 Hz). A camera system (Qualisys, Gothenburg, Sweden), consisting of eight high speed infrared cameras, was used to capture body kinematics (sampling frequency of 250 Hz). For further analysis, both force data and kinematic data were interpolated to a frequency of 500 Hz. We used three experimental terrain conditions: a level runway, a runway with a 4 cm drop and a runway with a 6 cm drop (figure 1(A)). Five guinea fowl (Numida meleagris) (body mass m~1:39+0:24 kg, touch down (TD) leg length during level running LTD,Level~0:21+0:02 m) were encouraged to run from one end of the runway to the other (running the step down). We wanted to understand the birds’ optimized strategy, as opposed to an unexpected perturbation response, so we trained the birds for a week before data collection. Before data collection, the birds were accustomed both to the task and to being handled by humans. Trials for each terrain condition (level, 4 cm drop, 6 cm drop) were collected in a single block (not randomized), to allow the birds to correctly anticipate the terrain. We collected 10 steady running trials per bird per condition, in which the approach up to the ‘22 step’ (before the middle of the runway) was in straight-line and approximately steady. Since we could not control the birds’ running speed, we also analyzed their velocity and acceleration during post-processing, as explained in further detail in 2. Neither surgery or anesthesia were used in this study because no invasive procedures were involved. The Royal Veterinary College Ethics and Welfare Committee approved all of the animal experiment protocols under the project title ‘Kinematics and kinetics in birds running over an uneven terrain’. Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 2 June 2014 | Volume 9 | Issue 6 | e100399 To approximate the CoM position and the foot point, two markers were attached to the birds’ back (cranial and caudal), one at digit III and one at the tarsometatarsophalangeal joint. The marker placement and techniques used to estimate the initial position and velocity of the CoM were the same as reported in [13]. The initial position of the CoM was determined by the average of the cranial and caudal marker position, and the initial velocity condition was derived from kinematics using the path- match optimization technique as described by [17]. We further corrected the initial position estimate based on the assumption that the birds’ pitch angular momentum during level running should be minimized (the body should not pitch forward or backward during steady running). This optimization led to an estimate of the true CoM location as positional offsets from the original markers placed on the birds back (horizontal offset xoffset~0:032+0:015 m, vertical offset yoffset~{0:040+0:016 m) [13]. We then calculated the CoM position trajectories by integrating the ground reaction forces twice. The following variables were extracted from the experimental data for further analysis: the length of the virtual leg L, which is defined as the distance between the CoM and the foot point, and its derivative, leg length velocity _L (figure 1(B)), the virtual leg angle a, which is measured anti-clockwise with respect to the horizontal, and its derivative, leg angular velocity _a, their corresponding TD conditions LTD and _LTD, aTD, _aTD, the axial (directed along the virtual leg) and fore-aft horizontal ground reaction force Faxial and Fx, respectively, the axial peak force Faxial,max, the axial and fore-aft impulse Iaxial and Ix, respectively, which are calculated by integrating the corresponding force trajectories over stance time, and the net CoM work DECoM, which is the net change in CoM Energy over the course of stance. 2 Statistical Analysis We made all parameters non-dimensional by normalizing them with respect to body mass m, gravitational acceleration g, body weight BW~mg, average TD leg length during level running L0~LTD,Level and periodic time of a pendulum T~ ffiffiffiffiffiffiffiffiffiffiffi L0=g p . The steps were categorized into four step types: level running (Level), two steps before drop (step 22), the pre-drop step (step 21), the drop step itself (step 0) and the first post-drop step (step +1). Since we could not control the birds’ running speed, we analyzed the fore-aft impulse of step 22 during post-processing and selected steady trials (i.e. DIxDƒ0:15 BW T, which corresponds to a change in fore-aft velocity of less than 0:22 m=s) [13]. Step 22 was used only to assess steadiness of the approach, and not further analyzed. We analyzed a total of 367 running steps at speeds between _x~½1:64,4:07m=s with following sample sizes: Level = 167, Step 21 = 73, Step 0 = 70, and Step +1 = 57. The statistical analysis of the experimental data was performed in Matlab (R2012a, Mathworks Inc., Natick, MA, USA). We ran a mixed model multi-way ANOVA on the entire dataset with fixed effects ‘step type’ nested within ‘drop height’, ‘individual’ as a random effect and ‘speed’ as a continuous effect (table 1). We then performed post-hoc pair-wise t-tests for the differences between the level mean values and the three step types (21, 0, and +1), separated into the two drop height conditions (4 cm, 6 cm) (table 2). As expected, some parameters of gait dynamics were signifi- cantly influenced by forward speed _x [32,33] (table 1). For comparison to simulation results, we were interested in under- standing the effect of the drop perturbation independent from variance in speed. For the factors that exhibited significant speed effect in the mixed model ANOVA, we further analyzed the speed effect using a simple regression analysis. We pooled the normalized data together (all birds, all trials and all step categories) and calculated each parameter’s linear regression with respect to _x, after confirming that the residuals from this regression were approximately normally distributed. If this analysis revealed a substantial speed effect by the criteria R2w0:15 and pv0:01, we recalculated the corresponding parameter (here, we use Y as a placeholder) by taking the residuals of the linear speed-regression (YRes) and adding the mean value of level running ( YLevel): ^Y~YResz YLevel: ð1Þ Based on these results, for further analysis we used the speed- corrected leg length velocity ^_L_LTD (R2~0:43, pv0:001) and the speed-corrected axial peak force ^Faxial,max (R2~0:29, pv0:001). 3 Model We used the passive, planar spring-loaded inverted pendulum (SLIP) as a reduced-order representation of whole-body dynamics of animal locomotion. This model is based on the observation that animals move with bouncing, spring-like gaits, with ground reaction forces approximated by a model with a point mass body and massless legs that resist only compressive loads [34–37]. This model has been widely used in biomechanics and robotics [38], because it qualitatively reproduces the dynamics of both walking [37] and running [34,35]. The SLIP model is a passive, energy conservative dynamic template of locomotion [39]. While active stance models have also been suggested as templates for legged locomotion [40–45], the most appropriate choice of active stance model for running animals remains unclear. In this study, we are focused specifically on the influence of swing-leg trajectory on landing conditions and, consequently, the peak force and impulse of the leg during stance. The passive SLIP model provides good prediction of the stance peak force, impulse and overall body dynamics given specified landing conditions [5,17,34,35,37]. Consequently, the SLIP model is the most appropriate dynamic template for this study, because it allows us to focus specifically on the effects of swing-leg trajectory on running dynamics. Figure 1. Illustration of experiment and modeling approach. A guinea fowl running a step down (A), and schematic drawing of the spring-loaded inverted pendulum (SLIP) model with swing-leg trajec- tory control applied as a function of fall time (B). The gray areas indicate the stance phases, and the line represents the body centre of mass (CoM) trajectory. The green dotted line indicates the time between apex and touch down (TD) during which the leg angle of the SLIP is adjusted according to the applied control strategy (see Methods). doi:10.1371/journal.pone.0100399.g001 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 3 June 2014 | Volume 9 | Issue 6 | e100399 The SLIP model has a multitude of possible solutions, depending on initial conditions (body position and velocity) and leg parameters (leg stiffness and leg length). In this model, the body is represented by a point mass m supported by a linear leg spring of stiffness k and resting leg length L0, touching the ground with the angle of attack aTD (figure 1(B)). During flight phase the CoM follows a ballistic curve, determined by the acceleration of gravity. The transition from flight to stance occurs when the landing condition y~L0 sin(aTD) is fulfilled. During stance phase the equation of motion is given by m€r~k L0 r {1   r{mg, ð2Þ with r~(x,y)T being the position of the point mass with respect to the foot point, r its absolute value and g~(0,g)T the gravitational acceleration, with g~9:81 m=s2. Take off occurs when the leg length (distance between the CoM and toe) exceeds the resting leg length L0. Since the system is energetically conservative, its state is fully described by the apex condition (y0, _x0 ð ÞT, with x0~0 and _y0~0 (the apex is the highest point of the CoM trajectory). To estimate appropriate SLIP model leg stiffness k and TD angle of the virtual leg aTD,SLIP, we optimized these model leg parameters to match the experimentally observed average ‘steady gait’ values for forward velocity, apex height, peak axial leg force and total axial leg impulse. As noted earlier, the peak force, impulse and body CoM dynamics of animal locomotion can be well approximated by the SLIP model [5,17,34,35,37]. In level terrain, all steady steps were included in the average used to fit a reference steady SLIP model. For the simulations in uneven terrain (4 cm and 6 cm drop), we used the step prior to the disturbance (step 21) to generate the reference steady gait, optimizing the leg parameters to match the peak force, axial impulse, apex height and forward velocity of this step. The model leg stiffness remained fixed within a terrain condition, and was therefore unchanged between step 21 and step 0, but was allowed to vary between terrains (level versus 4 cm, 6 cm drop runways), reflecting potential shifts in the reference ‘steady’ gait. The model was implemented in Matlab (R2012a, Mathworks Inc., Natick, MA, USA). 4 Running Simulations with Swing-Leg Trajectory To simulate running, we used the SLIP model with initial conditions and parameters of the reference steady gait (see above), and applied a prescribed swing-leg trajectory as a function of fall time during the flight phase to control TD conditions at the swing- stance transition. We assume our model has an anticipated time of ground contact for the nominal steady gait at a given speed, but no specific information about the terrain, including the size and location of the drop. We prescribe a continuous evolution of swing-leg angle as a function of time during the flight phase, from the instant of apex until the actual ground contact (figure 1). This means that if the ground is contacted early or late compared to the reference steady gait, the TD conditions are altered. During stance, no control was applied, and stance dynamics were solely determined by the TD conditions applied to the passive SLIP model. Thus, the only control applied to the model was the swing-leg trajectory as a function of fall time. We used the apex to initialize the swing-leg trajectory because it is a unique event that can be easily detected. Note, we do not assume any specific mechanisms of control to be analogous between the model and experiment. We are focused specifically on understanding how different swing-leg trajectories influence dynamics following a drop perturbation. A specified swing-leg trajectory could be achieved through a number of different control mechanisms, which are not the focus of the current study. We generated optimized swing-leg trajectories based on three different objective functions for the subsequent SLIP-modeled stance phase: i) constant peak force, ii) constant axial impulse, or iii) equilibrium (steady) gait. For each proposed objective function, we solved for a swing-leg trajectory as a function of fall time based on the relationship between landing conditions and predicted stance phase dynamics using the SLIP model. We focused our attention specifically on the effects of swing-leg trajectory because previous experimental studies have suggested leg geometry at contact as a primary control target in running [5,14,16,17,46]. We performed an initial simulation analysis to reveal the consequences of simultaneous adjustment of swing-leg length and angle on the predicted stance peak force and axial impulse of the SLIP running model. Figure 2 shows the contour lines of constant peak force (solid lines) and axial impulse (dashed lines) as a function of TD leg angle and TD leg length, predicted by the SLIP model for a single forward speed _x~2:84 m=s (average experi- mentally observed level running speed). Within the region of Table 1. Experimental Data. F-ratio Parameter Drop Height(Step Type) Individual Speed aTD [deg] 49.1* 140.6* 0.9 _aTD [deg/T] 29.8* 36.3* 106.9* LTD [L0] 15.2* 25.7* 24.6* _LTD [L0/T] 20.6* 145.3* 259.9* kLeg [BW/L0] 13.4* 92.2* 8.3* Faxial,max [BW] 9.1* 150.3* 67.8* Iaxial [BW T] 8.1* 250.0* 77.1* Ix [BW T] 15.6* 4.9* 16.5* DECoM [BW L0] 11.2* 15.9* 19.1* Analysis of variance (ANOVA) with four factors: Step type nested within drop height, individual as a random effect, and speed as a continuous effect. N = 367 steps. Significant differences (pƒ0:05) are indicated by asterisks. doi:10.1371/journal.pone.0100399.t001 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 4 June 2014 | Volume 9 | Issue 6 | e100399 Table 2. Experimental Data. Parameter Level Mean (s.d.) Drop Step Type Mean - Level Mean (s.d.) 21 0 +1 aTD [deg] 122.6 (5.4) 4 cm 0.83 (6.0) 27.1 (6.2)* 21.8 (6.8) 6 cm 20.3 (5.4) 29.7 (6.0)* 21.5 (5.6) _aTD [deg/T] 280.9 (15.5) 4 cm 7.0 (16.6)* 220.2 (9.6)* 21.8 (13.1) 6 cm 2.8 (17.6) 224.1 (14.2)* 20.3 (17.8) LTD [L0] 1.00 (0.03) 4 cm 20.01 (0.03)* 0.02 (0.03)* 20.01 (0.04) 6 cm 20.02 (0.04)* 0.03 (0.03)* 0.00 (0.04) ^_L_LTD [L0/T] 1.25 (0.22) 4 cm 0.03 (0.22) 20.18 (0.26)* 0.02 (0.24) 6 cm 0.01 (0.22) 20.24 (0.22)* 0.08 (0.26) kLeg [BW/L0] 11.9 (3.8) 4 cm 3.4 (5.3)* 4.1 (4.2)* 3.3 (4.1)* 6 cm 3.1 (4.7)* 4.3 (3.9)* 4.3 (4.0)* ^Faxial,max [BW] 2.21 (0.36) 4 cm 0.14 (0.47) 0.10 (0.44) 0.21 (0.49)* 6 cm 0.04 (0.45) 20.03 (0.47) 0.33 (0.51)* Iaxial [BW T] 1.01 (0.20) 4 cm 0.04 (0.19) 20.04 (0.21) 0.02 (0.22) 6 cm 0.00 (0.26) 20.14 (0.25)* 0.04 (0.28) Ix [BW T] 0.01 (0.08) 4 cm 20.03 (0.07)* 0.07 (0.08)* 20.03 (0.09) 6 cm 20.03 (0.06)* 0.08 (0.06)* 20.06 (0.08)* DECoM [BW L0] 0.02 (0.17) 4 cm 20.05 (0.17) 0.03 (0.17) 20.12 (0.20)* 6 cm 20.08 (0.14)* 0.01 (0.12) 20.21 (0.28)* Post-hoc t-test to compare the three step types 21, 0, and +1 to level running. Significant differences (pƒ0:05) are indicated by asterisks. doi:10.1371/journal.pone.0100399.t002 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 5 June 2014 | Volume 9 | Issue 6 | e100399 experimentally observed TD leg postures (gray square), the force and impulse contour lines are nearly vertically oriented (figure 2). This reveals that peak force and axial impulse are strongly influenced by leg angle at TD, whereas leg length at TD has a relatively small influence on SLIP-predicted stance leg loading. These observations suggest leg angle is the more effective target for swing-leg control of a SLIP running model. Furthermore, experimentally observed variation in leg length at touchdown is small in magnitude [17]. Animals tend to run with a consistent leg posture because variation in leg length influences gearing and muscle dynamics [10,25]. Consequently, for simplicity, we focused our predictions on simulations of leg angle adjustment only, without changes in leg length. Leg stiffness can also be adjusted as a function of fall time, as a potential control strategy for running [15]. However, leg stiffness is a stance parameter, and not a component of the ‘swing-leg trajectory’ per se. Although we did not directly investigate adjustment of leg stiffness as a function of fall time, we did nonetheless account for between-terrain shifts in leg stiffness, by fitting the nominal steady gait that observed in the ‘21 step’ position on the runway (see Methods section 3). This allowed the nominal steady gait to change between terrains, but not from step- to-step. We did, however, measure the experimentally observed step-by-step variance in effective leg stiffness (see Results and Discussion). In the ‘constant peak force policy’, we optimized the leg angle as a function of fall time such that the resulting peak force during stance remains constant for all steps. Specifically, we solve for a trajectory such that if the foot contacts the ground after apex, the landing conditions lead to a specified SLIP-modeled peak leg force. When this swing trajectory is applied to the model in the presence of a drop perturbation, the leg angle evolves until foot contact, and the peak force of the perturbed step (0) matches the peak force of the previous step (21). In the ‘constant axial impulse policy’, we regulate axial leg impulse rather than peak force, following similar methods. We solve for a swing-leg angular trajectory as a function of fall time to maintain a specific constant axial leg impulse achieved by the SLIP model. When this swing trajectory is applied to the model in the presence of a drop perturbation, the axial impulse of the perturbed step (0) matches that of the previous step (21). The ‘equilibrium gait policy’ has been suggested in theoretical literature as a method for achieving perfect disturbance rejection in uneven terrain [20,30,47]. This strategy ensures that the model achieves a steady gait (constant velocity and bounce height from apex to apex), with a symmetric CoM trajectories with respect to the vertical axis defined by mid-stance (TD and take off conditions are symmetrical). By choosing the appropriate TD leg angle for each velocity vector during the ballistic flight phase _r~(_x,_y)T, an equilibrium gait is obtained regardless of when the foot contacts the ground. We used this relationship to solve for a leg angle trajectory as a function of fall time to ensure steady gait of the SLIP model. While birds may not use a perfect equilibrium gait running, we consider the possible strategy that they optimize swing-leg trajectory to minimize deviations from an equilibrium gait for disturbance rejection. Figure 2. Swing-leg control strategy simulations of leg angle and leg length adjustment. Contours lines of constant peak force (blue solid lines) and constant axial impulse (green dashed lines) as a function of TD leg angle and TD leg length, predicted by the model simulations for one forward speed _x~2:84 m=s (experimentally observed average forward speed for level running). The gray square highlights the area of experimentally observed TD leg angles and TD leg lengths (lower and upper quartile). The slope of the contour lines reveals that TD leg angle has a much higher influence on both peak force and axial impulse than TD leg length. We subsequently focused our swing-leg control policies on leg angle adjustment only. doi:10.1371/journal.pone.0100399.g002 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 6 June 2014 | Volume 9 | Issue 6 | e100399 Further analysis of simulations using the methods above, as well as discussion of stability implications, can be found in Vejdani et al. 2013 [20]. Results We first report the experimentally observed changes in running dynamics (section 1), followed by a description of the simulation predictions (section 2), and comparison between experimental results and simulation predictions (section 3). 1 Experimental Data When guinea fowl negotiate an anticipated drop step, the trajectories over time of leg angle and axial leg force remain remarkably similar to level terrain locomotion. Figure 3 shows the measured trajectories over time of the birds’ leg angle (A), leg length (B), and leg force (C) for the different step types (Level, 21, 0, +1). Notable shifts occur in the stance fore-aft impulse and leg length trajectory of step 0 (figure 3). The results of the ANOVA for experimentally measured variables are listed in table 1, with post- hoc pairwise comparisons in table 2 and boxplots of data in figures 4 and 5. These findings are summarized below. Swing-Leg Kinematics in the Drop Step. The leg angle follows a consistent sinusoidal trajectory (figure 3(A), blue: stance leg, green: swing-leg), with little apparent change during negoti- ation of the drop step. Nonetheless, landing conditions vary in the drop step due to the extension of the ballistic flight phase at the transition between steps 21 and step 0. In the elongated flight phase, continuing leg retraction causes the bird to land with a steeper leg angle aTD at step 0 compared to level running (table 1). The leg length trajectory (figure 3(B), green line) also shows a relatively consistent trajectory across the the step types, but with a slowed rate of lengthening during the elongated flight phase. This results in a small but significant increase in leg length LTD, but a decrease in leg velocity ^_L_LTD at touchdown in step 0 compared to level running. Anticipatory Changes in Step 21. In step 21, preceding the drop, the leg length LTD, leg angular velocity ^_a_aTD and leg length velocity ^_L_LTD differ slightly but significantly from level terrain running (figure 4 and table 2). These findings suggest the birds tune their gait in anticipation of the drop step, which has also Figure 3. Experimental data: trajectories over time. Mean values (solid lines) and standard deviation (colored area) of leg angle (A), and leg length (B) (stance leg in blue, swing leg in green), and leg force (C) (axial force in red, fore-aft force in black) against step time for level running and the three step types 21, 0, and +1. The gray areas indicate the stance phases. The leg angle of both stance and swing leg follows a sinusoidal trajectory (A). Compared to the other step types, the compression of the stance leg is lower during the drop step (step 0) (B). In the drop step (step 0), the axial peak force is not significantly different from the previous step (step 21) or level running, but the fore-aft force indicates an acceleration (C). doi:10.1371/journal.pone.0100399.g003 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 7 June 2014 | Volume 9 | Issue 6 | e100399 been observed during negotiation of visible obstacles [13]. In step 21, the birds reduced leg retraction speed, adopted a 1–2% more crouched leg posture, and increased effective leg stiffness. The increase in leg stiffness was maintained across all three steps of the drop terrain (steps 21,0,+1), whereas the other leg parameters varied between steps (table 2) across the drop terrain. Nonetheless, the swing-leg trajectories remain very similar to level running (figure 3), suggesting that the overall task-level swing-leg control strategy may be maintained across step types within each terrain, with variation in the timing of ground contact causing step-by-step variations in landing conditions. Body Dynamics and Stance Leg Forces. The body dynamics during negotiation of the drop (step 0) are very similar to those observed by guinea fowl negotiating an unexpected pothole [17]. The axial peak force remains consistent in the step preceding (step 21) and during the perturbation (step 0), with no statistically significant change until step +1 (figure 3(C), and table 2). The total axial impulse ^Iaxial (integral of force over time) does not change in step 21, but decreases slightly in step 0, due to reduced stance duration. The net fore-aft impulse indicates acceleration in step 0 (figure 5 and table 2), but DECoM does not differ significantly from level terrain. This indicates that gravita- tional potential energy of the drop is passively converted to forward kinetic energy, increasing velocity, similar to unexpected pothole experiments [17]. The increased velocity is not main- tained, because the negative fore-aft impulse Ix and the negative net CoM work DECoM in the subsequent step (step +1) indicate that the bird actively absorbs energy, slowing down (table 2). In the step preceding the drop (step 21), the net fore-aft impulse Ix indicates slight deceleration, and the net change in body CoM energy DECoM is slightly negative (figure 5 and table 2). Thus, the results indicate a small active deceleration in anticipation of the drop. 2 Simulation Results We generated optimized swing-leg trajectories based upon three hypothesized task-level priorities: i) constant peak force, (ii) constant impulse, and iii) equilibrium (steady) gait. The optimized swing-leg trajectories were applied to a simple running model (see Methods section 3) to predict the swing and stance dynamics in ‘step 0’ of the drop perturbation. The simulations of swing-leg trajectory targeting constant peak force and constant impulse predict relatively similar dynamics during the drop step (table 3 and figure 5). As an illustration of the simulation results for a drop perturbation, figure 6 shows the CoM trajectories (A) and force profiles (B) of the SLIP model with two swing-leg control strategies—constant peak force (solid lines), and equilibrium gait (dashed lines). During level running the CoM trajectories and force profiles are identical, but when the flight phase duration differs from the expected nominal steady gait, the predictions of the two control strategies diverge. The predicted Figure 4. Experimental data: landing conditions. Boxplots of five TD parameters leg angle aTD (A), leg angular velocity _aTD (B), leg length LTD (C), speed-corrected leg length velocity ^_L_LTD (D), and leg stiffness kLeg (E) for level running and the three step types 21, 0, and +1. The boxes indicate the median (black line) and the range between the lower quartile (Q1) and the upper quartile (Q3). The whiskers show the range between the lowest and the highest value still within 1.56 IQR (inter quartile range IQR = Q3 - Q1). For simplicity, individuals and drop heights have been pooled together (see table 2 for more detailed information). Asterisks indicate a significant difference (pƒ0:05) compared to level running (post-hoc t-test). The drop step (step 0) differs significantly from level running for all five variables. doi:10.1371/journal.pone.0100399.g004 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 8 June 2014 | Volume 9 | Issue 6 | e100399 peak force and axial impulse in the drop step increase drastically for the equilibrium gait strategy. This lends further evidence to the trade-off suggested from previous theoretical studies (see Intro- duction). The constant peak force and constant impulse control strategies both result in a non-steady stance in the drop step, indicated by a positive fore-aft impulse (figure 5, green and blue lines). Thus, gravitational potential energy from the drop perturbation is converted into horizontal kinetic energy, and the running model accelerates. This forward acceleration is in agreement with the experimentally observed dynamics (figure 5C). 3 Comparison between Experimental Data and Simulation Results Simulations of constant peak force or constant impulse policies both result in a reasonably good match between measured and predicted dynamics. The constant peak force policy provides a slightly better match to median peak forces and axial impulse; however analysis of simulation fits across all drop perturbation trials suggest these two policies are equally good at predicting changes in landing conditions (table 3, figure 7). Consequently, we cannot conclusively distinguish between them. To quantitatively compare simulation predictions to experi- mental data, the most relevant parameters are TD leg angle in step 0 and the predicted changes in stance dynamics resulting from the altered landing conditions. The TD leg angle is predicted by applying the optimized swing-leg angular trajectory during the ballistic flight phase. The simulations allow us to evaluate the interaction between swing and stance dynamics, and identify aspects of bird running that match and deviate from the model predictions. To determine which swing-leg control policy was most consistent with guinea fowl behavior, we ran a simulation for each running trial, predicting the drop step dynamics by applying the three control policies to the SLIP model as described in the methods (Methods section 4). For each control policy, table 3 reports the average differences DaTD and root mean squared errors (RMSE) between the predicted touchdown virtual leg angle aTD,Policy and experimentally measured aTD (Methods section 3). Compared to equilibrium gait, both constant peak force and constant axial impulse control result in smaller deviations between predicted and measured TD leg angle DaTD and smaller RMSE, suggesting a more accurate prediction of the TD leg angle across all three step types simulated (level, 21 and 0). Stance phase peak force Faxial,max, axial impulse Iaxial, and fore- aft impulse Ix were simulated by applying the TD conditions resulting from each swing-leg control policy to the SLIP model (table 3). The simulation predictions are compared to experimen- tal data in figure 5, with boxplots showing the distribution of experimental data and colored lines indicating predictions of each control strategy. Simulations of the equilibrium gait policy predict considerable increases in ^Faxial,max and Iaxial during the drop step (step 0), which is not experimentally observed (figure 5). To further illustrate the divergence between the force and equilibrium gait policies, figure 7 shows the swing-leg trajectories predicted by the different control strategies for one constant forward speed _x~2:84 m=s (average experimentally observed forward speed). Contour lines of constant peak force (blue lines) and constant axial impulse (green lines) are plotted as a function of TD leg angle (y-axis) and fall time (x-axis), indicating the trajectories for each control policy (a single predicted swing-leg trajectory follows a single contour line). The red line indicates the leg angle trajectory that leads to equilibrium gait (here indicating swing-leg protraction as a function of fall time). The experimen- tally measured TD leg angles are shown for level running (white circle), 4 cm drop (gray circle) and 6 cm drop (black circle). The experimentally observed TD conditions lie between contour lines for constant peak force (blue) and constant axial impulse (green), Figure 5. Experimental measures of stance dynamics, overlaid with simulation predictions. Boxplots of three stance measures from the running birds: speed corrected axial peak force ^Faxial,max (A), axial impulse Iaxial (B), and fore-aft impulse Ix for level running and the three step types 21, 0, and +1. Asterisks indicate a significant difference (pƒ0:05) compared to level running. See tables 1 and 2 for more detailed statistical results. The colored lines show the simulation predictions for the three swing-leg control policies applied to the drop step: constant peak force (blue), constant impulse (green), and equilibrium gait (red). Swing-leg trajectories optimized for equilibrium gait predict higher ^Faxial,max (A) and Iaxial (B) during the drop step (step 0), which is not experimentally observed. Swing-leg trajectories optimized for constant peak force or constant impulse both result in a good match between measured and predicted dynamics. Analysis of simulation fits across all drop perturbation trials suggest these two policies are equally good at predicting changes in landing conditions (table 3, figure 7). doi:10.1371/journal.pone.0100399.g005 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 9 June 2014 | Volume 9 | Issue 6 | e100399 but differ markedly from the predictions of equilibrium gait. The approximate linearity of the contour lines for constant peak force and constant axial impulse indicate that these policies can be closely approximated by retracting the leg with a constant angular velocity (_a&28 deg=T for peak force control, and _a&26 deg=T for impulse control at the representative forward velocity shown). For the equilibrium gait policy, the simulation predicted swing- leg angular trajectory varies between late-swing retraction and protraction, depending on forward speed. Figure 8 shows the simulation predicted swing-leg angle trajectories resulting in equilibrium gait for forward speeds between _x~½0:5,3:5m=s, with constant leg length and leg stiffness. The simulations predict late-swing retraction for low speeds (_xv1:68 m=s), and protraction for higher speeds(_xw1:68 m=s). For a system with the body mass and virtual leg length of a guinea fowl, running at a forward speed of _x~1:68 m=s, an equilibrium gait can be achieved with a constant leg angle (a~120:1 deg), without adjusting the leg angle during swing (_a~0). Within the observed speed range of guinea fowl, the equilibrium gait policy predicts late-swing protraction. Yet, experimental data show that birds consistently retract their legs in late swing (table 1 and figure 4) across all speeds and step types. Although experimentally observed TD leg angles for steady level running (white circle) lie close to equilibrium gait predictions Table 3. Simulated control strategies compared to experimental data. Control Policy DaTD [deg] RMSE [deg] ^Faxial,max [BW] Iaxial [BW T] Ix [BW T] Level Constant Peak Force 20.6 5.4 2.33 1.00 0.03 Constant Impulse 20.5 3.1 2.46 1.08 0.02 Equilibrium Gait 2.0 5.9 2.56 1.20 0 Step 21 Constant Peak Force 20.2 5.0 2.42 1.03 0.04 Constant Impulse 20.4 3.0 2.53 1.10 0.03 Equilibrium Gait 2.5 5.6 2.80 1.22 0 Step 0 Constant Peak Force 20.3 4.5 2.42 0.92 0.08 Constant Impulse 2.0 4.2 2.75 1.10 0.06 Equilibrium Gait 9.0 10.7 4.79 2.21 0 Difference DaTD and root mean squared error (RMSE) of the predicted virtual leg angle at TD aTD,Policy and the experimentally measured virtual leg angle at TD aTD. Axial peak force ^Faxial,max, axial impulse Iaxial, and fore-aft impulse Ix are the predicted values of the corresponding control strategies. Compared to the equilibrium gait strategy, the RMSE suggest that both constant peak force and constant impulse control predict the TD leg angle more accurately. doi:10.1371/journal.pone.0100399.t003 Figure 6. Representative simulations illustrating the divergence between equilibrium gait (steady gait) and constant peak force control strategies. CoM trajectories (A) and force profiles (B) of the simulation results for two swing-leg control strategies: constant peak force (blue solid lines) and equilibrium gait (red dashed lines). The equilibrium gait strategy achieves steady dynamics but demands high forces; whereas the constant peak force strategy results in non-steady dynamics in the drop step, and requires adjustment in subsequent steps to return to a steady gait. doi:10.1371/journal.pone.0100399.g006 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 10 June 2014 | Volume 9 | Issue 6 | e100399 (red line), there is no evidence that the swing-leg trajectory directly targets equilibrium gait, because the drop perturbations lead to a sharp deviation from equilibrium gait predictions. Instead, the results suggest that the guinea fowl behavior more closely match predictions of swing-leg trajectory optimized to maintain constant peak leg force or constant leg impulse. Discussion Perturbation experiments [13,17] and theoretical models of walking and running [14–16,18–20,22] have suggested swing-leg trajectory as a critical target of control for legged locomotion because stance dynamics are highly sensitive to landing conditions. Swing-leg trajectory influences the timing of ground contact, the landing leg posture and body velocity at contact. These landing conditions, in turn, influence stability [14–16,18], robustness [19], leg work [19,20], disturbance rejection and collision impact energy losses [18]. Swing-leg trajectory can be optimised for consistent leg loading and economy, or alternatively, for steady body dynamics, but not all of these simultaneously [15,16,18–20]. We investigated how running guinea fowl manage this potential trade-off by measuring their ‘optimized’ locomotor strategy for negotiating a visible and well-practiced step down in terrain. The simulation results in figures 6 and 5 provide further evidence of the suggested trade-off in swing-leg trajectory. The specific swing-leg angular trajectory used by running guinea fowl is consistent with task-level priority to regulate leg loading (limiting fluctuations in peak force and impulse), rather than priority to maintain steady body dynamics. The birds’ swing-leg angular trajectory is consistent with both the constant peak force and constant impulse policies, but clearly deviates from the predictions of the equilibrium gait policy. The constant peak force and constant leg axial impulse policies both predict leg retraction in late swing with nearly constant angular velocity (figure 7). Previous studies have shown that running animals tend to retract the leg in late swing [14,17,48]; however, these studies could not explain the specific leg retraction velocities used by animals, because a wide range of retraction velocities can provide stability [14–16]. ‘Stability’ simply refers to whether or not the system recovers—whether a deviation in body dynamics decays (stable) or grows (unstable) over time [49,50]. Priority for stability alone is not sufficient to predict a specific leg angular trajectory. The equilibrium gait policy predicts a specific Figure 7. Model predicted late-swing leg angular trajectories, in comparison with experimental data. Predicted swing-leg trajectories, shown as leg angle against fall time (time from apex until TD), derived from SLIP simulations to achieve constant peak force (blue solid lines), constant axial impulse (green dashed lines), or equilibrium gait (red line) at touchdown. Predictions are for a single forward speed _x~2:84 m=s. The thick peak force (blue) and impulse (green dashed) contours indicate the predicted swing-leg trajectories, with thinner contours illustrating the gradient in force and impulse as the trajectory deviates from this. The mean measured leg trajectory is overlaid (dotted black line), along with the mean TD conditions for level running (white circle), 4 cm drop (gray circle) and 6 cm drop (black circle). The equilibrium gait trajectory (red) crosses loading contours, leading to increased force and impulse. The linearity of the constant peak force and impulse contours indicates that these strategies can be approximated by leg retraction with a constant angular velocity, whereas equilibrium gait requires leg protraction. The experimental data follows constant loading contours, suggest that guinea fowl do not use swing-leg trajectory to target equilibrium gait. doi:10.1371/journal.pone.0100399.g007 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 11 June 2014 | Volume 9 | Issue 6 | e100399 leg angular trajectory by targeting a perfectly steady gait, which can theoretically provide perfect disturbance rejection in the face of terrain height variation [20,47]. However, this policy can demand large increases in force and impulse in the stance phase. Furthermore, the equilibrium gait policy can predict either leg protraction or retraction of the leg in late swing (figure 8). While stable spring mass running with swing-leg protraction is possible (with appropriately tuned leg stiffness) [16], this strategy would result in higher leg impacts due to increased velocity of the foot with respect to the ground [15] (e.g., the opposite of ‘ground speed matching’, [48]). This might explain why, to our knowledge, only swing-leg retraction, never protraction, has been experimentally observed in bipedal locomotion of humans [51] and birds [12,16,17]. We found that stance dynamics immediately following the drop perturbation (step 0) are consistent with a passive energy- conservative leg model, albeit with a non-steady response in which gravitational potential energy is converted to kinetic energy, causing forward acceleration. In fact, the overall body dynamics of step 0 are remarkably similar to those of an unexpected drop step [17], despite evidence of anticipatory changes to gait in the drop terrain. The anticipatory adjustments include small but significant changes in the nominal gait of step 21 preceding the drop (table 2), and an increase in effective leg stiffness across all steps in the drop terrain (figure 4). These findings suggest that guinea fowl tune gait dynamics depending on context including the anticipated ‘rough- ness’ of terrain. Nonetheless, leg angular trajectory remains remarkably con- stant and rhythmic across steps within each terrain (figure 3), suggesting that birds target a consistent optimized trajectory within a terrain context and avoid step-by-step adjustments. A prescribed swing-leg trajectory has potential to be implemented through feed- forward control, with minimal feedback, circumventing neuro- muscular delays. However, our results do not reveal the underlying neural control mechanisms used to achieve the observed swing-leg trajectory. A consistent leg angular trajectory could be achieved through a combination of feedforward and feedback mechanisms, making use of internal models of dynamics as well as vestibular, visual and proprioceptive sensory information. Whatever the underlying control mechanisms, our findings are consistent with the idea that animals optimize swing-leg trajectory to achieve well- defined intrinsic-dynamic characteristics at the swing-stance transition, to bridge neuromuscular delays and minimize the need for rapid neural modulation. Although the results confirm that step 0 dynamics can be well approximated by a passive, energetically conservative leg model, the dynamics of the 2nd stance (step +1) clearly indicate net energy absorption, which cannot be achieved with a passive model. Consequently, a full dynamic model of the birds’ recovery over several steps requires a more sophisticated stance leg model that includes actuation. It will be interesting in future work to further investigate alternative task-level templates of running that allow for non-conservative stance dynamics following terrain perturbations. Actuated template models have been proposed and analyzed from a theoretical perspective [40–45], but it is not yet clear which of these is most appropriate for animal legged locomotion. Elabora- tions of stance models were not considered here because we were primarily focused on the effects of swing-leg trajectory on the swing-stance transition. Non-conservative stance models would have confounded the interpretation of swing-leg trajectories. The initial step down response (step 0) is energetically conservative and matches well with SLIP leg loading predictions, so we concluded that a more complex model was not justified for the current study. Nonetheless, future work should investigate more complex stance models to further explore the interactions between swing and stance dynamics in non-steady locomotion, in particular to understand the full time course of recovery from a perturbation. Additionally, we observed asymmetry in the force trajectory across all running conditions—which has also been noted previously [52] and likely reflects the complex underlying musculoskeletal structure and dynamics of animal legs. The passive SLIP model does not predict the precise shape of the biologically observed leg force trajectory, because it is also influenced by factors such as damping in tissues, muscle contractile properties and musculoskeletal gearing effects. The SLIP model serves only as a general ‘template’ of the overall body dynamics of running gaits [39], and does not reflect the specific underlying neuromuscular and musculoskeletal mechanisms. Nonetheless, template models such as SLIP provide a convenient approxima- tion of legged locomotion because animals tend to use periodic gaits with ground reaction forces and body dynamics that can be approximated by a point mass body with massless legs that resist only compressive loads [34–37]. The SLIP model is not the only model that provides a reductionist approximation of locomotor dynamics [40–45,49,53–56]; however, it is the most widely validated choice for simulations of running (see Methods section 3). These caveats aside, we have found that a simple reductionist model can reproduce many aspects of avian running dynamics during negotiation of a drop in terrain, by optimizing swing-leg angular trajectory to target landing conditions that meet the specific task-level priority of regulating stance leg loading. The observed strategy of minimizing fluctuations in peak force and impulse may also minimize energy cost of transport. Cost of transport is influenced by both muscular force and work [6,56], which is therefore strongly related to ground reaction force [57]. Additionally, a separate simulation study has compared swing-leg trajectories optimized for force, impulse and leg work, and found that all three of these policies predict similar swing-leg trajectories, yet diverge from the predictions of an equilibrium gait policy [20]. Thus, it appears that load regulation and economy are closely aligned priorities. A swing-leg trajectory optimized to regulate leg loading may have the dual benefits of minimizing injury risk and maximizing economy of uneven terrain locomotion. Figure 8. Late-swing leg angular trajectories predicted for the equilibrium gait policy, targeting steady gait. The equilibrium gait policy predicts a shift from late-swing retraction to protraction with increasing speed. Shown are the swing-leg angle trajectories predicted from simulations optimized for equilibrium gait, for a range of speeds _x~½0:5,3:5m=s. The simulations predict late-swing leg retraction for low speeds ( _xv1:68 m=s), and protraction for higher speeds (_xw1:68 m=s). doi:10.1371/journal.pone.0100399.g008 Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 12 June 2014 | Volume 9 | Issue 6 | e100399 The majority of animal locomotion studies have focused on steady-state locomotion, and many studies either implicitly or explicitly assumed that steady gait is an overriding priority and therefore a direct target of active control. While animals must avoid falling in uneven terrain, ‘stability’ and ‘disturbance rejection’ or ‘steadiness’ of gait may not be exceptionally pressing priorities for the control of swing-leg trajectory compared to other task-level demands, such as injury avoidance and economy. Applying a swing-leg trajectory that enforces a steady gait could dramatically increase the peak force and impulse experienced by the leg in the presence of a terrain drop. These forces could easily exceed the safety factors of animal musculoskeletal tissues, which are around 2–46 peak force of steady locomotion [25,26]. Therefore, minimizing fluctuations in peak force and impulse to prevent damage to musculoskeletal structures might be a more pressing priority than immediate recovery to a nominal steady gait following perturbations. Nonetheless, disturbance rejection is likely an important priority over slightly longer timescales. This conclusion is supported by the experimental finding that guinea fowl consistently recover from terrain perturbations within about 2–3 strides [10,11,13], but do not exhibit perfect, immediate disturbance rejection, even for small terrain perturbations [13]. We suggest that the immediate imperatives of swing-leg control in animal legged locomotion are related to injury avoidance and economy, not immediate stabilization to a nominal steady gait, while stance phase mechanisms (e.g., energy absorption/insertion) facilitate recovery to steady gait over multiple steps. Our simulations suggest a simple method for generating target swing-leg trajectories for implementation in legged robots to achieve performances similar to that of running animals. The ‘equilibrium gait’ policy has been suggested for legged robots for its disturbance rejection properties [30,47]; however, we suggest that it may be undesirable for systems with significant force limitations. A separate recent paper explores in more detail simulations of running dynamics with multiple alternative swing- leg control policies [20], and this paper also further discusses potential implications for bio-inspired robots. This systematic approach of comparing predictions based on multiple potential task-level priorities could help engineers design and control robots to benefit from passive-dynamic structures, minimize actuator demands and minimize control effort. Conclusions We have presented a novel approach combining simulations and experiment that allows us to investigate the task-level priorities in non-steady animal locomotion, including disturbance rejection, injury avoidance and economy. Guinea fowl negotiate a down- ward step using unsteady dynamics with forward acceleration, and recover to steady gait in subsequent steps. ‘Steadiness’ of gait does not appear to be the direct or immediate priority governing swing- leg trajectory used by running animals. Our results suggest, instead, that guinea fowl use swing-leg trajectories that reflect priority for load regulation, which may facilitate injury avoidance and economy in uneven terrain. Supporting Information Text S1 List of symbols, terms and definitions. (PDF) Acknowledgments The authors thank D. Renjewski, S. D. Wilshin and J. Gordon for fruitful discussions and feedback on the manuscript. Author Contributions Conceived and designed the experiments: MAD JWH. Performed the experiments: YB ABJ. Analyzed the data: YB HRV MAD. Contributed reagents/materials/analysis tools: ABJ HRV. Wrote the paper: YB HRV MAD JWH. Discussed and interpreted data: YB HRV ABJ CMH MAD JWH. References 1. Cavagna GA, Saibene FP, Margaria R (1964) Mechanical work in running. Journal of Applied Physiology 19: 249–256. 2. Fedak MA, Heglund NC, Taylor CR (1982) Energetics and mechanics of terrestrial locomotion ii. kinetic energy changes of the limbs and body as a function of speed and body size in birds and mammals. Journal of Experimental Biology 97: 23–40. 3. Heglund NC, Cavagna GA, Taylor CR (1982) Energetics and mechanics of terrestrial locomotion iii. energy changes of the centre of mass as a function of speed and body size in birds and mammals. Journal of Experimental Biology 97: 41–56. 4. Heglund NC, Fedak MA, Taylor CR, Cavagna GA (1982) Energetics and mechanics of terrestrial locomotion iv. total mechanical energy changes as a function of speed and body size in birds and mammals. Journal of Experimental Biology 97: 57–66. 5. Farley CT, Glasheen J, McMahon TA (1993) Running springs: speed and animal size. Journal of Experimental Biology 185: 71–86. 6. Roberts TJ, Kram R, Weyand PG, Taylor CR (1998) Energetics of bipedal running. I. Metabolic cost of generating force. Journal of Experimental Biology 201: 2745–2751. 7. Minetti AE, Alexander RM (1997) A theory of metabolic costs for bipedal gaits. Journal of Theoretical Biology 186: 467–476. 8. Rubenson J, Lloyd DG, Heliams DB, Besier TF, Fournier PA (2011) Adaptations for economical bipedal running: the effect of limb structure on three-dimensional joint mechanics. Journal of The Royal Society Interface 8: 740–755. 9. Moritz CT, Farley CT (2004) Passive dynamics change leg mechanics for an unexpected surface during human hopping. Journal of Applied Physiology 97: 1313–1322. 10. Daley MA, Voloshina A, Biewener AA (2009) The role of intrinsic muscle mechanics in the neuromuscular control of stable running in the guinea fowl. Journal of Physiology 587: 2693–2707. 11. Daley MA, Biewener AA (2011) Leg muscles that mediate stability: mechanics and control of two distal extensor muscles during obstacle negotiation in the guinea fowl. Philosophical Transactions of the Royal Society B 366: 1580–1591. 12. Daley MA, Felix G, Biewener AA (2007) Running stability is enhanced by a proximo-distal gradient in joint neuromechanical control. Journal of Experi- mental Biology 210: 383–394. 13. Birn-Jeffery A, Daley MA (2012) Birds achieve high robustness in uneven terrain through active control of landing conditions. Journal of Experimental Biology 215: 2117–2127. 14. Seyfarth A, Geyer H, Herr H (2003) Swing-leg retraction: A simple control model for stable running. Journal of Experimental Biology 206: 2547–2555. 15. Blum Y, Lipfert SW, Rummel J, Seyfarth A (2010) Swing leg control in human running. Bioinspiration & Biomimetics 5: 026006. 16. Blum Y, Birn-Jeffery A, Daley MA, Seyfarth A (2011) Does a crouched leg posture enhance running stability and robustness? Journal of Theoretical Biology 281: 97–106. 17. Daley MA, Biewener AA (2006) Running over rough terrain reveals limb control for intrinsic stability. Proceedings of the National Academy of Sciences of the USA 103: 15681–15686. 18. Karssen JGD, Haberland M, Wisse M, Kim S (2011) The optimal swing-leg retraction rate for running. In: IEEE International Conference on Robotics and Automation (ICRA). pp. 4000–4006. 19. Daley MA, Usherwood JR (2010) Two explanations for the compliant running paradox: Reduced work of bouncing viscera and increased stability in uneven terrain. Biology Letters 6: 418–421. 20. Vejdani HR, Blum Y, Daley MA, Hurst JW (2013) Bio-inspired swing leg control for spring-mass robots running on ground with unexpected height disturbance. Bioinspiration & Biomimetics 8: 046006. 21. Hobbelen DGE, Wisse M (2007) A disturbance rejection measure for limit cycle walkers: The gait sensitivity norm. IEEE Transactions on Robotics 23: 1213– 1224. Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 13 June 2014 | Volume 9 | Issue 6 | e100399 22. Wisse M, Schwab AL, Van der Linde RQ, Van der Helm FCT (2005) How to keep from falling forward: Elementary swing leg action for passive dynamic walkers. IEEE Transactions on Robotics 21: 393–401. 23. Byl K, Tedrake R (2009) Metastable walking machines. International Journal of Robotics Research 28: 1040–1064. 24. Ferris DP, Liang K, Farley CT (1999) Runners adjust leg stiffness for their first step on a new running surface. Journal of Biomechanics 32: 787–794. 25. Biewener AA (1989) Scaling body support in mammals: Limb posture and muscle mechanics. Science 245: 45–48. 26. Biewener AA (2005) Biomechanical consequences of scaling. Journal of Experimental Biology 208: 1665–1676. 27. Ker RF, Alexander RM, Bennett MB (1988) Why are mammalian tendons so thick? Journal of Zoology 216: 309–324. 28. Dow SM, Leendertz JA, Silver IA, Goodship AE (1991) Identification of subclinical tendon injury from ground reaction force analysis. Equine Veterinary Journal 23: 266–272. 29. Harrison SM, Whitton RC, Kawcak CE, Stover SM, Pandy MG (2010) Relationship between muscle forces, joint loading and utilization of elastic strain energy in equine locomotion. The Journal of Experimental Biology 213: 3998– 4009. 30. Ernst M, Geyer H, Blickhan R (2009) Spring-legged locomotion on uneven ground: a control approach to keep the running speed constant. In: International Conference on Climbing and Walking Robots. Istanbul, Turkey. 31. More HL, Hutchinson JR, Collins DF, Weber DJ, Aung SKH, et al. (2010) Scaling of sensorimotor control in terrestrial mammals. Proceedings of the Royal Society B: Biological Sciences 277: 3563–3568. 32. Cavagna GA, Franzetti P, Heglund NC, Willems P (1988) The determinants of the step frequency in running, trotting and hopping in man and other vertebrates. Journal of Physiology 399: 81–92. 33. Gatesy SM, Biewener AA (1991) Bipedal locomotion: Effects of speed, size and limb posture in birds and humans. Journal of Zoology 224: 127–147. 34. Blickhan R (1989) The spring-mass model for running and hopping. Journal of Biomechanics 22: 1217–1227. 35. McMahon TA, Cheng GC (1990) The mechanics of running: How does stiffness couple with speed? Journal of Biomechanics 23: 65–78. 36. Alexander RM (2002) Tendon elasticity and muscle function. Comparative Biochemistry and Physiology A 133: 1001–1011. 37. Geyer H, Seyfarth A, Blickhan R (2006) Compliant leg behaviour explains basic dynamics of walking and running. Proceedings of the Royal Society B 273: 2861–2867. 38. Poulakakis I, Papadopoulos E, Buehler M (2006) On the stability of the passive dynamics of quadrupedal running with a bounding gait. International Journal of Robotics Research 25: 669–687. 39. Full RJ, Koditschek DE (1999) Templates and anchors: Neuromechanical hypotheses of legged locomotion on land. Journal of Experimental Biology 202: 3325–3332. 40. Seipel JE, Holmes PJ, Full RJ (2004) Dynamics and stability of insect locomotion: a hexapedal model for horizontal plane motions. Biological Cybernetics 91: 76–90. 41. Seipel JE, Holmes P (2007) A simple model for clock-actuated legged locomotion. Regular and Chaotic Dynamics 12: 502–520. 42. Schmitt J, Clark J (2009) Modeling posture-dependent leg actuation in sagittal plane locomotion. Bioinspiration & Biomimetics 4: 046005. 43. Spence AJ, Revzen S, Seipel J, Mullens C, Full RJ (2010) Insects running on elastic surfaces. Journal of Experimental Biology 213: 1907–1920. 44. Andrews B, Miller B, Schmitt J, Clark JE (2011) Running over unknown rough terrain with a one-legged planar robot. Bioinspiration & Biomimetics 6: 026009. 45. Riese S, Seyfarth A (2012) Stance leg control: variation of leg parameters supports stable hopping. Bioinspiration & Biomimetics 7: 016006. 46. Grimmer S, Ernst M, Gu¨nther M, Blickhan R (2008) Running on uneven ground: Leg adjustment to vertical steps and self-stability. Journal of Experimental Biology 211: 2989–3000. 47. Ernst M, Geyer H, Blickhan R (2012) Extension and customization of self- stability control in compliant legged systems. Bioinspiration & Biomimetics 7: 046002. 48. Herr HM, McMahon TA (2001) A galloping horse model. International Journal of Robotics Research 20: 26–37. 49. McGeer T (1993) Dynamics and control of bipedal locomotion. Journal of Theoretical Biology 163: 277–314. 50. Dingwell JB, Kang HG (2007) Differences between local and orbital dynamic stability during human walking. Journal of Biomechanical Engineering 129: 586–593. 51. De Wit B, De Clercq D, Aerts P (2000) Biomechanical analysis of the stance phase during barefoot and shod running. Journal of Biomechanics 33: 269–278. 52. Cavagna GA (2006) The landing-take-off asymmetry in human running. Journal of Experimental Biology 209: 4051–4060. 53. Garcia MS, Chatterjee A, Ruina A (2000) Efficiency, speed, and scaling of two- dimensional passive-dynamic walking. Dynamics and Stability of Systems 15: 75–99. 54. Kuo AD (2002) Energetics of actively powered locomotion using the simplest walking model. Journal of Biomechanical Engineering 124: 113–120. 55. Srinivasan M, Ruina A (2006) Computer optimization of a minimal biped model discovers walking and running. Nature 439: 72–75. 56. Srinivasan M (2010) Fifteen observations on the structure of energy-minimizing gaits in many simple biped models. Journal of the Royal Society Interface 8: 74– 98. 57. Kram R, Taylor CR (1990) Energetics of running: a new perspective. Nature 346: 265–267. Swing Leg Trajectory of Running Guinea Fowl PLOS ONE | www.plosone.org 14 June 2014 | Volume 9 | Issue 6 | e100399

Swing-leg trajectory of running guinea fowl suggests task-level priority of force regulation rather than disturbance rejection.

06-30-2014

Blum, Yvonne,Vejdani, Hamid R,Birn-Jeffery, Aleksandra V,Hubicki, Christian M,Hurst, Jonathan W,Daley, Monica A

eng

PMC3782489

Anatomically Asymmetrical Runners Move More Asymmetrically at the Same Metabolic Cost Elena Seminati1*, Francesca Nardello2, Paola Zamparo2, Luca P. Ardigo` 2, Niccolo` Faccioli3, Alberto E. Minetti1 1 Department of Pathophysiology and Transplantation, Faculty of Medicine, University of Milan, Milan, Italy, 2 Department of Neurological and Movement Sciences, School of Exercise and Sport Sciences, University of Verona, Verona, Italy, 3 Department of Pathology and Diagnostics, Section of Radiology, University of Verona, Verona, Italy Abstract We hypothesized that, as occurring in cars, body structural asymmetries could generate asymmetry in the kinematics/ dynamics of locomotion, ending up in a higher metabolic cost of transport, i.e. more ‘fuel’ needed to travel a given distance. Previous studies found the asymmetries in horses’ body negatively correlated with galloping performance. In this investigation, we analyzed anatomical differences between the left and right lower limbs as a whole by performing 3D cross-correlation of Magnetic Resonance Images of 19 male runners, clustered as Untrained Runners, Occasional Runners and Skilled Runners. Running kinematics of their body centre of mass were obtained from the body segments coordinates measured by a 3D motion capture system at incremental running velocities on a treadmill. A recent mathematical procedure quantified the asymmetry of the body centre of mass trajectory between the left and right steps. During the same sessions, runners’ metabolic consumption was measured and the cost of transport was calculated. No correlations were found between anatomical/kinematic variables and the metabolic cost of transport, regardless of the training experience. However, anatomical symmetry significant correlated to the kinematic symmetry, and the most trained subjects showed the highest level of kinematic symmetry during running. Results suggest that despite the significant effects of anatomical asymmetry on kinematics, either those changes are too small to affect economy or some plastic compensation in the locomotor system mitigates the hypothesized change in energy expenditure of running. Citation: Seminati E, Nardello F, Zamparo P, Ardigo` LP, Faccioli N, et al. (2013) Anatomically Asymmetrical Runners Move More Asymmetrically at the Same Metabolic Cost. PLoS ONE 8(9): e74134. doi:10.1371/journal.pone.0074134 Editor: David Carrier, University of Utah, United States of America Received March 20, 2013; Accepted July 27, 2013; Published September 24, 2013 Copyright:  2013 Seminati et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: The project was supported by the Department of Pathophysiology and Transplantation -Human Physiology section- University of Milan. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: elena.seminati@unimi.it Introduction The symmetry between the left and right sides of the body plays an important role in legged locomotion. The symmetrical behaviour of lower limbs during gait has often been taken for granted, mainly for simplicity in data collection and analysis, while the lack of it was frequently considered as an indicator of gait pathology [1]. Differently from what expected, healthy human gait is rather asymmetrical [2,3]. This seems to reflect a functional difference inherently associated to the laterality of the dominant side characterising each individual [4,5]. This topic was intro- duced more than 80 years ago by Lund [6] who showed the effects of structural/anatomical asymmetry on lateral drift in human locomotion. The same experiments were recently repeated and supported the hypothesis of a relationship between leg length inequality and asymmetry in locomotion [7–9]. Body symmetry can be further modulated in sports: depending on the discipline, relevant muscles become asymmetrically different (tennis, fencing, throwing, etc.), or they are required to reach similar hypertrophy (ice-skating, downhill skiing, front crawl, etc.) on the two sides of the sagittal plane. Thus, body changes towards or from symmetry are not just the consequence of genetics and laterality, being also caused by specific training protocols. As the concept of symmetry has an important influence in human locomotion, it plays a key role in the design and maintenance of vehicles, which are periodically inspected and serviced to guarantee wheel balance and homogeneous tyre wearing, in order to reduce fuel consumption and ensure a safe drive. Would it be the same for human running? Can an anatomical/structural asymmetry of the human body cause kinematic/dynamic asymmetry of locomotion? Also, can structural or functional asymmetries be related to some increase of the metabolic cost of transport? Several authors studied symmetry in locomotion in humans [1– 4,10,11] and also in animals [12], but only few of them investigated the possible interaction between symmetry and energy saving. Manning and collaborators found negative correlations between anatomical symmetry and race time during competitions, both in human running and in galloping horses [13,14]. These preliminary findings encouraged us to study the possible interac- tions between different kinds of symmetry (anatomical and dynamical) and the human running performance, not only in term of race time, but also of energy saving. In the present study, we investigate the relationship between the cost of transport (C) PLOS ONE | www.plosone.org 1 September 2013 | Volume 8 | Issue 9 | e74134 while running at different increasing velocities and individual anatomical and dynamical symmetries in three differently trained groups of subjects, with the idea that ‘race cars’ should more strongly rely on symmetry than ordinary ‘automobiles’. Materials and Methods Subjects Nineteen healthy male subjects volunteered to participate this investigation. Exclusion criteria included neurological or muscu- loskeletal pathologies affecting running ability. The institutional ethics committee of the University of Milano had approved all methods and procedures, and subjects gave their written informed consent (approved by the same committee) prior to the start of testing. We clustered participants into three different groups, based on their specific running ability: N group 1, (n = 7): Untrained Runners (UR), who practiced sport (not specifically running) 3 times per week (less than 2 hours per week) N group 2, (n = 7): Occasional Runners (OR), fit athletes, who trained more than 3 times per week, (between 2 and 6 hours per week). Each of them had previously participated in a national competition (half marathon or 10 km competition) N group 3, (n = 5): Skilled Runners (SR), master athletes who trained more than 3 times per week (at least 6 hours per week); they were marathon runners, with a mean performance time of 2 h 44 min 24 s 610 min 12 s standard deviation (SD). Anthropometric characteristic of the different subject groups are shown in Table 1. MR Dataset and 3D Images Processing In order to evaluate the anatomical symmetries, each partici- pant underwent Magnetic Resonance (MR) imaging. Subjects were adjusted in a supine position as to preserve the maximal body symmetry in the sagittal plane. MR scans were performed with a 1.5-T superconductive magnet (Siemens, Erlangen, Germany). In all subjects multiplanar T1-weighted Spin-echo sequences were obtained (TE 11, TR 565, flip angle 90u), on a coronal plane for three different anatomical districts: Pelvis district (PD), Upper-Leg district (UD), including thigh and knee, Lower-Leg district (LD), including calf and ankle, with slice thickness of 4 mm. The matrix was 3206320 and the field of view (FOV) was 4606460. Total examination time was less than 7 minutes (36 coronal slices for each district). All the recorded images (saved in DICOM format) were subsequently analyzed with a custom, ad hoc program written in LabVIEW 8.6 (National Instrument, Austin, Texas, USA). The procedure we implemented exports, for each districts, 36 MR images (slices) as two-dimensional matrix of 3206320 pixels, each of which 1.4461.44 mm, and includes several post-processing steps, as shown in Figure 1. The 36 coronal slices, for every district, assembled together, re- create a three-dimensional (3D) volume, whose elements (voxel) are values corresponding to a grey level intensity (8 bit scale), reflecting proton density, (Figure 1a). In order to compare the subject’s left lower limb with the right one, firstly, the initial 3D volume has to be split in two separated volumes, right volume (Rv) and left volume (Lv), (Figure 1b). Successively the Lv is specularly reflected, with respect to the sagittal plane (Figure 1c), whilst the Rv is bordered by zero intensity voxel (Figure 1d), through a zero- padding operation, so that the left reflected volume (Lrv) can be virtually superimposed on the Rv (Figure 1e), and moved along the three axes in order to find the best matching overlap and to evaluate the ‘overall’ similarity (i.e. symmetry) between the two limbs. To achieve this aim the algorithm performs a 3D correlation between the contents of the two respective anatomical volumes. Table 1. Subject characteristics. UR OR SR Participants (n) 7 7 5 Age (years) 33.1613.2 31.9611.8 42.667.4 Body Mass (kg) 70.663.4 67.366.1 68.264.9 Height (cm) 175.964.7 177.364.0 177.864.4 Right leg length (cm) 83.163.6 84.064.1 85.866.3 Left leg length (cm) 82.863.7 83.063.7 84.867.2 Leg length discrepancy (cm) 1.160.7 1.060.8 1.361.0 Number of participants, mean 6 SD for age (yrs), body mass (kg), height (cm), right and left leg length (cm), and leg length discrepancy (LLD) (cm) for the 3 different groups of subjects: Untrained runners (UR), Occasional runners (OR) and Skilled runners (SR). doi:10.1371/journal.pone.0074134.t001 Figure 1. Principal steps involved in the 3D cross-correlation algorithm. a) The 36 slices of the MR sequence, create a 3D volume whose sizes are laterally indicated, b) right volume (Rv) and left volumes (Lv) separated, c) left reflected volume (Lrv) on the sagittal plane (in the mirror), d) zero-padding operation around right volume, e) Lrv superimposed to Rv in order to find the position that maximize the cross-correlation value. doi:10.1371/journal.pone.0074134.g001 Anatomical Asymmetries & Running Dynamics/Economy PLOS ONE | www.plosone.org 2 September 2013 | Volume 8 | Issue 9 | e74134 The correlation between two signals (cross-correlation) is a standard approach for signal processing and it has been recently designed in 3D in order to consider simultaneously the full anatomical volume information, to assist radiologists in providing correct diagnosis of metastases within the lungs [15,16] or brain [17], for instance. Following Lewis’ approach [18], a normalised cross-correlation coefficient (ri,j,k), was adopted to identify the symmetry degree between the 3D split volumes: ri,j,k~ P x,y,z Rv(x,y,z){Rvi,j,k  : Lrv(x{i,y{j,z{k){Lrv   ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P x,y,z Rv(x,y,z){Rvi,j,k  2: P x,y,z Lrv(x{i,y{j,z{k){Lrv  2 q where Lrv and Rvi,j,k are the voxel mean value of the left reflected volume and the right volume, respectively. The two volumes are virtually superimposed at coordinates i, j and k, and calculations are performed for all pairs of corresponding voxels along x, y and z axes. For every subject and each anatomical district we evaluated the maximal cross correlation value (rmax) (i.e. the value corresponding to the best overlap between right and left reflected volumes). This coefficient can assume a range of values between 21 and 1, depending upon the similarity of the 3D analyzed volumes, where a value of 1 indicates an exact matching of the Lrv with the Rv, a value of 21 indicates opposite grey values for voxels in Lrv with respect to Rv, and a value of 0 indicates no correlation between the two volumes. Figure 2. Examples of obtained cross-correlation values plotted versus iterations. Cross correlation values (r) of all iterations (134,400 overlap positions = 28 (i)660 (j)680 (k), between right volume and left reflected volume); a) comparison between two bottles, filled with the same volume of water (rmax =0.99); b) comparison between the right upper leg of a subject and the left upper leg of a different subject (rmax =0.51); c) comparison between right and left upper legs of the same subject. Inset: enlargement of cross correlation pattern showing the inner processing loop (i coordinates). doi:10.1371/journal.pone.0074134.g002 Anatomical Asymmetries & Running Dynamics/Economy PLOS ONE | www.plosone.org 3 September 2013 | Volume 8 | Issue 9 | e74134 Software reliability and accuracy were validated by comparing two identical bottles filled up with water (see Figure 2a), resulting in a maximal cross correlation value of rmax = 0.99. The algorithm provided a value of rmax = 1 only when the right volume of a specific subject was compared with itself, while the lowest value of rmax was obtained when the right volume of a specific subject was compared with the left reflected volume of an other different subject (rmax = 0.51, as shown in Figure 2b). We evaluated for every subject a single maximal cross correlation value (rmax) for each district, (rmax(PD) for Pelvis district, rmax(UD) for Upper-Leg district and rmax (LD) for Lower- Leg district) and secondly a ‘global’ anatomical cross correlation value (rmax) as the mean of the three districts: rmax~ rmax(PD)zrmax(UD)zrmax(LD) 3 Kinematics In order to capture kinematic functional symmetries on many steps, all the subjects performed level shod running on a treadmill (h/p/Cosmos Saturn 4.0, Germany). Human body has been modelled as a series of linked, rigid segments: 18 reflective markers were placed bilaterally on anatomical landmark points (immediately anterior to ear tragus, shoulder, elbow, wrist, greater trochanter, lateral epicondyle of femur, lateral malleolus, calcaneus, and 5th metatarsal head) and their 3D position was captured at 100 Hz, using an eight-camera Vicon MX optoelectronic system (Vicon, Oxford, UK). In this way, 12 body segments were defined [19]. After a brief period of familiarization on the treadmill, each subject ran at six different incremental speeds: from 2.22 m/s to 5 m/s, step 0.56 m/s. Each speed was maintained for at least 5 min, with a rest period of at least 5 min between successive trials. The 3D recorded coordinates of the 12 segments, together with the anthropometric tables [20,21], were used to compute the experimental trajectory of the Body Centre of Mass (BCOM). Successively, we adopted a recent mathematical method [22,23] simultaneously capturing the spatial and dynamical features of that 3D BCoM trajectory, which allows to quantify dynamical symmetry indices of locomotion in the 3 spatial axes; by having sampled the body motion on a treadmill, the trajectory of the BCOM can be represented by a closed 3D loops (Lissajous contours), representing its displacement with respect to the average position. The 3D trajectory is mathematically defined by a 6-harmonic Fourier series, whose coefficients are used to calculate the Dynamical Symmetry Indices SIx (for progression axis), SIy (for vertical axis) and SIz (for lateral axis). The motion of the BCOM is expected to exhibit perfect right–left symmetry if it contained just even harmonics in the progression and y vertical directions, and just odd harmonics in the lateral direction, as within a stride it oscillates twice in the sagittal (y–x) plane and only once in the horizontal (x–z) plane. Dynamical Symmetry indices are then averaged among the strides number (n) as to obtain for each velocity and each subject: SIx~ Pn j~1 SIx j n SIy~ Pn j~1 SIy j n SIz~ Pn j~1 SIz j n Table 2. Statistical correlation matrix results between variable pairs. Anatomical Symmetry Dynamical Symmetry Economy rmax (PD) rmax (UD) rmax (LD) rmax SIx SIy SIz GI C Anatomical Sym. rmax (PD) 1 0.501*; 0.040 0.427; 0.087 0.871**; 0.000 0.651**; 0.005 0.110; 0.675 20.094; 0.719 0.606**; 0.010 0.157; 0.547 rmax (UD) 1 0.507*; 0.038 0.782**; 0.000 0.322; 0.208 0.003; 0.990 0.020; 0.940 0.357; 0.160 20.114; 0.662 rmax (LD) 1 0.748**; 0.001 0.045; 0.863 0.103; 0.694 0.304; 0.236 0.046; 0.860 0.059; 0.822 rmax 1 0.487*; 0.048 0.095; 0.716 0.055; 0.834 0.473; 0.055 0.072; 0.785 Dynamical Sym. SIx 1 0.186; 0.447 0.009; 0.972 0.992**; 0.000 0.005; 0.983 SIy 1 0.617**; 0.005 0.186; 0.445 0.105; 0.668 SIz 1 0.012; 0.959 0.211; 0.385 GI 1 20.001; 0.995 Economy C 1 Pearson Correlation coefficient is presented together with the relative p-value for the following parameters: maximal cross correlation values for each anatomical district (rmax(PD), rmax(UD) and rmax(LD)), global anatomical cross correlation value (rmax), dynamical symmetry indices for each direction (SIx, SIy and SIz), Global Symmetry Index (GI) averaged among the different running speeds for each subject and metabolic Cost of transport (C). Values in bold indicate significant correlations (* = p,0.05, ** = p,0.01). doi:10.1371/journal.pone.0074134.t002 Anatomical Asymmetries & Running Dynamics/Economy PLOS ONE | www.plosone.org 4 September 2013 | Volume 8 | Issue 9 | e74134 (SI, 0: no symmetry between right and left steps, 1: complete symmetry). Successively, the three mean dynamic indices (SIx, SIy, SIz) are weighted according to the ‘real’ maximum displacement range of the BCOM, i.e. dx ( = running speedXstride frequency), dy and dz, respectively, and a Global symmetry Index (GI) is calculated as GI~ dx:SIxzdy:SIyzdz:SIz dxzdyzdz (GI, 0: no symmetry between right and left steps, 1: complete symmetry). Energy Cost Measurement Oxygen consumption ( _VO2) of running was measured with a breath-by-breath gas analyzer (Cosmed K4b2, Rome, Italy). Data, including heart rate (HR), were recorded at each progression speed, after the metabolic steady state had been achieved (3 min), for further 2 minutes. 5 minutes of testing was performed at each speed. Resting _VO2 was measured while standing. Respiratory Exchange Ratio (RER) was monitored in order to check for aerobic conditions (RER,1). We expressed the metabolic Cost of Transport (C), i.e. the oxygen consumed to move 1 kg of body mass 1 m distance, in J (kg m)21 by dividing the net _VO2 [measured - resting, [ml O2 (Kg min)21] by the progression speed (m min21), and by assuming an energy equivalent of 20.9 J ml O2 21. Statistical Analysis Relationships between variable pairs were investigated using Pearson’s correlation coefficient. To compare speed dependent variables (C, HR, SIx, SIy, SIz and GI), differences were analyzed using a two-ways ANOVA (groupxspeed) (with a post-hoc Bonferroni correction). For speed independent variables (rmax(PD), rmax(UD) and rmax(LD)), we performed a one-way ANOVA for repeated measures in order to detect difference among districts. Furthermore, Principal Component Analysis (PCA) was performed on the three anatomical indices, in order to estimate their relative contribution to the total variance. Statistical significance was accepted when p,0.05. Results Since only five OR and five SR subjects were able to complete all the running protocols up to 5.0 m/s, and UR subjects stopped at the speed of 4.44 m/s, we did not consider in the statistical analysis the highest speed level. Anatomical Symmetries Anatomical symmetries are described by the maximal cross- correlation value for each district (rmax(PD), rmax(UD) and rmax(LD)), and by the global anatomical cross correlation value (rmax). These values are limited to only 17 subjects, because two MR tests (one for the UR and one for the SR) had to be discarded due to technical problems. One-way ANOVA between the three groups of subject didn’t show any difference between UR, OR and SR for the cross- correlation values, while we found significantly lower values of anatomical symmetry for pelvis district, compared to the upper (p,0.05) and lower leg district (p,0.01) (rmax(PD) = 0.7760.09, rmax(UD) = 0.8260.05 and rmax(LD) = 0.8360.05). PCA showed that 65.8% of the total variance was explained by the first principal component, where the three considered parameters (rmax(LD), rmax(UD) and rmax(PD)) had almost the same weight. However UD seems to give the greatest contribution to the first principal component, with respect to the other two districts. Results regarding pairwise correlations between variables are summarized in Table 2: rmax(UD) is significantly correlated with rmax(PD) and rmax(LD) (p,0.05), also rmax(PD) and rmax(LD) seem to be positively correlated even if not significantly (p = 0.087). Significant results were found also between anatomical symme- tries and kinematics (mean values for the Global Symmetry Index (GI) were evaluated starting from the single values of SIx, SIy and SIz and averaged within each group of speeds for each subject); in particular rmax(PD) is positively correlated with Figure 3. Regression of the mean dynamic Global Symmetry Index (GI) versus the global anatomical cross correlation value (rmax). Each point represents the mean Global Symmetry Index averaged among the different running speeds for each subject; Untrained Runners (UR), Occasional Runners (OR) and Skilled Runners (SR) (r = 0.473; p = 0.055). doi:10.1371/journal.pone.0074134.g003 Figure 4. Mean values for the dynamic Global Symmetry Index (GI) plotted against running speed. Mean values for the dynamic Global Symmetry Index (GI) are evaluated starting from the single values of SIx, SIy and SIz and averaged within each group of subjects, 6 SD, in untrained runners (UR), occasional runners (OR) and skilled runners (SR). Two-way ANOVA (group6running speed) show that the group of UR had a mean GI always lower compared to the OR and SR, (* = p,0.01), independently from the running speed. doi:10.1371/journal.pone.0074134.g004 Anatomical Asymmetries & Running Dynamics/Economy PLOS ONE | www.plosone.org 5 September 2013 | Volume 8 | Issue 9 | e74134 SIx (p,0.01) and GI (p,0.05) and also rmax is positively and significantly correlated with SIx (p,0.05), while we observed a positive trend between rmax and GI, even if not significantly (p = 0.055) (see Figure 3). Kinematics Mean values for the Global Symmetry Index (GI), evaluated starting from the single values of SIx, SIy and SIz and averaged within each group of subjects, 6 SD, are shown in Figure 4. We performed a two-ways ANOVA, where independent variables were running speed and subject group and the dependent variable was GI. Results show that UR have a GI significantly lower than both OR and SR at each velocity (p,0.01). Also, GI for UR seems to decrease with increasing running velocity, even if not significantly. Statistical analysis did not show any difference between UR, OR and SR for the single kinematic symmetry indices, while one-way ANOVA for repeated measure shown significantly lower values for SIx, (0.7260.06) compared to SIy (0.8860.04) and SIz (0.8560.05), (p,0.01). Cost of Transport Results for the metabolic cost C and HR are presented in Figure 5. C is confirmed to be independent of speed, with no differences among running groups. At the same speed, HR decreased as runners’ ability increased, with values for SR significantly lower than for OR and UR. No significant correlation was found between the C and the previously analysed parameters, both for kinematics and for anatomical values (Table 2). Discussion The main aim of this project was to investigate the relationship among the anatomical/structural symmetry of the lower limbs, the dynamical symmetry of the 3D BCOM displacement and the metabolic cost of human running. C has been considered as an indirect index of running performance: at the same sustainable fraction of maximal _VO2, the lower the cost the higher the average speed [24]. While being aware of the speed and training level independency of C, as debated and reported in the literature [25229], our hypothesis was that more asymmetrical limbs, in subjects committed to run with symmetrical steps, would have involved a higher C. In other words, part of the inter-subject C variance could have been explained by different level of anatomical asymmetry. Differently from previous studies dealing with gross morpho- logical features (bones length [5,8], human face [13] and horse muzzle [14] landmarks) and isolated gait parameters (stride length and frequency [30,31], joint angles [3] and ground reaction forces [10]), we analysed the symmetry of the ‘whole’ (left and right) lower limb anatomy and of the global running kinematics (3D trajectory of BCOM), in three groups of differently trained athletes. Our hypothesis, inspired by the engineering of motor vehicles, was not completely verified. C was not significantly correlated either with anatomical symmetries or with dynamical symmetries in running, while we found significant correlations between the anatomical and dynamical symmetries indices (Table 2). This indicates that the more anatomically symmetrical are the subjects, the more symmetrical is their running gait (especially in the forward (x) direction). Figure 5. Mean values ± SD for the cost of transport (C) (lower curves) and for the heart rate (HR) (upper curves). C and HR are plotted against running speed for untrained runners (UR), occasional runners (OR) and skilled runners (SR). Results obtained with the two-way ANOVA (group6running speed) show no significant difference among groups of subjects across velocity for C, which results to be independent of the running speed. HR increased significantly with the running speed for all the three group of subjects. Furthermore we obtained significantly higher HR values for UR compared to OR and SR (* = p,0.01). doi:10.1371/journal.pone.0074134.g005 Anatomical Asymmetries & Running Dynamics/Economy PLOS ONE | www.plosone.org 6 September 2013 | Volume 8 | Issue 9 | e74134 This finding is in accordance with the recent literature, where high level of leg length discrepancy (LLD) is correlated with low symmetrical gait coefficients [7] in walking. In our work, individual LLD was always lower than 2 cm (Table 1), and had no effect on C, according to the studies of Gurney [32]. It is possible that some physiological adaptations of the human machinery compensate for small asymmetries typical of the mechanics of our legged system [1,2], with no influence on C. Rather, larger anatomical discrepancies, like a LLD higher than 2 cm [32] or a body mass not uniformly distributed [33,11], could influence economy. Similar adaptations behaviours might have occurred in runners wearing new and worn shoes [34], or on surfaces of different stiffness [35]. Despite of the changed properties of materials, runners modified their motion pattern as to retain their original dynamics of running. This could occur also in the subjects of this study, who seem to compensate their anatomical body asymmetries and minimize C, a strategy frequently adopted by animals [36]. With the main propulsive muscles operating close to isometric in running [37], tendons can store (stretching) and release (shortening) variable amounts of elastic energy during each step, in the attempt to adapt to different anatomical asymmetries. In this way the metabolic cost can be potentially kept unchanged. In addition, although HR results (Figure 5) witness the appropriateness of clustering subjects according to the different training status (most skilled runners reported the lowest HR, at the same speed, p,0.01), the almost speed-independent C values seem not to be influenced by the different fitness level, as also found by other investigators [27,29]. As also indicated in previous studies, training and experience seem to be important elements in the lower limb joint angle symmetry and in the stride variability of running, even at no apparent metabolic benefit [29231]. The most experienced and high performance athletes can maintain, even at high velocities, higher dynamical symmetry than untrained runners (Figure 4). As step frequency and muscles effort increase, the higher physical demand and peripheral fatigue could impair the maintenance of a symmetrical gait and a consistent locomotion pattern, as seen for the UR group. Furthermore, MRI measurements showed that the anatomical symmetry does not depend on the investigated district. PCA and correlation among lower limb districts could have been caused by misalignments of the two limbs during MRI test. However, due to the use of alignment tools during the tests, we feel confident that the intra-subject symmetry correlation among districts is not a measurement artefact. Similar eigenvalues from PCA suggest that the total variance of symmetry is equally explained by the three districts. This work brings developments in the study of locomotion symmetry, also by means of newly introduced methodologies (BCOM 3D trajectory analysis and 3D cross-correlation between ‘whole’ limb MRI voxels). Differently from the original hypothesis, asymmetrical limbs generate asymmetrical body running at no apparent additional metabolic cost. This suggests some plasticity of the human body in coping with structural changes, with the final result of preserving locomotion economy. Deeper insights have been obtained regarding the relationship between the symmetries correlation residuals and the cost of transport, with the idea that subjects would be less economic when their anatomical and dynamical symmetry values do not match. Supplemental analysis and discussion regarding this hypothesis have been reported in the Appendix S1. Even if statistical results in this perspective are weak, possibly due to the relatively small sample size and low asymmetry level, there are some hints suggesting that only the runners who fail to match their anatomy and dynamics features have an increased cost of locomotion. Therefore, the initial hypothesis embedded in the title ‘‘anatomically asymmetrical runners move more asym- metrically at the same metabolic cost’’ is still valid (i.e. the cost would increase when an anatomically asymmetrical runner attempts to move in a symmetrical way). Further studies focusing on adaptations of the muscle-tendon interplay could reveal how human machine compensate the small structural asymmetries that characterize our legged system. The anatomical asymmetry threshold, above which the now expected asymmetrical gait will also involve an increase in running cost, is the challenge for future investigations. Acknowledgments The authors would like to thank all the subjects for their participation in the study, and the Technician Lauro Dalla Chiara for his help during the MR scans performed at the University Hospital Polyclinic ‘‘Borgo Roma’’ in Verona (Italy). Statistical support from Carlo M. Biancardi is also greatly appreciated. Supporting Information Figure S1 Examples of univariate and bivariate regres- sions. Four different types of linear regressions are presented as examples of correlation between Dynamical Symmetry index in forward direction (SIx) and maximal cross-correlation value for Pelvis District (rmax(PD)): a) Univariate regression, b) Univariate regression with intercept forced to be equal to 0, c) Bivariate regression, d) Bivariate regression with intercept forced to be equal to 0. Untrained Runners (UR), Occasional Runners (OR) and Skilled Runners (SR) symbols as in Figure 3. N.B. The determination coefficient in regressions lines forced through the origin, differently from the general model, does not reflect the fraction of the variability in the dependent variable explained by the independent variable. This makes R2 values unrealistically high and not comparable with the ones obtained in the general models. (TIF) Appendix S1 (DOCX) Author Contributions Conceived and designed the experiments: AEM LPA PZ. Performed the experiments: ES FN NF LPA PZ. Analyzed the data: ES FN AEM. Contributed reagents/materials/analysis tools: NF. Wrote the paper: ES AEM. Designed the software used in analysis: AEM ES. Final approval of the paper version to be published: ES FN PZ LPA NF AEM. References 1. Sadeghi H, Allard P, Prince F, Labelle H (2000) Symmetry and limb dominance in able-bodied gait: a review. Gait & Posture 12: 34–45. 2. Nardello F, Ardigo` LP, Minetti AE (2009) Human locomotion: Right/left symmetry in 3D trajectory of body centre of mass. Gait & Posture 30(S): S802S81. 3. Forczek W, Staszkiewicz R (2012) An evaluation of symmetry in the lower limb joint during the able-bodied gait of women and men. Journal of Human Kinetics 35: 47–57. 4. Maupas E, Paysant J, Martinet N, Andre J (1999) Asymmetric leg activity in healthy subjects during walking, detected by electrogoniometry. Clin Biomech (Bristol, Avon); 14: 403–411. Anatomical Asymmetries & Running Dynamics/Economy PLOS ONE | www.plosone.org 7 September 2013 | Volume 8 | Issue 9 | e74134 5. Cuk T, Leben-Seljak P, Stefancic M (2001) Lateral asymmetry of human long bones. Variability and Evolution 9: 19–32. 6. Lund FH (1930) Physical asymmetries and disorientation. The American journal of Psychology 42(1): 51–62. 7. Gurney B, Mermier C, Robergs R, Gibson A, Rivero D (2001) Effects of limb- length discrepancy on gait economy and lower-extremity muscle activity in older adults. J Bone Joint Surg Am 83-A: 907–915. 8. Seeley MK, Umberger BR, Clasey JL, Shapiro R (2010) The relation between mild leg-lenght inequality and able-bodied gait asymmetry. J Sports Sci Med 9(4): 572–579. 9. Souman JL, Frissen I, Sreenivasa MN, Ernst MO (2009) Walking straight into circles Curr Biol 19: 1538–1542. 10. Herzog W, Nigg BM, Read LJ, Olsson E (1989) Asymmetries in ground reaction force patterns in normal human gait. Med Sci Sports Exerc 21: 110–114. 11. Mattes SJ, Martin PE, Royer TD (2000) Walking symmetry and energy cost in persons with unilateral transtibial amputations: matching prosthetic and intact limb inertial properties. Arch Phys Med Rehabil 81: 561–568. 12. Halling Thomsen M, Tolver Jensen A, Sorensen H, Lindegaard C, Haubro Andersen P (2010). Symmetry indices based on accelerometric data in trotting horses. J Biomech 43: 2608–2612. 13. Manning JT, Pickup LJ (1998) Symmetry and performance in middle distance runners. Int J Sports Med 19: 205–209. 14. Manning J, Ockenden L (1994) Fluctuating asymmetry in racehorses. Nature 370: 185–186. 15. Lee Y, Hara T, Fujita H, Itoh S, Ishigaki T (2001) Automated detection of pulmonary nodules in helical CT images based on an improved template- matching technique. IEEE Trans Med Imaging 20: 595–604. 16. Wang P, DeNunzio A, Okunieff P, O9Dell WG (2007) Lung metastases detection in CT images using 3D template matching. Med Phys 34: 915–922. 17. Ambrosini R, Wang P, O9Dell W (2010) Computer-aided detection of metastatic brain tumors using automated three-dimensional template matching. J Magn Reson Imaging 31: 85–93. 18. Lewis J (1996) Fast normalized cross-correlation. Industrial Light and Magic. 19. Minetti AE, Ardigo` LP, Saibene F (1994) Mechanical determinants of the minimum energy cost of gradient running in humans. J exp Biol 195: 211–225. 20. Dempster WT, Gabel WC, Felts WJ (1959) The anthropometry of the manual work space for the seated subject. Am J Phys Anthropol 17: 289–317. 21. Winter D (2005) Biomechanics and motor control of human movement. New York: John Wiley and Sons, Inc, Third Edition. 22. Minetti AE (2009) The mathematical description (Lissajous contour) of the 3D trajectory of the body centre of mass: A locomotor ‘signature’ for the physiology, biomechanics and pathology of human and animal gaits. Gait & Posture 30(S): S153–S153. 23. Minetti AE, Cisotti C, Mian OS (2011) The mathematical description of the body centre of mass 3D path in human and animal locomotion. J Biomech 44: 1471–1477. 24. Di Prampero PE (1986) The energy cost of human locomotion on land and in water. Int J Sports Med 7(2): 55–72. 25. Margaria R, Cerretelli P, Aghemo P, Sassi G (1963) Energy cost of running. J Appl Physiol 18: 367–370. 26. Daniels JT (1985) A physiologist’s view of running economy. Med Sci Sports Exerc 17: 332–338. 27. Slawinski JS, Billat VL (2004) Difference in mechanical and energy cost between highly, well, and nontrained runners. Med Sci Sports Exerc 36: 1440–1446. 28. Beneke R, Hutler M (2005) The effect of training on running economy and performance in recreational athletes. Med Sci Sports Exerc 37: 1794–1799. 29. McGregor SJ, Busa MA, Yaggie JA, Bollt EM (2009) High resolution MEMS accelerometers to estimate VO2 and compare running mechanics between highly trained inter-collegiate and untrained runners. PLoS One 4: e7355. 30. Cavanagh PR, Pollock ML, Landa J (1977) A biomechanical comparison of elite and good distance runners. Ann N Y Acad Sci 301: 328–45. 31. Nakayama Y, Kudo K, Ohtsuki T (2010) Variability and fluctuation in running gait cycle of trained runners and non-runners. Gait & Posture 31: 331–335. 32. Gurney B (2002) Leg length discrepancy. Gait & Posture 15(2): 195–206. 33. Saibene F, Minetti AE (2003) Biomechanical and physiological aspects of legged locomotion in humans. Eur J Appl Physiol 88: 297–316. 34. Kong PW, Candelaria NG, Smith DR (2009) Running in new and worn shoes: a comparison of three types of cushioning footwear. Br J Sports Med 43(10): 745– 9. 35. Hardin EC, van den Bogert AJ, Hamill J (2004) Kinematic adaptations during running: effects of footwear, surface, and duration. Med Sci Sports Exerc 36(5): 838–44. 36. Alexander RM (1989) Optimization and gaits in the locomotion of vertebrates. Physiol Rev 69: 1199–1227. 37. Srinivasan M (2011) Fifteen observations on the structure of energy-minimizing gaits in many simple biped models. J R Soc Interface 8: 74–98. Anatomical Asymmetries & Running Dynamics/Economy PLOS ONE | www.plosone.org 8 September 2013 | Volume 8 | Issue 9 | e74134

Anatomically asymmetrical runners move more asymmetrically at the same metabolic cost.

09-24-2013

Seminati, Elena,Nardello, Francesca,Zamparo, Paola,Ardigò, Luca P,Faccioli, Niccolò,Minetti, Alberto E

eng

PMC6448870

RESEARCH ARTICLE A comparison of match-physical demands between different tactical systems: 1-4-5-1 vs 1-3-5-2 Ivan BaptistaID1*, Dag Johansen2, Pedro Figueiredo3,4, Anto´nio RebeloID5, Svein Arne Pettersen1 1 School of Sport Sciences, University of Tromsø, the Arctic University of Norway, Tromsø, Norway, 2 Computer Science Department, University of Tromsø, the Arctic University of Norway, Tromsø, Norway, 3 Portugal Football School, Portuguese Football Federation, Lisboa, Portugal, 4 Research Center in Sports Sciences, Health Sciences and Human Development, CIDESD, University Institute of Maia, ISMAI, Maia, Portugal, 5 Faculty of Sport, University of Porto, Porto, Portugal * ivan.a.baptista@uit.no Abstract The team tactical system and distribution of the football players on the pitch is considered fundamental in team performance. The present study used time-motion analysis and triax- ial-accelerometers to obtain new insights about the impact of different tactical systems (1-4- 5-1 and 1-3-5-2) on physical performance, across different playing positions, in a profes- sional football team. Player performance data in fifteen official home matches was collected for analysis. The sample included twenty-two players from five playing positions (centre backs: n = 4; full-back/wide midfielder/ wing-back: n = 9; centre midfielder: n = 6 and centre forward: n = 3), making a total of 108 match observations. A novel finding was that general match physical demands do not differ considerably between these tactical formations, prob- ably because match-to-match variability (variation of players’ running profile from match-to- match) might be higher than the differences in physical performance between tactical sys- tems. However, change of formation had a different impact across playing positions, with centre backs playing in 1-4-5-1 performing significant more HIRcounts than in 1-3-5-2 (p = 0.031). Furthermore, a medium effect size (r = 0.33) was observed in HIRdist, with wide players covering higher distances when playing in 1-3-5-2 than in 1-4-5-1. These findings may help coaches to develop individualised training programs to meet the demands of each playing position according to the tactical system adopted. Introduction To better understand the constraints correlated with sporting success, match analysis has become an important tool in team sports. Nowadays it is well accepted among coaches and sport scientists that the match performance of a football team is, basically, based on four fac- tors: physical, technical, tactical and mental [1]. Even though, the majority of research has been executed within the physical and technical performance domain, previous studies have PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 1 / 12 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Baptista I, Johansen D, Figueiredo P, Rebelo A, Pettersen SA (2019) A comparison of match-physical demands between different tactical systems: 1-4-5-1 vs 1-3-5-2. PLoS ONE 14(4): e0214952. https://doi.org/10.1371/journal. pone.0214952 Editor: Luca Paolo Ardigò, Universita degli Studi di Verona, ITALY Received: November 23, 2018 Accepted: March 22, 2019 Published: April 4, 2019 Copyright: © 2019 Baptista et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Funding: The authors received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. started to establish connections between physiological demands and tactical behaviour in elite football [2–5]. The lack of research and information about this field can be observed in a systematic review (2012–2016) on match analysis in adult male football [6], where the contextual variables of research analysed (match half, quality of opposition, match location, scoring first, group stage vs knockout phase, substitutions, competitive level and different competitions) did not include the tactical systems used by teams. The team tactical system and the positioning and distribution of the players on the pitch is considered one of the most important strategic decisions in football [5, 7, 8] and, it is evident that player match-load is influenced by different factors, such as the playing position [2, 9, 10] and the tactical system [11]. This highlights the importance of understanding how physical demands may be affected by playing position in various tactical systems [6]. Despite some pre- vious research [12, 13] addressing the team global positioning on the field, using the measures of centre and dispersion, the role of the tactical system regarding the players’ physical perfor- mance, has not been fully described. Previous studies have concluded that the manipulation of playing formations in small sided games promotes changes in physical performance of teams and players in training [14]. Also, the success of different tactics and strategies depend on the capacities and abilities of the play- ers to perform specific actions during the match. Consequently, players must fulfil the neces- sary physiological requirements of their playing position inside the tactical system adopted [5, 15, 16]. Previous research has investigated the influence of opposition tactical formation on physio- logical performance variables and reported higher running distances when playing against a 1- 4-2-3-1 formation compared to a 1-4-4-2 formation [17]. In opposition, other studies [11, 18] using various teams and/or different players across different seasons have concluded that tacti- cal systems do not influence the match activity profiles of players. A pilot study with youth players [19] reported no correlation between physical/technical levels and tactical prominence in football matches. However, the identification of the tactical system adopted by a particular team is not a trivial step and previous studies have subjectively defined the tactical formations analysed by using qualified coaches to identify the different formations, as well as to verify if those formations were consistent throughout the game [17, 20]. To the best of our knowledge, no other study has examined the effect of playing formation on player load by position within the same team, in one full season. An in-depth analysis of match physical performance across playing positions, in different tactical formations, could provide a better understanding of position-specific demands and provide an useful insight to optimize training programs [11]. Therefore, the present study aimed to analyse how tactical systems affect the physical performance of a professional football team across different playing positions in all official home matches during one season. We hypothesize that, despite playing in their specific position, players will accumulate different external workload in matches, depending on the tactical formation deployed. Methods Participants and match analysis With institutional ethics approval from UiT The Arctic University of Norway Institutional Review Board, written informed consent from players and approval from the Norwegian Cen- tre for Research Data, data on performance in 15 official home matches from the professional team of a Norwegian elite football club, during one season (2017), was collected for analysis. The matches were all played on artificial grass surface, as described in detail previously [10]. Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 2 / 12 The sample included 22 players (25.2 ± 4.4 years of age; 76.2 ± 6.4 kg of body mass; and, 181.6 ± 5.6 cm of height) across four different playing positions: centre back, CB (n = 4, obser- vations[obs] = 37), full-back/wide midfielder/ wing-back, FB/WM/WB (n = 9, obs = 31), cen- tre midfielder, CM (n = 6, obs = 26), and centre forward, CF (n = 3, obs = 14), making a total of 139 match observations (Table 1). Playing-positions were chosen according to the two tacti- cal formations used by the team and previous research [9, 21, 22]. Team tactical systems and playing positions were determined by two UEFA-qualified coaches (one from the coaching staff of the team analysed) after visualizing video recordings of the sampled matches [17, 20]. These observers subjectively determined the tactical systems used at the beginning of the match and verified if the formations were consistent throughout the matches [17]. Further- more, 1-4-5-1 and 1-4-3-3 formations were combined, as well as 1-3-5-2 and 1-5-3-2. This procedure was applied due to difficulties in establishing specific differences between similar playing formations when in attacking and defending. When analysing the 1-3-5-2 formation the observers realized that the team often played in 1-5-3-2 formation when not in ball posses- sion (defending) and in 1-3-5-2 with ball possession (attacking). On the other hand, when observing the 1-4-5-1 formation, the observers concluded that the team played in 1-4-5-1 when defending and in 1-4-3-3 when attacking [11, 17]. No other changes in formations throughout the matches were noticed by the observers, therefor no matches were excluded from the analysis. Data was analysed only if: (a) players completed the full match (90 minutes), (b) the player played in the same position during all the match and (c) the team used 1-4-5-1 (1 goalkeeper; 2 CB + 2 FB; 3 CM + 2 WM; 1 CF) or 1-3-5-2 (1 goalkeeper; 3 CB; 3 CM + 2 WB; 2 CF) tactical formations during the entire match. To ensure players confidentiality, all data was anonymized before analyses. Procedures A stationary radio wave-based Local Positioning Measurement (LPM) tracking system (ZXY Sport Tracking System, Trondheim, Norway), with a default resolution of 20Hz, was used to characterize match activity profiles within the team. Each player wore a specially designed belt, wrapped tightly around the waist, with an electronic sensor system at the player’s lumbar spine, as reported previously [10]. At the stadium, where the matches occurred, there are 6 RadioEyes for optimal coverage, resulting in practically zero packet loss for transponders on the field. If packet loss occurred, the data was linearly interpolated. The accuracy and reliability of the system in measuring player movements in elite soccer competitions have been described in more detail in previous studies [23–25]. Physical performance variables Physical parameters analysed included: total distance (TotDist) number of accelerations (acccounts), acceleration distance (accdist), number of decelerations (deccounts), deceleration Table 1. Number of match observations per player and tactical system. Player 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 Observations per tactical system 1-4-5-1 3 5 7 1 3 5 1 0 6 2 4 2 6 0 6 0 1 1 1 0 1 1 1-3-5-2 0 7 7 7 2 0 0 6 5 0 5 3 0 1 6 1 0 0 0 2 0 0 Total observations 3 12 14 8 5 5 1 6 11 2 9 5 6 1 12 1 1 1 1 2 1 1 https://doi.org/10.1371/journal.pone.0214952.t001 Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 3 / 12 distance (decdist), number of HIR (HIRcounts), HIR distance (HIRdist), number of sprints (sprintcounts), sprint distance (sprintdist) and turns. The HIR (19.8 km
h−1) and sprinting (25.2 km
h−1) speed thresholds are similar to those reported in previous research [10, 22, 24, 26]. According to the ZXY Sport Tracking system accelerations were quantified through numer- ical derivation from positional data with a sampling frequency of 20Hz [25]. Furthermore, accelerations are defined by four event markers: (a) the start of the acceleration event is marked by the acceleration reaching the minimum limit of 1 m
s −2, (b) the acceleration reaches the acceleration limit of 2 m
s −2, (c) the acceleration remains above the 2 m
s −2 for at least 0.5 seconds and (d) the duration of the acceleration ends when it decreases below the minimum acceleration limit (1 m
s −2). Turns were counted only if the player performed a continuous and significant body rota- tion of more than 90˚ in one direction (derived from gyroscope and compass data). The end of a turn and the start of another occurs when a rotation in the opposite direction is measured. The angle threshold used by ZXY Sport Tracking system allowed us to analyse only angles 90˚. Statistical analysis The results are presented as mean and 95% confidence interval, unless otherwise stated. A lin- ear mixed-effects model with restricted maximum likelihood estimations was used to examine differences in Local Positioning Measurement-derived variables and match duration between 1-3-5-2 and 1-4-5-1 formations. Mixed models can account for unbalanced repeats per player and thus used to model the data. Tactical formation, playing position and their interaction was modelled as fixed effects (effects describing the association between the dependent variable and covariates), while ‘athlete ID’ was included as a random effect (effects generally represent- ing random deviations from the relationships of the fixed part of the model). An α-level of 0.05 was used as level of significance for statistical comparisons. Furthermore, multiple com- parisons were adjusted using the Tukey method. The t statistics from the mixed models were converted to effect size correlations [27]. Effect sizes were interpreted as <0.1, trivial; 0.1–0.3, small; 0.3–0.5, moderate; 0.5–0.7, large; 0.7–0.9, very large; 0.9–0.99, almost perfect; 1.0, perfect [28]. All statistical analyses were conducted using the lme4, lsmeans and psychometric pack- ages in R statistical software (version 3.4.1, R Foundation for Statistical Computing, Vienna, Austria). Results Centre-backs Slightly higher values, though not statistically significant, were found in HIRdist, Acc and Dec (counts and distance), sprintcounts and turns when playing in 1-4-5-1 compared to 1-3-5-2 for- mation (Table 2). Furthermore, CB playing in 1-4-5-1 were observed to perform significant more HIRcounts (36.1 ± 3.5) than in 1-3-5-2 (28.2 ± 3.5) (p = 0.008), with a correspondent medium effect size (r = 0.37). Wide positions No significant differences were observed between the tactical formations analysed from players playing in wide positions (Table 3). However, higher values in HIRdist (r = 0.19) and sprintdist (r = 0.16) were found when playing with 1-3-5-2 (977.2 ± 73.7; 236.9 ± 26.8) compared to 1-4- 5-1 (838.9 ± 62.5; 195.3 ± 22.7) formation. Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 4 / 12 Centre midfielders Small effect sizes were observed in HIRcounts (r = 0.12) and Acccounts (r = 0.14) (Table 4), with higher values being observed when playing in 1-4-5-1 (38.5 ± 3.2; 62.3 ± 5.5) than in 1-3-5-2 (35.7 ± 3.4; 55.9 ± 5.9). A similar effect size was also observed in turns (r = 0.15), with CM per- forming more turns when playing in 1-3-5-2 (40.3 ± 3.7) than in 1-4-5-1 (34.7 ± 3.4). Centre forwards No significant differences were found regarding any parameter analysed. However, higher val- ues, though with a trivial effect size, in HIRdist and sprintdist can be observed (Table 5) when playing in 1-3-5-2. Tactical system Significant differences were found in various parameters when comparing the physical perfor- mance of the whole team when playing with different tactical systems (Table 6). Significant higher values were observed in HIRcounts (r = 0.25) and sprintcounts (r = 0.22) when playing in Table 2. Mean and 95% confidence interval estimates of different physical parameters from centre backs, analysed according to the tactical system used, and respec- tive p-value and effect size of differences observed (n = 4; observations = 37). Variables CB p-value Effect Size (r) 1-4-5-1 1-3-5-2 TotDist (m) 10865.0 (227.6) 10591.8 (224.0) 0.825 0.15 HIR counts 36.1 (3.5) 28.2 (3.5) 0.008 0.37 HIR dist (m) 512.0 (81.5) 431.0 (81.3) 0.658 0.18 Sprint counts 6.6 (1.9) 5.4 (1.9) 0.871 0.15 Sprint dist (m) 64.4 (29.6) 74.2 (29.5) 0.999 0.06 Acc dist (m) 325.6 (37.6) 306.9 (37.6) 0.982 0.10 Acc counts 63.2 (6.1) 59.7 (6.1) 0.983 0.10 Dec dist (m) 321.2 (41.7) 278.5 (41.6) 0.543 0.20 Dec counts 60.3 (6.9) 53.6 (6.9) 0.680 0.18 Turns 32.2 (3.5) 25.8 (3.4) 0.437 0.21 https://doi.org/10.1371/journal.pone.0214952.t002 Table 3. Mean and 95% confidence interval estimates of different physical parameters from full-backs, wide midfielders and wing-backs analysed according to the tactical system used, and respective p-value and effect size of differences observed (n = 9; observations = 31). Variables FB/WM/WB p-value Effect Size (r) 1-4-5-1 1-3-5-2 TotDist 10842.6 (188.8) 11143.0 (233.0) 0.942 0.13 HIR counts 45.9 (2.7) 46.9 (3.2) 1.000 0.03 HIR dist 838.9 (62.5) 977.2 (73.7) 0.523 0.19 Sprint counts 14.1 (1.4) 14.0 (1.6) 1.000 0.01 Sprint dist 195.3 (22.7) 236.9 (26.8) 0.747 0.16 Acc dist 462.2 (28.5) 447.1 (33.2) 1.000 0.05 Acc counts 83.2 (4.7) 76.8 (5.7) 0.950 0.12 Dec dist 501.2 (31.5) 505.4 (36.9) 1.000 0.01 Dec counts 86.9 (5.3) 86.1 (6.2) 1.000 0.01 Turns 42.1 (2.9) 38.8 (3.7) 0.993 0.11 https://doi.org/10.1371/journal.pone.0214952.t003 Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 5 / 12 1-4-5-1 (43.6 ± 1.9; 11.4 ± 1.1) compared with 1-3-5-2 (40.0 ± 2.0; 10.0 ± 1.1) (p = 0.005 and p = 0.015, respectively). Furthermore, when playing in 1-4-5-1, the team was observed to per- form more Acccounts (75.8 ± 3.2) and Deccounts (77.8 ± 3.5), as well as covering higher distances in Decdist (440.3 ± 23.3) than when playing in 1-3-5-2 (71.1 ± 3.4; 72.5 ± 3.6; 413.7 ± 24.2; for Acccounts, Deccounts and Decdist) (p = 0.022; p = 0.014 and p = 0.032, respectively). Discussion Context The present study provides new insights into the physical demands of two common tactical formations, in elite football players across different playing positions. The context of this study appeared with the change of the head-coach, and consequently, the tactical formation and style of play used of the professional football team analysed. Since this replacement happened in the middle of the season, both tactical formations analysed were composed by an almost equal number of matches (7 and 8 home matches each). It is also important to refer that the change of head-coach led not only to a simple switch of the tactical structure used, but also to a change to a more complex style of play. A more possession and position-oriented style of play Table 4. Mean and 95% confidence interval estimates of different physical parameters from centre midfielders, analysed according to the tactical system used, and respective p-value and effect size of differences observed (n = 6; observations = 26). Variables CM p-value Effect Size (r) 1-4-5-1 1-3-5-2 TotDist 12009.0 (218.5) 11820.8 (238.7) 1.000 0.09 HIR counts 38.5 (3.2) 35.7 (3.4) 0.948 0.12 HIR dist 643.2 (73.1) 610.9 (78.1) 1.000 0.06 Sprint counts 7.0 (1.6) 7.0 (1.7) 1.000 0.05 Sprint dist 101.4 (26.6) 94.8 (28.4) 1.000 0.03 Acc dist 313.3 (33.4) 289.6 (35.5) 0.973 0.10 Acc counts 62.3 (5.5) 55.9 (5.9) 0.845 0.14 Dec dist 358.3 (37.0) 326.0 (39.4) 0.923 0.13 Dec counts 69.4 (6.2) 64.2 (6.6) 0.951 0.11 Turns 34.7 (3.4) 40.3 (3.7) 0.782 0.15 https://doi.org/10.1371/journal.pone.0214952.t004 Table 5. Mean and 95% confidence interval estimates of different physical parameters from centre forwards, analysed according to the tactical system used, and respective p-value and effect size of differences observed (n = 3; observations = 14). Variables CF p-value Effect Size (r) 1-4-5-1 1-3-5-2 TotDist 10724.4 (328.6) 10732.8 (328.6) 1.000 >0.01 HIR counts 48.6 (4.7) 47.1 (4.7) 1.000 0.05 HIR dist 835.2 (108.5) 930.5 (108.5) 0.881 0.14 Sprint counts 11.7 (2.4) 12.8 (2.4) 0.993 0.08 Sprint dist 164.5 (39.5) 208.5 (39.5) 0.689 0.18 Acc dist 483.4 (49.4) 477.7 (49.4) 1.000 0.02 Acc counts 82.9 (8.2) 80.2 (8.2) 1.000 0.05 Dec dist 461.4 (54.8) 470.8 (54.8) 1.000 0.03 Dec counts 78.3 (9.2) 73.4 (9.2) 0.992 0.09 Turns 36.8 (5.1) 29.7 (5.1) 0.810 0.16 https://doi.org/10.1371/journal.pone.0214952.t005 Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 6 / 12 were adopted (1-3-5-2) instead of the more direct play and counter-attack strategy used in the first half of the season (1-4-5-1). However, even with all these changes, the context remained the same (same players with similar physical capacities). Comparison according to playing position The results suggest that general match physical demands do not differ considerably between these two tactical formations when compared by playing position. Independent of formation and with few exceptions, players presented similar profiles in all the physical parameters ana- lysed. The most relevant exceptions were the higher HIRcounts in CB (1-4-5-1) and longer HIRdist in FB/WM/WB (1-3-5-2), with a medium and small effect size, respectively. CB playing in 1-4-5-1 performed more HIRcounts, probably due to the larger area they needed to cover when compared to the area covered by the three CBs when playing in 1-3-5-2. When in defensive organisation (without ball possession), the defensive line of three CBs became most of the time a defensive line composed by 5 players (three CBs and two WBs). The increased number of players playing in the defensive line leads to less m2 per player to cover. Players in wide positions covered more HIRdist when playing in 1-3-5-2 most likely because in this formation the team played with only two wide players (WB), and they needed to cover all the flank, while with 1-4-5-1 formation, those flanks were covered by a total of four players (two on each side). It has been speculated that match physical demands are higher for CF when playing “alone” in the offensive line (e.g. 1-4-5-1; 1-5-4-1), as they are very often isolated and marked by sev- eral opponents [29]. However, the results of the present study are slightly different, since higher, though not significant, values were found in HIRdist and sprintdist for CF, when playing with two attackers (1-3-5-2) compared with playing with only one (1-4-5-1). Furthermore, no differences in playing time (substitutions) were observed in any playing position between the two tactical systems analysed. Comparison according to team workload When playing position was not taken into consideration and the work-load of the whole team was analysed, the physical workload in some variables was significantly different between tacti- cal systems used. Small significant differences were observed in HIRcounts and sprintcounts, with the team performing more runs (>19,8 km/h) when playing in 1-4-5-1. The number of Acc Table 6. Mean and 95% confidence interval estimates of different physical parameters from the whole team, analysed according to the tactical system used, and respective p-value and effect size of differences observed. Variables Tactical system p-value Effect Size (r) 1-4-5-1 1-3-5-2 TotDist 11048.5 (140.2) 11091.2 (149.5) 0.705 0.03 HIR counts 43.6 (1.9) 40.0 (2.0) 0.005 0.25 HIR dist 779.9 (50.9) 762.8 (52.7) 0.541 0.06 Sprint counts 11.4 (1.1) 10.0 (1.1) 0.015 0.22 Sprint dist 156.9 (19.1) 158.6 (19.8) 0.867 0.02 Acc dist 420.7 (23.1) 401.1 (23.8) 0.085 0.16 Acc counts 75.8 (3.2) 71.1 (3.4) 0.022 0.20 Dec dist 440.3 (23.3) 413.7 (24.2) 0.032 0.19 Dec counts 77.8 (3.5) 72.5 (3.6) 0.014 0.22 Turns 36.9 (1.9) 33.5 (2.0) 0.057 0.16 https://doi.org/10.1371/journal.pone.0214952.t006 Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 7 / 12 and Dec was also higher when the 1-4-5-1 system was used. In general, almost all variables ana- lysed presented higher values during the first period of the season (1-4-5-1) than in the second (1-3-5-2). Previous research [30, 31] has suggested that teams who are winning the match tend to relax and decrease their work-rate. Alternatively, although teams who are losing the match may increase their work-rate during a specified period [32, 33], they may quickly lose the moti- vation to keep the elevated work rate, which may be especially evident when the goal difference increases negatively (conceding more goals) [34]. In fact, the differences observed between these two tactical systems might be, in part, justified by the significant discrepancy between the score line and match final results achieved during the first and second part of the season. While playing in 1-4-5-1 the team achieved one victory, four draws and three defeats in the eight home matches played. On the other hand, while playing in 1-3-5-2, the team had better results, with five victories, one draw and one defeat in the last seven home matches played. The match results (considerably more draws) and the differences in style of play may therefore, partly justify the higher work-rate of the 1-4-5-1 tactical system. Limitations Our initial hypothesis was that, despite playing in their specific positions, players would accumu- late different external workload in matches, depending on the preferred tactical formation. How- ever, the results presented in this study do not fully support the hypothesis, probably because the match-to-match variability might be larger than the differences in physical performance between tactical systems. Like most of the measures in team sports performance, the physical variables used in this study are not stable and are subject to a high variation between successive matches [35]. Furthermore, it has been proved that within-subject (player) and between-match variation in physical performance across the season might be experienced due to changes in the physical condition of the player [36, 37] and environmental conditions [38]. Previous studies have shown that match-to-match variability in performance characteristics of elite soccer players is high [35, 39, 40] and that future research based in match performance requires large sample sizes to iden- tify true systematic changes in workload. In fact, the sample size (22 players/108 observations) might be of such small numbers that true differences can be masked due to a statistical type 2 error, and such a consequence cannot be conclusively ruled out. Previous similar studies have analysed more matches [17] or used considerably larger sample sizes [11] than in the present study. However, they have not compared the physical demands of different tactical systems within the same players in the same context (same team and season) and to do so, a larger sample size than the one used in the present study becomes a difficult task to fulfil. Even though, the methodology used to determine the team formations is in line with previ- ous studies [11, 14, 17, 20, 41, 42], the process of defining team formations and controlling their consistency throughout the matches was based on the subjective assessment of observers. Further research is needed to attempt to define objectively team formations and to identify when changes occur [17]. Goalkeepers were not included in the present study, however their match activity profiles might be useful and interesting to analyse in different tactical systems and styles of play in future research. All these limitations should be taken into consideration when designing future studies. Perspectives and practical application Since previous research has shown that the players’ physical demands in matches are highly dependent on their positional role in the team [43, 44], analytics, in general, have become a Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 8 / 12 crucial component of team organization and content of training, to meet the position-specific requirements of physical conditioning [45]. This study goes beyond the individualization of training demands according to playing position, also suggesting that the change of tactical sys- tem might influence, specific variables of the team’s overall match activity profile, and those differences should be taken into consideration when designing training programs. On the other hand, differences are not notable in all playing positions and these findings should be interpreted with caution, as differences might be team dependent since other teams using the same tactical systems, probably appear with different styles of play. Change of formation had a different impact on different playing positions, with CB and wide positions presenting more substantial differences than CM and CF. As previously men- tioned, the present study and its findings may provide useful and novel insights for coaches on physical performance demands in different tactical formations across playing positions. The information provided should be taken into consideration when designing and implementing training program cycles, according to players’ playing position, the team’s tactical formation and style of play. The individualization and specialization of the training should, therefore, be a matter of reflection and analysis from practitioners. Supporting information S1 File. Data review. (XLSX) Author Contributions Conceptualization: Ivan Baptista. Data curation: Pedro Figueiredo. Formal analysis: Pedro Figueiredo. Investigation: Ivan Baptista. Methodology: Ivan Baptista, Anto´nio Rebelo, Svein Arne Pettersen. Project administration: Svein Arne Pettersen. Supervision: Dag Johansen, Svein Arne Pettersen. Visualization: Ivan Baptista. Writing – original draft: Ivan Baptista. Writing – review & editing: Dag Johansen, Pedro Figueiredo, Anto´nio Rebelo, Svein Arne Pettersen. References 1. Sarmento H, Marcelino R, Anguera MT, CampaniCo J, Matos N, LeitAo JC. Match analysis in football: a systematic review. J Sports Sci. 2014; 32(20):1831–43. Epub 2014/05/03. https://doi.org/10.1080/ 02640414.2014.898852 PMID: 24787442. 2. Bloomfield J, Polman R, O’Donoghue P. Physical Demands of Different Positions in FA Premier League Soccer. Journal of sports science & medicine. 2007; 6(1):63–70. Epub 2007/01/01. PMID: 24149226; PubMed Central PMCID: PMCPMC3778701. 3. Drust B, Atkinson G, Reilly T. Future perspectives in the evaluation of the physiological demands of soc- cer. Sports medicine (Auckland, NZ). 2007; 37(9):783–805. Epub 2007/08/29. https://doi.org/10.2165/ 00007256-200737090-00003 PMID: 17722949. Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 9 / 12 4. Moura FA, Martins LE, Anido Rde O, de Barros RM, Cunha SA. Quantitative analysis of Brazilian foot- ball players’ organisation on the pitch. Sports biomechanics. 2012; 11(1):85–96. Epub 2012/04/24. https://doi.org/10.1080/14763141.2011.637123 PMID: 22518947. 5. Rein R, Memmert D. Big data and tactical analysis in elite soccer: future challenges and opportunities for sports science. SpringerPlus. 2016; 5(1):1410. Epub 2016/09/10. https://doi.org/10.1186/s40064- 016-3108-2 PMID: 27610328; PubMed Central PMCID: PMCPMC4996805. 6. Sarmento H, Clemente FM, Araujo D, Davids K, McRobert A, Figueiredo A. What Performance Analysts Need to Know About Research Trends in Association Football (2012–2016): A Systematic Review. Sports medicine (Auckland, NZ). 2018; 48(4):799–836. Epub 2017/12/16. https://doi.org/10.1007/ s40279-017-0836-6 PMID: 29243038. 7. Kannekens R, Elferink-Gemser MT, Visscher C. Positioning and deciding: key factors for talent devel- opment in soccer. Scandinavian Journal of Medicine & Science in Sports. 2011; 21(6):846–52. https:// doi.org/10.1111/j.1600-0838.2010.01104.x PMID: 22126715 8. Sampaio J, Macas V. Measuring tactical behaviour in football. Int J Sports Med. 2012; 33(5):395–401. Epub 2012/03/02. https://doi.org/10.1055/s-0031-1301320 PMID: 22377947. 9. Schuth G, Carr G, Barnes C, Carling C, Bradley PS. Positional interchanges influence the physical and technical match performance variables of elite soccer players. J Sports Sci. 2016; 34(6):501–8. Epub 2015/12/25. https://doi.org/10.1080/02640414.2015.1127402 PMID: 26700131. 10. Baptista I, Johansen D, Seabra A, Pettersen SA. Position specific player load during match-play in a professional football club. PLoS One. 2018; 13(5). https://doi.org/10.1371/journal.pone.0198115 PMID: 29795703; PubMed Central PMCID: PMCPMC5967838. 11. Bradley P, Carling C, Archer D, Roberts J, Dodds A, Di Mascio M, et al. The effect of playing formation on high-intensity running and technical profiles in English FA Premier League soccer matches. J Sports Sci. 2011; 29(8):821–30. Epub 2011/04/23. https://doi.org/10.1080/02640414.2011.561868 PMID: 21512949. 12. Bartlett R, Button C, Robins M, Dutt-Mazumder A, Kennedy G. Analysing Team Coordination Patterns from Player Movement Trajectories in Soccer: Methodological Considerations. International Journal of Performance Analysis in Sport International Journal of Performance Analysis in Sport. 2017; 12 (2):398–424. 13. Moura FA, Martins LE, Anido RO, Ruffino PR, Barros RM, Cunha SA. A spectral analysis of team dynamics and tactics in Brazilian football. J Sports Sci. 2013; 31(14):1568–77. Epub 2013/05/02. https://doi.org/10.1080/02640414.2013.789920 PMID: 23631771. 14. Baptista J, Travassos B, Goncalves B, Mourao P, Viana JL, Sampaio J. Exploring the effects of playing formations on tactical behaviour and external workload during football small-sided games. J Strength Cond Res. 2018. Epub 2018/01/18. https://doi.org/10.1519/jsc.0000000000002445 PMID: 29337830. 15. Clemente FM, Couceiro MS, Martins FM, Ivanova MO, Mendes R. Activity profiles of soccer players during the 2010 world cup. J Hum Kinet. 2013; 38:201–11. Epub 2013/11/16. https://doi.org/10.2478/ hukin-2013-0060 PMID: 24235995; PubMed Central PMCID: PMCPMC3827759. 16. Silva JR, Magalhaes J, Ascensao A, Seabra AF, Rebelo AN. Training status and match activity of pro- fessional soccer players throughout a season. J Strength Cond Res. 2013; 27(1):20–30. Epub 2012/02/ 22. https://doi.org/10.1519/JSC.0b013e31824e1946 PMID: 22344051. 17. Carling C. Influence of opposition team formation on physical and skill-related performance in a profes- sional soccer team. European Journal of Sport Science. 2011; 11(3):155–64. https://doi.org/10.1080/ 17461391.2010.499972 18. Palucci Vieira LH, Aquino R, Lago-Penas C, Munhoz Martins GH, Puggina EF, Barbieri FA. Running Performance in Brazilian Professional Football Players During a Congested Match Schedule. J Strength Cond Res. 2018; 32(2):313–25. Epub 2018/01/26. https://doi.org/10.1519/JSC.0000000000002342 PMID: 29369952. 19. Clemente FM, Figueiredo AJ, Martins FM, Mendes RS, Wong DP. Physical and technical performances are not associated with tactical prominence in U14 soccer matches. Research in sports medicine (Print). 2016; 24(4):352–62. Epub 2016/10/30. https://doi.org/10.1080/15438627.2016.1222277 PMID: 27533018. 20. Aquino R, Carling C, Palucci Vieira LH, Martins G, Jabor G, Machado J, et al. Influence of Situational Variables, Team Formation, and Playing Position on Match Running Performance and Social Network Analysis in Brazilian Professional Soccer Players. J Strength Cond Res. 2018. Epub 2018/07/10. https://doi.org/10.1519/jsc.0000000000002725 PMID: 29985222. 21. Carling C, Le Gall F, Dupont G. Analysis of repeated high-intensity running performance in professional soccer. J Sports Sci. 2012; 30(4):325–36. Epub 2012/01/18. https://doi.org/10.1080/02640414.2011. 652655 PMID: 22248291. Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 10 / 12 22. Bradley P, Sheldon W, Wooster B, Olsen P, Boanas P, Krustrup P. High-intensity running in English FA Premier League soccer matches. J Sports Sci. 2009; 27(2):159–68. Epub 2009/01/21. https://doi.org/ 10.1080/02640410802512775 PMID: 19153866. 23. Bendiksen M, Pettersen SA, Ingebrigtsen J, Randers MB, Brito J, Mohr M, et al. Application of the Copenhagen Soccer Test in high-level women players—locomotor activities, physiological response and sprint performance. Human movement science. 2013; 32(6):1430–42. Epub 2013/09/11. https:// doi.org/10.1016/j.humov.2013.07.011 PMID: 24016711. 24. Ingebrigtsen J, Dalen T, Hjelde GH, Drust B, Wisloff U. Acceleration and sprint profiles of a professional elite football team in match play. Eur J Sport Sci. 2015; 15(2):101–10. Epub 2014/07/10. https://doi.org/ 10.1080/17461391.2014.933879 PMID: 25005777. 25. Pettersen SA, Johansen HD, Baptista IAM, Halvorsen P, Johansen D. Quantified Soccer Using Posi- tional Data: A Case Study. Frontiers in Physiology. 2018; 9(866). https://doi.org/10.3389/fphys.2018. 00866 PMID: 30034347 26. Dalen T, Ingebrigtsen J, Ettema G, Hjelde GH, Wisloff U. Player Load, Acceleration, and Deceleration During Forty-Five Competitive Matches of Elite Soccer. J Strength Cond Res. 2016; 30(2):351–9. Epub 2015/06/10. https://doi.org/10.1519/JSC.0000000000001063 PMID: 26057190. 27. Rosnow RL, Rosenthal R, Rubin DB. Contrasts and correlations in effect-size estimation. Psychological science. 2000; 11(6):446–53. Epub 2001/02/24. https://doi.org/10.1111/1467-9280.00287 PMID: 11202488. 28. Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive statistics for studies in sports medicine and exercise science. Medicine and science in sports and exercise. 2009; 41(1):3–13. Epub 2008/12/ 19. https://doi.org/10.1249/MSS.0b013e31818cb278 PMID: 19092709. 29. Bangsbo J, Peitersen B. Soccer systems and strategies. Champaign IL: Human Kinetics; 2000. 30. Lago C. The influence of match location, quality of opposition, and match status on possession strate- gies in professional association football. J Sports Sci. 2009; 27(13):1463–9. Epub 2009/09/17. https:// doi.org/10.1080/02640410903131681 PMID: 19757296. 31. Paul DJ, Bradley PS, Nassis GP. Factors affecting match running performance of elite soccer players: shedding some light on the complexity. Int J Sports Physiol Perform. 2015; 10(4):516–9. https://doi.org/ 10.1123/IJSPP.2015-0029 PMID: 25928752. 32. Castellano J, Blanco-Villasenor A, Alvarez D. Contextual variables and time-motion analysis in soccer. Int J Sports Med. 2011; 32(6):415–21. Epub 2011/05/19. https://doi.org/10.1055/s-0031-1271771 PMID: 21590641. 33. Lago-Penas C, Dellal A. Ball possession strategies in elite soccer according to the evolution of the match-score: the influence of situational variables. J Hum Kinet. 2010; 25:93–100. 34. Redwood-Brown AJ, O’Donoghue PG, Nevill AM, Saward C, Dyer N, Sunderland C. Effects of situa- tional variables on the physical activity profiles of elite soccer players in different score line states. Scand J Med Sci Sports. 2018. Epub 2018/07/29. https://doi.org/10.1111/sms.13271 PMID: 30055045. 35. Gregson W, Drust B, Atkinson G, Salvo VD. Match-to-match variability of high-speed activities in pre- mier league soccer. Int J Sports Med. 2010; 31(4):237–42. Epub 2010/02/17. https://doi.org/10.1055/s- 0030-1247546 PMID: 20157871. 36. Krustrup P, Bangsbo J. Physiological demands of top-class soccer refereeing in relation to physical capacity: effect of intense intermittent exercise training. J Sports Sci. 2001; 19(11):881–91. Epub 2001/ 11/07. https://doi.org/10.1080/026404101753113831 PMID: 11695510. 37. Mohr M, Krustrup P, Bangsbo J. Match performance of high-standard soccer players with special refer- ence to development of fatigue. J Sports Sci. 2003; 21(7):519–28. Epub 2003/07/10. https://doi.org/10. 1080/0264041031000071182 PMID: 12848386. 38. Ekblom B. Applied physiology of soccer. Sports medicine (Auckland, NZ). 1986; 3(1):50–60. Epub 1986/01/01. https://doi.org/10.2165/00007256-198603010-00005 PMID: 3633120. 39. Carling C, Bradley P, McCall A, Dupont G. Match-to-match variability in high-speed running activity in a professional soccer team. J Sports Sci. 2016; 34(24):2215–23. Epub 2016/05/05. https://doi.org/10. 1080/02640414.2016.1176228 PMID: 27144879. 40. Trewin J, Meylan C, Varley MC, Cronin J. The match-to-match variation of match-running in elite female soccer. J Sci Med Sport. 2018; 21(2):196–201. Epub 2017/06/10. https://doi.org/10.1016/j.jsams.2017. 05.009 PMID: 28595867. 41. Lacome M, Simpson BM, Cholley Y, Lambert P, Buchheit M. Small-Sided Games in Elite Soccer: Does One Size Fit All? International journal of sports physiology and performance. 2018; 13(5):568–76. Epub 2017/07/18. https://doi.org/10.1123/ijspp.2017-0214 PMID: 28714774. Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 11 / 12 42. Silva P, Chung D, Carvalho T, Cardoso T, Davids K, Araujo D, et al. Practice effects on intra-team syn- ergies in football teams. Human movement science. 2016; 46:39–51. Epub 2015/12/29. https://doi.org/ 10.1016/j.humov.2015.11.017 PMID: 26707679. 43. Di Salvo V, Baron R, Tschan H, Calderon Montero FJ, Bachl N, Pigozzi F. Performance characteristics according to playing position in elite soccer. Int J Sports Med. 2007; 28(3):222–7. Epub 2006/10/07. https://doi.org/10.1055/s-2006-924294 PMID: 17024626. 44. Mohr M, Krustrup P, Andersson H, Kirkendal D, Bangsbo J. Match activities of elite women soccer play- ers at different performance levels. J Strength Cond Res. 2008; 22(2):341–9. Epub 2008/06/14. https:// doi.org/10.1519/JSC.0b013e318165fef6 PMID: 18550946. 45. Carling C, Bloomfield J, Nelsen L, Reilly T. The role of motion analysis in elite soccer: contemporary per- formance measurement techniques and work rate data. Sports medicine (Auckland, NZ). 2008; 38 (10):839–62. Epub 2008/09/23. https://doi.org/10.2165/00007256-200838100-00004 PMID: 18803436. Match-physical demands across tactical systems PLOS ONE | https://doi.org/10.1371/journal.pone.0214952 April 4, 2019 12 / 12

A comparison of match-physical demands between different tactical systems: 1-4-5-1 vs 1-3-5-2.

04-04-2019

Baptista, Ivan,Johansen, Dag,Figueiredo, Pedro,Rebelo, António,Pettersen, Svein Arne

eng

PMC9601160

Citation: Cuenca-Martínez, F.; Sempere-Rubio, N.; Varangot-Reille, C.; Fernández-Carnero, J.; Suso-Martí, L.; Alba-Quesada, P.; Touche, R.L. Effects of High-Intensity Interval Training (HIIT) on Patients with Musculoskeletal Disorders: A Systematic Review and Meta- Analysis with a Meta-Regression and Mapping Report. Diagnostics 2022, 12, 2532. https://doi.org/10.3390/ diagnostics12102532 Academic Editor: Koichi Nishimura Received: 26 September 2022 Accepted: 13 October 2022 Published: 19 October 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). diagnostics Systematic Review Effects of High-Intensity Interval Training (HIIT) on Patients with Musculoskeletal Disorders: A Systematic Review and Meta-Analysis with a Meta-Regression and Mapping Report Ferran Cuenca-Martínez 1,† , Núria Sempere-Rubio 2,† , Clovis Varangot-Reille 1 , Josué Fernández-Carnero 3,4,* , Luis Suso-Martí 1,* , Patricio Alba-Quesada 1 and Roy La Touche 4,5,6 1 Exercise Intervention for Health Research Group (EXINH-RG), Department of Physiotherapy, University of Valencia, 46022 Valencia, Spain 2 UBIC, Department of Physiotherapy, Faculty of Physiotherapy, Universitat de València, 46010 Valencia, Spain 3 Department of Physical Therapy, Occupational Therapy, Rehabilitation and Physical Medicine, Rey Juan Carlos University, 28933 Madrid, Spain 4 Motion in Brains Research Group, Institute of Neuroscience and Sciences of the Movement (INCIMOV), Centro Superior de Estudios Universitarios La Salle, Universidad Autónoma de Madrid, 28049 Madrid, Spain 5 Departamento de Fisioterapia, Centro Superior de Estudios Universitarios La Salle, Universidad Autónoma de Madrid, 28049 Madrid, Spain 6 Instituto de Neurociencia y Dolor Craneofacial (INDCRAN), 28003 Madrid, Spain * Correspondence: josue.fernandez@urjc.es (J.F.-C.); luis.suso@uv.es (L.S.-M.); Tel.: +34-914-88-88-00 (J.F.-C.); +34-963-98-38-55 (L.S.-M.) † These authors contributed equally to this work. Abstract: The aim was to assess the impact of high-intensity interval training (HIIT) on patients with musculoskeletal disorders. We conducted a search of Medline, Embase, PEDro, and Google Scholar. We conducted a meta-analysis to determine the effectiveness of HIIT on pain intensity, maximal oxygen consumption (VO2 max), disability, and quality of life (QoL). We employed the GRADE and PEDro scales to rate the quality, certainty, and applicability of the evidence. Results showed significant differences in pain intensity, with a moderate clinical-effect (SMD = −0.73; 95% CI: −1.40–−0.06), and in VO2 max, with a moderate clinical-effect (SMD = 0.69; 95% CI: 0.42–0.97). However, the meta-analysis showed no statistically significant results for disability (SMD = −0.34; 95% CI: −0.92–0.24) and QoL (SMD = 0.40; 95% CI: −0.80–1.60). We compared HIIT against other exercise models for reducing pain intensity and increasing VO2 max. The meta-analysis showed no significant differences in favour of HIIT. Meta-regression analysis revealed that pain intensity scores were negatively associated with VO2 max (R2 = 82.99%, p = 0.003). There is low-moderate evidence that the HIIT intervention for patients with musculoskeletal disorders can reduce pain intensity and increase VO2 max but has no effect on disability and QoL. Results also showed that HIIT was not superior to other exercise models in reducing pain intensity and increasing VO2 max. Keywords: high-intensity interval training; musculoskeletal pain; pain intensity; VO2 max; disability; quality of life 1. Introduction Musculoskeletal pain is an important public health issue because of its impact on quality of life (QoL) and the disability it can represent [1]. More than 20% of the world’s population is affected by painful conditions, contributing to the high consumption of health- care resources [2]. Pain management can be approached from several perspectives, both pharmacological and non-pharmacological, the latter of which includes physical agents, manual therapy, psychosocial interventions, patient education, and exercise training [3,4]. Exercise therapy has been reported to be highly effective in managing patients with musculoskeletal pain [5] and has been shown to produce hypoalgesia by releasing beta- endorphins or endocannabinoids [6–8]. Exercise therapy also interacts with the autonomic, Diagnostics 2022, 12, 2532. https://doi.org/10.3390/diagnostics12102532 https://www.mdpi.com/journal/diagnostics Diagnostics 2022, 12, 2532 2 of 31 cognitive, and affective aspects of pain [9,10]. For example, a recent meta-analysis found that aerobic exercise led to reduced pain intensity, duration, and frequency as well as improved QoL for patients with migraines [11]. The effects of high-intensity interval training (HIIT) on pain tolerance and threshold have sparked interest among the scientific community concerned with pain [12,13]. As described by Andreato, HIIT is a form of training that alternates high-intensity exercises at 90% or more of the maximal oxygen consumption (VO2 max) (or ≥80% VO2 max for the clinical population) with recovery periods, repeating the exercise several times [14]. A number of articles have recently shown that HIIT could improve pain-related clinical variables in patients with musculoskeletal disorders [15–17]. To date, systematic reviews on HIIT have mainly focused on patients with cardiovascular diseases, cancer, or obesity, where HIIT has shown great effectiveness in modifying cardiorespiratory variables [18–20]. Picavet et al. found that disability and quality of life are commonly affected in patients with musculoskeletal disorders [1]. This work prompted us to include these two variables in our study, with the objective of evaluating the role of this therapeutic exercise model on this clinical population of patients with musculoskeletal disorders. In addition to this, we wanted to include the pain intensity variable because almost 1/5 of the world’s population lives with clinical conditions that involve pain [2]. Finally, we also wanted to include the variable VO2 max because it is an objective variable and, in addition, it is the gold standard for assessing cardiorespiratory fitness, which seems to be affected in patients with musculoskeletal disorders with associated pain [21]. As far as we know, no published review has assessed the effects of HIIT on clinical and cardiorespiratory variables in patients with musculoskeletal disorders and pain. Therefore, the main aim of the present study was to develop a systematic review and meta-analysis to assess the effectiveness of HIIT on pain intensity, maximal oxygen con- sumption, disability, and health-related QoL for patients with musculoskeletal disorders. 2. Materials and Methods This systematic review and the meta-analysis were performed according to the Pre- ferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) guidelines described by Moher [22]. The protocol of this systematic review and meta-analysis was reg- istered in an international registry prior to starting the review (Prospero: CRD42020216298 (5 November 2020)). 2.1. Inclusion Criteria The selection criteria used in this systematic review and meta-analysis were based on methodological and clinical factors, such as the Population, Intervention, Control, Outcomes, and Study Design (PICOS) described by Stone [23]. 2.1.1. Population The participants selected for the studies were patients older than 18 years with any kind of musculoskeletal disorder. The participants’ gender was irrelevant. 2.1.2. Intervention and Control The intervention was the HIIT exercise modality, which could be given as an indepen- dent treatment, added to an existing intervention, or embedded in an existing intervention (e.g., usual care and treatment). For the control group, the comparators were minimal intervention, no intervention, and usual care (e.g., maintenance of the habitual daily physi- cal activity profile, standard physical activity recommendations, physical exercise habits, and exercise intervention [excluding HIIT modality]) in combination or not with placebo interventions. In addition, we performed a sub-analysis to evaluate the effectiveness of HIIT compared with other therapeutic exercise models (e.g., moderate-intensity exercise, high-intensity continuous training, and home exercises) in those articles that, in addition Diagnostics 2022, 12, 2532 3 of 31 to a control or comparator with no intervention or minimal intervention, presented an additional group that performed an exercise model. 2.1.3. Outcomes The measures used to assess the results and effects were pain intensity, VO2 max, disability, and health-related QoL. 2.1.4. Study Design We selected randomised controlled trials (RCTs), randomised parallel-design con- trolled trials, randomised cross-over trials, and prospective controlled clinical trials. 2.2. Search Strategy The search for studies was performed using Medline (PubMed) (1950–2020), Embase (1950–2020), PEDro (1950–2020), and Google Scholar. The first search was run on the 8 November 2020 (however, the search was updated on 31 January 2022). We used a validated search filter for retrieving studies on measurement properties in PubMed; the same filter was adapted for all other databases [24]. In addition, the search was adapted and performed in Google Scholar due to its capacity to search for relevant articles and grey literature [25,26]. No restrictions were applied to any specific language as recommended by the international criteria [27]. The search strategy combined medical subject headings (MeSH) and non-MeSH terms, adding a Boolean operator (OR and/or AND) to combine them. The terms were as follows: “Hig- Intensity Interval Training”, “High-Intensity Inter- val Trainings”, “Interval Training, High-Intensity”, “Interval Trainings, High Intensity”, “Training, High-Intensity Interval”, “Trainings, High-Intensity Interval”, “High-Intensity Intermittent Exercise”, “Exercise, High-Intensity Intermittent”, “Exercises, High-Intensity Intermittent”, “High-Intensity Intermittent Exercises”, “Sprint Interval Training”, “Sprint Interval Trainings”, “Pain”, “Chronic Pain”, “Musculoskeletal Pain”, “Pain intensity”, “Dis- ability”, “Quality of Life”, “VO2 max”, “Maximal Oxygen Consumption”, and “Maximal Oxygen Uptake”. Two independent reviewers (F.C.-M. and J.F.-C.) conducted the search using the same methodology, and the differences were resolved by consensus. Additionally, meticulous manual searches were performed, including journals that have published articles related to the topic of this review as well as reference lists of the included studies. The reference sections of the original studies were screened manually. To remove duplicates, we employed the citation management software Mendeley (Mendeley desktop v1.17.4, Elsevier, New York, NY, USA) and hand-checked the citations [28]. 2.3. Selection Criteria and Data Extraction First, two independent reviewers (F.C.M. and L.S.M.), who assessed the relevance of the RCTs regarding the study questions and aims, performed a data analysis, which was performed based on information from the title, abstract, and keywords of each study. If there was no consensus or the abstracts did not contain sufficient information, the full text was reviewed. In the second phase of the analysis, the full text was used to assess whether the studies met all the inclusion criteria. Differences between the two independent reviewers were resolved by a consensus process moderated by a third reviewer [29]. Data described in the results were extracted by means of a structured protocol that ensured that the most relevant information was obtained from each study [30]. 2.4. Methodological Quality Assessment We used the Cochrane Handbook for Systematic Reviews of Interventions version 5.1.0 to assess the risk of bias in the included studies [30]. The assessment tool covers a total of 7 domains: (1) random sequence generation (selection bias), (2) allocation concealment (selection bias), (3) blinding of participants and personnel (performance bias), (4) blinding of outcome assessments (detection bias), (5) incomplete outcome data (attrition bias), Diagnostics 2022, 12, 2532 4 of 31 (6) selective reporting (reporting bias), and (7) other biases. Bias was assessed as low risk, high risk, or unclear risk. The studies’ methodological quality was assessed using the PEDro scale [31], which assesses the internal and external validity of a study and consists of 11 criteria: (1) spec- ified study eligibility criteria, (2) random allocation of patients, (3) concealed allocation, (4) measure of similarity between groups at baseline, (5) patient blinding, (6) therapist blinding, (7) assessor blinding, (8) fewer than 15% dropouts, (9) intention-to-treat analysis, 10) intergroup statistical comparisons, and 11) point measures and variability data. The methodological criteria were scored as follows: yes (1 point), no (0 points), or do not know (0 points). The PEDro score for each selected study provided an indicator of the methodological quality (9–10 = excellent; 6–8 = good; 4–5 = fair; 3–0 = poor) [32]. We used the data obtained from the PEDro scale to map the results of the quantitative analyses. Two independent reviewers (F.C.-M. and L.S.-M.) examined the quality of all the selected studies using the same methodology. Disagreements between the reviewers were resolved by consensus with a third reviewer. The concordance between the results (inter- rater reliability) was measured using Cohen’s kappa coefficient (κ) as follows: (1) κ > 0.7 indicated a high level of agreement between assessors; (2) κ = 0.5–0.7 indicated a moderate level of agreement; and (3) κ < 0.5 indicated a low level of agreement) [33]. 2.5. Evidence Map We created a visual map of the scientific evidence for each article to visually display the information as a bubble plot. The review information is based on 3 dimensions: 1. Type of outcome measure (bubble colour): The bubble colour represents the variables (pain intensity, blue; VO2 max, violet; disability, green; QoL, black). 2. Variable (x-axis): We employed the calculation of effect sizes. 3. Effect (y-axis): Each of the reviews was classified according to its methodological quality using the PEDro scale. 4. Statistically significant differences: Articles with statistically significant differences were marked with white dots. 2.6. Certainty of Evidence The certainty of evidence analysis was based on classifying the results into levels of evidence according to the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) framework, which is based on five domains: study design, imprecision, indirectness, inconsistency, and publication bias [34]. The assessment of the five domains was conducted according to GRADE criteria [35,36]. Evidence was categorised into the following four levels accordingly: (a) High quality. Further research is very unlikely to change our confidence in the effect estimate. All five domains are also met; (b) Moderate quality. Further research is likely to have an important impact on our confidence in the effect estimate and might change the effect estimate. One of the five domains is not met; (c) Low quality. Further research is very likely to have a significant impact on our confidence in the effect estimate and is likely to change the estimate. Two of the five domains are not met; and, finally, (d) Very low quality. Any effect estimates are highly uncertain. Three of the five domains are not met [35,36]. For the study design domain, the recommendations were downgraded one level in the event there was an uncertain or high risk of bias and serious limitations in the effect estimate (more than 25% of the participants were from studies with fair or poor methodological quality, as measured by the PEDro scale). In terms of inconsistency, the rec- ommendations were downgraded one level when the point estimates varied widely among studies, the confidence intervals showed minimal overlap, or when the I2 was substantial or large (greater than 50%). At indirectness domain recommendations were downgraded when severe differences in interventions, study populations or outcomes were found (the recommendations were downgraded in the absence of direct comparisons between the interventions of interest or when there are no key outcomes, and the recommendation is Diagnostics 2022, 12, 2532 5 of 31 based only on intermediate outcomes or if more than 50% of the participants were outside the target group). For the imprecision domain, the recommendations were downgraded by one level if there were fewer than 300 participants for the continuous data [37]. 2.7. Data Synthesis and Analysis The statistical analysis was conducted using MetaXL software (version 5.3 (EpiGear International, Sunrise Beach, Queensland, Australia) [38]. To compare the outcomes re- ported by the studies, we calculated the standardised mean difference (SMD) over time and the corresponding 95% confidence interval (CI) for the continuous variables. The statistical significance of the pooled SMD was examined as Hedges’ g to account for a possible overestimation of the true population effect size in the small studies [39]. We used the same inclusion criteria for the systematic review and the meta-analysis and included three additional criteria: (1) In the results, there was detailed information regarding the comparative statistical data of the exposure factors, therapeutic interventions, and treatment responses; (2) the intervention was compared with a similar control group; and (3) data on the analysed variables were represented in at least three studies. The estimated SMDs were interpreted as described by Hopkins et al. [40], that is, we considered that an SMD of 4.0 represented an extremely large clinical effect, 2.0–4.0 represented a very large effect, 1.2–2.0 represented a large effect, 0.6–1.2 represented a moderate effect, 0.2–0.6 represented a small effect, and 0.0–0.2 represented a trivial effect. We estimated the degree of heterogeneity among the studies using Cochran’s Q statistic test (a p-value < 0.05 was considered significant) and the inconsistency index (I2) [40]. We considered that an I2 > 25% represented small heterogeneity, I2 > 50% represented medium heterogeneity, and I2 > 75% represented large heterogeneity [41]. The I2 index is a complement to the Q test, although it has the same problems of power with a small number of studies [41]. When the Q-test was significant (p < 0.1) and/or the result of I2 was >75%, there was heterogeneity among the studies, and the random-effects model was conducted in the meta-analysis. To detect publication bias and to test the influence of each individual study, we performed a visual evaluation of the Doi plot [42], seeking asymmetry. We also performed a quantitative measure of the Luis Furuya-Kanamori (LFK) index, which has been shown to be more sensitive than the Egger test in detecting publication bias in a meta- analysis of a low number of studies [43]. An LFK index within ±1 represents no asymmetry, exceeding ±1 but within ±2 represents minor asymmetry, and exceeding ±2 involves major asymmetry. To test each study’s influence, we visually examined the forest plot and performed an exclusion sensitivity analysis. Lastly, we applied a meta-regression analysis to analyse the relationship between pain intensity and VO2 max variables using a random effects model employing the effect size statistic (Hedges’ g) of the pain intensity scores to correlate with the VO2 max scores [44]. 3. Results The study search strategy was presented in the form of a flow diagram (Figure 1). 3.1. Characteristics of the Included Studies The patients were diagnosed with a persistent musculoskeletal pain condition [2 knee osteoarthritis studies [45,46], two axial spondylarthritis studies [16,47], three studies on chronic nonspecific low back pain [17,48,49], one study on episodic migraineurs [50], one study on fibromyalgia [15], one study on subacromial pain syndrome [51], one study on rheumatoid arthritis and adult-juvenile idiopathic arthritis [52], and one study on general persistent pain condition with previous trauma [53], and all of them evaluated pain intensity, VO2 max, disability, and health-related QoL. Table 1 lists the descriptive characteristics of the included studies. Diagnostics 2022, 12, 2532 6 of 31 3.2. Interventions In all groups, HIIT was compared to other types of training or interventions (including controls and no interventions), with the exception of Bressel et al. [45], which studied a single HIIT and balance training group, and Sveaas et al. (2014 & 2019) [16,47], which included an HIIT and moderate-intensity continuous training (MICT) group and another no exercise group. Of the studies referred to above, three had two groups: one HIIT group and one MICT group [15,17,46]. Atan and Karavelio˘glu [15] included a third standard care group. Two other studies had only one HIIT and one standard care group [48,51]. Two studies had an HIIT group and another group that maintained the activities of daily living [52] and their usual physical activity [54]. Flehr et al. [53] had one HIIT group and one yoga group, while Verbrugghe et al. [49] studied four groups with different types of HIIT. The total duration of the intervention ranged from 6 to 12 weeks, with most studies having a frequency of two to three times per week, except for Keogh et al. [46] and Atan and Karavelio˘glu [15], which had frequencies of four and five times per week, respectively. Table 2 presents extensive details on the intervention characteristics of the included studies. , 12, x FOR PEER REVIEW 6 of 32 Figure 1. PRISMA Flowchart for selecting studies. 3.1. Characteristics of the Included Studies The patients were diagnosed with a persistent musculoskeletal pain condition [2 knee osteoarthritis studies [45,46], two axial spondylarthritis studies [16,47], three studies on chronic nonspecific low back pain [17,48,49], one study on episodic migraineurs [50], Figure 1. PRISMA Flowchart for selecting studies. Diagnostics 2022, 12, 2532 7 of 31 Table 1. Characteristics of the included studies. Author, Year Country Population Disease (n) Age (Years) Sex (%) Diagnostic Criteria Disease Duration (Years) Study Design—Duration Intervention(s) and Control Group (n) Outcome Measured (Instrument) Results Atan et al., 2020 [15] Turkey Fibromyalgia (n = 55) Age, 48.7 ± 9.1 y 100% F American College of Rheumatology 2016 diagnostic criteria Duration, 2.5 ± 1.6 y Pilot ROT—6 weeks Intervention - HIIT (n = 19) - MICT (n = 19) Control Usual care (n = 17) - Pain Intensity (VAS) - HRQoL (SF-36 PF, PRL, Pain, GH, V, SF, ER, MH, EWB, E/F, HC) - VO2 max (mL/kg/min) HIIT showed significant differences compared with a control group on pain intensity, VO2 max, and SF-36 PF, PRL, ER, E/F, EWB, GH, and HC (p < 0.05) but no significant difference compared with MCT. Berg et al., 2020 [50] Norway Chronic SAPS (n = 21) Age, 48.1 ± 12.5 y 48% F/52% M Clinical criteria Duration, 3.5 ± 4.8 y RCT—8 weeks Intervention HIIT + Home-exercise (n = 13) Control Home-exercise (n = 8) - Pain intensity (NPA) - Disability (SPADI) HIIT showed significant intragroup (p < 0.05) and intergroup differences (p < 0.05) compared with a control group in terms of disability but no significant difference in pain intensity. Bressel et al., 2014 [44] United States Knee OA (n = 18) Age, 64.5 ± 10.2 y 89% F/11% M Clinical and radiological criteria Duration, 6.8 ± 7.4 y Pre-post study—6 weeks Intervention - HIIT + Balance training (n = 18) Control No intervention (n = 18) Pain Intensity (VAS) HIIT showed a significant improvement in pain intensity (p < 0.05). Flehr et al., 2019 [52] Australia Persistent pain condition (n = 32) Age, 30.2 ± 8 y 100% F N/R Duration, More than 12 months RCT—8 weeks Intervention HIIT (n = 15) Control Bikram Yoga (n = 17) - Pain Intensity (BPI) - HRQoL (SF-36 PF, PRL, Pain, GH, V, SF, ER, MH) No significant difference between HIIT and Bikram Yoga in pain intensity. There was a significant intergroup difference on quality of life (SF-36 PF: p = 0.019; SF-36 MH: p = 0.005), with yoga showing higher improvement (SF-36 PF: M = 80.91; SF-36 MH: M= 63.94). Hanssen et al., 2018 [49] Switzerland Episodic migraine without aura (n= 36) Age, 36.8 ± 10.3 y 81% F/19% M International classification of headache disorders, 3rd ed. Duration, N/R RCT—12 weeks Intervention - HIIT (n = 13) - MICT (n = 11) Control Group No intervention (n = 12) VO2 max (mL/kg/min) No group × time interaction between the three groups (p = 0.14). Diagnostics 2022, 12, 2532 8 of 31 Table 1. Cont. Author, Year Country Population Disease (n) Age (Years) Sex (%) Diagnostic Criteria Disease Duration (Years) Study Design—Duration Intervention(s) and Control Group (n) Outcome Measured (Instrument) Results Keogh et al., 2018 [45] Australia Knee OA (n = 17) Age, 62.4 ± 8.3 y 76% F/24% M Diagnosis by an orthopaedic surgeon Duration, 4.7 ± 4.6 y Pilot RCT—8 weeks Intervention HIIT (n = 9) Control MICT (n = 8) - Disability (WOMAC, Lequesne Index) Both interventions demonstrated significant benefits on the WOMAC (HIIT: p = 0.05; MICT: p = 0.006) but without intergroup differences. No patient had significant improvement in the Lequesne index. Sandstad et al., 2015 [51] Norway RA and JIA (n = 27) Age, 33.0 ± 8.1 y 100% F Diagnosis by a rheumatologist Duration, N/R Cross-over trial—10 weeks Intervention HIIT (n = 12) Control No intervention (n = 15) - Pain Intensity (VAS) - Disability (MHAQ) - VO2 max (mL/kg/min) HIIT had a significant improvement in VO2 max (p < 0.001) but no difference in pain intensity and disability. Sveaas et al., 2014 [49] Norway axSpA (n = 24) Age, 48.5 ± 12.0 y 50% F/50% M Spondyloarthritis International Society criteria Duration, 24.9 ± 15.8 y Pilot RCT—12 weeks Intervention HIIT (n = 10) Control Usual care (n = 14) VO2 max (mL/kg/min) HIIT had a significantly higher VO2 max at 12 weeks than the control group (p < 0.001) Sveaas et al., 2019 [16] Norway axSpA (n = 97) Age, 46.2 ± N/R y 53% F/47% M Spondyloarthritis International Society criteria Duration, N/R RCT—12 weeks Intervention HIIT (n = 48) Control No intervention (n = 49) - Pain intensity (BASDAI neck/back/hip and peripheral pain) - VO2 max (mL/kg/min) HIIT significantly improves the neck/back/hips, and peripheral pain intensity, and the VO2 max more than the control group (p < 0.001; p = 0.016; p < 0.001). Thomsen et al., 2019 [53] Norway PsA (n = 67) Age, 48.0 ± 11.5 y 64% F/36% M Classification of psoriatic arthritis Study group criteria Duration, N/R RCT—11 weeks Intervention HIIT (n = 32) Control No intervention (n = 35) - Pain Intensity (VAS) HIIT showed no clear effect on pain intensity at the end of the intervention and at 9 months of follow-up. Diagnostics 2022, 12, 2532 9 of 31 Table 1. Cont. Author, Year Country Population Disease (n) Age (Years) Sex (%) Diagnostic Criteria Disease Duration (Years) Study Design—Duration Intervention(s) and Control Group (n) Outcome Measured (Instrument) Results Verbrugghe et al., 2018 [47] Belgium Nonspecific Chronic LBP (n = 20) Age, N/R 55% F/45% M Clinical criteria Duration, N/R CCT—6 weeks Intervention HIIT (n = 10) Control Usual care (n = 10) - Pain Intensity (NPRS) - Disability (RMDQ) - HRQoL (SF-36 PF, PRL, ER, E/F, EWB, SF, Pain, GH) - VO2 max (mL/kg/min) Both groups had a reduction in disability (p < 0.05) with no intergroup difference. HIIT improved significantly HRQoL (SF-36 PRL, ER, SF, and Pain) (p < 0.05) but with no intergroup differences. Verbrugghe et al., 2019 [17] Belgium Nonspecific Chronic LBP (n = 36) Age, 44.2 ± 9.8 y 68% F/32% M Clinical criteria Duration, 11.1 ± 7.7 y RCT—12 weeks Intervention HIIT (n = 18) Control MIT (n = 18) - Pain Intensity (NPRS) - Disability (MODI) - VO2 max (mL/kg/min) HIIT significantly improved disability and VO2 max more than MIT (p < 0.05). HIIT significantly reduced pain intensity (p < 0.05) but with no significant differences with MIT. Verbrugghe et al., 2020 [48] Belgium Nonspecific chronic LBP (n = 80) Age, 44.1 ± 9.7 y 58% F/42% M Clinical criteria Duration, 13.4 ± 9.1 y RCT—12 weeks Intervention - HITCOM (n = 19) - HITSTRE (n = 21) - HITSTAB (n = 20) - HITMOB (n = 20) - Pain Intensity (NPRS) - Disability (MODI) - VO2 max (mL/kg/min) All four HIIT groups significantly reduced pain intensity and disability and increased VO2 max (p < 0.05), with no intergroup differences. axSpA, axial spondyloarthritis; BPI, Brief Pain Inventory; CCT, Controlled clinical trial; E/F, energy/fatigue; ER, emotional role limitation; EWB, emotional well-being; GH, general health; HC, health change; HIIT, high-intensity interval training; HITCOM, high-intensity general resistance training, and high-intensity core strength training; HITMOB, trunk mobility exercises; HITSTAB, high-intensity core strength training; HITSTRE, high-intensity general resistance training; HRQoL, health-related quality of life; JIA, juvenile idiopathic arthritis; LBP, low back pain; MCT, moderate continuous training; MH, mental health; MHAQ, Modified Health Assessment Questionnaire; MICT, moderate-intensity continuous training; MIT, moderate-intensity training; MODI, Modified Oswestry Index; MPQ, McGill Pain Questionnaire; N/R, not reported; NPRS, Numeric Pain Rating Scale; OA, osteoarthritis; ODI, Oswestry Disability Index; PF, physical functioning; PRL, physical role limitation; PsA, psoriatic arthritis; RA, rheumatoid arthritis; RCT, randomised control trial; RMDQ, Roland-Morris Disability Questionnaire; SF-36, Short Form-36 Health Survey; SAPS, subacromial pain syndrome; SF, social functioning; SPADI, Shoulder Pain and Disability Index; V, vitality; VAS, visual analogue scale; WOMAC, Western Ontario and McMaster Universities Osteoarthritis Index. Diagnostics 2022, 12, 2532 10 of 31 Table 2. Prescription parameters extracted from each included study. Trial Group Exercise Protocol (Distribution and Exercise Type) Intensity (Pain Control during Training) Frequency and Duration Exercise Testing Atan et al., 2020 [15] HIIT (AerT) + StrT + Stretching Total exercise duration: 35 min Warmup and cooldown: 5 min stationary cycling. HIIT protocol: 4 × 4 min of high-intensity stationary cycling interval alternating with 3 min cycling recovery periods. Work/rest ratio: [1:0.75] Followed by 10 min full body (shoulder, arm, leg, and hip) StrT, using 1–3-kg weights (1 × 8–10 rep) and 5 min stretching (4–5 × 20–30 s for each muscle group). Measurement: HRmax (Monitorisation: N/R) Warmup and cooldown: 50% HRmax HIIT: Interval: 80–95% HRmax Active Rest: 70% HRmax StrT: N/R Pain: N/R 5×/week 6 weeks Maximal cardiopulmonary test on a cycloergometer at baseline and follow-up. HRmax, VO2 max, BP, workload, MET and duration-of-test were recorded. MICT (AerT) + StrT + Stretching Total exercise duration: 55 min. Warmup and cooldown: 5 min stationary cycling. MICT protocol: 45 min continuous stationary cycling Followed by 10 min full body (shoulder, arm, leg, and hip) StrT, using 1–3-kg weights (1 × 8–10 rep) and 5 min stretching (4–5 × 20–30 s for each muscle group). Measurement: HRmax (Monitorisation: N/R) Warmup and cooldown: 50% HRmax MICT: 65–70% HRmax StrT: N/R Pain: N/R Usual Care Recommendations regarding exercise for fibromyalgia. N/A Berg et al., 2020 [50] HIIT (StrT) + Usual Care HIIT protocol: 4 × 4 min shoulder abduction-adduction at 2 Hz intervals alternating with 3 min walking rest periods Work/Rest Ratio: [1:0.75] If the patient was able to continue the final interval for one additional minute, the workload was increased by 250 g in the following session. Home-based exercises: Scapular stabilising, rotator cuff, and pain-free ROM exercises. Measurement: WRmax Interval: 80% WRmax Rest: N/R Pain: When pain exceeds 5/10, session was ended. 3×/week 8 weeks Time to exhaustion test during shoulder abduction-adduction. WRmax was recorded. Usual Care Home-based exercises: Scapular stabilising, rotator cuff, and pain-free ROM exercises. N/R Bressel et al., 2014 [44] BalanceT + HIIT (AerT) Balance training: Perturbations with water jets. Followed by: HIIT protocol: (Progressive increase from 1st to 6th week) 3 to 6 × 0.5 to 2.5 min walking (1.3 to 2.1 m/s) on an underwater treadmill interval alternating with 1 to 2.5 min walking (1.3 to 1.8 m/s) rest periods. (depth: xiphoid process) Work/rest ratio: [1:2; 1:1.3; 1:1; 1:1; 1:1; 1:1] Measurement: RPE (Borg Scale/20) BalanceT: Progressive increase (from 1st to 6th week) from 11 to 18/20. HIIT: Interval: Progressive increase (from 1st to 6th week) from 13 to 19/20. Rest: 10/20. Pain: N/R 3×/week 6 weeks N/A Diagnostics 2022, 12, 2532 11 of 31 Table 2. Cont. Trial Group Exercise Protocol (Distribution and Exercise Type) Intensity (Pain Control during Training) Frequency and Duration Exercise Testing Flehr et al., 2019 [52] HIIT (StrT + AerT) 45 min functional training incorporating running, throwing, standing from a seated position, placing items overhead, and picking items up. Warmup and demonstration: 15 min. Movement learning: 15 min HIIT protocol: 15 min reproduction of the movement at high intensity. Four formats possible: As fast as possible, 8-exercises Tabata intervallic training followed by AerT, Maximum reps or load in a set time, or as many rounds as possible in 12 min followed by AerT N/R Interval: N/R Rest: N/R Pain: N/R 3×/week 8 weeks N/R Yoga 90 min Bikram Yoga class (Room at 40 ◦C and 40% humidity): Deep breathing, 45 to 50 min standing, stretching, and relaxation postures. Light to moderate (according to ACSM) and sometimes vigorous. Pain: N/R Hanssen et al., 2018 [49] HIIT (AerT) Warmup: 400 m of light running on a treadmill and 2 skipping exercises HIIT protocol: 4 × 4 min high-intensity running on a treadmill, interval alternating with 3 min running recovery periods. Work/rest ratio: [1:0.75] Cooldown: 400 m of light running and stretching Measurement: HRmax (HR checked using HR monitor) Interval: 90% to 95% HRmax (±5 bpm) Rest: 70% of HRmax Pain: N/R 2×/week 12 weeks Maximal Cardiopulmonary test on a treadmill. Anaerobic lactate-threshold, HRmax, RPE, and VO2 max were recorded. MICT (AerT) Warmup: 400 m of light running on a treadmill and 2 skipping exercises MICT protocol: 45 min continuous running on a treadmill. Cooldown: 400 m of light running and stretching Measurement: HRmax (HR checked using HR monitor) MICT: 70% HRmax (± 5 bpm) Pain: N/R Maintain their habitual daily physical activity N/A N/A Keogh et al., 2018 [45] HIIT (AerT) Warmup: 7 min stationary cycling, with progressively increasing intensity HIIT protocol: 5 × 45 s high-cadence stationary cycling interval alternating with 90 s low-intensity recovery cycling. Work/Rest Ratio: [1:2] Cooldown: 6–7 min of light to moderate cycling. HIIT: Interval: 110 rpm with a resistance similar or slightly higher than the rest. Intensity was defined as “an intensity at which you felt it was quite difficult to complete sentences during the exercise”. Rest: ∼70 rpm To avoid pain, progressive increase in initial sessions. 4×/week 8 weeks N/R MICT (AerT) Warmup and cooldown: Light intensity cycling for 3 min and 2 min, respectively. MICT protocol: 20 min continuous cycling. MICT: 60–80 rpm. Intensity was defined as “An intensity at which you are able to speak in complete sentences during the exercise”. To avoid pain, progressive increase in initial sessions Diagnostics 2022, 12, 2532 12 of 31 Table 2. Cont. Trial Group Exercise Protocol (Distribution and Exercise Type) Intensity (Pain Control during Training) Frequency and Duration Exercise Testing Sandstad et al., 2015 [51] HIIT (AerT) Warmup: 10 min stationary cycling at moderate intensity HIIT protocol: 4 × 4 min high-intensity stationary cycling interval alternating with 3 min cycling recovery periods. The speed and workload were adjusted continuously. Measurement: HRmax (HR checked using HR monitor) Warmup: ~70% Interval: 85–95% of HRmax Rest: ~70% of HRmax Pain: N/R 2×/week 10 weeks Maximal cardiopulmonary test on a bike. VO2 max and HRmax (defined as the highest HR during the test plus 5 bpm). Maintain daily life activities N/A N/A Sveaas et al., 2014 and 2019 [16,49] HIIT (AerT) + StrT + MICT (AerT) Twice a week, supervised HIIT and StrT: - HIIT protocol: 4 × 4 min walking/running on a treadmill interval alternating with 3 min of active resting. - StrT protocol: 20 min with external load (2–3 × 8–10 rep): Bench press or chest press machine, weighted squat or leg press machine, rowing with weights, triceps and biceps machine, and abdominal bridge. Once a week, individual interval training or MICT: 40 min of either interval training or MICT. Measurement: HRmax (HR checked using HR monitor) HIIT: Interval: 90–95% HRmax Rest: 70% HRmax MICT intensity: >70% HRmax Pain: Exercises were adapted if pain was ≥ 5/10 3×/week 12 weeks Cardiopulmonary test on a walking treadmill (modified Balke protocol). VO2 max and HRmax were recorded. Asked to not start exercise N/A N/A Thomsen et al., 2019 [53] HIIT (AerT) Warmup: 10 min. HIIT protocol: 4 × 4 min high-intensity stationary cycling interval alternating with a 3 min cycling recovery period. Work/rest ratio: [1:0.75] Supervised twice a week and individually once a week. Participants were instructed in using the HIIT concept by, for example, running, bicycling, or walking uphill. Measurement: HRmax (HR checked using HR monitor) Interval: 85–95% HRmax Rest: 70% HRmax Pain: N/R 3×/week 11 weeks Maximal cardiopulmonary test on a bike. VO2 max and HRmax (defined as the highest HR during the test more 5 bpm). Maintain daily physical activity N/A N/A Diagnostics 2022, 12, 2532 13 of 31 Table 2. Cont. Trial Group Exercise Protocol (Distribution and Exercise Type) Intensity (Pain Control during Training) Frequency and Duration Exercise Testing Verbrugghe et al., 2018 [47] HIIT (AerT) + High Intensity StrT HIIT protocol: -Warmup: 5 min -Followed by HIIT training: 5 × 1 min high-intensity stationary cycling interval alternating with 1 min of rest. Weekly increase of interval duration by 10 s until week 6. Work/rest ratio: [1:1; 1.2:1; 1.3:1; 1.5:1; 1.7:1; 1.8:1] High load whole body StrT training protocol: 3 upper body (pulley biceps curl, pulley chest press, and pulley vertical traction behind the neck) and 3 lower body exercises (leg press, leg extension, and leg curl) with external load: 1 to 2 × 8–12 rep. Measurement: VO2 max and 1RM (Monitorisation: N/R) Interval: VO2 max workload Rest: N/R StrT: 80% 1RM Pain: N/R 2×/week 6 weeks Maximal cardiopulmonary testing (Graded exercise test) on a bike. VO2 max, expiratory volume, respiratory exchange ratio, and HR were recorded A 1RM test was performed for every exercise. Usual Physiotherapy Care MICT protocol: 50 min continuous cycling, cross-training, and/or treadmill walking. Control motor exercise: Addressing lumbopelvic motor control impairments. Trunk StrT: Unstable posture corrections, plank, and bridge variations Measurement: HRmax (Monitorisation: N/R) MICT: 60–65% HRmax Pain: N/R Verbrugghe et al., 2019 [17] HIIT (AerT) + High-intensity Global and Core StrT HIIT protocol: -Warmup: 5 min cycling -HIIT Training: 5 × 1 min high-intensity cycling interval alternating with a 1 min cycling recovery period. Weekly increase of interval duration of 10 s until week 6. Work/rest ratio: [1:1; 1.2:1; 1.3:1; 1.5:1; 1.7:1; 1.8:1] High-intensity StrT: 3 upper body (vertical traction, chest press, arm curl) and 3 lower body exercises (leg curl, leg press, leg extension) executed with external load on machines: 1 × maximum 12 rep Core muscle training: 6 static core exercises [glute bridge, resistance band glute clam, lying diagonal back extension, adapted knee plank, adapted knee side plank, elastic band shoulder retraction with hip hinge): 1 × 10 rep of a 10 s static hold. Measurement: % VO2 max, %1RM and %MVC (Monitorisation: N/R) HIIT: Interval: 110 rpm at 100% VO2 max workload Rest: 75 rpm at 50% VO2 max workload StrT: 80% 1RM 5% workload increase when the participant was able to perform more than 10 reps on 2 consecutive sessions. Core: Between 17% and 100% MVC of m. transversus abdominis, m. multifidus, m. gluteus. Progressive increase of time and load (body weight bearing, elastic or weights). Pain: N/R 2×/week 12 weeks Maximal cardiopulmonary test on a bicycle. VO2 max, Maximal workload, LA, and HR were recorded. Workload was updated, with a complementary cardiopulmonary test, for the last 6 weeks. 1RM testing was performed for every exercise. MICT (AerT) + Moderate intensity Global and Core STrT MICT protocol: Cycling on a cycle ergometer. - Warmup: 5 min. - MICT: Continuous 14 min cycling at moderate intensity. Duration increased by 100 s every 2 sessions up to 22 min 40 s. Moderate intensity Global StrT: Same exercises as above, but at moderate intensity: 1 × 15 rep. Moderate intensity core training: Same exercises as above but at moderate intensity: 1 × 10 repetitions of a 10 s static hold. Measurement: % VO2 max, %1RM and %MVC (Monitorisation: N/R) MICT: 90 rpm at 60% VO2 max workload StrT: 60% of 1RM Core training: N/R Pain: N/R Diagnostics 2022, 12, 2532 14 of 31 Table 2. Cont. Trial Group Exercise Protocol (Distribution and Exercise Type) Intensity (Pain Control during Training) Frequency and Duration Exercise Testing Verbrugghe et al., 2020 [48] HIIT (AerT) + Global StrT HIIT protocol: - Warmup: 5 min cycling - HIIT Training: 5 × 1 min high-intensity cycling interval alternating with a 1 min cycling recovery period. Weekly increase of interval duration, of 10 s, until week 6. Work/rest ratio: [1:1; 1.2:1; 1.3:1; 1.5:1; 1.7:1; 1.8:1] High-intensity StrT: 3 upper body (vertical traction, chest press, arm curl) and 3 lower body exercises (leg curl, leg press, leg extension) executed with external load on machines: 2 × maximum 12 rep Measurement: % VO2 max and %1RM (Monitorisation: N/R) HIIT: Interval: 110 rpm at 100% VO2 max workload Rest: 75 rpm at 50% VO2 max workload StrT: 80% 1 RM Weight was increased when the participant was able to perform more than 10 reps on 2 consecutive sessions. Pain: N/R 2×/week 12 weeks Maximal cardiopulmonary test on a bicycle. VO2 max, expiratory volume, respiratory exchange ratio, and HR were recorded. Parameters were adapted at 6 weeks with another cardiopulmonary test. 1RM testing was performed for every exercise. HIIT (AerT) + Core StrT HIIT protocol: Same HIIT protocol as above. Core muscle training: 6 static core exercises [glute bridge, resistance band glute clam, lying diagonal back extension, adapted knee plank, adapted knee side plank, elastic band shoulder retraction with hip hinge): 2 × 10 rep of a 10 s static hold. Measurement: % VO2 max and %MVC (Monitorisation: N/R) HIIT: Interval: 110 rpm at 100% VO2 max workload Rest: 75 rpm at 50% VO2 max workload Core: 40–60% of the MVC of m. transversus abdominis, m. multifidus, m. gluteus. Progressive increase of time and load. Pain: N/R HIIT (AerT)+ Global and Core StrT HIIT protocol: Same HIIT protocol as above. High intensity StrT: Same exercise as above: 1 × maximum 12 rep Core muscle training: Same exercise as above: 1 × 10 rep of a 10 s static hold. Measurement: % VO2 max, %1RM and %MVC (Monitorisation: N/R) HIIT: Interval: 110 rpm at 100% VO2 max workload Rest: 75 rpm at 50% VO2 max workload StrT: 80% 1 RM Weight was increased when the participant was able to perform more than 10 reps on 2 consecutive sessions. Core: 40–60% of the MVC of m. transversus abdominis, m. multifidus, m. gluteus. Progressive increase in time and load Pain: N/R Diagnostics 2022, 12, 2532 15 of 31 Table 2. Cont. Trial Group Exercise Protocol (Distribution and Exercise Type) Intensity (Pain Control during Training) Frequency and Duration Exercise Testing HIIT (AerT)+ Mobility HIIT protocol: Same HIIT protocol as above. Mobility Training: 6 mobility exercises (hamstrings stretch, gluteus medius stretch, lower back rotation mobilisation, back extension stretch, hip flexor stretch, and mid-back extension mobilisation): Stretches were held on each side 2 × 30 s, and mobilisations were performed 2 × 10 rep. HIIT: Interval: 110 rpm at 100% VO2 max workload Rest: 75 rpm at 50% VO2 max workload Mobility: N/R Pain: N/R 1RM, one-repetition maximum; ACSM, American College of Sports Medicine; AerT, aerobic training; BalanceT, balance training; bpm, beats per min; HIIT, high-intensity interval training; HR, heart rate; HRmax, maximal heart rate; HRR, heart rate reserve; LA, lactate level; MICT, moderate-intensity continuous training; MVC, maximal voluntary contraction; N/A, not applicable; N/R, not reported; RPE, rating of perceived exertion; rpm, revolutions per minute; StrT, strength training; VO2 max, maximal oxygen uptake; WRmax, highest work rate. Diagnostics 2022, 12, 2532 16 of 31 3.3. Methodological Quality Results We evaluated the studies’ quality with the Cochrane assessment tool. Most of the studies had a low risk of selective reporting bias. The domain with the highest percentage of studies with a high risk of bias was the blinding of participants and personnel (performance bias). Figure 2 shows the risk of bias summary and risk of bias graph. The inter-rater reliability of the methodological quality assessment was high (κ = 0.787). All of the studies had an excellent or good methodological quality, except the one by Bressel et al. [45] Due to the nature of the interventions, none of the studies performed blinding of the patients or evaluators. Table 3 lists the PEDro scores for each study. The inter-rater reliability of the methodological quality assessment between assessors was high (κ = 0.815). Diagnostics 2022, 12, x FOR PEER REVIEW 17 of 32 3.3. Methodological Quality Results We evaluated the studies’ quality with the Cochrane assessment tool. Most of the studies had a low risk of selective reporting bias. The domain with the highest percentage of studies with a high risk of bias was the blinding of participants and personnel (perfor- mance bias). Figure 2 shows the risk of bias summary and risk of bias graph. The inter- rater reliability of the methodological quality assessment was high (κ = 0.787). All of the studies had an excellent or good methodological quality, except the one by Bressel et al. [45] Due to the nature of the interventions, none of the studies performed blinding of the patients or evaluators. Table 3 lists the PEDro scores for each study. The inter-rater relia- bility of the methodological quality assessment between assessors was high (κ = 0.815). Figure 2. Risk of bias summary. Review authors’ judgements about each risk of bias item for each included study (Risk of Bias scale) and risk of bias graph. Review authors’ judgements about each risk of bias item presented as percentages across all included studies (Risk of Bias scale). Table 3. Assessment of the studies’ quality based on the PEDro Scale. Items 1 2 3 4 5 6 7 8 9 10 11 Total Atan et al., 2020 [15] 1 1 1 1 0 0 1 1 1 1 1 8 Berg et al., 2020 [50] 1 1 1 0 0 0 0 1 1 1 1 6 Bressel et al., 2014 [44] 1 0 0 1 0 0 0 1 1 1 1 5 Flehr et al., 2019 [52] 1 1 1 1 0 0 1 1 1 1 1 8 Hanssen et al., 2018 [49] 1 1 1 1 0 0 1 1 1 1 1 8 Keogh et al., 2018 [45] 1 1 1 1 0 0 0 1 1 1 1 7 Sandstad et al., 2015 [51] 1 1 1 1 0 0 0 1 1 1 1 7 Figure 2. Risk of bias summary. Review authors’ judgements about each risk of bias item for each included study (Risk of Bias scale) and risk of bias graph. Review authors’ judgements about each risk of bias item presented as percentages across all included studies (Risk of Bias scale). Table 3. Assessment of the studies’ quality based on the PEDro Scale. Items 1 2 3 4 5 6 7 8 9 10 11 Total Atan et al., 2020 [15] 1 1 1 1 0 0 1 1 1 1 1 8 Berg et al., 2020 [50] 1 1 1 0 0 0 0 1 1 1 1 6 Bressel et al., 2014 [44] 1 0 0 1 0 0 0 1 1 1 1 5 Flehr et al., 2019 [52] 1 1 1 1 0 0 1 1 1 1 1 8 Hanssen et al., 2018 [49] 1 1 1 1 0 0 1 1 1 1 1 8 Keogh et al., 2018 [45] 1 1 1 1 0 0 0 1 1 1 1 7 Sandstad et al., 2015 [51] 1 1 1 1 0 0 0 1 1 1 1 7 Sveas et al., 2014 [16] 1 1 1 1 0 0 1 1 1 1 1 8 Sveas et al., 2019 [49] 1 1 1 1 0 0 1 1 1 1 1 8 Thomsen et al., 2019 [53] 1 1 1 1 0 0 1 1 1 1 1 8 Verbrugghe et al., 2018 [47] 1 0 0 1 0 0 1 1 1 1 1 6 Verbrugghe et al., 2019 [17] 1 1 1 1 0 0 0 1 1 1 1 7 Verbrugghe et al., 2020 [48] 1 1 1 1 0 0 0 1 1 1 1 7 1, patient choice criteria are specified; 2, random assignment of patients to groups; 3, hidden assignment; 4, groups were similar at baseline; 5, all patients were blinded; 6, all therapists were blinded; 7, all evaluators were blinded; 8, measures of at least one of the key outcomes were obtained from more than 85% of baseline patients; 9, intention- to-treat analysis was performed; 10, results from statistical intergroup comparisons were reported for at least one key outcome; 11, the study provides point and variability measures for at least one key outcome. Diagnostics 2022, 12, 2532 17 of 31 3.4. Evidence Map Figure 3 presents the results of the evidence map for the included studies. Verbrugghe et al., 2020 [48] 1 1 1 1 0 0 0 1 1 1 1 7 1, patient choice criteria are specified; 2, random assignment of patients to groups; 3, hidden assign- ment; 4, groups were similar at baseline; 5, all patients were blinded; 6, all therapists were blinded; 7, all evaluators were blinded; 8, measures of at least one of the key outcomes were obtained from more than 85% of baseline patients; 9, intention-to-treat analysis was performed; 10, results from statistical intergroup comparisons were reported for at least one key outcome; 11, the study pro- vides point and variability measures for at least one key outcome. 3.4. Evidence Map Figure 3 presents the results of the evidence map for the included studies. Figure 3. A mapping of included studies based on effect size. Blue, Pain intensity; Violet, VO2 max; Green, Disability; Black, Quality of Life. Bubbles marked with white dots indicate statistically sig- nificant differences (p < 0.05). 3.5. Meta-Analysis Results 3.5.1. Pain Intensity The meta-analysis showed statistically significant differences for the HIIT interven- tion, with a moderate clinical effect in seven studies (SMD: −0.73; 95% CI −1.40–−0.06; p < 0.05) but with evidence of significant heterogeneity (Q = 32.57, p < 0.001, I2 = 82%). The shape of the funnel and DOI plot did not present asymmetry, and the LFK index showed minor asymmetry (LFK, −1.73) indicating a low risk of publication bias (Figures 4A and Figure 3. A mapping of included studies based on effect size. Blue, Pain intensity; Violet, VO2 max; Green, Disability; Black, Quality of Life. Bubbles marked with white dots indicate statistically significant differences (p < 0.05). 3.5. Meta-Analysis Results 3.5.1. Pain Intensity The meta-analysis showed statistically significant differences for the HIIT intervention, with a moderate clinical effect in seven studies (SMD: −0.73; 95% CI −1.40–−0.06; p < 0.05) but with evidence of significant heterogeneity (Q = 32.57, p < 0.001, I2 = 82%). The shape of the funnel and DOI plot did not present asymmetry, and the LFK index showed minor asymmetry (LFK, −1.73) indicating a low risk of publication bias (Figures 4A and A1). The certainty of the evidence was low, showing that HIIT likely decreases pain intensity, having been downgraded due to imprecision (sample size < 300) and inconsistency (I2 = 82%) (Table 4). Regarding the sub-analysis comparing HIIT against other therapeutic exercise models, the meta-analysis showed no significant differences for the HIIT intervention in 3 studies (SMD: −0.35; 95% CI −0.76–0.06, p ≥ 0.05) with no evidence of significant heterogeneity (Q = 1.37, p = 0.5, I2 = 0%). The shape of the funnel and DOI plot did not present asym- metry, and the LFK index showed no asymmetry (LFK, 0.67) indicating a very low risk of publication bias (Figures 4B and A2). Diagnostics 2022, 12, 2532 18 of 31 Disability (4) RCT seri- ous Not serious ous Serious 35 33 - (−0.92– 0.24) ate (+) (+) (+) Critical Quality of life (4) RCT Not seri- ous Serious Not seri- ous Serious 53 44 - 0.40 (−0.80– 1.60) Low (+) (+) Critical * CI, confidence interval; RCT, randomised controlled trial. Figure 4. Synthesis forest plot of pain intensity variable. The forest plot summarises the results of the included studies (sample size, standardised mean differences [SMDs], and weight). The small boxes with the squares represent the point estimate of the effect size and sample size. The lines on either side of the box represent a 95% confidence interval (CI). Figure 4. Synthesis forest plot of pain intensity variable. The forest plot summarises the results of the included studies (sample size, standardised mean differences [SMDs], and weight). The small boxes with the squares represent the point estimate of the effect size and sample size. The lines on either side of the box represent a 95% confidence interval (CI). Table 4. Summary of findings and quality of evidence (GRADE). Certainty Assessment No. of Participants Effect Certainty Importance Outcome (No. of Studies) Study Design Risk of Bias Inconsistency Indirectness Imprecision HIIT Control Relative (95% CI) Absolute (95% CI) Pain intensity (7) RCT Not serious Serious Not serious Serious 119 120 - −0.73 (1.40–−0.06) Low (+) (+) Critical VO2 max (6) RCT Not serious Not serious Not serious Serious 112 118 - 0.69 (0.42–0.97) Moderate (+) (+) (+) Critical Disability (4) RCT Not serious Not serious Not serious Serious 35 33 - −0.34 (−0.92–0.24) Moderate (+) (+) (+) Critical Quality of life (4) RCT Not serious Serious Not serious Serious 53 44 - 0.40 (−0.80–1.60) Low (+) (+) Critical CI, confidence interval; RCT, randomised controlled trial. 3.5.2. VO2 max The meta-analysis showed significant differences for the HIIT intervention, with a moderate clinical effect in six studies (SMD: 0.69; 95% CI 0.42–0.97, p < 0.05), with no evidence of significant heterogeneity (Q = 4.06, p = 0.54, I2 = 0%). The shape of the funnel and DOI plot did not present asymmetry, and the LFK index showed minor asymmetry (LFK, 1.33) indicating a low risk of publication bias (Figures 5A and A2). The certainty of the evidence was moderate, showing that HIIT probably increases VO2 max, having been downgraded due to imprecision (sample size < 300) (Table 4). Diagnostics 2022, 12, 2532 19 of 31 dence of significant heterogeneity (Q = 4.06, p = 0.54, I2 = 0%). The shape of the funnel a DOI plot did not present asymmetry, and the LFK index showed minor asymmetry (LF 1.33) indicating a low risk of publication bias (Figures 5A and A2). The certainty of evidence was moderate, showing that HIIT probably increases VO2 max, having be downgraded due to imprecision (sample size < 300) (Table 4). Figure 5. Synthesis forest plot of VO2 max variable. The forest plot summarises the results of included studies (sample size, standardised mean differences [SMDs], and weight). The small bo with the squares represent the point estimate of the effect size and sample size. The lines on eit side of the box represent a 95% confidence interval (CI). Regarding the subanalysis comparing HIIT against other therapeutic exercise mo els, the meta-analysis showed no statistically significant differences for the HIIT interv tion in three studies (SMD: 0.28; 95% CI −0.31–0.87, p ≥ 0.05), with no evidence of sign cant heterogeneity (Q = 4.16, p = 0.13, I2 = 52%). The shape of the funnel and DOI plot d not present asymmetry, and the LFK index showed no asymmetry (LFK, −0.31) indicati a very low risk of publication bias (Figures 5B and A2). Figure 5. Synthesis forest plot of VO2 max variable. The forest plot summarises the results of the included studies (sample size, standardised mean differences [SMDs], and weight). The small boxes with the squares represent the point estimate of the effect size and sample size. The lines on either side of the box represent a 95% confidence interval (CI). Regarding the subanalysis comparing HIIT against other therapeutic exercise models, the meta-analysis showed no statistically significant differences for the HIIT intervention in three studies (SMD: 0.28; 95% CI −0.31–0.87, p ≥ 0.05), with no evidence of significant heterogeneity (Q = 4.16, p = 0.13, I2 = 52%). The shape of the funnel and DOI plot did not present asymmetry, and the LFK index showed no asymmetry (LFK, −0.31) indicating a very low risk of publication bias (Figures 5B and A2). 3.5.3. Disability The meta-analysis showed no statistically significant differences for the HIIT inter- vention in three studies (SMD: −0.34; 95% CI −0.92–0.24, p ≥ 0.05), with no evidence of significant heterogeneity (Q = 4.55, p = 0.21, I2 = 34%). The shape of the funnel and DOI plot did not present asymmetry, and the LFK index showed minor asymmetry (LFK, −1.68) indicating a low risk of publication bias (Figures 6A and A3). The certainty of the evidence was moderate, showing that HIIT probably does not decrease disability, being downgraded due to imprecision (sample size <300) (Table 4). Diagnostics 2022, 12, 2532 20 of 31 significant heterogeneity (Q = 4.55, p = 0.21, I2 = 34%). The shape of the funnel and DOI plot did not present asymmetry, and the LFK index showed minor asymmetry (LFK, −1.68) indicating a low risk of publication bias (Figures 6A and A3). The certainty of the evidence was moderate, showing that HIIT probably does not decrease disability, being downgraded due to imprecision (sample size <300) (Table 4). Figure 6. Synthesis forest plot of disability and quality-of-life variables. The forest plot summarises the results of the included studies (sample size, standardised mean differences [SMDs], and weight). The small boxes with the squares represent the point estimate of the effect size and sample size. The lines on either side of the box represent a 95% confidence interval (CI). 3.5.4. Quality of Life The meta-analysis showed no significant differences for the HIIT intervention in 4 studies (SMD: 0.40; 95% CI −0.80–1.60, p ≥ 0.05), with evidence of significant heterogeneity (Q = 24.01, p < 0.001, I2 = 88%). The shape of the funnel and DOI plot did not present asym- metry, and the LFK index showed minor asymmetry (LFK, 1.43), indicating a low risk of publication bias (Figures 6B and A3). The certainty of the evidence was low, showing that HIIT likely does not increase QoL, being downgraded due to imprecision (sample size < 300) and inconsistency (I2 = 88%) (Table 4). 3.6. Meta-Regression Analysis In the meta-regression analysis, we explored the role of pain intensity scores in im- proving VO2 max function. The results showed that pain intensity was significantly and negatively correlated with VO2 max (β = −0.91; Z = −3.02; p = 0.003 and R2 = 82.99%) (Figure 7). Figure 6. Synthesis forest plot of disability and quality-of-life variables. The forest plot summarises the results of the included studies (sample size, standardised mean differences [SMDs], and weight). The small boxes with the squares represent the point estimate of the effect size and sample size. The lines on either side of the box represent a 95% confidence interval (CI). 3.5.4. Quality of Life The meta-analysis showed no significant differences for the HIIT intervention in 4 stud- ies (SMD: 0.40; 95% CI −0.80–1.60, p ≥ 0.05), with evidence of significant heterogeneity (Q = 24.01, p < 0.001, I2 = 88%). The shape of the funnel and DOI plot did not present asymmetry, and the LFK index showed minor asymmetry (LFK, 1.43), indicating a low risk of publication bias (Figures 6B and A3). The certainty of the evidence was low, showing that HIIT likely does not increase QoL, being downgraded due to imprecision (sample size < 300) and inconsistency (I2 = 88%) (Table 4). 3.6. Meta-Regression Analysis In the meta-regression analysis, we explored the role of pain intensity scores in im- proving VO2 max function. The results showed that pain intensity was significantly and negatively correlated with VO2 max (β = −0.91; Z = −3.02; p = 0.003 and R2 = 82.99%) (Figure 7). Diagnostics 2022, 12, 2532 21 of 31 Diagnostics 2022, 12, x FOR PEER REVIEW 22 of 32 Figure 7. Meta-regression of pain intensity and VO2 max scores. The meta-regression approach uses regression analysis to determine the influence of selected variables (the independent variables) on the effect size (the dependent variable). The large bubbles, together with the line, indicate the rela- tionship of our model, and the small bubbles indicate their position, the relationship in the map of the effect size on the decrease in pain, on the score in the variable of maximal oxygen consumption. 4. Discussion Our main goal was to analyse the effect of HIIT on the VO2 max, pain intensity, disa- bility, and QoL of patients with musculoskeletal disorders. Our results suggest that HIIT has a significant moderate effect size on VO2 max and pain intensity but does not seem to improve the disability and QoL of patients with musculoskeletal disorders. We also found that pain intensity was negatively associated with VO2 max. We found a moderate certainty of evidence of a moderate effect size of HIIT on VO2 max when compared with no intervention. Several authors also found that HIIT was supe- rior to usual care or no intervention in improving VO2 max among patients with cardiovas- cular disorders or cancer [18,19,55]. We did not find that HIIT was superior to another exercise intervention on VO2 max; however, the results across systematic reviews differ [19,56,57]. It has been previously reported that HIIT induces muscular adaptations, such as mitochondrial biogenesis and increased intramuscular capillarisation [58,59] vascular adaptations, such as increased blood cell volume [60], and cardiac adaptations, such as increased cardiac output and contractility [59,61]. All of these mechanisms have been shown to play a role in VO2 max [62]. We found that the patients’ pain intensity scores were negatively associated with VO2 max, which is an important predictor of all-cause mortality and cardiovascular disease [63,64]. It should be noted that patients with chronic pain and musculoskeletal disorders have shown an increased risk of cardiovascular and chronic disease and an increased risk of mortality due to cardiac disease [65,66]. An improvement in cardiorespiratory capacity has been shown to decrease the mortality risk by up to 16% [67,68]. HIIT appears to be an effective solution for improving patients’ cardiorespiratory capacity. We found a low certainty of evidence of a moderate effect size of HIIT on pain inten- sity compared with no intervention. Geneen et al. found that physical activity appears to Figure 7. Meta-regression of pain intensity and VO2 max scores. The meta-regression approach uses regression analysis to determine the influence of selected variables (the independent variables) on the effect size (the dependent variable). The large bubbles, together with the line, indicate the relationship of our model, and the small bubbles indicate their position, the relationship in the map of the effect size on the decrease in pain, on the score in the variable of maximal oxygen consumption. 4. Discussion Our main goal was to analyse the effect of HIIT on the VO2 max, pain intensity, dis- ability, and QoL of patients with musculoskeletal disorders. Our results suggest that HIIT has a significant moderate effect size on VO2 max and pain intensity but does not seem to improve the disability and QoL of patients with musculoskeletal disorders. We also found that pain intensity was negatively associated with VO2 max. We found a moderate certainty of evidence of a moderate effect size of HIIT on VO2 max when compared with no intervention. Several authors also found that HIIT was superior to usual care or no intervention in improving VO2 max among patients with cardiovascular disorders or cancer [18,19,55]. We did not find that HIIT was superior to another exercise intervention on VO2 max; however, the results across systematic reviews differ [19,56,57]. It has been previously reported that HIIT induces muscular adaptations, such as mitochon- drial biogenesis and increased intramuscular capillarisation [58,59] vascular adaptations, such as increased blood cell volume [60], and cardiac adaptations, such as increased cardiac output and contractility [59,61]. All of these mechanisms have been shown to play a role in VO2 max [62]. We found that the patients’ pain intensity scores were negatively associated with VO2 max, which is an important predictor of all-cause mortality and cardiovascular dis- ease [63,64]. It should be noted that patients with chronic pain and musculoskeletal disorders have shown an increased risk of cardiovascular and chronic disease and an increased risk of mortality due to cardiac disease [65,66]. An improvement in cardiorespi- ratory capacity has been shown to decrease the mortality risk by up to 16% [67,68]. HIIT appears to be an effective solution for improving patients’ cardiorespiratory capacity. We found a low certainty of evidence of a moderate effect size of HIIT on pain intensity compared with no intervention. Geneen et al. found that physical activity appears to induce exercise-induced hypoalgesia in patients with chronic pain; however, the results Diagnostics 2022, 12, 2532 22 of 31 were inconsistent across the various exercise modalities [69]. When compared with another exercise intervention, HIIT did not show a greater effect. It has been shown that exercise- induced hypoalgesia acts through the activation of nociceptive inhibitory pathways that release endogenous opioids and endocannabinoids [70]; however, populations with chronic pain often have exercise-induced hypoalgesia dysfunction [70,71]. Nonetheless, we found that HIIT appeared to be an effective modality for decreasing pain intensity. Patients with musculoskeletal disorders often present central sensitisation, a facilitation of the nociceptive signal in the central nervous system [72]. Quantitative sensory testing is employed to evaluate central nervous system nociceptive modulation [72]. HIIT has shown an intensity- dependent [12,13] positive effect on pain tolerance [13] and pain thresholds [12,73]. In certain conditions, the presence of an inflammatory state can increase nociceptor activity and has been associated with pain intensity [71,74–76]. After performing HIIT, a number of authors have found a decrease in inflammatory markers [77–79], such as C-reactive protein, tumour necrosis factor-alpha and interleukin-6 (IL-6), and a release of anti-inflammatory cytokines, such as IL-10 [79]. In contrast, other authors have found that HIIT induced an acute increase in IL-6 levels [80,81]; however, Pedersen proposed that this acute liberation will then induce an anti-inflammatory response [82]. Shanaki et al. observed a decrease in pro-inflammatory M1-macrophage markers and an increase in anti-inflammatory M2- macrophage markers in mice after HIIT [83]. However, not all musculoskeletal conditions show reduced pain intensity in parallel with a decrease in pro-nociceptive or inflammatory serum markers [76,84], and not all musculoskeletal conditions progress with an increased inflammatory state [76]. We found a low level of evidence of no significant effect of HIIT on QoL compared with no intervention or usual care. Mugele et al. systematically reviewed the effect of HIIT on QoL, compared with usual care, and found unclear results [19]. QoL appears to be more closely related to interpretation and catastrophising than pain intensity [85], which might explain why we observed a decrease in pain intensity with no improvement in QoL. Monticone et al. found that a multidisciplinary treatment involving cognitive-behavioural therapy and exercise results in a significant improvement in QoL, while exercise alone resulted in little change [86]. We also found moderate certainty evidence of no significant effect of HIIT on disability compared with no intervention or usual care. Kamper et al. found that a treatment involving a physical and a psychological or social component had a greater effect on disability than physical therapy alone for patients with chronic low back pain. HIIT alone might be insufficient for improving disability or QoL in musculoskeletal disorders [87]. Time constraints and pain are two of the main barriers to physical activity for patients with musculoskeletal disorders [88–90]. Despite similar effects on VO2 max and pain inten- sity with other exercise types, HIIT requires less training volume to achieve similar effects in the included studies that provide the control group’s training duration [15,50]. Wewege et al. found that the most common adverse effects in patients with cardiovascular disease were musculoskeletal complaints; however, we observed that HIIT presented similar or almost no additional major or minor adverse events or pain flare-ups than no intervention or other exercise modalities [91]. Major cardiac adverse events during HIIT appear at a rate of 1 per 11,333 HIIT h in patients with cardiovascular disease [91] but with no significant difference in the overall adverse events rate between HIIT and MICT [91]. As recommended by Weston et al. if health professionals want to implement HIIT, they should evaluate patients on a case-by-case basis depending on their cardiac history [20]. Heisz et al. found that participants rated HIIT more enjoyable than MICT and that enjoyment increased with repeated HIIT when it remained constant with repeated MICT [92]. Health professionals should include HIIT in the management of musculoskeletal disorders, given that HIIT is a time-efficient, enjoyable, effective, and safe form of exercise. Finally, it is relevant to stress that it is important to prescribe exercise specifically for each patient and for each clinical condition, although in this work it has been grouped by variables, rather than by populations. Diagnostics 2022, 12, 2532 23 of 31 Limitations We found low-to-moderate quality evidence for our results. Further studies are needed on the effects of HIIT on musculoskeletal disorders to confirm our results. The sample sizes of the included studies were often very small. Future studies should include larger sample sizes to improve the quality of the evidence. Due to the lack of sufficient data and the heterogeneity among the interventions (e.g., frequency, intervention duration), we could not establish the specific effect on each musculoskeletal disorder and the optimal HIIT parameters. Due to the small number of trials, we pooled the aerobic and anaerobic HIIT training studies; future systematic reviews should evaluate them separately. Only a few studies compared the effect of HIIT against high-intensity continuous training or other types of exercise; future studies should include this type of high-intensity training. As recommended by the American Thoracic Society/American College of Chest Physicians Statement on Cardiopulmonary Exercise Testing, we included VO2 peak and VO2 max and used them interchangeably [93]. Quantitative sensory testing (e.g., pain pressure or thermal threshold, conditioned pain modulation, and temporal summation) is essential in pain research; future studies evaluating the effects of HIIT on musculoskeletal disorders should include these variables. In addition, no further meta-regression analysis could be performed due to the small number of articles sharing the outcomes of interest. Lastly, it is important to stress that there were 3 studies where HIIT was embedded in other exercise interventions such as balance exercise and continuous exercise. This is a clear limitation that should be considered when extrapolating the results [16,45,47]. 5. Conclusions There is low to moderate quality evidence that the HIIT intervention for patients with musculoskeletal disorders can improve pain intensity and VO2 max but not disability and QoL. The results of the subanalyses showed that HIIT was not superior to other exercise models in improving pain intensity and VO2 max. Clinically, this tells us that we can implement high-intensity interval exercise models if our goal is to improve pain intensity or increase cardiorespiratory fitness through maximal oxygen consumption. However, it is important to keep in mind two aspects: changes in pain intensity may not be accompanied by improvements in the subjective perception of quality of life or disability, at least, based on the data we currently have, and second, that this exercise model was not superior to other exercise models with respect to eliciting these clinical changes. This should be considered clinically. Low sample sizes and lack of prescription parameters emphasise the need for further research on HIIT in musculoskeletal disorders for its implementation in a clinical context. Author Contributions: Conceptualization, F.C.-M.; methodology, F.C.-M., N.S.-R., C.V.-R. and R.L.T.; software, F.C.-M. and R.L.T.; validation, N.S.-R., L.S.-M., P.A.-Q. and J.F.-C.; formal analysis, F.C.-M. and R.L.T.; investigation, all authors; resources, J.F.-C.; data curation, F.C.-M., C.V.-R. and R.L.T.; writing—original draft preparation, all authors; writing—review and editing, all authors; visualiza- tion, all authors; supervision, F.C.-M.; project administration, F.C.-M. and N.S.-R.; funding acquisition, not applicable. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: Not applicable. Informed Consent Statement: Not applicable. Data Availability Statement: Not applicable. Acknowledgments: The authors would like to thank the CSEU La Salle for its services in editing this manuscript. Conflicts of Interest: The authors declare no conflict of interest. Diagnostics 2022, 12, 2532 24 of 31 Appendix A Diagnostics 2022, 12, x FOR PEER REVIEW 25 of 32 Acknowledgments: The authors would like to thank the CSEU La Salle for its services in editing this manuscript. Conflicts of Interest: The authors declare no conflicts of interest. Appendix A (a) HIIT vs. control Figure A1. Cont. Diagnostics 2022, 12, 2532 25 of 31 Diagnostics 2022, 12, x FOR PEER REVIEW 26 of 32 (b) HIIT vs. Other therapeutic exercise models Figure A1. Synthesis funnel and Doi plot (LFK index) for pain intensity to assess the presence of publication bias. Appendix B Figure A1. Synthesis funnel and Doi plot (LFK index) for pain intensity to assess the presence of publication bias. Appendix B Diagnostics 2022, 12, x FOR PEER REVIEW 26 of 32 (b) HIIT vs. Other therapeutic exercise models Figure A1. Synthesis funnel and Doi plot (LFK index) for pain intensity to assess the presence of publication bias. Appendix B Figure A2. Cont. Diagnostics 2022, 12, 2532 26 of 31 Diagnostics 2022, 12, x FOR PEER REVIEW 27 of 32 (a) HIIT vs. control (b) HIIT vs. other therapeutic exercise models Figure A2. Synthesis funnel and Doi plot (LFK index) for VO2 max to assess the presence of publication bias. Figure A2. Synthesis funnel and Doi plot (LFK index) for VO2 max to assess the presence of publication bias. Diagnostics 2022, 12, 2532 27 of 31 Appendix C Diagnostics 2022, 12, x FOR PEER REVIEW 28 of 32 Appendix C (a) HIIT vs. control (disability) Figure A3. Cont. Diagnostics 2022, 12, 2532 28 of 31 Diagnostics 2022, 12, x FOR PEER REVIEW 29 of 32 (b) HIIT vs. control (quality of life) Figure A3. Synthesis Funnel and Doi plot (LFK index) for disability and quality-of-life variables to assess the presence of publication bias. References 1. Picavet, H.S.J.; Schouten, J.S.A.G. Musculoskeletal pain in the Netherlands: Prevalences, consequences and risk groups, the DMC3-study. Pain 2003, 102, 167–178. 2. Silva, A.G.; Alvarelhão, J.; Queirós, A.; Rocha, N.P. Pain intensity is associated with self-reported disability for several domains of life in a sample of patients with musculoskeletal pain aged 50 or more. Disabil. Health J. 2013, 6, 369–376. 3. Nielson, W.R.; Weir, R. Biopsychosocial Approaches to the Treatment of Chronic Pain. Clin. J. Pain 2001, 17, S114–S127. 4. Wright, A.; Sluka, K.A. Nonpharmacological Treatments for Musculoskeletal Pain. Clin. J. Pain 2001, 17, 33–46. 5. Babatunde, O.O.; Jordan, J.L.; Van Der Windt, D.A.; Hill, J.C.; Foster, N.E.; Protheroe, J. Effective treatment options for musculoskeletal pain in primary care: A systematic overview of current evidence. PLoS ONE 2017, 12. 6. Dietrich, A. Endocannabinoids and exercise. Br. J. Sports Med. 2004, 38, 536–541. 7. Goldfarb, A.H.; Jamurtas, A.Z. Beta-Endorphin Response to Exercise—An update. Sports Med. 1997, 24, 8–16. 8. Naugle, K.M.; Fillingim, R.B.; Riley, J.L., III. A meta-analytic review of the hypoalgesic effects of exercise. J. Pain 2012, 13, 1139– 1150. 9. Booth, J.; Moseley, G.L.; Schiltenwolf, M.; Cashin, A.; Davies, M.; Hübscher, M. Exercise for chronic musculoskeletal pain: A biopsychosocial approach. Musculoskelet. Care 2017, 15, 413–421. 10. Smith, B.E.; Hendrick, P.; Bateman, M.; Holden, S.; Littlewood, C.; Smith, T.O.; Logan, P. Musculoskeletal pain and exercise- challenging existing paradigms and introducing new. Br. J. Sports Med. 2019, 53, 907–912. 11. La Touche, R.; Fernández Pérez, J.J.; Proy Acosta, A.; González Campodónico, L.; Martínez García, S.; Adraos Juárez, D.; Serrano García, B.; Angulo-Díaz-Parreño, S.; Cuenca-Martínez, F.; Suso-Martí, L.; et al. Is aerobic exercise helpful in patients with migraine? A systematic review and meta-analysis. Scand. J. Med. Sci. Sports 2020, 30, 965–982. 12. Hakansson, S.; Jones, M.D.; Ristov, M.; Marcos, L.; Clark, T.; Ram, A.; Morey, R.; Franklin, A.; McCarthy, C.; Carli, L.D.; et al. Intensity-dependent effects of aerobic training on pressure pain threshold in overweight men: A randomized trial. Eur. J. Pain 2018, 22, 1813–1823. 13. O’Leary, T.J.; Collett, J.; Howells, K.; Morris, M.G. High but not moderate-intensity endurance training increases pain tolerance: A randomised trial. Eur. J. Appl. Physiol. 2017, 117, 2201–2210. 14. Andreato, L.V. High-Intensity Interval Training: Methodological Considerations for Interpreting Results and Conducting Research. Trends Endocrinol. Metab. 2020, 31, 812–817. 15. Atan, T.; Karavelioğlu, Y. Effectiveness of High-Intensity Interval Training vs Moderate-Intensity Continuous Training in Patients with Fibromyalgia: A Pilot Randomized Controlled Trial. Arch. Phys. Med. Rehabil. 2020, 101, 1865–1876. 16. Sveaas, S.H.; Bilberg, A.; Berg, I.J.; Provan, S.A.; Rollefstad, S.; Semb, A.G.; Hagen, K.B.; Johansen, M.W.; Pedersen, E.; Dagfinrud, H. High intensity exercise for 3 months reduces disease activity in axial spondyloarthritis (axSpA): A multicentre randomised trial of 100 patients. Br. J. Sports Med. 2019, 54, 292–297. 17. Verbrugghe, J.; Agten, A.; Stevens, S.; Hansen, D.; Demoulin, C.; Eijnde, B.O.; Vandenabeele, F.; Timmermans, A. Exercise Intensity Matters in Chronic Nonspecific Low Back Pain Rehabilitation. Med. Sci. Sports Exerc. 2019, 51, 2434–2442. 18. Batacan, R.B.; Duncan, M.J.; Dalbo, V.J.; Tucker, P.S.; Fenning, A.S. Effects of high-intensity interval training on cardiometabolic health: A systematic review and meta-analysis of intervention studies. Br. J. Sports Med. 2017, 51, 494–503. Figure A3. Synthesis Funnel and Doi plot (LFK index) for disability and quality-of-life variables to assess the presence of publication bias. References 1. Picavet, H.S.J.; Schouten, J.S.A.G. Musculoskeletal pain in the Netherlands: Prevalences, consequences and risk groups, the DMC3-study. Pain 2003, 102, 167–178. [CrossRef] 2. Silva, A.G.; Alvarelhão, J.; Queirós, A.; Rocha, N.P. Pain intensity is associated with self-reported disability for several domains of life in a sample of patients with musculoskeletal pain aged 50 or more. Disabil. Health J. 2013, 6, 369–376. [CrossRef] [PubMed] 3. Nielson, W.R.; Weir, R. Biopsychosocial Approaches to the Treatment of Chronic Pain. Clin. J. Pain 2001, 17, S114–S127. [CrossRef] [PubMed] 4. Wright, A.; Sluka, K.A. Nonpharmacological Treatments for Musculoskeletal Pain. Clin. J. Pain 2001, 17, 33–46. [CrossRef] 5. Babatunde, O.O.; Jordan, J.L.; Van Der Windt, D.A.; Hill, J.C.; Foster, N.E.; Protheroe, J. Effective treatment options for muscu- loskeletal pain in primary care: A systematic overview of current evidence. PLoS ONE 2017, 12. [CrossRef] 6. Dietrich, A. Endocannabinoids and exercise. Br. J. Sports Med. 2004, 38, 536–541. [CrossRef] 7. Goldfarb, A.H.; Jamurtas, A.Z. Beta-Endorphin Response to Exercise—An update. Sports Med. 1997, 24, 8–16. [CrossRef] 8. Naugle, K.M.; Fillingim, R.B.; Riley, J.L., III. A meta-analytic review of the hypoalgesic effects of exercise. J. Pain 2012, 13, 1139–1150. [CrossRef] 9. Booth, J.; Moseley, G.L.; Schiltenwolf, M.; Cashin, A.; Davies, M.; Hübscher, M. Exercise for chronic musculoskeletal pain: A biopsychosocial approach. Musculoskelet. Care 2017, 15, 413–421. [CrossRef] 10. Smith, B.E.; Hendrick, P.; Bateman, M.; Holden, S.; Littlewood, C.; Smith, T.O.; Logan, P. Musculoskeletal pain and exercise- challenging existing paradigms and introducing new. Br. J. Sports Med. 2019, 53, 907–912. [CrossRef] 11. La Touche, R.; Fernández Pérez, J.J.; Proy Acosta, A.; González Campodónico, L.; Martínez García, S.; Adraos Juárez, D.; Serrano García, B.; Angulo-Díaz-Parreño, S.; Cuenca-Martínez, F.; Suso-Martí, L.; et al. Is aerobic exercise helpful in patients with migraine? A systematic review and meta-analysis. Scand. J. Med. Sci. Sports 2020, 30, 965–982. [CrossRef] [PubMed] 12. Hakansson, S.; Jones, M.D.; Ristov, M.; Marcos, L.; Clark, T.; Ram, A.; Morey, R.; Franklin, A.; McCarthy, C.; Carli, L.D.; et al. Intensity-dependent effects of aerobic training on pressure pain threshold in overweight men: A randomized trial. Eur. J. Pain 2018, 22, 1813–1823. [CrossRef] [PubMed] 13. O’Leary, T.J.; Collett, J.; Howells, K.; Morris, M.G. High but not moderate-intensity endurance training increases pain tolerance: A randomised trial. Eur. J. Appl. Physiol. 2017, 117, 2201–2210. [CrossRef] [PubMed] 14. Andreato, L.V. High-Intensity Interval Training: Methodological Considerations for Interpreting Results and Conducting Research. Trends Endocrinol. Metab. 2020, 31, 812–817. [CrossRef] [PubMed] 15. Atan, T.; Karavelio˘glu, Y. Effectiveness of High-Intensity Interval Training vs Moderate-Intensity Continuous Training in Patients with Fibromyalgia: A Pilot Randomized Controlled Trial. Arch. Phys. Med. Rehabil. 2020, 101, 1865–1876. [CrossRef] 16. Sveaas, S.H.; Bilberg, A.; Berg, I.J.; Provan, S.A.; Rollefstad, S.; Semb, A.G.; Hagen, K.B.; Johansen, M.W.; Pedersen, E.; Dagfinrud, H. High intensity exercise for 3 months reduces disease activity in axial spondyloarthritis (axSpA): A multicentre randomised trial of 100 patients. Br. J. Sports Med. 2019, 54, 292–297. [CrossRef] 17. Verbrugghe, J.; Agten, A.; Stevens, S.; Hansen, D.; Demoulin, C.; Eijnde, B.O.; Vandenabeele, F.; Timmermans, A. Exercise Intensity Matters in Chronic Nonspecific Low Back Pain Rehabilitation. Med. Sci. Sports Exerc. 2019, 51, 2434–2442. [CrossRef] 18. Batacan, R.B.; Duncan, M.J.; Dalbo, V.J.; Tucker, P.S.; Fenning, A.S. Effects of high-intensity interval training on cardiometabolic health: A systematic review and meta-analysis of intervention studies. Br. J. Sports Med. 2017, 51, 494–503. [CrossRef] Diagnostics 2022, 12, 2532 29 of 31 19. Mugele, H.; Freitag, N.; Wilhelmi, J.; Yang, Y.; Cheng, S.; Bloch, W.; Schumann, M. High-intensity interval training in the therapy and aftercare of cancer patients: A systematic review with meta-analysis. J. Cancer Surviv. 2019, 13, 205–223. [CrossRef] 20. Weston, K.S.; Wisløff, U.; Coombes, J.S. High-intensity interval training in patients with lifestyle-induced cardiometabolic disease: A systematic review and meta-analysis. Br. J. Sports Med. 2014, 48, 1227–1234. [CrossRef] 21. Doury-Panchout, F.; Métivier, J.C.; Fouquet, B. VO2max in patients with chronic pain: The effect of a 4-week rehabilitation program. Ann. Phys. Rehabil. Med. 2014, 57, 1–10. [CrossRef] [PubMed] 22. Moher, D.; Liberati, A.; Tetzlaff, J.; Altman, D.G. Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. Int. J. Surg. 2009, 8, 6. 23. Stone, P.W. Popping the (PICO) question in research and evidence-based practice. Appl. Nurs. Res. 2002, 15, 197–198. [CrossRef] [PubMed] 24. Terwee, C.B.; Jansma, E.P.; Riphagen, I.I.; de Vet, H.C.W. Development of a methodological PubMed search filter for finding studies on measurement properties of measurement instruments. Qual. Life Res. 2009, 18, 1115–1123. [CrossRef] 25. Shariff, S.Z.; Bejaimal, S.A.; Sontrop, J.M.; Iansavichus, A.V.; Haynes, R.B.; Weir, M.A.; Garg, A.X. Retrieving clinical evidence: A comparison of PubMed and Google Scholar for quick clinical searches. J. Med. Internet Res. 2013, 15, e164. [CrossRef] 26. Haddaway, N.R.; Collins, A.M.; Coughlin, D.; Kirk, S. The Role of Google Scholar in Evidence Reviews and Its Applicability to Grey Literature Searching. PLoS ONE 2015, 10, e0138237. [CrossRef] 27. Moher, D.; Pham, B.; Jones, A.; Cook, D.J.; Jadad, A.R.; Moher, M.; Tugwell, P.; Klassen, T.P. Does quality of reports of randomised trials affect estimates of intervention efficacy reported in meta-analyses? Lancet 1998, 352, 609–613. [CrossRef] 28. Kwon, Y.; Lemieux, M.; McTavish, J.; Wathen, N. Identifying and removing duplicate records from systematic review searches. J. Med. Libr. Assoc. 2015, 103, 184–188. [CrossRef] 29. Furlan, A.D.; Pennick, V.; Bombardier, C.; van Tulder, M. 2009 updated method guidelines for systematic reviews in the Cochrane Back Review Group. Spine 2009, 34, 1929–1941. [CrossRef] 30. Higgins, J.; Green, S. Cochrane Handbook for Systematic Reviews of Interventions; Version 5.1.0[M]; Wiley-Blackwell: Hoboken, NJ, USA, 2008. 31. de Morton, N.A. The PEDro scale is a valid measure of the methodological quality of clinical trials: A demographic study. Aust. J. Physiother. 2009, 55, 129–133. [CrossRef] 32. Hariohm, K.; Prakash, V.; Saravankumar, J. Quantity and quality of randomized controlled trials published by Indian physiother- apists. Perspect. Clin. Res. 2015, 6, 91. [CrossRef] [PubMed] 33. Landis, J.R.; Koch, G.G. An Application of Hierarchical Kappa-type Statistics in the Assessment of Majority Agreement among Multiple Observers. Biometrics 1977, 33, 363. [CrossRef] [PubMed] 34. Guyatt, G.H.; Oxman, A.D.; Vist, G.E.; Kunz, R.; Falck-Ytter, Y.; Alonso-Coello, P.; Schünemann, H.J. GRADE Working Group GRADE: An emerging consensus on rating quality of evidence and strength of recommendations. BMJ 2008, 336, 924–926. [CrossRef] [PubMed] 35. Andrews, J.; Guyatt, G.; Oxman, A.D.; Alderson, P.; Dahm, P.; Falck-Ytter, Y.; Nasser, M.; Meerpohl, J.; Post, P.N.; Kunz, R.; et al. GRADE guidelines: 14. Going from evidence to recommendations: The significance and presentation of recommendations. J. Clin. Epidemiol. 2013, 66, 719–725. [CrossRef] 36. Balshem, H.; Helfand, M.; Schünemann, H.J.; Oxman, A.D.; Kunz, R.; Brozek, J.; Vist, G.E.; Falck-Ytter, Y.; Meerpohl, J.; Norris, S.; et al. GRADE guidelines: 3. Rating the quality of evidence. J. Clin. Epidemiol. 2011, 64, 401–406. [CrossRef] 37. Sanabria, A.J.; Rigau, D.; Rotaeche, R.; Selva, A.; Marzo-Castillejo, M.; Alonso-Coello, P. GRADE: Methodology for formulating and grading recommendations in clinical practice. Aten. Primaria 2015, 47, 48–55. [CrossRef] [PubMed] 38. Barendregt, J.J.; Doi, S.A. MetaXL User Guide Version 5.3; EpiGear International Pty Ltd.: Brisbane, Australia, 2016. 39. Hedges, L. Estimation of effect size from a series of independent experiments. Psychol. Bull. 1982, 92, 490–499. [CrossRef] 40. Hopkins, W.G.; Marshall, S.W.; Batterham, A.M.; Hanin, J. Progressive statistics for studies in sports medicine and exercise science. Med. Sci. Sports Exerc. 2009, 41, 3–13. [CrossRef] 41. Huedo-Medina, T.B.; Sánchez-Meca, J.; Marín-Martínez, F.; Botella, J. Assessing heterogeneity in meta-analysis: Q statistic or I2 index? Psychol. Methods 2006, 11, 193–206. [CrossRef] 42. Doi, S.A. Rendering the Doi plot properly in meta-analysis. Int. J. Evid. Based. Healthc. 2018, 16, 242–243. [CrossRef] 43. Furuya-Kanamori, L.; Barendregt, J.J.; Doi, S.A.R. A new improved graphical and quantitative method for detecting bias in meta-analysis. Int. J. Evid. Based. Healthc. 2018, 16, 195–203. [CrossRef] [PubMed] 44. Suurmond, R.; van Rhee, H.; Hak, T. Introduction, comparison, and validation of Meta-Essentials: A free and simple tool for meta-analysis. Res. Synth. Methods 2017, 8, 537–553. [CrossRef] [PubMed] 45. Bressel, E.; Wing, J.E.; Miller, A.I.; Dolny, D.G. High-intensity interval training on an aquatic treadmill in adults with osteoarthritis: Effect on pain, balance, function, and mobility. J. Strength Cond. Res. 2014, 28, 2088–2096. [CrossRef] [PubMed] 46. Keogh, J.W.; Grigg, J.; Vertullo, C.J. Is high-intensity interval cycling feasible and more beneficial than continuous cycling for knee osteoarthritic patients? Results of a randomised control feasibility trial. PeerJ 2018, 6, e4738. [CrossRef] [PubMed] 47. Sveaas, S.H.; Berg, I.J.; Provan, S.A.; Semb, A.G.; Hagen, K.B.; Vøllestad, N.; Fongen, C.; Olsen, I.C.; Michelsen, A.; Ueland, T.; et al. Efficacy of high intensity exercise on disease activity and cardiovascular risk in active axial spondyloarthritis: A randomized controlled pilot study. PLoS ONE 2014, 9, e108688. [CrossRef] [PubMed] Diagnostics 2022, 12, 2532 30 of 31 48. Verbrugghe, J.; Agten, A.; Eijnde, B.O.; Olivieri, E.; Huybrechts, X.; Seelen, H.; Vandenabeele, F.; Timmermans, A. Feasibility of high intensity training in nonspecific chronic low back pain: A clinical trial. J. Back Musculoskelet. Rehabil. 2018, 31, 657–666. [CrossRef] [PubMed] 49. Verbrugghe, J.; Agten, A.; Stevens, S.; Hansen, D.; Demoulin, C.; Eijnde, B.O.; Vandenabeele, F.; Timmermans, A. High Intensity Training to Treat Chronic Nonspecific Low Back Pain: Effectiveness of Various Exercise Modes. J. Clin. Med. 2020, 9, 2401. [CrossRef] 50. Hanssen, H.; Minghetti, A.; Magon, S.; Rossmeissl, A.; Rasenack, M.; Papadopoulou, A.; Klenk, C.; Faude, O.; Zahner, L.; Sprenger, T.; et al. Effects of different endurance exercise modalities on migraine days and cerebrovascular health in episodic migraineurs: A randomized controlled trial. Scand. J. Med. Sci. Sports 2018, 28, 1103–1112. [CrossRef] 51. Berg, O.K.; Paulsberg, F.; Brabant, C.; Arabsolghar, K.; Ronglan, S.; BjØrnsen, N.; TØrhaug, T.; Granviken, F.; Gismervik, S.; Hoff, J. High-Intensity Shoulder Abduction Exercise in Subacromial Pain Syndrome. Med. Sci. Sports Exerc. 2020, 53, 1–9. [CrossRef] 52. Sandstad, J.; Stensvold, D.; Hoff, M.; Nes, B.M.; Arbo, I.; Bye, A. The effects of high intensity interval training in women with rheumatic disease: A pilot study. Eur. J. Appl. Physiol. 2015, 115, 2081–2089. [CrossRef] 53. Flehr, A.; Barton, C.; Coles, J.; Gibson, S.J.; Lambert, G.W.; Lambert, E.A.; Dhar, A.K.; Dixon, J.B. #MindinBody—Feasibility of vigorous exercise (Bikram yoga versus high intensity interval training) to improve persistent pain in women with a history of trauma: A pilot randomized control trial. BMC Complement. Altern. Med. 2019, 19, 234. 54. Thomsen, R.S.; Nilsen, T.I.L.; Haugeberg, G.; Bye, A.; Kavanaugh, A.; Hoff, M. Impact of High-Intensity Interval Training on Disease Activity and Disease in Patients with Psoriatic Arthritis: A Randomized Controlled Trial. Arthritis Care Res. 2019, 71, 530–537. [CrossRef] [PubMed] 55. Palma, S.; Hasenoehrl, T.; Jordakieva, G.; Ramazanova, D.; Crevenna, R. High-intensity interval training in the prehabilitation of cancer patients—A systematic review and meta-analysis. Support. Care Cancer 2021, 29, 1781–1794. [CrossRef] 56. Milanovi´c, Z.; Sporiš, G.; Weston, M. Effectiveness of High-Intensity Interval Training (HIT) and Continuous Endurance Training for VO2max Improvements: A Systematic Review and Meta-Analysis of Controlled Trials. Sports Med. 2015, 45, 1469–1481. [CrossRef] 57. Ramos, J.S.; Dalleck, L.C.; Tjonna, A.E.; Beetham, K.S.; Coombes, J.S. The impact of high-intensity interval training versus moderate-intensity continuous training on vascular function: A systematic review and meta-analysis. Sports Med. 2015, 45, 679–692. [CrossRef] [PubMed] 58. Gibala, M. Molecular responses to high-intensity interval exercise. Appl. Physiol. Nutr. Metab. 2009, 34, 428–432. [CrossRef] [PubMed] 59. MacInnis, M.J.; Gibala, M.J. Physiological adaptations to interval training and the role of exercise intensity. J. Physiol. 2017, 595, 2915–2930. [CrossRef] [PubMed] 60. Belviranli, M.; Okudan, N.; Kabak, B. The Effects of Acute High-Intensity Interval Training on Hematological Parameters in Sedentary Subjects. Med. Sci. 2017, 5, 15. [CrossRef] 61. Huang, Y.-C.; Tsai, H.-H.; Fu, T.-C.; Hsu, C.-C.; Wang, J.-S. High-Intensity Interval Training Improves Left Ventricular Contractile Function. Med. Sci. Sports Exerc. 2019, 51, 1420–1428. [CrossRef] 62. Lundby, C.; Montero, D.; Joyner, M. Biology of VO(2) max: Looking under the physiology lamp. Acta Physiol. 2017, 220, 218–228. [CrossRef] 63. Kaminsky, L.A.; Arena, R.; Ellingsen, Ø.; Harber, M.P.; Myers, J.; Ozemek, C.; Ross, R. Cardiorespiratory fitness and cardiovascular disease—The past, present, and future. Prog. Cardiovasc. Dis. 2019, 62, 86–93. [CrossRef] [PubMed] 64. Kodama, S.; Saito, K.; Tanaka, S.; Maki, M.; Yachi, Y.; Asumi, M.; Sugawara, A.; Totsuka, K.; Shimano, H.; Ohashi, Y.; et al. Cardiorespiratory fitness as a quantitative predictor of all-cause mortality and cardiovascular events in healthy men and women: A meta-analysis. JAMA 2009, 301, 2024–2035. [CrossRef] [PubMed] 65. Williams, A.; Kamper, S.J.; Wiggers, J.H.; O’Brien, K.M.; Lee, H.; Wolfenden, L.; Yoong, S.L.; Robson, E.; McAuley, J.H.; Hartvigsen, J.; et al. Musculoskeletal conditions may increase the risk of chronic disease: A systematic review and meta-analysis of cohort studies. BMC Med. 2018, 16, 167. [CrossRef] [PubMed] 66. Fayaz, A.; Ayis, S.; Panesar, S.S.; Langford, R.M.; Donaldson, L.J. Assessing the relationship between chronic pain and cardiovas- cular disease: A systematic review and meta-analysis. Scand. J. pain 2016, 13, 76–90. [CrossRef] 67. Imboden, M.T.; Harber, M.P.; Whaley, M.H.; Finch, W.H.; Bishop, D.L.; Fleenor, B.S.; Kaminsky, L.A. The Association between the Change in Directly Measured Cardiorespiratory Fitness across Time and Mortality Risk. Prog. Cardiovasc. Dis. 2019, 62, 157–162. [CrossRef] 68. Laukkanen, J.A.; Zaccardi, F.; Khan, H.; Kurl, S.; Jae, S.Y.; Rauramaa, R. Long-term Change in Cardiorespiratory Fitness and All-Cause Mortality: A Population-Based Follow-up Study. Mayo Clin. Proc. 2016, 91, 1183–1188. [CrossRef] 69. Geneen, L.J.; Moore, R.A.; Clarke, C.; Martin, D.; Colvin, L.A.; Smith, B.H. Physical activity and exercise for chronic pain in adults: An overview of Cochrane Reviews. In Cochrane Database of Systematic Reviews; Geneen, L.J., Ed.; John Wiley & Sons, Ltd: Chichester, UK, 2017; Volume 4, p. CD011279. 70. Rice, D.; Nijs, J.; Kosek, E.; Wideman, T.; Hasenbring, M.I.; Koltyn, K.; Graven-Nielsen, T.; Polli, A. Exercise-Induced Hypoalgesia in Pain-Free and Chronic Pain Populations: State of the Art and Future Directions. J. Pain 2019, 20, 1249–1266. [CrossRef] 71. Sluka, K.A.; Frey-Law, L.; Hoeger Bement, M. Exercise-induced pain and analgesia? Underlying mechanisms and clinical translation. Pain 2018, 159 (Suppl. S1), S91–S97. [CrossRef] Diagnostics 2022, 12, 2532 31 of 31 72. Nijs, J.; George, S.Z.; Clauw, D.J.; Fernández-de-las-Peñas, C.; Kosek, E.; Ickmans, K.; Fernández-Carnero, J.; Polli, A.; Kapreli, E.; Huysmans, E.; et al. Central sensitisation in chronic pain conditions: Latest discoveries and their potential for precision medicine. Lancet Rheumatol. 2021, 3, e383–e392. [CrossRef] 73. Mijwel, S.; Backman, M.; Bolam, K.A.; Olofsson, E.; Norrbom, J.; Bergh, J.; Sundberg, C.J.; Wengström, Y.; Rundqvist, H. Highly favorable physiological responses to concurrent resistance and high-intensity interval training during chemotherapy: The OptiTrain breast cancer trial. Breast Cancer Res. Treat. 2018, 169, 93–103. [CrossRef] 74. van den Berg, R.; Jongbloed, E.M.; de Schepper, E.I.T.; Bierma-Zeinstra, S.M.A.; Koes, B.W.; Luijsterburg, P.A.J. The association between pro-inflammatory biomarkers and nonspecific low back pain: A systematic review. Spine J. 2018, 18, 2140–2151. [CrossRef] [PubMed] 75. Farrell, S.F.; de Zoete, R.M.J.; Cabot, P.J.; Sterling, M. Systemic inflammatory markers in neck pain: A systematic review with meta-analysis. Eur. J. Pain 2020, 24, 1666–1686. [CrossRef] [PubMed] 76. DeVon, H.A.; Piano, M.R.; Rosenfeld, A.G.; Hoppensteadt, D.A. The Association of Pain with Protein Inflammatory Biomarkers: A Review of the Literature. Nurs. Res. 2014, 63, 51–62. [CrossRef] 77. Adams, S.C.; DeLorey, D.S.; Davenport, M.H.; Stickland, M.K.; Fairey, A.S.; North, S.; Szczotka, A.; Courneya, K.S. Effects of high-intensity aerobic interval training on cardiovascular disease risk in testicular cancer survivors: A phase 2 randomized controlled trial. Cancer 2017, 123, 4057–4065. [CrossRef] [PubMed] 78. Khalafi, M.; Symonds, M.E. The impact of high-intensity interval training on inflammatory markers in metabolic disorders: A meta-analysis. Scand. J. Med. Sci. Sports 2020, 30, 2020–2036. [CrossRef] [PubMed] 79. Steckling, F.M.; Farinha, J.B.; da Cunha Figueiredo, F.; Dos Santos, D.L.; Bresciani, G.; Kretzmann, N.A.; Stefanello, S.T.; Courtes, A.A.; de Oliveira Beck, M.; Sangoi Cardoso, M.; et al. High-intensity interval training improves inflammatory and adipokine profiles in postmenopausal women with metabolic syndrome. Arch. Physiol. Biochem. 2019, 125, 85–91. [CrossRef] 80. Nunes, P.R.P.; Martins, F.M.; Souza, A.P.; Carneiro, M.A.S.; Orsatti, C.L.; Michelin, M.A.; Murta, E.F.C.; de Oliveira, E.P.; Orsatti, F.L. Effect of high-intensity interval training on body composition and inflammatory markers in obese postmenopausal women: A randomized controlled trial. Menopause 2019, 26, 256–264. [CrossRef] 81. Zwetsloot, K.A.; John, C.S.; Lawrence, M.M.; Battista, R.A.; Shanely, R.A. High-intensity interval training induces a modest systemic inflammatory response in active, young men. J. Inflamm. Res. 2014, 7, 9–17. [CrossRef] 82. Pedersen, B.K. Anti-inflammatory effects of exercise: Role in diabetes and cardiovascular disease. Eur. J. Clin. Investig. 2017, 47, 600–611. [CrossRef] 83. Shanaki, M.; Khosravi, M.; Khoshdooni-Farahani, A.; Dadashi, A.; Heydari, M.F.; Delfan, M.; Jafary, H.; Gorgani-Firuzjaee, S. High-Intensity Interval Training Reversed High-Fat Diet-Induced M1-Macrophage Polarization in Rat Adipose Tissue via Inhibition of NOTCH Signaling. J. Inflamm. Res. 2020, 13, 165–174. [CrossRef] 84. Eslami, R.; Parnow, A.; Pairo, Z.; Nikolaidis, P.; Knechtle, B. The effects of two different intensities of aerobic training protocols on pain and serum neuro-biomarkers in women migraineurs: A randomized controlled trail. Eur. J. Appl. Physiol. 2021, 121, 609–620. [CrossRef] [PubMed] 85. Lamé, I.E.; Peters, M.L.; Vlaeyen, J.W.S.; Kleef, M.v.; Patijn, J. Quality of life in chronic pain is more associated with beliefs about pain, than with pain intensity. Eur. J. Pain 2005, 9, 15–24. [CrossRef] [PubMed] 86. Monticone, M.; Ferrante, S.; Rocca, B.; Baiardi, P.; Farra, F.D.; Foti, C. Effect of a long-lasting multidisciplinary program on disability and fear-avoidance behaviors in patients with chronic low back pain: Results of a randomized controlled trial. Clin. J. Pain 2013, 29, 929–938. [CrossRef] [PubMed] 87. Kamper, S.J.; Apeldoorn, A.T.; Chiarotto, A.; Smeets, R.J.E.M.; Ostelo, R.W.J.G.; Guzman, J.; van Tulder, M.W. Multidisciplinary biopsychosocial rehabilitation for chronic low back pain: Cochrane systematic review and meta-analysis. BMJ 2015, 350, h444. [CrossRef] 88. Veldhuijzen van Zanten, J.J.C.S.; Rouse, P.C.; Hale, E.D.; Ntoumanis, N.; Metsios, G.S.; Duda, J.L.; Kitas, G.D. Perceived Barriers, Facilitators and Benefits for Regular Physical Activity and Exercise in Patients with Rheumatoid Arthritis: A Review of the Literature. Sports Med. 2015, 45, 1401–1412. [CrossRef] 89. Boutevillain, L.; Dupeyron, A.; Rouch, C.; Richard, E.; Coudeyre, E. Facilitators and barriers to physical activity in people with chronic low back pain: A qualitative study. PLoS ONE 2017, 12, e0179826. [CrossRef] 90. McPhail, S.M.; Schippers, M.; Marshall, A.L.; Waite, M.; Kuipers, P. Perceived barriers and facilitators to increasing physical activity among people with musculoskeletal disorders: A qualitative investigation to inform intervention development. Clin. Interv. Aging 2014, 9, 2113–2122. [CrossRef] 91. Wewege, M.A.; Ahn, D.; Yu, J.; Liou, K.; Keech, A. High-Intensity Interval Training for Patients with Cardiovascular Disease-Is It Safe? A Systematic Review. J. Am. Heart Assoc. 2018, 7, e009305. [CrossRef] 92. Heisz, J.J.; Tejada, M.G.M.; Paolucci, E.M.; Muir, C. Enjoyment for High-Intensity Interval Exercise Increases during the First Six Weeks of Training: Implications for Promoting Exercise Adherence in Sedentary Adults. PLoS ONE 2016, 11, e0168534. [CrossRef] 93. American Thoracic Society; American College of Chest Physicians. ATS/ACCP Statement on cardiopulmonary exercise testing. Am. J. Respir. Crit. Care Med. 2003, 167, 211–277. [CrossRef]

Effects of High-Intensity Interval Training (HIIT) on Patients with Musculoskeletal Disorders: A Systematic Review and Meta-Analysis with a Meta-Regression and Mapping Report.

10-19-2022

Cuenca-Martínez, Ferran,Sempere-Rubio, Núria,Varangot-Reille, Clovis,Fernández-Carnero, Josué,Suso-Martí, Luis,Alba-Quesada, Patricio,Touche, Roy La

eng

PMC5147982

RESEARCH ARTICLE Inter-Individual Variability in the Adaptive Responses to Endurance and Sprint Interval Training: A Randomized Crossover Study Jacob T. Bonafiglia1, Mario P. Rotundo1, Jonathan P. Whittall1, Trisha D. Scribbans1, Ryan B. Graham2, Brendon J. Gurd1* 1 School of Kinesiology and Health Studies, Queen’s University, Kingston, Ontario, Canada, 2 School of Human Kinetics, University of Ottawa, Ottawa, Ontario, Canada * gurdb@queensu.ca Abstract The current study examined the adaptive response to both endurance (END) and sprint interval training (SIT) in a group of twenty-one recreationally active adults. All participants completed three weeks (four days/ week) of both END (30 minutes at ~65% VO2peak work rate (WR) and SIT (eight, 20-second intervals at ~170% VO2peak WR separated by 10 sec- onds of active rest) following a randomized crossover study design with a three-month washout period between training interventions. While a main effect of training was observed for VO2peak, lactate threshold, and submaximal heart rate (HR), considerable variability was observed in the individual responses to both END and SIT. No significant positive rela- tionships were observed between END and SIT for individual changes in any variable. Non- responses were determined using two times the typical error (TE) of measurement for VO2peak (0.107 L/min), lactate threshold (15.7 W), and submaximal HR (10.7bpm). Non- responders in VO2peak, lactate threshold, and submaximal HR were observed following both END and SIT, however, the individual patterns of response differed following END and SIT. Interestingly, all individuals responded in at least one variable when exposed to both END and SIT. These results suggest that the individual response to exercise training is highly variable following different training protocols and that the incidence of non-response to exercise training may be reduced by changing the training stimulus for non-responders to three weeks of END or SIT. Introduction Considerable heterogeneity exists in the individual response in peak oxygen uptake (VO2peak) following exercise training [1–3]. Specifically, VO2peak can increase [2,4], decrease [5], or remain unchanged [6,7] following structured endurance training (END). Similarly, inter-indi- vidual variability in training responses have also been observed following supra-maximal sprint interval training (SIT) [8,9]. While variability in training responses has been demon- strated following both END and SIT, it is currently unknown whether individuals who fail to PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 1 / 14 a11111 OPEN ACCESS Citation: Bonafiglia JT, Rotundo MP, Whittall JP, Scribbans TD, Graham RB, Gurd BJ (2016) Inter- Individual Variability in the Adaptive Responses to Endurance and Sprint Interval Training: A Randomized Crossover Study. PLoS ONE 11(12): e0167790. doi:10.1371/journal.pone.0167790 Editor: Jose A. L. Calbet, Universidad de las Palmas de Gran Canaria, SPAIN Received: June 6, 2016 Accepted: November 20, 2016 Published: December 9, 2016 Copyright: © 2016 Bonafiglia et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its supporting information files. Funding: Funding was provided by the Natural Sciences and Engineering Council of Canada (grant number: 402635; http://www.nserc-crsng.gc.ca/ index_eng.asp) and the Canadian Foundation for Innovation (grant number: 25476;https://www. innovation.ca/). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. respond following one type of exercise training might respond to a different training stimulus (i.e. different exercise volume, intensity and metabolic demand). While END and SIT differ substantially in exercise volume, intensity, and metabolic demand, at the group level they induce strikingly similar adaptations in VO2peak [10,11], lac- tate threshold [12,13], and muscle oxidative potential [14–16]. Interestingly, limited evidence demonstrating that central adaptations following training may differ between END and SIT [17,18], supports the potential that the mechanisms underlying similar adaptations in VO2peak may differ following END and SIT. Further, individual variability in both peripheral [19,20] and central [17] adaptations following training have been observed. Together these results suggest that both central and peripheral adaptations may vary in an individual follow- ing END or SIT, supporting the hypothesis that an individual who fails to respond following END may respond following SIT (and vis versa). Therefore, in order to determine if individuals respond differently to END and SIT, the present study compared individual responses following three weeks of both END and SIT uti- lizing a randomized crossover study design with a three-month washout period between train- ing interventions. Individual changes in VO2peak, lactate threshold, and submaximal heart rate (HR) were compared and the incidence of response and non-response for all variables were classified using typical error (TE), an index of measurement error that considers both biological and technical variability [21]. We hypothesized that individual responses to END would not necessarily reflect responses to SIT (and vis versa), potentially due to differences in central and peripheral adaptations. Methods Twenty-one healthy recreationally active (self-reported < three hours of physical activity per week) men (n = 9) and women (n = 12) volunteered to participate in the study. Each partici- pant attended a preliminary screening session where they were briefed on the study, provided informed consent, and had their height and weight recorded. Participants were not previously trained in cycling and were not involved in a training program at the start of the study. Partici- pants were informed to maintain their regular physical activity and nutritional habits through- out the duration of the study. All experimental procedures performed on human participants were approved by the Health Sciences Human Research Ethics board at Queen’s University. Verbal and written explanation of the experimental protocol and associated risks was provided to all participants prior to obtaining written informed consent. Experimental Design The current study utilized a randomized crossover design (Fig 1) where participants com- pleted two, three-week training interventions separated by a three-month wash-out period during which participants were instructed to return to their pre-study levels of physical activ- ity. Physiological testing occurred in the week preceding, and the week following each three- week training intervention. All physiological testing and training for both experiments was performed on a Monark Ergomedic 874 E stationary ergometer (Vansbro, Sweden). Eight additional participants completed a supplemental experiment to determine typical error for all variables. All participants were asked to refrain from alcohol and caffeine 12 hours before, and nutritional supplements and exercise 24 hours before all physiological testing. Physiological Testing In the week preceding (pre) and the week following (post) training, participants reported to the lab on three separate occasions, separated by 24–48 hours. During each visit participants Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 2 / 14 Competing Interests: The authors have declared that no competing interests exist. completed a VO2peak incremental ramp test to volitional exhaustion as described previously [22]. Briefly, following a 20 minute warm-up of four alternating five minute periods of load- less and 80W pedalling at 80 RPM, work rate was increased by 25W per minute until volitional exhaustion. Gas exchange and heart rate (HR) were collected throughout each ramp test using a metabolic cart (Moxus AEI Technologies, Pittsburgh, PA) and Polar HR monitors (Polar Team2 Pro, Kempele, Finland). VO2peak was calculated for each test as the highest 30 second average VO2 value, whereas submaximal HR was calculated for each test as the 30 second aver- age HR value during the third stage of the ramp protocol (~156 W). Final pre- and post-train- ing VO2peak and submaximal HR were determined by averaging the three values obtained during each testing period. RPM was collected continuously throughout each test and peak aerobic power (WRpeak) was calculated using the average WR from the last 30 seconds of the test, whereas the WR at VO2peak was calculated using the average WR during the same 30 sec- onds used to calculate VO2peak. Lactate Threshold Fingertip capillary blood (~20 uL) was sampled at rest (baseline) and within the last 10 s of each successive one-minute stage during the first VO2peak ramp test of pre- and post-testing test using a Lactate Scout + (EFK Diagnostics, Magdeberg, Germany) as done previously [13]. Lactate threshold was determined as the first recorded work rate (WR) where lactate was >4 mmol/L [23,24], often referred to as the onset of blood lactate accumulation at 4 mmol/L [25–27]. Training Interventions Training consisted of two, three week training periods separated by ~three months. During each training period participants were instructed to either cycle for 30 minutes at ~65% of WR at VO2peak (END) or perform eight, 20-second intervals at ~170% of WR at VO2peak, sepa- rated by 10 seconds of rest (SIT). Both training interventions required participants to train four times per week and the order of training was counterbalanced such that 12 (six males; six females) participants completed END first. All training sessions were preceded by a one-min- ute loadless warm-up. Participants were instructed to maintain a cadence of 80RPM and received verbal encouragement throughout all training sessions. HR was collected during Fig 1. Overview of experimental protocol. doi:10.1371/journal.pone.0167790.g001 Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 3 / 14 training, at three minute intervals (END) and at the end of each interval (SIT), using Polar HR monitors (Polar Team2 Pro, Kempele, Finland). Ratings of perceived exertion (RPE) were col- lected immediately following each training session using a 6–20 Borg Scale [28]. HR and RPE were averaged over all training sessions to determine training HR and RPE for END and SIT. Determination of Typical Error In order to determine typical error (TE) for VO2peak, lactate threshold and submaximal HR, a supplemental experiment involving eight recreationally active participants (four males; four females, age, 21±1 yrs; BMI, 21±2 kg/m2; VO2peak, 44±6 mL/kg/min) reported to the lab on two separate occasions separated by at least a week as described previously [9]. On each visit to the lab participants performed identical incremental ramp tests to volitional fatigue as described above. VO2peak, lactate threshold and submaximal HR were determined for each test as described above and the resulting values were utilized to calculate TE. Typical error (TE) of measurement was calculated for VO2peak, lactate threshold, and sub- maximal HR as described previously [21] utilizing the following equation: TE ¼ SDdiff= ffiffiffi 2 p Where SDdiff is the variance (standard deviation) of the difference scores observed between the 2 repeats of each test. A non-responder for VO2peak, lactate threshold, or submaximal HR was defined as an individual who failed to demonstrate an increase or decrease that was greater than two times the TE away from zero. A change beyond two times the TE means there is high probability (i.e. 12 to 1 odds) that this response is a true physiological adaptation beyond what might be expected to result from technical and/or biological variability [21]. Statistical Analysis Data are expressed as means and standard deviation. To ensure efficacy of the washout period baseline and response measures of VO2peak, lactate threshold, and submaximal HR between training period one and two were compared using unpaired t-tests as described previously [29]. Effects of training protocol (END vs. SIT) and time (Pre vs. Post) for all variables were examined using a two-way, repeated measures ANOVA. Any significant main effects or inter- actions were subsequently analyzed using a Bonferroni post hoc test where appropriate. Unpaired t-tests were also used to assess differences in the training response for all variables between males and females following END and SIT separately, and to determine if responses following END or SIT differed between training periods. A simple linear regression was used to determine the relationship between baseline variables between training period one and two and between the magnitude of response between END and SIT. Differences in training HR and RPE between END and SIT were assessed using paired t-tests and simple linear regres- sions were used to determine if these variables were related to the magnitude of physiological responses following training. A McNemar’s test was used to determine whether END and SIT elicited similar rates of response for VO2peak, lactate threshold, and submaximal HR. Statisti- cal significance was accepted at p < 0.05. Results Attendance at training sessions was 100% and all data reported are solely from those partici- pants that completed the full study protocol. Three participants dropped out of the study following pre-training testing in the first training period and were not included in final analy- sis. Average HR during and RPE immediately following SIT (HR: 172.8 ± 7.8 bpm; RPE: Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 4 / 14 17.9 ± 0.9, mean ± SD) was significantly higher (p < 0.05) than END (HR: 166.2 ± 13.7 bpm; RPE: 15.5 ± 1.4, mean ± SD). Interestingly, RPE reported immediately following each SIT session was significantly related with the magnitude of change in VO2peak induced by SIT (r = 0.5, p < 0.05). Baseline measures, and the magnitude of response for all variables for train- ing periods one and two are presented in Table 1. Unpaired t-tests revealed no differences between baseline measures for VO2peak (p = 0.62), lactate threshold (p = 0.12), and submaxi- mal HR (p = 0.86) between training periods one and two. No differences were observed in the magnitude of response between training periods one and two for VO2peak (p = 0.20), lactate threshold (p = 0.55), or submaximal HR (p = 0.62). Additionally, there was no difference in the mean END or SIT response for any variable between training periods. Further, baseline measures for training periods one and two were significantly related for VO2peak (r = 0.94, p < 0.05), lactate threshold (r = 0.82, p < 0.05), and submaximal HR (r = 0.82, p < 0.05). Participant characteristics and pre- and post-training values for END and SIT are presented in Table 2. A main effect of training (p < 0.05) was observed for VO2peak (Fig 2A), lactate threshold (Fig 2B), submaximal HR (Fig 2C), and WRpeak (Fig 2D). No condition (END vs. SIT) or interaction (condition x time) effects were observed for any variable examined. While males had higher baseline VO2peak, lactate threshold, submaximal HR, and WRpeak than females (p < 0.05), there were no statistical differences in the magnitude of training responses between sexes (Table 2). No significant relationships were observed between END and SIT for individual changes in VO2peak (r = 0.14, p = 0.57; Fig 3A), lactate threshold (r = 0.10, p = 0.70; Fig 3B), or submaximal HR (r = 0.17, p = 0.46). Baseline VO2peak did not predict changes in VO2peak following END (r = 0.28, p = 0.22) but was negatively related with the change in VO2peak induced by SIT (r = -0.59, p < 0.01). Baseline lactate threshold, and submaximal HR were not related with training-induced changes following either END (lactate threshold: r = 0.0, p = 1.0; HR: r = 0.37, p = 0.10) or SIT (lactate threshold: r = 0.29, p = 0.23; HR: r = 0.11, p = 0.63). Unpaired t-tests revealed that the baseline characteristics of the participants used in the ancillary TE study did not statistically differ from the participants in the present study for all variables in Table 2. Two times TE was 0.107 L/min for VO2peak, 157 W for lactate threshold, and 10.0 bpm for submaximal HR. Individual patterns of response and rates of non-response for VO2peak, lactate threshold, and submaximal HR following both END and SIT are pre- sented in Fig 4. Following training six non-responders were observed where an individual par- ticipant failed to improve in one measured variable following either END or SIT; however, in all cases these non-responders improved at least one variable following training utilizing the other exercise protocol. McNemar’s tests did not reveal significant differences in the incidence of response for VO2peak (p = 0.6), lactate threshold (p = 0.1), and submaximal HR (p = 0.6) between END and SIT. Table 1. Pre-training and magnitude of response for training periods 1 and 2 for all participants. Training Period One Training Period Two Pre-training Response Pre-training Response VO2peak (L/min) 3.0 ± 0.9 +0.05 ± 0.2 2.9 ± 0.9 +0.15 ± 0.3 VO2peak (mL/kg/min) 42.7 ± 6.4 +0.7 ± 3.2 41.2 ± 6.9 +2.2 ± 3.2 Lactate Threshold (W) 165.9 ± 43.2 +20.2 ± 18.3 190.7 ± 51.4 +14.7 ± 34.9 HRsubmax (bpm) 153.9 ± 24.7 -5.8 ± 7.8 152.6 ± 21.5 -4.1 ± 12.8 Values are means ± standard deviation. doi:10.1371/journal.pone.0167790.t001 Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 5 / 14 Discussion The current study examined individual responses in VO2peak, lactate threshold and submaxi- mal exercise heart rate (HR) following three weeks of both END and SIT. Carryover effects from training period one to two were absent and the magnitude of responses between training periods was not different for any variable. These data highlight the effective implementation of our randomized cross-over study design [29]. In summary, the current study demonstrates inter-individual variability in the training responses to END and SIT and suggests that individ- ual patterns of response are dependent on the training protocol utilized. The major novel findings of the current study are that: 1) while END and SIT increased VO2peak, lactate threshold and submaximal HR at the group level with no differences observed between protocols, improvements within a given individual following END did not predict the improvement observed following SIT (and vice versa), 2) individual patterns of response were observed following both END and SIT, however these patterns varied within individuals between END and SIT, and 3) while our analysis revealed non-responses for one or more variables within most participants, we failed to observe a global non-response to END and SIT in any individual. Similar Group Responses in the Initial Adaptations to END and SIT At the group level, END protocols are effective at increasing VO2peak and lactate threshold [30], while SIT protocols at supra-maximal intensities also improve VO2peak [31] and lactate Table 2. Participant characteristics and group responses to END and SIT. END Pre Post Males (n = 9) Females (n = 12) Total (n = 21) Males (n = 9) Females (n = 12) Total (n = 21) Age (yrs) 20.4 ± 1.2 19.9 ± 1.2 20.3 ± 0.9 - - - Height (cm) † 181 ± 6 165 ± 7 172 ± 12 - - - Body mass (kg) † 81.9 ± 10.3 62.0 ± 11.6 70.0 ± 14.7 82.0 ± 10.9 61.2 ± 11.5 69.5 ± 15.6 VO2peak (L/min) † 3.7 ± 0.5 2.4 ± 0.5 3.0 ± 0.9 3.8 ± 0.5 2.5 ± 0.5 3.1 ± 0.9* VO2peak (mL/kg/min) † 46.0 ± 3.9 39.3 ± 6.7 42.2 ± 6.4 47.3 ± 5.4 41.4 ± 6.2 43.9 ± 6.4* Lactate Threshold (W) † 209 ± 38 149 ± 40 175 ± 47 233 ± 40 171 ± 46 199 ± 51* WRpeak (W) † 296 ± 42 196 ± 39 238 ± 64 309 ± 58 210 ± 32 252 ± 66* HRsubmax (bpm) † 135 ± 11 166 ± 19 152 ± 22. 129 ± 8 159 ± 18 146 ± 21* SIT Pre Post Males (n = 9) Females (n = 12) Total (n = 21) Males (n = 9) Females (n = 12) Total (n = 21) Age (yrs) 20.4 ± 1.2 19.9 ± 1.2 20.3 ± 0.9 - - - Height (cm) 181 ± 6 165 ± 7 172 ± 12 - - - Body mass (kg) † 82.8 ± 11.6 62.2 ± 12.4 70.4 ± 15.6 83.1 ± 11.2 61.7 ± 11.5 70.3 ± 15.6 VO2peak (L/min) † 3.7 ± 0.6 2.4 ± 0.5 3.0 ± 0.9 3.7 ± 0.5 2.6 ± 0.4 3.1 ± 0.9* VO2peak (mL/kg/min) † 45.0 ± 9.3 39.2 ± 5.5 41.7 ± 6.9 44.7 ± 5.5 41.6 ± 5.4 42.9 ± 5.5* Lactate Threshold (W) † 215 ± 36 154 ± 40 180 ± 47 230 ± 33.3 165 ± 41 192 ± 47* WRpeak (W) † 292 ± 45 202 ± 30 241 ± 57 314 ± 46 210 ± 32 255 ± 63* HRsubmax (bpm) † 133 ± 13 169 ± 19 155 ± 24 129 ± 9 167 ± 20 151 ± 25* Values are means ± standard deviation. WRpeak, peak aerobic power; HRsubmax, submaximal heart rate. †Significant baseline difference between males and females, p < 0.05. *Main effect of training, p < 0.05. doi:10.1371/journal.pone.0167790.t002 Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 6 / 14 threshold [12,13]. Consistent with these results and previous work form our lab utilizing the same protocols [16], a main effect of training was observed in the current study for VO2peak, lactate threshold, and submaximal HR. Also consistent with previous studies comparing END and SIT [14–16] no differences were observed between protocols for the magnitude of response at the group level. While the primary purpose of present study was not to determine if the group responses to END and SIT differ, the observation of similar group responses to END and SIT may suggest that a larger sample size is required to attain statistical power in order to detect potential interaction effects between training protocols. Individual Variability in Responsiveness to END and SIT While variability in the individual responses to END is established [2,6,7,19,32,33], we recently demonstrated similar variability in response to the SIT protocol utilized in the present study [9]. The major novel finding of the current study is our demonstration of variability in the indi- vidual responses following different training protocols (END and SIT). Specifically, our results demonstrated that exercise protocols which differ in intensity, time, and metabolic demand, like END and SIT, can induce different adaptive responses in VO2peak, lactate threshold and submaximal HR within a given individual. These findings confirm the hypothesis that individu- als who are not sensitive to a given exercise protocol may experience adaptation if exposed to a Fig 2. Group responses following 3 weeks of END and SIT. Group responses for VO2peak (A), lactate threshold (B), submaximal HR (C), and WRpeak (D). *Significant main effect of training, p < 0.05. doi:10.1371/journal.pone.0167790.g002 Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 7 / 14 Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 8 / 14 different protocol [5], potentially due to different sensitivities to training volume [7] and/or intensity [32]. While the mechanisms determining individual variability in sensitivities to differ- ing training protocols are unknown, genetic predispositions [34] may be responsible for vari- ance in the capacity of central [17] and peripheral [14,35] adaptations to training. A similar disassociation between individual changes in VO2peak has previously been observed following END and resistance training [33], but to our knowledge we are the first to demonstrate inter- individual variability in the response to two protocols known to induce equivalent improve- ments in aerobic capacity at the group level. Importantly, while the current data suggests that individuals may respond favorably to a change in training stimulus, we cannot rule out the pos- sibility that the different individual responses observed between END and SIT were a result of simply training twice at two different times (i.e. it is possible that an individual completing END twice may not demonstrate identical responses), differences in external physical activity between training periods and/or changes in nutritional habits caused by different training pro- tocols (i.e. END vs. SIT), or training at different times of the year (i.e. fall vs. winter). Addition- ally, the present study only examined individual variability in the initial response to training (i.e. the response to three weeks of training), and it remains possible that individual differences Fig 3. Correlations of individual responses following 3 weeks of END and SIT. Relationship between individual responses in VO2peak (A) and lactate threshold (B). Dashed lines represent the typical error cut-offs. Individuals falling within the shaded area failed to improve either VO2peak or lactate threshold following both END and SIT, while the hashed area represents an adverse response following both training protocols. doi:10.1371/journal.pone.0167790.g003 Fig 4. Individual patterns of response following three weeks of training. Positive responses (white boxes), non-responses (grey boxes) and adverse responses (black boxes) are shown for all participants across all variables following END (A) and SIT (B). A dashed box indicates that data was unavailable for a given variable. Individuals who failed to improve any variables for either END or SIT, “Overall non-responders” are indicated by diamond filled boxes. The percentage of participants demonstrating a non-response (NR; including both non- and adverse responses) for each variable, and overall, is also provided. doi:10.1371/journal.pone.0167790.g004 Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 9 / 14 observed following three weeks of END and SIT may not persist following longer training peri- ods. Thus, while our results support the consideration of multiple training protocols when attempting to optimize individual exercise prescription [36], there remains very little data, and much future work still needed, before we fully understand inter-individual responsiveness to different training protocols. Consistent with previous observations of heterogeneity in the individual response to END [7,32,33,37] and SIT [8,9] we have also observed significant rates of non-response following both END and SIT in the current study (Fig 4). The present finding that END and SIT elicited similar rates of non-response for VO2peak, lactate threshold, and submaximal HR agrees with previous observations that END and SIT Importantly, while inter-individual variability in the response to training has been repeatedly demonstrated [7,33,37], attempts to quantify individ- uals as responders or non-responders are relatively recent [5–9,32,38]. In the current study, the use of two times the typical error (TE) to identify responders and non-responders [21] may have led to higher incidences of non-responses than previously reported [6–8,32]. How- ever, despite the use of this conservative method of identifying responders, we have observed a subset of adverse responders to VO2peak, lactate threshold, and submaximal HR following both END and SIT that is consistent with previous observations of adverse responses to exer- cise for a variety of cardiovascular risk factors [5]. Interestingly, a non- or adverse response to VO2peak, lactate threshold, or submaximal HR following one training protocol did not pre- clude a positive response following the other training protocol. Recently, several reports have recommended that before individuals are classified as responders or non-responders, it is important to determine if variability in the individual responses within the experimental con- dition are greater than within-subject variation [39–41]. While we were unable to conduct this analysis due to our current study lacking a time-matched control group, it is important that future studies examining rates of response/non-response to exercise training consider the recently recommended approach to performing these analyses [39–41]. This the limitation aside, the current study adds to a growing body of literature that identifies a portion of the population that either does not respond, or responds adversely to exercise training and sug- gests that these non-/adverse-responders may respond more positively to different training protocols. Mechanisms Underlying Individual Variability to END and SIT Despite marked differences in the physiological stress they impose, a single bout of END or SIT elicits analogous molecular responses in skeletal muscle [16], leading to similar peripheral adaptations including changes in fibre-type distribution [16], increased skeletal muscle oxida- tive capacity [14–16,42] and resting muscle glycogen content [14–16]. Interestingly, the central adaptations elicited by END or SIT are inconsistent [17,18], however, only central adaptations associated with six weeks of END prevails as independent predictors of the VO2peak responses [43]. Few studies have compared both central and peripheral adaptations to multiple training protocols and significant differences in training duration, frequency, and volume limits the ability to compare and interpret findings from different studies [43,44]. While recent research has elucidated mechanisms that primarily explain the adaptive responses to training [43], future research is needed to determine if variability in the mechanisms that underlie changes in exercise capacity/performance explain individual response variability following training. At the individual level, heterogeneity in both central [17] and peripheral adaptations are present following END [19,20] and SIT [14], which suggests that variability in individual responses to END and SIT may be due in part to individual variance in the magnitude of peripheral and central adaptations. Why variance in central and/or peripheral adaptations Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 10 / 14 may exist within an individual following different training protocols is currently unknown, however, evidence from the HERITAGE study suggests that much of this variability may results from a genetic predisposition to a specific type of training stimulus [2]. Interestingly, recent evidence has found associations between several genetic markers and individual train- ing responses [34,45,46], however, while these findings are a step towards optimizing individ- ual exercise prescriptions [34] whether genetic signatures exist that may predict which type of training an individual is most likely to respond to is unknown. This remains an interesting and important area for future investigation. Individual Patterns of Response Following END and SIT we observed individual patterns of response, where improvements in VO2peak were not necessarily associated with improvements in lactate threshold or submaxi- mal HR (Fig 4). The existence of individual patterns of response is consistent with previous studies demonstrating that non-responders in VO2peak can be responders to other variables associated with END [6,19] and SIT [8,9]. An additional novel finding of the present study is that individual patterns of response were different following END and SIT. This variability in individual patterns of response meant that even though several individuals failed to improve any variable following either END or SIT, no “global non-responders” (i.e. individuals that failed to improve following either protocol) were observed. These results further support the consideration of multiple training protocols when prescribing exercise, and raise the possibil- ity that an individual who does not appear to be responding to an initial exercise prescription may respond more favourably if an alternative mode of training is prescribed. As continuing the training stimulus beyond initial exposure (four weeks) reduces the incidence of non- response in VO2peak [32], whether switching training protocols after initial exposure or extending the amount of training prescription is equally effective at diminishing non- responses remains an area for future research. Conclusion The current study assessed individual responses in VO2peak, lactate threshold, and submaxi- mal exercise heart rate (HR) following three weeks of both END and SIT. While training elic- ited significant improvements in all variables at the group level, considerable heterogeneity was observed in the individual responses including a number of non-/adverse-responders. Further, individual patterns of response were not related across END and SIT and appear to be training protocol dependent. All participants demonstrated a positive response in at least one variable following the completion of both END and SIT suggesting that the existence of true non-responders to exercise training is unlikely and that different training protocols should be considered when optimizing individual exercise prescription. Supporting Information S1 Table. Raw data used for all tables and figures. (XLSX) Acknowledgments The authors would like to thank Elizabeth Mathew and Wendy Fu for their help with HR data analysis and a dedicated group of volunteers for their help in conducting training sessions. Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 11 / 14 Author Contributions Conceptualization: JB MR JW TS RG BG. Data curation: JB MR JW TS RG BG. Formal analysis: JB MR JW TS RG BG. Funding acquisition: BG. Methodology: JB MR JW TS RG BG. Writing – original draft: JB MR JW TS RG BG. Writing – review & editing: JB MR JW TS RG BG. References 1. Kohrt WM, Malley MT, Coggan AR, Spina RJ, Ogawa T, Ehsani AA, et al. Effects of gender, age, and fit- ness level on response of VO2max to training in 60–71 yr olds. J Appl Physiol [Internet]. 1991 Nov [cited 2016 Feb 28]; 71(5):2004–11. Available from: http://jap.physiology.org/content/71/5/2004. abstract PMID: 1761503 2. Bouchard C, Rankinen T. Individual differences in response to regular physical activity. Med Sci Sports Exerc. 2001; 33(6 Suppl):S446–51; discussion S452–3. PMID: 11427769 3. Skinner JS, Jasko´lski a, Jasko´lska a, Krasnoff J, Gagnon J, Leon a S, et al. Age, sex, race, initial fitness, and response to training: the HERITAGE Family Study. J Appl Physiol. 2001; 90(5):1770–6. PMID: 11299267 4. Karavirta L, Ha¨kkinen K, Kauhanen A, Arija-Bla´zquez A, Sillanpa¨a¨ E, Rinkinen N, et al. Individual responses to combined endurance and strength training in older adults. Med Sci Sports Exerc. 2011; 43 (3):484–90. doi: 10.1249/MSS.0b013e3181f1bf0d PMID: 20689460 5. Bouchard C, Blair SN, Church TS, Earnest CP, Hagberg JM, Ha¨kkinen K, et al. Adverse metabolic response to regular exercise: Is it a rare or common occurrence? PLoS One. 2012; 7(5). 6. Scharhag-Rosenberger F, Walitzek S, Kindermann W, Meyer T. Differences in adaptations to 1 year of aerobic endurance training: Individual patterns of nonresponse. Scand J Med Sci Sport. 2012; 22 (1):113–8. 7. Sisson SB, Katzmarzyk PT, Earnest CP, Bouchard C, Blair SN, Church TS. Volume of exercise and fit- ness nonresponse in sedentary, postmenopausal women. Med Sci Sports Exerc. 2009; 41(3):539–45. doi: 10.1249/MSS.0b013e3181896c4e PMID: 19204597 8. Astorino TA, Schubert MM. Individual responses to completion of short-term and chronic interval train- ing: A retrospective study. PLoS One. 2014; 9(5). 9. Gurd BJ, Giles MD, Bonafiglia JT, Raleigh JP, Boyd JC, Ma JK, et al. Incidence of nonresponse and individual patterns of response following sprint interval training. Appl Physiol Nutr Metab [Internet]. 2016; 41(3):229–34. doi: 10.1139/apnm-2015-0449 PMID: 26854820 10. Gist NH, Fedewa M V., Dishman RK, Cureton KJ. Sprint Interval Training Effects on Aerobic Capacity: A Systematic Review and Meta-Analysis. Sport Med [Internet]. 2014; 44(2):269–79. Available from: http://link.springer.com/10.1007/s40279-013-0115-0 11. Sloth M, Sloth D, Overgaard K, Dalgas U. Effects of sprint interval training on VO 2max and aerobic exer- cise performance: A systematic review and meta-analysis. Scand J Med Sci Sports [Internet]. 2013; 23 (6):e341–52. doi: 10.1111/sms.12092 PMID: 23889316 12. Esfarjani F, Laursen PB. Manipulating high-intensity interval training: Effects on, the lactate threshold and 3000m running performance in moderately trained males. J Sci Med Sport [Internet]. 2007; 10 (1):27–35. Available from: http://linkinghub.elsevier.com/retrieve/pii/S1440244006001149 doi: 10. 1016/j.jsams.2006.05.014 PMID: 16876479 13. Zelt JGE, Hankinson PB, Foster WS, Williams CB, Reynolds J, Garneys E, et al. Reducing the volume of sprint interval training does not diminish maximal and submaximal performance gains in healthy men. Eur J Appl Physiol [Internet]. 2014; Available from: http://www.ncbi.nlm.nih.gov/pubmed/25091854 14. Gibala MJ, Little JP, van Essen M, Wilkin GP, Burgomaster KA, Safdar A, et al. Short-term sprint interval versus traditional endurance training: similar initial adaptations in human skeletal muscle and exercise performance. J Physiol. 2006; 575(Pt 3):901–11. doi: 10.1113/jphysiol.2006.112094 PMID: 16825308 Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 12 / 14 15. Burgomaster K a, Howarth KR, Phillips SM, Rakobowchuk M, Macdonald MJ, McGee SL, et al. Similar metabolic adaptations during exercise after low volume sprint interval and traditional endurance training in humans. J Physiol. 2008; 586(1):151–60. doi: 10.1113/jphysiol.2007.142109 PMID: 17991697 16. Scribbans TD, Edgett BA, Vorobej K, Mitchell AS, Joanisse SD, Matusiak JBL, et al. Fibre-specific responses to endurance and low volume high intensity interval training: Striking similarities in acute and chronic adaptation. PLoS One. 2014; 9(6). 17. MacPherson REK, Hazell TJ, Olver TD, Paterson DH, Lemon PWR. Run sprint interval training improves aerobic performance but not maximal cardiac output. Med Sci Sports Exerc. 2011; 43(1):115– 22. doi: 10.1249/MSS.0b013e3181e5eacd PMID: 20473222 18. Matsuo T, Saotome K, Seino S, Shimojo N, Matsushita A, Iemitsu M, et al. Effects of a low-volume aero- bic-type interval exercise on _VO 2max and cardiac mass. Med Sci Sports Exerc. 2014; 46(1):42–50. doi: 10.1249/MSS.0b013e3182a38da8 PMID: 23846165 19. Vollaard NBJ, Constantin-Teodosiu D, Fredriksson K, Rooyackers O, Jansson E, Greenhaff PL, et al. Systematic analysis of adaptations in aerobic capacity and submaximal energy metabolism provides a unique insight into determinants of human aerobic performance. J Appl Physiol. 2009; 106(5):1479–86. doi: 10.1152/japplphysiol.91453.2008 PMID: 19196912 20. McPhee JS, Williams AG, Perez-Schindler J, Degens H, Baar K, Jones DA. Variability in the magnitude of response of metabolic enzymes reveals patterns of co-ordinated expression following endurance training in women. Exp Physiol. 2011; 96(7):699–707. doi: 10.1113/expphysiol.2011.057729 PMID: 21571817 21. Hopkins WG. Measures of reliability in sports medicine and science. Sports Med. 2000; 30(1):1–15. PMID: 10907753 22. Edgett BA, Foster WS, Hankinson PB, Simpson CA, Little JP, Graham RB, et al. Dissociation of Increases in PGC-1α and Its Regulators from Exercise Intensity and Muscle Activation Following Acute Exercise. PLoS One [Internet]. 2013; 8(8):e71623. Available from: http://dx.plos.org/10.1371/journal. pone.0071623 doi: 10.1371/journal.pone.0071623 PMID: 23951207 23. Bishop D, Jenkins DG, Mackinnon LT. The relationship between plasma lactate parameters, Wpeak and 1-h cycling performance in women. Med Sci Sports Exerc [Internet]. 1998 Aug [cited 2015 Nov 25]; 30(8):1270–5. Available from: http://www.ncbi.nlm.nih.gov/pubmed/9710868 PMID: 9710868 24. Heck H, Mader A, Hess G, Mucka S, Muller R, Hollmann W. Justification of the 4-mmol/l Lactate Threshold. Int J Sports Med. 1985; 6(3):117–30. doi: 10.1055/s-2008-1025824 PMID: 4030186 25. Bentley DJ. Incremental Exercise Test Design and Analysis. Sport Med. 2007; 37(7):575–86. 26. Svedahl K, MacIntosh BR. Anaerobic threshold: the concept and methods of measurement. Can J Appl Physiol. 2003; 28(2):299–323. PMID: 12825337 27. Faude O, Kindermann W, Meyer T. Lactate threshold concepts: How valid are they? Sport Med [Inter- net]. 2009; 39(6):469–90. Available from: file:///C:/Users/mitch_000/Downloads/Faude-SportsMed- 2009-SG-Fisiologia.pdf 28. Borg GA V. Psychophysical bases of percieved exertion. Med Sci Sports Exerc. 1982; 14(5):377–81. PMID: 7154893 29. Wellek S, Blettner M. On the Proper Use of the Crossover Design in Clinical Trials. Dtsch Arztebl Int. 2012; 109(15):276–81. doi: 10.3238/arztebl.2012.0276 PMID: 22567063 30. Jones AM, Carter H. The effect of endurance training on parameters of aerobic fitness. Sports Med. 2000; 29(6):373–86. PMID: 10870864 31. Bacon AP, Carter RE, Ogle EA, Joyner MJ. VO2max Trainability and High Intensity Interval Training in Humans: A Meta-Analysis. PLoS One. 2013; 8(9). 32. Ross R, de Lannoy L, Stotz PJ. Separate Effects of Intensity and Amount of Exercise on Interindividual Cardiorespiratory Fitness Response. Mayo Clin Proc [Internet]. Elsevier Inc; 2015; 90(11):1–9. Avail- able from: http://linkinghub.elsevier.com/retrieve/pii/S0025619615006400 33. Hautala AJ, Kiviniemi AM, Ma¨kikallio TH, Kinnunen H, Nissila¨ S, Huikuri H V., et al. Individual differ- ences in the responses to endurance and resistance training. Eur J Appl Physiol. 2006; 96(5):535–42. doi: 10.1007/s00421-005-0116-2 PMID: 16369817 34. Timmons JA, Knudsen S, Rankinen T, Koch LG, Jensen T, Keller P, et al. Using molecular classification to predict gains in maximal aerobic capacity following endurance exercise training in humans programs Using molecular classification to predict gains in maximal aerobic capacity following endurance exercise training in human. J Appl Physiol. 2012; 35. Granata C, Oliveira RSF, Little JP, Renner K, Bishop DJ. Training intensity modulates changes in PGC- 1 and p53 protein content and mitochondrial respiration, but not markers of mitochondrial content in human skeletal muscle. FASEB J [Internet]. 2015 Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 13 / 14 36. Buford TW, Roberts MD, Church TS. Toward exercise as personalized medicine. Sport Med. 2013; 43 (3):157–65. 37. Bouchard C, An P, Rice T, Skinner JS, Wilmore JH, Gagnon J, et al. Familial aggregation ofV_o 2 max response to exercise training: results from the HERITAGE Family Study. J Appl Physiol [Internet]. 1999; 87(3):1003–8. Available from: http://jap.physiology.org/content/87/3/1003.abstract PMID: 10484570 38. Wolpern AE, Burgos DJ, Janot JM, Dalleck LC. Is a threshold-based model a superior method to the rel- ative percent concept for establishing individual exercise intensity? a randomized controlled trial. BMC Sports Sci Med Rehabil [Internet]. BMC Sports Science, Medicine and Rehabilitation; 2015; 7(1):16. Available from: http://www.biomedcentral.com/2052-1847/7/16 39. Hopkins WG. Individual responses made easy. J Appl Physiol [Internet]. 2015; 118(12):1444–6. doi: 10. 1152/japplphysiol.00098.2015 PMID: 25678695 40. Hecksteden A, Kraushaar J, Scharhag-Rosenberger F, Theisen D, Senn S, Meyer T. Individual response to exercise training -a statistical perspective. J Appl Physiol. 2015; 118:1450–9. doi: 10.1152/ japplphysiol.00714.2014 PMID: 25663672 41. Atkinson G, Batterham AM. True and false inter-individual differences in the physiological response to an intervention. Exp Physiol [Internet]. 2015; 6:1–29. Available from: http://www.ncbi.nlm.nih.gov/ pubmed/25823596 42. Ma JK, Scribbans TD, Edgett BA, Boyd CJ, Simpson CA, Little JP, et al. Extremely low-volume, high- intensity interval training improves exercise capacity and increases mitochondrial protein content in human skeletal muscle. Open J Mol Integr Physiol [Internet]. 2013; 03(04):202–10. Available from: http://www.scirp.org/journal/PaperInformation.aspx?PaperID=39842&#abstract 43. Montero D, Cathomen A, Jacobs RA, Flu¨ck D, de Leur J, Keiser S, et al. Haematological rather than skeletal muscle adaptations contribute to the increase in peak oxygen uptake induced by moderate endurance training. J Physiol [Internet]. 2015; 593(20):4677–88. doi: 10.1113/JP270250 PMID: 26282186 44. Jacobs RA, Flu¨ck D, Bonne TC, Bu¨rgi S, Christensen PM, Toigo M, et al. Improvements in exercise per- formance with high-intensity interval training coincide with an increase in skeletal muscle mitochondrial content and function. J Appl Physiol [Internet]. 2013; 115(6):785–93. Available from: http://www.ncbi. nlm.nih.gov/pubmed/23788574 doi: 10.1152/japplphysiol.00445.2013 PMID: 23788574 45. Bouchard C, Sarzynski MA, Rice TK, Kraus WE, Church TS, Sung YJ, et al. Genomic predictors of the maximal O2 uptake response to standardized exercise training programs. J Appl Physiol. 2011; 110 (5):1160–70. doi: 10.1152/japplphysiol.00973.2010 PMID: 21183627 46. He Z, Hu Y, Feng L, Li Y, Liu G, Xi Y, et al. NRF-1 genotypes and endurance exercise capacity in young Chinese men. Br J Sports Med. 2008; 42(5):361–6. doi: 10.1136/bjsm.2007.042945 PMID: 18184751 Individual Responses to Endurance and Sprint Interval Training PLOS ONE | DOI:10.1371/journal.pone.0167790 December 9, 2016 14 / 14

Inter-Individual Variability in the Adaptive Responses to Endurance and Sprint Interval Training: A Randomized Crossover Study.

12-09-2016

Bonafiglia, Jacob T,Rotundo, Mario P,Whittall, Jonathan P,Scribbans, Trisha D,Graham, Ryan B,Gurd, Brendon J

eng

PMC10185608

Vol.:(0123456789) Sports Medicine (2023) 53:1255–1271 https://doi.org/10.1007/s40279-023-01816-1 ORIGINAL RESEARCH ARTICLE Variability in Running Economy of Kenyan World‑Class and European Amateur Male Runners with Advanced Footwear Running Technology: Experimental and Meta‑analysis Results Melanie Knopp1,3  · Borja Muñiz‑Pardos2  · Henning Wackerhage3  · Martin Schönfelder3  · Fergus Guppy4  · Yannis Pitsiladis5  · Daniel Ruiz1 Accepted: 26 January 2023 / Published online: 2 March 2023 © The Author(s) 2023 Abstract Background Advanced footwear technology improves average running economy compared with racing flats in sub-elite athletes. However, not all athletes benefit as performance changes vary from a 10% drawback to a 14% improvement. The main beneficiaries from such technologies, world-class athletes, have only been analyzed using race times. Objective The aim of this study was to measure running economy on a laboratory treadmill in advanced footwear technol- ogy compared to a traditional racing flat in world-class Kenyan (mean half-marathon time: 59:30 min:s) versus European amateur runners. Methods Seven world-class Kenyan and seven amateur European male runners completed a maximal oxygen uptake assess- ment and submaximal steady-state running economy trials in three different models of advanced footwear technology and a racing flat. To confirm our results and better understand the overall effect of new technology in running shoes, we conducted a systematic search and meta-analysis. Results Laboratory results revealed large variability in both world-class Kenyan road runners, which ranged from a 11.3% drawback to a 11.4% benefit, and amateur Europeans, which ranged from a 9.7% benefit to a 1.1% drawback in running economy of advanced footwear technology compared to a flat. The post-hoc meta-analysis revealed an overall significant medium benefit of advanced footwear technology on running economy compared with traditional flats. Conclusions Variability of advanced footwear technology performance appears in both world-class and amateur runners, suggesting further testing should examine such variability to ensure validity of results and explain the cause as a more per- sonalized approach to shoe selection might be necessary for optimal benefit. Key Points Running economy of world-class Kenyan and amateur European runners with next-generation long-distance running shoes that contain advanced footwear technol- ogy varies greatly, with a range from a 11.4% benefit to a 11.3% detriment. Meta-analysis results reveal an overall statistically sig- nificant medium benefit of advanced footwear technol- ogy on running economy when compared with tradi- tional racing flats and confirmed the variability we report when examining the performance benefits of advanced footwear technology. Our results suggest a more personalized approach to new footwear technology. * Melanie Knopp Melanie.Knopp@adidas.com 1 adidas Innovation, adidas AG, Herzogenaurach, Germany 2 GENUD Research Group, Faculty of Health and Sport Sciences, University of Zaragoza, Saragossa, Spain 3 Department of Sport and Health Sciences, Technical University of Munich, Munich, Germany 4 Institute of Life and Earth Sciences, Heriot Watt University, Edinburgh, UK 5 School of Sport and Health Sciences, University of Brighton, Eastbourne, UK 1256 M. Knopp et al. 1 Introduction Kenyan elite runners win many international track and road distance races, which has stimulated research into the causes of this success [1–6]. When examining the geographical dis- tribution of the top 20 running performances for male and female athletes in both middle- and long-distance events (800 m, 1500 m, 3000 m, 5000 m, 10,000 m, 5 km, 10 km, half-marathon, and marathon) in the past 5 years (since the last Olympic cycle: 5 August, 2016 to 29 August, 2021), 41.6% have been achieved by Kenyan athletes [7]. Such running performances depend on three main physiological factors: (1) an athletes’ maximal oxygen uptake ( ̇VO2max), (2) their fractional utilization of ̇VO2max or the ability of an athlete to sustain a high percentage of their ̇VO2max for long periods of time, and (3) their running economy [8–11]. Previous research examining the uniqueness specifically of Kenyan or other elite East African runners has suggested that of these, it is running economy that is particularly unique in this population [6, 10, 12]. Various studies have further attributed this especially to the anthropometric char- acteristics of East Africans with smaller body size, thinner lower legs, and a greater Achilles tendon moment arm with a shorter forefoot length [1, 10, 12–14]. Running economy can be defined as the ability to move efficiently in terms of energy demand while running at a specified submaximal velocity and can be measured as the rate of oxygen uptake per kilogram body weight and min- ute ( ̇VO2 in mL O2/kg/min) at that speed [10, 11, 15, 16]. Previous work has reported that among elite runners with similar ̇VO2max levels, running economy can account for 65.4% of the variation observed in a 10-km race perfor- mance [17]. Running economy is affected by many factors including anthropometric, biomechanical, metabolic, neu- romuscular, and cardiorespiratory efficiency [11]. One ele- ment that has gained interest in recent years is an athlete’s mechanical efficiency being affected by different footwear characteristics such as weight, cushioning, and longitudinal bending stiffness, all of which are included in recent tech- nological advances in long-distance running shoes [18–21]. Previously published work has attributed the improvements of performance of such advanced footwear technology to various mechanisms [20, 22]. The advances in shoe tech- nology themselves have been designed to maximize run- ning economy while minimizing energy loss and consist of a curved stiff element component and a high midsole stack height made of a compliant, resilient, and lightweight foam (Fig. 1). The curved rigid element increases the longitudinal bending stiffness of the shoe and thereby creates a mecha- nism with a teeter-totter effect on the running mechanics, which occurs when a runner’s center of pressure overcomes the bending point of the curved structure and causes the reaction force to act on the heel perpendicular to the stiff ele- ment providing leverage during push-off [20, 23]. The high midsole stack height enhances this mechanism and allows for a more curved plate to be inserted into the midsole [20]. The compliant, resilient, lightweight foam material for the midsole ensures that the shoe weight remains light while still having a soft foam with a high-energy return as these have all been suggested to also affect performance [18–20]. The impact of advanced footwear technology on running events is reflected in the progression of world records, with every male and female world record starting from 5 km to the marathon broken by athletes wearing different versions of these shoes since their release [24]. Previous research completed on such footwear technology in the field quanti- fies this impact on performance, with data from the Strava fitness app on more than a million marathon and half-mar- athons revealing that shoes containing this new technology could improve race performance in sub-elite athletes, as individuals ran 4–5% faster in advanced footwear technology than runners wearing an average racing flat [25]. Similarly, Rodrigo-Carranza et al. showed that in a sub-cohort of top- 100 men’s marathon performances from 2015 to 2019 that completed races in both advanced footwear technology and traditional flats, 29 of 40 athletes (72.50%) improved their Fig. 1 Schematic of different long-distance running shoes, including A a traditional racing flat, which is classically low to the floor with relatively thin soles with the focus here being to keep the shoes lightweight, and B advanced footwear technology, which consists of a curved stiff element in the forefoot of the shoe, as well as a high midsole stack height made up of a resilient, compliant, and lightweight foam 1257 Variability in Advanced Footwear Technology performance with this type of footwear [26]. This is also supported by various laboratory-based running economy studies comparing advanced footwear technology to tradi- tional racing flats in sub-elite athletes, suggesting that the design of these shoes reduces the energy cost of running on average by about 2.7–4.4%, thereby benefiting overall run- ning performance [15, 27–30]. While previous studies have compared the running econ- omy of non-elite runners wearing different shoe technologies in relatively controlled laboratory settings [15, 27–30], no study has examined the variability in running economy of the main beneficiaries (i.e., world-class athletes). Knowing this, the primary aim of this study was to answer the research question: how does the variability in physiological response in terms of running economy on a laboratory treadmill in advanced footwear technology compare to a traditional rac- ing flat in world-class Kenyan distance runners (half-mar- athon mean time: 59:30 min:s) versus European amateur runners? Based on the obtained results, we decided to sys- tematically search the literature for similar relevant studies and conducted a post-hoc meta-analysis to confirm the found range of variability, and better understand the overall effect of advanced footwear technology. 2 Materials and Methods 2.1 Participants Fifteen subjects volunteered to participate in this study and were classified as either world class or amateur. Run- ners with current or recent injuries that prevented them from training were excluded, as well as those uncomforta- ble with running on a treadmill. Shoe size was also part of the inclusion criteria because of shoe cost considerations. One participant dropped out as he struggled to run on a treadmill, meaning 14 participants were finally included for analysis in this study. The world-class cohort comprised seven male world-class Kenyan runners (mean ± standard devia- tion, age: 22.7 ± 3.2 years, height: 1.7 ± 0.05 m, mass: 59.9 ± 4.8  kg, body mass index: 19.7 ± 0.6  kg/m2, ̇VO2peak: 75.9 ± 3.5 mL/kg/min) (Table 1) [31]. These runners were recruited through sponsorship deals with collaborating companies and were all professional road racing athletes who had an official mean personal record for the half-marathon of 59:30 ± 0:48 min:s, and a 10-km personal best of 27:33 ± 0:41 min:s. The amateur cohort consisted of seven well-trained male amateur European runners, who at the time of measurement were training daily, (mean ± standard deviation, age: 28.1 ± 4.2 years, height: 1.8 ± 0.03 m, mass: 72.1 ± 7.0 kg, body mass index: 21.9 ± 1.8 kg/m2, ̇VO2peak: 62.3 ± 5.1 mL/kg/min) and volunteered to take part in this research (Table 1). All participants gave written informed consent to being a part of this study after they understood the experimental procedures, potential injury risks, and possible benefits. 2.2 Shoes Throughout the experimental protocol, analyzed shoe con- ditions included a commercially available traditional rac- ing shoe (FLAT) used by the subjects regularly for their own training, as well as three different commercially avail- able models of AdvFootTech (1–3) that differed in their geometry and weight (Table 2). As all athletes were the same shoe size, everyone tested in UK 8.5 (US 9/EU 42 2/3). 2.3 Experimental Protocol This study comprised two laboratory visits occurring on separate days, with a 24-h pause for recovery, at the adidas Sports Science Research Laboratory in Herzogenaurach, Germany located close to sea level at an altitude of 300 m (Fig. 2). During the first session, we collected ̇VO2peak and baseline measurements. In the subsequent session, we measured running economy in different footwear conditions at either 75% (world class) or 70% (amateur) of the corre- sponding velocity to the measured ̇VO2peak, (v ̇VO2peak) [32]. We chose the 75/70% of v ̇VO2peak as this was a sub- maximal speed related to speeds these subjects would use when running at a marathon pace. To ensure consistency and avoid any confounding effects of circadian rhythm [33], we tested participants at the same time of day and encouraged them to match their diet, sleep, and training patterns prior to each session. Furthermore, to ensure the athletes felt comfortable being in a foreign Table 1 Participant descriptive and physiological characteristics for each of the measured cohorts Data shown are mean ± standard deviation ̇VO2peak maximal oxygen uptake, v ̇VO2peak velocity at ̇VO2peak, Student’s t test *Significance (p < 0.05) Variable World class Amateur p-value n = 7 n = 7 Age (years) 22.7 ± 3.2 28.1 ± 4.2 0.020* Height (cm) 174.3 ± 4.9 181.4 ± 2.6 0.008* Weight (kg) 59.9 ± 4.8 72.1 ± 7.0 0.003* ̇VO2peak (mL/kg/min) 75.9 ± 3.5 62.3 ± 5.1 < 0.001* ̇VO2peak (L/min) 4.53 ± 0.43 4.49 ± 0.48 0.870 v ̇VO2peak (km/h) 22.3 ± 0.6 18.8 ± 1.2 < 0.001* 1258 M. Knopp et al. environment and understood all that was asked of them, their coach as well as manager traveled with them and helped with testing. This favored a clearer communication between the research team and the athletes. 2.3.1 Visit 1 In this preliminary visit, we collected physiological baseline and anthropometric measurements. Throughout the whole experiment, all treadmill sessions were conducted in the same standardized laboratory chamber (mean ± standard deviation, temperature: 25.5 ± 1.1 °C, humidity: 60.2 ± 8.8%, pressure: 980.7 ± 4.9 mBar) on a HP Cosmos motorized treadmill (Venus 200/75; h/p/cosmos sports and medical GmbH, Nussdorf-Traunstein, Germany) set at a 1% gradient to mimic the energetic cost of running outdoors [34]. Given that some runners were not accustomed to treadmill running or using a ̇VO2peak protocol, we familiarized subjects dur- ing a 15-min session on the treadmill with increasing speeds. Once they felt comfortable running on a treadmill, we fitted each athlete with a heart rate monitor (Polar H7; Polar Elec- tro Oy, Kempele, Finland) and face mask (7450 Series V2 Mask; Hans Rudolph, Inc., Shawnee, KS, USA), connected to the MetaMax 3B portable cardiopulmonary gas exchange measuring device (CORTEX Biophysik GmbH, Leipzig, Germany). We then collected respiratory parameters from the subjects using an automated breath-by-breath method, via the measurement and evaluation software, MetaSoft Stu- dio (CORTEX Biophysik GmbH, Leipzig, Germany). Before each testing session, we calibrated this system according to the manufacturer’s instructions [35, 36]. To assess maximal aerobic capacity, athletes completed a ̇VO2peak ramp test using an incremental speed protocol with a continuous 1% incline. For this, athletes ran in the new pairs of the traditional racing FLAT test condition. For the world-class athletes, this test started at 10 km/h for 2 min and increased progressively at 1 km/h/min until volitional exhaustion. Amateurs completed the same protocol starting at 8 km/h. During this test, we verbally encouraged all ath- letes to ensure a maximal output was reached. Table 2 Descriptive characteristics of the AdvFootTech and FLAT NShoe characteristics based on size UK 8.5/US 9 Energy return classification: low: < 70%; medium: 70–80%; high: > 80% AdvFootTech advanced footwear technology, FLAT traditional racing flat Shoe label Mass (g) Forefoot stack height (mm) Rearfoot stack height (mm) Heel-to-toe drop (mm) Energy return (%) Stiff element? AdvFootTech 1 225 31.5 39 8.5 High Yes AdvFootTech 2 210 29.5 39.5 10 High Yes AdvFootTech 3 196 31 39.5 8.5 High Yes FLAT 197 19 24 5 Low No Fig. 2 Illustration of the methods protocol of the present study. A For visit 1, we collected baseline information of the subjects, which included conducting a maximal oxygen uptake ( ̇VO2peak) assess- ment. B On the second day of testing, we then assessed the run- ning economy of both traditional racing flat (FLAT) and different advanced footwear technology (AdvFootTech) models. v ̇VO2peak velocity at ̇VO2peak 1259 Variability in Advanced Footwear Technology Upon completion, two experienced exercise physiolo- gists detected and agreed upon ventilatory thresholds and ̇V O2peak values. For all cardiorespiratory data, we cleaned the breath-by-breath raw data by removing outlying data points that were more than two standard deviations away from the mean of a seven-breath window. After these outliers were removed, data were smoothed further by taking a moving seven-breath average. The ̇VO2max value was recorded as the highest cleaned and smoothed value during the test. As we did not repeat a verification test to confirm these val- ues, the highest recorded ̇VO2 value will be defined as a ‘ ̇VO2peak’ [37]. The measured v ̇VO2peak (km/h) was also recorded and used to prescribe the running speed for the run- ning economy tests during visit 2. Ventilatory threshold data as well as previously recorded personal bests of each athlete were used to ensure the selected speeds were sufficient in obtaining testing data that are relevant to racing and would not be affected by fatigue. 2.3.2 Visit 2 During visit 2, we assessed running economy for each of the different shoes at 75% of v ̇VO2peak (17.0 ± 0.4 km/h) for world-class athletes and 70% (13.1 ± 1.0 km/h) for ama- teur athletes. When subjects arrived, they first completed a 6-min standardized warm-up in the FLAT. This was then followed by a 12-min break during which we prepared the equipment for the test that consisted of 6-min bouts with a 12-min rest between bouts. Before each new treadmill trial, athletes changed their shoes for the next bout. The last 30 s of this break were recorded on the treadmill to obtain rest- ing values. From the recorded measurements, we calculated run- ning economy, oxygen cost of transport, and energetic cost using the Péronnet and Masicotte equation expressed in mL/kg/min, mL/kg/km, and W/kg, respectively, from the ̇VO2 data during the 60-s period from minute 4 to 5 of each test [38]. 2.4 Data and Statistical Analysis All data analysis and statistical tests were performed using RStudio [39]. Statistical analyses of the data were performed using the R package ‘stats’ (version 4.0.0) in RStudio [39, 40] using the traditional level of signifi- cance (p < 0.05). Power and sample size calculations were performed using the R package ‘pwr’ (version 1.3-0) in RStudio also using the traditional level of significance (p < 0.05), 80% power, and four different groups for the four different shoes. We conducted a Student’s t test on the descriptive characteristics to analyze population dif- ferences between the measured world-class and amateur athletes. Additionally, an analysis of variance test with repeated measures and a Bonferroni post-hoc correction were conducted on the steady-state physiological data [41, 42]. 2.5 Systematic Review and Meta‑analysis To confirm the found range of variability with the previ- ously published literature, and better understand the overall effect of advanced footwear technology, we conducted a sys- tematic electronic search of relevant studies and a related meta-analysis. For this retrospective systematic literature search, Sco- pus, SPORT-Discus, PubMed, Web of Science, and Foot- wear Science databases were searched using the terms “Racing Shoes” and “Running Shoes + Running Economy” through 21 November, 2021. Inclusion criteria for this review were studies that (1) examined the running perfor- mance effect of different versions of advanced footwear tech- nology for road running compared to a traditional racing flat control condition; and (2) measured the running economy (mL/kg/min) of this comparison. Additional secondary out- come measures including oxygen cost of transport (mL/kg/ km) and energetic cost (W/kg) were also analyzed to pro- vide a bigger picture of the effects of such new technology on running performance. These results were then pooled using Hedge’s g for a standardized effect size [43] and the inverse heterogeneity (IVhet) model using the Epigear Meta XL software (version 5.3) [44]. We further analyzed out- comes of the meta-analysis using a z-score for significance, Cochran’s Q statistic for heterogeneity, and I-squared for inconsistency [45] and assessed the risk of bias using the Cochrane Risk of Bias Instrument for RCTs (RoB 2) [46]. 3 Results 3.1 Running Economy From the available dataset (n = 14), for running economy there was a significant difference between shoe types in the amateur athletes (F(3) = 8.308, p = 0.001) where running economy in the advanced footwear technology was significantly lower than in the FLAT. Compared to the FLAT shoe, amateur athletes saw running economy improved by 3.5 ± 3.7% (pBonferroni = 0.042) with AdvFoot- Tech 1, 4.6 ± 2.7% (pBonferroni = 0.005) with AdvFootTech 2, and 5.0 ± 3.4% (pBonferroni = 0.002) with AdvFootTech 3 (Fig. 3B, Table 3), with no significant differences between the three advanced footwear technology conditions. Both the world-class and amateur athletes showed a large inter-individual variability with individual trials 1260 M. Knopp et al. showing a ± 11.4% variation in performance (Fig. 3). When examining the individual advanced footwear tech- nology conditions for the world-class population, the inter-individual range in overall performance changes of all included subjects vary by 14.6% on average for the dif- ferent shoes. A similar pattern is also seen in the amateur population where values here range from a 9.7% benefit to a 1.1% drawback for advanced footwear technology when compared to the flat for a narrower inter-individual total range of 10.8% (Fig. 3B). For this population, the individual advanced footwear technology range in perfor- mance changes was narrower than that of the world-class population for an average of a 9.5% difference between the maximum and minimum percent change per shoe. Via a time and running economy interaction analysis, we ensured the shoe order did not have a significant effect on the described results (world-class: p = 0.61; amateur: p = 0.67). In Table 3, we present the results for running economy, oxygen consumption, and percentage change in running economy in the advanced footwear technology models compared to a traditional running flat for both the world- class and amateur cohorts. Here, we compare the different shoes among cohorts, stratifying the data according to the amateur or world-class athlete results, as well as global effects comparing all tested subjects. 3.2 Systematic Review Study Characteristics From the initial search that resulted in 929 studies, 30 were selected for a full-text analysis after excluding by duplicates, title, and abstract, and five studies were finally included after fulfilling the inclusion criteria (Fig. 4). All examined stud- ies were randomized crossover trials investigating a range of recreational to highly trained runners with a combined average measured ̇VO2peak of 67.1 ± 8.2 mL/kg/min. All studies examined a steady-state running analysis on a tread- mill with different advanced footwear technology shoes compared to traditional racing flats, with Hébert-Losier et al. also including participants’ own shoes and spray painting the others to blind participants to model details [27]. Of the five studies, Barnes and Kilding was the only experiment to also include a female cohort [15]. Examined footwear conditions of the studies included in the meta-analysis are described in Table 4, please note data of shoe conditions irrelevant for this study, such as track spikes, were excluded in the meta-analysis [15]. When repeated conditions were used for the meta-analysis comparison, the corresponding conditions were divided by the number of repeated com- parisons to ensure no double counting of effects. The test- ing was conducted at a variety of different speeds either between 14 and 18 km/h or in the case of Hébert-Losier et al., at different speeds relative to ̇VO2peak [27]. Hereby, we decided to subgroup the analysis based on the speed at which physiological variables were measured according to the protocols. We included four different speed categoriza- tions starting with a very low speed that included 60% of v ̇V O2peak where the speed was 11.0 ± 0.6 km/h; the low speed category included those conditions measured at 14 km/h for both men and women or 70% of v ̇VO2peak with a speed of 12.9 ± 0.7 km/h; the medium-speed category included 16 km/h for men, 15 km/h for women, and 80% of v ̇VO2peak with a speed of 14.7 ± 0.8 km/h; finally, the high-speed cat- egory included 18 km/h for men, and 16 km/h for women. Considering the risk of bias assessment of the included studies, all studies had some concerns for the category of bias arising from period and carryover effects, given the unknown effect of the physiological starting point between the trials and what carryover or how long a car- ryover might be with regard to running in advanced foot- wear technology. The overall risk of bias across all stud- ies was of some concern owing to the similarities in the protocol of the study and the period and carryover effects. Fig. 3 Percentage change in steady-state running economy oxygen consumption (mL/kg/ min) relative to a traditional running flat (FLAT) in different shoe conditions for both A world-class and B amateur pop- ulations. These shoes include a FLAT on the far left as well as three different advanced foot- wear technology (AdvFootTech) conditions. Here, a negative percentage change indicates less oxygen consumption at a given speed and therefore a better run- ning economy 1261 Variability in Advanced Footwear Technology Table 3 Steady-state physiological results for each of the different AdvFootTech and FLAT models separated between the world-class and amateur cohorts as well as statistical findings of the whole combined sample AdvFootTech advanced footwear technology, ANOVA analysis of variance, FLAT traditional racing flat, SD standard deviation *Significant difference (p < 0.05) † Shoes with value significantly different to the FLAT Variable World class (mean ± SD) Among world-class subjects Amateur (mean ± SD) Among ama- teur subjects Combined sample n = 7 n = 7 Main effect shoes within subjects Main popula- tion effect between subjects Interaction effect within subjects FLAT AdvFoot- Tech 1 AdvFootTech 2 AdvFootTech 3 Repeated- measures ANOVA FLAT AdvFootTech 1 AdvFootTech 2 AdvFootTech 3 Repeated- measures ANOVA Running economy (mL O2/kg/ min) 54.5 ± 2.0 54.9 ± 1.6 54.7 ± 2.8 53.5 ± 3.1 F = 0.743 p = 0.541 47.7 ± 2.6 46.1 ± 3.2† pBonf = 0.043 45.5 ± 2.1† pBonf = 0.004 45.3 ± 1.9† pBonf = 0.002 F = 8.308 p = 0.001* F = 3.360 p = 0.030* F = 46.608 p < 0.001* F = 1.741 p = 0.177 Oxygen cost of transport (mL O2/kg/ km) 192.3 ± 8.1 193.8 ± 6.6 192.9 ± 11.8 188.7 ± 9.1 F = 0.875 p = 0.474 220.2 ± 12.3 212.3 ± 5.0† pBonf = 0.047 209.9 ± 8.8† pBonf = 0.006 208.9 ± 10.4† pBonf = 0.003 F = 7.511 p = 0.002* F = 4.245 p = 0.012* F = 20.757 p < 0.001* F = 2.478 p = 0.077 Energetic cost (W/kg) 19.4 ± 0.7 19.6 ± 0.6 19.5 ± 1.0 19.0 ± 1.3 F = 0.836 p = 0.493 16.9 ± 0.9 16.2 ± 1.2† pBonf = 0.018 16.0 ± 0.8† pBonf = 0.002 15.9 ± 0.7† pBonf = < 0.001 F = 10.007 p < .001* F = 3.572 p = 0.024* F = 47.887 p < 0.001* F = 1.886 p = 0.150 Respiratory exchange ratio 0.92 ± 0.02 0.93 ± 0.02 0.93 ± 0.02 0.90 ± 0.05 F = 1.001 p = 0.416 0.91 ± 0.03 0.88 ± 0.02† pBonf = 0.029 0.88 ± 0.03† pBonf = 0.016 0.88 ± 0.03† pBonf = 0.005 F = 6.518 p = 0.004* F = 2.741 p = 0.058 F = 4.935 p = 0.048* F = 1.663 p = 0.193 Heart rate (bpm) 158.4 ± 8.8 157.7 ± 8.5 157.3 ± 10.1 155.6 ± 11.2 F = 0.919 p = 0.453 160.3 ± 5.9 157.2 ± 7.2 160.1 ± 6.5 158.8 ± 7.5 F = 1.527 p = 0.242 F = 1.542 p = 0.221 F = 0.278 p = 0.609 F = 1.072 p = 0.373 % Change in running economy to traditional running FLAT 0.0 ± 0.0 0.8 ± 5.0 0.3 ± 3.9 − 1.9 ± 5.6 F = 0.74 p = 0.543 0.0 ± 0.0 − 3.5 ± 3.7† pBonf = 0.042 − 4.6 ± 2.7† pBonf = 0.005 − 5.0 ± 3.4† pBonf = 0.002 F = 7.969 p = 0.001* F = 3.579 p = 0.023* F = 4.170 p = 0.066 F = 2.039 p = 0.126 1262 M. Knopp et al. 3.3 Meta‑analysis Primary Outcome Measure: Running Economy The meta-analysis of running economy (mL/kg/min) in all five examined studies comparing different advanced foot- wear technology to racing flat conditions revealed a statisti- cally significant benefit of advanced footwear technology on running economy measures with an overall medium effect of − 0.58 [mean (95% confidence interval); g = − 0.58 (− 0.75, − 0.42), Z = − 6.86 (p < 0.001)], where a negative value indicates improved efficiency when running (Fig. 5). When sub-grouped by speed, the analysis showed a small effect [g = − 0.29 (− 0.87, 0.31)] at very low speeds, a medium effect [g = − 0.58 (− 0.90, − 0.26)] at low speeds, a medium effect [g = − 0.54 (− 0.79, − 0.28)] at medium speeds, and a large effect [g = − 0.92 (− 1.31, − 0.52)] at high speeds. Incorporating the data presented in this study, results are Fig. 4 Flow chart showing study selection. Adapted from the PRISMA flow diagram [60] Table 4 Descriptive characteristics of shoe products included in the meta-analysis Shoe characteristics based on size UK 8.5/US 9 and obtained from original journal articles used in the meta-analysis or measurements conducted from RunningWarehouse.com. FLAT 6 varies (mean ± standard deviation) as it is a combination of the participants own footwear and includes sizes varying from US 8.5 to 12. Missing information (n/a) is because of the confidentiality of midsole material or missing information in the examined studies AdvFootTech advanced footwear technology, EVA ethylene–vinyl acetate, FLAT traditional racing flat, n/a not available, PEBA polyether block amide, TPU thermoplastic polyurethane Shoe label Mass (g) Forefoot stack height (mm) Rearfoot stack height (mm) Heel-to-toe drop (mm) Midsole material Stiff element? AdvFootTech 1 225 31.5 39 8.5 n/a Yes AdvFootTech 2 210 29.5 39.5 10 n/a Yes AdvFootTech 3 196 31 39.5 8.5 n/a Yes AdvFootTech 4 [15, 27–29] 195 21 31 10 PEBA Yes AdvFootTech 5 [30] 196 32 40 8 PEBA Yes AdvFootTech 6 [30] 210 27 35 8 n/a Yes AdvFootTech 7 [30] 207 24 34 10 TPU Yes AdvFootTech 8 [30] 213 30 35 5 EVA Yes AdvFootTech 9 [30] 207 33 38 5 n/a Yes AdvFootTech 10 [30] 213 31 39 8 PEBA Yes AdvFootTech 11 [30] 210 36 40 4 PEBA Yes FLAT 197 19 24 5 TPU No FLAT 2 [28, 29] 181 15 23 8 EVA No FLAT 3 [28] 221 13 23 10 TPU No FLAT 4 [15, 29] 224 13 23 10 TPU No FLAT 5 [27] 130 13 13 1 TPU No FLAT 6 [27] 313 ± 44 n/a 26.0 ± 7.9 9.4 ± 6.7 Varies No FLAT 7 [30] 210 21 30 9 EVA No 1263 Variability in Advanced Footwear Technology showing an overall medium effect [g = − 0.39 (− 1.01, 0.23)]. When this sub-analysis is further distributed by population, the world-class subgroup showed a small effect [g = − 0.02 (− 0.88, 0.85)], and the amateur subgroup showed a large effect [g = − 0.80 (− 1.70, 0.10)]. In this analysis, no statisti- cally significant heterogeneity, as assessed via Q, was found (Q = 14.42, p = 1.00) and inconsistency, as assessed using I2 as an extension of Q, was very low (I2 = 0%) [45]. 3.4 Meta‑analysis Secondary Outcome Measures: Oxygen Cost of Transport and Energetic Cost The meta-analysis of oxygen cost of transport (mL/kg/km) of the three studies that included this data revealed a statis- tically significant benefit of advanced footwear technology on the oxygen cost of transport measures [mean (95% CI); g = − 0.67 (− 0.87, − 0.47), Z = − 6.60 (p = < 0.001), Fig. 6]. Considering the subgroup analysis by speed, a medium effect [g = − 0.58 (− 0.96, − 0.20)] was found at low speeds, a medium effect [g = − 0.62 (− 0.95, − 0.30)] at medium speeds, and a large effect [g = − 0.92 (− 1.31, − 0.52)] at high speeds. Incorporating the data presented in this study, an overall medium effect [g = − 0.47 (− 1.10, 0.16)] was found. Here as well, no statistically significant heterogeneity was found (Q = 14.03, p = 0.99) and inconsistency was very low (I2 = 0%) among the examined studies [45]. Finally, the meta-analysis of energetic cost (W/kg) of the four studies showed a statistically significant benefit of advanced footwear technology on energetic cost measures [mean (95% CI); g = − 0.54 (− 0.71, − 0.37), Z = − 6.28 (p = < 0.001), Fig. 7]. Further examination of the subgroup speed analysis shows a small effect [g = − 0.27 (− 0.86, 0.31)] at very low speeds, a medium effect [g = − 0.53 (− 0.85, − 0.21)] at low speeds, a medium effect [g = − 0.55 (− 0.82, − 0.27)] at medium speeds, and a large effect [g = − 0.69 (− 1.07, − 0.31)] at high speeds. Analysis of the present study shows an overall medium effect [g = − 0.41 (− 1.04, 0.21)]. Again, here, no statistically significant het- erogeneity was found (Q = 8.44, p = 1.00) and inconsistency was very low (I2 = 0%) between the subgroups [45]. 4 Discussion In this study, we aimed to assess the variability in running economy in advanced footwear technology compared to a traditional racing flat on a treadmill in world-class Kenyan versus European amateur runners at speeds proportional to a marathon pace. Our laboratory results revealed ± 11.4% variability of the running economy of different advanced footwear technology running shoes in world-class Kenyan road runners, while for amateur Europeans, results range from a 9.7% benefit to a 1.1% drawback. The post-hoc meta-analysis revealed an overall statistically significant medium benefit of advanced footwear technology on run- ning economy when compared with traditional flats. 4.1 Running Economy and Running Performance Inter‑Individual Variability The running economy of the measured advanced footwear technology compared to a traditional racing flat of all tested subjects revealed a large inter-subject variability with overall values that ranged from an 11.4% benefit to an 11.3% draw- back (Fig. 3). To compare this variation of running economy to other studies, we conducted a systematic literature search. Interestingly, this revealed similar variability in the found research considering the obtained confidence intervals in the conducted meta-analysis (Figs. 5, 6, 7). Hoogkamer et al. examined for the first time advanced footwear technology versus previously established marathon racing flats, all mass neutralized, in high-caliber athletes at three distinct speeds. The results found a range of 1.97–6.26% benefit in energetic cost (W/kg) of the new advanced footwear technology versus flats [28]. A similar study conducted by Barnes and Kild- ing showed a 1.72–7.15% running economy benefit (mL/ kg/min) in highly trained runners in favor of the advanced footwear technology with only trivial-to-small differences between the tested men and women [15]. On average, this study found a 4.2% running economy benefit of advanced footwear technology versus the flat, which decreased to 2.9% when these conditions were weight matched, indicat- ing the effect weight might have on such testing [15]. In an additional study, Hunter et al. found a response range of a 0.0–6.4% improvement in running economy (mL/kg/min) for advanced footwear technology and further suggested that different runners may require individualized shoe stiffnesses to enhance performance [29]. Hébert-Losier et al. examined both running economy and performance during a 3-km time trial and found a variability in running economy (mL/kg/ min) of a worsening by a 10.3–13.3% improvement across conditions in recreational runners, and a time trial variability of a worsening by a 4.7–9.3% improvement [27]. To com- pare seven different models of advanced footwear technol- ogy, Joubert et al. conducted running economy tests (mL/ kg/min) with trained distance runners and found that when all advanced footwear technology shoes are combined, the responses, as calculated from presented mean and stand- ard deviations as well as described values, ranged from a 1% disadvantage to a 5.3% advantage [30]. An additional group of research studies also conducted a similar analysis by examining race performance measures instead of physi- ological data obtained in a laboratory. Considering these as well, Guinness et al. examined marathon race performance results from hundreds of elite marathoners who switched to advanced footwear technology and found that 74.5% of 1264 M. Knopp et al. g 1 0 -1 -2 -3 Study or Subgroup Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 18, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 18, Men Present Study, AdvFootTech 3 vs FLAT, 70%, Amateur Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 16, Men High Speed subgroup Present Study, AdvFootTech 2 vs FLAT, 70%, Amateur Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 16, Women Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 14, Men Joubert et al., 2021 - AdvFootTech 11 vs FLAT 7, 16, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 16, Men Joubert et al., 2021 - AdvFootTech 5 vs FLAT 7, 16, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 2, 18, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 2, 14, Men Joubert et al., 2021 - AdvFootTech 9 vs FLAT 7, 16, Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 14, Women Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 2, 16, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 3, 18, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 3, 14, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 3, 16, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 16, Women Very Low Speed Q=0.36, p=0.55, I2=0% Low Speed Q=1.33, p=0.99, I2=0% Medium Speed Q=2.76, p=1.00, I2=0% High Speed Q=3.68, p=0.60, I2=0% Present Study Q=2.04, p=0.84, I2=0% Overall Q=14.42, p=1.00, I2=0% Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 14, Men Low Speed subgroup Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 15, Women Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 6, 70%, Men Medium Speed subgroup Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 6, 80%, Men Hunter et al., 2019 - AdvFootTech 4 vs FLAT 4, 16, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 14, Women Present Study, AdvFootTech 1 vs FLAT, 70%, Amateur Men Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 6, 60%, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 15, Women Joubert et al., 2021 - AdvFootTech 10 vs FLAT 7, 16, Men Present Study subgroup Joubert et al., 2021 - AdvFootTech 7 vs FLAT 7, 16, Men Hunter et al., 2019 - AdvFootTech 4 vs FLAT 2, 16, Men Present Study, AdvFootTech 3 vs FLAT, 75%, World-Class Men Very Low Speed subgroup Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 5, 80%, Men Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 5, 70%, Men Joubert et al., 2021 - AdvFootTech 6 vs FLAT 7, 16, Men Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 5, 60%, Men Joubert et al., 2021 - AdvFootTech 8 vs FLAT 7, 16, Men Present Study, AdvFootTech 2 vs FLAT, 75%, World-Class Men Present Study, AdvFootTech 1 vs FLAT, 75%, World-Class Men g (95% CI) % Weight -1.88 ( -3.08, -0.68) 1.9 -1.27 ( -2.35, -0.18) 2.3 -1.08 ( -2.69, 0.54) 1.1 -1.01 ( -2.06, 0.04) 2.5 -0.92 ( -1.31, -0.52) 17.6 -0.89 ( -2.46, 0.68) 1.1 -0.86 ( -1.89, 0.17) 2.6 -0.82 ( -1.85, 0.20) 2.6 -0.79 ( -2.43, 0.85) 1.0 -0.76 ( -1.78, 0.26) 2.6 -0.76 ( -2.40, 0.87) 1.0 -0.74 ( -1.57, 0.09) 4.0 -0.72 ( -1.55, 0.11) 4.0 -0.71 ( -2.34, 0.92) 1.0 -0.71 ( -1.73, 0.30) 2.7 -0.68 ( -1.50, 0.14) 4.0 -0.68 ( -1.50, 0.15) 4.1 -0.67 ( -1.49, 0.16) 4.1 -0.64 ( -1.46, 0.18) 4.1 -0.63 ( -1.64, 0.38) 2.7 -0.58 ( -0.75, -0.42) 100.0 -0.58 ( -1.59, 0.42) 2.7 -0.58 ( -0.90, -0.26) 26.3 -0.56 ( -1.56, 0.44) 2.7 -0.54 ( -1.40, 0.33) 3.7 -0.54 ( -0.79, -0.28) 41.0 -0.53 ( -1.49, 0.43) 3.0 -0.52 ( -1.31, 0.27) 4.4 -0.48 ( -1.48, 0.52) 2.8 -0.47 ( -1.98, 1.04) 1.2 -0.47 ( -1.30, 0.37) 3.9 -0.40 ( -1.39, 0.59) 2.8 -0.40 ( -2.01, 1.21) 1.1 -0.39 ( -1.01, 0.23) 7.0 -0.37 ( -1.97, 1.24) 1.1 -0.36 ( -1.15, 0.42) 4.5 -0.30 ( -1.83, 1.22) 1.2 -0.29 ( -0.87, 0.30) 8.0 -0.21 ( -1.16, 0.73) 3.1 -0.18 ( -1.03, 0.67) 3.8 -0.16 ( -1.76, 1.45) 1.1 -0.11 ( -0.93, 0.71) 4.0 -0.02 ( -1.62, 1.58) 1.1 0.04 ( -1.44, 1.52) 1.3 0.20 ( -1.29, 1.68) 1.2 1265 Variability in Advanced Footwear Technology the men ran faster with an estimate of a 1.4–2.8% improve- ment in performance, while 71.4% of the women ran faster with an estimate of a 0.6–2.2% performance improvement [47]. Similarly, Senefeld et al. further examined performance and racing shoes in elite racers in four major marathons and found that in a subgroup of athletes with subsequent race performance of a flat then advanced footwear technology, the between-race change in performance for female athletes had a 95% confidence interval range from a 6.9% hindrance to a 13.8% advantage and a 5.4% hindrance to an 11.4% advantage in male athletes, suggesting that observed find- ings in a laboratory setting translate to real improvements in racing conditions [48]. Finally, Bermon et al. analyzed sea- sonal best times throughout the years to determine the effect of switching to advanced footwear technology, and found that in half-marathon and marathon races of a subgroup of athletes who competed in the same event with and without these shoes, all athletes (except male half-marathon runners) significantly improved their performance times with calcula- tions on presented data showing that on average the female athletes showed a greater benefit of 1.9% faster in both races when compared with a 0.8% better performance found in the male athletes [49]. Overall, comparable to the present study, the variability in previously published data range from a 13.8% benefit to a 10.3% drawback in an overall change in performance of advanced footwear technology versus traditional racing flats as measured both in the laboratory with steady-state running physiology tests, and in the field examining race times. Additional results from the five studies included after a retrospective systematic review and meta-analysis revealed that advanced footwear technology had an overall significant medium effect of − 0.58 when compared with a flat in terms of running economy, oxygen cost of transport, and energetic cost, even when accounting for the large individual vari- ability found in these individual studies [15, 27–30]. Inter- estingly, as revealed via the subgroup analysis, the effect changed with the speed sub-groups where very low speeds showed a small effect and high speeds showed a greater effect, aligning with what has previously been shown in the literature [50]. This suggests that mechanisms involved in the advanced footwear technology might be proportional to the other biomechanical aspects such as changes in stride or gait cycle that alter with speed, with the mechanism reduc- ing the energy required for running bouts proportionally higher when running at higher speeds [51]. Despite the findings of the meta-analysis, it remains important to consider the great inter-individual differences in the response to footwear conditions with individuals in the presented study as well as subjects in previous research showing significant inter-individual differences. Such results suggest possible methodological limitations of measuring the performance of running shoes (e.g., laboratory-based studies, insufficient familiarization protocols), as well as the importance of an individualized approach for athletes considering different biomechanical or anthropometrics that could be contributing to optimize their response to advanced footwear technology. 4.2 Intra‑Individual Running Economy Differences in Shoe Conditions When examining the individual cases, some subjects showed meaningful effects depending on the specific advanced footwear technology shoe being tested, and others were not always trending the same way among all advanced footwear technology models. For example, given the results here, one of the world-class Kenyan runners showed a range from an 11.4% to a 0.2% benefit in the different advanced footwear technology models (Fig. 3A). For the aforementioned ath- lete, comparing personal best half-marathon times, this individual did indeed improve a sub-1-h half-marathon time by over 1:20 (min:s) in a shoe where this athlete was more economical during testing [52]. However, for another world- class subject who exhibited a running economy range of a 2.5% benefit to a 6.6% drawback for different advanced footwear technology, comparing marathon seasonal best times, this athlete was able to set a new personal record by reducing 2 min off a time already under 2:10 (h:min) in shoes that they, according to our test, should have performed worse in. This further affirms possible limitations of testing shoe performance in this way, particularly with a world-class Kenyan running population where further confounders such as a lack of familiarization to treadmill running and testing conditions might be playing a role. 4.3 Populations Running Economy Differences When examining in our study the differences in variability ranges between the world-class (an 11.4% benefit to a 11.3% drawback) and the amateur (a 9.7% benefit to a 1.1% draw- back) populations, further exploration into the data revealed possible explanations. As we did not measure the running economy of all participants at the same speed, we are unable to conclude how the running efficiency of these two popula- tions compared as a baseline in the same traditional racing flat. However, previously published research established that East Africans have a running economy advantage when compared with their Spanish counterparts [12]. Therefore, Fig. 5 Forest plot displaying running economy (mL/kg/min) com- parisons between advanced footwear technology (AdvFootTech) and traditional racing flats (FLAT) sub-categorized into different speeds. Study labels consist of the study name, the examined AdvFootTech versus FLAT condition where + indicates conditions that are weight matched, the speed either in km/h or as a % of peak, and the exam- ined population. CI confidence interval ◂ 1266 M. Knopp et al. g 1 0 -1 -2 -3 Study or Subgroup Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 18, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 18, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 3, 14, Men Present Study, AdvFootTech 1 vs FLAT, 70%, Amateur Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 16, Men Present Study, AdvFootTech 2 vs FLAT, 70%, Amateur Men Present Study, AdvFootTech 3 vs FLAT, 70%, Amateur Men High Speed subgroup Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 16, Women Joubert et al., 2021 - AdvFootTech 11 vs FLAT 7, 16, Men Joubert et al., 2021 - AdvFootTech 5 vs FLAT 7, 16, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 16, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 2, 18, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 2, 14, Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 14, Women Joubert et al., 2021 - AdvFootTech 9 vs FLAT 7, 16, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 2, 16, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 3, 18, Men Low Speed Q=4.01, p=0.55, I2=0% Medium Speed Q=1.79, p=1.00, I2=0% High Speed Q=3.67, p=0.60, I2=0% Present Study Q=2.42, p=0.79, I2=0% Overall Q=14.03, p=0.99, I2=0% Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 3, 16, Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 15, Women Medium Speed subgroup Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 16, Women Low Speed subgroup Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 14, Women Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 15, Women Present Study subgroup Joubert et al., 2021 - AdvFootTech 10 vs FLAT 7, 16, Men Joubert et al., 2021 - AdvFootTech 7 vs FLAT 7, 16, Men Present Study, AdvFootTech 3 vs FLAT, 75%, World-Class Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 14, Men Joubert et al., 2021 - AdvFootTech 6 vs FLAT 7, 16, Men Joubert et al., 2021 - AdvFootTech 8 vs FLAT 7, 16, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 14, Men Present Study, AdvFootTech 2 vs FLAT, 75%, World-Class Men Present Study, AdvFootTech 1 vs FLAT, 75%, World-Class Men g (95% CI) % Weight -1.86 ( -3.06, -0.66) 2.8 -1.27 ( -2.36, -0.18) 3.4 -1.17 ( -2.04, -0.30) 5.2 -1.02 ( -2.62, 0.58) 1.5 -1.01 ( -2.05, 0.04) 3.6 -0.97 ( -2.56, 0.62) 1.6 -0.94 ( -2.52, 0.65) 1.6 -0.92 ( -1.31, -0.52) 25.4 -0.86 ( -1.90, 0.17) 3.7 -0.78 ( -2.42, 0.85) 1.5 -0.75 ( -2.39, 0.88) 1.5 -0.75 ( -1.77, 0.27) 3.8 -0.73 ( -1.56, 0.10) 5.8 -0.72 ( -1.54, 0.11) 5.8 -0.72 ( -1.73, 0.30) 3.8 -0.71 ( -2.34, 0.92) 1.5 -0.68 ( -1.51, 0.14) 5.8 -0.68 ( -1.50, 0.15) 5.8 -0.67 ( -0.87, -0.47) 100.0 -0.66 ( -1.48, 0.16) 5.9 -0.64 ( -1.65, 0.37) 3.9 -0.62 ( -0.95, -0.30) 37.6 -0.62 ( -1.62, 0.39) 3.9 -0.58 ( -0.96, -0.20) 27.1 -0.48 ( -1.48, 0.52) 4.0 -0.48 ( -1.47, 0.52) 4.0 -0.47 ( -1.10, 0.16) 10.0 -0.40 ( -2.01, 1.21) 1.5 -0.36 ( -1.97, 1.25) 1.5 -0.35 ( -1.89, 1.18) 1.7 -0.23 ( -1.22, 0.75) 4.1 -0.16 ( -1.76, 1.45) 1.5 -0.02 ( -1.62, 1.58) 1.5 0.04 ( -0.94, 1.02) 4.1 0.05 ( -1.43, 1.53) 1.8 0.19 ( -1.29, 1.68) 1.8 Fig. 6 Forest plot displaying oxygen cost of transport (mL/kg/km) comparisons between advanced footwear technology (AdvFootTech) and traditional racing flats (FLAT) sub-categorized into different speeds. Study labels consist of the study name, the examined Adv- FootTech versus FLAT condition where + indicates conditions that are weight matched, the speed either in km/h or as a % of peak, and the examined population. CI confidence interval 1267 Variability in Advanced Footwear Technology one consideration could be that our world-class cohort was already more economical when running in the traditional racing flat and therefore would not benefit as much when compared to the amateur European population. Additionally, regarding the methodology, certain dif- ferences between the two populations are also apparent. First, while the relative effort between populations might be comparable, the speed at which they attained such effort differed with the average submaximal velocity for the world-class runners being 17.1 ± 0.4 km/h compared with 13.1 ± 1.0 km/h of the amateurs. These differences could be affecting the percentage benefits of advanced footwear technology in regard to running economy [53]. Moreover, even with a brief warm-up and familiarization session, some world-class runners were not used to running on a treadmill, which as Colino et al. has suggested, changes the mechanics compared with overground running [54, 55]. Furthermore, of note, at the point of testing, the world-class population had already been training in a version of the advanced foot- wear technology and were therefore familiar with the high- stack height and the feel of running with this technology. In contrast, the amateurs were not regularly running in such shoes outside of the present study. Previous research con- ducted has suggested injury risks and possible biomechani- cal changes when transitioning to novel footwear (e.g., mini- malist shoes) too quickly, recommending a longer adaptation period [56–58]. Both considerations could have biased the results of the present study. 4.4 Limitations Several limitations to this study must also be acknowledged. First, we acknowledge the present study is underpowered. As no previous study had been conducted examining a world- class cohort, we had to do power and sample size calcu- lations post-hoc. To start with the amateur cohort, using the smallest found effect size of 0.47 for running economy, sample size calculations revealed that 14 participants should be considered for such an analysis, consistent with the 14 total participants we had recruited at the start of the experi- ment. Using this same effect size for the amateur cohort, calculations revealed a power of 46.2%. When considering each cohort separately, as with most other studies examin- ing sub-elite populations, we were able to see differences in advanced footwear technology for the amateurs. For the world-class cohort, the effect sizes for running economy of advanced footwear technology shoes compared to the flat varied from 0.04 to − 0.30. Considering this range in effect size, the power calculation here revealed a 5.2% up to a 20.4%. As this signifies our study as being underpow- ered, we also calculated the necessary sample size that would be needed for the world-class cohort to achieve the desired power of 80%. Based on which effect size, results here revealed 32–1705 participants would be needed, which is a challenge to maintain the high level required in such a large group of participants. This is a common issue that stud- ies using world-class athletes are often underpowered given the singularity and inaccessibility to this sample, resulting rather in case studies or studies with a limited sample size [59]. With the world-class athletes, we must also consider the margin of the examined population, where even a mini- mal improvement in efficiency can reduce the finishing time over the duration of a marathon and could be the difference between a podium place or not. Furthermore, the results reflect that we must consider the large inter-subject vari- ability and therefore the individuality of the athletes. The question remains of how to detect the marginal changes in an elite population. To further examine this, future studies should also consider examining the test–retest reliability of steady-state running economy laboratory assessments con- ducted on world-class athletes. Additional limitations must also be considered owing to the athletes’ schedules and availability. More time would have also allowed us to repeat testing measures with the athletes, which would have ensured further reliability of the testing. An additional limitation was that no female athletes were tested within the scope of this study as we only had access to male athletes. Previous results considering both sexes range from only trivial to small differences in labora- tory testing to significant differences in performance finish- ing times for female athletes [15, 48, 49]. Furthermore, it is important to note that because the intention was to test with shoes readily available on the market, it was impossible to blind the participants as to the shoe they were testing. As mentioned, because some athletes were already familiar with and training in versions of these shoes, athletes may have had pre-established opinions that could have influenced the results and the placebo effect cannot be excluded [29]. It must be noted, however, that related research comparing the running economy of different shoes where subjects were blinded to the shoes that were painted in black still revealed similar results [27]. Limitations related to the systematic review and meta- analysis include methodological and characterization varia- tions. For example, some studies manipulated the shoe con- ditions in terms of weight matching or spray painting for blinding. Additionally, the ambiguity in subject definition related to the caliber of runners makes it difficult to place the results according to populations. Finally, with respect to the described shoe conditions, the specific model or version of a shoe within a franchise was not always clearly labeled, thus we had to make an informed categorization based on the information available. 1268 M. Knopp et al. g 1.8 0.9 0 -0.9 -1.8 -2.7 Study or Subgroup Present Study, AdvFootTech 3 vs FLAT, 70%, Amateur Men Present Study, AdvFootTech 2 vs FLAT, 70%, Amateur Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 16, Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 14, Men Joubert et al., 2021 - AdvFootTech 11 vs FLAT 7, 16, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 16, Women Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 2, 18, Men Joubert et al., 2021 - AdvFootTech 5 vs FLAT 7, 16, Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 16, Women Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 2, 14, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 3, 18, Men Joubert et al., 2021 - AdvFootTech 9 vs FLAT 7, 16, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 16, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 2, 16, Men High Speed subgroup Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 3, 14, Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 18, Men Hoogkamer et al., 2018 - AdvFootTech 4+ vs FLAT 3, 16, Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 14, Women Medium Speed subgroup Very Low Speed Q=0.43, p=0.51, I2=0% Low Speed Q=1.46, p=0.98, I2=0% Medium Speed Q=2.41, p=1.00, I2=0% High Speed Q=0.25, p=1.00, I2=0% Present Study Q=2.33, p=0.80, I2=0% Overall Q=8.44, p=1.00, I2=0% Present Study, AdvFootTech 1 vs FLAT, 70%, Amateur Men Low Speed subgroup Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 6, 80%, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 14, Men Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 6, 70%, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 18, Men Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 6, 60%, Men Barnes et al., 2019 - AdvFootTech 4 vs FLAT 4, 15, Women Present Study subgroup Joubert et al., 2021 - AdvFootTech 10 vs FLAT 7, 16, Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 15, Women Joubert et al., 2021 - AdvFootTech 7 vs FLAT 7, 16, Men Present Study, AdvFootTech 3 vs FLAT, 75%, World-Class Men Barnes et al., 2019 - AdvFootTech 4+ vs FLAT 4, 14, Women Very Low Speed subgroup Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 5, 80%, Men Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 5, 70%, Men Joubert et al., 2021 - AdvFootTech 6 vs FLAT 7, 16, Men Hebert-Losier et al., 2020 - AdvFootTech 4 vs FLAT 5, 60%, Men Joubert et al., 2021 - AdvFootTech 8 vs FLAT 7, 16, Men Present Study, AdvFootTech 2 vs FLAT, 75%, World-Class Men Present Study, AdvFootTech 1 vs FLAT, 75%, World-Class Men g (95% CI) % Weight -1.14 ( -2.77, 0.49) 1.1 -0.94 ( -2.53, 0.64) 1.1 -0.88 ( -1.91, 0.15) 2.7 -0.80 ( -1.83, 0.22) 2.7 -0.80 ( -2.44, 0.84) 1.1 -0.78 ( -1.80, 0.24) 2.7 -0.76 ( -1.59, 0.07) 4.1 -0.75 ( -2.39, 0.88) 1.1 -0.73 ( -1.75, 0.28) 2.8 -0.73 ( -1.55, 0.10) 4.2 -0.70 ( -1.53, 0.12) 4.2 -0.70 ( -2.33, 0.93) 1.1 -0.69 ( -1.70, 0.32) 2.8 -0.69 ( -1.37, -0.02) 6.2 -0.69 ( -1.07, -0.31) 19.4 -0.67 ( -1.49, 0.16) 4.2 -0.66 ( -1.67, 0.35) 2.8 -0.65 ( -1.32, 0.02) 6.3 -0.55 ( -1.55, 0.45) 2.8 -0.55 ( -0.82, -0.27) 37.6 -0.54 ( -0.71, -0.37) 100.0 -0.53 ( -2.05, 0.98) 1.2 -0.53 ( -0.85, -0.21) 27.4 -0.51 ( -1.47, 0.45) 3.1 -0.51 ( -1.51, 0.49) 2.9 -0.48 ( -1.34, 0.38) 3.8 -0.47 ( -1.47, 0.52) 2.9 -0.47 ( -1.31, 0.36) 4.1 -0.46 ( -1.45, 0.54) 2.9 -0.41 ( -1.04, 0.21) 7.2 -0.39 ( -2.00, 1.22) 1.1 -0.34 ( -1.33, 0.65) 2.9 -0.34 ( -1.95, 1.27) 1.1 -0.33 ( -1.86, 1.20) 1.2 -0.33 ( -1.32, 0.66) 2.9 -0.27 ( -0.86, 0.31) 8.3 -0.19 ( -1.14, 0.75) 3.2 -0.17 ( -1.02, 0.68) 3.9 -0.15 ( -1.76, 1.45) 1.1 -0.08 ( -0.90, 0.75) 4.2 -0.02 ( -1.62, 1.58) 1.1 0.07 ( -1.41, 1.55) 1.3 0.22 ( -1.27, 1.71) 1.3 1269 Variability in Advanced Footwear Technology 5 Conclusions Next-generation long-distance running shoes that contain advanced footwear technology result in large inter- and intra- subject variability when measured for changes in running economy in both world-class Kenyan and amateur Euro- pean runners with overall values that range from an 11.3% hindrance to an 11.4% benefit. Similar variability was also found in the literature as measured both in the laboratory and with real race performance. Additionally, meta-analy- sis results reveal an overall significant medium benefit of advanced footwear technology on running economy when compared with traditional flats. Such results have impor- tant indications. First of all, while testing the performance of shoes with running economy tests has become standard practice, further research should consider other methods that ensure ecological validity, which could include repeated economy tests or field-based tests. Furthermore, perfor- mance testing should be standardized to get a better com- parison between studies. This is particularly important for the world-class athletes where additional constraints could be affecting their results as well as the acknowledgment that they may already have a better running economy. Second, this study acknowledges that a more personalized approach is necessary and that, when confirmed with additional test- ing, the inter- as well as intra-subject variability should be considered by stakeholders involved in elite sport. First, among others, it could affect athletes and coaches regarding their shoe selection; sport associations should acknowledge the importance of individualization in sport; shoe manufac- turers should consider this when implementing new technol- ogy; and governing bodies should consider what impact this might have on the sport, with regard to which magnitude of effect is acceptable and fair. Acknowledgements This study was conducted at the adidas Sports Sci- ence Research Laboratory in Herzogenaurach, Germany. The authors acknowledge the runners for their voluntary participation in this study and the coaches for allowing the athletes to participate. Additional thanks go to members of the adidas research team including Harry Miles, Julian Fritz, Alejandro Alcañiz, Mario Fleiter, Tobias Luckfiel, and Heiko Schlarb and athlete servicing team including Fabian Sch- weizer who all supported this research. Declarations Funding This study was supported by adidas AG. Open Access fund- ing enabled and organized by Projekt DEAL. Conflict of interest MK and DR are both employees of adidas AG. YP is the founding member of the Sub2 marathon project (http:// www. sub2h rs. com). BM-P, FG, HW, and MS have no conflicts of interest that are directly relevant to the content of this article. Ethics approval This experiment was submitted to the Technical Uni- versity of Munich Ethics Committee, who advised that formal approval was not required. This study was conducted in accordance with the ethical standards of the Declaration of Helsinki. Consent to participate All participants gave written informed consent to being a part of this study after they were informed of and under- stood the experimental procedures, potential injury risks, and possible benefits. Consent for publication Not applicable. Data availability Considering the inherent characteristics of this research, the participants of this study did not agree to publicly share the obtained individual data. Code availability Not applicable. Author contributions MK and DR conceived and designed the research. MK and DR performed and supported the experiments with the help of additional colleagues. MK, DR, BM-P, HW, MS, FG, and YP analyzed the data. MK and FG conducted the statistical analysis. MK, DR, BM-P, HW, MS, and YP interpreted the results of the experi- ment. MK drafted the manuscript. DR, BM-P, HW, MS, FG, and YP edited and revised the manuscript. Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. References 1. Mooses M, Mooses K, Haile DW, Durussel J, Kaasik P, Pitsi- ladis YP. Dissociation between running economy and running performance in elite Kenyan distance runners. J Sports Sci. 2015;33(2):136–44. https:// doi. org/ 10. 1080/ 02640 414. 2014. 926384. 2. Tucker R, Onywera VO, Santos-Concejero J. Analysis of the Ken- yan distance-running phenomenon. Int J Sports Physiol Perform. 2015;10(3):285–91. https:// doi. org/ 10. 1123/ ijspp. 2014- 0247. 3. Hamilton B. East African running dominance: what is behind it? BJSM. 2000;34(5):391–4. https:// doi. org/ 10. 1136/ bjsm. 34.5. 391. 4. Tawa N, Louw Q. Biomechanical factors associated with run- ning economy and performance of elite Kenyan distance run- ners: a systematic review. J Bodyw Mov Ther. 2018;22(1):1–10. https:// doi. org/ 10. 1016/j. jbmt. 2017. 11. 004. 5. Saltin B, Kim CK, Terrados N, Larsen H, Svedenhag J, Rolf CJ. Morphology, enzyme activities and buffer capacity in leg Fig. 7 Forest plot displaying energetic cost (W/kg) comparisons between advanced footwear technology (AdvFootTech) and tradi- tional racing flats (FLAT) sub-categorized into different speeds. Study labels consist of the study name, the examined AdvFootTech versus FLAT condition where + indicates conditions that are weight matched, the speed either in km/h or as a % of peak, and the exam- ined population. CI confidence interval ◂ 1270 M. Knopp et al. muscles of Kenyan and Scandinavian runners. Scand J Med Sci Sports. 1995;5(4):222–30. https:// doi. org/ 10. 1111/j. 1600- 0838. 1995. tb000 38.x. 6. Saltin B, Larsen H, Terrados N, Bangsbo J, Bak T, Kim CK, et al. Aerobic exercise capacity at sea level and at altitude in Kenyan boys, junior and senior runners compared with Scan- dinavian runners. Scand J Med Sci Sports. 1995;5(4):209–21. https:// doi. org/ 10. 1111/j. 1600- 0838. 1995. tb000 37.x. 7. World Athletics. All time outdoor lists; 2021. https:// www. world athle tics. org/ recor ds/ topli sts/. Accessed Sep 2021. 8. Joyner MJ. Modeling: optimal marathon performance on the basis of physiological factors. J Appl Physiol (1985). 1991;70(2):683–7. https:// doi. org/ 10. 1152/ jappl. 1991. 70.2. 683. 9. Conley DL, Krahenbuhl GS, Burkett LN. Training for aerobic capacity and running economy. Phys Sportsmed. 1981;9(4):107–46. https:// doi. org/ 10. 1080/ 00913 847. 1981. 11711 060. 10. Foster C, Lucia A. Running economy: the forgotten factor in elite performance. Sports Med. 2007;37(4–5):316–9. https:// doi. org/ 10. 2165/ 00007 256- 20073 7040- 00011. 11. Barnes KR, Kilding AE. Running economy: measurement, norms, and determining factors. Sports Med Open. 2015;1(1):8. https:// doi. org/ 10. 1186/ s40798- 015- 0007-y. 12. Lucia A, Esteve-Lanao J, Oliván J, Gómez-Gallego F, SanJuan AF, Santiago C, et al. Physiological characteristics of the best Eritrean runners: exceptional running economy. Appl Physiol Nutr Metab. 2006;31(5):530–40. https:// doi. org/ 10. 1139/ h06- 029. 13. Mooses M, Hackney AC. Anthropometrics and body composition in East African runners: potential impact on performance. Int J Sports Physiol Perform. 2017;12(4):422–30. https:// doi. org/ 10. 1123/ ijspp. 2016- 0408. 14. Kunimasa Y, Sano K, Oda T, Nicol C, Komi PV, Locatelli E, et al. Specific muscle-tendon architecture in elite Kenyan distance run- ners. Scand J Med Sci Sports. 2014;24(4):e269–74. https:// doi. org/ 10. 1111/ sms. 12161. 15. Barnes KR, Kilding AE. A randomized crossover study investi- gating the running economy of highly-trained male and female distance runners in marathon racing shoes versus track spikes. Sports Med. 2019;49(2):331–42. https:// doi. org/ 10. 1007/ s40279- 018- 1012-3. 16. Saunders PU, Pyne DB, Telford RD, Hawley JA. Factors affect- ing running economy in trained distance runners. Sports Med. 2004;34(7):465–85. https:// doi. org/ 10. 2165/ 00007 256- 20043 4070- 00005. 17. Conley DL, Krahenbuhl GS. Running economy and distance running performance of highly trained athletes. Med Sci Sports Exerc. 1980;12(5):357–60. https:// doi. org/ 10. 1249/ 00005 768- 19801 2050- 00010. 18. Hoogkamer W, Kipp S, Spiering BA, Kram R. Altered running economy directly translates to altered distance-running perfor- mance. Med Sci Sports Exerc. 2016;48(11):2175–80. https:// doi. org/ 10. 1249/ mss. 00000 00000 001012. 19. Worobets J, Wannop JW, Tomaras E, Stefanyshyn D. Softer and more resilient running shoe cushioning properties enhance run- ning economy. Footwear Sci. 2014;6(3):147–53. https:// doi. org/ 10. 1080/ 19424 280. 2014. 918184. 20. Nigg BM, Cigoja S, Nigg SR. Effects of running shoe construc- tion on performance in long distance running. Footwear Sci. 2020;12(3):133–8. https:// doi. org/ 10. 1080/ 19424 280. 2020. 17787 99. 21. Stefanyshyn D, Fusco C. Increased shoe bending stiffness increases sprint performance. Sports Biomech. 2004;3(1):55–66. https:// doi. org/ 10. 1080/ 14763 14040 85228 30. 22. Hoogkamer W, Kipp S, Kram R. The biomechanics of competi- tive male runners in three marathon racing shoes: a randomized crossover study. Sports Med. 2019;49(1):133–43. https:// doi. org/ 10. 1007/ s40279- 018- 1024-z. 23. Farina EM, Haight D, Luo G. Creating footwear for performance running. Footwear Sci. 2019;11:S134–5. https:// doi. org/ 10. 1080/ 19424 280. 2019. 16061 19. 24. Muniz-Pardos B, Sutehall S, Angeloudis K, Guppy FM, Bosch A, Pitsiladis Y. Recent improvements in marathon run times are likely technological, not physiological. Sports Med. 2021;51(3):371–8. https:// doi. org/ 10. 1007/ s40279- 020- 01420-7. 25. Quealy K, Katz J. Nikeʼs fastest shoes may give runners an even bigger advantage than we thought. The New York Times; 2019. 26. Rodrigo-Carranza V, González-Mohíno F, Santos del Cerro J, Santos-Concejero J, González-Ravé JM. Influence of advanced shoe technology on the top 100 annual performances in men’s marathon from 2015 to 2019. Sci Rep. 2021;11(1): 22458. https:// doi. org/ 10. 1038/ s41598- 021- 01807-0. 27. Hébert-Losier K, Finlayson SJ, Driller MW, Dubois B, Esculier J-F, Beaven CM. Metabolic and performance responses of male runners wearing 3 types of footwear: Nike Vaporfly 4%, Saucony Endorphin racing flats, and their own shoes. J Sport Health Sci. 2020. https:// doi. org/ 10. 1016/j. jshs. 2020. 11. 012. 28. Hoogkamer W, Kipp S, Frank JH, Farina EM, Luo G, et al. A comparison of the energetic cost of running in marathon racing shoes. Sports Med. 2018;48(4):1009–19. https:// doi. org/ 10. 1007/ s40279- 017- 0811-2. 29. Hunter I, McLeod A, Valentine D, Low T, Ward J, Hager R. Run- ning economy, mechanics, and marathon racing shoes. J Sports Sci. 2019;37(20):2367–73. https:// doi. org/ 10. 1080/ 02640 414. 2019. 16338 37. 30. Joubert DP, Jones GP. A comparison of running economy across seven highly cushioned racing shoes with carbon-fibre plates. Footwear Sci. 2022. https:// doi. org/ 10. 1080/ 19424 280. 2022. 20386 91. 31. McKay AKA, Stellingwerff T, Smith ES, Martin DT, Mujika I, Goosey-Tolfrey VL, et al. Defining training and performance caliber: a participant classification framework. Int J Sports Phys- iol Perform. 2022;17(2):317–31. https:// doi. org/ 10. 1123/ ijspp. 2021- 0451. 32. Tam E, Rossi H, Moia C, Berardelli C, Rosa G, Capelli C, et al. Energetics of running in top-level marathon runners from Kenya. Eur J Appl Physiol. 2012;112(11):3797–806. https:// doi. org/ 10. 1007/ s00421- 012- 2357-1. 33. Atkinson G, Reilly T. Circadian variation in sports performance. Sports Med. 1996;21(4):292–312. https:// doi. org/ 10. 2165/ 00007 256- 19962 1040- 00005. 34. Jones AM, Doust JH. A 1% treadmill grade most accurately reflects the energetic cost of outdoor running. J Sports Sci. 1996;14(4):321–7. https:// doi. org/ 10. 1080/ 02640 41960 87277 17. 35. Macfarlane DJ, Wong P. Validity, reliability and stability of the portable Cortex Metamax 3B gas analysis system. Eur J Appl Physiol. 2012;112(7):2539–47. https:// doi. org/ 10. 1007/ s00421- 011- 2230-7. 36. Vogler AJ, Rice AJ, Gore CJ. Validity and reliability of the Cortex MetaMax3B portable metabolic system. J Sports Sci. 2010;28(7):733–42. https:// doi. org/ 10. 1080/ 02640 41090 35827 76. 37. Poole DC, Jones AM. Measurement of the maximum oxygen uptake ̇VO2max: ̇VO2peak is no longer acceptable. J Appl Phys- iol. 2017;122(4):997–1002. https:// doi. org/ 10. 1152/ jappl physi ol. 01063. 2016. 38. Pérronet F, Massicotte D. Table of nonprotein respiratory quotient: an update. Can J Sport Sci. 1991;16:23–9. 39. RStudio Team. RStudio: integrated development environment for R. Boston: RStudio Inc.; 2016. 40. R Core Team. R: a language and environment for statistical com- puting. Vienna: R Foundation for Statistical Computing; 2020. 1271 Variability in Advanced Footwear Technology 41. Chambers JM, Freeny A, Heiberger RM. Analysis of variance; designed experiments. In: Chambers JM, Hastie TJ, editors. Statistical models in S. Wadsworth & Brooks/Cole. New York: Springer; 1992. 42. Schmider E, Ziegler M, Danay E, Beyer L, Bühner M. Is it really robust? Methodology. 2010;6(4):147–51. https:// doi. org/ 10. 1027/ 1614- 2241/ a0000 16. 43. Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. London: Routledge; 1988. 44. Doi SAR, Barendregt JJ, Khan S, Thalib L, Williams GM. Advances in the meta-analysis of heterogeneous clinical trials I: the inverse variance heterogeneity model. Contemp Clin Trials. 2015;45:130–8. https:// doi. org/ 10. 1016/j. cct. 2015. 05. 009. 45. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses. BMJ. 2003;327(7414):557–60. https:// doi. org/ 10. 1136/ bmj. 327. 7414. 557. 46. Sterne JAC, Savović J, Page MJ, Elbers RG, Blencowe NS, Boutron I, et al. RoB 2: a revised tool for assessing risk of bias in randomised trials. BMJ. 2019;366: l4898. https:// doi. org/ 10. 1136/ bmj. l4898. 47. Guinness J, Bhattacharya D, Chen J, Chen M, Loh A. An obser- vational study of the effect of Nike Vaporfly shoes on marathon performance. arXiv: Applications; 2020. 48. Senefeld JW, Haischer MH, Jones AM, Wiggins CC, Beilfuss R, Joyner MJ, et al. Technological advances in elite marathon per- formance. J Appl Physiol. 2021;130(6):2002–8. https:// doi. org/ 10. 1152/ jappl physi ol. 00002. 2021. 49. Bermon S, Garrandes F, Szabo A, Berkovics I, Adami PE. Effect of advanced shoe technology on the evolution of road race times in male and female elite eunners. Front Sports Act Living. 2021. https:// doi. org/ 10. 3389/ fspor. 2021. 653173. 50. Day E, Hahn M. Optimal footwear longitudinal bending stiffness to improve running economy is speed dependent. Footwear Sci. 2020;12(1):3–13. https:// doi. org/ 10. 1080/ 19424 280. 2019. 16968 97. 51. Novacheck TF. The biomechanics of running. Gait Posture. 1998;7(1):77–95. https:// doi. org/ 10. 1016/ S0966- 6362(97) 00038-6. 52. World Athletics. Athlete profiles; 2021. https:// www. world athle tics. org/ athle tes. Accessed 21 Nov 2021. 53. Kipp S, Kram R, Hoogkamer W. Extrapolating metabolic sav- ings in running: implications for performance predictions. Front Physiol. 2019;10:79. https:// doi. org/ 10. 3389/ fphys. 2019. 00079. 54. Colino E, Felipe JL, Van Hooren B, Gallardo L, Meijer K, Lucia A, et al. Mechanical properties of treadmill surfaces compared to other overground sport surfaces. Sensors (Basel). 2020. https:// doi. org/ 10. 3390/ s2014 3822. 55. Gidley AD, Lankford DE, Bailey JP. The construction of com- mon treadmills significantly affects biomechanical and metabolic variables. J Sports Sci. 2020;38(19):2236–41. https:// doi. org/ 10. 1080/ 02640 414. 2020. 17769 29. 56. Anderson LM, Bonanno DR, Hart HF, Barton CJ. What are the benefits and risks associated with changing foot strike pattern dur- ing running? A systematic review and meta-analysis of injury, run- ning economy, and biomechanics. Sports Med. 2020;50(5):885– 917. https:// doi. org/ 10. 1007/ s40279- 019- 01238-y. 57. Hardin EC, van den Bogert AJ, Hamill J. Kinematic adaptations during running: effects of footwear, surface, and duration. Med Sci Sports Exerc. 2004;36(5):838–44. https:// doi. org/ 10. 1249/ 01. mss. 00001 26605. 65966. 40. 58. Ridge ST, Johnson AW, Mitchell UH, Hunter I, Robinson E, Rich BS, et al. Foot bone marrow edema after a 10-wk tran- sition to minimalist running shoes. Med Sci Sports Exerc. 2013;45(7):1363–8. https:// doi. org/ 10. 1249/ MSS. 0b013 e3182 874769. 59. Cejuela R, Sellés-Pérez S. Road to Tokyo 2020 Olympic Games: training characteristics of a world class male triathlete. Front Physiol. 2022. https:// doi. org/ 10. 3389/ fphys. 2022. 835705. 60. Page MJ, McKenzie JE, Bossuyt PM, Boutron I, Hoffmann TC, Mulrow CD, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ. 2021;372: n71. https:// doi. org/ 10. 1136/ bmj. n71.

Variability in Running Economy of Kenyan World-Class and European Amateur Male Runners with Advanced Footwear Running Technology: Experimental and Meta-analysis Results.

03-02-2023

Knopp, Melanie,Muñiz-Pardos, Borja,Wackerhage, Henning,Schönfelder, Martin,Guppy, Fergus,Pitsiladis, Yannis,Ruiz, Daniel

eng

PMC10356687

Full title: The Acute Demands of Repeated-Sprint Training on Physiological, Neuromuscular, Perceptual and Performance Outcomes in Team Sport Athletes: A Systematic Review and Meta-Analysis Running heading: Acute Demands of Repeated-Sprint Training Authors: Fraser Thurlow1,3, Jonathon Weakley1,2,3, Andrew Townshend1,3, Ryan G. Timmins1,3, Matthew Morrison1,3, Shaun J. McLaren4,5 Affiliations: 1 School of Behavioural and Health Sciences, Australian Catholic University, Brisbane, Australia 2 Carnegie Applied Rugby Research (CARR) Centre, Carnegie School of Sport, Leeds Beckett University, United Kingdom 3 Sports Performance, Recovery, Injury and New Technologies (SPRINT) Research Centre, Australian Catholic University, Queensland, Australia 4 Newcastle Falcons Rugby Club, Newcastle upon Tyne, United Kingdom 5 Institute of Sport, Manchester Metropolitan University, Manchester, UK ORCID Identifiers: Fraser Thurlow: 0000-0002-0234-9615 Jonathon Weakley: 0000-0001-7892-4885 Andrew D. Townshend: 0000-0002-6714-8304 Ryan G. Timmins: 0000-0003-4964-1848 Matthew Morrison: 0000-0002-3535-6707 Shaun J. McLaren: 0000-0003-0480-3209 Corresponding Author: Fraser Thurlow School of Behavioural and Health Sciences, Australian Catholic University, 1100 Nudgee Road, Banyo 4014, Queensland, AUSTRALIA E: fraser.thurlow@acu.edu.au Supplementary Table S1. Modified Downs and Black scale outcomes for the assessment of reporting quality and risk of bias. Study Item number Total score (out of 14) 1 2 3 6 7 10 12 15 16 18 20 22 23 25 Abt et al. [118] 1 1 0 1 1 1 0 0 1 1 1 0 0 1 9 AbuMoh’D et al. [180] 1 1 0 1 1 1 0 1 1 1 1 0 1 1 11 Aguiar et al. [95] 1 1 1 1 1 0 0 0 1 1 1 0 1 0 9 Akenhead et al. [57] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Alemdaroğlu et al. [23] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Almansba et al. [96] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Alizadeh et al. [167] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Altimari et al. [181] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Archiza et al. [182] 1 1 1 1 1 1 0 1 1 1 1 0 1 1 12 Attene et al. [115] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Ayarra et al. [183] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Aziz et al. [184] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Baldi et al. [185] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Balsalobre-Fernández et al. [186] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Beato et al. [187] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 10 Beato et al. [188] 1 1 1 1 1 0 0 0 1 1 1 0 1 0 9 Beato & Drust [162] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 9 Beaven et al. [189] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Binnie et al. [190] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Binnie et al. [191] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Binnie et al. [192] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Blasco-Lafarga et al. [108] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Borges et al. [193] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Brahim et al. [97] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Brini et al. [154] 1 1 0 1 1 0 0 0 1 1 1 0 0 1 8 Brini et al. [98] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Brini et al. [194] 1 1 1 1 1 0 0 0 1 1 1 0 1 0 9 Brini et al. [195] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Brini et al. [46] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Brocherie et al. [196] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Brocherie et al. [54] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Brocherie et al. [197] 1 1 1 1 1 1 0 1 1 1 1 0 1 1 12 Broderick et al. [141] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Buchheit [198] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Buchheit et al. [59] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Buchheit et al. [60] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Buchheit et al. [99] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Campa et al. [168] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Campos et al. [199] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Campos-Vazquez et al. [200] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 10 Caprino et al. [201] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Castagna et al. [153] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Castagna et al. [202] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Chaouachi et al. [203] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Charlot et al. [204] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Chen et al. [205] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Clifford et al. [34] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Corrêa et al. [206] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Costello et al. [127] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Cuadrado-Peñafiel et al. [207] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Da Silva et al. [208] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Dal Pupo et al. [116] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Study Item number Total score (out of 14) 1 2 3 6 7 10 12 15 16 18 20 22 23 25 Dal Pupo et al. [209] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Daneshfar et al. [210] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Dardouri et al. [211] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 de Andrade et al. [212] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Delextrat et al. [213] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Delextrat et al. [214] 1 1 1 1 1 1 1 0 1 1 1 0 0 1 11 Delextrat et al. [175] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Dellal et al. [61] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Dellal & Wong [100] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Dent et al. [131] 1 1 0 1 1 0 0 0 1 1 1 0 0 1 8 Donghi et al. [215] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Doyle et al. [216] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Dupont et al. [217] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Dupont et al. [86] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 10 Eliakim et al. [124] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Elias et al. [218] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Elias et al. [219] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Eniseler et al. [220] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Eryilmaz & Kaynak [221] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Eryilmaz et al. [37] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Essid et al. [222] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Farjallah et al. [223] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Figueira et al. [119] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Fornasier-Santos et al. [224] 1 1 1 1 1 0 0 1 1 1 1 0 1 0 10 Fort-Vanmeerhaeghe et al.[225] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Fortin & Billaut [226] 1 1 1 1 1 0 0 1 1 1 1 0 0 1 10 Freitas et al. [227] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Gabbett [228] 1 1 0 1 1 1 0 0 1 1 1 0 0 0 9 Gabbett et al. [89] 1 1 0 1 1 1 0 0 1 1 1 0 0 0 8 Gabbett et al. [229] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Gabbett et al. [230] 1 1 0 1 1 1 0 0 1 1 1 0 0 0 9 Galvin et al. [231] 1 1 1 1 1 1 0 1 1 1 1 0 1 1 12 Galy et al. [177] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Gantois et al. [232] 1 1 0 1 1 0 0 0 1 1 1 0 0 0 7 Gantois et al. [14] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Gantois et al. [233] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 García-Unanue et al. [169] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Gatterer et al. [234] 1 1 1 1 1 1 0 1 1 1 1 0 1 1 12 Gharbi et al. [83] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Gharbi et al. [235] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Gibson et al. [101] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Girard et al. [236] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Girard et al. [149] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 González-Frutos et al. [237] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Gonzalo-skok et al. [102] 1 1 1 1 1 0 0 0 1 1 1 0 1 1 10 Goodall et al. [238] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Hamlin et al. [239] 0 1 1 1 1 0 0 0 1 1 1 0 1 1 9 Hamlin et al. [240] 1 1 1 1 1 0 0 1 1 1 1 0 1 1 11 Hammami et al. [241] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Haugen et al. [62] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Haugen et al. [128] 1 1 1 1 1 0 0 0 1 1 1 0 1 1 10 Hermassi et al. [242] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Higham et al. [90] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Hollville et al. [243] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Howatson et al. [35] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Study Item number Total score (out of 14) 1 2 3 6 7 10 12 15 16 18 20 22 23 25 Iaia et al. [120] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Iaia et al. [19] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Impellizzeri et al. [170] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Ingebrigtsen et al. [244] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Ingebrigtsen et al. [171] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Iacono et al. [42] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Izquierdo et al. [140] 1 1 1 1 1 0 0 1 1 1 1 0 1 1 11 Jang & Joo [245] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Jiménez-Reyes et al. [246] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Johnston & Gabbett [40] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Joo [110] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Jorge et al. [247] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Kaplan [109] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Keir et al. [25] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Keogh et al. [172] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Kilduff et al. [248] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Klatt et al. [36] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Krakan et al. [249] 1 1 0 1 1 1 0 0 1 1 1 0 0 0 8 Krueger et al. [250] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Lakomy et al. [78] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Lapointe et al. [251] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Le Rossignol et al. [173] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Little & Williams [121] 1 1 0 1 1 0 0 0 1 1 1 0 0 0 7 Lockie et al. [252] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Lockie et al. [253] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Lockie et al. [254] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Lombard et al. [255] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Madueno et al. [24] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Maggioni et al. [16] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Mancha-Triguero et al. [139] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Marcelino et al. [256] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Matzenbacher et al. [257] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 McGawley & Andersson [258] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Meckel et al. [259] 1 1 0 1 1 0 0 0 1 1 1 0 0 1 8 Meckel et al. [260] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Meckel et al. [261] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Meckel et al. [262] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Meckel et al. [263] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Michalsik et al. [264] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Mohr et al. [265] 1 1 1 1 1 0 0 0 1 1 1 0 1 1 10 Mohr et al. [266] 1 1 1 1 1 0 0 0 1 1 1 0 1 1 10 Mohr et al. [267] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Moncef et al. [268] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Morcillo et al. [48] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Moreira et al. [269] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Mujika et al. [164] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Müller et al. [270] 1 1 0 1 1 1 0 0 1 1 1 0 0 0 8 Okuno et al. [271] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Nakamura et al. [272] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Nascimento et al [273] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Nedrehagen & Saeterbakken [274] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Nikolaidis et al. [275] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Padulo et al. [276] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Padulo et al. [277] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Padulo et al. [114] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Notes: 0 = no; 1 = yes; U = unable to determine. Item 1: clear aim/hypothesis; Item 2: outcome measures clearly described; Item 3: patient characteristics clearly described; Item 6: main findings clearly described; Item 7: measures of random variability provided; Item 10: actual probability values reported; Item 12: participants prepared to participate representative of the entire population; Item 15: blinding of outcome measures; Item 16: analysis completed was planned; Item 18: appropriate statistics; Item 20: valid and reliable outcome measures; Item 22: participants recruited over the same period; Item 23: randomised; Item 25: adjustment made for confounding variables. Study Item number Total score (out of 14) 1 2 3 6 7 10 12 15 16 18 20 22 23 25 Padulo et al [156] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Padulo et al. [150] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Paulauskas et al. [122] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Perroni et al. [103] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Petisco et al. [278] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Purkhús et al. [279] 1 1 1 1 1 0 0 0 1 1 1 0 1 0 9 Pyne et al. [280] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Ramírez-Campillo et al. [281] 1 1 1 1 1 0 0 1 1 1 1 0 1 1 11 Rampinini et al. [282] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Rampinini et al. [174] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Rey et al. [283] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Rodríguez-Fernández et al. [165] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Rodríguez-Fernández et al. [284] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Røksund et al. [285] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Ruscello et al. [286] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Ruscello et al. [104] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Russell et al. [123] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Salleh et al. [287] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Sánchez-Sánchez et al. [117] 1 1 1 1 1 0 0 0 1 1 1 0 0 1 9 Sánchez-Sánchez et al. [288] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Sánchez-Sánchez et al. [289] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Sanders et al. [290] 1 1 1 1 1 1 0 0 1 1 0 0 0 0 8 Scanlan et al. [291] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Scanlan et al. [292] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Selmi et al. [58] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Selmi et al. [293] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Shalfawi et al. [294] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Shalfawi et al. [295] 1 1 1 1 1 0 0 0 1 1 1 0 1 0 9 Shalfawi et al. [296] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Silva et al. [297] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Soares-Caldeira et al. [298] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Spineti et al. [299] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Stojanovic et al. [300] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Suarez-Arrones et al. [105] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Taylor et al. [2] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Teixeira et al. [301] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Thomassen et al. [302] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Tønnessen et al. [303] 1 1 1 1 1 1 0 0 1 1 1 0 1 0 10 Torreblanca-Martinez et al. [304] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Tounsi et al. [176] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Trecroci et al. [305] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Turki et al. [111] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Ulupinar et al. [126] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Ulupinar et al. [125] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Van den Tillaar et al. [306] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Vasquez-Bonilla et al. [307] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Wadley & Le Rossignol [308] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 West et al. [309] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Woolley et al. [33] 1 1 1 1 1 0 0 0 1 1 1 0 0 0 8 Yanci et al. [310] 1 1 1 1 1 0 0 0 1 1 1 0 1 0 9 Zagatto et al. [106] 1 1 1 1 1 1 0 0 1 1 1 0 0 1 10 Zagatto et al. [311] 1 1 1 1 1 1 0 0 1 1 1 0 1 1 11 Zagatto et al. [107] 1 1 1 1 1 1 0 0 1 1 1 0 0 0 9 Supplementary Table S2. Summary of participant and study characteristics from all included studies. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Abt et al. [118] 11 (NR) SOC TRA NR NR NR NC CRO (ran) 6 different, time-matched RS protocols (∼60 s), performed twice each on an indoor synthetic sports floor, separated by 3−7 days. AbuMoh’d et al. [180] 18 SOC NAT NR NR NR C PAG (r) Baseline RS test on an athletics track, before an intervention. Aguiar et al. [95] 34 SOC NAT INT: 27 ± 5 CON: 27 ± 5 INT: 175 ± 5 CON: 175 ± 6 INT: 73 ± 5 CON: 73 ± 7 NC PAG (r) RS test before an intervention Akenhead et al. [57] 9 SOC NAT 26 ± 3 172 ± 6 71 ± 7 NC OBS RS test performed in an indoor sports hall. Test ends when Sdec = 5% for 2 consecutive trials. Alemdaroğlu et al. [23] 9 SOC TRA 18 ± 1 177 ± 5 74 ± 7 NC CRO (ran) 4 different RS tests performed twice each on an AG pitch, separated by 48 hrs. Alizadeh et al. [167] 41 SOC NAT High: 17 ± 1 Med: 18 ± 1 Low: 17 ± 1 High: 177 ± 3 Med: 174 ± 5 Low: 171 ± 5 High: 71 ± 4 Med: 66 ± 5 Low: 67 ± 5 NC OBS Single RS test. Results according to the criterion of VO2max Almansba et al. [96] 17 SOC NAT 16 ± 0 175 ± 1 67 ± 9 NC CRO (ran) 2 RS tests performed on AG, separated by 72 hrs. Altimari et al. [181] 46 SOC NAT 18 ± 0 174 ± 5 64 ± 4 NC OBS RS test on a SOC field. U17 group only, birth tertiles combined. Archiza et al. [182] 18 (0%) SOC NAT Sham: 20 ± 2 INT: 22 ± 4 Sham: 160 ± 0 INT: 160 ± 0 Sham: 55 ± 5 INT: 56 ± 6 C PAG (r) Baseline RS test on a grass field, before an intervention. Attene et al. [115] 36 (39%) BB NAT M: 16 ± 1; F: 16 ± 1 M: 178 ± 1 F: 165 ± 1 M: 66 ± 6 F: 56 ± 7 NC PAG (r) 2 different baseline RS tests on an indoor court, as part of a testing battery, before a RST intervention. Ayarra et al. [183] 40 FUT TRA 22 ± 5 176 ± 7 70 ± 10 NC OBS Single RS test on an indoor wooden surface. Aziz et al. [184] 40 MIX INTL 23 ± 4 173 ± 1 64 ± 6 NC OBS RS test on NG, as part of a testing battery. Baldi et al. [185] 26 SOC NAT 23 ± 4 178 ± 6 72 ± 8 NC OBS RS test on outdoor NG, as part of a testing battery. Balsalobre- Fernández et al. [186] 11 BB NAT 25 ± 6 200 ± 11 99 ± 9 NC OBS RS test in an indoor hall. Beato et al. [187] 36 SOC TRA 21 ± 2 179 ± 7 74 ± 7 NC PAG (r) Baseline RS test before an intervention and RS training data Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Beato et al. [188] 20 SOC NAT 18−21 177 ± 6 71 ± 7 NC PAG (r) Baseline RS test before an intervention and RS training data Beato & Drust [162] 16 SOC TRA 21 ± 1 179 ± 8 71 ± 8 NC CRO (ran) RS test on a synthetic outdoor track. Sub-maximal RS test excluded from the review. Beaven et al. [189] 12 RUG NAT 22 ± 1 185 ± 4 96 ± 9 C CRO (ran) RS test on an indoor running track. Binnie et al. [190] 24 (0%) HOC NR SAN: 19 ± 7 GRA: 21 ± 4 SAN: 168 ± 12 GRA: 167 ± 67 SAN: 66 ± 9 GRA: 63 ± 6 NC PAG (r) Baseline RS test in a gymnasium, before an intervention. Participant’s pair-matched by VO2max. Binnie et al. [191] 10 (70%) HOC/ NET NAT M: 23 ± 3 F: 20 ± 3 M: 182 ± 5 F: 176 ± 11 M: 83 ± 6 F: 69 ± 15 NC CRO (ran) Baseline RS test in a gymnasium. Binnie et al. [192] 10 (80%) HOC/ NET NR M: 22 ± 2 F: 21 ± 1 M; 181 ± 5 F: 179 ± 14 M: 78 ± 6 F: 74 ± 18 NC CRO (ran) Baseline RS test in a gymnasium Blasco-Lafarga et al. [108] 13 SOC NAT 18 ± 1 172 ± 4 68 ± 6 NC CRO RS test on a SOC pitch. Borges et al. [193] 20 SOC NAT 17 ± 1 175 ± 7 69 ± 9 NC PAG (r) Baseline RS test before an intervention. Brahim et al. [97] 27 SOC NAT DEF: 18 ± 1 MID: 18 ± 1 FWD: 17 ± 1 DEF: 183 ± 6 MID: 178 ± 5 FWD: 180 ± 5 DEF: 75 ± 9 MID: 70 ± 7 FWD: 72 ± 4 NC OBS 3 different RS tests on NG, separated by > 1 day. Brini et al. [154] 16 BB NR 23 ± 3 186 ± 10 78 ± 8 NC CRO (ran) 4 different RS protocols, separated by 48-hrs. Brini et al. [98] 16 BB NAT 22 ± 3 186 ± 10 78 ± 8 C PAG (r) RS test before an intervention. Brini et al. [194] 16 BB NR 23 ± 2 186 ± 9 78 ± 11 C PAG (r) RS test before an intervention. Brini et al. [195] 16 BB NAT 23 ± 2 186 ± 10 78 ± 8 NC CRO (ran) 2 different RS tests on a BB court, separated by > 48-hrs. Brini et al. [46] 40 BB NAT 27 ± 3 192 ± 9 88 ± 9 NC OBS RS test on a wooden BB court. Brocherie et al. [196] 16 SOC INTL 27 ± 4 177 ± 4 72 ± 5 NC OBS RS test on indoor AG, as part of a testing battery. Brocherie et al. [54] 8 SOC INTL 28 ± 5 176 ± 4 72 ± 3 NC OBS RS test on indoor AG. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Brocherie et al. [197] 36 HOC NAT 25 ± 5 178 ± 6 76 ± 8 C PAG (r) Baseline RS test on an indoor synthetic floor, before an intervention. Broderick et al. [141] 19 MIX TRA 21.0 ± 2.0 178.8 ± 7.2 8.1 ± 8.9 C CRO (ran) RS tests in an indoor gymnasium, separated by 7 days. Buchheit [198] 27 MIX NAT HB: 23 ± 3 TS3: 23 ± 4 HB: 188 ± 7 TS3: 180 ± 8 HB: 88 ± 11 TS3: 77 ± 9 NC OBS RS tests were performed by different groups of athletes on an indoor synthetic track. Buchheit et al. [59] 13 MIX NR 22 ± 3 179 ± 5 75 ± 5 NC CRO (ran) 4 different RS protocols on an indoor synthetic track, separated by > 48-hrs. Buchheit et al. [60] 13 MIX NR 22 ± 3 179 ± 5 75 ± 5 NC CRO (ran) 2 different RS protocols on an indoor synthetic track, separated by > 48-hrs. Buchheit et al. [99] 12 MIX NAT 22 ± 2 178 ± 8 76 ± 4 NC CRO (ran) 4 different RS protocols on an indoor synthetic track, separated by > 48-hrs. Campa et al. [168] 36 SOC NAT 17 ± 1 EL: 177 ± 6 S-EL: 178 ± 6 EL: 69 ± 4 S-EL: 70 ± 7 NC OBS RS test on NG. Campos et al. [199] 11 FUT NAT 19 ± 1 178 ± 7 71 ± 6 NC PAG Baseline RS test on an indoor FUT court before an intervention. Campos- Vazquez et al. [200] 21 SOC NAT 18 ± 1 177 ± 6 70 ± 7 NC PAG (r) Baseline RS test on AG, before an intervention. Caprino et al. [201] 10 BB TRA 16 ± 1 184 + 7 77 + 8 NC OBS RS test before an official BB match. Castagna et al. [153] 16 BB TRA 17 ± 1 181 ± 6 73 ± 10 NC OBS (ran) 2 different RS tests on an indoor wooden BB court, separated by > 48-hrs, as part of a testing battery. Castagna et al. [202] 18 BB TRA 17 ± 1 181 ± 6 73 ± 10 NC OBS RS test on an indoor wooden BB court, separated by > 48-hrs. Chaouachi et al. [203] 23 SOC NAT 19 ± 1 181 ± 6 73 ± 4 NC CRO (ran) RS test on an indoor synthetic track. Charlot et al. [204] 10 FUT NAT 26 ± 4 170 ± 7 70 ± 9 NC OBS* RS test before a FUT tournament Chen et al. [205] 26 SOC NAT 21 ± 1 173 ± 4 65 ± 5 C PAG (r) RS test on an indoor synthetic surface. Clifford et al. [34] 20 MIX NAT CON: 21 ± 2 INT: 23 ± 3 CON: 177 ± 1 INT: 183 ± 1 CON: 73 ± 12 INT: 77 ± 10 C PAG (r) Baseline RS test before an intervention period. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Corrêa et al. [206] 10 SOC TRA 19 ± 1 179 ± 0 71 ± 7 NC OBS* Baseline RS test on outdoor NG. Costello et al. [127] 24 RUG NAT 21 ± 2 182 ± 5 88 ± 9 C CRO (ran) RS protocol (session 1 & day 1 only) Cuadrado- Peñafiel et al. [207] 37 SOC/ FUT NAT SOC: 29 ± 1 FUT: 27 ± 5 SOC: 178 ± 1 FUT: 179 ± 1 SOC: 73 ± 12 FUT: 75 ± 7 NC OBS Single RS test Da Silva et al. [208] 29 SOC NAT 18 ± 1 179 ± 5 74 ± 7 NC OBS Single RS test. Dal Pupo [116] 14 FUT TRA U17 170 ± 6 63 ± 8 NC OBS (ran) 2 different RS tests on a FUT court, separated by 48-hrs. Dal Pupo et al. [209] 7 FUT TRA 16 ± 1 172 ± 9 65 ± 8 NC OBS (ran) 2 different RS tests on a concrete floor, separated by 48-hrs. Daneshfar et al. [210] 20 HB INTL 16 ± 1 185 ± 5 83 ± 6 NC OBS (ran) 2 different RS tests were performed indoors, separated by 48- hrs, as part of a testing battery. Dardouri et al. [211] 29 MIX NR 23 ± 2 180 ± 10 69 ± 9 NC OBS RS test, indoors, as part of a testing battery. de Andrade et al. [212] 16 MIX NAT 22 ± 3 186 ± 10 79 ± 23 NC OBS Single RS test on an indoor rigid surface. Delextrat et al. [213] 17 (53%) BB TRA M: 22 ± 3 F: 21 ± 3 M: 19 ± 9 F: 176 ± 8 M: 91 ± 10 F: 74 ± 10 C CRO (ran) (r) Baseline RS test, before an intervention. Delextrat et al. [214] 31 BB TRA FWD: 16 ± 1 G: 17 ± 1 CEN: 16 ± 1 FWD: 183 ± 5 G: 175 ± 6 CEN: 191 ± 8 FWD: 75 ± 7 G: 69 ± 5 CEN: 81 ± 3 NC OBS (ran) RS test, as part of a testing battery. Delextrat et al. [175] 16 (50%) BB TRA M: 23 ± 3 F: 22 ± 2 M: 191 ± 9 F: 179 ± 9 M: 90 ± 10 F: 78 ± 9 C PAG (ran) (r) Baseline RS test, before an intervention. Dellal et al. [61] 22 SOC INTL 24 ± 4 178 ± 6 80 ± 6 NC OBS 3 different RS protocols performed indoors, separated by > 48- hrs, as part of a testing battery. Dellal & Wong [100] 39 SOC NAT Open age to U17 PRO: 180 ± 4 U19: 178 ± 7 U17: 180 ± 6 PRO: 72 ± 4 U19: 69 ± 6 U17: 67 ± 5 NC OBS 2 different RS tests on AG, separated by 1 week. Dent et al. [131] 15 (47%) SOC TRA M: 20 ± 2 F: 19 ± 2 NR M: 79 ± 11 F: 62 ± 7 NC CRO Single RS protocol. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Donghi et al. [215] 12 SOC NAT 17 ± 1 178 ± 6 69 ± 4 C CRO (ran) Baseline RS test in an indoor gym, Doyle et al. [216] 25 (0%) SOC INTL 19 ± 3 167 ± 6 63 ± 7 NC OBS RS test performed on an indoor surface. Dupont et al. [217] 12 SOC TRA 23 ± 4 179 ± 6 72 ± 7 NC OBS RS test on an indoor tartan track. Dupont et al. [86] 11 SOC TRA 25 ± 4 176 ± 6 68 ± 4 NC OBS RS test on an indoor tartan track. Eliakim et al. [124] 12 BB NAT 16 ± 1 186 ± 10 76 ± 6 C CRO (ran) RS test on a BB court, CON condition only. Elias et al. [218] 14 ARF NAT 21 ± 3 186 ± 7 80 ± 7 NC CRO (ran) Baseline RS test on an indoor, wooden surface. Elias et al. [219] 24 ARF NAT 20 ± 3 186 ± 6 81 ± 8 NC PAG (r) Baseline RS test on an indoor wooden sprung floor, before an intervention. Eniseler et al. [220] 19 SOC NAT 17 ± 1 174 ± 5 66 ± 6 C PAG (r) Baseline RS test on NG before an intervention Eryilmaz & Kaynak [221] 16 VB TRA 21 ± 1 184 ± 5 74 ± 8 NC OBS RS test on an indoor VB court. Eryilmaz et al. [37] 12 MIX TRA 24 ± 4 179 ± 6 73 ± 9 NC SG Data extracted from one session during a RST intervention. Essid et al. [222] 18 HB NAT 17 ± 0.3 190 ± 10 78 ± 10 NC CRO (ran) RS test (morning session only) Farjallah et al. [223] 20 SOC NAT 19 ± 1 180 ± 10 70 ± 11 C PAG RS test on a SOC field, before an intervention. Figueira et al. [119] 12 BB NAT 21 ± 2 190 ± 7 86 ± 6 NC CRO (ran) 2 different RS tests. Fornasier-Santos et al. [224] 35 RUG NAT 18 ± 1 182 ± 7 95 ± 15 C PAG (r) Baseline RS test on an indoor, concrete floor and RS training data from the control group. Fort- Vanmeerhaeghe et al.[225] 11 HB (0%) NAT 17 ± 1 182 ± 7 70 ± 8 NC OBS RS test on a BB court Fortin & Billaut [226] 15 AF TRA 21 ± 2 188 ± 19 82 ± 3 NC PAG Baseline RS test before an intervention. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Freitas et al. [227] 9 BB NAT 21 ± 3 198 ± 8 93 ± 15 NC CRO (r) Baseline RS test in an indoor centre. Gabbett [228] 19 (0%) SOC NAT / INTL 18 ± 3 NR NR NC OBS Same RS test, repeated twice. Gabbett et al. [89] 58 RUG NAT 24 ± 4 184 ± 6 97 ± 10 NC OBS RS test on a synthetic surface, as part of a testing battery. Gabbett et al. [229] 86 RUG NAT ST: 25 ± 4 N-ST: 23 ± 4 N-SEL: 22 ± 4 ST: 185 ± 5 N-ST: 182 ± 6 N-SEL: 183 ± 7 ST: 96 ± 8 N-ST: 99 ± 12 N-SEL: 96 ± 11 NC OBS RS test on a synthetic surface, as part of a testing battery. Gabbett et al. [230] 16 (0%) SOC NAT / INTL 18.3 ± 2.8 NR NR NC PAG Baseline RS test, before an intervention. Galvin et al. [231] 42 RUG NAT 18 ± 2 183 ± 7 88 ± 9 C PAG (r) RS test performed outdoors, before an intervention. Galy et al. [177] 22 FUT INTL MG: 24 ± 4 N-MG: 23 ± 5 MG: 173 ± 5 | N-MG: 180 ± 8 MG: 72 ± 7 N-MG: 74 ± 12 NC OBS RS test on an indoor synthetic court, as part of a testing battery. Gantois et al. [232] 20 BB NAT 18-24 180 ± 6 81 ± 13 NC OBS RS test on a BB court. Gantois et al. [14] 20 BB NAT 21 ± 2 181 ± 8 74 ± 9 C PAG (r) RS test on a BB court, before an intervention. Gantois et al. [233] 12 BB NAT 22 ± 3 180 ± 2 81 ± 14 NC SG Baseline RS test, before an intervention. García-Unanue et al. [169] 33 FUT NAT / TRA 23 ± 4 176 ± 6 73 ± 6 NC OBS RS test on a FUT field. Results according to playing level. Gatterer et al. [234] 14 SOC TRA 24 ± 2 178 ± 7 77 ± 7 C PAG Baseline RS test, before an intervention. Gharbi et al. [83] 20 MIX TRA 22 ± 3 178 ± 7 71 ± 8 NC CRO (ran) Series of RS protocols on an indoor synthetic surface, separated by >24 hrs. Gharbi et al. [235] 16 MIX TRA 23 ± 2 178 ± 4 72 ± 3 C OBS (ran) RS test on an indoor synthetic surface Gibson et al. [101] 32 SOC TRA 18 ± 1 179 ± 5 177 ± 5 NC OBS RS test on an indoor synthetic surface Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Girard et al. [236] 12 SOC INTL 28 ± 5 176 ± 4 64 ± 5 NC OBS RS test on indoor AG, wearing normal football boots with plantar pressure insoles inserted. Girard et al. [149] 13 SOC NAT 18 ± 1 190 ± 10 83 ± 10 NC OBS RS test on indoor AG, wearing normal football boots with plantar pressure insoles inserted. González-Frutos et al. [237] 13 (0%) HOC INTL 25 ± 6 167 ± 4 59 ± 4 NC OBS Single RS test Gonzalo-skok et al. [102] 22 BB NAT 16 ± 1 180 ± 6 81 ± 13 C PAG (r) 2 different RS tests were performed on an indoor BB court, as part of a testing battery, before an intervention. Goodall et al. [238] 12 MIX NR 25 ± 6 180 ± 7 77 ± 7 NC OBS Single RS protocol. Hamlin et al. [239] 20 (85%) RUG NAT 19 ± 1 180 ± 10 85 ± 14 NC CRO (r) Baseline RS protocol, before an intervention. Hamlin et al. [240] 19 RUG TRA CON: 22 ± 4 INT: 20 ± 2 CON: 178 ± 5 INT: 174 ± 5 CON: 88 ± 14 INT: 77 ± 10 C PAG (r) Baseline RS test in an indoor stadium, on 2 separate occasions, 4−5 days apart. Hammami et al. [241] 28 HB NAT INT: 17 ± 0 CON: 17 ± 0 INT: 163 ± 4 CON: 164 ± 4 INT: 61 ± 5 CON: 60 ± 4 C PAG (r) Baseline RS test before an intervention Haugen et al. [62] 25 (52%) SOC TRA INT: 17 ± 1 CON: 17 ± 1 INT: 174 ± 8 CON: 173 ± 6 INT: 65 ± 8 CON: 62 ± 7 C PAG (r) Baseline RS test before an intervention Haugen et al. [128] 42 SOC TRA 17 ± 1 178 ± 6 66 ± 9 C PAG (r) Baseline RS test before an intervention Hermassi et al. [242] 22 HB NAT 19 ± 0 179 ± 2 83 ± 1 NC OBS (ran) 2 different RS tests, separated by 3−7 days, as part of a testing battery. Higham et al. [90] 18 RUG INTL 22 ± 2 183 ± 6 90 ± 8 NC OBS RS test on an indoor synthetic track, as part of a testing battery. Hollville et al. [243] 10 HOC NAT 19 ± 1 180 ± 6 72 ± 5 NC OBS RS test on AG. Results from the 1st set only. Howatson et al. [35] 20 MIX NAT 22 ± 2 178 ± 7 85 ± 14 NC OBS Single RS protocol performed on an outdoor track. Iaia et al. [120] 18 SOC NAT 19 ± 1 180 ± 7 74 ± 7 NC PAG (r) Baseline RS test on AG, as part of a testing battery, performed by 2 different groups. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Iaia et al. [19] 29 SOC NAT 17 ± 1 178 ± 10 69 ± 8 C PAG (r) Data extracted from baseline RS tests on AG and the 1st RS training session of an intervention. Impellizzeri et al. [170] 22 SOC NAT 22 ± 1 177 ± 4 73 ± 5 NC OBS Same RS test on NG, performed twice on different occasions Impellizzeri et al. [170] 30 SOC NAT 25 ± 5 181 ± 5 78 ± 8 NC OBS* RS test on NG, performed at different timepoints across a regular season. Impellizzeri et al. [170] 108 SOC NAT / TRA 24 ± 4 75 ± 7 179 ± 5 NC OBS* RS test on NG. Results according to player level. Ingebrigtsen et al. [244] 57 SOC NAT 22 ± 5 181 ± 5 75.2 ± 7.6 NC OBS RS test on indoor AG, as part of a testing battery Ingebrigtsen et al. [171] 51 SOC NAT PRO: 26 ± 7 SEMI: 20 ± 3 PRO: 183 ± 5 SEMI: 181 ± 5 NR NC OBS RS test. Results according to player level. Iacono et al. [42] 18 HB NAT 25 ± 4 188 ± 7 91 ± 9 NC PAG (r) RS test on an indoor court before an intervention Izquierdo et al. [140] 19 HB NAT INT: 21 ± 5 PLA: 24 ± 5 INT: 182 ± 8 PLA: 190 ± 8 INT: 79 ± 8 PLA: 87 ± 12 C PAG (r) Baseline RS test on an indoor HB court, before an intervention. Jang & Joo [245] 12 SOC NAT 23 ± 2 175 ± 6 71 ± 5 NC CRO (r) Single RS test. Jiménez-Reyes et al. [246] 20 RUG INTL 24 ± 4 188 ± 5 96 ± 7 NC OBS RS test on an indoor synthetic athletics track. Johnston & Gabbett [40] 12 RUG NR 23 ± 2 179 ± 10 85 ± 11 NC CRO (ran) The same RS test was performed twice on different occasions. Joo [140] 11 SOC TRA 22 ± 2 174 ± 6 NR NC SG Baseline RS test before an intervention. Jorge et al. [247] 43 SOC NAT 18 ± 1 178 ± 8 74 ± 10 NC OBS* RS test on NG, performed at different timepoints across a season. Kaplan [109] 85 SOC TRA 21 ± 3.8 176 ± 6 69 ± 7 NC OBS RS test on NG as part of a testing battery. Keir et al. [25] 8 SOC NAT 21 ± 2 176 ± 5 75 ± 4 NC OBS (ran) Single RS test. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Keogh [172] 74 (0%) HOC TRA REP: 19 ± 1 Club: 20 ± 2 REP: 165 ± 1 Club: 164 ± 1 REP: 59 ± 1 Club: 57 ± 1 NC OBS RS test as part of a testing battery Kilduff et al. [248] 20 RUG NAT 26 ± 2 185 ± 4 96 ± 8 C CRO (ran) Baseline RS test on an indoor synthetic track, before an intervention. Klatt et al. [36] 29 HB NAT U20: 18 ± 1 SEN: 27 ± 6 U20: 182 ± 8 SEN: 192 ± 9 U20: 79 ± 9 SEN 90 ± 14 NC OBS* Single RS protocol Krakan et al. [249] 41 (NR) MIX TRA NR RS-G, 181 ± 7 PLY, 175 ± 6 RS-G, 81 ± 8 PLY, 77 ± 9 NC PAG RS test before an intervention Krueger et al. [250] 18 HOC INTL 17 ± 1 182 ± 6 74 ± 8 C PAG (r) Baseline RS test, before an intervention. Lakomy et al. [78] 18 HOC NAT 24 ± 4 179 ± 5 77 ± 4 C CRO (ran) (r) 2 different RS protocols on AG Lapointe et al. [251] 17 (71%) BB NAT 22 186 ± 12 89 ± 17 C PAG (r) Baseline RS test before an intervention. Le Rossignol et al. [173] 20 ARF NAT 22 ± 2 188 ± 6 88 ± 8 NC OBS RS test on an outdoor synthetic track, as part of a testing battery Little & Williams [121] 6 SOC NAT 18−27 NR NR NC CRO (ran) 4 different RS protocols, performed on non-consecutive days. Lockie et al. [252] 17 SOC INTL 20 ± 2 181 ± 6 78 ± 7 NC OBS RS test on outdoor NG, as part of a testing battery. Lockie et al. [253] 19 (0%) SOC INTL 20 ± 1 164 ± 6 61 ± 8 NC OBS RS test on outdoor NG, as part of a testing battery. Lockie et al. [254] 18 SOC INTL 21 ± 2 181 ± 6 78 ± 6 NC OBS RS test on outdoor NG, as part of a testing battery. Results are for all players. Lombard et al. [255] 23 HOC NAT / INTL 24 ± 3 178 ± 3 77 ± 5 NC OBS RS test on AG, as part of a testing battery. Results are for all players. Madueno et al. [24] 8 (75%) BB NAT 20 ± 2 183 ± 10 78 ± 17 NC CRO (ran) 2 different RS protocols on an indoor hardwood floor, separated by 2−7 days. Maggioni et al. [16] 36 BB NAT 19 ± 1 182 ± 7 74 ± 10 C PAG (r) RS training data from an intervention. Mancha-Triguero et al. [139] 61 BB NAT U18 M: 195 F: 168 M: 85 F: 57 NC OBS RS test on BB court. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Marcelino et al. [256] 12 BB TRA 19 ± 1 193 ± 7 89 ± 15 NC CRO Same 2 baseline RS tests, separated by 24-hrs, before an intervention. Matzenbacher et al. [152] 9 FUT TRA 17 ± 0 176 ± 7 68 ± 9 NC OBS * RS test performed at the beginning and end of the season. McGawley & Andersson [258] 18 SOC NAT 23 ± 4 180 ± 8 76 ± 6 NC PAG Baseline RS test on AG, before an intervention. Meckel et al. [259] 18 SOC NAT 22-32 NR 77 ± 8 NC OBS * RS test performed at different timepoints across a season. Meckel et al. [260] 12 BB NAT 17 ± 1 187 ± 9 78 ± 6 NC CRO (ran) RS test on a BB court, after a game day warm-up. Meckel et al. [261] 33 SOC NAT 16-18 175 ± 4 67 ± 7 NC OBS (ran) 2 different RS tests on NG, separated by ~1 week, as part of a testing battery. Meckel et al. [262] 16 VB NAT 26 ± 5 192 ± 6 84 ± 7 NC OBS (ran) RS test in a sports arena, as part of a testing battery. Meckel et al. [263] 20 SOC NAT 17 ± 1 174 ± 7 67 ± 7 NC CRO (ran) RS test on a SOC pitch, after a match warm up. Michalsik et al.[264] 26 HB INTL 26 ± 3 189 ± 6 91 ± 9 NC OBS RS test on an indoor HB court. Results are all players combined. Mohr et al. [265] 40 SOC NAT 22 ± 0 177 ± 1 73 ± 1 C PAG (r) Baseline RS test on NG, before an intervention. Mohr et al. [266] 18 SOC TRA 19 ± 1 179 ± 6 79 ± 4 NC PAG (r) Baseline RS test on AG, before an intervention. Mohr et al. [267] 17 SOC NAT 27 ± 1 184 ± 1 80 ± 2 C CRO Baseline RS test on indoor AG, before an intervention. Moncef et al. [268] 44 HB NAT 22 ± 3 182 ± 6 85 ± 2 NC OBS RS test, as part of a testing battery. Morcillo et al. [48] 18 SOC NAT 27 ± 4 180 ± 5 78 ± 5 NC OBS Single RS test. Moreira et al. [269] 10 FUT NAT 24 ± 3 174 ± 5 73 ± 9 C CRO (ran) Baseline RS test before an intervention. Mujika et al. [164] 28 SOC TRA U17 & U18 U17: 178 ± 6 U18: 179 ± 9 U17: 70 ± 7 U18: 72 ± 8 NC OBS RS test on indoor AG. Müller et al. [270] 12 RUG TRA 25 ± 4 177 ± 5 92 ± 12 NC CRO (ran) Single RS test. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Nakamura et al. [272] 13 HB NAT 24 ± 4 187 ± 7 88 ± 3 NC OBS RS test in a gymnasium. Nascimento et al [273] 18 FUT TRA 17 ± 1 177 ± 5 69 ± 7 C PAG (r) Baseline RS test before a long-term RS intervention. Nedrehagen &Saeterbakken [274] 22 (41%) SOC TRA INT: 20 ± 3 CON: 22 ± 3 INT: 20 ± 3 CON: 22 ± 3 69 ± 10 C PAG (r) Baseline RS test on indoor AG before an intervention Nikolaidis et al. [275] 36 SOC TRA 22 ± 5 180 ± 6 75 ± 8 NC OBS RS test on AG, as part of a testing battery. Okuno et al. [271] 12 HB NAT 19 ± 2 185 ± 87 85 ± 10 NC CRO Single RS test Padulo et al. [276] 18 SOC NAT 16 ± 0 174 ± 10 65 ± 10 NC CRO (ran) Same RS test, repeated twice, on AG, separated by > 6 days. Padulo et al. [277] 17 SOC NAT 17 ± 1 179 ± 5 69 ± 7 NC CRO (ran) Same 2 RS tests and 1 different RS test on AG, separated by 3 days. Padulo et al. [114] 18 BB NAT 16 ± 1 178 ± 10 66 ± 9 NC CRO 2 different RS tests on an indoor BB court, repeated twice, separated by > 48-hrs, as part of a testing battery. Padulo et al [156] 18 SOC NAT 16 ± 0 174 ± 10 65 ± 10 NC CRO The same RS test was repeated twice, and 1 different RS test, on AG, separated by 1-week. Padulo et al. [150] 17 SOC INTL 16 ± 0 181 ± 10 66 ± 10 NC CRO 3 different RS tests on AG, separated by 5 days. Paulauskas et al. [122] 12 BB NAT 21 ± 2 190 ± 7 86 ± 6 NC CRO (ran) 2 different RS protocols, on an indoor wooden BB court, separated by 1-week. Perroni et al. [103] 12 SOC TRA 23 ± 6 177 ± 6 75 ± 7 NC SG Baseline RS test on AG, before an intervention. Petisco et al. [278] 10 SOC NAT 22 ± 3 178 ± 4 70 ± 3 C CRO (ran) RS test following the regular warm-up protocol. Purkhús et al. [279] 25 (0%) VB NAT 18 ± 4 172 ± 7 63 ± 11 C PAG (r) Baseline RS test on an indoor HB court, before an intervention. Pyne et al. [280] 60 ARF NAT 18 ± 0 188 ± 7 82 ± 8 NC OBS RS test on an indoor sprung wooden floor, as part of a testing battery. Ramírez- Campillo et al. [281] 30 (0%) SOC TRA CON: 23 ± 2 PLA: 23 ± 2 INT: 23 ± 3 CON: 161 ± 6 PLA: 164 ± 9 CR: 162 ± 4 CON: 60 ± 8 PLA: 57 ± 5 INT: 60 ± 8 C PAG (r) Baseline RS test, before an intervention. Rampinini et al. [282] 18 SOC NAT 26 ± 5 182 ± 4 81 ± 8 NC OBS RS test on outdoor NG. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Rampinini et al. [174] 23 SOC NAT / TRA PRO: 25 ± 4 AM: 26 ± 6 PRO: 180 ± 3 AM: 177 ± 5 PRO: 74 ± 5 AM: 71 ± 8 NC OBS RS test on outdoor NG. Rey et al. [283] 19 SOC TRA INT: 24 ± 3 CON: 24 ± 2 INT: 179 ± 5 CON: 178 ± 5 INT: 74 ± 7 CON: 75 ± 7 C PAG (r) Baseline RS test on an indoor court, before an intervention. Rodríguez- Fernández et al. [165] 33 SOC NAT PRO: 24 ± 3 YTH: 18 ± 1 PRO: 180 ± 2 YTH: 174 ± 10 PRO: 75 ± 5 YTH: 65 ± 1 NC SG Baseline RS test before an intervention. Rodríguez- Fernández et al. [284] 24 SOC TRA 19 ± 2 176 ± 6 67 ± 9 NC SG Baseline RS test before an intervention Røksund et al. [285] 75 SOC NAT 19 ± 3 181 ± 6 75 ± 10 NC OBS Single RS test as part of a testing battery. Ruscello et al. [286] 15 (0%) SOC NAT 23 ± 6 165 ± 6 59 ± 9 NC CRO (r) 2 different RS tests on AG, separated by > 48-hrs. Ruscello et al. [104] 17 SOC NAT 22 ± 4 177 ± 6 72 ± 10 NC CRO (r) 2 different RS tests on AG, separated by > 48-hrs. Russell et al. [123] 14 SOC NAT 18 ± 2 178 ± 5 75 ± 6 NC CRO (ran) Baseline RS test before an intervention. Salleh et al. [287] 24 SOC TRA 21 ± 2 173 ± 3 65 ± 3 NC OBS Single RS test. Sánchez-Sánchez et al. [117] 18 SOC TRA 22 ± 2 175 ± 6 74 ± 9 NC OBS RS test on 4 different AG pitches, separated by 72 hrs. Sánchez-Sánchez et al. [288] 21 SOC NAT U18 NR NR NC OBS Single RS test. Sánchez-Sánchez et al. [289] 16 SOC NAT / TRA 21 ± 1 69 ± 5 177 ± 5 C PAG Baseline RS test before an intervention Sanders et al. [290] 20 (50%) SOC INTL M: 21 ± 1 F: 20 ± 1 M: 178 ± 7 F: 168 ± 6 M: 75 ± 5 F: 63 ± 5 NC OBS Single RS test. Scanlan et al. [291] 9 (67%) MIX TRA 22 ± 4 171 ± 6 73 ± 12 NC CRO (ran) Two different RS protocols on an indoor, sprung, hardwood surface. Scanlan et al. [292] 8 (75%) BB TRA 20 ± 1 183 ± 10 78 ± 17 NC CRO RS protocol an indoor, hardwood BB court. Selmi et al. [58] 24 SOC NAT 17 ± 0 172 ± 9 68 ± 7 NC CRO (ran) 3 different RS tests on outdoor AG, separated by > 48-hrs. Selmi et al. [293] 30 SOC NAT 18 ± 1 178 ± 5 70 ± 7 C PAG (r) Baseline RS test before an intervention Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Shalfawi et al. [294] 30 (0%) SOC NAT 19 ± 4 167 ± 4 58 ± 7 NC OBS RS test in an indoor arena. Shalfawi et al. [295] 15 SOC NAT 16 ± 1 179 ± 7 68 ± 9 C PAG (r) RS test on indoor AG before an intervention Shalfawi et al. [296] 17 (0%) SOC TRA 21 ± 3 1769 ± 5 64 ± 6 C PAG (r) RS test on an indoor Mondo track Silva et al. [297] 22 SOC NAT 18 ± 1 175 ± 6 71 ± 5 NC SG Baseline RS test before an intervention. Soares-Caldeira et al. [298] 14 FUT NAT INT: 25 ± 8 CON: 21 ± 5 172 ± 6 72 ± 9 C PAG (r) RS test on an indoor synthetic floor, before an intervention. Spineti et al. [299] 22 SOC NAT 18 ± 0 180 ± 8 70 ± 9 NC PAG (r) RS test before an intervention Stojanovic et al. [300] 24 BB NAT 22. ± 3 197 ± 6 96 ± 9 NC OBS RS test on a BB court, as part of a testing battery. Suarez-Arrones et al. [105] 16 RUG TRA 27 ± 5 180 ± 7 91 ± 16 C PAG (r) Data extracted from baseline RS tests (both groups) and training data (RST group). Taylor et al. [2] 15 SOC TRA 24 ± 4 179 ± 6 77 ± 8 NC PAG Data extracted from a RST intervention. Teixeira et al. [301] 20 (0%) FUT NAT 19 ± 2 162 ± 5 59 ± 8 NC PAG (r) Baseline RS test on an indoor FUT court, as part of a testing battery, before a long-term training intervention. Thomassen et al. [302] 18 SOC NAT 23 ± 1 182 ± 2 79 ± 2 NC PAG (r) Baseline RS test on an indoor wooden surface. Tønnessen et al. [303] 20 SOC NAT 16 ± 1 176 ± 7 67 ± 9 C PAG (r) Baseline RS test before an intervention. Torreblanca- Martinez et al.[304] 18 (0%) SOC NAT 18 ± 2 162 ± 5 56 ± 7 NC SG RS test on outside AG. Tounsi et al. [176] 33 SOC NAT 17 ± 0 NR NR NC CRO (ran) RS test on NG Trecroci et al. [305] 9 SOC NAT 17−19 177 ± 2 66 ± 6 NC CRO (r) Baseline RS test on NG, before an intervention. Turki et al. [111] 19 SOC NR 18 ± 1 175 ± 7 70 ± 8 C CRO (ran) (r) Baseline RS test. Study Participants Experimental Approach N# Sport Level Age (yrs) Stature (cm) Body mass (kg) Design Type Details Ulupinar et al. [126] 18 SOC TRA 20 ± 2 178 ± 5 72 ± 6 NC CRO (ran) 2 different RS protocols on outdoor NG, separated by > 48 hrs Ulupinar et al. [125] 16 SOC TRA 19 ± 2 176 ± 5 70 ± 6 NC CRO (ran) 4 different RS protocols on indoor AG, separated by > 48 hrs Van den Tillaar et al. [306] 17 (0%) SOC NR 17 ± 1 168 ± 5 62 ± 7 NC OBS Single RS test on a track. Vasquez-Bonilla et al. [307] 38 (0%) SOC NAT 23 ± 4 165 ± 11 61 ± 7 NC OBS Single RST test on an indoor court Wadley & Le Rossignol [308] 17 ARF NAT 21 ± 2 182 ± 5 81 ± 10 NC OBS RS test on an asphalt surface, as part of a testing battery. West et al. [309] 15 RUG NAT 28 ± 3 188 ± 6 99 ± 9 C CRO (ran) RS test on an indoor sprint track. Woolley et al. [33] 10 MIX NR 27 ± 3 178 ± 6 78 ± 8 NC CRO (ran) RS protocol on a non-slip indoor surface Yanci et al. [310] 39 FUT TRA 23 ± 5 170 ± 10 69 ± 10 C PAG (r) Baseline RS test before an intervention Zagatto et al. [106] 20 BB NAT 17 ± 1 191 ± 8 84 ± 12 NC CRO (ran) 2 different RS tests on an indoor court, separated by 2−4 days. Zagatto et al. [311] 12 BB NAT 25 ± 7 200 ± 10 97 ± 9 C CRO (r) RS test on a BB court, CON condition only. Zagatto et al. [107] 10 BB NAT 17 ± 1 191 ± 7 87 ± 15 C CRO (ran) Single RS protocol on a BB court Data are presented as mean ± standard deviation. Abbreviations: N# = number of participants (unless stated, the proportion of males was 100%). M = male; F = female; NR = not reported; NA = not applicable; OBS = observational design; CRO = crossover design; SG = single group pre-test post-test design; ran = experimental treatment or measurements delivered in a randomised order; r = random assignment of participants to experimental groups; C = controlled study; NC = non-controlled study; PLA = placebo; SOC = soccer, FUT = futsal; RUG = rugby; HOC = field hockey; BB = basketball; AF = American football; ARF = Australian rules football; VB = volleyball; HB = handball; NET = netball; MIX = mixture of team sports; TRA = trained/developmental athletes; INT = international/elite athletes; NAT = national/highly trained athletes; PRO = professional; SEMI = semi-professional; AM = amateur; YTH = youth; CON = control group; INT = intervention group; Sham = sham group; RS = repeated-sprint; RS-G = repeated- sprint group; PLY = plyometric group; REP = representative players; Club = club players; MID = midfielders; FWD = forwards; DEF = defenders; G = guards; CEN = centres; U17 = under 17 players; U18 = under 18 players; U20 = under 20 players; SEN = senior players; VO2max = maximal oxygen consumption; High = high V02max group; Med = medium VO2max group; Low = low VO2max group; SAN = sand training group; GRA = grass training group; TS3 = team sport 3; ST = starting players; N-ST = non-starting players; N-SEL = non-selected players; MG = Melanesian group; N-MG = Non-Melanesian group; Sdec = percentage sprint decrement; yrs = years; hrs = hours; AG = artificial grass; NG = natural grass; cm = centimetre; kg = kilogram; ~ = approximately; * = single group time series. Supplementary Table S3. Summary of exercise protocol information and outcomes from all studies. Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Abt et al. [118] STR 1 × 22 15 m 1:10N (~26 s) AH - Savg: 2.64 ± 0.06 s - - B[La]peak: 1.3 ± 0.2 to 7.6 ± 0.6 mmol·L-1 STR 1 × 22 15 m 1:10N (~26 s) P - Savg: 2.63 ± 0.07 s - - B[La]peak: 1.0 ± 0.1 to 8.7 ± 0.9 mmol·L-1 STR 1 × 22 30 m 1:10N (~45 s) AH - Savg: 4.57 ± 0.22 s - - B[La]peak: 1.2 ± 0.2 to 10.6 ± 0.7 mmol·L-1 STR 1 × 22 30 m 1:10N (~45 s) P - Savg: 4.59 ± 0.15 - - B[La]peak: 1.3 ± 0.2 to 11.1 ± 0.8 mmol·L-1 AbuMoh’d [180] STR 1 × 7 30 m 30 s P - INT, Savg: 3.71 ± 0.05; PLA, Savg: 3.70 ± 0.05 - - INT, B[La] 5’: 9.0 ± 0.1 mmol·L-1; PLA, B[La] 5’: 9.2 ± 0.2 mmol·L-1 Akenhead et al. [57] SHU 1 × 12 25 m (12.5 + 12.5) 20 s P - Sdec: 5.3% - - - Aguiar et al. [95] MDA 1 × 7 34.2 m 25 s AK - INT, Savg: 6.69 ± 0.20 s CON, Savg: 7.31 ± 0.34 s - - - Alemdaroğlu et al. [23] SHU 1 × 6 40 m On 25 s (~17 s) AH - Sbest: 7.35 ± 0.17 s; Stotal: 45.93 ± 0.84 s; Sdec: 4.13 ± 1.81% - - B[La]3’: 9.3 ± 2.5 mmol·L-1 STR 1 × 6 40 m On 25 s (~19 s) AH - Sbest: 5.68 ± 0.20 s; Stotal: 34.90 ± 1.21 s; Sdec: 2.42 ± 1.43% - - B[La]3’: 7.6 ± 1.4 mmol·L-1 SHU 1 × 8 30 m (15 + 15) On 25 s (~19 s) AH - Sbest: 5.64 ± 0.16 s; Stotal: 46.41 ± 1.32 s; Sdec: 2.85 ± 1.51% - - B[La]3’: 7.9 ± 2.1 mmol·L-1 STR 1 × 8 30 m On 25 s (~20 s) AH - Sbest: 4.50 ± 0.15 s; Stotal: 37.21 ± 1.23 s; Sdec: 3.29 ± 0.91% - - B[La]3’: 8.1 ± 1.4 mmol·L-1 Alizadeh et al. [167] STR 1 × 6 35 m 10 s P - High, Sbest: 5.34 ± 0.13 s; Stotal: 33.47 ± 0.99 s; Sdec: 9.6 ± 0.1%; Med, Sbest: 5.39 ± 0.14 s; Stotal: 34.77 ± 0.56 s; Sdec: 9.3 ± 0.2%; Low, Sbest: 6.22 ± 0.39 s; Stotal: 40.56 ± 3.50 s; Sdec: 9.2 ± 0.3% - - High,  B[La]3’: 1.73 to 6.97 mmol·L-1; MED,  B[La]3’: 1.9 to 9.0 mmol·L-1 Almansba et al. [96] MDY 1 × 6 40 20 s P - Sbest: 7.97 ± 0.39 s; Savg: 8.37 ± 0.30 s; Sdec: 4.8 ± 2.0% 6−20: 15.2 ± 1.6 au - B[La]2’: 12.9 ± 1.5 mmol·L-1 ; HRpeak: 189 ± 7 b·min-1 HRav: 195 ± 8 b·min-1  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological STR 1 × 6 40 20 s P - Sbest: 5.75 ± 0.28 s; Savg: 6.16 ± 0.29 s; Sdec: 6.7 ± 3.1% 6−20: 13.9 ± 1.8 au B[La]2’: 11.6 ± 1.2 mmol·L-1; HRpeak: 185 ± 6 b·min-1; HRavg: 178 ± 9 b·min-1 Altimari et al. [181] SHU 1 × 6 40 m (20 + 20) 20 s P - 1TR, Savg: 7.08 ± 0.27 s; Sdec: 5.3 ± 1.3%; 2TR, Savg: 7.16 ± 0.25 s; Sdec: 5.4 ± 1.2%; 3TR, Savg: 7.08 ± 0.27 s; Sdec: 5.6 ± 1.5% - - - Archiza et al. [182] SHU 1 × 6 40 m (20 + 20) 20 s P - Sham, Sbest: 7.50 ± 0.20 s; Savg: 7.90 ± 0.20 s; Sdec: 6.3 ± 3.0%; INT, Sbest: 7.60 ± 0.30 s; Savg: 8.20 ± 0.30 s; Sdec: 7.9 ± 2.4% - - - Attene et al. [115] SHU 1 × 10 30 m (15 + 15) 30 s P - Sbest: 6.41 ± 0.43 s; Stotal: 67.27 ± 4.43 s; Sdec: 10.9 ± 4.3% CR10: 8.6 ± 0.5 au - B[La]3’: 9.5 ± 1.6 mmol·L-1 Ayarra et al. [183] STR 1 × 6 30 m 25 s A - Stotal: 26.03 ± 2.09 s; Sdec: 1.7 ± 3% - - - Aziz et al. [184] STR 1 × 8 40 m 30 s AI - Sbest: 5.45 ± 0.23 s; Stotal: 45.90 ± 1.64 s; Sdec: 5.4 ± 2.7% - - Baldi et al. [185] SHU 1 × 6 40 m (20 + 20) 20 s P - Sbest: 7.13 ± 0.24 s; Sdec: 5.2 ± 1.6% - B[La]peak:17.6 ± 2.6 mmol·L-1 Balsalobre- Fernández et al. [186] STR 1 × 6 35 m 10 s P - - -  CMJAA: -4.2 cm (-9.2 ± 4.8%) - Beato et al. [187] SHU 1 × 6 40 m (20 + 20) 20 s P - STR-G, Sbest: 7.13 ± 0.17 s, Savg:7.46 ± 0.19 s; SHU-G group, Sbest: 7.14 ± 0.18 s, Savg:7.50 ± 0.21 s - - - STR 3 × 7 30 m 20 s P 4 min P - CR10: 6.3 ± 0.5 au - - SHU 3 x 7 40 m 20 s P 4 min P - CR10: 6.4 ± 0.6 au - - Beato et al. [188] SHU 1 × 6 40 m (20 + 20) 20 s P - STR-G, Sbest: 7.30 ± 0.15 s; Savg: 7.56 ± 0.20 s SHU-G, Sbest: ± 7.23 ± 0.32 s; Savg: 7.46 ± 0.31 s - - - STR 3 × 7 30 m 20 s P 4 min P - CR10: 6.1 ± 0.8 au - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological SHU 3 × 7 40 m (20 + 20) 20 s P 4 min p - CR10: 6.4 ± 0.7 - - Beato & Drust [162] STR 3 × 7 30 m 25 s AQ 3 min P - - - HRpeak: 192 ± 12 b·min-1 Beaven et al. [189] STR 1 × 5 40 m On 30 s (~24 s) P - Stotal: 27.58 ± 1.58 s - HRpost: 139 ± 8 b·min-1 Binnie et al. [190] STR 1 × 8 20 m 20 s AW - SAN, Stotal: 30.97 ± 1.58 s; Sdec: 4.8 ± 2.1%; GRA, Stotal: 29.56 ± 1.69 s; Sdec: 4.5 ± 2.2% - - SAN, B[La]peak: 6.5 ± 2.3 mmol·L-1; GRA, B[La]peak: 5.7 ± 2.5 mmol·L-1 Binnie et al. [191] STR 1 × 8 20 m 20 s AK - Sbest: 3.31 s; Stotal: 27.46 s; Sdec: 3.7% - B[La]post: 8.2 mmol·L-1 HRpeak: 160 b·min-1 Binnie et al. [192] STR 1 × 8 20 m 20 s AK - Sbest: 3.34 s; Stotal: 27.94 s; Sdec: 4.4% - - B[La]post: 7.5 mmol·L-1 HRpeak: 163 b·min-1 Blasco- Lafarga et al. [108] MDC 1 × 7 34.2 m 25 s AK - Sbest: 5.72 ± 0.13 s; Savg: 5.91 ± 0.14 s; Stotal: 41.41 ± 0.99 s; Sdec: 3.5 ± 1.6% CR10: 9.1 ± 2.2 au - B[La]3’: 8.5 ± 1.4 mmol·L-1 Borges et al. [193] SHU 1 × 6 40 m (20 + 20) 20 s P - RES, Sbest: 7.35 ± 0.07 s; Savg: 7.70 ± 0.14 s; PLY, Sbest: 7.21 ± 0.18 s; Savg: 7.55 ± 0.22 s - - - Brahim et al. [97] SHU 1 × 6 40 m (20 + 20) 20 s P - Sdec: 2.7 ± 1.3% - - MDB 1 × 12 20 m 40 s P - Sdec: 3.8 ± 2.3% - - MDA 1 × 7 34.2 m 25 s AK - Sdec: 4.3 ± 3.4% - - Brini et al. [154] SHU 1 × 10 30 m 30 s P - Sbest: 5.80 ± 0.21 s; Stotal: 58.99 ± 1.67 s CR10: 4.3 ± 0.5 au - B[La]post: 5.3 ± 1.7 mmol·L-1; HRpeak: 194 ± 2 b·min-1 SHU 1 × 10 30 m 30 s AX - Sbest: 5.88 ± 0.15 s; Stotal: 59.58 ± 1.36 s CR10: 5.0 ± 0.6 au B[La]post: 5.5 ± 2.1 mmol·L-1; HRpeak: 195 ± 2 b·min-1 SHU 1 × 10 30 m 30 s AY - Sbest: 5.91 ± 0.15 s; Stotal: 60.02 ± 1.11 s CR10: 7.4 ± 0.8 au - B[La]post: 6.8 ± 2.2 mmol·L-1; HRpeak: 195 ± 2 b·min-1 SHU 1 × 10 30 m 30 s AZ - Sbest: 5.92 ± 0.11 s; Stotal: 60.10 ± 0.94 s CR10: 8.2 ± 0.8 au - B[La]post: 6.6 ± 2.1 mmol·L-1; HRpeak: 196 ± 2 b·min-1  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Brini et al. [194] SHU 1 × 10 30 m (15 + 15) 30 s P - SSG, Sbest: 5.90 ± 0.11 s; Savg: 5.98 ± 0.68 s; Stotal: 59.78 ± 0.68 s; RS, Sbest: 5.88 ± 0.13 s; Savg: 5.97 ± 1.14 s; Stotal: 59.72 ± 1.14 s - - SSG, HRpeak: 186 ± 4 b·min-1; RS, HRpeak: 189 ± 3 b·min-1 Brini et al. [195] SHU 1 × 10 30 m (15 + 15) 30 s P - Sbest: 5.89 ± 0.10 s; Stotal: 59.60 ± 0.90 s; Sdec: 1.2 ± 0.5% CR10: 7 ± 1 au - B[La]3’: 6.6 ± 2.1 mmol·L-1; HRpeak: 191 ± 1 b·min-1 MD 1 × 10 30 m 30 s P - Sbest: 5.90 ± 0.10 s; Stotal: 59.80 ± 0.90 s; Sdec: 1.3 ± 0.5% CR10: 8 ± 1 au - B[La]3’: 6.8 ± 2.2 mmol·L-1; HRpeak: 195 ± 1 Brini et al. [98] MD 1 × 10 30 m 30 s P - INT, Sbest: 6.91 ± 0.1s; Stotal: 70.90 ± 0.98 s; CON, Sbest: 6.87 ± 0.12 s; CON, Stotal: 69.81 ± 0.62 s INT, CR10: 5.6 ± 1.4 au; CON, CR10: 6.0 ± 1.3 au - INT, B[La]3’: 5.4 ± 2.1 mmol·L-1; HRpeak: 187 ± 3 b·min-1; CON, B[La]3’: 5.8 ± 2.4 mmol·L-1; HRpeak: 187 ± 6 b·min-1 Brini et al. [46] MDA 1 × 10 30 m 30 s P - PRO, Sbest: 8.07 ± 0.03 s; Stotal: 83.35 ± 2.19 s; SEMI, Sbest: 8.21 ± 0.16 s; Stotal: 83.56 ± 2.17 s; PRO, CR10: 6.8 ± 0.6 SEMI, CR10: 6.9 ± 0.6 - PRO, B[La]3’: 8.0 ± 2.0 mmol·L-1; HRpeak: 187 ± 2 b·min-1 SEMI, B[La]3’: 9.5 ± 0.6 mmol·L-1; HRpeak: 189 ± 1 b·min-1 Brocherie et al. [196] STR 1 × 6 35 m 10 s P - Sbest: 4.87 ± 0.14 s; Stotal 31.73 ± 1.13 s; Sdec: 8.7 ± 2.3% - - Brocherie et al. [54] STR 1 × 6 35 m 10 s P - Savg: 5.34 ± 0.25 s; Sdec: 9.5 ± 2.4% 6−20: 15.9 ± 0.9 au  sprint 1−6: ΔL: 17.5 ± 2.5 to 18.0 ± 2.9 cm; Δz: 1.9 ± 0.3 to 2.7 ± 0.3 cm; Fzmax: 2.36 ± 0.18 to 2.41 ± 0.14 N; Kvert: 127.6 ± 17.7 to 91.4 ± 10.4 kN·m-1; Kleg: 13.7 ± 1.7 to 13.8 ± 2.7 kN·m- 1 B[La]4’: 10.5 ± 2.0 mmol·L-1 Brocherie et al. [197] STR 1 × 8 20 m On 20 s (~17 s) P - HYP, Stotal: 27.23 ± 1.15 s; Sdec: 4.0 ± 1.7%; NOR, 27.05 ± 0.81 s; Sdec: 4.3 ± 1.9%; CON, 26.98 ± 1.03 s; Sdec: 5.2 ± 2.1% - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Broderick et al. [141] STR 1 × 3 15 m 20 s P - INT, Sbest: 2.58 ± 0.10 s; Stotal: 7.82 ± 0.32 s; CON, Sbest: 2.58 ± 0.10 s; Stotal: 7.84 ± 0.31 s - - - Buchheit [198] STR 1 × 6 30 m 20 s P - Sbest: 5.73 ± 0.27 s; Savg: 5.90 ± 0.27 s; Sdec: 2.8 ± 0.9% - - - SHU 1 × 6 25 m 25 s AL - Sbest: 3.96 ± 0.15 s; Savg: 4.09 ± 0.17 s; Sdec: 3.2 ± 1.3% Buchheit et al. [59] STR 1 × 6 25 m On 25s (~21 s) AL - Sbest: 3.97 ± 0.15 s; Savg: 4.09 ± 0.16 s Sdec: 2.8 ± 1.2% CR10: 7 ± 1 au - B[La]3’: 9.4 ± 2.4 mmol·L-1; VO2avg: 38.1 ± 5.0 ml·min- 1·kg-1 (% VO2max: 76 ± 10%); HRpeak: 175 ± 11 b·min-1 (% HRmax: 95 ± 6%) SHU 1 × 6 25 m (12.5 + 12.5) On 25 s (~20 s) AL - Sbest: 5.186 ± 0.16 s; Savg: 5.29 ± 0.17 s; Sdec: 2.5 ± 1.0% CR10: 7 ± 1 au - B[La]3’: 9.9 ± 2.0 mmol L-1; VO2avg: 39.7 ± 5.0 ml·min- 1·kg-1 (% VO2max: 79 ± 10%); HRpeak: 177 ± 11.0 b·min-1 (% HRmax: 96 ± 6%) STR 1 × 6 25 m On 25 s (~21 s) AM - Sbest: 3.98 ± 0.14 s; Savg: 4.14 ± 0.17 s Sdec: 3.9 ± 1.5% CR10: 8 ± 1 au - B[La]3’: 10.2 ± 2.4 mmol·L-1; VO2avg: 40.2 ± 4.5 ml·min- 1·kg-1 (% VO2max: 80 ± 9%); HRpeak: 176 ± 11 b·min-1 (% HRmax: 96 ± 6%) SHU 1 × 6 25 m (12.5 + 12.5) On 25 s (~20 s) AM - Sbest: 5.18 ± 0.18 s; Savg: 5.43 ± 0.18 s Sdec: 3.4 ± 2.3% CR10: 8 ± 1 au - B[La]3’: 10.4 ± 2.1 mmol·L-1; VO2avg: 42.2 ± 5.0 ml·min- 1·kg-1 (% VO2max: 84 ± 10%); HRpeak: 178 ± 11 b·min-1 (% HRmax: 97 ± 6%) Buchheit et al. [60] STR 1 × 6 25 m On 25 s (~21 s) AL - Sbest: 3.96 ± 0.15 s; Savg: 4.09 ± 0.17 s Sdec: 3.2 ± 1.3% CR10: 7.2 ± 1.4 au -  B[La]3’: 2.2 ± 0.2 to 9.3 ± 2.4 mmol·L-1; VO2avg: 35.8 ± 4.7 ml·min-1·kg-1 (% VO2max: 77.4 ± 9.3%); HRpeak: 173 ± 9 b·min-1 (% HRmax: 94 ± 5%)  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological SHU 1 × 6 25 m (12.5 + 12.5) On 25 s (~20 s) AL - Sbest: 5.16 ± 0.17 s; Savg: 5.30 ± 0.17 s Sdec: 2.6 ± 1.2% CR10: 7.2 ± 0.8 au -  B[La]3’: 2.2 ± 0.2 to 10.0 ± 1.7 mmol·L-1; VO2avg: 40.4 ± 5.2 ml·min-1·kg-1 (% VO2max: 80.5 ± 10.3%); HRpeak: 173 ± 10 b·min-1 (% HRmax: 94 ± 5%) Buchheit et al. [99] STR 1 × 6 30 m On 25 s (~20 s) AL - Sbest: 4.37 ± 0.17 s; Savg: 4.69 ± 0.20 s Sdec: 6.7 ± 2.5% CR10: 7.4 ± 1.5 au -  B[La]3’:  10.1 ± 2.2 mmol·L-1; HRpeak: 184 ± 7 b·min-1 MDD 1 × 6 ~27.6 m On 25 s (~20 s) AL - Sbest: 4.38 ± 0.17 s; Savg: 4.61 ± 0.29 s Sdec: 4.8 ± 3.6% CR10: 6.9 ± 1.7 au -  B[La]3’:  8 ± 2.3 mmol·L- 1; HRpeak: 181 ± 8 b·min-1 MDE 1 × 6 ~21.2 m On 25 s (~20 s) AL - Sbest: 4.36 ± 0.15 s; Savg: 4.69 ± 0.16 s Sdec: 7.0 ± 3.2% CR10: 6.0 ± 1.6 au -  B[La]3’:  6.1 ± 2.5 mmol·L-1; HRpeak: 178 ± 9 b·min-1 MDF 1 × 6 ~ 19.2 m On 25 s (~20 s) AL - Sbest: 4.39 ± 0.19 s; Savg: 4.73 ± 0.19 s Sdec: 7.1 ± 3.0% CR10: 6.0 ± 1.1 au -  B[La] 3’:  7.4 ± 2.3 mmol·L-1; HRpeak: 180 ± 8 b·min-1 Campa et al. [168] SHU 1 × 6 40 m (20 + 20) 20 s P - EL, Sbest: 7.00 ± 0.30 s; Savg: 7.50 ± 0.40 s; Sdec: 6.3 ± 3.1%; S-EL, Sbest: 7.70 ± 0.20 s; Savg: 7.90 ± 0.20 s; Sdec: 3.4 ± 1.1% - - - Campos et al. [199] SHU 1 × 8 40m (10 + 20 + 10) 20 s P - IT100, Sbest: 8.12 ± 0.20 s; Savg: 8.69 ± 0.36 s; IT86, Sbest: 8.28 ± 0.24 s; Savg: 8.50 ± 0.18 s - - - Campos- Vazquez et al. [200] SHU 1 × 6 40 m (20 + 20) 20 s P - SQ, Sbest: 6.99 ± 0.11 s; Savg: 7.40 ± 0.18 s; TG, Sbest: 7.07 ± 0.18 s; Savg: 7.42 ± 0.15 s - - - Caprino et al. [201] SHU 1 × 10 30 m (15 + 15) 30 s P - Stotal: 58.80 ±2.10 s; Sdec: 2.3 ± 1.0% - -  B[La] 3’: 5.1 ± 1.4 mmol·L-1 to 12.4 ± 2.8 mmol·L-1 Castagna et al. [153] SHU 1 × 10 30 m (15 + 15) 30 s P - Savg: 6.17 ± 0.10 s; Stotal: 60.56 ± 1.60 s; Sdec: 3.4 ± 2.3% - -  B[La]3’: 2.5 ± 0.7 mmol L- 1 to 14.1 ± 3.5 mmol·L-1 SHU 1 × 10 30 m (15 + 15) 30 s AZ - Savg: 6.32 ± 0.10 s; Stotal: 62.15 ± 2.99 s Sdec: 5.0 ± 2.4% - -  B[La]3’: 2.4 ± 0.5 to 13.2 ± 2.9 mmol·L-1  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Castagna et al. [202] SHU 1 × 10 30 m (15 + 15) 30 s P - Sdec: 3.4 ± 2.3% - -  B[La]post: 2.5 ± 0.7 to 13.6 ± 3.1 mmol·L-1;  B[La]3’: 2.5 ± 0.7 to 14.2 ± 3.5 mmol·L-1 Chaouachi et al. [203] STR 1 × 7 30 m 25 s AQ - Savg: 4.50 ± 0.13 s; Stotal: 31.21 ± 1.13 s; Sdec: 6.0 ± 2.5% - - - Charlot et al. [204] STR 1 × 6 25 m 25 s AK - Savg: 3.84 ± 0.17 s: Stotal: 23.10 ± 1.10 s; Sdec: 7.4 ± 3.9% - - - SHU 1 × 6 25 m (12.5 + 12.5) 25 s AK - Savg: 5.32 ± 0.17 s: Stotal: 30.50 ± 2.30 s; Sdec: 4.1 ± 1.3% - - - Chen et al. [205] SHU 1 × 6 40 m (20 + 20) 20 s P - Sbest: 7.50 ± 0.50 s; Stotal: 45.9 ± 3.34 s; Sdec: 3.5 ± 2.5% 6−20: 15 ± 3.6 au - B[La]post: 9.8 ± 2.1 mmol·L-1; HRpeak:171 ± 12 b·min-1 Clifford et al. [34] STR 1 × 20 30 m 30 s P - INT, Sbest: 4.41 ± 0.23 s; Savg: 4.65 ± 0.25 s; PLA: 4.48 ± 0.14 s; Savg: 4.70 ± 0.15 s INT, 6−20: 15 ± 1 au; PLA, 6−20: 14 ± 2 au INT,  CMJAA: -11.8 ± 8.9%; PLA,  CMJAA: -9.6 ± 4.8% INT, CK 24 h: 188 ± 62 to 542 ± 461 u·L-1 (188%); PLA, CK 24 h: 318 ± 145 to 592 ± 321 u·L-1 (86%) Corrêa et al. [206] STR 1 × 6 35 m 10 s P - Stotal: 31.17 ± 1.03 s; Sdec: 8.2 ± 2.77% - - - Costello et al. [127] STR 1 × 20 20 m 20 s A - Savg: 3.43 ± 0.2 s CR10: 9 ± 1.1 - B[La]post: 12.4 ± 2.6 mmol·L- 1: HRavg: 178 ± 8 b·min-1 Cuadrado- Peñafiel et al. [207] SHU 1 × 6 40 m (20 + 20) 30 s P - SOC, Sbest: 7.01 ± 0.22 s; Sdec: 2.7 ± 0.6%; FUT: 7.26 ± 0.19 s; Sdec: 4.4 ± 1.2% - - SOC, B[La]post: 13.7 ± 2.8 mmol·L-1 ; FUT, B[La]post: 14.3 ± 3.4 mmol·L-1 Da Silva et al. [208] SHU 1 × 7 34.2 m 25 s P - Sbest: 6.30 ± 0.24 s; Savg: 6.56 ± 0.23 s; Sdec: 4.0 ± 1.9% - - B[La]peak: 15.4 ± 2.2 mmol·L- 1 Dal Pupo [116] STR 1 × 6 25 m 15 s A - Sbest: 3.80 ± 0.18 s; Savg: 3.98 ± 0.20 s; Sdec: 4.7 ± 2.0% - CMJ AB: 43.52 ± 1.48 to 41.68 ± 1.25 cm (-4.2%) B[La]peak: 11.1 ± 2.4 mmol·L- 1 SHU 1 × 6 25 m (12.5 + 12.5) 15 s A - Sbest: 5.17 ± 0.23 s; Savg: 5.34 ± 0.23 s; Sdec: 3.2 ± 1.4% -  CMJAB: 43.52 ± 1.48 to 40.37 ± 1.28 cm (-7.2%) B[La]peak: 12.2 ± 3.3 mmol·L- 1  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Dal Pupo et al. [209] STR 1 × 6 25 m 15 s A - Sbest: 3.73 ± 0.12 s; Savg: 3.91 ± 0.15 s; Sdec: 4.7 ± 1.8% - - - SHU 1 × 6 25 m (12.5 + 12.5) 15 s A - Sbest: 5.13 ± 0.22 s; Savg: 5.30 ± 0.20 s; Sdec: 3.3 ± 0.9% - - - Daneshfar et al. [210] SHU 1 × 10 30 m (15 + 15) 30 s P - Test, Sbest: 6.35 ± 0.08 s; Stotal: 68.97 ± 0.23 s; Sdec: 9.1 ± 1.1% Retest, Sbest: 6.30 ± 0.08 s; Stotal: 69.25 ± 0.24 s; Sdec: 9.3 ± 1.1% CR10: 8.8 ± 0.1 au - B[La]3’: 10.0 ± 0.1 mmol·L-1 Dardouri et al. [211] SHU 1 × 10 30 m (15 + 15) 30 s P - Sbest: 6.15 ± 0.25 s; Stotal: 63.90 ± 2.50 s; Sdec: 4.1 ± 1.4% - - B[La]3’: 14.8 ± 0.4 mmol·L-1 De Andrade et al. [212] STR 1 × 6 35 m 10 s P - Sbest: 4.43 ± 0.17 s; Savg: 4.91 ± 0.23 s; Stotal: 29.45 ± 1.39 s; Sdec: 11.3 ± 7.6% - - B[La]peak: 13.7 ± 2.4 mmol·L- 1 Delextrat et al. [213] SHU 1 × 10 30 m (15 + 15) 30 s P - M, Stotal: 58.40 ± 2.80 s; Sdec: 4.3 ± 0.4%; F, Stotal: 63.50 ± 2.20 s; Sdec: 3.6 ± 0.9% - - - Delextrat et al. [214] SHU 1 × 6 20 m (10 + 10) On 20 s (~15 s) P - Stotal: 29.00 ± 2.10 s; Sdec: 4.0 ± 2.7% - - - Delextrat et al. [175] SHU 1 × 10 30 m (15 + 15) 30 s P - M, Stotal: 58.01± 3.01 s; Sdec: 4.3 ± 1.5%; F, Stotal: 63.34 ± 2.38 s; Sdec: 3.6 ± 0.3% - - - Dellal et al. [61] STR 1 × 10 20 m 30 s AK - - - - HRpeak:191 b·min-1 (% HRmax: 91%) STR 1 × 10 30 m 30 s AK - - - - HRpeak:198 b·min-1 (% HRmax: 95%) STR 1 × 15 20 m 30 s AK - - - - HRpeak:198 b·min-1; (% HRmax: 95%) Dellal & Wong [100] MDX 1 × 10 20 m 25 s AK - U17, Sbest: 5.39 ± 0.03 s; Savg: 5.47 ± 0.04 s; Stotal: 32.76 ± 0.24 s; Sdec: 1.4 ± 0.6%; U19, Sbest: 5.34 ± 0.03 s; 5.39 ± 0.04 s; Stotal: 32.25 ± 0.26 s; Sdec: 1.0 ± 0.4%; PRO, Sbest: 5.31 ± 0.05 s; Savg: 5.37 ± 0.07 s; Stotal: 32.22 ± 0.42 s; Sdec: 1.2 ± 0.5% - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Dent et al. [131] STR 4 × 6 30 m On 30 s (~25 s) AK 7 min P M, Sbest: set 1, 4.29 ± 0.05 s; set 2, 4.35 ± 0.02 s; set 3, 4.45 ± 0.10 s; set 4, 4.49 ± 0.11 s; Savg: set 1, 4.47 ± 0.9 s; set 2, 4.54 ± 0.12 s; set 3, 4.60 ± 0.13 s; set 4, 4.54 ± 0.12 s; Sdec: set 1, 4.7 ± 1.4%; set 2, 4.9 ± 1.4%; set 3, 5.4 ± 2.0%; set 4, 4.3 ± 1.1% F, Sbest: set 1, 4.74 ± 0.18 s; set 2, 4.87 ± 0.14 s; set 3, 4.96 ± 0.27 s; set 4, 4.97 ± 0.22 s Savg: set 1, 5.09 ± 0.21 s; set 2, 5.17 ± 0.31 s; set 3, 5.24 ± 0.27 s; set 4, 5.23 ± 0.31 s; Sdec: set 1, 7.1 ± 2.1%; set 2, 6.6 ± 2.8%; set 3, 7.2 ± 1.3%; set 4, 7.2 ± 2.8% - - M: set 1,  B[La]3’: 0.9 ± 0.4 to 10.0 ± 1.6 mmol·L-1; set 2, B[La]3’: 11.9 ± 2.9 mmol·L-1; set 3, 11.6 ± 3.3 mmol·L-1; set 4, 11.6 ± 4.0 mmol·L-1; HRpost: set 1, 179 ± 20 b·min- 1; set 2, 175 ± 38 b·min-1, set 3, 188 ± 10 b·min-1; set 4, 189 ± 10 b·min-1; F: set 1,  B[La]3’: 0.8 ± 0.3 to 10.0 ± 3.5 mmol·L-1; set 2, B[La]3’: 12 ± 3.6 mmol·L-1; set 3, 12.0 ± 3.3 mmol·L-1; set 4, 12.2 ± 3.7 mmol·L-1 HRpost: set 1, 189 ± 9 b·min-1; set 2, 190 ± 8 b·min-1; set 3, 191 ± 6 b·min-1; set 4, 190 ± 8 b·min-1 Donghi et al. [215] SHU 1 × 6 40 m (20 + 20) 20 s P - - CR10: 5 ± 1.2 au - - Doyle et al. [216] STR 1 × 6 20 m On 15 s (~12 s) AW - Sbest: 3.43 ± 0.16 s; Stotal: 21.42 ± 0.97 s; Sdec: 4.4 ± 0.3% - - - Dupont et al. [217] STR 1 × 7 30 m 20 s A - Savg: 4.60 ± 0.14 s - - - Dupont et al. [86] STR 1 × 15 40 m 25 s AZ - Savg: 6.41 ± 0.31 s; Sdec: 8.6 ± 3.2% - - B[La]3’: 13.8 ± 3.1 mmol·L-1; VO2avg: 60.5 ± 4.3 ml·min- 1·kg-1 Eliakim et al. [124] STR 1 × 12 20 m On 20 s (~ 17 s) P - Sbest: 3.23 ± 0.17 s; Savg: 3.24 ± 0.04 s; Stotal: 38.91 ± 0.52 s; Sdec: 2.3 ± 0.6 CR10: 7 ± 1 au - HRavg: 177 ± 6 b·min-1 HRpost: 181 ± 4 b·min-1 Elias et al. [218] STR 1 × 6 20 m On 30 s (~27 s) P - PAS, Stotal: 18.53 ± 0.28 s; COL, Stotal: 18.62 ± 0.46 s; CWT, Stotal: 18.63 ± 0.45 s - - - Elias et al. [219] STR 1 × 6 20 m On 30 s (~27 s) P - PAS, Stotal: 18.66 ± 0.37 s; COL, Stotal: 18.50 ± 0.47 s; CWT, Stotal: 18.68 ± 0.39 s - - - Eniseler et al. [220] SHU 1 × 6 40 m (20 + 20) 20 s P - RS, Sbest: 6.75 ± 0.19 s; Savg: 7.13 ± 0.17 s; Sdec: 5.5 ± 0.8% SSG, Sbest: 6.73 ± 0.19 s; Savg: 7.12 ± 0.17 s; Sdec: 5.8 ± 1.1% - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Eryilmaz & Kaynak [221] STR 1 × 10 20 m 20 s AK - Sbest: 2.97 ± 0.10 s; Savg: 3.21 ± 0.10 s; Sdec: 8.0 ± 2.7% - - - Eryilmaz et al. [37] STR 1 × 10 20 m 20 s AK - Savg: 4.28 ± 0.10 s - - - Essid et al. [222] SHU 1 × 6 30 m (15 + 15) On 20 s (~14 s) P - Sbest: 6.19 ± 0.03 s; Savg: 6.78 ± 0.03; Sdec: 8.7 ± 0.0% - - - Farjallah et al. [223] SHU 1 × 6 40 m (20 + 20) 20 s P - INT, Savg: 7.18 ± 0.23 s; PLA, Savg: 7.34 ± 0.03 s - - - Figueira et al. [119] SHU 3 × 10 30 m (15 + 15) 30 s P 5 min P Stotal: 59.22 ± 2.10 s; Sdec: 3.6 ± 1.6% - - B[La]3’: 13.0 ± 2.3 mmol·L-1 ; HRpeak: 174 ± 7 b·min-1 STR 3 × 20 15 m 15 s P 5 min P Stotal: 53.66 ± 1.56 s; Sdec: 4.9 ± 2.1% - - B[La]3’: 8.5 ± 3.4 mmol·L-1 HRpeak: 174 ± 7 b·min-1 Freitas et al. [227] SHU 1 × 10 30 m (15 + 15) 30 s P - Stotal: 57. 50 ± 2.89 s; Sdec: 2.9 ± 1.0% - - - Fornasier-Santos et al. [224] STR 1 × 10 40 m On 30 s (~25 s) P - - HYP, CR10: 9.2 ± 0.7 au NOR, CR10: 9.2 ± 0.7 au - HYP: 13.7 ± 4.3 mmol·L-1 NOR: 13.0 ± 4.2 mmol·L-1 STR 2 × 8 40 m NR P 3 min P - CR10: 8.3 ± 0.5 - 10.2 ± 3.3 mmol·L- Fort- Vanmeerhaeghe et al. [225] SHU 1 × 10 30 m (15 + 15) 30 s P - Sbest: 6.20 ± 0.20 s; Savg: 6.34 ± 0.19 s - - - Fortin & Billaut [226] STR 1 × 12 20 m 20 s AK - Sham, Sbest: 3.07 ± 0.13 s; Stotal: 39.69 ± 1.34 s; INT, Sbest: 3.05 ± 0.08 s; Stotal: 39.79 ± 1.65 s - - - Gabbett [228] STR 1 × 6 20 m On 15 s (~12 s) AJ - Stotal: 21.50 ± 1.20 s; Sdec: 5.6 ± 1.6% - - B[La]post: 9.3 ± 2.0 mmol·L-1 HRpeak: 182 ± 6 b·min-1 Gabbett et al. [89] STR 1 × 12 20 m On 20 s (~17 s) P - Stotal: 38.70 ± 2.30 s - - - Gabbett et al. [229] STR 1 × 12 20 m On 20 s (~17 s) P - ST, Stotal: 38.30 ± 2.80 s; N-ST, Stotal: 38.90 ± 3.20; N-SEL, Stotal: 39.10 ± 3.30 - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Gabbett et al. [230] STR 1 × 6 20 m On 15 s (~12 s) AJ - INT, Stotal: 21.16 ± 1.06 s; CON, Stotal: 20.71 ± 0.52 s - - - Galvin et al. [231] STR 1 × 10 30 m 30 s P - HYP, Stotal: 32.20 ± 1.10 s; Sdec: 4.0 ± 3.0%; NOR, Stotal: 32.70 ± 1.20 s; Sdec: 5.1 ± 3 .9% - - - Galy et al. [177] STR 1 × 6 25 m 25 s AK - MG, Sbest: 3.77 ± 0.19 s; Savg: 3.99 ± 0.17 s; Stotal: 23.96 ± 1.05 s; Sdec: 5.9 ± 3.1%; N-MG, Sbest: 3.92 ± 0.19 s; Savg: 4.09 ± 0.17 s; Stotal: 24.55 ± 1.01 s; Sdec: 4.4 ± 1.8% - - - SHU 1 × 6 25 m (12.5 + 12.5) 25 s AK - MG, Sbest: 5.29 ± 0.19 s; Savg: 5.47 ± 0.19 s; Stotal: 32.79 ± 1.14 s; Sdec: 3.4 ± 1.0%; N-MG, Sbest: 5.31 ± 0.18 s; Savg: 5.53 ± 0.15 s; Stotal: 33.21 ± 0.92 s; Sdec: 4.3 ± 0.8% - - - Gantois et al. [232] STR 1 × 6 30 m 20 s P - Sbest: 4.59 ± 0.24 s; Savg: 4.82 ± 0.31 s; Stotal: 27.60 ± 6.77 s; Sdec: 5.3 ± 2.9% - - - Gantois et al [14] STR 1 × 6 30 m 20 s P - RS, Sbest: 4.56 ± 0.24 s; Savg: 4.83 ± 0.38 s; Stotal: 29.00 ± 2.30; Sdec: 6.4 ± 3.5%; CON, Sbest: 4.64 ± 0.24 s; Savg: 4.87 ± 0.22; Stotal: 29.08 ± 1.56 s; Sdec: 4.1 ± 1.8% - - - Gantois et al. [233] STR 1 × 6 30 m 20 s P - Sbest: 4.58 ± 0.21 s; Savg: 4.84 ± 0.31; Stotal: 29.00 ± 1.91 s; Sdec: 7.6 ± 5.8% - - - García- Unanue et al. [169] STR 1 × 7 30 m 20 s P - ELT, Savg: 4.37 ± 0.15 s; Sdec: 4.2 ± 1.4%; AM, Savg: 4.67 ± 0.18 s; Sdec: 6.4 ± 2.2% - ELT,  CMJ AA: 35.7 ± 6.0 to 34.0 ± 4.3 cm (-4.8%) AM,  CMJ AA: 33.8 ± 4.2 to 31.8 ± 3.6 cm (-5.9%) - Gatterer et al. [234] SHU 1 × 6 40 m (20 + 20) 20 s P - NOR, Sbest: 7.18 ± 0.24 s; Savg: 7.60 ± 0.19 s; Sdec: 5.8 ± 1.9%; HYP, 7.28 ± 0.21 s; Savg: 7.66 ± 0.32 s; Sdec: 5.2 ± 2.6% - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Gharbi et al. [83] SHU 1 × 2 30 m (15 + 15) 30 s P - Sbest: 6.26 ± 0.24 s; Stotal: 12.63 ± 0.47 s; Sdec: 1.0 ± 0.7% - - B[La]3’: 1.8 ± 0.6 to 5.7 ± 1.2 mmol·L-1 SHU 1 × 3 30 m (15 + 15) 30 s P - Sbest: 6.18 ± 0.23 s; Stotal: 18.75 ± 0.61 s; Sdec: 1.5 ± 1.0% - - B[La]3’: 1.8 ± 0.6 to 9.4 ± 1.7 mmol·L-1 SHU 1 × 4 30 m (15 + 15) 30 s P - Sbest: 6.17 ± 0.21 s; Stotal: 25.05 ± 0.81 s; Sdec: 2.0 ± 1.1% - - B[La]3’: 1.8 ± 0.6 to 9.6 ± 1.9 mmol·L-1 SHU 1 × 5 30 m (15 + 15) 30 s P - Sbest: 6.29 ± 0.20 s; Stotal 32.36 ± 1.23 s; Sdec: 2.6 ± 1.4% - - B[La]3’: 1.8 ± 0.6 to 10.5 ± 2.6 mmol·L-1; SHU 1 × 9 30 m (15 + 15) 30 s P - Sbest: 6.28 ± 0.23 s; Stotal: 58.68 ± 2.38 s; Sdec: 3.9 ± 1.3% - - B[La]3’: 1.8 ± 0.6 to 12.6 ± 2.3 mmol·L-1; SHU 1 × 10 30 m (15 + 15) 30 s P - Sbest: 6.23 ± 0.23 s; Stotal: 64.96 ± 2.57 s; Sdec: 4.5 ± 1.4% - - B[La]3’: 1.8 ± 0.6 to 12.7 ± 1.0 mmol·L-1 Gharbi et al. [235] SHU 1 × 10 30 m (15 + 15) 30 s P - Sbest: 6.10 ± 0.20 s; Stotal: 63.20 ± 2.20 s; Sdec: 3.5 ± 1.1% - - B[La]3’: 15.3 ± 2.1 mmol·L-1 Gibson et al. [101] MDA 1 × 6 40 m 25 s P - Sbest: 7.11 ± 0.25 s; Stotal: 44.40 ± 1.62 s; Sdec: 3.6 ± 1.2% - - - Girard et al. [236] STR 1 × 6 35 m 10 s P - Savg: 5.36 ± 0.29 s; Sdec: 8.6 ± 2.8% - - - Girard et al. [149] STR 1 × 6 20 m 20 s P - Savg: 3.23 ± 0.13 s; Sdec: 2.8 ± 1.7% - ΔL: 13.6 ± 2.1 to 15.4 ± 2.7 cm; Δz: 1.7 ± 0.4 to 2.2 ± 0.4 cm; Fzmax: 2.0 ± 0.28 to 2.1 ± 0.26 N; Kvert: 120 ± 9.3 to 97 ± 5.2 kN·m-1; Kleg: 15.0 ± 10.0 to 13.7 ± 7.0 kN·m- 1 - González- Frutos et al. [237] STR 1 × 6 30 m 30 s AK Savg: 4.89 ± 0.07 s - - - Gonzalo-skok et al. [102] SHU 1 × 6 40 m (20 + 20) 20 s P - INT, Sbest: 7.16 ± 0.23 s; Savg: 7.52 ± 0.23 s; Sdec: 5.1 ± 1.8%; CON, Sbest: 7.17 ± 0.24 s; 7.50 ± 0.24 s; Sdec: 4.6 ± 1.8% - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological MDG 1 × 5 25 m (5 m per turn) 20 s P - INT, Sbest: 6.58 ± 0.21 s; Savg: 6.86 ± 0.25 s; Sdec: 2.0 ± 0.7; CON, Sbest: 6.56 ± 0.3; Savg: 6.84 ± 0.22 s; Sdec: 2.3 ± 1.5% - - - Goodall et al. [238] STR 1 × 12 30 m 30 s P - Sbest: 4.23 ± 0.13 s; Savg: 4.68 ± 0.08 s; - -  B[La]3’: 3.1 ± 1.4 to 12.8 ± 3.0, mmol∙L-1; B[La]post sprint 1: 2.7 mmol∙L-1; sprint 3: 4.8 mmol∙L1; sprint 5: 7.2 mmol∙L-1; sprint 7: 9.1 mmol∙L-1; sprint 9: 10.4 mmol∙L-1; sprint 11: 11.6 mmol∙L-1 Hamlin et al. [239] STR 1 × 10 40 m On 30 s (~24 s) P - CWT, Savg: 6.36 ± 0.40 s; ARC: 6.38 ± 0.50 s - - CTWI, B[La]3’: 13.6 ± 2.6 mmol·L-1; HRavg: 171 ± 9 b·min-1; AR, B[La]3’: 14.2 ± 2.3 mmol·L-1; HRavg: 173 ± 11 b·min-1 Hamlin et al. [240] STR 1 × 8 20 m On 20 s (~17 s) P - NOR, Stotal: 27.40 ± 3.20; Sdec: 3.5 ± 1.2%; HYP, Stotal: 27.50 ± 3.90 s; Sdec: 3.5 ± 1.3% - - - Hammami et al. [241] SHU 1 × 6 40 m (20 + 20) 20 s P - INT, Sbest: 7.13 ± 0.32 s; Savg: 7.39 ± 0.33 s, Stotal: 44.4 ± 2.0 s; Sdec: 3.7 ± 1.4% INT, Sbest: 7.21 ± 0.13 s; Savg: 7.44 ± 0.15 s, Stotal: 44.6 ± 0.9 s; Sdec: 3.1 ± 1.7% - - - Haugen et al. [62] STR 1 × 12 20 m 60 s P - INT, Sbest: 3.11 ± 0.17 s; Savg: 3.16 ± 0.17 s CON, Sbest: 3.02 ± 0.17 s; Savg: 3.07 ± 0.17 s - - INT, B[La]post: 3.8 ± 1.3 mmol·L-1; HRpeak (% HRmax): 85 ± 4% CON, B[La]post: 3.5 ± 1.4 mmol·L-1; HRpeak (% HRmax): 86 ± 4% Haugen et al. [128] STR 1 × 15 20 m 60 s P - Sbest: 2.94 ± 0.15; Savg: 2.98 ± 0.15 CR10: 3.8 ± 1.2 au - B[La]post: 4.4 ± 1.8 mmol·L-1 Hermassi et al. [242] STR 1 × 6 30 m On 20 s (~16 s) P - Sbest: 4.42 ± 0.14; Savg: 4.57 ± 0.12; Stotal: 27.40 ± 0.70 s; 3.4 ± 1.6% - - - SHU 1 × 6 30 m (15 + 15) On 20 s (~14 s) P - Sbest: 5.97 ± 0.36; Savg: 6.23 ± 0.25; Stotal: 37.40 ± 1.50 s; Sdec: 4.5 ± 3.3% - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Higham et al. [90] STR 1 × 6 30 m On 20 s (~16 s) P - Stotal: 24.76 ± 0.62 s - - - Hollville et al. [243] STR 1 × 6 20 m On 20 s (~17 s) P - Sbest: 3.14 s± 0.12 s; Stotal: 19.30 ± 0.60 s; Sdec: 2.4 ± 1.3% CR10: 4.5 ± 1.6 au HRpost (% HRmax): 88 ± 5% Howatson et al. [35] STR 1 × 15 30 m 60 s P - Sbest: 4.33 ± 0.21 s; Savg: 4.49 ± 0.09 s; Sdec: 4.5 ± 1.5% - -  CK 24 h: 158 ± 56 to 776 ± 312 u·L-1 (385%) Iaia et al. [120] STR 1 × 15 40 m 30 s P - SEP, Stotal: 86.09 ± 6.30 s; Sdec: 5.0 ± 2.3%; SEM, Stotal: 83.81 ± 2.37 s; Sdec: 4.1 ± 1.3% - - - Iaia et al. [19] STR 1 × 6 5 s (~30 m) 15 s P - - - - B[La]post : 3.1 ± 0.8 to 9.3 ± 1.6 mmol·L-1 STR 1 × 6 5 s (~30 m) 30 s P - - - - B[La]post : 3.5 ± 1.1 to 6.6 ± 1.8 mmol·L-1 STR 1 × 15 40 m 30 s P - RS15, Stotal: 92.91 ± 4.66 s; Sdec: 5.9 ± 2.2%; RS30, Stotal: 91.45 ± 4.35 s; Sdec: 5.2 ± 2.1% - - - Impellizzeri et al. [170] SHU 1 × 6 40 m (20 + 20) 20 s P - Test, Sbest: 6.90 ± 0.09 s; Savg: 7.20 ± 0.11; Sdec: 4.3 ± 1.2%; Retest, Sbest: 6.92 ± 0.10 s; Savg: 7.19 ± 0.14 s; Sdec: 3.8 ± 1.4% - - - Impellizzeri et al. [170] SHU 1 × 6 40 m (20 + 20) 20 s P - PRE, Sbest: 6.94 ± 0.15 s; Savg: 7.32 ± 0.13 s; Sdec: 5.4 ± 2.2%; ELY, Sbest: 6.87 ± 0.17 s; Savg: 7.16 ± 0.15 s; Sdec: 4.3 ± 1.7%; MID, Sbest: 6.93 ± 0.15 s; Savg: 7.22 ± 0.14 s; Sdec: 4.2 ± 1.6%; END, Sbest: 6.92 ± 0.15 s; Savg: 7.20 ± 0.13 s; Sdec: 4.0 ± 1.7% - - - Impellizzeri et al. [170] SHU 1 × 6 40 m (20 + 20) 20 s P - PRO, Sbest: 6.88 ± 0.19 s; Savg: 7.12 ± 0.17 s; Sdec: 3.3 ± 1.5%; M-PRO, Sbest: 6.83 ± 0.18 s; Savg: 7.20 ± 0.19 s; Sdec: 5.1 ± 1.8%; AM, Sbest: 7.08 ± 0.23 s; Savg: 7.55 ± 0.25 s; Sdec: 6.1 ± 2.0% - - - Ingebrigtsen et al. [244] STR 1 × 7 35 m 25 s A - Savg: 5.25 ± 0.19 s - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Ingebrigtsen et al. [171] STR 1 × 7 35 m 25 s A - EL, Savg:.5.24 ± 0.24 s; Sdec: 8.3 ± 5.3%; S-EL, Savg: 5.26 ± 0.18 s; Sdec: 6.4 ± 3.7% - - EL, HRpeak: 179 ± 9; S-EL, HRpeak: 188 ± 7 Iacono et al. [42] SHU 1 × 6 40 m (20 + 20) On 20 s (~14 s) P - SSG, Sbest: 5.30 ± 0.15 s; Savg: 5.48 ± 0.15; Sdec: 3.4 ± 0.5% RS, Sbest: 5.31 ± 0.22 s; Savg: 5.48 ± 0.18; Sdec: 3.3 ± 1.0% - - - Izquierdo et al. [140] STR 1 × 6 15 m 60 s P - PLA, Savg: 2.45 ± 0.06 s; INT, Savg: 2.39 ± 0.06 s - - - Jiménez- Reyes et al. [246] STR 1 × 10 40 m 30 s P - - -  sprint 1−10: V0:  15.1 ± 1.3%, F0:  5.9 ± 4.5%; P0:  20.1 ± 3.3%, RF:  6.8 ± 2.0%, DRF:  14.0 ± 6.0 - Johnston & Gabbett [40] STR 1 × 12 20 m On 20 s (~17 s) AW - Sbest: 3.09 ± 0.04 s; Savg: 3.49 ± 0.14 s; Stotal: 41.89 ± 0.20 s; Sdec: 11.4 ± 4.5% 6−20: 12.3 ± 1.2 au - HRpeak: 166 ± 9 b·min-1 HRavg: 154 ± 9 b·min-1 Joo [110] MDA 1 × 7 34.2 m 25 s A - Stotal: 45.7 ± 2.6 s - - - Jorge et al. [247] MDA 1 × 7 34.2 m 25 AL - U20 ELY, Savg: 6.68 ± 0.16 s; Sdec: 4.3 ± 1.0%; U20 MID, Savg: 6.20 ± 0.13 s; Sdec: 4.1 ± 1.0%; U20 END, Savg: 6.40 ± 0.14 s; Sdec: 4.0 ± 1.0%; U17 ELY, Savg: 7.01 ± 0.21 s; Sdec: 5.3 ± 2.0%; U17 MID, Savg: 6.25 ± 0.16 s; Sdec: 4.5 ± 1.8%; U17 END: Savg: 6.32 ± 0.13 s; Sdec: 3.8 ± 1.3% - - - Kaplan [109] MDA 1 × 7 34.2 m 25 AL - Sbest: 7.37 ± 0.26 s; Savg: 7.57 ± 0.25 s; Sdec: 4.4 ± 1.7% - - - Keir et al. [25] STR 1 × 6 35 m 10 s P - - - - B[La]peak: 14.8 ± 2.8 mmol·L- 1; VO2avg: 45.6 ± 9.4 ml·min- 1·kg-1; HRpeak: 182 ± 10 b·min-1 Keogh [172] STR 1 × 6 40 m On 30 s (~25 s) AK - REP, Sdec: 13.1 ± 1.0; Club, Sdec: 12.7 ± 1.4 - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Kilduff et al. [248] SHU 1 × 6 40 m 20 s P - Sbest: 6.72 ± 0.16 s; Savg: 7.01 ± 0.16 s; Stotal: 42.09 ± 0.94 s - - - Klatt et al. [36] SHU 4 × 6 40 m (20 + 20) 30 s P 5 min P U20, Sbest: 6.99 ± 0.17 s; Savg: 7.39 ± 0.26 s SEN, Sbest: 7.12 ± 0.29 s; Savg: 7.65 ± 0.32 s; U20, CR10: 8.7 ± 1.2 au SEN, CR10: 8.3 ± 2.0 au U20, CMJAC: 37.5 ± 5.1 cm to 39 ± 4.7 cm (4.0%) SEN, CMJAC: 31.6 ± 3.9 cm to 34.0 ± 3.9 cm (7.6%) B[La]post: 10.2 ± 2.6 mmol·L-1 U20,  CK 24 h: 285 ± 155 to 354 ± 134 u·L-1 (24%) SEN,  CK 24 h: 214 ± 82 to 443 ± 207 u·L-1 (47%) Krakan et al. [249] STR 1 × 6 25 m 25 s P - RS, Sbest: 3.78 ± 0.08 s; Savg: 3.97 ± 0.10 s; Sdec: 5.0 ± 3.2% PLY, Sbest: 3.74 ± 0.11 s; Savg: 3.96 ± 0.14 s, Sdec: 5.8 ± 0.1 RS, CR10: 7.3 ± 1.5 au PLY, CR10: 8 ± 1.1 au - RS, B[La]post: 13.1 ± 2.5 mmol·L-1 PLY, B[La]post: 14.8 ± 2.3 mmol·L-1 Krueger et al. [250] STR 1 × 6 30 m On 25 s (~21 s) P - CWI, Stotal: 26.23 ± 1.06 s; CON, Stotal: 26.05 ± 0.69 s - - - Lakomy et al. [78] STR 1 × 6 40 m 30 s AW - Savg: 5.97 ± 0.40 s; Sdec: 4.2 ± 2.4% - - - STR 1 × 6 40 m 30 s PR - Savg: 6.03 ± 0.52 s; Sdec: 3.9 ± 1.3% - - - Lapointe et al. [251] STR 1 × 12 30 m 20 s AK - CON, Sbest: 4.83 ± 0.36 s; Savg: 5.18 ± 0.51 s; Sdec: 7.1 ± 3.1%; INT, Sbest: 4.80 ± 0.35 s; Savg: 5.16 ± 0.47 s; Sdec: 7.3 ± 3.2% CON, CR10: 8 ± 1.2 au INT, CR10: 7.5 ± 1.1 au - CON, B[La]1’: 14.0 ± 2.4 mmol·L-1; INT, B[La]1’ : 13.5 ± 1.5 mmol·L-1 Le Rossignol et al. [173] STR 1 × 6 30 m On 20 s (~16 s) P - SEL, Stotal: 25.26 ± 0.55 s; N-SEL, Stotal: 25.92 ± 0.8 s - - - Little & Williams [121] STR 1 × 15 40 m 1:6N (~34 s) P - Savg: 5.73 ± 0.07 s 6−20: 14.4 ± 1.0 au - B[La]2’ : 9.6 ± 0.6 mmol·L-1 ; HRavg (% HRmax): 85.8 ± 0.8% STR 1 × 15 40 m 1:4N (~22 s) P - Savg: 5.93 ± 0.19 s 6−20: 17.1 ± 0.4 au - B[La]2’ : 14.1 ± 1.0 mmol·L- 1 ; HRavg (% HRmax): 89.2 ± 1.9% STR 1 × 40 15 m 1:6N (~16 s) P - Savg: 2.59 ± 0.05 s 6−20: 17.3 ± 0.5 au - B[La]2’ : 8.8 ± 1.1 mmol·L-1 ; HRavg (% HRmax) : 86.8 ± 1.0% STR 1 × 40 15 m 1:4N (~10 s) P - Savg: 2.65 ± 0.10 s 6−20: 18.8 ± 0.4 au - B[La]2’ : 13.0 ± 1.7 mmol·L- 1 ; HRavg (% HRmax): 89.3 ± 1.2%  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Lockie et al. [252] STR 1 × 7 30 m On 20 s (~16 s) AX - FSH, Savg: 32.08 ± 1.31 s; EXP, Savg: 31.67 ± 0.76 s - - - Lockie et al. [253] STR 1 × 6 20 m On 15 s (~11s) AX - Stotal: 31.95 ± 1.06 s - - - Lockie et al. [254] STR 1 × 7 20 m On 20 s (~15s) AX - Stotal: 31.95 ± 1.06 s - - - Lombard et al. [255] STR 1 × 6 30 m On 25 s (~21 s) AX - Stotal: 26.77 ± 0.96 s - - - Madueno et al. [24] SHU 1 × 12 20 m (15 + 5) 20 s P - - CR10: 6.5 ± 0.5 au -  B[La]post: 2.0 to 6.8 mmol·L-1;  B[La]5’: 4.8 mmol·L-1; VO2avg: 33.3 ± 4.0 mL·kg-1· min-1; VO2avg (% VO2max): 73.1 ± 9.8%; HRavg: 166 ± 8 b·min-1 (% HRmax: 83 ± 6%) SHU 1 × 12 20 m (15 + 5) 20 s AZ - - CR10: 6.0 ± 0.5 au -  B[La]post: 2.0 to 8.6 mmol·L-1; B[La]5’: 6.3 mmol·L-1; VO2avg: 37.7 ± 7.1 mL·kg-1·min-1 (% VO2max: 82.5 ± 14.9%); HRavg: 173 ± 5 b·min-1 (% HRmax: 86 ± 2%) Maggioni et al. [16] SHU 3 × 6 40 m (20 + 20) 20 s P 3 min P - CR10: 6.1 ± 2.7 au - - Mancha- Triguero et al. [139] STR 1 × 5 14 m 30 s A - M, Sbest: 2.48 ± 0.18 s; Savg: 2.65 ± 0.16 s; Stotal: 13.27 ± 0.83 s; F, Sbest: 2.70 ± 0.16 s; Savg: 2.99 ± 0.15 s; Stotal: 14.98 ± 0.73 s - - - Marcelino et al. [256] STR 1 × 12 20 m 20 s Ak - SSG 1, Sbest: 3.20 ± 0.10 s; Savg: 3.36 ± 0.10; Sdec: 5.3 ± 3.9%; SSG 2, Sbest: 3.18 ± 0.07 s; Savg: 3.37 ± 0.07 s; Sdec: 6.1 ± 3.3% - - - Matzenbacher et al. [257] SHU 1 × 6 40 m (20 + 20) 20 s P - PRE, Sbest: 7.13 ± 0.26 s; Savg: 7.49 ± 0.34 s; Sdec: 4.9 ± 1.7%; END, Sbest:7.15 ± 0.24 s; Savg: 7.42 ± 0.27 s; Sdec: 3.8 ± 1.9% - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological McGawley & Andersson [258] STR 1 × 6 30 m On 20 s (~16 s) P - Condition 1, Sbest: 27.70 ± 0.50 s; Sdec: 4.7 ± 1.6%; Condition 2, Sbest: 26.70 ± 0.90 s; Sdec: 5.2 ± 1.1% - - - Meckel et al. [259] STR 1 × 6 30 m 30 s P - PRE, Stotal: 22.50 ± 0.60 s; Sdec: 2.9 ± 0.3%; MID, Stotal: 23.70 ± 0.63 s; Sdec: 2.3 ± 0.2%; END, Stotal: 23.51 ± 0.62 s; Sdec: 2.2 ± 0.2% - - - Meckel et al. [260] STR 1 × 12 20 m On 20 s (~17 s) P - Stotal: 39.70 ± 0.60 s; Sdec: 5.0 ± 0.5% CR10: 6.9 ± 0.4 au -  B[La]2’: 2.0 ± 0.1 to 8.8 ± 0.7 mmol·L-1; HRpeak: 182 ± 2 b·min-1 Meckel et al. [261] STR 1 × 6 40 m On 30 s (~24 s) P - Sbest: 5.60 ± 0.26 s; Stotal: 35.10 ± 1.50 s; Sdec: 4.8 ± 1.9% CR10: 4.9 ± 1.4 au - B[La]2’ : 11.3 ± 2.5 mmol·L-1; HRpeak: 179 ± 8 b·min-1 STR 1 × 12 20 m On 20 s (~17 s) P - Sbest: 3.10 ± 0.10 s; Stotal: 38.80 ± 1.20 s; Sdec: 5.0 ± 2.0% CR10: 4.0 ± 1.3 au - B[La]2’ : 10.5 ± 1.8 mmol·L-1; HRpeak: 184 ± 8 b·min-1 Meckel et al. [262] STR 1 × 6 30 m 30 s P - Stotal: 27.71 ± 1.40 s; Sdec: 1.6 ± 0.7% CR10: 5.4 ± 1.5 au - B[La]2’ : 10.1 ± 2.1 mmol·L-1; HRpeak: 171 ± 7 b·min-1 Meckel et al. [263] STR 1 × 12 20 m On 20 s (~17 s) P - Stotal: 37.80 ± 1.40 s; Sdec: 4.4 ± 1.5% CR10: 5.2 ± 1.3 au - B[La]2’ : 6.7 ± 1.1 mmol·L-1; HRpeak: 174 ± 9 b·min-1 Michalsik et al. [264] STR 1 × 7 30 m 25 s AQ - Sbest: 4.09 ± 0.12 s; Savg: 4.30 ± 0.13 s - - - Mohr et al. [265] STR 1 × 5 30 m 25 s AK - Savg: 4.58 ± 0.15 s - - - Mohr et al. [266] STR 1 × 5 30 m 25 s AK - SEP, Sbest: 4.34 ± 0.05 s; Savg: 4.45 ± 0.05 s; SEM, Sbest: 4.32 ± 0.06 s; Savg: 4.41 ± 0.07 s - - - Mohr et al. [267] STR 1 × 3 30 m 25 s AK - Stotal: 13.36 ± 0.11 s - - - Moncef et al. [268] SHU 1 × 6 40 m (20 + 20) On 20 s (~14 s) P - Savg: 6.38 ± 0.86 s - - - Morcillo et al. [48] STR 1 × 12 30 m 30 s P - Sbest: 4.09 ± 0.05 s; Sdec: 3.7 ± 1.5% - - B[La]peak: 9.5 ± 2.3 mmol·L-1 Moreira et al. [269] STR 1 × 5 30 m 25 s AQ - Stotal: 4.65 ± 0.68 s - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Mujika et al. [164] STR 1 × 6 30 m On 30 s (~26 s) AL - U17, Savg: 4.43 ± 0.11 s; Stotal: 26.61 ± 0.53 s; Sdec: 4.1 ± 1.1%; U18, Savg: 4.39 ± 0.12 s; Stotal: 26.34 ± 0.94 s; Sdec: 4.6 ± 1.1% - - U17, B[La]peak: 10.9 ± 1.7 mmol·L-1; U18, B[La]peak: 12.3 ± 1.5 mmol·L-1 Müller et al. [270] STR 1 × 6 35 m 10 s P - - -  CMJ AC: 36.1 ± 5.7 to 34.4 ± 4.9 cm (-4.8%) B[La]post: 11.2 ± 4.4 mmol·L- 1; B[La]5’: 15.0 ± 3.9 HRpeak: 174 ± 20 b·min-1 Nakamura et al. [272] SHU 1 × 6 30 m (15 + 15) On 20 s (~15 s) P - Sbest: 5.62 ± 0.16 s; Savg: 6.03 ± 0.18 s; Sdec: 7.4 ± 2.5% - - B[La]3’: 10.6 ± 2.1 mmol·L-1 ; HRpeak: 180 ± 6 b·min-1 Nascimento et al [273] SHU 1 × 8 40 m (10 + 20 + 10) 20 s (~14 s) P - CON, Sbest: 8.53 ± 0.34 s; Savg: 9.09 ± 0.39 s; Sdec: 6.5 ± 1.1%; INT, Sbest: 8.14 ± 0.18 s; Savg: 8.53 ± 0.15 s; Sdec: 4.8 ± 0.8% - - CON, B[La]peak: 13.2 ± 2.7 mmol·L-1 INT, B[La]peak: 16.2 ± 2.8 mmol·L-1 Nedrehagen & Saeterbakken [274] SHU 1 × 6 40 m (20 + 20) 30 s P - INT, Savg: 7.79 ± 0.37 s CON, Savg: 7.79 ± 0.5 - - - Nikolaidis et al. [275] STR 1 × 10 20 m On 30 s (~27 s) AQ - Sbest: 3.14 ± 0.11 s; Savg: 3.24 ± 0.11 s; Sdec: 3.4 ± 1.6% - - - Okuno et al. [271] SHU 1 × 6 30 m (15 + 15) On 20 s (~14 s) P - Sbest: 5.82 ± 0.15 s; Savg: 6.06 ± 0.18; Sdec: 4.2 ± 1.1% - - - Padulo et al. [276] SHU 1 × 6 40 m (20 + 20) 20 s P - Test, Sbest: 7.09 ± 0.18 s; Stotal: 44.84 ± 1.09 s; Sdec: 5.5 ± 1.6%; Retest, Sbest: 7.06 ± 0.15 s; Stotal: 44.76 ± 1.09 s; Sdec: 5.7 ± 1.7% Test, CR10: 7.2 ± 0.9 au; Retest, CR10: 7.2 ± 0.4 au - Test, B[La]3’: 11.3 ± 2.0 mmol·L-1; Retest, B[La]3’: 11.7 ± 1.7 mmol·L-1 Padulo et al. [277] SHU 1 × 6 40 m (20 + 20) 20 s P - Test, Sbest: 6.97 ± 0.12 s; Stotal: 43.76 ± 0.90 s; Sdec: 4.6 ± 1.5%; Retest, Sbest: 7.03 ± 0.15 s; Stotal: 44.08 ± 0.75 s; Sdec: 4.5 ± 1.1% - - Test, B[La]3’: 11.6 ± 2.2 mmol·L-1; Retest, B[La]3’: 11.6 ± 2.1 mmol·L-1 Padulo et al. [114] SHU 1 × 10 30 m (15 + 15) 30 s P - Test, Sbest: 5.81 ± 0.32 s; Stotal: 60.19 ± 3.57 s; Sdec: 3.5 ± 1.7%; Retest, Sbest: 5.82 ± 0.31 s; Stotal: 60.50 ± 3.56 s; Sdec: 3.8 ± 1.6%; Test, CR10: 7.8 ± 1.3 au; Retest, CR10: 8.0 ± 1.2 au - Test, B[La]3’: 11.9 ± 2.5 mmol·L-1; Retest, B[La]3’: 11.9 ± 2.1 mmol·L-1  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological SHU 1 × 10 30 m (10 + 10 + 10) 30 s P - Test, Sbest: 7.02 ± 0.44 s; Stotal: 72.49 ± 4.82 s; Sdec: 3.3 ± 1.3%; Retest, Sbest 7.01 ± 0.44 s; Stotal: 72.51 ± 4.77 s; Sdec: 3.4 ± 1.4% Test, CR10: 7.8 ± 1.6 au Retest, CR10: 8.1 ± 1.5 au - Test, B[La]3’: 11.3 ± 2.8 mmol·L-1 Retest, B[La]3’: 11.4 ± 2.5 mmol·L-1 Padulo et al. [156] SHU 1 × 6 40 m (20 + 20) 20 s P - Test, Sbest: 7.10 ± 0.20 s; Stotal: 44.89 ± 1.14 s; Sdec: 5.5 ± 1.9%; Retest, Sbest: 7.09 ± 0.20; Stotal: 44.79 ± 1.13 s; Sdec: 5.3 ± 1.7% Test, CR10: 7.0 ± 1.2 au; Retest, CR10: 7.2 ± 0.7 au - Test, B[La]3’: 11.2 ± 2.1 mmol·L-1 Retest, B[La]3’: 11.3 ± 2.0 mmol·L-1 SHU 1 × 6 40 m (20 + 20) 20 s AP - Sbest: 7.16 ± 0.23; Stotal: 45.77 ± 1.34 s; Sdec: 6.6 ± 1.6% CR10: 7.9 ± 1.2 au - B[La]3’: 13.1 ± 2.1 mmol·L-1 Padulo et al. [150] SHU 1 × 6 40 m (20 + 20) 15 s P Sbest: 7.36 ± 0.10 s; Stotal: 46.12 ± 0.85 s; Sdec: 4.5 ± 1.2% -  CMJ AA: 39.2 cm to 35.6 ± 0.9 cm (-9.0%) B[La]3’: 14.5 ± 0.4 mmol·L-1 SHU 1 × 6 40 m (20 + 20) 20 s P Sbest: 7.35 ± 0.16 s; Stotal: 45.41 ± 0.94 s; Sdec: 3.0 ± 0.9% -  CMJ AA: 39.2 cm to 37.5 ± 2.7 cm (-4.3%) B[La]3’: 12.7 ± 1.2 mmol·L-1 SHU 1 × 6 40 m (20 + 20) 25 s P Sbest: 7.33 ± 0.13 s; Stotal: 44.82 ± 0.90 s; Sdec: 1.9 ± 0.7% -  CMJ AA: 39.2 cm to 38.3 ± 3.7 cm (-2.3%) B[La]3’: 8.0 ± 1.5 mmol·L-1 Paulauskas et al. [122] SHU 3 × 10 30 m (15 + 15) 30 s P 5 min P Sbest: set 1, 58.45 ± 1.63 s; set 2, 59.25 ± 2.03 s; set 3, 60.02 ± 2.41 s; - - B[La]3’: 13.02 ± 2.28 mmol·L-1; HRpeak: set 1, 175 ± 8 b·min-1; set 2, 178 ± 5 b·min-1; set 3, 182 ± 10 b·min-1; HRavg: set 1, 163 ± 9.1 b·min-1; set 2, 169 ± 7 b·min-1; set 3, 169 ± 6 b·min-1 SHU 3 × 20 15 m (7.5 + 7.5) 15 s P 5 min P Sbest: set 1, 53.37 ± 1.64 s; set 2, 53.58 ± 1.48 s; set 3, 54.04 s - - B[La]3’: 8.5 ± 3.4 mmol·L-1; HRpeak: set 1, 174 ± 9 b·min-1; set 2, 178 ± 8 b·min-1; set 3, 179 ± 7 b·min-1; HRavg: set 1, 161 ± 10 b·min-1; set 2, 170 ± 9 b·min-1; 171 ± 8 b·min-1 Perroni et al. [103] MDA 1 × 7 30 m 25 s AK - Savg: 6.12 ± 0.04 s; Stotal: 42.84 ± 1.96 s; Sdec: 3.7 ± 1.2% - - - Petisco et al. [278] SHU 1 × 6 30 m (15 + 15) 20 s P - Sbest: 5.77 ± 0.15 s; Stotal: 35.70 ± 0.65 s - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Purkhús et al. [279] STR 1 × 5 30 m 25 s AK - CON, Savg: 5.46 ± 0.38 s INT, Savg: 5.64 ± 0.29 s - - - Pyne et al. [280] STR 1 × 6 30 m On 20 s (~16 s) P - Stotal: 25.83 ± 0.60 s; Sdec: 3.8 ± 1.1% - - - Ramírez- Campillo et al. [281] STR 1 × 6 35 m 10 s P - CON, Savg: 7.35 ± 0.50 s; PLA, Savg: 7.08 ± 0.60 s; INT, Savg: 7.48 ± 1.00 s - - - Rampinini et al. [312] SHU 1 × 6 40 m (20 + 20) 20 s P - Sbest: 7.00 ± 0.19 s; Savg: 7.25 ± 0.17 s; Sdec: 3.3 ± 1.6% - - - Rampinini et al. [174] SHU 1 × 6 40 m (20 + 20) 20 s P - PRO, Sbest: 6.86 ± 0.13 s; Savg: 7.17 ± 0.09 s; Sdec: 4.5 ± 1.9%; AM, Sbest: 6.97 ± 0.15 s; Savg: 7.41 ± 0.19 s; Sdec: 6.0 ± 1.9% - - - Rodríguez- Fernández et al. [165] STR 1 × 8 30 m 25 s A - YTH, Sbest: 4.03 ± 0.15 s; Savg: 4.19 ± 0.12 s; Stotal: 33.52 ± 0.97 s; Sdec: 3.9 ± 1.6% PRO, Sbest: 3.92 ± 0.11 s; Savg: 4.12 ± 0.12 s; Stotal: 32.91 ± 0.91 s; Sdec: 5.2 ± 1.9% - - - Rodríguez- Fernández et al. [284] STR 1 × 8 30 m 25 s AK - Sbest: 3.87 ± 0.04 s; Savg: 4.03 ± 0.04 s; Stotal: 32.26 ± 0.31 s; Sdec: 4.3 ± 0.3% - - - Rey et al. [283] STR 1 × 6 25 m 25 s AK - INT, Sbest: 3.21 ± 0.08 s; Savg: 3.29 ± 0.07 s; Stotal: 19.77 ± 0.46 s; Sdec: 2.4 ± 1.5% CON, Sbest: 3.15 ± 0.12 s; Savg: 3.25 ± 0.15 s; Stotal: 19.53 ± 0.95 s; Sdec: 3.1 ± 1.9% - - - Røksund et al.[285] STR 1 × 8 30 m On 30 s (~27 s) P - Savg: 3.14 ± 0.10 s - - - Ruscello et al. [286] STR 1 × 7 30 m 1:5N (~26 s) P - Savg: 5.24 ± 0.33 s - - B[La]3’: 10.9 ± 1.8 mmol·L-1 SHU 1 × 7 30 m (15 + 15) 1:3N (~21 s) P - Savg: 6.84 ± 0.44 s - - B[La]3’: 7.9 ± 2.4 mmol·L-1  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Ruscello et al. [104] STR 1 × 7 30 m 1:5N (~22 s) P - Savg: 4.53 ± 0.28 s; Sdec: 4.8% -  CMJ AD: 46.8 ± 4.5 to 43.3 ± 5.0 cm (-7.5%) - SHU 1 × 7 30 m (15 + 15) 1:5N (~30 s) P - Savg: 5.89 ± 0.35 s; Sdec: 3.4% -  CMJ AD: 46.9 ± 4.5 to 43.0 ± 5.1 cm (-8.3%) - MDC 1 × 7 30 m (5 m per turn) 1:5N (~42 s) P - Savg: 8.51 ± 0.41 s; Sdec: 2.5% -  CMJ AD: 46.9 ± 4.4 to 43.5 ± 5.0 cm (-7.1%) - Russell et al. [123] STR 1 × 15 30 m 60 s P - CON, Savg: 4.34 ± 0.17 s; Stotal: 65.08 ± 2.56 s; INT, Savg: 4.37 ± 0.23 s; Stotal: 65.56 ± 3.38 s - - CON,  CK 24 h: 232 ± 44 u·L-1 to 785 ± 129 u·L-1 (238%); INT  CK 24 h: 232 ± 49 u·L-1 to 799 ± 141 u·L-1 (244%) Salleh et al. [287] MDC 1 × 5 40 m 60 s AU - Savg: 7.54 ± 0.65 s; Sdec: 1.9 ± 1.6% - - - Sánchez- Sánchez et al. [117] SHU 1 × 6 40 m (20 + 20) 20 s A - Sys1, Sbest: 7.38 ± 0.25 s; Savg: 7.93 ± 0.30 s; Stotal: 47.55 ± 1.74 s; Sys2, Sbest: 7.5 ± 0.26 s; Savg: 7.97 ± 0.26 s; Stotal: 47.85 ± 1.59 s; Sys3, Sbest: 7.74 ± 0.29 s; Savg: 8.24 ± 0.29 s; Stotal: 49.46 ± 1.75 s; Sys4, Sbest: 7.51 ± 0.32 s; Savg: 8.02 ± 0.25 s; Stotal: 48.14 ± 1.48 s - Sys1,  CMJAA: 36.5 ± 4.4 to 28.3 ± 4.5cm (-22.5%); Sys2,  CMJAA: 35.5 ± 5.4 to 26.0 ± 4.9 cm (-26.1%); Sys3,  CMJAA: 36.4 ± 5.7 to 26.5 ± 5.2 cm (-27.1%); Sys4,  CMJAA: 36.9 ± 5.1 to 30.1 ± 5.9 cm (-18.5%) Sys1, B[La]1’: 12.9 ± 2.3 mmol·L-1 ; B[La]3’: 13.0 ± 2.5 mmol·L-1; HRpeak 184 ± 13 b·min-1; Sys2, B[La]1’: 12.4 ± 2.4 mmol·L-1 ; B[La]3’: 13.0 ± 3.0 mmol·L-1; HRpeak 185 ± 12 b·min-1; Sys3, B[La]1’: 11.0 ± 2.3 mmol·L-1 ; B[La]3’: 11.0 ± 1.9 mmol·L-1; HRpeak 183 ± 13 b·min-1; Sys4, B[La]1’: 11.8 ± 2.5 mmol·L-1 ; B[La]3’: 11.1 ± 2.5 mmol·L-1; HRpeak 185 ± 12 b·min-1 Sánchez- Sánchez et al. [288] STR 1 × 7 30 m 20 s A - Savg: 4.46 ± 0.17 s; Sdec: 4.7 ± 2.0% - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Sánchez- Sánchez et al. [289] STR 1 × 6 20 m 20 s P - Sbest: 3.19 ± 0.11 s; Savg: 3.29 ± 0.08 s - - - Sanders et al. [290] STR 1 × 10 30 m 25 s P - - - - HRpost (% HRmax): 93% Scanlan et al. [291] STR 1 × 10 20 m 30 s P - Stotal: 35.02 ± 2.1 s; Sdec: 2.7 ± 1.2% 6−20: 15.2 ± 2.1 - B[La]post: 4.6 ± 0.8 to 11.0 ± 1.6 mmol·L-1; HRpeak: 169 ± 12 b·min-1 STR 1 × 10 20 m 30 s AAG - Stotal: 37.73 ± 2.5 s; Sdec: 9.4 ± 5.2% 6−20: 18.4 ± 1.3 au - B[La]post: 5.0 ± 1.1 to 16.5 ± 4.5 mmol·L-1; HRpeak: 187 ± 9 b·min-1 Scanlan et al. [292] SHU 1 × 12 20 m AE 20 s P - Sdec: 2.8 ± 0.8% - - - Selmi et al. [58] STR 2 × 5 20 m 15 s AJ 1 min P Sbest: set 1, 3.31 ± 0.14 s; set 2, 3.38 ± 0.12 s; Stotal: set 1, 16.97 ± 0.69 s; set 2, 17.69 ± 0.58 s; Sdec: Set 1, 2.9 ± 1.6%; Set 2, 5.1 ± 2.8% CR10: 6.3 ± 1.4 au -  B[La]3’: 1.8 ± 0.6 to 8.1 ± 2.2 mmol·L-1 HRpeak: 186 ± 14 b·min-1 HRavg: 137 ± 12 b·min-1 STR 2 × 5 20 m 15 s AJ 2 min P Sbest: set 1, 3.28 ± 0.10 s; set 2, 3.33 ± 0.11 s; Stotal: set 1, 16.90 ± 0.57 s; set 2, 17.11 ± 0.47 s; Sdec: Set 1, 3.2 ± 1.6%; Set 2, 2.8 ± 1.6% CR10: 3.2 ± 1.5 au -  B[La]3’: 1.5 ± 0.2 to 8.2 ± 1.0 mmol·L-1 HRpeak: 182 ± 9 b·min-1 HRavg: 125 ± 11 b·min-1 STR 2 × 5 20 m 15 s AJ 4 min P Sbest: set 1, 3.31 ± 0.11 s; set 2, 3.31 ± 0.11 s; Stotal: set 1, 16.97 ± 0.64 s; set 2, 17.06 ± 0.55 s; Sdec: Set 1, 2.7 ± 1.3%; Set 2, 3.1 ± 1.4% CR10: 3.4 ± 1.2 au -  B[La]3’: 1.6 ± 0.3 to 8.5 ± 1.8 mmol·L-1 HRpeak: 180 ± 10 b·min-1 HRavg: 114 ± 5 b·min-1 Selmi et al. [293] SHU 1 × 20 40 m (20 + 20) 20 s P - INT, Sbest: 7.53 ± 0.48 s; Stotal: 47.86 ± 2.81 s; Sdec: 6.0 ± 1.9% CON, Sbest: 7.69 ± 0.31 s; Stotal: 49.05 ± 1.52 s; Sdec: 6.3 ± 2.0% - - - Shalfawi et al. [294] STR 1 × 7 30 m 30 s P - Sbest: 4.93 ± 0.20 s; Savg: 5.04 ± 0.20 s; Stotal: 35.35 ± 1.40 s; Sdec: 2.2 ± 1.0% - - - Shalfawi et al. [295] STR 1 × 10 40 m 60 s P - INT, Savg: 5.92 ± 0.26 s CON, Savg: 5.84 ± 0.27 s - - - Shalfawi et al. [296] STR 1 × 10 40 m 60 s P - ATG, Savg: 6.15 ± 0.4 s RS, Savg: 6.19 ± 0.25 s - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Silva et al. [297] SHU 1 × 6 40 m (20 + 20) 20 s P - Sbest: 6.44 ± 0.14 s; Savg: 6.57 ± 0.26 s; Stotal: 44.20 ± 0.40 s; Sdec: 9.8 ± 1.4% - - - Soares- Caldeira et al. [298] SHU 1 × 6 40 m (20 + 20) 20 s P - INT, Sbest: 7.17 ± 0.37 s; Savg: 7.62 ± 0.35 s; Sdec: 6.3 ± 2.0%; CON, Sbest: 6.95 ± 0.16 s; Savg: 7.49 ± 0.20 s; Sdec: 7.8 ± 4.3% - - - Spineti et al. [299] SHU 1 × 6 40 m (20 + 20) 20 s P - CCT, Sbest: 6.93 ± 0.15 s; Savg: 7.43 ± 0.10; Sdec: 7.2 ± 2.2% TST, Sbest: 7.11 ± 0.19 s; Savg: 7.54 ± 0.23; Sdec: 6.1 ± 1.9% - - - Suarez- Arrones et al. [105] MDAF 1 × 6 40 m (20 + 20) 20 s P - RS, Sbest: 7.60 ± 0.20 s; Savg: 8.00 ± 0.20 s; Sdec: 5.3 ± 1.3%; SQ, Sbest: 7.50 ± 0.30 s; Savg: 7.90 ± 0.30 s: Sdec: 5.0 ± 2.0% - - - SHU 3 × 6 40 m (20 + 20) 20 s P 4 min P - 6−20: 13.9 ± 0.4 au - - Stojanovic et al. [300] SHU 1 × 10 30 m (15 + 15) 30 s P - Savg: 5.77 ± 0.18 s; Sdec: 3.5 ± 1.1% - - - Taylor et al. [2] STR 3−4 × 7 30 m 20 s P 4 min P - - - HRpeak (% HRmax): 92 ± 5% SHU 3−4 × 7 30 m 20 s P 4 min P - - - HRpeak (% HRmax): 89 ± 11% Teixeira et al. [301] STR 1 × 8 40 m 20 s P - IT7.5, Sbest: 8.86 ± 0.25 s; Savg: 9.39 ± 0.26 s; Sdec: 6.5 ± 1.4%; IT15, Sbest: 8.83 ± 0.36 s; Savg: 9.33 ± 0.36 s; Sdec: 5.7 ± 3.2% - - - Thomassen et al. [302] STR 1 × 10 20 m 15 s AK - INT, Savg: 3.35 ± 0.07 s; Stotal: 33.44 ± 0.44 s; Sdec: 5.8 ± 1.0% NT, Savg: 3.34 ± 0.09 s; Stotal: 33.41 ± 0.32 s; Sdec: 5.9 ± 0.8% - - - Tønnessen et al. [303] STR 1 × 10 40 m 60 s - P INT, Savg: 5.42 ± 0.18 s; CON, Savg: 5.41 ± 0.19 s - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Torreblanca- Martinez et al. [304] STR 1 × 12 30 m 30 s P - Sdec: 6.5 ± 3.0% 6−20: 15.2 ± 2.5 au - HRpost: 179 ± 12 b·min-1 Tounsi et al. [176] SHU 1 × 6 40 m (20 + 20) 20 s P - M, Sbest: 7.09 ± 0.24 s; Savg: 7.32 ± 0.28; Sdec: 3.2 ± 1.2% F, Sbest: 8.42 ± 0.47 s; Savg: 8.85 ± 0.45; Sdec: 5.1 ± 2.5% - - - Trecroci et al. [305] STR 1 × 5 30 m 25 s P - SST, Sbest: 4.26 ± 0.11 s; Stotal: 21.94 ± 0.67 s; ARC, Sbest: 4.25 ± 0.07 s; Stotal: 21.91 ± 0.58 s - - - Turki et al. [111] MDX 1 × 6 20 m (4 m per turn) 25 s AK - PRO, Sbest: 5.39 ± 0.18 s; Savg: 5.52 ± 0.17 s; Stotal: 33.09 ± 1.00 s; Sdec: 2.4 ± 1.0%; COL, Sbest: 5.49 ± 0.26 s; Savg: 5.62 ± 0.27 s; Stotal: 33.70 ± 1.60 s; Sdec: 2.4 ± 0.6% - - - Ulupinar et al. [126] STR 1 × 10 40 m 30 s P - Sbest: 5.43 ± 0.03 s; Stotal: 56.7 ± 1.6 s; Sdec: 4.8 ± 1.7% 6−20: 17 ± 1 au - B[La]peak: 18.6 ± 1.7 mmol·L- 1; HRpeak: 184 ± 8 b·min-1 HRavg: 164 ± 7 b·min-1 STR 1 × 20 20 m 15 s P - Sbest: 3.18 ± 0.03 s; Stotal: 67.3 ± 3.0 s; Sdec: 6.9 ± 2.8% 6−20: 19 ± 1 au - B[La]peak: 16.6 ± 2.2 mmol·L- 1: HRpeak: 188 ± 8 b·min-1 HRavg: 168 ± 9 b·min-1 Ulupinar et al. [125] STR 1 × 20 15 m 30 s P - Stotal: 49.9 ± 1.2; Sdec: 3.6 ± 1.8% 6−20: 11.5 ± 2.9 au - B[La]peak: 9.1 ± 3.0 mmol·L-1: HRpeak: 186 ± 9 b·min-1 HRavg: 168 ± 9 b·min-1 STR 1 × 20 15 m 1:5N (~12 s) P - Stotal: 52.7 ± 1.3; Sdec: 8.7 ± 2.8% 6−20: 16.3 ± 1.9 au - B[La]peak: 14.9 ± 3.7 mmol·L- 1: HRpeak: 190 ± 11 b·min-1 HRavg : 178 ± 11 b·min-1 STR 1 × 10 30 m 30 s P - Stotal: 44.9 ± 1.2; Sdec: 7.1 ± 3.8% 6−20: 13.9 ± 2.4 au - B[La]peak: 15.0 ± 4.1 mmol·L- 1: HRpeak: 191 ± 13 b·min-1 HRavg: 172 ± 10 b·min-1 STR 1 × 10 30 m 1:5N (~22 s) P - Stotal: 45.8 ± 1.1; Sdec: 9.3 ± 2.3% 6−20: 15.8 ± 2.9 au - B[La]peak: 16.9 ± 3.5 mmol·L- 1: HRpeak: 190 ± 12 b·min-1 HRavg: 177 ± 8 b·min-1 Van den Tillaar et al. [306] STR 1 × 7 30 m On 30 s (~25 s) AK - Savg: 5.46 ± 0.33 s - - -  Exercise protocol Outcomes Study RST Mode Sets × Reps Distance / Duration Rest Time Rest Mode I-set Rest Performance Perceptual Neuromuscular Physiological Vasquez- Bonilla et al. [307] STR 1 × 8 20 m 20 s AK - Sbest: 3.81 ± 0.17 s; Savg: 4.08 ± 0.21 s; Stotal: 32.64 ± 1.75 s; Sdec: 7 ± 3% - - - Wadley & Le Rossignol [308] STR 1 × 12 20 m 20 s P - Stotal: 39.31 ± 0.12 s; Sdec: 5.5 ± 3.3% - - - West et al. [309] SHU 1 × 6 40 m (20 + 20) 20 s P - Sbest: 6.60 ± 0.16 s; Savg: 6.87 ± 0.15 s; Stotal: 41.23 ± 0.92 s - - - Woolley et al. [33] STR 1 × 40 15m 30s PR - - 6−20: 16.7 ± 1.8 au -  CK 24 h: 279 ± 322 to 1121 ± 1362 u·L-1 (302%) Yanci et al. [310] STR 1 × 6 30 m 25 s A - CON, Savg: 4.57 ± 0.20 s PLY1: Savg: 4.47 ± 0.22 s PLY2: Savg: 4.45 ± 0.23 s - - - Zagatto et al. [106] SHU 1 × 10 30m 30 s P - Sbest: 6.56 ± 0.30 s; Savg: 6.84 ± 0.30 s; Stotal: 68.40 ± 2.91 s; Sdec: 4.2 ± 1.8% - - B[La]peak: 9.8 ± 2.5 mmol·L-1; VO2avg: 37.0 ± 2.9 ml·min- 1·kg-1; HRpeak: 185 ± 9 b·min-1 MDC 1 × 10 30 m (5 m per turn) 30 s P - Sbest: 8.14 ± 0.36 s; Savg: 8.39 ± 0.36 s; Stotal: 83.99 ± 3.60 s; Sdec: 3.0 ± 1.1% - - B[La]peak: 8.2 ± 1.9 mmol·L-1; VO2avg: 36.1 ± 3.2 ml·min- 1·kg-1; HRpeak: 186 ± 9 b·min-1 Zagatto et al. [311] SHU 2 × 10 30 m (10 + 10 + 10) 30 s P P 5.50 min Set 1, Sbest: 6.85 ± 0.35 s; Savg: 7.01 ± 0.31 s; Stotal: 70.15 ± 3.07 s; Sdec: 2.4 ± 1.5% Set 2, Sbest: 6.88 ± 0.32 s; Savg: 7.13 ± 0.36 s; Stotal: 71.31 ± 3.59 s; Sdec: 3.6 ± 1.58% -  CMJ AB: 43.2 ± 9.7 to 37.6 ± 4.0 cm (-9.4 ± 18.0%) - Zagatto et al. [107] MDC 1 × 10 30 m 30 s P - Sbest: 7.09 ± 0.57 s; Savg: 7.30 ± 0.63 s; Stotal: 72.84 ± 6.42 s - - - Data are presented as mean ± SD. Abbreviations: I-set = inter-set; RST = repeated-sprint training; sRPE = session ratings of perceived exertion; au = arbitrary units; CR10 = category ration 0−10 rating of perceived exertion scale; 6−20 = 6−20 rating of perceived exertion scale; SHU = shuttle repeated-sprint; STR = straight-line repeated-sprint; MD = multi-directional repeated-sprint; A = active recovery; P = passive recovery; M = male; F = female; B[La]post = blood lactate measured immediately post-exercise; B[La]peak = highest blood lactate value measured from two or more time-points between 0−10 min post-exercise; B[La]1’ = blood lactate measured 1 minutes post-exercise; B[La]2’ = blood lactate measured 2 minutes post-exercise; B[La]3’ = blood lactate measured 3 minutes post-exercise; B[La]4’ = blood lactate measured 4 minutes post exercise; B[La]5’ = blood lactate measured 5 minutes post-exercise; CK 24 h = serum creatine kinase measured 24 hours post-exercise Sdec = percentage sprint decrement; Savg = average sprint time; Sbest = best sprint time; Stotal = total sprint time; CMJ = counter movement jump height; HRavg = average heart rate; HRpeak = peak heart rate; HRpost = end-set heart rate recorded immediately post-exercise; % HRmax = percentage of maximal heart rate; VO2avg = average oxygen consumption; % VO2max = percentage of maximal oxygen consumption; V0 = theoretical maximal velocity F0 = theoretical maximal force; P0 = theoretical maximal power; RFpeak = maximal ratio of force; DRF = slope/rate of decrease in ratio of force with increasing velocity; Kvert vertical stiffness;; Kleg = leg stiffness; ΔL = leg compression; Δz = centre of mass vertical displacement; Fzmax = maximal vertical force; PLA = placebo group = CON = control group; STR-G = straight-line repeated-sprints groups; SHU-G = shuttle repeated-sprints group; High = high VO2 max group; Med = medium VO2 max group; Low = low VO2 max group; INT = intervention group; U17 = under 17 players; U18 = under 18 players; U19 = under 19 players; U20 = under 20 players; PRE = pre-season; ELY = early/start of season; MID = mid-season; END = end/post of season; YTH = youth players; SEN = senior players; PRO = professional players; SEMI = semi-proffessional players; COL = college players; REP = representative players; Club = club players; AM = amateur players; EL = elite players; S-EL = sub-elite players; M-PRO = mid-proffessional players; EXP = experienced players; FSH = freshman players; FUT = futsal players; SOC = soccer players; SAN = sand training group; GRA = grass training group; NOR = normoxia group; HYP = hypoxia group; MG = Melaneysian group; N-MG = non-Meleynesian group; ARC = active recovery condition; SSG = small sided games group; SEM = speed endurance maintenance group; SEP = speed endurance production group; RS15 = repeated-sprint group with 15 s rest; RS30 = repeated-sprint group with 30 s rest; Sys1 = turf system 1; Sys2 = turf system 2; Sys3 = turf system 3; Sys4 = turf system 4; IT7.5 = interval training 7.5 seconds group; IT15 = interval training 15 seconds group; RS = repeated sprint group; ATG = agility training group; 1TR = under 17 group born 1st tertile; 2TR = under 17 group born 2nd tertile; 3TR = under 17 group born 3rd tertile; Sham = sham group; RES = resisted sprint training group; PLY = plyometric group; PLY1 = plyometrics one day per week group; PLY2 plyometrics two days per week group; LLTL = live low-train low group; IT100 = interval training at 100% group; IT86 = interval training at 86% group; SQ = squat group; TG = take-off group; PAS = passive recovery group; COL = cold water recovery group; CWT = contrast water therapy group; NT = Non-training group; ST = starting players; N-ST = non-starting players; N-SEL = non-selected players; SST = soccer specific training condition;  = change from baseline; - = not applicable. A 3 × multi-angle turns B 4 × multi-angle turns C 5 × multi-angle turns D 2 × 45° turns E 2 × 90° turns F 2 × 135° turns G 4 × 45° turns H Run at 8 km·h-1 back to one way start line I Light stretching J 10 m deceleration zone + 10m run zone at either end K Jog back to one way start line L Jogging at 2−2.1 m·s-1 M Single counter-movement jump following each sprint N Exercise to rest ratio O Walking or running to maintain 60-65% of HR maximum P 3 × counter-movement jumps following each sprint Q Self-paced jogging R Short enforced deceleration zone (<10m) S Run at 6 km·h-1 T 4 × 90° turns (quadrangle) U Walk for 40 s, stationary rest for 20 s V 4 × 100° turns W 10m zone at both ends to decelerate, then jog back to two-way start line. X Run at 20% maximal aerobic speed Y Run at 35% maximal aerobic speed Z Run at 50% maximal aerobic speed AA Measured via an Optojump AB Measured via force-platforms AC Measured via a contact mat AD Measured via FreePower Jump AE Repeated 5-0-5 Agility test: total rep distance = 20 m, timed distance = 10 m AF Change of direction performed around a cone AG Run at 50% maximal speed Supplementary Table S4. Influence of programming variables on the variance of meta-analysed acute physiological, perceptual and performance demands of repeated-sprint training in team sport athletes. Total Variance (σ2) Variance Explained by Moderators (R2META) Observed (no moderators) With Moderators HRavg b∙min-1 335 - - % HRmax 19 - - HRpeak b∙min-1 59 55 0.07 VO2avg ml∙kg-1∙min-1 89.6 - - B[La] mmol∙L-1 9.3 6.3 0.32 sRPE au (deciMax) 3.1 3.0 0.03 Sbest s 2.71 1.10 0.60 Savg s 2.68 0.69 0.74 Sdec % 4.8 3.5 0.27 Dashed lines indicate outcome measure where moderator analysis could not be performed.

The Acute Demands of Repeated-Sprint Training on Physiological, Neuromuscular, Perceptual and Performance Outcomes in Team Sport Athletes: A Systematic Review and Meta-analysis.

05-24-2023

Thurlow, Fraser,Weakley, Jonathon,Townshend, Andrew D,Timmins, Ryan G,Morrison, Matthew,McLaren, Shaun J

eng

PMC3575608

Hindawi Publishing Corporation The Scientific World Journal Volume 2013, Article ID 189149, 11 pages http://dx.doi.org/10.1155/2013/189149 Review Article Exercise-Induced Muscle Damage and Running Economy in Humans Cláudio de Oliveira Assumpção, Leonardo Coelho Rabello Lima, Felipe Bruno Dias Oliveira, Camila Coelho Greco, and Benedito Sérgio Denadai Human Performance Laboratory, UNESP, Avenue 24 A, Bela Vista-Rio, 13506-900 Rio Claro, SP, Brazil Correspondence should be addressed to Benedito S´ergio Denadai; bdenadai@rc.unesp.br Received 18 December 2012; Accepted 18 January 2013 Academic Editors: L. Guimar˜aes-Ferreira, H. Nicastro, J. Wilson, and N. E. Zanchi Copyright © 2013 Cl´audio de Oliveira Assumpc¸˜ao et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Running economy (RE), defined as the energy demand for a given velocity of submaximal running, has been identified as a critical factor of overall distance running performance. Plyometric and resistance trainings, performed during a relatively short period of time (∼15–30 days), have been successfully used to improve RE in trained athletes. However, these exercise types, particularly when they are unaccustomed activities for the individuals, may cause delayed onset muscle soreness, swelling, and reduced muscle strength. Some studies have demonstrated that exercise-induced muscle damage has a negative impact on endurance running performance. Specifically, the muscular damage induced by an acute bout of downhill running has been shown to reduce RE during subsequent moderate and high-intensity exercise (>65% VO2max). However, strength exercise (i.e., jumps, isoinertial and isokinetic eccentric exercises) seems to impair RE only for subsequent high-intensity exercise (∼90% VO2max). Finally, a single session of resistance exercise or downhill running (i.e., repeated bout effect) attenuates changes in indirect markers of muscle damage and blunts changes in RE. 1. Introduction Running economy (RE), defined as the energy demand for a given velocity of submaximal running, is an important predictor of aerobic running performance, particularly in elite runners who have a similar aerobic power (i.e., max- imal oxygen uptake, VO2max) [1]. Runners with high RE demonstrate lower energetic cost at submaximal velocity and consequently tend to run faster at given distance or longer at a constant velocity. A number of biomechanical (e.g., gait patterns, kinemat- ics, and the kinetics of running) and physiological factors (e.g., oxidative muscle capacity) seem to influence RE in trained athletes [2, 3]. Moreover, some interventions (plyo- metric, resistance and altitude training) performed during relatively short periods of time (∼15–30 days) have been successfully used to improve RE [4–6]. Plyometric and resistance trainings lead to neuromuscular adaptations such as increased neural drive to the muscles and changes in muscle stiffness and muscle fiber composition, which might reduce the energetic cost during submaximal exercise. How- ever, plyometric and resistance trainings, especially when they are unaccustomed activities, may cause delayed onset muscle soreness (DOMS), swelling, and reduced muscle strength. The negative effect of muscle-damaging exercises on endurance running performance has been experimentally demonstrated in both animal [7, 8] and human experi- ments [9]. However, studies that have investigated the effect of exercise-induced muscle damage (EIMD) on RE have produced equivocal results [9–12]. This review discusses the effects of EIMD induced by different exercise types (strength, long-distance running, and downhill running) on RE. Different recovery strategies aiming to enhance the RE after EIMD are also addressed. 2. Running Economy Aerobic fitness, as well as running performance, can be mea- sured by different variables, for example, maximal oxygen uptake (VO2max), lactate threshold (LT), onset of blood 2 The Scientific World Journal lactate accumulation (OBLA), movement economy (ME), or running economy (RE). The VO2max, which reflects an indi- vidual’s maximal rate of aerobic energy expenditure, has been considered the gold standard for measuring aerobic power [13]. Indeed, the VO2max has a positive association with aerobic running performance obtained during middle- and long-distance events (1,500 m–42,195 m) [14–16]. However, some studies [17–20] have shown that subjects with similar VO2max values may attain different aerobic running perfor- mance or VO2 values during exercise of similar duration and intensity. These differences are most likely due to variations in ME among subjects. The ME is defined as the amount of energy necessary (Kcal⋅min−1) to perform a given task [21]. However, due to the difficulty to determine the external work performed during running, RE, expressed as the volume of oxygen uptake (mL⋅Kg−1⋅min−1) during a specific submaximal running intensity (Km⋅h−1), has been adopted. There is a strong asso- ciation between RE and aerobic running performance, with RE being a better predictor of performance than VO2max, particularly in athletes who have similar VO2max [19, 22]. Several factors have been proposed to influence RE in trained subjects. These include oxidative muscle capacity and muscle stiffness. Muscle stiffness corresponds to the ability of the muscles to store and release elastic energy. Moreover, some interventions such as training, environment, and muscle damage can modify the oxygen cost over a range of running speeds [6, 11, 23–25]. Improved oxidative muscle capacity can be associated with reduced oxygen consumption per mitochondrial respi- ratory chain during submaximal exercise. Trained subjects are known to have better RE than untrained individuals, and long-distance runners are more economical than middle- distance runners [26]. Additionally, a high weekly volume of training has also been associated with better RE. However, a short period (4–6 weeks) of high-intensity aerobic training (near or above VO2max) can also lead to improvement in the RE of trained runners [27]. In addition to aerobic characteristics and adaptations, neuromuscular profile factors have also been considered important aspects of RE. Type II muscle fibres seem to be positively correlated with submaximal energy consumption, especially at lower speeds [28]. Furthermore, both muscular stiffness and the ability to rapidly develop muscular force (i.e., rate of force development (RFD)) have demonstrated significant correlations with RE [29, 30]. Stiffer muscle- tendon complexes may increase elastic energy storage by reducing the ground contact time, thus decreasing the running oxygen cost. Similar to stiffness, a higher RFD is associated with a shorter time to generate a contraction. This effect could also diminish the ground contact time and running oxygen cost. Heavy weight and plyometric training associated with endurance training have improved RE in well-trained runners [5, 31]. Basically, these types of strength training induce neuromuscular adaptations such as increased neural drive to the muscles, altered muscle-tendon complex stiffness, and changed muscle fibre composition (i.e., I → IIA ← IIX). Environmental variables can also be used to reduce the energetic cost of running. RE can be improved (2-3%) after relative short periods (∼15–20 days) of altitude exposure (∼2.000–4.500 m). Altitude exposure during daily activities, sleeping, or training can enhance RE at sea level altitude through haematological and muscle changes in favour of oxygen transport [32–34]. Moreover, heat exposure during training sessions can also improve RE by enhancing the thermoregulatory process, thus reducing the cardiovascular and muscle work for a given exercise intensity [23, 35, 36]. More recently, EIMD has also proposed to generate important modifications in RE. Muscular damage induced by an acute bout of downhill running has shown to reduce RE in the days following the intervention (24–120 hours) [9, 11]. Specific aspects of this intervention are addressed hereafter. 3. Muscle Damage and the Repeated Bout Effect Skeletal muscle damage has been considered an important factor contributing to DOMS and strength loss after eccentric exercise [37]. Basically, the exercise conditions at which muscle damage can be induced are unaccustomed exercises and exercises with higher intensity or longer duration than those to which the subject is adapted [37, 38]. The result- ing metabolic overload and mechanical strain have been suggested the main factors generating muscle damage [38]. Warren et al. [39] have suggested that measures of muscle function such as strength and power are effective indicators of both the magnitude and time course of muscle damage. Depending on the magnitude of muscle damage, muscle force at isometric, and dynamic testing conditions may be impaired for 1–7 days after the exercise [40–43]. Other important symptoms of muscle damage are disruption of the sarcolemma and extracellular matrix [44, 45], increased blood levels of creatine kinase (CK) and myoglobin (MB), stiffness, and swelling [46–48]. In general, muscle damage can be induced by both static (isometric) and dynamic (concentric and eccentric) muscle contractions. However, there is substantial evidence that eccentric muscle actions result in greater muscle damage than isometric or concentric actions [49–52]. The magnitude of strength loss after EIMD may vary between 5–10% and ∼60% [43, 52], depending on the characteristics of the protocol and the type of muscle actions (i.e., isometric, concentric or eccentric) used during the posttest. The different effects of eccentric versus isometric or concentric actions have also been verified in the context of whole body exercises (i.e., run- ning, cycling, and cross-country skiing) [43, 53]. For example, muscle damage and strength loss are higher during running (∼20–30%), which involves concentric and eccentric actions, when compared with cycling (∼10–15%), which involves mainly concentric actions [53]. In accordance with Millet and Lepers [53], although concentric and eccentric actions are present during cross-country skiing, muscular damage is considerably lesser during this exercise than in running because shock waves are present only during running. The main factors attributed to the greater effect of eccentric contractions on muscle damage are the higher peak torque The Scientific World Journal 3 values [54] and reduced motor unit activation for a given force [54–56], both of which induce a higher mechanical stress on the muscles [54]. Other important aspects of the greater muscle damage induced by eccentric muscle actions are that no energy (ATP) is necessary to detach the cross- bridges formed during muscle contraction [57] and that the longer length of the muscles during the contraction generates greater muscle damage. In addition to the main mechanical factors (i.e., the force level produced and the change in muscle length) [37, 58, 59], some metabolic factors such as substrate depletion, calcium influx, and reactive oxygen species have also been proposed to influence muscle damage [38, 60]. The effects of the different mechanical and metabolic factors that would contribute to muscle damage do not occur at the same time. The time course of the events involves damage in components of excitation-contraction system and sarcomeres [59] and degeneration and regeneration of muscle fibres, during which DOMS, stiffness, and swelling occur [37]. Additionally, there is an inflammatory response generating a transfer of fluid and cells to remove damaged contractile proteins and cellular debris from the damaged muscles [61]. Thereafter, the muscle regeneration process is initiated [61]. Although some of these effects may appear only some hours after the exercise, muscle strength may be impaired during and immediately after the exercise. Thus, mechanisms other than muscle damage can also explain the muscle fatigue (i.e., strength loss). It has been suggested that the magnitude of muscle damage and the loss of muscle function might be attenuated after one bout of eccentric exercise [62–64]. This concept is known as the repeated bout effect (RBE). The RBE has been demonstrated after both eccentric muscle actions [65] and downhill running [66]. In general, this protective effect is confirmed by the reduced decrements and faster recovery of muscle strength, less swelling and DOMS, and attenuated changes in CK and MB in the blood [62, 67, 68]. In addition, alterations in muscle circumference or echo intensity (inflam- mation) are also smaller after the first eccentric exercise bout [68]. This protective effect has been demonstrated after a few days of the eccentric exercise [66] and may last up to 6 months (circumference, DOMS, and inflammation) or 9 months (maximal isometric force, CK), depending on the marker of muscle damage [65]. It has been hypothesized that the RBE is mediated by neu- ral, cellular, and mechanical mechanisms [63, 64]. The neural changes proposed to contribute to the RBE are increased slow-twitch fibre recruitment and synchronisation of motor unit firing, better distribution of the workload among muscle fibres, higher participation of synergist muscles to torque production, and increased motor unit activity relative to torque produced [69–71]. Neural mechanisms have been suggested based on the reduced median frequency [69], which reflects some central aspects related to motor unit recruitment. Howatson et al. [69] have demonstrated a 10% decrease in median frequency 14 days after a bout consisting of either 10 or 45 maximal eccentric actions. RBE has also been observed in the untrained contralateral limb, referred to as the contralateral RBE [72]. These studies [69, 72] con- firm that, in addition to intramuscular adaptations, central aspects regarding motor unit recruitment are also involved in RBE. The main mechanical adaptations associated with RBE are increased muscle stiffness and intramuscular connective tissue and changes in the intermediate filament system (main- tenance of structural integrity of sarcomeres) [63]. Cellular adaptations are associated with higher number of sarcom- eres in myofibrils [59, 73], which might decrease myofib- rillar disruption in the next exercise bout, strengthened plasma membranes, increased protein synthesis, removal of stress-susceptible fibres [59, 74, 75], and remodelling of the cytoskeleton, including effects on proteins such as titin and desmin, talin and vinculin [76], which might improve the strength and the stability of sarcomeres and protect muscle fibers against injuries. Other adaptation that has been hypothesized to explain the RBE is the lesser inflammatory response. Since the mechanical disruption is decreased after the first eccentric exercise bout, the stimulus for the inflam- matory response is also reduced after the exercise [73, 74]. Some of these alterations have been associated with reduced muscle damage (strengthened extracellular matrix) and a change in the optimal angle for torque production toward a longer muscle length (increases in number of sarcomeres). The magnitude of muscle damage induced by eccentric exercise is greater at longer muscle lengths [65, 77]. When the muscles are elongated, the sarcomere length is also greater. Because the severity of muscle damage is influenced by the muscle strain generated [73, 78], it has been suggested that the RBE would be greater under conditions of longer muscle lengths. Nosaka et al. [73] have investigated the effect of the range of motion of the exercise used to induce muscle damage on the RBE. The protocol used to induce muscle damage involved 24 maximal eccentric contractions of the elbow joint, using amplitudes of 50–100∘ or 130–180∘. Although the changes (maximal isometric strength, range of motion, upper arm circumference, muscle soreness, and CK) induced by the first bout were significantly greater using the higher amplitude, both exercise conditions induced RBE. However, the effect generated by the short range of motion was lesser than that promoted by the higher amplitude. Other factor that can modify this protective effect (i.e., RBE) of eccentric exercise is the magnitude of muscle damage [65, 66], which is influenced by the exercise intensity of the first bout. Chen et al. [66] showed that 30 eccentric con- tractions performed at 40% of maximal isometric strength generated a smaller attenuation of the changes in indirect markers of muscle damage (20–60%) than maximal eccentric exercise (65–100%). It has been also demonstrated that both the muscle damage level (i.e., CK) and strength impairment and recovery (i.e., isometric torque) are progressively greater with increases in the number of bouts (1–4). However, Chen et al. [68] have demonstrated that repetitive submaximal eccentric exercise bouts (40% MVC) performed every two weeks promote a protective effect similar to that induced by one maximal eccentric exercise bout. In this study, the main indirect markers of muscle damage were less affected by the second to the fourth bouts of submaximal eccentric exercise than the first; that is, the protective effect is promoted under 4 The Scientific World Journal Table 1: Comparison of the effects of the resistance exercise on running economy. Study Subjects EIMD Muscle damage VO2max (%) RE (%) Paschalis et al. [10] 10 healthy males 120 eccentric actions ↑ CK, ↑ DOMS, and ↓ ROM, and ↓ strength 55 and 75 √ Burt et al. [12] 9 healthy men 100 squats at 80% body mass √ CK, ↑ DOMS, and ↓ strength 90 ↓ 4-5 Vassilis et al. [87] 24 young healthy men 120 eccentric actions ↑ CK, ↑ DOMS, ↓ strength 70 √ Scott et al. [88] 8 active men and 8 active women 3-4 × 10 repetitions of squat, lunges, step up and step down, and stiff-legged deadlift ↑ DOMS 70 √ EIMD: exercise-induced muscle damage; %VO2max: exercise intensity at which running economy was measured; RE: running economy; CK: creatine kinase; DOMS: delayed onset muscle soreness; ROM: range of motion; ↓ indicates decrease; √ indicates no change; ↑ indicates increase. conditions of reduced levels of induced muscle damage. Even after repeated submaximal bouts the magnitude of muscle damage was still smaller than that induced by one maximal bout. The authors suggested that the effect of exercise intensity on the protective effect of the first bout does not apply when some bouts of low-intensity exercise are performed. Therefore, the magnitude of muscle damage does not necessarily affect the protective effect of eccentric exercise. Moreover, Howatson et al. [69] have compared two protocols of maximal eccentric contractions to induce muscle damage with 45 or 10 contractions. After 14 days, subjects performed the same protocol with 45 contractions. Although the effect of the higher volume of the first bout on damage markers (CK, DOMS, and isometric torque) was greater, the protective effects of both protocols were similar. Therefore, the intensity of the first bout seems to be the main aspect of the magnitude of muscle damage and RBE. Because one exercise bout is enough to generate the RBE, some studies have also investigated whether resistance training could also reduce the effects of eccentric exercise on muscle damage markers [67, 79]. Specifically, Newton et al. [67] found that resistance-trained subjects demonstrated smaller RBE when compared with untrained subjects. More- over, Falvo et al. [79] did not find changes in indirect markers of muscle damage (maximal isometric torque and CK) in resistance-trained men. The authors attributed the absence of RBE to a lack of neural adaptation. Thus, it is likely that strength training induces to adaptations that reduce the RBE. The majority of studies that have analysed the RBE used relatively short time periods after the eccentric exercise (i.e., from approximately 7–14 days to 6–9 weeks). However, some studies [80, 81] have reported that the RBE induced by 24 maximal eccentric actions of the elbow flexors may last up to 6 months. Nosaka et al. [80] aimed to investigate the responses of the main indirect markers of muscle damage (CK, maximal isometric torque, DOMS, and swelling) five days after the eccentric exercise bout, with sessions six, nine, and twelve months apart. The main finding of this study was that the RBE for strength, swelling, DOMS, and CK lasted up to six months. 4. Strength Exercise, Muscle Damage, and Running Economy A variety of studies have investigated the influence of EIMD and DOMS on neuromuscular performance indicators (i.e., strength and rate of force development) [82–84]. These studies verified that the isomeric peak torque is compromised immediately after the damaging exercise that causes DOMS, with a gradual recovery in subsequent days. The magnitude and the recovery rate from strength loss seem to be related to the training history of the muscle group. For instance, when performing maximal eccentric contractions, upper limb muscles (less active) demonstrate greater loss of strength (50–70%) and slower recovery (60–90 days) when compared to lower limb muscles (locomotory muscles) (20–30% and 10–30 days, resp.) [37, 85]. However, only a few studies have investigated the effects of EIMD and DOMS on aerobic per- formance indexes (e.g., VO2max, lactate response to exercise, VO2 kinetics, and movement economy) [86]. These studies analysed the effects of EIMD on RE [10, 12] and VO2 kinetics during submaximal cycling exercise [86]. In this context, studies that investigated the effects of strength exercises (i.e., jumps, isoinertial and isokinetic eccentric exercises) on RE will be addressed (Table 1). Paschalis et al. [10] analysed the effects of eccentric exercises on indirect muscle damage markers (CK, DOMS, ROM, and isometric force) and RE in active individuals who were not engaged in strength training programs. The eccentric exercise protocol consisted of 120 (12 × 10) maximal voluntary contractions (MVC) at an angular velocity of 1.05 rad⋅s−1. Although indirect muscle damage markers were significantly altered in the subsequent days (24–72 h), the RE (assessed at 55 and 75% VO2max) was not modified. Similar data were obtained by Vassilis et al. [87], who analysed The Scientific World Journal 5 Table 2: Comparison of the effects of the downhill running on running economy. Study Subjects EIMD Muscle damage VO2max (%) RE (%) Chen et al. [11] 50 male students 30 min DHR at −15% ↑ CK, ↑ DOMS, ↓ strength, and ↑ LDH 70, 80, and 90 ↓ 5 Hamill et al. [92] 10 recreational female runners 30 min DHR at −15% ↑ CK, ↑ DOMS 80 √ Braun and Dutto [93] 9 endurance trained men 30 min DHR at −10% ↑ DOMS 65, 75, and 85 ↓ 3 Chen et al. [94] 10 soccer trained men 30 min DHR at −15% ↑ CK, ↑ DOMS, ↓ strength, and ↑ MB 65, 75, and 85 ↓ 4–7 EIMD: exercise-induced muscle damage; DHR: downhill running; %VO2max: exercise intensity at which running economy was measured; RE: running economy; CK: creatine kinase; DOMS: delayed onset muscle soreness; MB: myoglobin; LDH: lactate dehydrogenase; ↓ indicates decrease; √ indicates no change; ↑ indicates increase. the effects of eccentric exercise (120 MVC at a 60∘⋅s−1) on RE in recreational athletes with no previous experience in resistance training. The RE (assessed at 70% VO2max) was not changed 48 hours after the damaging bout. Therefore, EIMD induced by isokinetic eccentric contractions do not seem to interfere on RE measured at moderate intensities (55– 75% VO2max). Using closed kinetic-chain exercises, Scott et al. [88] have also analysed the effects of EIMD on RE. The vol- unteers performed a series of lower extremity resistance exercises designed to induce DOMS. RE was analysed at 70% VO2max, 24–30 hours after the EIMD. Although the subjects demonstrated a higher rate of perceived exertion values, RE was maintained unaltered throughout the days after EIMD. In another study, Marcora and Bosio [9] did not find any alteration in RE (70% VO2max) after 100 drop jumps, although DOMS, CK, and knee extensors strength were significantly affected by EIMD. Therefore, the evidence suggests that muscle damage induced by both open and closed kinetic-chain exercises dose not alter RE at moderate intensities (55–75% VO2max). However, in a recent study, Burt et al. [12] presented con- flicting data regarding the effect of EIMD on RE. In this study, indirect markers of muscle damage and RE were measured, 24–48 h after EIMD (10 sets of 10 squats at 80% body mass). Significant increases in all indirect markers of muscle damage, kinematic parameters (stride length and stride frequency), and oxygen uptake during submaximal running (∼90% VO2max) were observed at 24–48 h following the initial bout of EIMD. Some authors [82, 89] have suggested that the changes in RE are associated with decrements in neuromuscular function (i.e., MVC) after EIMD. However, both the magnitude and the time course of the changes in muscular function (MVC) and RE can be different. Therefore, changes during submaximal exercise (i.e., RE) may not be strictly associated with neuromuscular function. As a whole, these data suggest that the effects of muscle damage induced by strength exercise (i.e., jumps, isoinertial and isokinetic eccentric exercises) on RE are intensityd- dependent. During moderate exercises (55–75% VO2max), RE is not altered by EIMD. However, during high-intensity exercise (∼90% VO2max), RE is impaired. During high intensity exercise, additional type II fibres, which are the most affected by EIMD, are recruited. Moreover, at these intensities, the VO2 either attains a delayed steady state (heavy domain) or continues to increase slowly (i.e., VO2 slow component (VO2SC)) reaching its maximal values at the end of exercise (severe domain) [90]. Although the physio- logical determinants of VO2SC remain poorly understood, some authors have proposed that an increased ATP and/or O2 cost of power production in fatigued fibres, rather than the additional recruitment of poorly efficient muscle fibres, is responsible for the VO2SC [91]. Therefore, the effects of strength exercise-induced muscle damage on RE seem to depend on the fibre recruitment pattern and/or on the mechanisms determining the VO2SC. 5. Downhill and Long-Distance Running, Muscle Damage, and Running Economy Adopting a more specific approach, some studies have inves- tigated the influence of muscle damage induced by strenuous exercise (e.g., long-distance running) or downhill running on neuromuscular parameters, and RE. This aspect and the effect of some interventions on RE during the recovery period after EIMD are also addressed in this topic (Table 2). As mentioned previously, muscle damage is usually induced by maximal and submaximal eccentric contractions, but it can also be observed when a high volume of eccen- tric/concentric contractions are performed, due to the eccen- tric contractions per se [95] or because of metabolite accu- mulation that may lead to stress and impairment of the muscle fibres [96]. Because a high number of concentric and, particularly, eccentric contractions are performed during long-distance running, the symptoms of muscle damage 6 The Scientific World Journal are usually observed immediately and a few days after the running bout. In a study conducted by Millet et al. [97], changes in muscle function and muscle damage markers from 22 expe- rienced marathon runners were collected and analysed after they had run an international extreme mountain ultrama- rathon. The race consisted of a 166 km marathon through mountainous terrain with the final destination set at 9500 m below the starting point. This predominately downhill con- figuration required a high number of eccentric contractions particularly for the knee extensors. Indirect muscle damage markers (strength, CK, LDH, and MB) were analysed before, immediately after, and 2, 5, 7, 9, and 16 days after the marathon. The authors found higher decreases in force production immediately after the ultramarathon, most likely because of the fatigue experienced during the race. However, some of the strength markers remained altered until 5 days after exercise, as usually occurs after muscle damaging activities. Blood markers also demonstrated the highest value immediately after the race, returning to baseline values 5 days later. The authors found that even though this type of activity can induce extreme muscle damage, after 16 days, all the alterations induced by muscle damage and/or fatigue had returned to normal. Considering that force production is intimately related to RE, these findings may indicate that an extremely damaging activity may induce high levels of force loss and decreases in RE. Force production is usually fully recovered 5 days after the damaging activity. However, RE may recover at a faster rate than force. To investigate factors that could influence RE, Kyrolainen et al. [89] subjected 7 experienced runners to a protocol sim- ulating a marathon. RE and kinematic variables (stride length and frequency, mean contact time, external mechanical work and power, and angular displacements and velocities of the hip, knee, and ankle joints) were collected before, during (at the 1st, 13th, 26th, and 42nd kilometres), two hours after, and in the days (2, 4, and 6) after the marathon. Muscle damage markers (CK and SOR) were also collected after the exercise. The impairment of RE (i.e., higher oxygen consumption) was observed only at the end of the marathon (42nd kilometre and two hours afterward). CK and SOR were significantly increased immediately after the marathon and returned to baseline values only at the 6-day postexercise time point. These data may indicate that alterations in RE after marathon running may not be exclusively in the result of muscle damage but may be affected by other factors such as thermal stress. To better understand the time course of recovery of the various parameters of muscle function following marathon running, it is important to investigate other indirect EIMD markers, such as force and inflammatory response. Muscle damage induced by downhill running has also been widely studied in the last decades. This type of exercise has been proven to induce muscle damage even when performed for relatively short periods (e.g., 30 minutes) due to higher mechanical stress applied to the lower limb muscles during the contact with the ground phase [95]. Some studies have shown that downhill running can lead to muscle damage of the same magnitude as plyometric or maximal eccen- tric exercises [92, 98]. Considering that downhill running induces muscle damage, a series of studies have investigated its influence on neuromuscular and metabolic markers in animals [99] and humans [11, 93, 94, 98]. In animals, downhill running has been utilised to induce overtraining [100] as well as a training method to increase the number of sarcomeres [101]. In humans, this exercise model has been recently studied in attempt to understand its influence on specific running and aerobic parameters, such as RE [93, 98] and running kinematics. To the best of our knowledge, Hamill et al. [92] performed the first study investigating the influence of downhill run- ning on RE. In this study, 10 recreational female runners underwent a 30 min downhill running bout (DRB) with −15% slope at 73.5% of maximal heart rate. Indirect markers of muscle damage (SOR and CK), RE (80% VO2max), and kinematic parameters were measured before and 2 and 5 days after the DRB. SOR and CK levels increased 2 days after the DRB, returning to baseline values 5 days after the exercise. Although kinematic parameters were modified, the DRB did not alter RE. The authors proposed that changes in kinematics might be due to increases in SOR, which compromises the range of motion, and thus alters the movement patterns. In another model to investigate the influence of downhill running-induced muscle damage on RE, Braun and Dutto [93] conducted a study in which 9 endurance-trained subjects underwent a DRB (30 minutes at 70% VO2peak with a −10% slope). Assessments of SOR, RE (65%, 75% and 85% VO2max), and stride length were performed before and 48 hours after the DRB. SOR was increased and RE was impaired 48 hours after the DRB, suggesting that muscle damage might have increased the energy cost of running. The authors stated that the muscle damage decreased the range of motion and strength, thus compromising running kinematics, which is known to be related to RE. To better describe the time course of changes in RE, Chen et al. [94] subjected 10 soccer-trained volunteers to a downhill running protocol similar to that proposed by Braun and Dutto [93]. Muscle damage (MVC, SOR, CK and MB) and RE (65%, 75%,d and 85% VO2max) were assessed before and 1 hour and 1–5 days after the DRB. Alterations in muscle damage markers were consistent with those found in the literature, including increases in CK, MB, and SOR, with peak values attained 48 hours after the DRB. Strength loss was also maximal immediately after the DRB. All muscle damage markers returned to baseline values 5 days after the DRB. The magnitude of change was smaller, and the time course recovery was faster for RE (4–7% and 4 days, resp.) than for the indirect markers (i.e., isometric peak torque (IPT)) of muscle damage (7–21% and 4 days, resp.). The authors suggested that the alterations in running kinematics, the need to recruit more muscle fibres, the impairment in the stretch- shortening cycle, and the reduced levels of muscle glycogen might impair RE following a DRB. Because alterations in the muscular tissue due to EIMD have been shown to affect RE because of differences in muscle fibre recruitment and other neuromuscular properties, Chen et al. [11] assessed RE at 3 different intensities (70, 80, and 90%VO2peak) after a DRB. Muscle damage markers showed the expected alterations, peaking 2 days after DRB. The The Scientific World Journal 7 alteration in RE measured at 90% VO2peak was significantly higher than at 80% VO2peak. No significant change in RE was found at 70% VO2peak. Previous studies have indicated that fast-twitch motor units are progressively recruited with increased levels of exercise intensity [102]. Because several investigations have reported selective damage to type II muscle fibres after eccentric muscle actions in humans [40, 103], there appears to be a relationship between the motor unit recruitment pattern and impairment in RE. Therefore, the effects of strength exercises and down- hill running on RE seem to be different. While strength exercises seem to affect RE only during high intensity sub- maximal exercises (∼90% VO2max), downhill running also increases the energetic cost during moderate exercises (>65% VO2max). It is important to note that during running exercise, the VO2SC is attenuated (heavy intensity) and/or nonexistent (moderate). Thus, the effect of strength exercises on RE seems to occur only at running intensities at which the VO2SC is present. Greater muscle mass and/or the magnitude or specificity of muscle damage induced by downhill running may partially explain these results. A variety of interventions have been proposed to enhance recovery from EIMD, that is, to reduce the severity and duration of injury and SOR. It is a common belief that low- intensity training (i.e., active recovery) enhances the recovery process by accelerating the return to homeostasis after EIMD. To investigate whether submaximal running would influence the recovery from DRB, Chen et al. [98] analysed the effect of 30-minute daily running exercises performed at different intensities (40%, 50%, 60%, and 70% VO2peak) by different groups on the recovery of muscle damage and RE. Muscle damage was induced by a DRB (30 minutes at 70% VO2peak with −10% slope). The authors found that the time-course recovery of muscle damage markers and RE was similar for all groups, regardless of whether submaximal running was performed. Thus, low-to-moderate-intensity running seems not to improve the recovery from muscle damage and/or RE impairment. Performing a similar subsequent bout of eccentric exer- cise results in significantly less change in the markers of muscle damage. This phenomenon is known as the RBE [104]. In fact, Byrnes et al. [105] and Chen et al. [66] demonstrated that when a DRB was repeated 1–6 weeks after the first bout, the indirect markers of muscle damage (isometric peak torque, CK, SOR, and range of motion) were significantly reduced. Moreover, Chen et al. [106] verified that the RBE was also observed in RE and running kinematics parameters. In this study, 12 male subjects underwent the same downhill running protocol adopted by Chen et al. [11] except the interval allowed between bouts that was twice as long, to allow full recovery from the first bout. The authors found significant changes in all markers of muscle damage after both protocols. However, the RE, kinematics parameters, and SOR were less affected after the second DRB. Therefore, one bout of DRB might induce a protective effect, leading to reduced levels of SOR and blunted changes in RE and biomechanical parameters. In another study, Burt et al. [12] subjected 9 subjects to repeated bouts of 100 squats and measured muscle damage markers (isometric peak torque, vertical jump height, CK, and SOR) and RE before, immediately after the bouts, and 1-2 days after the bouts. The bouts were separated by enough time to recover from the EIMD symptoms. All muscle damage and RE markers were significantly affected by the first bout. However, no alterations in some muscle damage markers and RE were observed after the second bout. Thus, a previous damaging activity leads to blunted or nonexistent alterations in muscle damage markers and RE. 6. Supplementation and Muscle Damage Recovery A series of recent studies has investigated the influence of different types of supplementation on recovery from and prevention of muscle damage. The main supplements that seem to protect against muscle damage are the flavonoids, which are known for their efficient anti-inflammatory and antioxidant properties. Studies investigating supplementa- tion with flavonoid rich substances and their influence on muscle damage will be discussed. Howatson et al. [107] conducted a study in which muscle damage, inflammatory response, and oxidative stress were measured before, immediately after, and 24 and 48 hours after a marathon. The purpose of this study was to inves- tigate whether a tart cherry juice supplement would affect recovery from muscle damage after marathon running in 20 recreational marathon runners, using a double-blind placebo intervention. Muscle damage markers determined in this study were CK, LDH, DOMS, and IPT. Other parameters (total antioxidant status, thiobarbituric acid reactive species,d and protein carbonyls) were measured to identify inflam- matory response, and oxidative stress. Both groups (control versus supplemented) demonstrated similar decreases in IPT. However, IPT was higher for the supplementation group at all time points, showing faster recovery. Moreover, the authors found that the supplementation enhanced the anti- inflammatory response as well as reduced the oxidative stress. Kuehl et al. [108] investigated the effects of tart cherry juice supplementation on SOR immediately after a long- distance running (∼26 km) bout. The study was performed in a randomised, double-blind placebo fashion in 54 experi- enced runners. The subjects were separated in two groups: placebo and tart cherry supplement. Both groups started ingesting their supplements 7 days prior to the running bout. The increase in SOR was significantly greater for the placebo group when compared to the tart cherry group. These findings are similar to those of Howatson et al. [107] and indicate that the anti-inflammatory and antioxidant properties of the tart cherry supplement might reduce SOR after EIMD. Supplements containing flavonoid compounds have been shown to confer a protective effect against muscle damage either by attenuating a vast number of markers or by accel- erating their recovery after the EIMD. This type of protection has been hypothesised to be due to the anti-inflammatory and antioxidant properties present in these types of compounds [107]. A number of studies have shown a direct relationship between flavonoid supplementation and protection against 8 The Scientific World Journal muscle damage. Because muscle damage may affect RE, it would be interesting to analyse if this type of supplementation can protect against RE impairment after EIMD. 7. Conclusion Despite the systematic implications of RE on aerobic running performance, only a few experiments have specifically stud- ied the response of this index after EIMD. Recent studies have analysed RE after strength exercises (i.e., jumps, isoinertial and isokinetic eccentric exercises) and downhill running. These studies have found that the magnitude of reduction in muscle function (MVC) after EIMD is greater than the RE and kinematic parameters. Moreover, the time course for the changes in muscle function, RE, and kinematic parameters are not similar. As a whole, these data suggest that the putative mechanisms underlying muscle function and RE during the recovery from EIMD are not completely shared. The effects of muscle damage on RE seem to depend of the interaction between the type of eccentric exercise and the intensity at which the RE is measured. Strength exercises seem to modify RE preferentially during high-intensity exercise (∼ 90% VO2max). However, the effects of downhill running can also be observed at moderate intensities (>65% VO2max). Finally, a single session of strength exercise or downhill running attenuates changes in indirect markers of muscle damage and blunts changes in RE. Acknowledgments This research was supported by grants from Conselho Nacional de Desenvolvimento Cient´ıfico e Tecnol´ogico and Fundac¸˜ao de Amparo a Pesquisa do Estado de S˜ao Paulo, and the authors declar that they have no conflict of interests. References [1] J. Daniels and N. Daniels, “Running economy of elite male and elite female runners,” Medicine and Science in Sports and Exercise, vol. 24, no. 4, pp. 483–489, 1992. [2] T. Anderson, “Biomechanics and running economy,” Sports Medicine, vol. 22, no. 2, pp. 76–89, 1996. [3] P. U. Saunders, D. B. Pyne, R. D. Telford, and J. A. Hawley, “Fac- tors affecting running economy in trained distance runners,” Sports Medicine, vol. 34, no. 7, pp. 465–485, 2004. [4] A. M. Turner, M. Owings, and J. A. Schwane, “Improvement in running economy after 6 weeks of plyometric training,” Journal of Strength and Conditioning Research, vol. 17, no. 1, pp. 60–67, 2003. [5] L. G. A. Guglielmo, C. C. Greco, and B. S. Denadai, “Effects of strength training on running economy,” International Journal of Sports Medicine, vol. 30, no. 1, pp. 27–32, 2009. [6] P. U. Saunders, R. D. Telford, D. B. Pyne, A. G. Hahn, and C. J. Gore, “Improved running economy and increased hemoglobin mass in elite runners after extended moderate altitude expo- sure,” Journal of Science and Medicine in Sport, vol. 12, no. 1, pp. 67–72, 2009. [7] M. D. Carmichael, J. M. Davis, E. A. Murphy et al., “Recovery of running performance following muscle-damaging exercise: relationship to brain IL-1𝛽,” Brain, Behavior, and Immunity, vol. 19, no. 5, pp. 445–452, 2005. [8] M. D. Carmichael, J. M. Davis, E. A. Murphy et al., “Role of brain IL-1𝛽 on fatigue after exercise-induced muscle damage,” American Journal of Physiology, vol. 291, no. 5, pp. R1344–R1348, 2006. [9] S. M. Marcora and A. Bosio, “Effect of exercise-induced muscle damage on endurance running performance in humans,” Scan- dinavian Journal of Medicine and Science in Sports, vol. 17, no. 6, pp. 662–671, 2007. [10] V. Paschalis, Y. Koutedakis, V. Baltzopoulos, V. Mougios, A. Z. Jamurtas, and V. Theoharis, “The effects of muscle damage on running economy in healthy males,” International Journal of Sports Medicine, vol. 26, no. 10, pp. 827–831, 2005. [11] T. C. Chen, K. Nosaka, M. J. Lin, H. L. Chen, and C. J. Wu, “Changes in running economy at different intensities following downhill running,” Journal of Sports Sciences, vol. 27, no. 11, pp. 1137–1144, 2009. [12] D. Burt, K. Lamb, C. Nicholas, and C. Twist, “Effects of repeated bouts of squatting exercise on sub-maximal endurance running performance,” European Journal of Applied Physiology, vol. 113, no. 2, pp. 285–293, 2013. [13] A. M. Jones and H. Carter, “The effect of endurance training on parameters of aerobic fitness,” Sports Medicine, vol. 29, no. 6, pp. 373–386, 2000. [14] P. A. Farrell, J. H. Wilmore, and E. F. Coyle, “Plasma lactate accumulation and distance running performance,” Medicine and Science in Sports and Exercise, vol. 11, no. 4, pp. 338–344, 1979. [15] R. R. Pate, C. A. Macera, S. P. Bailey, W. P. Bartoli, and K. E. Powell, “Physiological, anthropometric, and training correlates of running economy,” Medicine and Science in Sports and Exercise, vol. 24, no. 10, pp. 1128–1133, 1992. [16] D. W. Morgan and J. T. Daniels, “Relationship between VO2max and the aerobic demand of running in elite distance runners,” International Journal of Sports Medicine, vol. 15, no. 7, pp. 426– 429, 1994. [17] E. F. Coyle, A. R. Coggan, M. K. Hopper, and T. J. Walters, “Determinants of endurance in well-trained cyclists,” Journal of Applied Physiology, vol. 64, no. 6, pp. 2622–2630, 1988. [18] D. L. Costill, H. Thomason, and E. Roberts, “Fractional utiliza- tion of the aerobic capacity during distance running,” Medicine and Science in Sports and Exercise, vol. 5, no. 4, pp. 248–252, 1973. [19] D. L. Conley and G. S. Krahenbuhl, “Running economy and dis- tance running performance of highly trained athletes,” Medicine and Science in Sports and Exercise, vol. 12, no. 5, pp. 357–360, 1980. [20] D. W. Morgan, P. E. Martin, G. S. Krahenbuhl, and F. D. Baldini, “Variability in running economy and mechanics among trained male runners,” Medicine and Science in Sports and Exercise, vol. 23, no. 3, pp. 378–383, 1991. [21] W. A. Sparrow and K. M. Newell, “Metabolic energy expendi- ture and the regulation of movement economy,” Psychonomic Bulletin and Review, vol. 5, no. 2, pp. 173–196, 1998. [22] D. W. Morgan, P. E. Martin, and G. S. Krahenbuhl, “Factors affecting running economy,” Sports Medicine, vol. 7, no. 5, pp. 310–330, 1989. [23] L. W. Armstrong and C. M. Maresh, “The induction and decay of heat acclimatisation in trained athletes,” Sports Medicine, vol. 12, no. 5, pp. 302–312, 1991. The Scientific World Journal 9 [24] P. U. Saunders, R. D. Telford, D. B. Pyne et al., “Short- term plyometric training improves running economy in highly trained middle and long distance runners,” Journal of Strength and Conditioning Research, vol. 20, no. 4, pp. 947–954, 2006. [25] Ø. Støren, J. Helgerud, E. M. Støa, and J. Hoff, “Maximal strength training improves running economy in distance run- ners,” Medicine and Science in Sports and Exercise, vol. 40, no. 6, pp. 1087–1092, 2008. [26] L. J. Brandon, “Physiological factors associated with middle distance running performance,” Sports Medicine, vol. 19, no. 4, pp. 268–277, 1995. [27] B. S. Denadai, M. J. Ortiz, C. C. Greco, and M. T. De Mello, “Interval training at 95% and 100% of the velocity at VO2 max: effects on aerobic physiological indexes and running performance,” Applied Physiology, Nutrition and Metabolism, vol. 31, no. 6, pp. 737–743, 2006. [28] C. Bosco, G. Montanari, R. Ribacchi et al., “Relationship between the efficiency of muscular work during jumping and the energetics of running,” European Journal of Applied Physi- ology and Occupational Physiology, vol. 56, no. 2, pp. 138–143, 1987. [29] J. R. Fletcher, S. P. Esau, and B. R. MacIntosh, “Changes in ten- don stiffness and running economy in highly trained distance runners,” European Journal of Applied Physiology, vol. 110, no. 5, pp. 1037–1046, 2010. [30] A. Arampatzis, G. De Monte, K. Karamanidis, G. Morey-Klaps- ing, S. Stafilidis, and G. P. Br¨uggemann, “Influence of the muscle-tendon unit’s mechanical and morphological properties on running economy,” The Journal of Experimental Biology, vol. 209, no. 17, pp. 3345–3357, 2006. [31] R. W. Spurrs, A. J. Murphy, and M. L. Watsford, “The effect of plyometric training on distance running performance,” Euro- pean Journal of Applied Physiology, vol. 89, no. 1, pp. 1–7, 2003. [32] B. D. Levine and J. Stray-Gundersen, “’Living high-training low: effect of moderate-altitude acclimatization with low-altitude training on performance,” Journal of Applied Physiology, vol. 83, no. 1, pp. 102–112, 1997. [33] E. Y. Robertson, P. U. Saunders, D. B. Pyne, C. J. Gore, and J. M. Anson, “Effectiveness of intermittent training in hypoxia combined with live high/train low,” European Journal of Applied Physiology, vol. 110, no. 2, pp. 379–387, 2010. [34] J. P. Wehrlin, P. Zuest, J. Hall´en, and B. Marti, “Live high-train low for 24 days increases hemoglobin mass and red cell volume in elite endurance athletes,” Journal of Applied Physiology, vol. 100, no. 6, pp. 1938–1945, 2006. [35] J. Svedenhag, “Running economy,” in Running and Science, J. Bangsbo and H. Larsen, Eds., pp. 85–105, Munksgaard, Copenhagen, Denmark, 2000. [36] J. A. Houmard, D. L. Costill, J. A. Davis, J. B. Mitchell, D. D. Pascoe, and R. A. Robergs, “The influence of exercise intensity on heat acclimation in trained subjects,” Medicine and Science in Sports and Exercise, vol. 22, no. 5, pp. 615–620, 1990. [37] C. Byrne, C. Twist, and R. Eston, “Neuromuscular function after exercise-induced muscle damage theoretical and applied implications,” Sports Medicine, vol. 34, no. 1, pp. 49–69, 2004. [38] J. C. Tee, A. N. Bosch, and M. I. Lambert, “Metabolic conse- quences of exercise-induced muscle damage,” Sports Medicine, vol. 37, no. 10, pp. 827–836, 2007. [39] G. L. Warren, D. A. Lowe, and R. B. Armstrong, “Measurement tools used in the study of eccentric contraction-induced injury,” Sports Medicine, vol. 27, no. 1, pp. 43–59, 1999. [40] J. Friden, M. Sjostrom, and B. Ekblom, “Myofibrillar damage following intense eccentric exercise in man,” International Journal of Sports Medicine, vol. 4, no. 3, pp. 170–176, 1983. [41] M. J. Gibala, J. D. MacDougall, M. A. Tarnopolsky, W. T. Stauber, and A. Elorriaga, “Changes in human skeletal muscle ultra- structure and force production after acute resistance exercise,” Journal of Applied Physiology, vol. 78, no. 2, pp. 702–708, 1995. [42] T. Hortob´agyi, J. Houmard, D. Fraser, R. Dudek, J. Lambert, and J. Tracy, “Normal forces and myofibrillar disruption after repeated eccentric exercise,” Journal of Applied Physiology, vol. 84, no. 2, pp. 492–498, 1998. [43] C. Byrne, R. G. Eston, and R. H. T. Edwards, “Characteristics of isometric and dynamic strength loss following eccentric exercise-induced muscle damage,” Scandinavian Journal of Medicine and Science in Sports, vol. 11, no. 3, pp. 134–140, 2001. [44] W. T. Stauber, P. M. Clarkson, V. K. Fritz, and W. J. Evans, “Extracellular matrix disruption and pain after eccentric muscle action,” Journal of Applied Physiology, vol. 69, no. 3, pp. 868–874, 1990. [45] J. Frid´en and R. L. Lieber, “Eccentric exercise-induced injuries to contractile and cytoskeletal muscle fibre components,” Acta Physiologica Scandinavica, vol. 171, no. 3, pp. 321–326, 2001. [46] J. B. Rodenburg, P. R. Bar, and R. W. De Boer, “Relations between muscle soreness and biochemical and functional out- comes of eccentric exercise,” Journal of Applied Physiology, vol. 74, no. 6, pp. 2976–2983, 1993. [47] D. B. Pyne, “Exercise-induced muscle damage and inflamma- tion: a review,” Australian Journal of Science and Medicine in Sport, vol. 26, no. 3-4, pp. 49–58, 1994. [48] T. C. Chen, H. L. Chen, A. J. Pearce, and K. Nosaka, “Attenuation of eccentric exercise-induced muscle damage by precondition- ing exercises,” Medicine and Science in Sports and Exercise, vol. 44, no. 11, pp. 2090–2098, 2012. [49] P. V. Komi and J. T. Viitasalo, “Changes in motor unit activity and metabolism in human skeletal muscle during and after repeated eccentric and concentric contractions,” Acta Physiolog- ica Scandinavica, vol. 100, no. 2, pp. 246–254, 1977. [50] P. M. Clarkson, W. C. Byrnes, K. M. McCormick, L. P. Turcotte, and J. S. White, “Muscle soreness and serum creatine kinase activity following isometric, eccentric, and concentric exercise,” International Journal of Sports Medicine, vol. 7, no. 3, pp. 152–155, 1986. [51] K. Vissing, K. Overgaard, A. Nedergaard, A. Fredsted, and P. Schjerling, “Effects of concentric and repeated eccentric exer- cise on muscle damage and calpain-calpastatin gene expression in human skeletal muscle,” European Journal of Applied Physiol- ogy, vol. 103, no. 3, pp. 323–332, 2008. [52] D. M. DiPasquale, R. J. Bloch, and R. M. Lovering, “Deter- minants of the repeated-bout effect after lengthening contrac- tions,” American Journal of Physical and Medicine Rehabilita- tion, vol. 90, no. 10, pp. 816–824, 2011. [53] G. Y. Millet and R. Lepers, “Alterations of neuromuscular function after prolonged running, cycling and skiing exercises,” Sports Medicine, vol. 34, no. 2, pp. 105–116, 2004. [54] R. M. Enoka, “Eccentric contractions require unique activation strategies by the nervous system,” Journal of Applied Physiology, vol. 81, no. 6, pp. 2339–2346, 1996. [55] S. H. Westing, A. G. Cresswell, and A. Thorstensson, “Muscle activation during maximal voluntary eccentric and concentric knee extension,” European Journal of Applied Physiology and Occupational Physiology, vol. 62, no. 2, pp. 104–108, 1991. 10 The Scientific World Journal [56] E. Kellis and V. Baltzopoulos, “Muscle activation differences between eccentric and concentric isokinetic exercise,” Medicine and Science in Sports and Exercise, vol. 30, no. 11, pp. 1616–1623, 1998. [57] R. M. Enoka, Neuromechanics of Human Movementedition, Human Kinetics Books, Champaign, Ill, USA, 3rd edition, 2002. [58] D. A. Jones, D. J. Newham, and C. Torgan, “Mechanical influ- ences on long-lasting human muscle fatigue and delayed-onset pain,” The Journal of Physiology, vol. 412, pp. 415–427, 1989. [59] U. Proske and D. L. Morgan, “Muscle damage from eccentric exercise: mechanism, mechanical signs, adaptation and clinical applications,” The Journal of Physiology, vol. 537, no. 2, pp. 333– 345, 2001. [60] D. G. Allen and H. Westerblad, “Role of phosphate and calcium stores in muscle fatigue,” The Journal of Physiology, vol. 536, no. 3, pp. 657–665, 2001. [61] R. Child, S. Brown, S. Day, A. Donnelly, H. Roper, and J. Saxton, “Changes in indices of antioxidant status, lipid peroxidation and inflammation in human skeletal muscle after eccentric muscle actions,” Clinical Science, vol. 96, no. 1, pp. 105–115, 1999. [62] K. Nosaka and P. M. Clarkson, “Muscle damage following repeated bouts of high force eccentric exercise,” Medicine and Science in Sports and Exercise, vol. 27, no. 9, pp. 1263–1269, 1995. [63] M. P. McHugh, “Recent advances in the understanding of the repeated bout effect: the protective effect against muscle damage from a single bout of eccentric exercise,” Scandinavian Journal of Medicine and Science in Sports, vol. 13, no. 2, pp. 88–97, 2003. [64] M. P. McHugh, D. A. J. Connolly, R. G. Eston, and G. W. Gleim, “Exercise-induced muscle damage and potential mechanisms for the repeated bout effect,” Sports Medicine, vol. 27, no. 3, pp. 157–170, 1999. [65] K. Nosaka and K. Sakamoto, “Effect of elbow joint angle on the magnitude of muscle damage to the elbow flexors,” Medicine and Science in Sports and Exercise, vol. 33, no. 1, pp. 22–29, 2001. [66] T. C. Chen, K. Nosaka, and P. Sacco, “Intensity of eccentric exer- cise, shift of optimum angle, and the magnitude of repeated- bout effect,” Journal of Applied Physiology, vol. 102, no. 3, pp. 992–999, 2007. [67] M. J. Newton, G. T. Morgan, P. Sacco, D. W. Chapman, and K. Nosaka, “Comparison of responses to strenuous eccentric exercise of the elbow flexors between resistance-trained and untrained men,” Journal of Strength and Conditioning Research, vol. 22, no. 2, pp. 597–607, 2008. [68] T. C. Chen, H. L. Chen, M. J. Lin, C. J. Wu, and K. Nosaka, “Potent protective effect conferred by four bouts of low- intensity eccentric exercise,” Medicine and Science in Sports and Exercise, vol. 42, no. 5, pp. 1004–1012, 2010. [69] G. Howatson, K. Van Someren, and T. Hortob´agyi, “Repeated bout effect after maximal eccentric exercise,” International Jour- nal of Sports Medicine, vol. 28, no. 7, pp. 557–563, 2007. [70] T. C. Chen and K. Nosaka, “Responses of elbow flexors to two strenuous eccentric exercise bouts separated by three days,” Journal of Strength and Conditioning Research, vol. 20, no. 1, pp. 108–116, 2006. [71] G. L. Warren, K. M. Hermann, C. P. Ingalls, M. R. Masselli, and R. B. Armstrong, “Decreased EMG median frequency during a second bout of eccentric contractions,” Medicine and Science in Sports and Exercise, vol. 32, no. 4, pp. 820–829, 2000. [72] G. Howatson and K. A. Van Someren, “Ice massage: effects on exercise-induced muscle damage,” Journal of Sports Medicine and Physical Fitness, vol. 43, no. 4, pp. 500–505, 2003. [73] K. Nosaka, M. Newton, P. Sacco, D. Chapman, and A. Lavender, “Partial protection against muscle damage by eccentric actions at short muscle lengths,” Medicine and Science in Sports and Exercise, vol. 37, no. 5, pp. 746–753, 2005. [74] D. J. Newham, D. A. Jones, and P. M. Clarkson, “Repeated high- force eccentric exercise: effects on muscle pain and damage,” Journal of Applied Physiology, vol. 63, no. 4, pp. 1381–1386, 1987. [75] H. S. Thompson, P. M. Clarkson, and S. P. Scordilis, “The repeated bout effect and heat shock proteins: intramuscular HSP27 and HSP70 expression following two bouts of eccentric exercise in humans,” Acta Physiologica Scandinavica, vol. 174, no. 1, pp. 47–56, 2002. [76] T. M. Lehti, R. Kalliokoski, and J. Komulainen, “Repeated bout effect on the cytoskeletal proteins titin, desmin, and dystrophin in rat skeletal muscle,” Journal of Muscle Research and Cell Motility, vol. 28, no. 1, pp. 39–47, 2007. [77] R. B. Child, J. M. Saxton, and A. E. Donnelly, “Comparison of eccentric knee extensor muscle actions at two muscle lengths on indices of damage and angle-specific force production in humans,” Journal of Sports Sciences, vol. 16, no. 4, pp. 301–308, 1998. [78] S. V. Brooks and J. A. Faulkner, “Severity of contraction-induced injury is affected by velocity only during stretches of large strain,” Journal of Applied Physiology, vol. 91, no. 2, pp. 661–666, 2001. [79] M. J. Falvo, B. K. Schilling, R. J. Bloomer, and W. A. Smith, “Repeated bout effect is absent in resistance trained men: an electromyographic analysis,” Journal of Electromyography and Kinesiology, vol. 19, no. 6, pp. e529–e535, 2009. [80] K. Nosaka, K. Sakamoto, M. Newton, and P. Sacco, “How long does the protective effect on eccentric exercise-induced muscle damage last?” Medicine and Science in Sports and Exercise, vol. 33, no. 9, pp. 1490–1495, 2001. [81] P. M. Clarkson, K. Nosaka, and B. Braun, “Muscle function after exercise-induced muscle damage and rapid adaptation,” Medicine and Science in Sports and Exercise, vol. 24, no. 5, pp. 512–520, 1992. [82] R. G. Eston, S. Finney, S. Baker, and V. Baltzopoulos, “Muscle tenderness and peak torque changes after downhill running following a prior bout of isokinetic eccentric exercise,” Journal of Sports Sciences, vol. 14, no. 4, pp. 291–299, 1996. [83] D. L. Macintyre, W. D. Reid, D. M. Lyster, I. J. Szasz, and D. C. Mckenzie, “Presence of WBC, decreased strength, and delayed soreness in muscle after eccentric exercise,” Journal of Applied Physiology, vol. 80, no. 3, pp. 1006–1013, 1996. [84] R. Molina and B. S. Denadai, “Dissociated time course recovery between rate of force development and peak torque after eccentric exercise,” Clinical Physiology and Functional Imaging, vol. 32, no. 3, pp. 179–184, 2012. [85] L. C. R. Lima and B. S. Denadai, “The repeated bout effect: a comparison between upper and lower limbs,” Motriz, vol. 17, no. 4, pp. 738–747, 2011. [86] R. Molina and B. S. Denadai, “Muscle damage slows oxy- gen uptake kinetics during moderate-intensity exercise per- formed ate high pedal rate,” Applied Physiology Nutrition and Metabolism, vol. 36, no. 6, pp. 848–855, 2011. [87] P. Vassilis, B. Vassilios, M. Vassilis et al., “Isokinetic eccentric exercise of quadriceps femoris does not affect running econ- omy,” Journal of Strength and Conditioning Research, vol. 22, no. 4, pp. 1222–1227, 2008. [88] K. E. Scott, R. Rozenek, A. C. Russo, J. A. Crussemeyer, and M. G. Lacourse, “Effects of delayed onset muscle soreness The Scientific World Journal 11 on selected physiological responses to submaximal running,” Journal of Strength and Conditioning Research, vol. 17, no. 4, pp. 652–658, 2003. [89] H. Kyrolainen, T. Pullinen, R. Candau, J. Avela, P. Huttunen, and P. V. Komi, “Effects of marathon running on running economy and kinematics,” European Journal of Applied Physiology, vol. 82, no. 4, pp. 297–304, 2000. [90] G. A. Gaesser and D. C. Poole, “The slow component of oxygen uptake kinetics in humans,” Exercise and Sport Sciences Reviews, vol. 24, pp. 35–71, 1996. [91] D. T. Cannon, A. C. White, M. F. Andriano, F. W. Kolkhorst, and H. B. Rossiter, “Skeletal muscle fatigue precedes the slow com- ponent of oxygen uptake kinetics during exercise in humans,” The Journal of Physiology, vol. 589, no. 3, pp. 727–739, 2011. [92] J. Hamill, P. S. Freedson, P. M. Clarkson, and B. Braun, “Muscle soreness during running: biomechanical and physiological considerations,” Journal of Applied Biomechanics, vol. 7, no. 2, pp. 125–137, 1991. [93] W. A. Braun and D. J. Dutto, “The effects of a single bout of downhill running and ensuing delayed onset of muscle soreness on running economy performed 48 h later,” European Journal of Applied Physiology, vol. 90, no. 1-2, pp. 29–34, 2003. [94] T. C. Chen, K. Nosaka, and J. H. Tu, “Changes in running econ- omy following downhill running,” Journal of Sports Sciences, vol. 25, no. 1, pp. 55–63, 2007. [95] R. G. Eston, J. Mickleborough, and V. Baltzopoulos, “Eccentric activation and muscle damage: biomechanical and physiologi- cal considerations during downhill running,” British Journal of Sports Medicine, vol. 29, no. 2, pp. 89–94, 1995. [96] H. J. Appell, J. M. C. Soares, and J. A. R. Duarte, “Exercise, muscle damage and fatigue,” Sports Medicine, vol. 13, no. 2, pp. 108–115, 1992. [97] G. Y. Millet, K. Tomazin, S. Verges et al., “Neuromuscular consequences of an extreme mountain ultra-marathon,” PLoS ONE, vol. 6, no. 2, Article ID e17059, 2011. [98] T. C. Chen, K. Nosaka, and C. C. Wu, “Effects of a 30-min running performed daily after downhill running on recovery of muscle function and running economy,” Journal of Science and Medicine in Sport, vol. 11, no. 3, pp. 271–279, 2008. [99] S. A. Hahn, L. F. Ferreira, J. B. Williams et al., “Downhill treadmill running trains the rat spinotrapezius muscle,” Journal of Applied Physiology, vol. 102, no. 1, pp. 412–416, 2007. [100] B. C. Pereira, L. A. Filho, G. F. Alves et al., “A new overtraining protocol for mice based on downhill running sessions,” Clinical and Experimental Pharmacology and Physiology, vol. 39, no. 9, pp. 793–798, 2012. [101] R. Lynn and D. L. Morgan, “Decline running produces more sarcomeres in rat vastus intermedius muscle fibers than does incline running,” Journal of Applied Physiology, vol. 77, no. 3, pp. 1439–1444, 1994. [102] B. Essen, “Glycogen depletion of different fibre types in human skeletal muscle during intermittent and continuous exercise,” Acta Physiologica Scandinavica, vol. 103, no. 4, pp. 446–455, 1978. [103] D. A. Jones, D. J. Newham, J. M. Round, and S. E. J. Tolfree, “Experimental human muscle damage: morphological changes in relation to other indices of damage,” The Journal of Physiology, vol. 375, pp. 435–448, 1986. [104] T. C. Chen, “Effects of a second bout of maximal eccentric exercise on muscle damage and electromyographic activity,” European Journal of Applied Physiology, vol. 89, no. 2, pp. 115– 121, 2003. [105] W. C. Byrnes, P. M. Clarkson, J. S. White et al., “Delayed onset muscle soreness following repeated bouts of downhill running,” Journal of Applied Physiology, vol. 59, no. 3, pp. 710–715, 1985. [106] T. C. Chen, H. L. Chen, C. J. Wu et al., “Changes in running economy following a repeated bout of downhill running,” Journal of Exercise Science and Fitness, vol. 5, no. 2, pp. 109–117, 2007. [107] G. Howatson, M. P. McHugh, J. A. Hill et al., “Influence of tart cherry juice on indices of recovery following marathon running,” Scandinavian Journal of Medicine and Science in Sports, vol. 20, no. 6, pp. 843–852, 2010. [108] K. S. Kuehl, E. T. Perrier, D. L. Elliot, and J. C. Chesnutt, “Efficacy of tart cherry juice in reducing muscle pain during running: a randomized controlled trial,” Journal of the Interna- tional Society of Sports Nutrition, vol. 7, article 17, 2010.

Exercise-induced muscle damage and running economy in humans.

02-04-2013

Assumpção, Cláudio de Oliveira,Lima, Leonardo Coelho Rabello,Oliveira, Felipe Bruno Dias,Greco, Camila Coelho,Denadai, Benedito Sérgio

eng

PMC6835892

nutrients Article Consumption of An Anthocyanin-Rich Antioxidant Juice Accelerates Recovery of Running Economy and Indirect Markers of Exercise-Induced Muscle Damage Following Downhill Running Leonardo C. R. Lima 1,2,3,4,*, Renan V. Barreto 1, Natália M. Bassan 1,2, Camila C. Greco 1 and Benedito S. Denadai 1 1 Human Performance Laboratory, São Paulo State University, Av. 24-A, 1515, Rio Claro, SP 13506-900, Brazil 2 Centro Universitário Hermínio Ometto, Av. Dr. Maximiliano Baruto, 500, Araras, SP 13607-339, Brazil 3 Centro Universitário Salesiano de São Paulo, R. Baronesa Geraldo de Resede, 330, Campinas, SP 13075-270, Brazil 4 Centro Universitário UniMetrocamp, R. Dr. Sales de Oliveira, 1661, Campinas, SP 13035-500, Brazil * Correspondence: leonardocrlima@gmail.com Received: 18 July 2019; Accepted: 6 August 2019; Published: 23 September 2019



 Abstract: This study examined the effects of anthocyanin-rich antioxidant juice (AJ) on the recovery of exercise-induced muscle damage (EIMD) and the running economy (RE) following downhill running (DHR). Thirty healthy young men were randomly divided into two blinded groups and consumed either AJ or placebo (PLA) for nine days (240 mL twice-a-day). On day 5, the participants from both groups ran downhill (−15%) for 30 min at 70% of their maximal oxygen uptake (VO2max) speeds. The changes in RE (oxygen uptake (VO2) and perceived effort (PE) during 5-min runs at 80%VO2max) and EIMD (isometric peak torque (IPT), muscle soreness (SOR) and serum creatine kinase activity (CK)) were compared over time and between the groups on the 4 days following DHR. VO2 and PE increased (p < 0.05) immediately following DHR for both groups and remained elevated for PLA until 48h post-DHR while fully recovering 24 h post-DHR for AJ. SOR was greater (p < 0.05) for PLA throughout the study. CK increased for both groups and was greater (p < 0.05) for PLA at 96 h post-DHR. IPT decreased for both groups but recovered faster for AJ (72 h) compared to PLA (no full recovery). AJ accelerated recovery of RE and EIMD and should be used in specific contexts, but not chronically. Keywords: running economy; antioxidant supplementation; anthocyanins; exercise-induced muscle damage; recovery; muscle soreness 1. Introduction The running economy (RE) is an important predictor of performance in endurance events. It is defined as the amount of oxygen required to sustain running at a fixed submaximal speed [1]. RE represents, therefore, how efficient athletes are during running. Athletes with similar maximal oxygen consumptions (VO2max) may present different performances in long-distance runs due to the differences in RE [2]. Several factors influence RE acutely and chronically. A relationship between RE and neuromuscular aspects exists. A growing body of literature investigated the effects of exercise-induced muscle damage (EIMD) on parameters associated with RE [3–5]. EIMD occurs when muscle tissue is damaged following strenuous exercise, leading to compromised force production capacity, muscle soreness and leakage of intracellular proteins to the circulation [6]. Nutrients 2019, 11, 2274; doi:10.3390/nu11102274 www.mdpi.com/journal/nutrients Nutrients 2019, 11, 2274 2 of 12 Downhill running (DHR) has been reported as a damaging activity due to the high volume of eccentric contractions performed during the breaking phase of running associated with oxidative stress produced in the muscle by prolonged mitochondrial activity [3,4,7]. Assumpção et al. [8] have reviewed the literature and showed that DHR compromises RE as much as countermovement jumps and heavy-load squatting exercises [8]. Anthocyanins are phenolic compounds found in dark-colored fruits that act as a pigment in nature [9]. However, evidence suggests that anthocyanin rich foods have powerful antioxidant and anti-inflammatory properties when consumed by humans [10]. In fact, there is evidence that consuming anthocyanin-rich juices leads to faster recovery of markers of EIMD following resistance training [11] and endurance events [12]. The authors have recently reviewed the literature on the effects of consuming tart cherry juice (which is rich in anthocyanins and other phenolic compounds) in recovery from EIMD [13]. However, to the best of the authors’ knowledge, there are no studies investigating if the consumption of anthocyanin-rich juice accelerates the recovery of RE following damaging bouts. The aim of the present study was to investigate the effects of anthocyanin-rich juice consumption on the magnitude of changes and time-course of recovery of markers of EIMD following a DHR bout. Our hypothesis was that the consumption of an antioxidant juice rich in anthocyanins would promote faster recovery of markers of EIMD and RE when compared to a placebo treatment. 2. Materials and Methods 2.1. Participants Thirty healthy male physical education students (age: 22.3 ± 2.6 years; height: 176.6 ± 6.4 cm; body mass 77.1 ± 10.5 kg) participated in the present study. The inclusion criteria for the present study were: Aged between 18 and 30 years-old; not having any experience with strength or aerobic training in the last six months; being a non-smoker and not having had lower-limb injuries in the last six months. They were instructed to refrain from intense physical activity, to keep their regular dietary habits and to drink plenty of water during the experimental period. All the participants read and signed an informed consent prior to their participation in the study, which was approved by the institution’s ethical board. A total of thirty participants were enrolled for the study and all of them completed the experimental protocol (15 per group). All the interactions and procedures in the present study were in accordance to the Declaration of Helsinki for research involving humans. 2.2. Experimental Design The study was conducted under double-blind, placebo-controlled conditions. The participants were allocated to either experimental (EXP) or placebo (PLA) groups in a randomized fashion. Randomization was performed by the lead examiner using a draw application for smartphone. Before being assigned to each group, the participants were familiarized to the experimental procedures. The familiarization sessions included performing maximal isometric contractions on an isokinetic dynamometer (System 3, Biodex Systems, Shirley, NY, USA). The participants also had the opportunity to run for one minute on a treadmill (Pulsar, h/p/Cosmos, Germany) with 0% inclination to familiarize with treadmill running. In the second familiarization visit, the participants’ VO2max were determined. After at least five days following the last familiarization session, the participants ran downhill (−15%) for 30 min at 70% of their VO2max speed. Chen et al. [3] showed that this protocol leads to significant damage to lower limb muscles and compromised RE. The knee extensors isometric peak torque (IPT), and markers of RE (oxygen uptake (VO2) and perceived effort (PE) during submaximal running bouts) were assessed 15 min before, 15 min after, and 1–4 days following DHR. Lower limb muscle soreness was assessed 15 min before and 1–4 days following DHR. Serum creatine kinase (CK) activity was assessed 15 min before, 2 and 4 days following DHR. Nutrients 2019, 11, 2274 3 of 12 The participants in the EXP group consumed 240 mL of an anthocyanin-rich antioxidant juice (Antiox, Juxx, Brazil) twice a day with a 12 h interval between doses at the day of the DHR bout and on the 4 days preceding and 4 days following it. Participants in the PLA group consumed a placebo consisting of water mixed with maltodextrin. The antioxidant juice and the placebo solution were isocaloric (106 Kcal per dose), isovolumetric (240 mL) and had the same amount of carbohydrates per dose (26 g). The experiment was conducted in a double-blinded fashion with all subjects and examiners blinded for the treatment being administered. The treatment bottles were opaque and the participants from both groups did not have contact with each other to avoid cross-contamination. The experimental design is illustrated in Figure 1. Nutrients 2019, 11, x FOR PEER REVIEW 3 of 13 The participants in the EXP group consumed 240 mL of an anthocyanin-rich antioxidant juice (Antiox, Juxx, Brazil) twice a day with a 12 h interval between doses at the day of the DHR bout and on the 4 days preceding and 4 days following it. Participants in the PLA group consumed a placebo consisting of water mixed with maltodextrin. The antioxidant juice and the placebo solution were isocaloric (106 Kcal per dose), isovolumetric (240 mL) and had the same amount of carbohydrates per dose (26 g). The experiment was conducted in a double-blinded fashion with all subjects and examiners blinded for the treatment being administered. The treatment bottles were opaque and the participants from both groups did not have contact with each other to avoid cross-contamination. The experimental design is illustrated in Figure 1. Figure 1. The experimental design of the study. The biggest dash represents downhill running. AJ: antioxidant juice; PLA: placebo; CK: creatine kinase. 2.3. Antioxidant Juice The participants in the EXP group consumed an anthocyanin-rich antioxidant juice that consisted of a mixture of clarified apple juice with prum, blueberry, maquiberry, raspberry and cranberry concentrates. Each dose of the juice (240 mL) contained 58 mg of anthocyanins and an antioxidant capacity of 67,680 μmoL/mL of Trolox equivalent, as assessed by the oxygen radical absorbance capacity (ORAC5) scale to identify antioxidant capacity. The evidence suggests that consuming tart cherry juice with equivalent levels of anthocyanins attenuates muscle soreness and accelerates the recovery of muscle function following damaging bouts [11,12,14]. The daily intake and timing of antioxidant juice consumption in the present study were planned based on previous studies that showed enhanced recovery of indirect markers of EIMD due to the consumption of similar anthocyanin-rich juices [11,12,15,16]. 2.4. Maximal Oxygen Uptake VO2max was determined through a treadmill incremental test. The participants ran on a treadmill (Pulsar, h/p/Cosmos, Nussdorf-Traunstein, Germany) wearing a mask attached to a breath- by-breath gas analyzer (Quark PFT Ergo, Cosmed, Pavona, Italy) that recorded oxygen uptake (VO2) and carbon dioxide (CO2) production during exercise until test cessation. The incremental test started with a three-minute warm-up at 7 km/h followed by continuous 1 km/h increments every minute. The treadmill inclination was constant and set at 1%. The criteria adopted for cessation of the test were: (1) Heart rate of 95% of the predicted maximal (220-age); (2) the respiratory exchange ratio greater than 1,15; (3) voluntary fatigue. Following filtering of the data, VO2max was considered as the greatest VO2 value recorded and sustained for at least 15 s during the test. The speed at which VO2max was reached (sVO2max) was also recorded. Figure 1. The experimental design of the study. The biggest dash represents downhill running. AJ: antioxidant juice; PLA: placebo; CK: creatine kinase. 2.3. Antioxidant Juice The participants in the EXP group consumed an anthocyanin-rich antioxidant juice that consisted of a mixture of clarified apple juice with prum, blueberry, maquiberry, raspberry and cranberry concentrates. Each dose of the juice (240 mL) contained 58 mg of anthocyanins and an antioxidant capacity of 67,680 µmoL/mL of Trolox equivalent, as assessed by the oxygen radical absorbance capacity (ORAC5) scale to identify antioxidant capacity. The evidence suggests that consuming tart cherry juice with equivalent levels of anthocyanins attenuates muscle soreness and accelerates the recovery of muscle function following damaging bouts [11,12,14]. The daily intake and timing of antioxidant juice consumption in the present study were planned based on previous studies that showed enhanced recovery of indirect markers of EIMD due to the consumption of similar anthocyanin-rich juices [11,12,15,16]. 2.4. Maximal Oxygen Uptake VO2max was determined through a treadmill incremental test. The participants ran on a treadmill (Pulsar, h/p/Cosmos, Nussdorf-Traunstein, Germany) wearing a mask attached to a breath-by-breath gas analyzer (Quark PFT Ergo, Cosmed, Pavona, Italy) that recorded oxygen uptake (VO2) and carbon dioxide (CO2) production during exercise until test cessation. The incremental test started with a three-minute warm-up at 7 km/h followed by continuous 1 km/h increments every minute. The treadmill inclination was constant and set at 1%. The criteria adopted for cessation of the test were: (1) Heart rate of 95% of the predicted maximal (220-age); (2) the respiratory exchange ratio greater than 1,15; (3) voluntary fatigue. Following filtering of the data, VO2max was considered as the greatest VO2 value recorded and sustained for at least 15 s during the test. The speed at which VO2max was reached (sVO2max) was also recorded. Nutrients 2019, 11, 2274 4 of 12 2.5. Running Economy The running economy was assessed at a fixed-speed of 5 min runs at 80% of individual sVO2max. This intensity was chosen based on the findings of Chen et al. [4] that RE at 80% and 90% sVO2max, but not at 70% sVO2max, is compromised following DHR. Therefore, this study adopted the lowest intensity at each RE which is compromised following DHR. The breath-to-breath gas exchanges were registered during RE tests and the mean VO2 was recorded during the fifth minute of each test. At the end of each RE test, the participants rated their perceived effort in a scale that varied from 6 to 20 [17]. The VO2 at the last minute of the RE tests and the perception of effort were recorded as metabolic and perceptual indices of RE, respectively. 2.6. Indirect Markers of Exercise-Induced Muscle Damage IPT, muscle soreness and serum CK activity were assessed as indirect markers of EIMD. To assess IPT, the participants performed two 5 s maximal voluntary isometric contractions in an isokinetic dynamometer with a 180 s recovery interval between contractions. A signal acquisition device with a sampling frequency of 1000 Hz (Miotool, 200/400, Miotec, Porto Alegre, Brazil) was synchronized with the dynamometer (System 3, Biodex Systems, Shirley, NY, USA) during the maximal voluntary isometric contractions for greater precision during data acquisition. The participants were seated at the dynamometer following the manufacturer’s guidelines, with their trunks, hips and dominant thighs firmly secured to the chair, their knees flexed at 70◦, and their legs firmly attached to the dynamometer shaft. They were instructed to perform knee extensions as quickly and forcefully as possible during 5 s with strong verbal encouragement being provided by the examiners. The acquired data was saved and stored for subsequent analyses. The data obtained during the maximal voluntary isometric contractions was filtered (Butterworth filter, low pass, 4th order, with a 15 Hz cut-off frequency) and analyzed in a MatLab environment (MatLab 6.5, Mathworks, Natick, MA, USA). IPT was considered as the greatest value in the torque-time curve. The contraction with the greatest IPT was used for further analyses. Muscle soreness was quantified using a 1000 mm visual analogs scale with the saying “not sore at all” and “very, very sore” at the extremities. The participants were instructed to rate their perceived soreness after climbing up and down from a 45 cm chair with their dominant limb without external assistance. They performed this test following 5 min of seated rest and could perform as many repetitions as necessary. They were instructed to mark the visual analogs scale according to the soreness they felt on their knee extensors after completing the stepping exercise. Serum CK activity was quantified by spectrophotometric analyses. Further, 500 µL of blood was extracted from the participant’s earlobes 5 min after the application of a vasodilator ointment (Finalgon, Pharma GmbH & Co. KG, Aachen, Germany) to avoid hemolysis. The blood samples were allowed to clot for 10 min and centrifuged for 10 min at 56,000 rpm (Microhemato, Modelo 2410, Fanem, São Paulo, Brazil) and the serum samples were extracted and stored at −70 ◦C for further analyses. Serum CK activity was determined using a commercial kit (CK-NAC UV, Wiener Lab, Rosário, Argentina) and a spectrophotometer (Bio-2000, Bioplus, São Paulo, Brazil). The reference values for healthy men for the method used ranged between 24 and 195 U/l. 2.7. Statistics Data normality was confirmed using the Shapiro-Wilk test. Data sphericity and homogeneity were confirmed using the Mauchly and Levene tests, respectively. The differences between groups in baseline values for all dependent variables as well as anthropometric data were tested by the student’s t-test. The changes over time and between groups were compared using the mixed model ANOVAs for repeated (time) and non-repeated (groups) measures with Bonferroni post-hoc tests. All analyses were performed in a professional software (Statistical Package for Social Sciences 17, IBM, Armonk, Nutrients 2019, 11, 2274 5 of 12 NY, USA). The significance levels were set at p <0.05. The data are expressed as the means ± standard deviation unless otherwise stated. 3. Results The participants’ mean age, body mass, height, body mass index, VO2max, sVO2max, DHR speeds and RE-test speeds are presented in Table 1. No significant differences in such variables were found. Table 1. The characteristics of the sample for the experimental (EXP) and placebo (PLA) groups. PLA (n = 15) EXP (n = 15) Age (years) 22.8 ± 2.8 21.9 ± 2.3 Body mass (kg) 79.5 ± 11.8 74.6 ± 8.7 Height (m) 1.74 ± 0.07 1.77 ± 0.06 BMI (kg/m2) 26.2 ± 3.2 23.7 ± 2.2 VO2max (mL/kg/min) 41.8 ± 5.7 43.7 ± 4.3 sVO2max (km/h) 13.9 ± 1.4 14.7 ± 1.2 Downhill Running Speed (km/h) 9.7 ± 1 10.3 ± 0.9 Running Economy Test Speed (km/h) 10.1 ± 1.2 10.5 ± 1 PLA: Placebo group; EXP: Experimental group; BMI: Body mass index; VO2max: Maximal oxygen uptake; sVO2max: Speed at which the maximal oxygen uptake was achieved. No significant differences between groups were found for baseline values of VO2 (CON: 33 ± 3.8 mL·kg−1·min−1; EXP: 35.1 ± 3.3 mL·kg−1·min−1), perceived effort (CON: 11.5 ± 1.2; EXP: 12.3 ± 1.2), IPT (CON: 290 ± 34 Nm; EXP: 278 ± 36 Nm), knee extensor muscle soreness (CON: 0 ± 0 mm; EXP: 0 ± 0 mm) and serum CK activity (CON: 106 ± 49 U.l−1; EXP: 126 ± 40 U.l−1). The significant group versus the time interactions were found for VO2 (F(5) = 20.05, p < 0.01), perceived effort (F(5) = 4.86, p < 0.01), IPT (F(5) = 3.80, p = 0.003), knee extensor muscle soreness (F(4) = 3.82, p < 0.01) and serum CK activity (F(2) = 3.70, p = 0.31). The pairwise comparisons showed that VO2 significantly increased for both groups immediately following DHR and fully recovered 24 h and 72 h post-exercise for the experimental and control groups, respectively. VO2 was significantly greater for the control group than the experimental group 24 h post-DHR. This was also the case for perceived effort. The absolute changes in VO2 and perceived effort over time following DHR are presented in Figure 2. Nutrients 2019, 11, x FOR PEER REVIEW 6 of 13 Figure 2. Cont. Nutrients 2019, 11, 2274 6 of 12 Figure 2. The absolute changes in oxygen uptake (A) and perceived effort (B) over time following downhill running. * p < 0.05 compared to the baseline values in the same group. ‡ p < 0.05 compared to the experimental group at the same time-point. CON: control group; EXP: experimental group. The isometric peak torque significantly decreased (p < 0.05) for both groups immediately after DHR and remained so throughout the entire experimental period for the control group, but fully recovered 72 h post-DHR for the experimental group (Figure 3). Figure 2. The absolute changes in oxygen uptake (A) and perceived effort (B) over time following downhill running. * p < 0.05 compared to the baseline values in the same group. ‡ p < 0.05 compared to the experimental group at the same time-point. CON: control group; EXP: experimental group. The isometric peak torque significantly decreased (p < 0.05) for both groups immediately after DHR and remained so throughout the entire experimental period for the control group, but fully recovered 72 h post-DHR for the experimental group (Figure 3). Nutrients 2019, 11, x FOR PEER REVIEW 7 of 13 Figure 3. The relative changes in isometric peak torque over time following downhill running. * p <0.05 compared to the baseline values in the same group. CON: control group; EXP: experimental group. Knee extensor muscle soreness significantly increased (p < 0.05) 24 h following DHR and remained so during the whole experiment for both groups. However, knee extensor muscle soreness was significantly (p < 0.05) greater for the control group at all time-points. Serum CK activity significantly increased 48 h following DHR and remained elevated until 96 h following DHR for both groups. Serum CK activity was significantly (p < 0.05) greater for the control group at 96 h post-DHR. The changes in knee extensor muscle soreness and serum CK activity are presented in Figure 4. Figure 3. The relative changes in isometric peak torque over time following downhill running. * p <0.05 compared to the baseline values in the same group. CON: control group; EXP: experimental group. Knee extensor muscle soreness significantly increased (p < 0.05) 24 h following DHR and remained so during the whole experiment for both groups. However, knee extensor muscle soreness was significantly (p < 0.05) greater for the control group at all time-points. Serum CK activity significantly increased 48 h following DHR and remained elevated until 96 h following DHR for both groups. Serum CK activity was significantly (p < 0.05) greater for the control group at 96 h post-DHR. The changes in knee extensor muscle soreness and serum CK activity are presented in Figure 4. Nutrients 2019, 11, 2274 7 of 12 Nutrients 2019, 11, x FOR PEER REVIEW 8 of 13 Figure 4. The changes in knee extensor muscle soreness (A) and serum creatine kinase (CK) activity (B) over time following downhill running. * p <0.05 compared to baseline values in the same group; ‡ p <0.05 compared to the experimental group at the same time-point. CON: control group; EXP: experimental group. 4. Discussion The aim of the present study was to investigate the acute impact of antioxidant juice consumption on changes in RE and the recovery of indirect markers of EIMD following DHR. It was hypothesized that phenolic compounds present in the antioxidant juice—especially anthocyanins— would accelerate recovery of muscle soreness, serum CK activity, muscle function and RE following DHR, and this was confirmed by the obtained data. The data from the control group showed that DHR significantly compromises RE at 80% sVO2max—measured as VO2 and the perceived effort—with full recovery reached 3 days following the exercise bout. DHR also led to significant changes in the indirect markers of EIMD (IPT, muscle soreness and serum CK activity) with full recovery not being reached within four days following the exercise bout for the control group. These findings corroborate what has been previously reported in the literature and confirm that recovery kinetics are different between RE and the indirect markers of EIMD [8]. Figure 4. The changes in knee extensor muscle soreness (A) and serum creatine kinase (CK) activity (B) over time following downhill running. * p <0.05 compared to baseline values in the same group; ‡ p <0.05 compared to the experimental group at the same time-point. CON: control group; EXP: experimental group. 4. Discussion The aim of the present study was to investigate the acute impact of antioxidant juice consumption on changes in RE and the recovery of indirect markers of EIMD following DHR. It was hypothesized that phenolic compounds present in the antioxidant juice—especially anthocyanins—would accelerate recovery of muscle soreness, serum CK activity, muscle function and RE following DHR, and this was confirmed by the obtained data. The data from the control group showed that DHR significantly compromises RE at 80% sVO2max—measured as VO2 and the perceived effort—with full recovery reached 3 days following the exercise bout. DHR also led to significant changes in the indirect markers of EIMD (IPT, muscle soreness and serum CK activity) with full recovery not being reached within four days following the exercise bout for the control group. These findings corroborate what has been previously reported in the literature and confirm that recovery kinetics are different between RE and the indirect markers of EIMD [8]. Nutrients 2019, 11, 2274 8 of 12 The present study found that antioxidant juice consumption results in faster recovery of muscle function as well as attenuated muscle soreness and serum CK activity following an exercise bout consisting of 30 min of DHR. Our findings are similar to those which showed that consuming tart cherry (Prunus cerasus L.) juice accelerates recovery of the indirect markers of EIMD following the different types of exercise bouts (i.e., resistance exercise training, maximal isokinetic eccentric contractions, DHR, marathon running and stochastic cycling) [11,12,16,18,19]. However, little is known about the mechanisms underlying downhill running-induced muscle damage. It is generally accepted that EIMD is characterized by two distinct events. In the first event, the mechanical strain imposed by unaccustomed exercise damages the sarcolemma and ultrastructural sarcomere proteins [20]. This results in compromised muscle function due to the disrupted contractile and structural proteins as well as the compromised excitation-contraction coupling [6]. The mechanical damaging event is usually aggravated when eccentric contractions are performed during unaccustomed exercise bouts due to the unique motor unit recruitment patterns during such contractions [21]. The mechanical damage is followed by cellular signaling for repair, which triggers an inflammatory response consisting of the migration of neutrophils and monocytes (which differentiate into macrophages once in the damaged site) [22]. The immune cells promote the degradation of cellular debris through phagocytosis by producing oxygen reactive species. However, this degradation is not exclusive to cellular debris, but also affects healthy, functioning, structures of adjacent myocytes. This is referred to as the second event of EIMD and leads to muscle soreness, increased CK release to the blood stream and, possibly, further loss of muscle function [13]. It is yet to be determined if the oxidative stress produced by the mitochondrial respiratory chain during DHR anticipates and/or aggravates the second event of EIMD. It is, however, well established that DHR significantly affects the indirect markers of EIMD [8,23]. The data obtained in the present study suggests that muscle function was significantly compromised immediately following DHR for both groups. This was expected, since the consumption of antioxidant juice is not expected to strengthen the sarcolemma nor impact the motor unit recruitment patterns, attenuating the first EIMD event. However, accelerated recovery kinetics were observed for both RE (VO2 and perceived effort) and muscle function (IPT) (Figures 2 and 3). The perceived effort and VO2 fully recovered two days earlier for the experimental group with significant differences between the groups observed 1 day following DHR. Similarly, IPT reached full recovery during the study in the experimental group while it remained compromised throughout the entire study for the control group. No significant differences were found between the groups for the IPT values. Although previous studies have reported attenuated changes in IPT following damaging bouts when associated with consumption of tart cherry juice [11], accelerated recovery kinetics are also important when investigating strategies to attenuate EIMD [16,24]. The differences between changes in RE and muscle function observed among the groups in the present study might be explained by the antioxidant properties of anthocyanins in the antioxidant juice. Previous studies showed that anthocynin-rich tart cherry juice reduced total oxidative stress and circulating levels of C-reactive protein other oxygen reactive species [12,16,25]. It has been reported that anthocyanins (as well as other phenolic compounds) scavenge free radicals secreted by lymphocytes and produced in the mitochondrial respiratory chain [26]. It has also been reported that consuming foods as rich in anthocyanins as the antioxidant juice used in the present study decreases circulating levels of pro-inflammatory cytokines such as interleukin-6 and tumor necrosis factor-α following damaging bouts [12,19], potentially attenuating the second event of EIMD. This might also explain the attenuated muscle soreness for the experimental group observed at all assessment points in our study. The delayed-onset muscle soreness is frequently described as a symptom of the inflammation that occurs in the muscle and fascia following eccentric-biased activities due to the interaction of algesic pro-inflammatory substances such as histamines, bradykinins and prostaglandins with nociceptors [27]. Hence, if a treatment attenuates inflammation, it also attenuates the ensuing soreness caused by it. Nutrients 2019, 11, 2274 9 of 12 Serum CK activity increased for both groups following downhill running, peaking 4 days after it. However, serum CK activity was greater for the control group at its peak. As an intracellular enzyme, the increased CK activity in the bloodstream is a sign of membrane and tissue damage [28]. Peak CK activity in the bloodstream occurs later than other indirect markers of EIMD since it must be transported from the lymph to the circulation [28]. Significantly greater serum CK activity for the control group suggests either greater mechanical stress (which was not the case, since both groups exercised at identical volumes and intensities) or that membrane damage caused by lipolytic enzymes such as phospholipase A2—which is activated by pro-inflammatory cytokines [29]—was greater in the absence of the treatment investigated in the present study. It can, therefore, be assumed that the anti-inflammatory properties of the treatment investigated in the present study attenuated secondary damage to healthy myocytes induced by inflammation. Although oxidative-stress or inflammatory markers were not assessed in the present study, it is important to notice that the concentration of anthocyanins in the antioxidant juiced consumed by the participants in the experimental group is similar to those of cherry juices used in studies that found attenuated inflammatory responses and total oxidative status. Associated with the observation of similar effects on the indirect markers of EIMD between our treatment and those previously reported, this adds to the presented rationale. To the best of the authors’ knowledge, no previous study investigated the impact of consuming antioxidant/anti-inflammatory treatments in the magnitude of changes and recovery kinetics of RE following damaging bouts. This study found that not only does consuming antioxidant juice accelerate recovery of RE markers, but it also attenuates changes 1 day following DHR. In fact, our results suggest that RE is only compromised immediately following DHR when consuming antioxidant juice. This has important implications regarding training protocols and, especially, competitive schedules. In specific contexts, athletes are submitted to short-term competitions at which they are expected to perform in subsequent days. In such contexts, it is important for endurance athletes to maintain their efficiency, and EIMD from previous days might be an issue. Training camps are also an example of condensed endurance events during which it is of the best interest of athletes to perform as well as possible. Our findings indicate that consuming an antioxidant juice rich in anthocyanins might be a good strategy to maintain efficiency and attenuate muscle soreness in such contexts. Caution is warranted when transferring the finding of the present study to trained athletes. The participants in our study presented VO2max values between 40−45 mL·kg−1·min−1, which are not compatible with trained athletes. The evidence suggests that antioxidant status is greater for athletes compared to sedentary controls [30]. Hence, there is a possibility that additional antioxidant properties of anthocyanin-rich foods do not further improve the already-high antioxidant response to damaging exercises in trained athletes. However, Howatson et al. [12] showed that the consumption of Prunus cersasus L. accelerates recovery of muscle function and soreness following marathon running in experienced runners. Further studies are warranted to investigate if anthocyanin-rich foods accelerate recovery of RE in elite athletes. The continuous use of antioxidant juices during endurance training programs should not be encouraged. The evidence suggests that, despite the beneficial effects of antioxidant supplementation in the recovery from EIMD and RE, oxidative stress might be an important component for training adaptation [31]. Merry and Ristow [32] reviewed the literature on this topic and concluded that a balance in redox signaling is the key to optimal endurance adaptation and long-term antioxidant supplementation can blunt the physiological stress imposed by exercise, consequently compromising optimal training adaptation. This should be taken into consideration when planning nutritional strategies for training and competitions. The assessment of oxidative status and markers of inflammation is important, and the absence of these variables is a limitation of our study. While this study did find a significant impact of the treatment in our main outcome, the direct markers of secondary damage to better elucidate the mechanisms of faster RE recovery associated with antioxidant juice consumption were not assessed. The authors encourage further investigation of such mechanisms. This study also did not carry out dietary analyses. Nutrients 2019, 11, 2274 10 of 12 This might have implications regarding an already high antioxidant status due to the high anthocyanin intake by the participants prior to the experiment. Since this study only recommended the participants to maintain their regular dietary habits but did not control nor assess them during the experimental period, this might also be considered a limitation of the present study. Another limitation of the present study is the intensity at which RE was assessed. Future studies should assess the impacts of antioxidant juice consumption on the changes in RE at different intensities following DHR and other damaging bouts. Finally, there seems to be a lack of standardization for the composition of antioxidant products investigated in the literature. this study focused on anthocyanin concentration and antioxidant capacity when choosing our treatment, as well as the commercial availability in our country. Therefore, the findings presented in the present study represent the impact of this specific antioxidant juice on changes in RE and the indirect markers of EIMD, which may not be the same when using antioxidant juices with other compositions. It is recommended that practitioners and colleagues focus on anthocyanin concentration when choosing which juice to prescribe/investigate. 5. Conclusions In conclusion, this study shows that consuming an anthocyanin-rich antioxidant juice four days prior to, at the day and four days following DHR resulted in the accelerated recovery of RE and muscle function as well as attenuated muscle soreness. These data suggest that this nutritional strategy might be useful to maintain satisfactory performance in condensed competitions and training camps. Caution is warranted when planning long-term antioxidant supplementation, as training adaptations might be blunted. Future studies are needed to clarify the mechanisms underlying faster recovery of RE when consuming antioxidant juice. Author Contributions: Conceptualization, L.C.R.L. and B.S.D.; data curation, L.C.R.L., R.V.B. and N.M.B.; formal analysis, C.C.G. and B.S.D.; funding acquisition, L.C.R.L. and B.S.D.; investigation, L.C.R.L., R.V.B. and N.M.B.; methodology, L.C.R.L. and B.S.D.; project administration, L.C.R.L. and B.S.D.; resources, L.C.R.L.; Supervision, C.C.G. and B.S.D.; writing–original draft, L.C.R.L. and R.V.B.; writing–review & editing, N.M.B, C.C.G. and B.S.D. Funding: This work was funded by the São Paulo Research Foundation (FAPESP) grant number 2013/23585-4 and the APC was funded by the São Paulo Research Foundation (FAPESP) grant number 2019/17202-1. Acknowledgments: The authors would like to thank Professor Glyn Howatson for his contributions to this work and all participants for their effort. Conflicts of Interest: The authors declare no conflicts of interest. References 1. Fletcher, J.R.; MacIntosh, B.R. Running economy from a muscle energetics perspective. Front. Phyisol. 2017, 8, 1–15. [CrossRef] [PubMed] 2. Hoogkamer, W.; Kipp, S.; Spiering, B.A.; Kram, R. Altered running economy directly translates to altered distance-running performance. Med. Sci. Sports Exerc. 2016, 48, 2175–2180. [CrossRef] [PubMed] 3. Chen, T.C.; Nosaka, K.; Tu, J.H. Changes in running economy following downhill running. J. Sports Sci. 2007, 25, 55–63. [CrossRef] [PubMed] 4. Chen, T.C.; Nosaka, K.; Lin, M.J.; Chen, H.L.; Wu, C.J. Changes in running economy at different intensities following downhill running. J. Sports Sci. 2009, 27, 1137–1144. [CrossRef] [PubMed] 5. Braun, W.; Paulson, S. The effects of a downhill running bout on running economy. Res. Sports. Med. 2012, 20, 274–285. [CrossRef] [PubMed] 6. Hyldahl, R.D.; Chen, T.C.; Nosaka, K. Mechanisms and mediators of the skeletal muscle repeated bout effect. Exerc. Sport Sci. Rev. 2017, 45, 24–33. [CrossRef] [PubMed] 7. Braun, W.A.; Dutto, D.J. The effects of a single bout of downhill running and ensuing delayed onset of muscle soreness on running economy performed 48 h later. Eur. J. Appl. Physiol. 2003, 90, 29–34. [CrossRef] [PubMed] 8. Assumpção, C.O.; Lima, L.C.; Olvieira, F.B.; Greco, C.C.; Denadai, B.S. Exercise-induced muscle damage and running economy in humans. Sci. World J. 2013, 2013, 1–11. [CrossRef] Nutrients 2019, 11, 2274 11 of 12 9. Damar, I.; Eksi, A. Antioxidant capacity and anthocyanin profile of sour cherry (Prunus cersus L.) juice. Food Chem. 2012, 135, 2910–2914. [CrossRef] 10. Kuehl, K.S. Cherry juice targets antioxidant potential and pain relief. Med. Sport. Sci. 2012, 59, 86–93. 11. Connolly, D.A.; McHugh, M.P.; Padilla-Zakour, O.I.; Carlson, L.; Sayers, S.P. Efficacy of a tart cherry juice blend in preventing the symptoms of muscle damage. Br. J. Sports Med. 2006, 40, 679–683. [CrossRef] [PubMed] 12. Howatson, G.; McHugh, M.P.; Hill, J.A.; Brouner, J.; Jewell, A.P.; van Someren, K.A.; Shave, R.E.; Howatson, S.A. Influence of tart cherry juice on indices of recovery following marathon running. Scand. J. Med. Sci. Sports 2010, 20, 843–852. [CrossRef] [PubMed] 13. Lima, L.C.R.; Assumpção, C.O.; Prestes, J.; Denadai, B.S. Consumption of cherries as a strategy to attenuate exercise-induced muscle damage and inflammation in humans. Nutr. Hosp. 2015, 32, 1885–1893. 14. McLeay, Y.; Barnes, M.J.; Mundel, T.; Hurst, S.M.; Hurst, R.D.; Stannard, S.R. Effect of New Zealand blueberry consumption on recovery from eccentric exercise-induced muscle damage. J. Int. Soc. Sports Nutr. 2012, 9, 2–12. [CrossRef] [PubMed] 15. Traustadóttir, T.; Davies, S.S.; Stock, A.A.; Su, Y.; Heward, C.B.; Roberts, L.J.; Harman, S.M. Tart cherry juice decreases oxidative stress in healthy older men and women. J. Nutr. 2009, 139, 1896–1900. [CrossRef] [PubMed] 16. Bowtell, J.L.; Summers, D.P.; Dyer, A.; Fox, P.; Mileva, K.N. Montmorrency cherry juice reduced muscle damage caused by intensive strength exercise. Med. Sci. Sports Exerc. 2011, 43, 1544–1551. [CrossRef] [PubMed] 17. Borg, G.A.V. Psychophysical bases of perceived exertion. Med. Sci. Sports Exerc. 1982, 14, 377–381. [CrossRef] [PubMed] 18. Kuehl, K.S.; Perrier, E.T.; Elliot, D.L.; Chesnutt, J.C. Efficacy of tart cherry juice in reducing muscle pain during running: A randomized controlled trial. J. Int. Soc. Sports Nut. 2010, 7, 1–6. [CrossRef] 19. Bell, P.G.; Walshe, I.H.; Davison, G.W.; Stevenson, E.J.; Howatson, G. Recovery facilitation with Montmorrency cherries following high-intensity, metabolically challenging exercise. Appl. Physiol. Nutr. Metab. 2015, 40, 414–423. [CrossRef] 20. Morgan, D.L.; Proske, U. Popping sarcomere hypothesis explains stretch-induced muscle damage. Clin. Exp. Pharmacol. Physiol. 2004, 31, 541–545. [CrossRef] 21. Enoka, R. Eccentric contractions require unique activation strategies by the nervous system. J. Appl. Physiol. 1996, 81, 2339–2346. [CrossRef] 22. Paulsen, G.; Mikkelsen, U.R.; Raastadt, T.; Peake, J.M. Leucocytes, cytokines and satellite cells: What role do they play in muscle damage and regeneration following eccentric exercise? Exerc. Immunol. Rev. 2012, 18, 42–97. [PubMed] 23. Vernillo, G.; Giandolini, M.; Edwards, W.B.; Morin, J.B.; Samozino, P.; Horvais, N.; Millet, G.Y. Biomechanics and physiology of uphill and downhill running. Sports Med. 2017, 47, 615–629. [CrossRef] [PubMed] 24. Lima, L.C.R.; Bassan, N.M.; Cardozo, A.C.; Gonçalves, M.; Greco, C.C.; Denadai, B.S. Isometric pre-conditioning blunts exercise-induced muscle damage but does not attenuate changes in running economy following downhill running. Hum. Mov. Sci. 2018, 60, 1–9. [CrossRef] [PubMed] 25. Kelley, D.S.; Rasooly, R.; Jacob, R.A.; Kader, A.A.; Mackey, B.E. Consumption of Bing sweet cherries lowers circulating concentrations of inflammation markers in healthy men and women. J. Nutr. 2006, 136, 981–986. [CrossRef] [PubMed] 26. Mastaloudis, A.; Morrow, J.D.; Hopkins, D.W.; Devaraj, S.; Traber, M.G. Antioxidant supplementation prevents exercise-induced lipid peroxidation, but not inflammation, in ultramarathon runners. Free. Radic. Biol. Med. 2004, 36, 1329–1341. [CrossRef] 27. Hotfiel, T.; Freiwald, J.; Hoppe, M.W.; Lutter, C.; Forst, R.; Grim, C.; Bloch, W.; Hüttel, M.; Heiss, R. Advances in delayed-onset muscle soreness (DOMS): Part I: Pathogenesis and diagnostics. Sportverletz Sportschaden 2018, 32, 243–250. [CrossRef] 28. Brancaccio, P.; Maffulli, N.; Limongelli, F.M. Creatine kinase monitoring in sport medicine. Br. Med. Bull. 2007, 81, 209–230. [CrossRef] 29. Verbourg, E.; Murphy, R.M.; Stephenson, D.G.; Lamb, G.D. Disruption of excitation-contraction coupling and titin by endogenous Ca2+–activated proteases in toad muscle fibres. J. Physiol. 2005, 564, 775–790. [CrossRef] Nutrients 2019, 11, 2274 12 of 12 30. Dékány, M.; Nemeskéri, V.; Györe, I.; Harbula, I.; Malomsoki, J.; Puscok, J. Antioxidant status of interval-trained athletes in various sports. Int. J. Sports Med. 2006, 27, 112–116. [CrossRef] 31. Slattery, K.; Bentley, D.; Coutts, A.J. The role of oxidative, inflammatory and neuroendocrinological systems during exercise stress in athletes: Implications of antioxidant supplementation on physiological adaptation during intensified physical training. Sports Med. 2015, 45, 453–471. [CrossRef] [PubMed] 32. Merry, T.L.; Ristow, M. Do antioxidant supplements interfere with skeletal muscle adaptation to exercise training? J. Physiol. 2016, 594, 5135–5146. [CrossRef] [PubMed] © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Consumption of An Anthocyanin-Rich Antioxidant Juice Accelerates Recovery of Running Economy and Indirect Markers of Exercise-Induced Muscle Damage Following Downhill Running.

09-23-2019

Lima, Leonardo C R,Barreto, Renan V,Bassan, Natália M,Greco, Camila C,Denadai, Benedito S

eng

PMC10405799

Resultant equations for training load monitoring during a standard microcycle in sub-elite youth football: a principal components approach José Eduardo Teixeira1,2,3, Pedro Forte1,2,4, Ricardo Ferraz1,5, Luís Branquinho1,4, Ryland Morgans6, António José Silva1,7, António Miguel Monteiro1,2 and Tiago M. Barbosa1,2 1 Research Centre in Sports, Health and Human Development, Covilhã, Portugal 2 Department of Sport Sciences, Instituto Politécnico de Bragança, Bragança, Portugal 3 Department of Sport Sciences, Polytechnic Institute of Guarda, Guarda, Portugal 4 CI-ISCE Douro, Higher Institute of Educational Sciences of the Douro, Penafiel, Portugal 5 Department of Sport Sciences, University of Beira Interior, Covilhã, Portugal 6 Institute for Coaching and Performance, University of Central Lancashire, Preston, United Kingdom 7 Sport Sciences, University of Trás-os-Montes and Alto Douro, Vila Real, Portugal ABSTRACT Applying data-reduction techniques to extract meaningful information from electronic performance and tracking systems (EPTS) has become a hot topic in football training load (TL) monitoring. The aim of this study was to reduce the dimensionality of the internal and external load measures, by a principal component approach, to describe and explain the resultant equations for TL monitoring during a standard in-season microcycle in sub-elite youth football. Additionally, it is intended to identify the most representative measure for each principal component. A principal component analysis (PCA) was conducted with a Monte Carlo parallel analysis and VariMax rotation to extract baseline characteristics, external TL, heart rate (HR)-based measures and perceived exertion. Training data were collected from sixty sub-elite young football players during a 6-week training period using 18 Hz global positioning system (GPS) with inertial sensors, 1 Hz short-range telemetry system, total quality recovery (TQR) and rating of perceived exertion (RPE). Five principal components accounted for 68.7% of the total variance explained in the training data. Resultant equations from PCA was subdivided into: (1) explosiveness, accelerations and impacts (27.4%); (2) high-speed running (16.2%); (3) HR-based measures (10.0%); (4) baseline characteristics (8.3%); and (5) average running velocity (6.7%). Considering the highest factor in each principal component, decelerations (PCA 1), sprint distance (PCA 2), average HR (PCA 3), chronological age (PCA 4) and maximal speed (PCA 5) are the conditional dimension to be considered in TL monitoring during a standard microcycle in sub-elite youth football players. Current research provides the first composite equations to extract the most representative components during a standard in-season microcycle in sub-elite youth football players. Futures research should expand the resultant equations within training days, by considering other well-being measures, technical-tactical skills and match-related contextual factors. How to cite this article Teixeira JE, Forte P, Ferraz R, Branquinho L, Morgans R, Silva AJ, Monteiro AM, Barbosa TM. 2023. Resultant equations for training load monitoring during a standard microcycle in sub-elite youth football: a principal components approach. PeerJ 11:e15806 DOI 10.7717/peerj.15806 Submitted 26 January 2023 Accepted 7 July 2023 Published 4 August 2023 Corresponding author Tiago M. Barbosa, barbosa@ipb.pt Academic editor Silvia Comani Additional Information and Declarations can be found on page 15 DOI 10.7717/peerj.15806 Copyright 2023 Teixeira et al. Distributed under Creative Commons CC-BY 4.0 Subjects Kinesiology, Sports Injury, Sports Medicine Keywords Youth, Workload, Soccer, Global positioning system, PCA INTRODUCTION Training load (TL) monitoring has become a research hot topic in youth football (Impellizzeri et al., 2022; Staunton et al., 2021). This is largely due to the growing access to electronic performance and tracking systems (EPTS) that provides valid TL measures (de Dios-Álvarez et al., 2021; Oliva-Lozano & Muyor, 2022). In recent years, the weekly TL variation has been extensively analyzed in elite and sub-elite football contexts (Teixeira et al., 2022a). Training monitoring has been extensively performed using objective and subjective methods to monitor internal training load (ITL) and external training load (ETL) (Impellizzeri et al., 2022). Global positioning system (GPS) devices have become a customary, low-cost and optimal navigation satellite system to extract valid and reliable ETL outcomes (e.g., distances, sprints, accelerations (ACC), change of directions or body impacts) (Beato et al., 2018; Buchheit et al., 2021). Otherwise, the ITL has been usually monitored by heart rate (HR) and perceived exertion using non-invasive wearable sensor systems, rating perceived exertion (RPE) and total quality recovery (TQR) scales (Haddad et al., 2017; Brink et al., 2010). The research has shown a significant correlation between ETL and ITL in young athletes, however it is still difficult to interpret fitness-recovery status (Impellizzeri et al., 2022). Combining ETL and ITL has been reported as a valid strategy to analyse dose-response dissonances, however the major influencing factor remain to be defined (Bourdon et al., 2017; Teixeira et al., 2021a). Additionally, the emergent tracking tools appears to have created confusion in dose-response considerations given the data analysis requirement to extract relevant information from large amounts of data (Griffin et al., 2021; Scantlebury et al., 2020). This kind of tracking device can provide big datasets express as a thousand data per second expressed by a large number of variables depending on the time-motion technology used (Rojas-Valverde et al., 2020; Ruan et al., 2022). Otherwise, another challenge has been to standardize the physical and psychophysiological data in meaningful information (Impellizzeri et al., 2022; Staunton et al., 2021; Vanrenterghem et al., 2017). As well, capturing the training frequency, intensity, time/duration, type, volume, and progression (FITT-VP) variables is another critical challenge created by tracking systems (Staunton et al., 2021; Scantlebury et al., 2020). Thus, it is more critical than ever to turning datasets into relevant information for athlete-monitoring cycle (Teixeira et al., 2021a; Weaving et al., 2019). Afterwards, the data-reduction techniques has been applied to explain the dimensionality of the TL variables in different football codes such as futsal (Rico-González et al., 2022a), Australian football (Sheehan et al., 2020), rugby (Scantlebury et al., 2020; Weaving et al., 2020) and Gaelic football (Gamble et al., 2019). Principal component analysis (PCA) is one of the most used data-reduction techniques to extract redundant information from TL data in football (Rico-González et al., 2022b; Rojas-Valverde et al., 2020). Using a PCA approach, a significant percentage of the total variance in a dataset can be extracted (Warmenhoven et al., 2019). Thus, PCA analysis allows to reduce the complexity in a large group of correlated variables by determining the Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 2/21 principal components (O’Donoghue, 2008; Rojas-Valverde et al., 2020). Recently, a systematic review conducted in football reported a 77.1% of explained variance in 12.8 extracted variables out of 51.4 variables distributed over 6.4 principal components (Rojas- Valverde et al., 2020). However, the studies with PCA approaches has focused mainly on TL monitoring in professional and elite youth football (Casamichana et al., 2019; Scantlebury et al., 2020; Sheehan et al., 2020). Until now, PCA approaches were only applied in elite football contexts to simplify the TL having regard to different game formats (Casamichana et al., 2019; Zurutuza et al., 2020), contextual factors (Gonçalves et al., 2019; Oliva-Lozano et al., 2021), competition level (Ricotti et al., 2013), positional role (Moura et al., 2015), tactical behaviour (Ric et al., 2016; Rico-González et al., 2022b) and motor skills (Los Arcos, Mendiguchia & Javier, 2017). Recently, some studies have described the application of TL monitoring strategies during a weekly microcycle in sub-elite youth football, expressing by a low seasonal variation and a high weekly variation (Teixeira et al., 2021b, 2022b). Therefore, it is important to establish the major influencing factor for an accurate training monitoring and manipulation during a standard microcyle. Also, an optical TL monitoring can enhance a proper long-term athlete development, injury prevention and training design (Pino-Ortega et al., 2021; Rico-González et al., 2022c; Rojas- Valverde et al., 2020). More specifically, this can help research, practitioners and coaches to prescribe adequate training intensity over a standard microcycle in youth football (Rico- González et al., 2022a). Therefore it is critical to standardize and reduce the dimensionality of the weekly training data in young football players from sub-elite contexts (Teixeira et al., 2022c; Trecroci et al., 2018). Thus, the aim of this study was to reduce the dimensionality of the internal and external load measures, by a PCA approach, in order to describe and explain the resultant equations for TL monitoring during a standard microcycle in a sub-elite youth football players. Additionally, it is intended to identify the most representative measure for each principal component. METHODS Participants Sixty sub-elite youth and male football players were included this study from an under (U) 15 (n = 20), U17 (n = 20) and U19 (n = 20) sub-elite youth football academy (Table 1). All parents or legal guardians were written briefed about research aims and risks, providing a written consent for participant’s inclusion. The research was developed in accordance with the Declaration of Helsinki (Winter & Maughan, 2009) with an ethical approval from the local Ethical Committee from the University of Trás-os-Montes e Alto Douro (3379- 5002PA67807). Quasi-experimental approach Current research has a prospective, observational and cross-sectional design, by applying an individual TL strategy via GPS technology, HR monitoring system, RPE and TQR scales. Resultant equations for TL monitoring in sub-elite youth football was obtained by a PCA approach. The weekly TL was continuously monitored during 2019–2020 in-season, representing a total of 6-week period from 18 training sessions and 324 observation cases Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 3/21 (Teixeira et al., 2021b, 2022d). A minimum of 150 observation cases (i.e., 5 to 10 cases per variable) was assured to perform PCA analysis (Jolliffe & Cadima, 2016). Figure 1 summarizes the procedures for quasi-experimental approach. Procedures The training data eligibility considered the following inclusion criteria: (a) youth football players aged between 13 and 20 years old (i.e., U15, U17 and U19) (Teixeira et al., 2021a); (b) young football players should have at least 5 years of competitive experience in football (Ford et al., 2020); (c) training data featured at least 35 consecutive playing minutes without any break for injury, abandonment or other arbitrary reason (de Dios-Álvarez et al., 2021); (d) training data considered a competitive one-game week schedule and three training sessions per week (Teixeira et al., 2021b, 2022a). The exclusion of cases occurred when the following exclusion criteria were met: (a) event of absence, injury, illness and abandonment during monitored training sessions; (b) players that were not integrated in the common team session due to rehabilitation, complementary and/or individual training sessions; (c) the match data was not included in the analysis (Teixeira et al., 2022d). For ETL and ITL monitoring, each participant wore the micro-technology (i.e., GPS and HR) within a little pocket on the upper back between both scapulae of a custom-made vest Table 1 Description baseline characteristics of participants. Variables U15 (n = 20) U17 (n = 20) U19 (n = 20) Overall (n = 60) Age (y) 13.28 ± 0.49 15.39 ± 0.51 17.29 ± 0.55 15.19 ± 1.75 RA (a.u.) 0.25 ± 0.17 0.25 ± 0.17 0.24 ± 0.20 0.25 ± 0.18 MO (a.u.) −0.42 ± 0.76 2.02 ± 1.09 2.23 ± 1.49 1.33 ± 1.67 Height (m) 1.69 ± 0.78 1.76 ± 0.48 1.76 ± 0.70 1.74 ± 0.08 Weight (kg) 55.67 ± 9.41 64.28 ± 6.61 68.90 ± 8.39 62.48 ± 10.03 BMI (kg/m2) 19.29 ± 1.99 20.68 ± 1.79 22.11 ± 1.50 20.61 ± 2.14 Sitting height (cm) 81.96 ± 5.78 92.02 ± 7.61 90.73 ± 8.06 88.36 ± 8.51 PHV (cm) 14.18 ± 0.80 13.90 ± 1.09 14.46 ± 1.87 14.20 ± 1.39 Experience (y) 4.82 ± 0.90 6.64 ± 1.65 8.81 ± 1.70 6.76 ± 1.42 Note: Abbreviations: a.u., arbitrary unit; BMI, body mass index; MO, maturity offset; PHV, peak high velocity; RA, relative age; y, years. Figure 1 Training load monitoring using a prospective, observational and cross-sectional design. Full-size  DOI: 10.7717/peerj.15806/fig-1 Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 4/21 (Beato et al., 2018). All methodological procedures for ETL and ITL were previously applied for 2 weeks to familiarize players with data collection (de Dios-Álvarez et al., 2021). Using a “match day minus format” (MD), the weekly microcycle included the training sessions MD-3 (Tuesday), MD-2 (Wednesday), and MD-1 (Friday). The number of observation for each training day was: MD-3 (n = 41), MD-2 (n = 38), and MD-1 (n = 44) (Teixeira et al., 2022d, 2021b). The training days for the three age groups were the same following this order: U15—6 to 7:30 PM; U17—7:30 to 9:00 PM; U19—9:00 PM to 10:30 PM. The average duration of training sessions had the following lengths for each age group: U15 = 148.99 min; U17 = 132.46 min; U19 = 195.95 min. Medical and logistical staff members ensured that all training classes had standardized clothes, nutrition and medical care during training sessions (Teixeira et al., 2022d). All training sessions were performed on a synthetic turf outdoor pitch with official dimensions (FIFA standard; 100 m × 70 m) and similar environment conditions (i.e., 14–20
C; relative humidity 52–66%) (Coutinho et al., 2015). Weekly standard microcycle Table 2 showed the weekly training overview in the studied sub-elite youth football academy. The standard microcycle was planned in accordance with the following key points: (i) training aims, time duration and pitch dimensions; (ii) physiological target and speed, agility and quickness (SAQ) emphasis; (iv) training tasks and exercises. Weekly training overview was designed according to field notes and academy training model. Also, current typical microcycle was designed during an in-season standard microcycle with aforementioned training days (i.e., MD-3, MD-2 and MD-1) (Branquinho, Ferraz & Marques, 2021; Rago et al., 2020). Small, medium, large-sided, and simulated games (i.e., Table 2 Weekly standard microcycle in the sampled sub-elite youth football academy. Construct MD-3 (Tuesday) MD-2 (Wednesday) MD-1 (Friday) Aim (tactical) Recovery/technical skills Acquisitive training focused on game principles (collective behaviour and organization) Finishing situations and tactical schemes Duration 90 min 90 min 90 min Dimensions 50 m × 60 m (half field) 100 m × 60 m (entire field) 50 m × 60 m (half field) Physiological set 75–80% HRmax 90–95% HRmax >85% MRS SAQ Strength (Quickness, COD and agility) Endurance/Aerobic Speed Warm up Technical and coordination skills Dynamic stretching Plyometric exercises and SSC Training tasks (1) SSG, MSG, and ball possession (small areas); (1) Ball possession, LSG and simulated games; (1) Finishing exercises (i.e., individual, sectional and intersecional situations: 1 × 0 + GK to 11 × 0 + GK); (2) Individual enrichment training (i.e., 1v1 to 3v3). (2) Game strategy. (2) Tactical schemes (i.e., outsides and corners). Note: Abbreviations: COD, change of direction speed; GK, goalkeeper; HRmax, maximal heart rate; LSG, large-side games; MD, “match day minus” format; MSG, medium-sided games; MRS, maximum running speed; PHV, peak high velocity; SAQ, speed, agility and quickness; SSC, stretch-shortening cycle; SSG, small-sided games. Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 5/21 SSG, MSG, LSG) was categorized in accordance with Zurutuza et al. (2020). The SAQ training was classified by Trecroci et al. (2016) for sub-elite football players. Training load measures Table 3 described the construct, measurement unit, and formula for each ETL and ITL measure. All constructs were considered according to previous TL-based reports, specifically: (i) total distance (TD); (ii) average running velocity; (iii) high-speed running (HSR); (iv) explosiveness, ACC and body impacts; (v) HR-based measures; and (vi) perceived exertion and recovery (Rico-González et al., 2022a; Sheehan et al., 2020; Teixeira et al., 2021a). External load measures The ETL was tracked using a 18 Hz global positioning system (GPS) coupled with accelerometer (100 Hz), magnetometer (10 Hz) and gyroscope (100 Hz) (STATSports Apex
, Northern Ireland) (Buchheit et al., 2021). With a reliable satellite signal, all devices were turned on 30 min before the training data collection (Beato et al., 2018; Buchheit et al., Table 3 Construct, description and formulas from external and internal training load. TL Constructs Variable Description and formula ETL Total distance TD (m) Total distance covered (in meters) Average running velocity AvS (m·min−1) Game pace or average speed distance in meter per minutes. MRS (m·s−1) Maximal speed in meter per seconds High intensity running rHSR (m) Relative high-speed running (rHSR) distance (m) covered at 19.8–25.1 km·h−1. SPR (n | m) The sprints were measured by number and average sprint distance (m) in a velocity >25.1 km·h−1. Explosiveness, accelerations and impacts HMLD (m) High metabolic load distance (HMLD) is a metabolic variable defined as the distance, expressed in meters, covered by player when the metabolic power exceeds 25.5 W·kg−1. DSL (au) The DSL was computed by measuring the sum of the accelerations in the three orthogonal axes of movement (expressed as a G force > 2G). ACC | DEC (m·s−2) Number of accelerations (>3 m·s−2) and decelerations. ITL HR HRmax (bpm) Maximum heart rate (HRmax) AvHR (bpm) Average heart rate (AvHR). %HRmax Percentage of HRmax (%HRmax) TRIMP (au) Akubat TRIMP (iTRIMP) = Training duration × 0.2053e3.5179x. Among which e = Napierian logarithms, 3.5179 is the exponent, and x = HRratio. Perceived exertion and recovery RPE (au) Perceived exertion was measured by 15-point Portuguese Borg Rating of Perceived Exertion 6–20 Scale (Borg RPE 6–20). sRPE The sRPE was obtained by multiplying total duration of training sessions for each individual RPE score. TQR (au) To monitor recovery, each player was asked to report the TQR score on a scale from 6 to 20. Note: Abbreviations: ACC, acceleration; AvHR, average heart rate; AvS, average speed; DEC, deceleration; HMLD, high metabolic load distance; HRmax, maximal heart rate; MRS, maximum running speed; SPR, average sprint distance; SPR_N, number of sprints; sRPE, session ratings of perceived exertion; TD, total distance; TL, Training load; TQR, total quality recovery; TRIMP, training impulse. Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 6/21 2021). The accuracy of GPS Apex
devices was good (bias 5%) (Beato et al., 2018). The ETL measures were as follows: TD covered (m), average speed (AvS (m·min−1)), maximal running speed (MRS (m·s−1)), relative high-speed running (rHSR (m): 19.8–25.1 km·h−1) distance (m), high metabolic load distance (HMLD (m) > 25.5 W·kg−1), number sprints (n) and average sprint distance (SPR (m) (>25.1 km·h−1)) (m), dynamic stress load (DSL (a.u.)), number of ACC (>3 m·s−2) and number of decelerations (DEC < 3 m·s−2) (Teixeira et al., 2021b, 2022a) (Table 3). Internal training load measures The ITL were obtained by RPE, TQR, and the HR monitors. A Garmin
TM HR band (Garmin Ltd
, International Ltd., Olathe, KS, USA) was used to capture HR-based measurements utilizing a 1 Hz short-range telemetry system (Gómez-Carmona et al., 2020). Maximum heart rate (HRmax), average heart rate (HRmean), percentage of HRmax (% HRmax) and individual players’ training impulse (TRIMP) were monitored (Akubat et al., 2012; Branquinho, Ferraz & Marques, 2021). The Yo-Yo intermittent recovery test level 1 (YYIR1) was used to determine HRmax (Bangsbo, Iaia & Krustrup, 2008). The 15-point Portuguese Borg’s RPE 6-20 scale (Cabral et al., 2020) and TQR 6-20 score (Brink et al., 2010; Kenttä & Hassmén, 1998) were used to evaluate perceived effort. The entire time of training sessions for each participant was multiplied to get the session RPE (sRPE = RPE × session duration). Individual RPE’s and TQR’s were taken 30 min after and before each training session, respectively. Players were already familiarized with the RPE procedures by reporting in a Microsoft Excel
spreadsheet (Microsoft Corporation
, Redmond, WA, USA) (Teixeira et al., 2021b, 2022a) (Table 3). Baseline characteristics Players’ individual characteristics were collected by height (m), weight (kg), chronological age (years), sitting height (cm) and experience level (years). Anthropometric measures were measured using standard the International Society for the Advancement of Kinanthropometry (ISAK) guidelines (Marfell-Jones et al., 2006). Body mass (kg) was evaluated by an electronic scale Tanita MC 780-P MA
(Tanita Corporation, Tokyo, Japan) with minimum clothing. Height (cm) was collected using an electronic stadiometer (Seca, Hamburg, Germany). Players’ height (m), weight (kg) and sitting height (cm) were recorded by the average of three measurements to the nearest 0.1 using international units (IU). Body mass index (BMI) was calculated by dividing weight by the square of height (kg/m2). BMI’s cut-offs used were: underweight < 18.5 kg/m2, normal 18.50–24.99 kg/m2, overweight ≥ 25 kg/m2, obese ≥ 30 kg/m2 (Suarez-Arrones et al., 2018). Relative age (a.u.) was calculated as the difference between the player’s birthdate and the cut-off date (31st August) was divided by the number of 365 days a year (Hill et al., 2020). Based on a predictive set of Mirwald’s equations, maturity offset and peak high velocity (PHV) were calculated (Mirwald et al., 2002; Teixeira et al., 2022a). Sub-elite young football was divided into pre-PHV (n = 52), mid-PHV (n = 65) and post-PHV (n = 207). Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 7/21 Resultant equations for training load monitoring The individual-based principal component in the resultant equations for TL monitoring were: low-moderate volume, high intensity, explosiveness, change of direction, collisions and body impacts (Rico-González et al., 2022a; Sheehan et al., 2020; Teixeira et al., 2021a). Also, the resultant equations added the baseline characteristics (i.e., anthropometric and maturational status) and the ITL measures (de Dios-Álvarez et al., 2021; Suarez-Arrones et al., 2018). Thus, the resultant equations was computed by a PCA approach can be expressed by the following algorithm (Jolliffe & Cadima, 2016): PCAn ¼ X Φi1  xi þ Φi2  x2 . . . ð ÞΦin  xn where the PCAn is the n principal component, Φ is the loading vector comprising loadings (i1, i1…) of the first principal component. The loadings must have a sum of squares of exactly one. This is due to the possibility of a considerable variation when loadings are of a great magnitude. It also specifies how the major component will move (PCAn), along which data varies the most (Jokiniemi, Pietilä & Mikkonen, 2021). The outcome is a line that is closest to the n observations in p-dimensional space. Euclidean distance squared is used to gauge proximity; xn are normalized predictors. Normalized predictors (xn) have mean values equal to zero and standard deviations equal to one (Jokiniemi, Pietilä & Mikkonen, 2021; Jolliffe & Cadima, 2016). Resultant equation to quantify the weighted TL was expressed by: TLweekly ¼ X PCA1 þ PCA2 . . . ð ÞPCAn where the TLWeekly is the sum of each PCA (p) and its weighted load vector (Jolliffe & Cadima, 2016). Statistical analysis A data reduction technique was conducted using a principal component analysis (PCA) with 95% confidence intervals (95% CI) (Pino-Ortega et al., 2021; Rojas-Valverde et al., 2020). Monte Carlo parallel analysis were conducted to determine the number of extracted factors (Jokiniemi, Pietilä & Mikkonen, 2021). Z score were computed to scaled and centered final selection variables for PCA using Kaiser–Meyer–Olkin (KMO) values for measure of sampling adequacy and the Bartlett Sphericity test to ensure the sampled training data was suitable for data reduction. Factor analysis was acceptable when KMO values are greater than 0.6 and Bartlett Sphericity less than 0.05 (Pino-Ortega et al., 2021). The number of PCA to be retained was determined using the scree plot for the derived factor eigenvalues, considering eigenvalues greater than 1 (Rojas-Valverde et al., 2020). Factor’s components loading was computed using an orthogonal rotation with a VariMax method due to perpendicularity in the correlation matrix of the interest variables (Warmenhoven et al., 2019). Selection criteria for extraction of non-correlated variables was performed in r < 0.4 (Rojas-Valverde et al., 2020). Weightings (eigenvectors) are represented by a 2D plot and the results of the PCA are presented in a path analysis. The sample size was calculated by GPower, Version 3.1.5.1 (Institut für Experimentelle Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 8/21 Psychologie, Düsseldorf, Germany) with an effect size ß of 0.4, an a of 0.05, and a power of 0.8 (1−ß) (Teixeira et al., 2022a). Kolmogorov–Smirnov and Levene’s test were used to assess the normality and homogeneity. Statistical significance was set at p < 0.05. Data are presented as the mean ± SD using JASP software (JASP Team, 2022; jasp-stats.org). RESULTS Data-reduction procedure, eigenvalue and component number Figure 2 presents the eigenvalue ranged between 1.44% and 5.21%. Overall, five PCA accounted for 68.6% of the total explained variance. The five extracted PCA explained 27.4%, 16.2%, 10.0%, 8.3% and 6.7% of the variance in TL dataset, respectively. Thus, the first PC explained 27.4% of the TL by TD, HMLD, DSL, ACC and DEC. The second PCA explained 16.2% of the TL thought HSRr and SPR. The thirty PCA explained 10.0% of the TL via HRmax, AvHR, %HR and TRIMP. The fourth PCA explained 8.3% of the baseline outset (i.e., sRPE, TQR, maturation offset and chronological age). The fifth PCA explained 6.7% of the accumulated TL (i.e., AvS and MRS). Constantly, PHV, relative age, experience level and BMI were excluded from the PCA (r < 0.4). Table 4 also shows the data-reduction procedure resulting from rotated component matrix for accumulated TL variables with factor component loadings (eigenvectors). Four variables were excluded from the PCA due to the communalities below 0.4 (i.e., PHV, relative age, experience level and BMI). Also, KMO’s criteria reported a sampling adequacy of sampled data, reporting a considerable proportion of the variance as result of the underlying factors (KMO = 0.73). Furthermore, significant Barlett Sphericity test was significant (p < 0.001). Resultant equations and paths from principal components analysis The weightings (eigenvectors) of the PCA analysis are represented by a path graph in Fig. 3. Overall, the weightings ranged between −0.52 to 0.97. The highest weightings were observed in AvHR (bpm) (PCA 3) and the lowest weightings in sRPE (au) (PCA 4). 0 1 2 3 4 5 6 0 5 10 15 20 Component Eigenvalue Data Simulated (95th quantile) Figure 2 Scree plot for principal component analysis representing the component, explained variance and eigenvalues. Full-size  DOI: 10.7717/peerj.15806/fig-2 Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 9/21 Considering the highest factor in each principal component, the variables considered were TD (0.698), SPR (0.940), AvHR (0.967), Age (0.836) and MRS (0.790) for PCA 1 to PCA 5. The resultant equations from extracted principal component are presented in Table 5. On this basis, the resultant equations for TL monitoring during a weekly microcycle can be expressed into five principal components determine the equations for the baseline variables: (1) explosiveness and impacts; (2) HSR; (3) HR measures; (4) baseline characteristics; (5) average running velocity. DISCUSSION The aim of this study was to reduce the dimensionality of the internal and external load measures, by a PCA approach, in order to describe and explain the resultant equations for TL monitoring during a standard microcycle in a sub-elite youth football players. Additionally, it is intended to identify the most representative measure for each principal component. After data reduction, five principal components were extracted from TL dataset explaining 68.7% of the total variance. The TL measures with the highest weight in each PCA were DEC, SPR distance, average HR, chronological age and MRS. Resultant equations for TL monitoring during a standard microcycle in sub-elite youth football was split into: (1) explosiveness, ACC and impacts (27.4%); (2) HSR (16.2%); (3) heart bate-based measures (10.0%); (4) baseline characteristics (8.3%); (5) average running Table 4 Principal component analysis: data reduction procedure using varimax for rotated component matrix with factor loadings (eigenvectors) >0.4. Variables PC1 PC2 PC3 PC4 PC5 Uniqueness TD (m) 0.698 0.365 AvS (m·min−1) 0.680 0.321 MRS (m·s−1) 0.790 0.259 HSRr (m) 0.928 0.041 HMLD (m) 0.788 0.501 0.123 SPR (n) 0.895 0.088 SPR (m) 0.940 0.066 DSL (au) 0.705 0.465 ACC (m·s−2) 0.844 0.233 DEC (m·s−2) 0.877 0.184 HRmax (bpm) 0.763 0.366 HRAv (bpm) 0.967 0.055 %HRmax 0.953 0.081 TRIMP (au) 0.692 0.501 sRPE (au) −0.516 0.629 TQR (au) −0.553 0.676 OFFSET (y) 0.669 0.343 Age (y) 0.836 0.261 Note: Abbreviations: ACC, acceleration; AvHR, average heart rate; AvS, average speed; DEC, deceleration; HMLD, high metabolic load distance; HRmax, maximal heart rate; MRS, maximum running speed; SPR, average sprint distance; SPR_N, number of sprints; sRPE, session ratings of perceived exertion; TD, total distance; TQR, total quality recovery; TRIMP, training impulse. Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 10/21 Table 5 Resultant equations from extracted principal component analysis. PCA Construct Variables Calculation 1 Explosiveness, accelerations and impacts TD (m), HMLD (m), DSL (au), ACC (>3 m·s−2), DEC (<3 m·s−2) 0.698 × TD (m) + 0.788 × HMLD (m) + 0.705 × DSL (au) + 0.844 × ACC (m·s−2) + 0.877 × DEC (m·s−2) 2 High intensity running rHSR (19.8–25.1 km · h−1), SPR (n), SPR (m) 0.928 × rHSR (km · h−1) + 0.895 × SPR (n) + 0.940 × SPR (m) 3 Heart rate HRmax (bpm), AvHR (bpm), %HRmax, TRIMP (au) 0.763 × HRmax (bpm) + 0.967 × AvHR (bpm) + 0.953 × %HRmax + 0.692 × AkubatTRIMP (au) 4 Baseline characteristics TQR (au), sRPE (au), Offset (y), Age (y) −0.553 × TQR (au) + −0.516 × sRPE (au) + 0.669 × Offset (y) + 0.836 × Age (y) 5 Average running velocity AvS (m·min−1), MRS (m·s−1) 0.680 × AvS (m · min−1) + 0.790 × MRS (m·s−1) Note: Abbreviations: ACC, acceleration; AvHR, average heart rate; AvS, average speed; DEC, deceleration; HMLD, high metabolic load distance; HRmax, maximal heart rate; MRS, maximum running speed; SPR, average sprint distance; SPR_N, number of sprints; sRPE, session ratings of perceived exertion; TD, total distance; TQR, total quality recovery; TRIMP, training impulse. Figure 3 Principal component analysis and weightings (eigenvectors) were presented with a path. Full-size  DOI: 10.7717/peerj.15806/fig-3 Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 11/21 velocity (6.7%). Considering the highest representative factor in each principal component, the variables considered were DEC (PCA 1), SPR distance (PCA 2), average HR (PCA 3), chronological age (PCA 4) and MRS (PCA 5). In football, Pino-Ortega et al. (2021) also determined conditional dimensions such as angular velocity, speed displacements, HMLD, HSR, SPR, TD covered, metabolic power, DSL, jumps, impacts, ACC and DEC. The first PCA complies TD, HMLD, DSL, ACC and DEC, being grouped as explosiveness, ACC and impacts. Although there is a definite correlation between body impacts, ACC, and DEC. Otherwise, the TD may be due to an inverse relationship between training volume and intensity (Castillo et al., 2020). Also, the metabolic power was rather than speed-based zones to express running intensity (Osgnach et al., 2010). Nevertheless, the TD could fall outside this construct at first sight. An interaction effect between TD and DEC had already been documented for sub-elite football players (Teixeira et al., 2021b). The second PCA extracted HSRr and SPR, wherefore the HSR is an excellent variable to give meaning about training intensity (Harper et al., 2020). Zurutuza et al. (2020) combined peak velocity and distance covered at different velocities in the same principal component, confirming our results on high intensity demands. The third PCA complied the HR-based measures (i.e., HRmax, AvHR, %HRmax and TRIMP), confirming the correlation between HR-based measures and ETL outcomes (de Dios-Álvarez et al., 2021; Ellis et al., 2021). The fourth PCA was explained by TQR, sRPE, maturation offset and chronological age. Although the fourth PCA has a lower variance explained it is fundamental to consider the influence of chronological age, biological age and perceived exertion (Teixeira et al., 2022a). In line with this component, the perceived exertion seems to be better explained with trainability, maturation and stage of development (Malina et al., 2019). Also, the TL could be influenced by acute: chronic workload ratio, training monotony and well-being variations (Clemente et al., 2021a, 2021b; Rico-González et al., 2022c). Indeed, the literature reported that greater acute: chronic workload ratio and training monotony levels are normally associated with an increased risk of injury or health issues. These measurements might be utilized to comprehend how the data changes throughout in-season phases (Rico-González et al., 2022a). Effectively, perceived exertion in young football players may be also influenced psychophysiological determinants as self-perception of competence and practice experience (Branquinho et al., 2021; Ferraz et al., 2017, 2018). Leading biological maturation in youth sports has become a research-practice gap still lacking knowledge about sub-elite environments using data reduction approaches (Cumming, 2018; Teixeira et al., 2021b, 2022c). Finally, the fifth PCA explained 6.7% of the accumulated TL thought AvS and MRS. Pacing behavior was also reported as a key point to football performance (Ferraz et al., 2018, 2020). Research findings was slightly small than previous research in futsal (Rico-González et al., 2022a), Australian football (Sheehan et al., 2020), rugby (Scantlebury et al., 2020; Weaving et al., 2020) and Gaelic football (Gamble et al., 2019). Wherefore, the comparisons with current research would consider the differences between football codes. Scantlebury et al. (2020) reported a cumulative explained variance of 91%, 96% and 91% variance in TL in rugby union, field hockey and soccer. Casamichana et al. (2019) reported an explained variance of the external training intensity between 39% and 44%. Also, the Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 12/21 eigenvalue of this study ranged between 1.44% to 5.21% by setting up values of accumulated TL substantially lower compared to other studies (i.e., eigenvalues between 1.0% and 68.0%) (Pino-Ortega et al., 2021; Scantlebury et al., 2020). Albeit, current research represents the first time that this statistical approach has been used in a sub-elite youth football, specifically using training data (Rico-González et al., 2022b; Rojas-Valverde et al., 2020). Current applied PCA determine the resultant equations from individual-based principal components, expressing by major component weightings (Rico-González et al., 2022a; Sheehan et al., 2020; Teixeira et al., 2021a). Indeed, this is the traditional PCA algorithm that computes the principal components based on the covariance matrix or the singular value decomposition the data. It is widely used methods in team sports for dimensionality reduction, data visualization, and feature extraction (Pino-Ortega et al., 2021; Rico- González et al., 2022b; Rojas-Valverde et al., 2020). Other ratios, scores and equivalent equations have already been proposed to measure the TL, by emphasizing training intensity, volume or locomotion profile (Clemente et al., 2019; Owen et al., 2017; Rago et al., 2019). However, the PCA algorithms are diverse and some have not yet been implemented in football (Rico-González et al., 2022b; Rojas-Valverde et al., 2020). Hence, future perspective can explore other PCA algorithms such as incremental, Kernel, sparse and robust PCA approaches (Rojas-Valverde et al., 2020). Incremental PCA allows for incremental updates to the principal components as new data points are added in large datasets or when new data is continuously acquired, such as in real-time monitoring of football players’ performance or training data (Jokiniemi, Pietilä & Mikkonen, 2021). Kernel, sparse and robust PCA has been mainly applied for nonlinear dimensionality reduction, sparsity constraints and noisy or incomplete data (Teixeira et al., 2022c). Futures research should expand the resultant equations by considering other well-being, technical-tactical and match-related contextual factors. Also, PCA approach must also consider the principal component in TL monitoring when considering training mode (i.e., small-sided and conditioned games), training day (i.e., MD-3, MD-2, and MD-1), age group (i.e., U15, U17, and U19) and maturational bands (i.e., pre-, mid- and post-PHV) (Teixeira et al., 2021a). Additionally, the training data represents only a specific sub-elite football academy and must be considered carefully when applied to another to other teams and contexts. As study limitations, the sample size and number of factors was rather small than previous studies with longer monitoring period (Rojas-Valverde et al., 2020). Also, the total variance was also relatively smaller for this PCA paths than other reports in football codes (Pino-Ortega et al., 2021; Rojas-Valverde et al., 2020). However, it must be ensured that football had the lowest percentage of the variance comparing with other football codes (Rojas-Valverde et al., 2020). Furthermore, choosing a higher threshold for total variance (%) may result in fewer retained principal components and a higher degree of data reduction with a consequent loss, noise or redundant information (Jokiniemi, Pietilä & Mikkonen, 2021; Jolliffe & Cadima, 2016). In general, there is no strict rule for the minimum value for percentage of total variance in PCA, as it depends on the specific application and the trade-off between data reduction and information retention (Rojas- Valverde et al., 2020). Furthermore, a commonly used threshold for retaining a principal Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 13/21 component is to choose those components that explain at least 60–80% of the total variance, depending on the specific data analysis requirements (Jokiniemi, Pietilä & Mikkonen, 2021; Jolliffe & Cadima, 2016). Finally, the TL strategies applied in this quasi-experimental approach for only compiles GPS, HR and perceived exertion, however more objective measure of fatigue and recovery should be considered in futures reports, such as HR variability, electromyography signal intensity, biochemical markers and other well-being measures (Clemente et al., 2021a, 2021b). Also, further PCA approaches are needed to consider the principal components when integrating physical, physiological and tactical factors in football under an integrative perspective (Teixeira et al., 2022c). PRACTICAL APPLICATIONS  Current resultant composite equations can be applied to relative contribution of the ITL and ETL measures for monitoring and management load in sub-elite youth football.  Data reduction techniques decrease the redundant information and dimensionality of the training data, expressing in the following principal components: explosiveness and impacts, high-speed running, heart bate-based measures, baseline characteristics and average running velocity.  Considering the highest factor in each principal component, DEC (PCA 1), sprint distance (PCA 2), average HR (PCA 3), chronological age (PCA 4) and maximal speed (PCA 5) are the conditional dimension to be considered in TL monitoring during a standard microcycle in sub-elite youth football players.  Maturational status should be carefully considered in the TL monitoring together with relative age effect, chronological and baseline characteristics.  Self-perception and practice experience may affect the variance explained by perceived exertion and pacing behavior.  Training intensity and volume can be more accurately measured by current resultant composite equations and/or most representative factor for a standard microcycle in sub-elite youth football players.  Futures research should expand the resultant equations for TL monitoring in sub-elite youth football with well-being, technical-tactical and match-related contextual factors. CONCLUSION Using a PCA approach, five principal components could be applied to extract to describe and explain resultant equations for TL monitoring during an in-season standard microcycle in sub-elite youth football. Current research provides the first composite equations to extract the TL in this specific population expressed as explosiveness and impacts, high-speed running, HR-based measures, baseline characteristics and average running velocity. Considering the highest factor in each principal component, DEC (PCA 1), SPR distance (PCA 2), average HR (PCA 3), chronological age (PCA 4) and maximal SPR (PCA 5) are the conditional dimension to be considered in TL monitoring during a standard microcycle in sub-elite youth football players. Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 14/21 Future research should expand the resultant equations within the microcycle, by considering other well-being measures, technical-tactical factors and match-related contextual factors. ADDITIONAL INFORMATION AND DECLARATIONS Funding This project was supported by the National Funds through FCT—Portuguese Foundation for Science and Technology (UIDB/DTP/04045/2020). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Grant Disclosures The following grant information was disclosed by the authors: National Funds through FCT—Portuguese Foundation for Science and Technology: UIDB/DTP/04045/2020. Competing Interests Tiago M. Barbosa is an Academic Editor for PeerJ. Author Contributions  José Eduardo Teixeira conceived and designed the experiments, performed the experiments, analyzed the data, prepared figures and/or tables, authored or reviewed drafts of the article, and approved the final draft.  Pedro Forte conceived and designed the experiments, prepared figures and/or tables, authored or reviewed drafts of the article, and approved the final draft.  Ricardo Ferraz performed the experiments, analyzed the data, authored or reviewed drafts of the article, and approved the final draft.  Luís Branquinho performed the experiments, analyzed the data, authored or reviewed drafts of the article, and approved the final draft.  Ryland Morgans analyzed the data, authored or reviewed drafts of the article, and approved the final draft.  António José Silva conceived and designed the experiments, authored or reviewed drafts of the article, and approved the final draft.  António Miguel Monteiro conceived and designed the experiments, authored or reviewed drafts of the article, and approved the final draft.  Tiago M. Barbosa conceived and designed the experiments, authored or reviewed drafts of the article, and approved the final draft. Human Ethics The following information was supplied relating to ethical approvals (i.e., approving body and any reference numbers): Informed consent was obtained from all subjects involved in the study. The experimental approach was approved and followed by the local Ethical Committee from University of Trás-os-Montes & Alto Douro (3379-5002PA67807). Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 15/21 Data Availability The following information was supplied regarding data availability: The raw measurements are available in the Supplemental File. Supplemental Information Supplemental information for this article can be found online at http://dx.doi.org/10.7717/ peerj.15806#supplemental-information. REFERENCES Akubat I, Patel E, Barrett S, Abt G. 2012. Methods of monitoring the training and match load and their relationship to changes in fitness in professional youth soccer players. Journal of Sports Sciences 30(14):1473–1480 DOI 10.1080/02640414.2012.712711. Bangsbo J, Iaia FM, Krustrup P. 2008. The Yo-Yo intermittent recovery test: a useful tool for evaluation of physical performance in intermittent sports. Sports Medicine 38(1):37–51 DOI 10.2165/00007256-200838010-00004. Beato M, Coratella G, Stiff A, Iacono AD. 2018. The validity and between-unit variability of GNSS units (STATSports Apex 10 and 18 Hz) for measuring distance and peak speed in team sports. Frontiers in Physiology 9:1288 DOI 10.3389/fphys.2018.01288. Bourdon P, Cardinale M, Murray A, Gastin P, Kellmann M, Varley M, Gabbett T, Coutts A, Burgess D, Gregson W, Cable N. 2017. Monitoring athlete training loads: consensus statement. International Journal of Sports Physiology and Performance 12(s2):S2-161 DOI 10.1123/IJSPP.2017-0208. Branquinho L, Ferraz R, Marques MC. 2021. 5-a-side game as a tool for the coach in soccer training. Strength & Conditioning Journal 43(5):96–108 DOI 10.1519/SSC.0000000000000629. Branquinho L, Ferraz R, Travassos B, Marinho DA, Marques MC. 2021. Effects of different recovery times on internal and external load during small-sided games in soccer. Sports Health 13(4):324–331 DOI 10.1177/1941738121995469. Brink MS, Nederhof E, Visscher C, Schmikli SL, Lemmink KA. 2010. Monitoring load, recovery, and performance in young elite soccer players. The Journal of Strength & Conditioning Research 24(3):597 DOI 10.1519/JSC.0b013e3181c4d38b. Buchheit M, Simpson BM, Hader K, Lacome M. 2021. Occurrences of near-to-maximal speed-running bouts in elite soccer: insights for training prescription and injury mitigation. Science and Medicine in Football 5(2):105–110 DOI 10.1080/24733938.2020.1802058. Cabral LL, Nakamura FY, Stefanello JMF, Pessoa LCV, Smirmaul BPC, Pereira G. 2020. Initial validity and reliability of the Portuguese Borg rating of perceived exertion 6–20 scale. Measurement in Physical Education and Exercise Science 24:103–114 DOI 10.1080/1091367X.2019.1710709. Casamichana D, Castellano J, Díaz AG, Martín-García A. 2019. Looking for complementary intensity variables in different training games in football. Journal of Strength and Conditioning Research. Publish Ahead of Print DOI 10.1519/JSC.0000000000003025. Castillo D, Raya-González J, Clemente FM, Yanci J. 2020. The influence of youth soccer players’ sprint performance on the different sided games’ external load using GPS devices. Research in Sports Medicine 28(2):194–205 DOI 10.1080/15438627.2019.1643726. Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 16/21 Clemente FM, Afonso J, Costa J, Oliveira R, Pino-Ortega J, Rico-González M. 2021a. Relationships between sleep, athletic and match performance, training load, and injuries: a systematic review of soccer players. Healthcare 9(7):808 DOI 10.3390/healthcare9070808. Clemente FM, González-Fernández FT, Ceylan HI, Silva R, Younesi S, Chen YS, Badicu G, Wolanski P, Murawska-Ciałowicz E. 2021b. Blood biomarkers variations across the pre-season and interactions with training load: a study in professional soccer players. Journal of Clinical Medicine 10(23):5576 DOI 10.3390/jcm10235576. Clemente FM, Rabbani A, Conte D, Castillo D, Afonso J, Truman Clark CC, Nikolaidis PT, Rosemann T, Knechtle B. 2019. Training/match external load ratios in professional soccer players: a full-season study. International Journal of Environmental Research and Public Health 16(17):3057 DOI 10.3390/ijerph16173057. Coutinho D, Gonçalves B, Figueira B, Abade E, Marcelino R, Sampaio J. 2015. Typical weekly workload of under 15, under 17, and under 19 elite Portuguese football players. Journal of Sports Sciences 33(12):1229–1237 DOI 10.1080/02640414.2015.1022575. Cumming SP. 2018. A game plan for growth: how football is leading the way in the consideration of biological maturation in young male athletes. Annals of Human Biology 45(5):373–375 DOI 10.1080/03014460.2018.1513560. de Dios-Álvarez V, Suárez-Iglesias D, Bouzas-Rico S, Alkain P, González-Conde A, Ayán- Pérez C. 2021. Relationships between RPE-derived internal training load parameters and GPS-based external training load variables in elite young soccer players. Research in Sports Medicine 31(1):1–16 DOI 10.1080/15438627.2021.1937165. Ellis M, Penny R, Wright B, Noon M, Myers T, Akubat I. 2021. The dose-response relationship between training-load measures and aerobic fitness in elite academy soccer players. Science and Medicine in Football 5(2):128–136 DOI 10.1080/24733938.2020.1817536. Ferraz R, Gonçalves B, Coutinho D, Marinho DA, Sampaio J, Marques MC. 2018. Pacing behaviour of players in team sports: influence of match status manipulation and task duration knowledge. PLOS ONE 13(2):e0192399 DOI 10.1371/journal.pone.0192399. Ferraz R, Gonçalves B, Coutinho D, Oliveira R, Travassos B, Sampaio J, Marques MC. 2020. Effects of knowing the task’s duration on soccer players’ positioning and pacing behaviour during small-sided games. International Journal of Environmental Research and Public Health 17(11):3843 DOI 10.3390/ijerph17113843. Ferraz R, Gonçalves B, Tillaar R, Saiz S, Sampaio J, Marques M. 2017. Effects of knowing the task duration on players’ pacing patterns during soccer small-sided games. Journal of Sports Sciences 36(1):116–122 DOI 10.1080/24733938.2017.1283433. Ford PR, Bordonau JLD, Bonanno D, Tavares J, Groenendijk C, Fink C, Gualtieri D, Gregson W, Varley MC, Weston M, Lolli L, Platt D, Di Salvo V. 2020. A survey of talent identification and development processes in the youth academies of professional soccer clubs from around the world. Journal of Sports Sciences 38(11–12):1269–1278 DOI 10.1080/02640414.2020.1752440. Gamble D, Bradley J, McCarren A, Moyna NM. 2019. Team performance indicators which differentiate between winning and losing in elite Gaelic football. International Journal of Performance Analysis in Sport 19(4):478–490 DOI 10.1080/24748668.2019.1621674. Gómez-Carmona CD, Bastida-Castillo A, González-Custodio A, Olcina G, Pino-Ortega J. 2020. Using an inertial device (WIMU PRO) to quantify neuromuscular load in running: reliability, convergent validity, and influence of type of surface and device location. The Journal of Strength & Conditioning Research 34(2):365–373 DOI 10.1519/JSC.0000000000003106. Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 17/21 Gonçalves B, Coutinho D, Exel J, Travassos B, Peñas C, Sampaio J. 2019. Extracting spatial-temporal features that describe a team match demands when considering the effects of the quality of opposition in elite football. PLOS ONE 14(8):e0221368 DOI 10.1371/journal.pone.0221368. Griffin A, Kenny IC, Comyns TM, Purtill H, Tiernan C, O’Shaughnessy E, Lyons M. 2021. Training load monitoring in team sports: a practical approach to addressing missing data. Journal of Sports Sciences 39(19):2161–2171 DOI 10.1080/02640414.2021.1923205. Haddad M, Stylianides G, Djaoui L, Dellal A, Chamari K. 2017. Session-RPE method for training load monitoring: validity, ecological usefulness, and influencing factors. Frontiers in Neuroscience 11:612 DOI 10.3389/fnins.2017.00612. Harper DJ, Morin J-B, Carling C, Kiely J. 2020. Measuring maximal horizontal deceleration ability using radar technology: reliability and sensitivity of kinematic and kinetic variables. Sports Biomechanics 1–17 DOI 10.1080/14763141.2020.1792968. Hill M, Scott S, Malina RM, McGee D, Cumming SP. 2020. Relative age and maturation selection biases in academy football. Journal of Sports Sciences 38(11–12):1359–1367 DOI 10.1080/02640414.2019.1649524. Impellizzeri FM, Jeffries AC, Weisman A, Coutts AJ, McCall A, McLaren SJ, Kalkhoven J. 2022. The ‘training load’ construct: why it is appropriate and scientific. Journal of Science and Medicine in Sport 25(5):445–448 DOI 10.1016/j.jsams.2021.10.013. JASP Team. 2022. JASP (Version 0.16.3) [Computer Software]. Available at https://jasp-stats.org/ download/ (accessed 1 December 2022). Jokiniemi K, Pietilä AM, Mikkonen S. 2021. Construct validity of clinical nurse specialist core competency scale: an exploratory factor analysis. Journal of Clinical Nursing 30(13–14):1863–1873 DOI 10.1111/jocn.15587. Jolliffe IT, Cadima J. 2016. Principal component analysis: a review and recent developments. Philosophical Transactions of the Royal Society A: Mathematical, Physical and Engineering Sciences 374(2065):20150202 DOI 10.1098/rsta.2015.0202. Kenttä G, Hassmén P. 1998. Overtraining and recovery: a conceptual model. Sports Medicine 26(1):1–16 DOI 10.2165/00007256-199826010-00001. Los Arcos A, Mendiguchia J, Javier Y. 2017. Specificity of jumping, acceleration and quick change of direction motor abilities in soccer players. Kinesiology 49(1):22–27 DOI 10.26582/k.49.1.12. Malina R, Cumming S, Rogol A, Coelho-e-Silva M, Figueiredo A, Konarski J, Koziel S. 2019. Bio-banding in youth sports: background, concept, and application. Sports Medicine 49(11):1671–1685 DOI 10.1007/s40279-019-01166-x. Marfell-Jones M, Olds T, Stewart A, Carter L. 2006. ISAK accreditation handbook. Montreal: International Society for the Advancement of Kinanthropometry (ISAK). Mirwald RL, Baxter-Jones ADG, Bailey DA, Beunen GP. 2002. An assessment of maturity from anthropometric measurements. Medicine and Science in Sports and Exercise 34(4):689–694 DOI 10.1249/00005768-200204000-00020. Moura FA, Santana JE, Vieira NA, Santiago PRP, Cunha SA. 2015. Analysis of soccer players’ positional variability during the 2012 UEFA European championship: a case study. Journal of Human Kinetics 47(1):225–236 DOI 10.1515/hukin-2015-0078. O’Donoghue P. 2008. Principal components analysis in the selection of key performance indicators in sport. International Journal of Performance Analysis in Sport 8(3):145–155 DOI 10.1080/24748668.2008.11868456. Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 18/21 Oliva-Lozano JM, Muyor JM. 2022. Understanding the FIFA quality performance reports for electronic performance and tracking systems: from science to practice. Science and Medicine in Football 6(3):398–403 DOI 10.1080/24733938.2021.1984557. Oliva-Lozano JM, Rojas-Valverde D, Gómez-Carmona CD, Fortes V, Pino-Ortega J. 2021. Impact of contextual variables on the representative external load profile of Spanish professional soccer match-play: a full season study. European Journal of Sport Science 21(4):497–506 DOI 10.1080/17461391.2020.1751305. Osgnach C, Poser S, Bernardini R, Rinaldo R, Di Prampero PE. 2010. Energy cost and metabolic power in elite soccer: a new match analysis approach. Medicine & Science in Sports & Exercise 42(1):170–178 DOI 10.1249/MSS.0b013e3181ae5cfd. Owen AL, Djaoui L, Newton M, Malone S, Mendes B. 2017. A contemporary multi-modal mechanical approach to training monitoring in elite professional soccer. Science and Medicine in Football 1(3):216–221 DOI 10.1080/24733938.2017.1334958. Pino-Ortega J, Rojas-Valverde D, Gómez-Carmona CD, Rico-González M. 2021. Training design, performance analysis, and talent identification—a systematic review about the most relevant variables through the principal component analysis in soccer, basketball, and rugby. International Journal of Environmental Research and Public Health 18(5):2642 DOI 10.3390/ijerph18052642. Rago V, Brito J, Figueiredo P, Krustrup P, Rebelo A. 2019. Relationship between external load and perceptual responses to training in professional football: effects of quantification method. Sports 7(3):68 DOI 10.3390/sports7030068. Rago V, Rebelo A, Krustrup P, Mohr M. 2020. Contextual variables and training load throughout a competitive period in a top-level male soccer team. The Journal of Strength & Conditioning Research, Publish Ahead of Print 35(11):3177–3183 DOI 10.1519/JSC.0000000000003258. Ric A, Torrents C, Gonçalves B, Sampaio J, Hristovski R. 2016. Soft-assembled multilevel dynamics of tactical behaviors in soccer. Frontiers in Psychology 7(35):1513 DOI 10.3389/fpsyg.2016.01513. Rico-González M, Oliveira R, González Fernández FT, Clemente FM. 2022a. Acute: chronic workload ratio and training monotony variations over the season in youth soccer players: a systematic review. International Journal of Sports Science & Coaching 11:174795412211045 DOI 10.1177/17479541221104589. Rico-González M, Pino-Ortega J, Praça GM, Clemente FM. 2022b. Practical applications for designing soccer’ training tasks from multivariate data analysis: a systematic review emphasizing tactical training. Perceptual and Motor Skills 129(3):892–931 DOI 10.1177/00315125211073404. Rico-González M, Puche-Ortuño D, Clemente FM, Aquino R, Pino-Ortega J. 2022c. The most demanding exercise in different training tasks in professional female futsal: a mid-season study through principal component analysis. Healthcare 10(5):838 DOI 10.3390/healthcare10050838. Ricotti L, Rigosa J, Niosi A, Menciassi A. 2013. Analysis of balance, rapidity, force and reaction times of soccer players at different levels of competition. PLOS ONE 8(10):e77264 DOI 10.1371/journal.pone.0077264. Rojas-Valverde D, Pino-Ortega J, Gómez-Carmona CD, Rico-González M. 2020. A systematic review of methods and criteria standard proposal for the use of principal component analysis in team’s sports science. International Journal of Environmental Research and Public Health 17(23):8712 DOI 10.3390/ijerph17238712. Ruan L, Ge H, Gómez MA, Shen Y, Gong B, Cui Y. 2022. Analysis of defensive playing styles in the professional Chinese Football Super League ahead of print. Science and Medicine in Football 1–9 DOI 10.1080/24733938.2022.2099964. Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 19/21 Scantlebury S, Till K, Beggs C, Dalton-Barron N, Weaving D, Sawczuk T, Jones B. 2020. Achieving a desired training intensity through the prescription of external training load variables in youth sport: more pieces to the puzzle required. Journal of Sports Sciences 38(10):1124–1131 DOI 10.1080/02640414.2020.1743047. Sheehan WB, Tribolet R, Spurrs R, Fransen J, Novak AR, Watsford ML. 2020. Simplifying the complexity of assessing physical performance in professional Australian football. Science and Medicine in Football 4(4):285–292 DOI 10.1080/24733938.2020.1745264. Staunton CA, Abt G, Weaving D, Wundersitz DWT. 2021. Misuse of the term ‘load’ in sport and exercise science. Journal of Science and Medicine in Sport 25(5):439–444 DOI 10.1016/j.jsams.2021.08.013. Suarez-Arrones L, Petri C, Maldonado RA, Torreno N, Munguía-Izquierdo D, Di Salvo V, Méndez-Villanueva A. 2018. Body fat assessment in elite soccer players: cross-validation of different field methods. Science and Medicine in Football 2(3):203–208 DOI 10.1080/24733938.2018.1445871. Teixeira JE, Alves AR, Ferraz R, Forte P, Leal M, Ribeiro J, Silva AJ, Barbosa TM, Monteiro AM. 2022a. Effects of chronological age, relative age, and maturation status on accumulated training load and perceived exertion in young sub-elite football players. Frontiers in Physiology 13:832202 DOI 10.3389/fphys.2022.832202. Teixeira JE, Branquinho L, Ferraz R, Leal M, Silva AJ, Barbosa TM, Monteiro AM, Forte P. 2022b. Weekly training load across a standard microcycle in a sub-elite youth football academy: a comparison between starters and non-starters. International Journal of Environmental Research and Public Health 19(18):11611 DOI 10.3390/ijerph191811611. Teixeira JE, Forte P, Ferraz R, Branquinho L, Silva AJ, Monteiro AM, Barbosa TM. 2022c. Integrating physical and tactical factors in football using positional data: a systematic review. PeerJ 10(3):e14381 DOI 10.7717/peerj.14381. Teixeira JE, Forte P, Ferraz R, Leal M, Ribeiro J, Silva AJ, Barbosa TM, Monteiro AM. 2021a. Monitoring accumulated training and match load in football: a systematic review. International Journal of Environmental Research and Public Health 18(8):3906 DOI 10.3390/ijerph18083906. Teixeira JE, Forte P, Ferraz R, Leal M, Ribeiro J, Silva AJ, Barbosa TM, Monteiro AM. 2021b. Quantifying sub-elite youth football weekly training load and recovery variation. Applied Sciences 11(11):4871 DOI 10.3390/app11114871. Teixeira JE, Forte P, Ferraz R, Leal M, Ribeiro J, Silva AJ, Barbosa TM, Monteiro AM. 2022d. The association between external training load, perceived exertion and total quality recovery in sub-elite youth football. The Open Sports Sciences Journal 15(1):235 DOI 10.2174/1875399X-v15-e2207220. Trecroci A, Milanović Z, Frontini M, Iaia FM, Alberti G. 2018. Physical performance comparison between under 15 elite and sub-elite soccer players. Journal of Human Kinetics 61(1):209–216 DOI 10.1515/hukin-2017-0126. Trecroci A, Milanović Z, Rossi A, Broggi M, Formenti D, Alberti G. 2016. Agility profile in sub-elite under-11 soccer players: is SAQ training adequate to improve sprint, change of direction speed and reactive agility performance? Research in Sports Medicine 24(4):331–340 DOI 10.1080/15438627.2016.1228063. Vanrenterghem J, Nedergaard NJ, Robinson MA, Drust B. 2017. Training load monitoring in team sports: a novel framework separating physiological and biomechanical load-adaptation pathways. Sports Medicine 47(11):2135–2142 DOI 10.1007/s40279-017-0714-2. Warmenhoven J, Cobley S, Draper C, Harrison A, Bargary N, Smith R. 2019. Bivariate functional principal components analysis: considerations for use with multivariate movement Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 20/21 signatures in sports biomechanics. Sports Biomechanics 18(1):10–27 DOI 10.1080/14763141.2017.1384050. Weaving D, Beggs C, Dalton-Barron N, Jones B, Abt G. 2019. Visualizing the complexity of the athlete-monitoring cycle through principal-component analysis. International Journal of Sports Physiology and Performance 14(9):1304–1310 DOI 10.1123/ijspp.2019-0045. Weaving D, Dalton-Barron N, McLaren S, Scantlebury S, Cummins C, Roe G, Jones B, Beggs C, Abt G. 2020. The relative contribution of training intensity and duration to daily measures of training load in professional rugby league and union. Journal of Sports Sciences 38(14):1674–1681 DOI 10.1080/02640414.2020.1754725. Winter EM, Maughan RJ. 2009. Requirements for ethics approvals. Journal of Sports Sciences 27(10):985 DOI 10.1080/02640410903178344. Zurutuza U, Castellano J, Echeazarra I, Guridi I, Casamichana D. 2020. Selection of training load measures to explain variability in football training games. Frontiers in Psychology 10:2897 DOI 10.3389/fpsyg.2019.02897. Teixeira et al. (2023), PeerJ, DOI 10.7717/peerj.15806 21/21

Resultant equations for training load monitoring during a standard microcycle in sub-elite youth football: a principal components approach.

08-04-2023

Teixeira, José Eduardo,Forte, Pedro,Ferraz, Ricardo,Branquinho, Luís,Morgans, Ryland,Silva, António José,Monteiro, António Miguel,Barbosa, Tiago M

eng

PMC8541599

medicina Article Comparison and Performance Validation of Calculated and Established Anaerobic Lactate Thresholds in Running Sanghyeon Ji 1,2 , Aldo Sommer 1,3, Wilhelm Bloch 1,3 and Patrick Wahl 4,*



 Citation: Ji, S.; Sommer, A.; Bloch, W.; Wahl, P. Comparison and Performance Validation of Calculated and Established Anaerobic Lactate Thresholds in Running. Medicina 2021, 57, 1117. https://doi.org/ 10.3390/medicina57101117 Academic Editor: Jan Bilski Received: 15 September 2021 Accepted: 12 October 2021 Published: 16 October 2021 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). 1 The German Research Centre of Elite Sport, German Sport University Cologne, 50933 Cologne, Germany; hyeon7748@gmail.com (S.J.); a.sommer@dshs-koeln.de (A.S.); w.bloch@dshs-koeln.de (W.B.) 2 Department of Sports Medicine and Exercise Physiology, Institute of Sport Sciences, Goethe University Frankfurt, 60487 Frankfurt, Germany 3 Department of Molecular and Cellular Sport Medicine, Institute of Cardiology and Sports Medicine, German Sport University Cologne, 50933 Cologne, Germany 4 Institute of Interdisciplinary Exercise Science and Sports Medicine, Medical School Hamburg, 20457 Hamburg, Germany * Correspondence: patrick.wahl@medicalschool-hamburg.de; Tel.: +49-40-361-226-43209 Abstract: Background and Objectives: This study aimed to compare the calculated running veloc- ity at the anaerobic lactate threshold (cLTAn), determined by a mathematical model for metabolic simulation, with two established threshold concepts (onset of blood lactate accumulation (OBLA; 4 mmol·L−1) and modified maximal deviation method (mDmax)). Additionally, all threshold con- cepts were correlated with performance in different endurance running events. Materials and Methods: Ten sub-elite runners performed a 30 s sprint test on a cycle ergometer adjusted to an isokinetic mode set to a cadence of 120 rpm to determine maximal lactate production rate (VLamax), and a graded exercise test on a treadmill to determine maximal oxygen uptake (VO2max). Running velocities at OBLA, mDmax, and cLTAn were then compared with each other, and further correlated with running performance over various distances (3000 m, 5000 m, and 10,000 m). Results: The mean difference in cLTAn was −0.13 ± 0.43 m·s−1 and −0.32 ± 0.39 m·s−1 compared to mDmax (p = 0.49) and OBLA (p < 0.01), respectively. cLTAn indicated moderate to good concordance with the established threshold concepts (mDmax: ICC = 0.87, OBLA: ICC = 0.74). In comparison with other threshold concepts, cLTAn exhibited comparable correlations with the assessed running performances (cLTAn: r = 0.61–0.76, mDmax: r = 0.69–0.79, OBLA: r = 0.56–0.69). Conclusion: Our data show that cLTAn can be applied for determining endurance performance during running. Due to the consideration of individual physiological profiles, cLTAn offers a physiologically justified approach to assess an athlete’s endurance performance. Keywords: aerobic capacity; anaerobic capacity; maximal lactate production rate; exercise testing; endurance performance; metabolism 1. Introduction Determination of the blood lactate response during exercise is among the most widely used performance diagnostic tools [1,2]. Blood lactate concentration increases above the resting value with increasing exercise intensity. However, as long as exercise is performed at a constant exercise intensity under a certain intensity threshold, blood lactate concentration remains constant, physiologically known as a steady-state condition [3,4]. At a certain exercise intensity, a minor increment in the workload induces an accelerated blood lactate accumulation and subsequent fatigue-related metabolic consequences, such as the negative impact of hydrogen ion accumulation (acidosis) on muscle function and performance [4–7]. This considerable point has been defined as the anaerobic lactate threshold (LTAn), which is generally considered to be a good indicator of individual aerobic endurance performance and can be used for prescribing endurance training intensities [8,9]. Medicina 2021, 57, 1117. https://doi.org/10.3390/medicina57101117 https://www.mdpi.com/journal/medicina Medicina 2021, 57, 1117 2 of 12 In recent decades, researchers have developed several concepts to determine LTAn. Most LTAn concepts are usually applied to lactate performance curves derived from graded incremental exercise tests [8]. Most existing LTAn concepts use either fixed lactate con- centrations [4,10] or inflection points [11,12] as their determination criteria. However, these criteria are derived either arbitrarily or empirically from the graphical analysis of the lactate performance curve. Moreover, LTAn has shown to be strongly dependent on the applied test protocol [13,14] and on the athlete’s training status [15], which is critical because there is no clear standardized test procedure defined, which thus hinders accurate data interpretation and comparison. Therefore, the physiological background and the validity/reliability/comparability of these LTAn concepts have been questioned [8]. Lactate production and removal are ongoing processes, which are closely related to metabolic rate but not necessarily to oxygen delivery [5,6,16,17]. There is a continual exchange of lactate between various organs and cells, which can be used as an energy source for oxidative energy production and/or as a major precursor to gluconeogenesis [5,17]. This emphasizes the complexity of metabolic processes behind blood lactate concentrations during exercise or other conditions. Limiting interpretation solely to blood lactate kinetics in response to graded exercise tests allows only scarce insight into the complex metabolic processes of total energy production [18,19]. In 1984, Mader [20] suggested that the lactate performance curve and the correspond- ing exercise intensity at LTAn may be influenced by aerobic (maximal oxygen uptake; VO2max) or anaerobic (glycolytic) capacity (maximal lactate production rate; VLamax) sepa- rately [20]. Further research confirmed this assumption and showed that different com- binations of VO2max and VLamax can result in two identical lactate performance curves with equal LTAn [18]. In a more differentiated approach, Mader and Heck [3] proposed a mathematical simulation model of energy production processes in skeletal muscle. Using Michaelis–Menten kinetics, these researchers described the activation of glycolysis as a lactate production system and the oxidative phosphorylation as a combustion system, both depending on the total metabolic rate [3]. Based on this theoretical construct, the term “maximal steady-state of blood lactate (MLSS)” was introduced (as another concept of LTAn), at which the extent of lactate formation by glycolysis is exactly equal to the maximal elimination rate of lactate by combustion. Thus, no lactate accumulation in blood lactate over time occurs (Figure 1) [3]. Thereby, it was suggested that accelerated accumulation of blood lactate during exercise is due to the saturation of the combustion system (oxidative phosphorylation) [3], which was later verified by subsequent investigations of lactate kinetics during exercise [6,21]. As this mathematical model considers both the maximal aerobic and anaerobic capacities for the determination of LTAn, it provides differentiated information about the energetic background of LTAn, as well as the physiological profile of an athlete [18]. Based on Mader’s approach, Hauser et al. [22] applied the mathematical model to calculate the power output at MLSS during cycling using individual VO2max- and VLamax- values and demonstrated a significant correlation with the experimental determined MLSS, and high reliability in the estimation of MLSS [23]. However, there is a lack of knowledge regarding the transferability of the model to running. Furthermore, the calculation method in the previous study [22] has only been compared to the empirically determined MLSS, but not to the actual athlete’s competition performance, which is an essential aspect for a practical application of a laboratory testing parameter [8]. Therefore, this study aimed to calculate running velocity at LTAn using individual VO2max and VLamax and an adapted mathematical method initially described by Mader and Heck [3] and Hauser et al. [22]. The calculated LTAn (cLTAn) was then compared with other established experimentally determined LTAn concepts. Additionally, we aimed to validate cLTAn against the athlete’s recent performance in endurance running events. Medicina 2021, 57, 1117 3 of 12 Medicina 2021, 57, x FOR PEER REVIEW 3 of 12 Figure 1. An exemplary description of the mathematical model for metabolic simulation, presenting the gross lactate formation (VLass) and the maximal lactate elimination rate (VLaoxmax) depending on exercise intensity [22]. Maximal lactate steady state (MLSS) is defined as the exercise intensity at which the lactate formation is exactly equal to elimination. VO2max = maximal oxygen uptake; VLamax = maximal lactate production rate; Ks4 = individual constant value of the relationship between oxy- gen demand and running velocity. Based on Mader’s approach, Hauser et al. [22] applied the mathematical model to calculate the power output at MLSS during cycling using individual VO2max- and VLamax- values and demonstrated a significant correlation with the experimental determined MLSS, and high reliability in the estimation of MLSS [23]. However, there is a lack of knowledge regarding the transferability of the model to running. Furthermore, the calcu- lation method in the previous study [22] has only been compared to the empirically de- termined MLSS, but not to the actual athlete’s competition performance, which is an es- sential aspect for a practical application of a laboratory testing parameter [8]. Therefore, this study aimed to calculate running velocity at LTAn using individual VO2max and VLamax and an adapted mathematical method initially described by Mader and Heck [3] and Hauser et al. [22]. The calculated LTAn (cLTAn) was then compared with other established experimentally determined LTAn concepts. Additionally, we aimed to validate cLTAn against the athlete’s recent performance in endurance running events. 2. Materials and Methods 2.1. Subjects Ten sub-elite male middle- and long-distance runners (age = 19.2 ± 3.5 years, body mass = 65.8 ± 5.8 kg, height = 181.7 ± 5.2 cm, VO2max = 69.8 ± 6.7 mL∙kg−1 min−1, VLamax = 0.39 ± 0.09 mmol L−1 s−1) participated in this study. Prior to signing the written informed con- sent of the investigation, all participants were informed about the experimental proce- dures. The protocols used in this investigation were approved by the Ethics Committee of the university and are in line with the Declaration of Helsinki. Figure 1. An exemplary description of the mathematical model for metabolic simulation, presenting the gross lactate formation (VLass) and the maximal lactate elimination rate (VLaoxmax) depending on exercise intensity [22]. Maximal lactate steady state (MLSS) is defined as the exercise intensity at which the lactate formation is exactly equal to elimination. VO2max = maximal oxygen uptake; VLamax = maximal lactate production rate; Ks4 = individual constant value of the relationship between oxygen demand and running velocity. 2. Materials and Methods 2.1. Subjects Ten sub-elite male middle- and long-distance runners (age = 19.2 ± 3.5 years, body mass = 65.8 ± 5.8 kg, height = 181.7 ± 5.2 cm, VO2max = 69.8 ± 6.7 mL·kg−1 min−1, VLamax = 0.39 ± 0.09 mmol L−1 s−1) participated in this study. Prior to signing the written informed consent of the investigation, all participants were informed about the experimental procedures. The protocols used in this investigation were approved by the Ethics Committee of the university and are in line with the Declaration of Helsinki. 2.2. Design The present investigation consisted of two different performance tests completed on a single day. The body mass was measured before the performance testing (Tanita Corp., Tokyo, Japan). Participants were instructed to arrive in the laboratory in a rested, 2 h postprandial, and well-hydrated state. They were ordered to avoid strenuous exercise for at least 24 h before the test. First, the participants performed a 30 s isokinetic sprint test on a cycle ergometer with subsequent measurements of whole-blood lactate concentration for the determination of VLamax. After a 60 min break, a graded exercise test on a treadmill (second test) was performed to determine VO2max and running velocity at the onset of blood lactate accumu- lation (OBLA; 4 mmol·L−1) [4] and at the modified maximal deviation point (mDmax) [24]. The cLTAn was determined according to the calculation scheme described by Mader and Heck [3], as well as by Hauser et al. [22], and subsequently compared with OBLA and mDmax. To evaluate the validity of cLTAn, OBLA, and mDmax as indicators of endurance performance, running velocities at each concept were compared with the participant’s per- formance (average velocity (m·s−1)) over various distances (3000 m, 5000 m, and 10,000 m). Medicina 2021, 57, 1117 4 of 12 One participant did not provide performance data, so only data from nine participants were included in correlation analysis. 2.3. Isokinetic Sprint Test and VLamax Determination (Performance Capacity of Glycolysis) The participants first performed a 10 min standardized warm-up at 1.5 W kg−1 body mass. After an additional passive rest for 5 min, a 30 s sprint test was performed on a cycle ergometer adjusted to an isokinetic mode set to a cadence of 120 rpm [25,26]. Participants were instructed to perform the test in a sitting position and were verbally encouraged throughout the test to achieve and maintain maximal effort. After the sprint, participants took a rest in a sitting position for 10 min. Immediately before sprint testing, as well as every minute after the sprint bout (1′–10′), 20 µL of capillary blood was taken from the earlobe for lactate analysis (Biosen C-line; EKF Diagnostic Sales, Magdeburg, Germany). The VLamax was calculated using the following equation [27]: VLamax (mmol L−1 s−1) = ([La]peak − [La]rest) · (texerc − talac)−1 (1) where Lapeak (mmol L−1) is the peak post-exercise lactate concentration, Larest (mmol L−1) is the resting lactate concentration, texerc (s) is the duration of exercise, and talac (s) is the period at the beginning of exercise in which no lactate formation is assumed. According to Heck et al. [27], talac was set to 5.5 s for all participants. 2.4. Graded Exercise Running Test and VO2max and LTAn Determination The graded exercise test was performed on a treadmill (Woodway, Weil am Rhein, Germany), which started at 2.4 m s−1 and increased by 0.4 m s−1 every 5 min until volitional exhaustion was reached. After each step of the graded exercise test, a 30 s rest was given for blood sampling. Furthermore, heart rate (HR) (H7, Polar Electro Oy, Kempele, Finland) and breath-by-breath expired gases (Cortex Metalyzer II, Leipzig, Germany) were continuously measured throughout the test. The VO2max corresponded to the highest value measured (moving average of 30 s) during the test. Blood lactate concentrations during the incremental tests were plotted against running velocity and then fitted by a third-order polynomial function. Running velocity at OBLA was set as the point at which blood lactate concentration reached 4 mmol·L−1 [4]. mDmax was identified as the point on the third-order polynomial curve that yielded the maximal perpendicular distance to a straight line formed by the peak lactate point, and by the point of the first rise in blood lactate concentration at which the slope of the fitted lactate curve was equal to 1.00 [24]. 2.5. Calculation of Running Velocity at cLTAn To determine cLTAn, the oxidative and glycolytic energy production depending on exercise intensity must initially be known, which can be expressed as the activity of oxidative phosphorylation (VO2ss) and glycolysis (VLass), respectively [3]. The theoretical background of the applied equations and constants is explained in detail by previous publications [3,22]. According to Mader and Heck [3], the implementation of the metabolic simulation model requires knowing the free ADP concentration, which is the main regulating substrate for the activation of VO2ss and VLass. Since there is no simple and practical procedure for measuring free ADP concentration, the ADP-dependent equations in the previous study were transposed into VO2ss-dependent equations [22]. On this occasion, the term “VO2ss” represents the steady-state oxygen consumption at a constant work rate [3,22]. Hauser et al. [22] calculated the VO2ss in relation to exercise intensity based on the assump- tion of a linear relationship between oxygen demand (VO2) and workload. Thereby, a constant value for VO2 per 1 W (Ks4 = 11.7 mL O2 W−1) was used for all participants based on the data of previous cycling experiments [3,28]. However, it should be noted that VO2 in running is more affected by an athlete’s exercise economy (i.e., metabolic cost at a given workload) than in cycling. Running economy was shown to be influenced by several physi- Medicina 2021, 57, 1117 5 of 12 ological and biomechanical factors [29], which can lead to greater inter-individual variation in comparison to the cycling economy due to weight-bearing activity [30]. Therefore, it is necessary to determine Ks4 (mL kg−1·min−1 per 1 m·s−1 running velocity) individually, by plotting VO2 during incremental tests against running velocity. The Ks4 corresponded to the slope of linear regression (y = mx + b) between VO2 and running speed (Figure 2). study were transposed into VO2ss dependent equations [22]. On this occasion, the term “VO2ss” represents the steady-state oxygen consumption at a constant work rate [3,22]. Hauser et al. [22] calculated the VO2ss in relation to exercise intensity based on the assump- tion of a linear relationship between oxygen demand (VO2) and workload. Thereby, a con- stant value for VO2 per 1 W (Ks4 = 11.7 mL O2 W−1) was used for all participants based on the data of previous cycling experiments [3,28]. However, it should be noted that VO2 in running is more affected by an athlete’s exercise economy (i.e., metabolic cost at a given workload) than in cycling. Running economy was shown to be influenced by several phys- iological and biomechanical factors [29], which can lead to greater inter-individual varia- tion in comparison to the cycling economy due to weight-bearing activity [30]. Therefore, it is necessary to determine Ks4 (mL kg−1∙min−1 per 1 m∙s−1 running velocity) individually, by plotting VO2 during incremental tests against running velocity. The Ks4 corresponded to the slope of linear regression (y = mx + b) between VO2 and running speed (Figure 2). Figure 2. An exemplary description of the determination of the individual Ks4 (constant value of the relationship between oxygen demand and running velocity). The slope of the regression line corresponds to Ks4. From this equation, Ks4 for this runner is 12.1 mL∙kg−1∙min−1 per 1 m∙s−1. After determining the individual Ks4, the VO2ss in relation to running velocity was calculated with Equation (2). VO2ss (mL kg−1 min−1) = V Ks4 + VO2rest (2) where v (m s−1) is the running velocity, VO2rest (mL kg−1 min−1) is the resting oxygen uptake, and Ks4 is the constant value of the relationship between oxygen demand and the running velocity (i.e., mL kg−1 per 1 m s−1 running velocity). Figure 2. An exemplary description of the determination of the individual Ks4 (constant value of the relationship between oxygen demand and running velocity). The slope of the regression line corresponds to Ks4. From this equation, Ks4 for this runner is 12.1 mL·kg−1·min−1 per 1 m·s−1. After determining the individual Ks4, the VO2ss in relation to running velocity was calculated with Equation (2). VO2ss (mL kg−1 min−1) = V Ks4 + VO2rest (2) where v (m s−1) is the running velocity, VO2rest (mL kg−1 min−1) is the resting oxygen uptake, and Ks4 is the constant value of the relationship between oxygen demand and the running velocity (i.e., mL kg−1 per 1 m s−1 running velocity). By knowing VO2ss (from resting level to VO2max), it is possible to calculate VLass (lactate formation) as a function of VO2ss, as demonstrated in the following equation: VLass (mmol L−1 min−1) = 60 · . VLamax 1 + ( Ks2 s Ks1· . VO2ss . VO2max− . VO2ss 3 ) (3) where VLamax (mmol L−1·s−1) is the maximal glycolytic rate, VO2max (mL kg−1 min−1) is the maximal oxygen uptake, VO2ss (mL kg−1 min−1) is the steady-state oxygen con- sumption, and Ks1 and Ks2 are the 50% activity rate constant of oxidative phosphorylation (0.0631) and glycolysis (1.331), respectively [22]. Furthermore, the maximal lactate elimination rate (VLaoxmax) which depends on VO2ss can also be calculated based on the experimentally estimated value of lactate equivalent (i.e., the amount of oxidized lactate per unit O2), lactate distribution volume [3], and using the following equation: VLaoxmax  mmol L−1 min−1 = lactate-equivalent lactate distribution volume ·VO2ss = 0.02049 0.4 ·VO2ss (4) where VLaoxmax (mmol L−1 min−1) is the maximal lactate elimination rate as a function of the steady-state oxygen consumption (VO2ss; mL kg−1 min−1) [22]. Medicina 2021, 57, 1117 6 of 12 According to Hauser et al. [22], lactate-equivalent and lactate distribution volume were set to 0.02049 mmol lactate per 1 mL O2 and 0.4 L H2O per kg body weight, respectively. Thus, simulating the simultaneous lactate formation and elimination depending on the metabolic rate or running speed can be carried out based on the individual VO2max and VLamax value, as well as body weight. cLTAn is defined as the running velocity at which the lactate formation is exactly equal to elimination (i.e., VLass = VLaoxmax). 2.6. Statistical Analysis For statistical analysis of the data, the software IBM SPSS version 24 (Chicago, IL, USA) was used. Descriptive statistics of the data are presented as means ± standard devi- ation (±SD). The normal distribution and the variance homogeneity were verified using the Shapiro–Wilk test and Mauchly test of sphericity, respectively. Statistically relevant differences between the three LTAn concepts were determined using one-way repeated measure ANOVA with Bonferroni correction for post hoc tests. Statistical differences were considered to be significant for p ≤ 0.05. To estimate the practical relevance, effect sizes (partial eta squared, ηp2) were calculated for the main effect. According to Cohen [31], a ηp2 ≥ 0.01 indicates small effects, ≥0.059 medium effects, and ≥0.138 large effects. To display the concordance between the LTAn concepts, Bland–Altman plots were constructed. Furthermore, the intra-class correlation coefficients (ICC) were calculated based on a single-measure two-way mixed-effects model. For evaluating the degree of agreement between cLTAn vs. OBLA or mDmax, the “absolute agreement” type of analysis (ICC (2,1)) was chosen. For the comparison of each LTAn concept vs. 3000 m, 5000 m, or 10,000 m, we chose the “consistency” type of analysis (ICC (3,1)). According to Koo and Li [32], the degree of agreement was interpreted as follows: <0.50 = poor, 0.50–0.75 = moderate, 0.75–0.90 = good, and >0.90 = excellent. Pearson’s correlations were also calculated and interpreted as follows: 0.0–0.3 = negligible, 0.3–0.5 = low, 0.5–0.7 = moderate, 0.7–0.9 = high, and 0.9–1.0 = very high [33]. 3. Results Individual values of maximal metabolic performance tests and individual running velocities at each LTAn concept are presented in Table 1. Table 1. Body mass, maximal oxygen uptake (VO2max), maximal lactate production rate (VLamax), constant value of the relationship between oxygen demand and running velocity (Ks4), and running velocity at the onset of blood lactate accumulation (OBLA), at the modified maximal deviation method (mDmax) and the calculated anaerobic lactate threshold (cLTAn) for each participant. Participant Body Mass (kg) VO2max (mL kg−1·min−1) VLamax (mmol L−1 s−1) Ks4 (mL kg−1·min−1 per 1 m s−1) OBLA (m s−1) mDmax (m s−1) cLTAn (m s−1) 1 59.2 74.6 0.38 12.1 5.19 4.93 4.70 2 64.4 70.0 0.32 12.1 4.97 4.65 4.42 3 64.4 80.3 0.33 13.2 - 5.12 4.87 4 72.6 68.0 0.33 11.8 4.43 4.16 4.38 5 72.3 65.4 0.42 11.3 4.54 4.43 4.20 6 68.5 62.3 0.46 10.7 4.31 4.37 3.99 7 59.1 80.1 0.31 14.2 4.53 4.33 4.53 8 73.2 67.4 0.55 10.8 4.77 4.52 4.31 9 58.4 68.7 0.33 11.0 4.94 4.61 4.72 10 65.4 60.8 0.50 11.0 4.13 3.97 3.66 Mean ± SD 65.8 ± 5.8 69.8 ± 6.7 0.39 ± 0.09 11.8 ± 1.1 4.65 ± 0.35 4.44 ± 0.28 * 4.32 ± 0.34 * * significantly different compared to OBLA (p < 0.01). Repeated measures ANOVA showed a significant difference between LTAn concepts with a large effect (p < 0.01, η2p = 0.63). Post hoc analysis using Bonferroni correction revealed that running velocity at OBLA was significantly higher compared to cLTAn and mDmax (p < 0.01). No significant difference was found between running velocity at cLTAn and mDmax (p = 0.49). The cLTAn indicated a high correlation with OBLA (r = 0.83, Medicina 2021, 57, 1117 7 of 12 r2 = 0.70, p < 0.01) and mDmax (r = 0.81, r2 = 0.65, p < 0.01). Between OBLA and mDmax, there was a very high correlation (r = 0.94, r2 = 0.89, p < 0.001). According to the Bland–Altman Plots (Figure 3), the mean difference in cLTAn was −0.13 ± 0.43 m·s−1 and −0.32 ± 0.39 m·s−1 compared to mDmax and OBLA, respectively. The intraclass correlation coefficient comparing cLTAn with mDmax showed a good agree- ment (ICC = 0.87), whereas a moderate agreement was shown between cLTAn and OBLA (ICC = 0.74). 0 65. 60.8 0.50 .0 . 3 3.9 3.66 Mean ± SD 65.8 ± 5.8 69.8 ± 6.7 0.39 ± 0.09 11.8 ± 1.1 4.65 ± 0.35 4.44 ± 0.28 * 4.32 ± 0.34 * * significantly different compared to OBLA (p < 0.01). Repeated measures ANOVA showed a significant difference between LTAn concepts with a large effect (p < 0.01, η2p = 0.63). Post hoc analysis using Bonferroni correction re- vealed that running velocity at OBLA was significantly higher compared to cLTAn and mDmax (p < 0.01). No significant difference was found between running velocity at cLTAn and mDmax (p = 0.49). The cLTAn indicated a high correlation with OBLA (r = 0.83, r2 = 0.70, p < 0.01) and mDmax (r = 0.81, r2 = 0.65, p < 0.01). Between OBLA and mDmax, there was a very high correlation (r = 0.94, r2 = 0.89, p < 0.001). According to the Bland–Altman Plots (Figure 3), the mean difference in cLTAn was −0.13 ± 0.43 m∙s−1 and −0.32 ± 0.39 m∙s−1 compared to mDmax and OBLA, respectively. The intraclass correlation coefficient comparing cLTAn with mDmax showed a good agreement (ICC = 0.87), whereas a moderate agreement was shown between cLTAn and OBLA (ICC = 0.74). (a) (b) Figure 3. Bland–Altman Plots: differences in running velocity at calculated anaerobic lactate thresh- old (cLTAn) vs. modified maximal deviation method (mDmax; (a)) and onset of blood lactate accu- mulation (OBLA; (b)). The solid lines indicate the mean difference; the dotted lines indicate the limits of agreement (mean ± 1.96 SD); the dashed lines represent the fitted linear regression. The mean running velocities over the distances of 3,000 m, 5,000 m, and 10,000 m were 5.65 ± 0.29 m s−1, 5.37 ± 0.26 m s−1, and 5.03 ± 0.26 m s−1, respectively. cLTAn and mDmax indicated moderate to high correlations with running performance over all Figure 3. Bland–Altman Plots: differences in running velocity at calculated anaerobic lactate threshold (cLTAn) vs. modified maximal deviation method (mDmax; (a)) and onset of blood lactate accumulation (OBLA; (b)). The solid lines indicate the mean difference; the dotted lines indicate the limits of agreement (mean ± 1.96 SD); the dashed lines represent the fitted linear regression. The mean running velocities over the distances of 3000 m, 5000 m, and 10,000 m were 5.65 ± 0.29 m s−1, 5.37 ± 0.26 m s−1, and 5.03 ± 0.26 m s−1, respectively. cLTAn and mDmax indicated moderate to high correlations with running performance over all distances observed (cLTAn: 0.61 < r < 0.76, 0.37 < r2 < 0.58, p < 0.05; mDmax: 0.69 < r < 0.79, 0.48 < r2 < 0.62, p < 0.05), whereby OBLA had the poorest correlations (0.56 < r < 0.69, 0.32 < r2 < 0.48, p ≤ 0.09) compared to other concepts in most cases (Figure 4a). The intraclass correlation (Figure 4b) also revealed good concordance of cLTAn (ICC = 0.75–0.86) and mDmax (ICC = 0.82–0.88) with running performance over all distances observed, whereas OBLA showed only moderate concordance (ICC = 0.68–0.80) in most cases. (a) (b) Figure 4. Correlations (a) and intraclass correlation coefficients (b) of the running velocity at onset of blood lactate accumulation (OBLA), modified maximal deviation method (mDmax), and calculated anaerobic lactate threshold (cLTAn) compared to average running velocity over 3000 m (v3000), 5000 m (v5000), and 10,000 m (v10,000); * p < 0.05. Medicina 2021, 57, 1117 8 of 12 4. Discussion The purpose of this study was to determine cLTAn in running by adapting the mathe- matical model for metabolic simulation previously described by Mader and Heck [3] and Hauser et al. [22]. cLTAn demonstrated moderate to good concordance with the established concepts in determining the running velocity at LTAn. Although cLTAn provided lower running velocity compared to mDmax and OBLA, the correlation of cLTAn with the en- durance running performance was similar compared to mDmax and even better compared to OBLA. One of the relevant criteria for the practical application of a laboratory test parameter is its relationship with competitive performance. A comprehensive review by Faude et al. [8] demonstrated moderate to high correlations (r = 0.66–0.92) between various LTAn concepts and performance in endurance running competitions and therefore justified the practical application of those concepts in sports diagnostics. Even though cLTAn did not indicate significantly superior results, its good concordance (ICC = 0.75–0.86) with mDmax and OBLA, as well as comparable correlations (r = 0.61–0.76) with competition performance, can support its applicability as a valid indicator to assess an athlete’s endurance performance. The metabolic simulation model (cLTAn) incorporates the influence of individual VO2max, VLamax, and Ks4 on LTAn, as well as their combined effects [18,22]. This could enable a more differentiated approach in the interpretation of the endurance performance of an athlete. The individually determined Ks4 values are dependent on individual exercise economy, expressed by the relationship between energy demand and running velocity [29]. Especially in well-trained athletes with similar VO2max, running economy has been shown to be a crucial indicator of distance running performance [29,34,35]. The consideration of individual physiological profiles allows specific explanations of how equal and/or differ- ent endurance performance can be achieved regarding the interplay of single metabolic parameters [18]. For instance, participants 3 and 7 in our study showed similar aero- bic and anaerobic capacities (VO2max: 80.3 vs. 80.1 mL kg−1·min−1, VLamax: 0.33 vs. 0.31 mmol L−1 s−1); however, participant 3 displayed a much higher speed at LTAn regard- less of the used LTAn concept (Table 1) and, consequently, better performance compared to participant 7 (e.g., 10,000 m running time: 30 vs. 32 min). In this case, the performance dif- ferences could be explained by much lower Ks4 (13.2 vs. 14.2 mL kg−1 min−1 per 1 m s−1). A recent training study used the metabolic simulation-model-detected training-induced changes in single performance capacities (i.e., VO2max and VLamax). The authors reported specific explanations of changes in endurance performance (MLSS) [36], which highlights the potential for the practical application of the model. Despite the moderate to good agreement with other LTAn concepts, cLTAn systemat- ically provides lower running velocities in our study (Figure 3). This discrepancy could be attributed to the underrated VO2max by using a graded exercise test. The main reason we used a graded incremental protocol, instead of a ramp protocol, was to concurrently determine OBLA and mDmax, as well as the relationship between steady-state oxygen demand and running velocity (i.e., individual Ks4). However, the mean time to exhaustion of our test protocol was ~38 min, which is significantly longer than the “optimal” test duration for assessing VO2max, as suggested by previous studies [37–39]. Sperlich et al. [40] reported that VO2max, achieved with the same graded exercise test protocol as in our study, was significantly lower (on average 2 mL min−1 kg−1) than assessed by incremental tests with shorter test duration (ranged from 7–11 min). Hauser [28] showed that a theoretical 25% increase in VO2max (and constant VLamax, and Ks4) leads to a 44% increase in calcu- lated MLSS in cycling. Indeed, cLTAn is increased by ~0.2 m s−1 when the participant’s VO2max is increased by 2 mL min−1 kg−1 (and constant values of VLamax, and Ks4), and thus the difference between running speed at cLTAn and the other LTAn is reduced (data not presented). To solve the underestimation of VO2max, further work should use a VO2max verification bout [41,42] or a combined step- and ramp-exercise protocol [43]. Such proto- cols could ensure the appropriate determination of VO2max and the individual Ks4 at the same time, as two core parameters of the metabolic simulation model. Medicina 2021, 57, 1117 9 of 12 Another potential contributing factor to the difference between cLTAn and other LTAn concepts could be the run-nonspecific test procedure for the assessment of VLamax and its influence on cLTAn. The cycling sprint test is an established anaerobic test for nearly all sports disciplines. Thus, we determined the participant’s VLamax using an isokinetic cycle sprint [22,23,36,44]. However, the peak post-exercise lactate concentration, which is a key parameter for the estimation of VLamax, is dependent on the exercise modality used in tests [44]. Unfortunately, up to now no established running-specific test procedure for VLamax determination exists. Just recently, Quittmann et al. [45] attempted to measure VLamax and sprint performance parameters using a running sprint test. However, this study used fixed distances, rather than a fixed time for the sprint test, which might influence VLamax determination. Whether and how VLamax estimation and cLTAn determination would be affected by applying a running-specific anaerobic test procedure remain to be clarified. Since VO2ss contributes as a core parameter to the calculation of both the lactate formation and elimination rate at any given running velocity, it is necessary to determine VO2ss (from resting level to VO2max) as precisely as possible. For the determination of VO2ss, the relationship between oxygen demand and running velocity (Ks4) plays an important role [3]. In contrast to the previous study in cycling [22], we individually determined the Ks4 value considering the inter-individual variation in the running economy. Typically, it is assumed that there is a linear relationship between VO2 and workload. This has been supported by several investigations indicating a nearly invariant oxygen cost of transport (calculated by dividing oxygen uptake by running velocity, mL kg−1 km−1) over a range of running speeds (2.0–4.0 m s−1) [46,47]. However, these studies investigated the individual running energetics only from the start of exercise until LTAn intensity and not till exhaustion. Daniels and Daniels [48] suggested that the metabolic demand of running is not exclusively dependent on running speed and can vary with an athlete’s specialized background. They found that most of the 800–1500 m specialists in their study showed an equal oxygen cost of transport over all intensities examined. In contrast to that, the specialists in longer distances (3000 m—marathon) mostly showed an increased oxygen cost of transport at exercise intensities above 70% of VO2max [48]. These findings emphasize the importance of considering the individual running energetics over all possible test speeds to assess the performance difference between athletes. To what extent the running energetics, especially near the LTAn intensity, differ between athletes, and how they affect the LTAn, is unclear. With respect to the previous model in cycling [22], we, therefore, decided to use the Ks4 from a linear fit to calculate VO2ss in our study. However, there is abundant space for further progress in analyzing the relationship between metabolic rate and running velocity and its influence on cLTAn determination. For instance, a curvilinear fit suggested by Batliner et al. [49] might better assess the inter-individual difference in running energetics, especially around and above the LTAn intensity, which might consequently lead to an improved performance prediction of cLTAn. In addition to the above methodological limitations, it is important to note that our data did not address the basic variability and reproducibility of each physiological measure (VO2max, VLamax, and Ks4), which are also relevant quality criteria for the application of the cLTAn. However, previous research in cycling already demonstrated a very high reliability for both VO2max and VLamax, as well as the calculated MLSS from these two parameters [23]. Further studies with a longitudinal analysis in running should be carried out to investigate the reliability and sensitivity of the single performance tests and metabolic simulation model for detecting performance changes. 5. Practical Applications The present study suggests that the mathematical model for metabolic simulation could be applied to assess an athlete’s endurance performance in running by considering multiple physiological parameters. Considering multiple physiological measures, the metabolic simulation model (cLTAn) provides an insight into the complex interplay of Medicina 2021, 57, 1117 10 of 12 single metabolic systems and their influence on endurance performance. This allows a differentiated interpretation of the athlete’s performance, which could be useful for establishing training interventions targeting and eliminating specific weaknesses in the physiological profile of an athlete. 6. Conclusions The metabolic simulation model considers different metabolic parameters to evaluate an athlete’s performance profile. In determining running velocity at LTAn, the metabolic simulation model (cLTAn) revealed a moderate to good agreement with other established concepts. However, the velocity at cLTAn was lower with regard to the other LTAn concepts. With regard to the compared LTAn concepts, comparable and partially better correlations between cLTan and the endurance performance of sub-elite middle- and long-distance runners were found. Author Contributions: Conceptualization, P.W.; methodology, S.J., A.S. and P.W.; formal analysis, S.J.; investigation, S.J., A.S. and P.W.; resources, P.W. and W.B.; data curation, S.J., A.S. and P.W.; writing—original draft preparation, S.J.; writing—review and editing, S.J., A.S. and P.W.; visualization, S.J.; supervision, P.W. and W.B. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethics Committee of German Sport University Cologne (approval code: 146/2021; approval date: 4 October 2021). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: The data presented in this study are available on request from the corresponding author. Conflicts of Interest: The authors declare no conflict of interest. References 1. Billat, V.L. Use of Blood Lactate Measurements for Prediction of Exercise Performance and for Control of Training. Recommenda- tions for Long-Distance Running. Sports Med. 1996, 22, 157–175. [CrossRef] 2. Jacobs, I. Blood Lactate. Sports Med. 1986, 3, 10–25. [CrossRef] 3. Mader, A.; Heck, H. A Theory of the Metabolic Origin of “Anaerobic Threshold”. Int. J. Sports Med. 1986, 7, 45–65. [CrossRef] 4. Heck, H.; Mader, A.; Hess, G.; Mücke, S.; Müller, R.; Hollmann, W. Justification of the 4-Mmol/l Lactate Threshold. Int. J. Sports Med. 1985, 6, 117–130. [CrossRef] [PubMed] 5. Brooks, G.A. Lactate glycolytic end product and oxidative substrate during sustained exercise in mammals—The “lactate shuttle”. In Circulation, Respiration, and Metabolism; Springer: Berlin/Heidelberg, Germany, 1985; pp. 208–218. 6. Messonnier, L.A.; Emhoff, C.A.W.; Fattor, J.A.; Horning, M.A.; Carlson, T.J.; Brooks, G.A. Lactate Kinetics at the Lactate Threshold in Trained and Untrained Men. J. Appl. Physiol. 2013, 114, 1593–1602. [CrossRef] [PubMed] 7. Hering, G.O.; Hennig, E.M.; Riehle, H.J.; Stepan, J. A Lactate Kinetics Method for Assessing the Maximal Lactate Steady State Workload. Front. Physiol. 2018, 9, 1–11. [CrossRef] [PubMed] 8. Faude, O.; Kindermann, W.; Meyer, T. Lactate Threshold Concepts: How Valid Are They? Sports Med. 2009, 39, 469–490. [CrossRef] 9. Svedahl, K.; MacIntosh, B.R. Anaerobic Threshold: The Concept and Methods of Measurement. Can. J. Appl. Physiol. 2003, 28, 299–323. [CrossRef] 10. Kindermann, W.; Simon, G.; Keul, J. The Significance of the Aerobic-Anaerobic Transition for the Determination of Work Load Intensities during Endurance Training. Eur. J. Appl. Physiol. Occup. Physiol. 1979, 42, 25–34. [CrossRef] 11. Cheng, B.; Kuipers, H.; Snyder, A.C.; Keizer, H.A.; Jeukendrup, A.; Hesselink, M. A New Approach for the Determination of Ventilatory and Lactate Thresholds. Int. J. Sports Med. 1992, 13, 518–522. [CrossRef] 12. Bishop, D.; Jenkins, D.G.; Mackinnon, L.T. The Relationship between Plasma Lactate Parameters, Wpeak and 1-h Cycling Performance in Women. Med. Sci. Sports Exerc. 1998, 30, 1270–1275. [CrossRef] [PubMed] 13. Riboli, A.; Rampichini, S.; Cè, E.; Limonta, E.; Coratella, G.; Esposito, F. Effect of Ramp Slope on Different Methods to Determine Lactate Threshold in Semi-Professional Soccer Players. Res. Sports Med. 2019, 27, 326–338. [CrossRef] [PubMed] Medicina 2021, 57, 1117 11 of 12 14. Jamnick, N.A.; Botella, J.; Pyne, D.B.; Bishop, D.J. Manipulating Graded Exercise Test Variables Affects the Validity of the Lactate Threshold and VO2peak. PLoS ONE 2018, 13, e0199794. [CrossRef] [PubMed] 15. Riboli, A.; Rampichini, S.; Cè, E.; Limonta, E.; Borrelli, M.; Coratella, G.; Esposito, F. Training Status Affects between-Protocols Differences in the Assessment of Maximal Aerobic Velocity. Eur. J. Appl. Physiol. 2021, 121, 3083–3093. [CrossRef] [PubMed] 16. Gladden, L.B. Lactate Metabolism: A New Paradigm for the Third Millennium. J. Physiol. 2004, 558, 5–30. [CrossRef] 17. Brooks, G.A. The Science and Translation of Lactate Shuttle Theory. Cell Metab. 2018, 27, 757–785. [CrossRef] [PubMed] 18. Bleicher, A.; Mader, A.; Mester, J. Zur Interpretation von Laktatleistungskurven—Experimentelle Ergebnisse Mit Comput- ergestützten Nachber. Spectr. Sportwiss. 1998, 10, 92–104. 19. Olbrecht, J. Lactate production and metabolism in swimming. In World Book of Swimming. From Science to Performance; Nova: New York, NY, USA, 2011; pp. 255–276. 20. Mader, A. Eine Theorie zur Berechnung der Dynamik und des Steady State von Phosphorylierungszustand und Stoffwechselaktivität der Muskelzelle Als Folge des Energiebedarfs; Deutsche Sporthochschule Köln: Köln, Germany, 1984. 21. Stainsby, W.N.; Brooks, G.A. Control of Lactic Acid Metabolism in Contracting Muscles and during Exercise. Exerc. Sport Sci. Rev. 1990, 18, 29–64. [CrossRef] 22. Hauser, T.; Adam, J.; Schulz, H. Comparison of Calculated and Experimental Power in Maximal Lactate-Steady State during Cycling. Theor. Biol. Med. Model. 2014, 11, 1–12. [CrossRef] 23. Adam, J.; Öhmichen, M.; Öhmichen, E.; Rother, J.; Müller, U.M.; Hauser, T.; Schulz, H. Reliability of the Calculated Maximal Lactate Steady State in Amateur Cyclists. Biol. Sport 2015, 32, 97–102. [CrossRef] 24. Zwingmann, L.; Strütt, S.; Martin, A.; Volmary, P.; Bloch, W.; Wahl, P. Modifications of the Dmax Method in Comparison to the Maximal Lactate Steady State in Young Male Athletes. Physician Sportsmed. 2019, 47, 174–181. [CrossRef] [PubMed] 25. Faria, E.W.; Parker, D.L.; Faria, I.E. The Science of Cycling Physiology and Training-Part 1. Sports Med. 2005, 35, 285–312. [CrossRef] [PubMed] 26. Wahl, P.; Zinner, C.; Grosskopf, C.; Rossmann, R.; Bloch, W.; Mester, J. Passive Recovery Is Superior to Active Recovery during a High-Intensity Shock Microcycle. J. Strength Cond. Res. 2013, 27, 1384–1393. [CrossRef] [PubMed] 27. Heck, H.; Schulz, H.; Bartmus, U. Diagnostics of Anaerobic Power and Capacity. Eur. J. Sport Sci. 2003, 3, 1–23. [CrossRef] 28. Hauser, T. Untersuchungen zur Validität und Praktikabilität des Mathematisch Bestimmten Maximalen Laktat-Steady-States Bei Rader- gometrischen Belastungen; Chemnitz University of Technology: Chemnitz, Germany, 2013. 29. Saunders, P.U.; Pyne, D.B.; Telford, R.D.; Hawley, J.A. Factors Affecting Running Economy in Trained Distance Runners. Sports Med. 2004, 34, 465–485. [CrossRef] [PubMed] 30. Swinnen, W.; Kipp, S.; Kram, R. Comparison of Running and Cycling Economy in Runners, Cyclists, and Triathletes. Eur. J. Appl. Physiol. 2018, 118, 1331–1338. [CrossRef] 31. Cohen, J. Statistical Power Analysis for the Behavioral Sciences; Taylor and Francis: Hoboken, NJ, USA, 1988. 32. Koo, T.K.; Li, M.Y. A Guideline of Selecting and Reporting Intraclass Correlation Coefficients for Reliability Research. J. Chiropr. Med. 2016, 15, 155–163. [CrossRef] 33. Mukaka, M.M. A Guide to Appropriate Use of Correlation Coefficient in Medical Research. Malawi Med. J. 2012, 24, 69–71. 34. Barnes, K.R.; Kilding, A.E. Running Economy: Measurement, Norms, and Determining Factors. Sports Med.-Open 2015, 1, 1–15. [CrossRef] 35. Morgan, D.; Baldini, F.; Martin, P.; Kohrt, W. Ten Kilometer Performance and Predicted Velocity at VO2max among Well-Trained Male Runners. Med. Sci. Sports Exerc. 1989, 21, 78–83. [CrossRef] 36. Hommel, J.; Öhmichen, S.; Rudolph, U.M.; Hauser, T.; Schulz, H. Effects of Six-Week Sprint Interval or Endurance Training on Calculated Power in Maximal Lactate Steady State. Biol. Sport 2019, 36, 47–54. [CrossRef] [PubMed] 37. Buchfuhrer, M.J.; Hansen, J.E.; Robinson, T.E.; Sue, D.Y.; Wasserman, K.; Whipp, B.J. Optimizing the Exercise Protocol for Cardiopulmonary Assessment. J. Appl. Physiol. Respir. Environ. Exerc. Physiol. 1983, 55, 1558–1564. [CrossRef] 38. González-Alonso, J.; Calbet, J.A.L. Reductions in Systemic and Skeletal Muscle Blood Flow and Oxygen Delivery Limit Maximal Aerobic Capacity in Humans. Circulation 2003, 107, 824–830. [CrossRef] [PubMed] 39. Bentley, D.J.; Newell, J.; Bishop, D. Incremental Exercise Test Design and Analysis: Implications for Performance Diagnostics in Endurance Athletes. Sports Med. 2007, 37, 575–586. [CrossRef] [PubMed] 40. Sperlich, P.F.; Holmberg, H.-C.; Reed, J.L.; Zinner, C.; Mester, J.; Sperlich, B. Individual versus Standardized Running Protocols in the Determination of VO2max. J. Sports Sci. Med. 2015, 14, 386. [PubMed] 41. Poole, D.C.; Jones, A.M. Measurement of the Maximum Oxygen Uptake VO2max: VO2peak Is No Longer Acceptable. J. Appl. Physiol. 2017, 122, 997–1002. [CrossRef] 42. Schaun, G.Z. The Maximal Oxygen Uptake Verification Phase: A Light at the End of the Tunnel? Sports Med.-Open 2017, 3, 44. [CrossRef] 43. Krüger, M.; de Marees, M.; Dittmar, K.-H.; Sperlich, B.; Mester, J. Whole-Body Cryotherapy’s Enhancement of Acute Recovery of Running Performance in Well-Trained Athletes. Int. J. Sports Physiol. Perform. 2015, 10, 605–612. [CrossRef] 44. Nitzsche, N.; Baumgärtel, L.; Schulz, H. Comparison of Maximum Lactate Formation Rates in Ergometer Sprint and Maximum Strength Loads. Dtsch. Z. Sportmed. 2018, 69, 13–17. [CrossRef] 45. Quittmann, O.J.; Appelhans, D.; Abel, T.; Strüder, H.K. Evaluation of a Sport-Specific Field Test to Determine Maximal Lactate Accumulation Rate and Sprint Performance Parameters in Running. J. Sci. Med. Sport 2020, 23, 27–34. [CrossRef] Medicina 2021, 57, 1117 12 of 12 46. Helgerud, J.; Støren, Ø.; Hoff, J. Are There Differences in Running Economy at Different Velocities for Well-Trained Distance Runners? Eur. J. Appl. Physiol. 2010, 108, 1099–1105. [CrossRef] [PubMed] 47. Helgerud, J. Maximal Oxygen Uptake, Anaerobic Threshold and Running Economy in Women and Men with Similar Perfor- mances Level in Marathons. Eur. J. Appl. Physiol. Occup. Physiol. 1994, 68, 155–161. [CrossRef] [PubMed] 48. Daniels, J.; Daniels, N. Running Economy of Elite Male and Elite Female Runners. Med. Sci. Sports Exerc. 1992, 24, 483–489. [CrossRef] [PubMed] 49. Batliner, M.; Kipp, S.; Grabowski, A.; Kram, R.; Byrnes, W. Does Metabolic Rate Increase Linearly with Running Speed in All Distance Runners? Sports Med. Int. Open 2018, 2, E1–E8. [CrossRef] [PubMed]

Comparison and Performance Validation of Calculated and Established Anaerobic Lactate Thresholds in Running.

10-16-2021

Ji, Sanghyeon,Sommer, Aldo,Bloch, Wilhelm,Wahl, Patrick

eng

PMC4619465

RESEARCH ARTICLE Footwear Decreases Gait Asymmetry during Running Stefan Hoerzer1*, Peter A. Federolf2,3, Christian Maurer1,4, Jennifer Baltich1, Benno M. Nigg1 1 Human Performance Laboratory, Faculty of Kinesiology, University of Calgary, Calgary, Alberta, Canada, 2 Institute for Sport Science, University of Innsbruck, Innsbruck, Tyrol, Austria, 3 Department of Neuroscience, Norwegian University of Science and Technology, Trondheim, Norway, 4 Red Bull Diagnostic and Training Center, Thalgau, Salzburg, Austria * stefan.hoerzer@ucalgary.ca Abstract Previous research on elderly people has suggested that footwear may improve neuromus- cular control of motion. If footwear does in fact improve neuromuscular control, then such an influence might already be present in young, healthy adults. A feature that is often used to assess neuromuscular control of motion is the level of gait asymmetry. The objectives of the study were (a) to develop a comprehensive asymmetry index (CAI) that is capable of detecting gait asymmetry changes caused by external boundary conditions such as foot- wear, and (b) to use the CAI to investigate whether footwear influences gait asymmetry dur- ing running in a healthy, young cohort. Kinematic and kinetic data were collected for both legs of 15 subjects performing five barefoot and five shod over-ground running trials. Thirty continuous gait variables including ground reaction forces and variables of the hip, knee, and ankle joints were computed for each leg. For each individual, the differences between the variables for the right and left leg were calculated. Using this data, a principal compo- nent analysis was conducted to obtain the CAI. This study had two main outcomes. First, a sensitivity analysis suggested that the CAI had an improved sensitivity for detecting changes in gait asymmetry caused by external boundary conditions. The CAI may, there- fore, have important clinical applications such as monitoring the progress of neuromuscular diseases (e.g. stroke or cerebral palsy). Second, the mean CAI for shod running (131.2 ± 48.5; mean ± standard deviation) was significantly lower (p = 0.041) than the CAI for bare- foot running (155.7 ± 39.5). This finding suggests that in healthy, young adults gait asymme- try is reduced when running in shoes compared to running barefoot, which may be a result of improved neuromuscular control caused by changes in the afferent sensory feedback. Introduction Falls are one of the main causes for fatal injury and hospitalization in older adults [1–3]. Identi- fying factors that contribute to falls has become an important objective in clinical geriatric research. The absence of footwear was identified as an important risk factor for the occurrence of falls in elderly adults [4]. The reduced risk of falls reported in the mentioned study concurs PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 1 / 12 OPEN ACCESS Citation: Hoerzer S, Federolf PA, Maurer C, Baltich J, Nigg BM (2015) Footwear Decreases Gait Asymmetry during Running. PLoS ONE 10(10): e0138631. doi:10.1371/journal.pone.0138631 Editor: Jose Manuel Garcia Aznar, University of Zaragoza, SPAIN Received: May 29, 2015 Accepted: August 31, 2015 Published: October 21, 2015 Copyright: © 2015 Hoerzer et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: The authors would like to acknowledge NSERC Create, Biomechanigg Sport & Health Research, and the da Vinci Foundation for financial support. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. In addition, the Red Bull Diagnostic and Training Center provided support in the form of a salary for CM, but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of this author are articulated in the 'author contributions' section. with other studies that assessed the effect of footwear on the likelihood of falls or balance [5–7]. In addition to mechanical factors potentially causing a reduced risk of falls when wearing foot- wear [8], it is also possible that footwear may alter the type or amount of afferent sensory feed- back causing improved neuromuscular control. If footwear does in fact improve neuromuscular control, then such an influence might already be present in young, healthy adults, long before it may become clinically relevant in the prevention of falls. A feature that is often used to assess neuromuscular control of motion is the level of asymmetry between the contra-lateral limbs during gait. In fact, in many neurophysiological disorders such as stroke [9, 10], Parkinson’s disease [11], or cerebral palsy [12], gait asymmetry can be seen as one of the indicators of the severity of the condition. One challenge when assessing gait asymmetry in healthy, young adults is that the kinematic and kinetic differences between the left and right lower limbs are rather small compared to the inherent movement variability. In addition, one could argue that gait asymmetry is a character- istic that applies to several body segments simultaneously [13–15], especially when investigat- ing changes caused by external boundary conditions such as footwear. Therefore, a new asymmetry index, a comprehensive asymmetry index (CAI), is required that is especially sensi- tive to changes in gait asymmetry caused by external boundary conditions. Three actions can be taken in order to increase the sensitivity of the CAI: First, all available kinematic and kinetic data should be incorporated to provide an all-encompassing assessment of an individual’s lower limb gait asymmetry. This allows considering the moving human body as a whole system rather than analysing individual variables [16, 17]. Second, the waveforms of all gait variables should be normalized to their standard deviation waveform to account for asymmetry caused by the natural variability of the movement. This should be done since previous studies indi- cated that gait asymmetry may only be relevant when it exceeds the inherent variability of a gait variable [13, 18]. Third, a principal component analysis (PCA) can be used to filter out the covariate structure of gait asymmetry [16, 19]. This is based on the assumption that gait asym- metry observed in one variable can only occur if it is accompanied by asymmetries in other var- iables [19]. To give a simplified example: contra-lateral asymmetries in the knee joint angle can only occur within a given motion task, if ankle and/or hip angles change accordingly. In summary, a CAI with enhanced sensitivity to detect gait asymmetry changes is required in order to investigate whether footwear influences the level of asymmetry between the contra- lateral limbs during gait. A reduction in gait asymmetry may support previous research indicat- ing that footwear improves neuromuscular control. The new CAI should be tested on a highly automated movement, i.e. running, rather than more complex movements in which higher cognitive functions are more likely to interfere with the movement pattern and may potentially affect gait asymmetry. Therefore, the objectives of the study were (a) to develop a comprehensive asymmetry index (CAI) that can be used to study changes in gait asymmetry caused by external boundary condi- tions such as footwear, and (b) to use the CAI to investigate whether footwear influences gait asymmetry during running in a healthy, young cohort. Based on the aforementioned studies, it was hypothesized that footwear decreases gait asymmetry as compared to barefoot running. Methods Study participants Fifteen subjects were recruited for this study, seven females and eight males: age: 25.4 (SD 4.4) years; height: 1.74 (SD 0.07) m; mass: 71.2 (SD 8.4) kg. The subjects were healthy, with no neu- romuscular or neurological disorders, and had no lower-extremity pain at the time of testing. All study participants provided written informed consent in accordance with the University of Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 2 / 12 Competing Interests: The authors of this manuscript have the following competing interests: CM has a commercial affiliation (Red Bull Diagnostic and Training Center). However, this does not alter the authors' adherence to PLOS ONE policies on sharing data and materials. Abbreviations: CAI, Comprehensive Asymmetry Index; EV, Eigenvalue; PCA, Principal Component Analysis; PC-vector, Principal Component Vector. Calgary’s policy on research using human subjects. The study protocol was approved by the Conjoint Health Research Ethics Board of the University of Calgary. Data collection Kinematic and kinetic data were collected while the subjects performed for each leg five bare- foot and five shod heel-toe over-ground running trials (running speed: 4.00 ± 0.6 ms−1). A standard, neutral running shoe, without unique design features that potentially could have influenced gait asymmetry, was provided for each subject (New Balance 506; New Balance Ath- letic Shoe Inc., USA). A running trial was considered successful when the subject’s foot that was being tested landed within the edges of a force platform (Kistler Instrumente AG, Switzer- land). The force platform was used to record ground reaction forces (GRFs) at a sampling rate of 2,400 Hz. At the same time, kinematic data were collected by means of a marker-based motion capture system having eight synchronized, digital, high-speed, infrared cameras (Motion Analysis Corporation, USA). Twenty-two retro-reflective markers were mounted on each study participant. Marker locations included the right and left anterior superior iliac spine, the right and left posterior superior iliac spine, and proximal, lateral, and distal aspects of the thigh and shank. To describe the foot motion, markers were placed at proximal and dis- tal, and lateral locations of the test shoe and on corresponding locations on the bare foot. For the purpose of a neutral standing trial, additional markers were also placed on (and after the neutral trial removed from) the right and left greater trochanters, the medial and lateral knee joint, and the medial and lateral malleoli to define joint centres. A sampling rate of 240 Hz was used to record the trajectories of the markers. Data pre-processing Cortex motion analysis software (Motion Analysis Corporation, USA) was used to reconstruct the trajectories of the 22 markers for each running trial. A fourth-order, low-pass, Butterworth filter was applied to the kinematic and kinetic data to filter out movement artefacts and mea- surement noise with cut-off frequencies of 6 Hz for kinematic data and 50 Hz for kinetic data [20]. Standard motion analysis software (KinTrak 7.0; Human Performance Laboratory, Cal- gary, Canada) was used to compute 30 time-continuous gait variables. The 30 variables included joint angles, joint moments, and joint angular velocities of the ankle, knee, and hip, as well as ground reaction forces in all three planes of motion: frontal, sagittal, and transverse (Table 1). Joint moments and GRFs were normalized to body weight. All variables were resam- pled to 101 time points representing 0 to 100% of the stance phase. Comprehensive asymmetry index The following data-processing steps were conducted for each subject and shoe condition (i.e. barefoot and shod). First, the mean waveform for each of the 30 variables was calculated based Table 1. Gait variables. Segment Variables (frontal, sagittal, and transverse planes) Hip joint Angles [°] Moments [BWm] Angular velocities [°s−1] Knee joint Angles [°] Moments [BWm] Angular velocities [°s−1] Ankle joint Angles [°] Moments [BWm] Angular velocities [°s−1] Centre of pressure Ground reaction forces [BW] Time-continuous gait variables that were computed over the stance phase for each subject, leg, and shoe condition. These variable types were used for the comprehensive asymmetry index. doi:10.1371/journal.pone.0138631.t001 Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 3 / 12 on the five collected trials. Second, the mean waveform for each variable was divided by the average of the corresponding standard deviation waveforms. This was done to normalize the variables to account for asymmetry caused by the natural variability of the movement [13, 18]. Third, all normalized waveforms were vectorized into a 3,030-dimensional (30 variables x 101 time points) row vector, q, by horizontally appending the waveforms. Hence, qleft_leg and qright_leg incorporated all available information about an individual’s movement during the stance phase. Finally, a difference vector, Δq = qright_leg—qleft_leg, between the multi-dimen- sional row vectors of the right and left legs was calculated for each participant and shoe condi- tion. The difference vector Δq quantified all measured aspects of asymmetry of the participants’ gait. Therefore, the vector norm of Δq (i.e. the Euclidean distance from the origin to Δq) may serve as a single CAI of the study participants’ overall gait asymmetry. However, Δq is a complex high-dimensional (3,030 dimensions) construct. It is possible that some components of Δq contain artefacts that appear to indicate asymmetry. These arte- facts are actually the result of random fluctuations of the data due to the natural variability of the movement. The expected gait asymmetry changes within an individual were rather small and the signal-to-noise ratio is unfavourable. Relevant changes in the gait pattern and, there- fore, in gait asymmetry between shoe conditions in one variable have to be interrelated with changes in the asymmetry of other variables [19]. It was speculated that the use of a PCA would allow increasing the sensitivity of the CAI to detect small changes in gait asymmetry. For the PCA, an input matrix M was created containing the difference vector for each individ- ual with each shoe condition: M ¼ Dq1 ... Dq30 2 6664 3 7775 ð1Þ The input matrix contained 3,030 columns (30 variables x 101 time points) and 30 rows (15 subjects x 2 shoe conditions). The PCA comprised the following steps: (1) calculation of the covariance matrix of M; and (2) calculation of the eigenvectors and eigenvalues of the covari- ance matrix [21]. The eigenvectors represent the orthogonal principal component vectors (PC- vectors), p. The PC-vectors are defined by the direction of the highest correlated variance in the data. Since in the current study the input matrix for the PCA contained the difference vec- tors (right-left) for each of the individuals, the variance in the matrix and the definition of the PC-vectors were due to the asymmetry of the individuals’ gait. The eigenvalue (EV) spectrum was assessed to determine a suitable number k of PC-vectors for the definition of the CAI. Within the first 15 EVs a drop is visible between EV8 and EV9 (Fig 1). Therefore, the first eight PC-vectors (k = 8) were expected to provide the best compro- mise between retaining as much correlated asymmetry as possible and filtering out uncorre- lated noise [16]. The difference vectors Δq were then represented in a subspace spanned by the eight selected PC-vectors by projecting each difference vector Δq onto the PC-vectors: Psi ¼ Dqs  pi ð2Þ where s indicates the study participants and i represents the number of the PC-vector. A sub- ject- and condition-specific CAI was then calculated as the Euclidean distance from the origin Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 4 / 12 using the projections (Psi): CAIs ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi X k i¼1 ðPsiÞ 2 v u u t ð3Þ Sensitivity analysis and statistics To assess the sensitivity of the CAI, it was determined whether the difference vectors by them- selves would be able to confirm the hypothesized difference in gait asymmetry between shod and barefoot running and how the CAI depended on the number k of PC-vectors used. There- fore, different variations of the CAI for each individual and shoe condition were calculated: (1) CAIs without PCA, using the vector norm (i.e. Euclidean distance) of the raw Δq only; (2) CAIs with PCA, based on all possible numbers of PC-vectors (k = 1. . .30). A paired samples t- test (p0.05; IBM SPSS Statistics 20, IBM Corporation, USA) was then used to assess the sig- nificance of the difference between the different mean CAIs for barefoot and shod running. Relevant asymmetry variables The relevant asymmetry variables and their correlations were identified by analysing the load- ings of the eight PC-vectors. The loading magnitude indicates the amount of variance in a vari- able that is captured by the corresponding PC-vector [22]. Since this variance was caused by gait asymmetry, variables with higher loadings contributed more to an asymmetrical gait. The loadings were multiplied with their corresponding EVs to weight the loadings according to the amount of variance/asymmetry covered by each PC-vector. Results The eight PC-vectors that were used for the calculation of the CAI contained 76.4% of the over- all asymmetry in all gait variables (Fig 1). The subject-specific CAIs for barefoot running ran- ged from 103.9 to 210.9, whereas the range for shod running was from 48.4 to 212.1 (Fig 2). Fig 1. Eigenvalue spectrum. Eigenvalue spectrum of the first 15 principal component vectors that was used to determine the number of principal component vectors for the definition of the comprehensive asymmetry index (CAI). After the first eight eigenvalues (black bars) a drop can be seen. Hence, the first eight principal component vectors (k = 8) were used for the definition of the CAI. doi:10.1371/journal.pone.0138631.g001 Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 5 / 12 Averaged over all participants the CAI (k = 8) for running barefoot was 155.7 ± 39.5 (mean ± standard deviation) and for running in the shoe condition was 131.2 ± 48.5 (Table 2). The difference between the two conditions was significant (p = 0.041). Comparing barefoot and shod running using the CAI calculated as the direct Euclidean distance of the raw Δq to the origin (i.e. without filtering out uncorrelated asymmetries by the PCA) revealed no signifi- cant difference (p = 0.067; Table 2). The evaluation of how the CAI depended on the number k of PC-vectors used for the definition of the CAI showed that k  3 was not sufficient to detect significant asymmetry differences between barefoot and shod running (Table 2). For 4  k  8 and 12  k  13 the differences between the mean CAIs for barefoot and shod running were significant. The relevant asymmetry variables (i.e. variables with the highest PC-vector loadings) were mainly located in the ankle and knee joint (Fig 3). The frontal knee angle had the highest PC- vector loading (1.73) followed by the frontal ankle moment (1.50) and the frontal ankle angle (1.39). The PC-vector loadings showed correlations particularly between the frontal ankle angle/moment and the frontal knee angle/moment (PC-vector 1, PC-vector 2). Discussion The current study had two main outcomes. First, a novel approach to quantify gait asymmetry was proposed that combined correlated asymmetries in multiple gait variables into one com- prehensive asymmetry index, the CAI. The sensitivity analysis suggested that considering corre- lated asymmetries improves the sensitivity for detecting changes in gait asymmetry caused by external boundary conditions. This would be particularly useful when assessing the progression of clinical conditions such as cerebral palsy or the progress of rehabilitation treatments. The proposed method allowed to examining the structure of gait asymmetry by assessing the indi- vidual loadings of principal component vectors. Again, this has potential for clinical gait analy- sis and may contribute to a better understanding of the specific manifestations of a patient’s underlying condition, for example, in stroke and cerebral palsy patients. Second, the result of the CAI supported the hypothesis that even in healthy, young adults, gait asymmetry is reduced when running in shoes compared to running barefoot. This suggests that footwear seems to Fig 2. Subject-specific comprehensive asymmetry index (CAI) for barefoot and shod running. Study participants are arranged by increasing CAI for barefoot running. All CAIs calculated using eight principal component vectors (k = 8). doi:10.1371/journal.pone.0138631.g002 Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 6 / 12 affect certain aspects of the neuromuscular control system that are involved in the coordination of the movements of left and right lower limbs. Comprehensive asymmetry index The development of the CAI was motivated by the goal to provide a comprehensive asymmetry index with enhanced sensitivity for changes in gait asymmetry. Considering this main goal and the way it was implemented led to advantageous and disadvantageous characteristics of the proposed method, which will be discussed in the following paragraphs. Since the CAI is a single value representing the totality of gait asymmetry of an individual (based on the measured variables), it facilitates direct comparisons between individuals with Table 2. Mean comprehensive asymmetry indexes (CAI) for barefoot and shod running. k Mean CAI Barefoot Mean CAI Shod p-Value Δq 177.7 (SD 33.7) 157.9 (SD 39.1) 0.067 1 61.5 (SD 40.1) 68.6 (SD 48.2) 0.302 2 106.3 (SD 46.4) 84.0 (SD 43.5) 0.084 3 123.2 (SD 43.3) 98.6 (SD 41.2) 0.060 4 136.0 (SD 39.3) 104.0 (SD 45.4) 0.020 5 141.4 (SD 40.8) 113.9 (SD 48.6) 0.045 6 147.0 (SD 42.9) 121.4 (SD 48.4) 0.042 7 152.9 (SD 40.7) 126.3 (SD 47.6) 0.031 8 155.7 (SD 39.5) 131.2 (SD 48.5) 0.041 9 157.8 (SD 39.8) 135.1 (SD 46.8) 0.061 10 161.0 (SD 39.2) 136.9 (SD 47.0) 0.052 11 163.3 (SD 38.1) 139.6 (SD 46.2) 0.059 12 165.9 (SD 37.3) 142.0 (SD 43.5) 0.042 13 167.2 (SD 37.2) 144.2 (SD 42.9) 0.050 14 168.4 (SD 37.2) 146.2 (SD 42.8) 0.064 15 169.8 (SD 36.4) 147.4 (SD 42.8) 0.060 16 171.2 (SD 35.2) 148.7 (SD 42.4) 0.058 17 171.8 (SD 35.3) 150.1 (SD 42.6) 0.069 18 172.6 (SD 35.6) 151.0 (SD 42.8) 0.072 19 173.4 (SD 35.9) 151.8 (SD 42.3) 0.073 20 174.2 (SD 35.1) 152.8 (SD 42.1) 0.072 21 174.5 (SD 35.3) 153.7 (SD 42.0) 0.078 22 175.2 (SD 35.4) 154.3 (SD 41.6) 0.075 23 175.4 (SD 35.4) 155.3 (SD 41.1) 0.083 24 176.1 (SD 34.6) 155.7 (SD 40.9) 0.071 25 176.5 (SD 34.5) 156.2 (SD 40.8) 0.071 26 176.8 (SD 34.5) 156.7 (SD 40.4) 0.072 27 177.1 (SD 34.4) 157.1 (SD 40.0) 0.072 28 177.4 (SD 34.1) 157.4 (SD 39.7) 0.069 29 177.6 (SD 33.8) 157.6 (SD 39.5) 0.067 30 177.7 (SD 33.7) 157.9 (SD 39.1) 0.067 Mean comprehensive asymmetry indexes (CAI) and p-values (paired samples t-test) for comparisons between barefoot and shod running based on different CAIs calculated with the raw difference vector (Δq) and different numbers of principal component vectors (k = 1. . .30). doi:10.1371/journal.pone.0138631.t002 Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 7 / 12 Fig 3. Weighted loadings of the first eight principal component vectors. These eight principal component vectors (PC-vectors) were used to calculate the comprehensive asymmetry index (CAI). Y-axes indicate the magnitude of the loading. X-axes represent the analysed biomechanical variables: V-Vertical; ML-Medial lateral; AP-Anterior posterior; GRF-Ground reaction force; CoP-Centre of pressure; S-Sagittal plane; F-Frontal plane; T-Transverse plane; A-Angle; M-Moment; V-Velocity. doi:10.1371/journal.pone.0138631.g003 Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 8 / 12 respect to overall gait asymmetry. The CAI offers no advantage, however, when it is necessary to quantify gait asymmetries of isolated variables (e.g. sagittal knee joint angle) at a specific time-point (e.g. at mid stance). In this case, other methods may provide a faster and more pre- cise assessment of gait asymmetry [15, 23–26]. It is important to realize that CAIs can only be compared among individuals when they have been calculated using the same variables. Another limitation of the current method is that it is possible that unique gait asymmetries present in only one individual may not contribute sufficiently to be represented in the lower order PC-vectors. Therefore, if this method is applied as a diagnostic tool to quantify asymme- try in an individual patient, then both the PCA-filtered and direct Euclidean distance-based CAI should be assessed to ensure that the patient does not exhibit an unusual asymmetry pattern. The results of the sensitivity analysis (Table 2) suggested that the PCA acted as a filter sepa- rating correlated from uncorrelated gait asymmetry variables [16]. Correlated asymmetries are more likely to contain actual differences in the movement pattern while uncorrelated asymme- tries are more likely to contain a high proportion of noise [19]. Another advantage of determin- ing the correlation structure of gait asymmetry using a PCA is that the resultant PC-vector loadings show the relevant asymmetry variables and their correlations. In fact, investigating the relevant asymmetry variables and their correlations suggested that the ankle and knee joint seemed to have the highest importance for the generation and compensation of gait asymmetry (Fig 3). Gait variables of the hip seemed to be less involved. Determining the relevant asymme- try variables and their correlation has potential for clinical gait analysis and may contribute to a better understanding of the specific manifestations of a patient’s underlying condition. PCA has been used before when investigating gait asymmetry [14, 15, 24]. However, to the best knowledge of the authors, it has not yet been applied in the all-encompassing form that was set up in this study. The CAI was based on data measured with a 3D motion capture system and a force platform during over-ground running. This experimental setup limits the amount of strides that can be measured and may also reduce the applicability of the CAI to monitor gait asymmetry in spe- cific cases (i.e. a laboratory setting is required). Therefore, future studies should investigate the sensitivity of the CAI to detect gait asymmetry changes using data acquired with wearable sen- sors (e.g. accelerometers) to increase the amount of data that can be collected and the applica- bility of the CAI. Because of the small sample size (15 study participants) and the recruitment of healthy indi- viduals only, a systematic discussion of CAI values is not possible, and an actual non-patholog- ical asymmetry range was not identified. Further studies should determine specific pathological and non-pathological ranges, as well as investigate how limb dominance, gender, or other external boundary conditions affect the CAI. Effect of footwear on gait asymmetry Gait asymmetry in a healthy population has been documented in several studies [14, 15, 27]. Previous research has also reported an impact of footwear on the running kinematics and kinetics of healthy adults [28–30]. From a purely mechanical perspective, one would expect that wearing footwear, which may not be manufactured perfectly symmetrical, would either not affect or increase gait asymmetry. However, as pointed out in the introduction, previous studies indicated that footwear may improve neuromuscular control of motion. This might lead to a decrease in gait asymmetry as suggested by Vagenas and Hoshizaki [31] based on a limited set of isolated kinematic variables of the foot. The findings of the comprehensive analy- sis of this study support this hypothesis (Table 2). Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 9 / 12 Improved motor control mechanisms associated with wearing footwear might be a result of altered cutaneous sensory information of the plantar or dorsal surface of the feet [32–34]. Two recent review studies attested to the significance of plantar sensory feedback for the control of movement and supported the utilization of textured materials for improving perceptual-motor performance [35, 36]. The magnitude of the effect of footwear on gait asymmetry was subject-dependent (Fig 2). In fact, a few study participants (3 out of 15) even demonstrated an increase in gait asymmetry when running in shoes. De Wit et al. [28] reported a subject-depended impact of footwear on the kinematics and kinetics during running. However, it remains unknown which mechanisms cause these subject-dependent responses to footwear. One mechanism might be related to sub- ject-specific sensitivity thresholds of the plantar or dorsal surface of the feet that may influence the afferent feedback to the neuromuscular control system [33]. Conclusion Footwear seems to reduce gait asymmetry during running in healthy, young individuals. Changes in the afferent sensory feedback to the neuromuscular control system may be a possi- ble explanation for this observation. Supporting Information S1 File. Supplementary Data. Subject demographics, eigenvalue spectrum, subject-specific comprehensive asymmetry index (CAI) for barefoot and shod running calculated using the raw Δq and different numbers of principal component vectors (k = 1. . .30), and weighted load- ings of the first eight principal component vectors. (XLSX) Acknowledgments The authors wish to thank María Fernanda Frías for data acquisition and Beatrix Vereijken for helpful feedback on the manuscript. Author Contributions Conceived and designed the experiments: BMN. Performed the experiments: JB BMN. Ana- lyzed the data: SH PAF CM. Contributed reagents/materials/analysis tools: SH PAF CM. Wrote the paper: SH PAF CM JB BMN. References 1. Alexander BH, Rivara FP, Wolf ME. The cost and frequency of hospitalization for fall-related injuries in older adults. Am J Public Health. 1992; 82: 1020–1023. PMID: 1609903 2. Ayoung-Chee P, McIntyre L, Ebel BE, Mack CD, McCormick W, Maier RV. Long-term outcomes of ground-level falls in the elderly. J Trauma Acute Care Surg. 2014; 76: 498–503. doi: 10.1097/TA. 0000000000000102 PMID: 24458057 3. Sattin RW. Falls among older persons: A public health perspective. Annu Rev Public Health. 1992; 13: 489–508. PMID: 1599600 4. Koepsell TD, Wolf ME, Buchner DM, Kukull WA, LaCroix AZ, Tencer AF, et al. Footwear style and risk of falls in older adults. J Am Geriatr Soc. 2004; 52: 1495–1501. PMID: 15341551 5. Federolf PA, Roos L, Nigg BM. The effect of footwear on postural control in bipedal quiet stance. Foot- wear Sci. 2012; 4: 115–122. 6. Larsen ER, Mosekilde L, Foldspang A. Correlates of falling during 24 h among elderly Danish commu- nity residents. Prev Med. 2004; 39: 389–398. PMID: 15226051 Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 10 / 12 7. Menz HB, Morris ME, Lord SR. Footwear characteristics and risk of indoor and outdoor falls in older people. Gerontology. 2006; 52: 174–180. PMID: 16645298 8. Shultz R, Birmingham TB, Jenkyn TR. Differences in neutral foot positions when measured barefoot compared to in shoes with varying stiffnesses. Med Eng Phys. 2011; 33: 1309–1313. doi: 10.1016/j. medengphy.2011.05.009 PMID: 21700484 9. Hendrickson J, Patterson KK, Inness EL, McIlroy WE, Mansfield A. Relationship between asymmetry of quiet standing balance control and walking post-stroke. Gait Posture. 2014; 39: 177–181. doi: 10. 1016/j.gaitpost.2013.06.022 PMID: 23877032 10. Patterson KK, Gage WH, Brooks D, Black SE, McIlroy WE. Evaluation of gait symmetry after stroke: a comparison of current methods and recommendations for standardization. Gait Posture. 2010; 31: 241–246. doi: 10.1016/j.gaitpost.2009.10.014 PMID: 19932621 11. Yogev G, Plotnik M, Peretz C, Giladi N, Hausdorff JM. Gait asymmetry in patients with Parkinson's dis- ease and elderly fallers: when does the bilateral coordination of gait require attention? Exp Brain Res. 2007; 177: 336–346. PMID: 16972073 12. Böhm H, Döderlein L. Gait asymmetries in children with cerebral palsy: do they deteriorate with run- ning? Gait Posture. 2012; 35: 322–327. doi: 10.1016/j.gaitpost.2011.10.003 PMID: 22055251 13. Exell TA, Gittoes MJ, Irwin G, Kerwin DG. Gait asymmetry: composite scores for mechanical analyses of sprint running. J Biomech. 2012; 45: 1108–1111. doi: 10.1016/j.jbiomech.2012.01.007 PMID: 22296935 14. Sadeghi H. Local or global asymmetry in gait of people without impairments. Gait Posture. 2003; 17: 197–204. PMID: 12770633 15. Sadeghi H, Allard P, Prince F, Labelle H. Symmetry and limb dominance in able-bodied gait: a review. Gait Posture. 2000; 12: 34–45. PMID: 10996295 16. Daffertshofer A, Lamoth CJ, Meijer OG, Beek PJ. PCA in studying coordination and variability: a tuto- rial. Clin Biomech. 2004; 19: 415–428. 17. Maurer C, Federolf PA, von Tscharner V, Stirling L, Nigg BM. Discrimination of gender-, speed-, and shoe-dependent movement patterns in runners using full-body kinematics. Gait Posture. 2012; 36: 40– 45. doi: 10.1016/j.gaitpost.2011.12.023 PMID: 22304784 18. Giakas G, Baltzopoulos V. Time and frequency domain analysis of ground reaction forces during walk- ing: an investigation of variability and symmetry. Gait Posture. 1997; 5: 189–197. 19. Federolf PA, Boyer K, Andriacchi T. Application of principal component analysis in clinical gait research: identification of systematic differences between healthy and medial knee-osteoarthritic gait. J Biomech. 2013; 46: 2173–2178. doi: 10.1016/j.jbiomech.2013.06.032 PMID: 23910389 20. Angeloni C, Riley PO, Krebs DE. Frequency content of whole body gait kinematic data. IEEE Trans Rehabil Eng. 1994; 2: 40–46. 21. Moore JK, Kooijman JDG, Schwab AL, Hubbard M. Rider motion identification during normal bicycling by means of principal component analysis. Multibody Syst Dyn. 2011; 25: 225–244. 22. Chau T. A review of analytical techniques for gait data. Part 1: Fuzzy, statistical and fractal methods. Gait Posture. 2001; 13: 49–66. PMID: 11166554 23. Crenshaw SJ, Richards JG. A method for analyzing joint symmetry and normalcy, with an application to analyzing gait. Gait Posture. 2006; 24: 515–521. PMID: 16427288 24. Olney SJ, Griffin MP, McBride ID. Multivariate examination of data from gait analysis of persons with stroke. Phys Ther. 1998; 78: 814–828. PMID: 9711207 25. Robinson RO, Herzog W, Nigg BM. Use of force platform variables to quantify the effects of chiropractic manipulation on gait symmetry. J Manipulative Physiol Ther. 1987; 10: 172–176. PMID: 2958572 26. Zifchock RA, Davis I, Higginson J, Royer T. The symmetry angle: a novel, robust method of quantifying asymmetry. Gait Posture. 2008; 27: 622–627. PMID: 17913499 27. Sadeghi H, Allard P, Duhaime M. Functional gait asymmetry in able-bodied subjects. Hum Mov Sci. 1997; 16: 243–258. 28. De Wit B, De Clercq D, Aerts P. Biomechanical analysis of the stance phase during barefoot and shod running. J Biomech. 2000; 33: 269–278. PMID: 10673110 29. Bishop M, Fiolkowski P, Conrad B, Brunt D, Horodyski M. Athletic footwear, leg stiffness, and running kinematics. J Athl Train. 2006; 41: 387–392. PMID: 17273463 30. Hamill J, Russell EM, Gruber AH, Miller R. Impact characteristics in shod and barefoot running. Foot- wear Sci. 2011; 3: 33–40. 31. Vagenas G, Hoshizaki B. A multivariable analysis of lower extremity kinematic asymmetry in running. Int J Sports Biomech. 1992; 8: 11–29. Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 11 / 12 32. Muise SB, Lam CK, Bent LR. Reduced input from foot sole skin through cooling differentially modulates the short latency and medium latency vestibular reflex responses to galvanic vestibular stimulation. Exp Brain Res. 2012; 218: 63–71. doi: 10.1007/s00221-012-3002-2 PMID: 22278107 33. Yi Y, Park S. Effect of reduced cutaneous cues on motion perception and postural control. Exp Brain Res. 2009; 195: 361–369. doi: 10.1007/s00221-009-1796-3 PMID: 19404630 34. Zhang S, Li L. The differential effects of foot sole sensory on plantar pressure distribution between bal- ance and gait. Gait Posture. 2013; 37: 532–535. doi: 10.1016/j.gaitpost.2012.09.012 PMID: 23063479 35. Alfuth M, Rosenbaum D. Effects of changes in plantar sensory feedback on human gait characteristics: a systematic review. Footwear Sci. 2012; 4: 1–22. 36. Orth D, Davids K, Wheat J, Seifert L, Liukkonen J, Jaakkola T, et al. The role of textured material in sup- porting perceptual-motor functions. PLOS ONE. 2013; 8: e60349. doi: 10.1371/journal.pone.0060349 PMID: 23565232 Footwear Decreases Gait Asymmetry during Running PLOS ONE | DOI:10.1371/journal.pone.0138631 October 21, 2015 12 / 12

Footwear Decreases Gait Asymmetry during Running.

10-21-2015

Hoerzer, Stefan,Federolf, Peter A,Maurer, Christian,Baltich, Jennifer,Nigg, Benno M

eng

PMC8073231

medicina Article Pacing in Long-Distance Running: Sex and Age Differences in 10-km Race and Marathon Ivan Cuk 1 , Pantelis T. Nikolaidis 2,3 , Elias Villiger 4 and Beat Knechtle 4,5,*



 Citation: Cuk, I.; Nikolaidis, P.T.; Villiger, E.; Knechtle, B. Pacing in Long-Distance Running: Sex and Age Differences in 10-km Race and Marathon. Medicina 2021, 57, 389. https://doi.org/10.3390/medicina 57040389 Academic Editor: Edgaras Stankeviˇcius Received: 25 February 2021 Accepted: 15 April 2021 Published: 17 April 2021 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). 1 Faculty of Physical Education and Sports Management, Singidunum University, 11000 Belgrade, Serbia; ivan_cuk84@yahoo.com 2 Exercise Physiology Laboratory, 18450 Nikaia, Greece; pademil@hotmail.com 3 School of Health and Caring Sciences, University of West Attica, 10679 Athens, Greece 4 Institute of Primary Care, University of Zurich, 8006 Zurich, Switzerland; eviliger@gmail.com 5 Medbase St. Gallen Am Vadianplatz, 9000 St. Gallen, Switzerland * Correspondence: beat.knechtle@hispeed.ch; Tel.: +41-(0)-71-226-93-00 Abstract: Background and objective: The recent availability of data from mass-participation run- ning events has allowed researchers to examine pacing from the perspective of non-elite distance runners. Based on an extensive analysis of the literature, we concluded that no study utilizing mass-participation events data has ever directly compared pacing in the 10-km race, with other long-distance races. Therefore, the main aim of this study was to assess and compare pacing between 10-km runners and marathoners, in regards to their sex and age. Materials and methods: For the purpose of this study, official results from the Oslo marathon (n = 8828) and 10-km race (n = 16,315) held from 2015 to 2018 were included. Results: Both 10-km runners and marathoners showed positive pacing strategies. Moreover, two-way analysis of variance showed that women were less likely to slow in the marathon than men (9.85% in comparison to 12.70%) however, not in the 10-km race (3.99% in comparison to 3.38%). Finally, pace changing is more prominent in youngest and oldest marathoners comparing to the other age groups (12.55% in comparison to 10.96%). Conclusions: Based on these findings, practitioners should adopt different training programmes for marathoners in comparison to shorter long-distance runners. Keywords: running; endurance; health; marathoners; recreation 1. Introduction Pacing can be defined as the distribution of exercise intensity during a prolonged time [1]. Optimal pacing is one of the most important contributors to achieving the best results in long-distance running [1,2], while significantly decreasing the risk of muscu- loskeletal injuries [3]. Although a pacing strategy that can classify as optimal depends on many factors (e.g., race length and profile, altitude, or weather conditions [4,5]) an even pacing strategy with an end-spurt has often been the best choice for long-distance events [6]. This strategy was best seen in the recent successful sub-2-h marathon challenge by Eliud Kipchoge, where the pacing was artificially controlled to be even throughout the race, with the spontaneous end-spurt by Kipchoge in the last several hundred meters [7]. Different pacing strategies in long-distance running were extensively investigated by sev- eral influential studies, both on track [4] and road racing [8,9]. However, those studies only involved a small sample size of professional athletes. On the other hand, the recent availability of mass-participation events data has allowed researchers to examine pacing from the perspective of a wider range of athletes including recreational distance runners of all ages [10,11]. The first studies using mass-participation events data were focused on independent long-distance events, primarily marathons and half-marathons. Contrary to elite runners, when a wider range of athletes was investigated, previous research reported positive pacing Medicina 2021, 57, 389. https://doi.org/10.3390/medicina57040389 https://www.mdpi.com/journal/medicina Medicina 2021, 57, 389 2 of 11 in endurance running races, where pace (time/distance) increased after approximately 3/4 of the race [12,13]. This decline in running speed was more prominent in men than in women due to several physiological [12] and psychological factors [14,15], such as a riskier and faster start by men or women’s better fat utilization to obtain energy. Moreover, studies investigating pacing in age group endurance runners proved to be somewhat inconsistent in their findings, with either no differences between age groups [11,12] or with more even pacing in older age groups [13]. Regarding the 10-km race, several studies examined pacing in this long-distance event, mainly in a small samples of track runners [1,4], time trial runners on either track [16] or treadmill [17], as well as in triathletes [18]. Pacing in the aforementioned elite 10-km runners proved to be rather even with an end-spurt. However, in recent years, 10-km races became increasingly popular in recreational runners, especially among women and beginners of both sexes [19,20]. Recreational athletes would participate in a 10-km race in the context of their preparation for a subsequent longer race [19], as well as a training tool to ameliorate pace time [21]. Therefore, further investigation of pacing in recreational 10-km runners is crucial to better understand the mechanisms controlling the pacing in long-distance events. This might help runners to enjoy running more as well as to achieve better results. In recent years, several studies using new methodological approaches attempted to compare pacing in mass-participation events [10,20]. Nevertheless, the performance of different distances and events was not adjusted in the abovementioned studies; e.g., some events might be under different environmental conditions, opposition fields, or race profiles. On the other hand, a new methodological approach allowed researchers to directly compare pacing between half-marathon and marathon on the same race and track, with rather similar weather conditions [10,22], thus providing more detailed and comparable results. For example, a novel finding was that women’s pacing was similar to men’s in half-marathon, whereas in a marathon women had more even pacing compared to men. Specifically, physiological rather than psychological factors can influence the additional lack of speed in marathoners (and not half-marathoners), such as better utilization of fat by women or men’s muscle glycogen depletion [10,22]. Accordingly, women did not differ by age group in pace variability, whereas youngest and oldest men, showed larger variability in pace [10]. However, further proof is needed that the observed sex and age differences are not specific to only a few races (i.e., Vienna and Ljubljana) and only two long-distance running events (i.e., half-marathon and marathon). This can be achieved by investigating and comparing pacing strategies in other long-distance races, such as longer ultra-races, or shorter and increasingly popular 10km races with already popular and investigated half-marathon or marathon. Based on an extensive analysis of the literature, we concluded that no study utilizing mass-participation events data has ever directly compared pacing in a 10-km race, with other long-distance races. Such a comparison could shed additional light on the importance of the mechanisms underlying pacing behaviour of the long distance runners, as well as to better understand potential training requirements for both recreational and proficient runners. Therefore, the main aim of this study was to assess and compare pacing between the increasingly popular 10-km race and the most popular long-distance race—marathon, in regards to their sex and age. We hypothesized that 10-km runners will show more even pacing than marathoners, particularly women and middle age runners. 2. Materials and Methods For this study, official results from the Oslo marathon and Oslo 10-km race held from 2015 to 2018 were included [23]. Split times from the middle of the race were also included (i.e., 5 km and 21.0975 km for the 10-km race and marathon respectively). The Oslo marathon was chosen as an officially certified race because it was held on a rather flat track (elevation difference 60 m). For reference, the Berlin Marathon considered “the fastest marathon”, has an elevation difference of 21 m [24]. Moreover, both the 10-km race and marathon were held on the same day, whereas the 10-km race was entirely Medicina 2021, 57, 389 3 of 11 contained within the marathon race. Finally, note that hyperthermia can significantly affect pacing in both elite and recreational runners [5,12]. The Oslo Marathon is traditionally held in Norway at the end of September, usually in colder weather conditions, which can reduce the chances of hyperthermia in runners. 2.1. Participants In total, 25,143 participants of all performance levels were considered for this study (10-km race, n = 16,315; Marathon, n = 8828), however, most of them were recreational runners. Participants who did not finish any of the races, or did not have recorded any of the split times were excluded from the initial sample. The present research was approved by the Institutional Review Board of Kanton St. Gallen, Switzerland, with a waiver of the requirement for informed consent of the par- ticipants as the research concerned the study of publicly available data (Ethical Committee St. Gallen 1 June 2010). This research was conducted in accordance with ethical standards derived from the Declaration of Helsinki adopted in 1964 and revised in 2013. 2.2. Data Acquisition All data was acquired from the official Oslo Marathon results page [23]. First, overall times, athlete details and a link to each athlete’s split times were copied from the results page and pasted into an Excel document. This was done separately per year, distance and gender. The split times were then added in a second step using custom Python scripts that followed the official link to each athlete’s split times and extracted all available split times. 2.3. Procedures In the first step of data analysis, the average running speed of all runners was cal- culated for the first and second half of both the 10-km race and marathon. A particular novelty of this study was the use of time (i.e., minutes and seconds) per kilometre as a unit of speed. This “runners friendly” measurement of speed was chosen as a very practical tool for both professional and recreational runners as well as their coaches. For example, GPS watches, often utilized by runners to monitor running speed, presents minutes per kilometre by default. Moreover, in long-distance races, each kilometre is usually marked. Therefore, participants can see the time they consumed running between each kilometre. As a result, runners and running coaches often rely on the time needed to run one kilometre when assessing and comparing someone’s running speed or pace maintenance. Considering that even pacing is the best choice for long-distance running [6], pacing assessment from the aspect of speed maintenance was selected for this study. Thereafter, speed variation was calculated based on the percentage difference in speed observed between the second and the first half of the race (i.e., % change = (second half time − first half time)/first half time). Percentage change was considered as a continuous variable [14]. Applying this method, it was possible to normalize pace and compare it between different race distances [10]. Criteria for inclusion in the final data set were having timing data for the halfway mark and the full race in proper sequence (e.g., finishing time greater than split time); a net time less than the gun time; and a slowing less than 400% [14]. 2.4. Statistical Analysis Prior to all statistical tests, descriptive statistics were calculated as mean and standard deviation. Since the Kolmogorov-Smirnov or similar data normality tests are not sensi- tive when using a large sample size, data distribution normality was verified by visual inspection of histograms and QQ plots [10,22]. To assess pacing differences between the first and second half of the 10-km race and marathon, two 2-way between-within ANOVAs were performed (separately for women and men). The main effect of pace (first half and second half), race (10-km race and marathon), and the interaction pace x race were observed. Medicina 2021, 57, 389 4 of 11 To assess pace change between women and men in 10-km race and marathon, two-way ANOVA with between factors was performed. The main effect of sex (women and men), race (10-km race and marathon), and the interaction sex × race were observed. Finally, to assess pace change between age groups in 10-km race and marathon, two 2-way between-within ANOVAs were performed (separately for women and men). The main effect of age group (18–23; 24–34; 34–39; 40–44; 45–49; 50–54; 55–59; 60–64; 65+), race (10-km race and marathon) and the interaction age × race were observed. In addition, for all ANOVAs, Bonferroni post-hoc test was performed. The effect size was calculated as eta squared ( 2), where the values of 0.01, 0.06, and above 0.14 were considered small, medium, and large, respectively [25]. Alpha level was set at p ≤ 0.05. All statistical tests were performed using Microsoft Office Excel 2007 (Microsoft Corporation, Redmond, WA, USA) and SPSS 20 (IBM, Armonk, NY, USA). 3. Results The first and second half pacing of participants is presented in Table 1. Regardless of their sex and age, both 10-km runners and marathoners showed a positive pacing strategy (i.e., second half of the race was slower than the first half). Further examination of pacing between 10-km runners and marathoners, in regards to their sex and age, is presented in Figures 1–3. Table 1. Speed indicators (in min/km) of 10-km and marathon runners showed as mean ± standard deviation. Women (n10-km = 9932; nmarathon = 2048) Men (n10-km = 6383; nmarathon = 6780) 10-km Race (min/km) Marathon (min/km) 10-km Race (min/km) Marathon (min/km) Mean SD Mean SD Mean SD Mean SD Age: 18–23 n = 1589 First half 6:03.2 1:05.9 5:50.0 0:38.9 5:03.1 1:06.3 5:17.8 0:42.4 Second half 6:16.0 1:12.9 6:27.7 1:04.9 5:12.7 1:10.0 6:06.0 1:11.6 Total 6:09.6 1:08.5 6:08.9 0:49.4 5:07.9 1:07.1 5:41.9 0:54.3 Age: 24–34 n = 7777 First half 6:07.7 1:02.7 5:41.2 0:40.7 5:13.2 1:06.0 5:10.0 0:42.5 Second half 6:19.1 1:09.6 6:15.0 1:02.9 5:22.9 1:10.7 5:49.1 1:07.1 Total 6:13.4 1:05.2 5:58.1 0:49.8 5:18.1 1:07.4 5:29.6 0:52.4 Age: 34–39 n = 3695 First half 6:14.7 1:02.2 5:36.1 0:37.5 5:17.7 1:06.3 5:09.9 0:43.7 Second half 6:27.7 1:08.2 6:04.4 0:52.4 5:28.4 1:10.1 5:47.5 1:04.0 Total 6:21.2 1:04.4 5:50.2 0:43.4 5:23.1 1:07.4 5:28.7 0:51.9 Age: 40–44 n = 3784 First half 6:17.9 1:01.1 5:46.3 0:42.6 5:19.3 1:05.2 5:12.5 0:39.4 Second half 6:32.3 1:07.1 6:17.4 0:59.5 5:29.3 1:09.5 5:51.4 1:00.4 Total 6:25.1 1:03.3 6:01.8 0:49.6 5:24.3 1:06.6 5:32.0 0:47.6 Age: 45–49 n = 3514 First half 6:18.7 1:01.4 5:47.7 0:37.4 5:29.0 1:06.4 5:14.6 0:40.1 Second half 6:34.0 1:08.5 6:21.5 0:55.4 5:41.5 1:11.9 5:51.0 0:59.8 Total 6:26.4 1:04.2 6:04.6 0:44.6 5:35.3 1:08.4 5:32.8 0:47.9 Age: 50–54 n = 2342 First half 6:23.4 1:04.8 5:52.8 0:43.1 5:34.1 1:07.2 5:23.2 0:41.6 Second half 6:41.4 1:11.5 6:29.3 0:58.5 5:48.3 1:13.0 6:05.0 1:02.4 Total 6:32.4 1:07.4 6:11.0 0:49.2 5:41.2 1:09.3 5:44.1 0:49.9 Age: 55–59 n = 1256 First half 6:45.1 1:06.2 6:00.3 0:39.7 5:43.7 1:06.4 5:23.6 0:39.1 Second half 7:05.0 1:12.7 6:40.1 0:51.9 5:58.0 1:13.7 6:03.0 0:58.0 Total 6:55.1 1:08.5 6:20.2 0:44.5 5:50.8 1:09.4 5:43.3 0:46.7 Age: 60–64 n = 646 First half 6:52.6 1:09.2 6:29.6 0:50.1 5:53.3 1:17.4 5:37.0 0:47.1 Second half 7:10.9 1:11.8 7:26.8 1:07.5 6:09.4 1:22.8 6:17.3 1:05.3 Total 7:01.7 1:09.7 6:58.2 0:56.8 6:01.3 1:19.0 5:57.2 0:53.5 Age: 65+ n = 540 First half 7:03.6 1:03.8 6:44.0 0:41.5 6:18.7 1:09.1 6:02.3 0:44.0 Second half 7:28.2 1:11.2 7:27.9 1:02.7 6:35.4 1:16.1 6:54.7 1:06.9 Total 7:15.9 1:06.7 7:06.0 0:50.5 6:27.1 1:11.6 6:28.5 0:53.1 n = number of participants, SD = standard deviation of data, min/km = minutes per kilometer. Medicina 2021, 57, 389 5 of 11 Medicina 2021, 57, x FOR PEER REVIEW 5 of 11 Figure 1. Women’s (upper panel) and men’s (lower panel) running time in the first and second half of 10-km race and marathon. Data showed as mean ± standard deviation. **—Significant dif- ferences at p< 0.01. 3.1. Pacing in 10-km and Marathon For women (Figure 1, upper panel), the two-way ANOVA showed significant main effects of pace(F(3,11978) = 6513.1, ŋ2 = 0.02, p< 0.01), race(F(3,11978) = 187.7, ŋ2= 0.01, p< 0.01) as well as pace × raceinteraction (F(3,11978) = 1086.1, ŋ2< 0.01, p< 0.01), whereas for men (Figure Figure 1. Women’s (upper panel) and men’s (lower panel) running time in the first and second half of 10-km race and marathon. Data showed as mean ± standard deviation. **—Significant differences at p < 0.01. 3.1. Pacing in 10-km and Marathon For women (Figure 1, upper panel), the two-way ANOVA showed significant main effects of pace(F(3,11978) = 6513.1, 2 = 0.02, p < 0.01), race(F(3,11978) = 187.7, 2 = 0.01, p < 0.01) Medicina 2021, 57, 389 6 of 11 as well as pace × raceinteraction (F(3,11978) = 1086.1, 2 < 0.01, p < 0.01), whereas for men (Figure 1, lower panel), the two-way ANOVA showed significant main effects of pace(F(3,13161) = 8720.9, 2 = 0.04, p < 0.01), race(F(3,13161) = 14.6, 2 < 0.01, p < 0.01) as well as sex × race interaction (F(3,13161) = 2619.0, 2 = 0.01, p < 0.01). 3.2. Pace Change in Men and Women in 10-km and Marathon Regarding pace change (Figure 2), the two-way ANOVA showed significant main effects of sex(F(3,25139) = 129.6, 2 < 0.01, p < 0.01), race(F(3,25139) = 3717.2, 2 = 0.13, p < 0.01) as well as sex × race interaction (F(3,25139) = 149.3, 2 = 0.01, p < 0.01). 3.3. Age Group Pace Change in 10-km and Marathon For women (Figure 3, upper panel), the two-way ANOVA showed significant main effects of age(F(17,11962) = 359.3, 2 = 0.01, p < 0.01), race(F(17,11962) = 441.1, 2 = 0.04, p < 0.01) as well as age × race interaction (F(17,11962) = 189.3, 2 < 0.01, p < 0.01), whereas for men (Figure 3, lower panel), the two-way ANOVA showed significant main effects of age(F(17,13145) = 3.9, 2 < 0.01, p < 0.01), race(F(17,13145) = 1678.2, 2 = 0.11, p < 0.01) as well as age × race interaction (F(17,13145) = 4.0, 2 < 0.01, p < 0.01). Medicina 2021, 57, x FOR PEER REVIEW 6 of 1 1, lower panel), the two-way ANOVA showed significant main effects of pace(F(3,13161) = 8720.9, ŋ2 = 0.04, p< 0.01), race(F(3,13161) = 14.6, ŋ2< 0.01, p< 0.01) as well as sex × race interaction (F(3,13161) = 2619.0, ŋ2= 0.01, p< 0.01). 3.2. Pace Change in Men and Women in 10-km and Marathon Regarding pace change (Figure 2), the two-way ANOVA showed significant main effects of sex(F(3,25139) = 129.6, ŋ2 < 0.01, p< 0.01), race(F(3,25139) = 3717.2, ŋ2 = 0.13, p< 0.01) a well as sex × race interaction (F(3,25139) = 149.3, ŋ2 = 0.01, p< 0.01). 3.3. Age Group Pace Change in 10-km and Marathon For women (Figure 3, upper panel), the two-way ANOVA showed significant main effects of age(F(17,11962) = 359.3, ŋ2 = 0.01, p< 0.01), race(F(17,11962) = 441.1, ŋ2 = 0.04, p< 0.01) a well as age × race interaction (F(17,11962) = 189.3, ŋ2< 0.01, p< 0.01), whereas for men (Figure 3 lower panel), the two-way ANOVA showed significant main effects of age(F(17,13145) = 3.9 ŋ2 < 0.01, p< 0.01), race(F(17,13145) = 1678.2, ŋ2 = 0.11, p< 0.01) as well as age × race interaction (F(17,13145) = 4.0, ŋ2 < 0.01, p< 0.01). Figure 2. Pace change in 10-km race and marathon for women and men. Data showed as mean ± standard deviation. **—Significant differences at p< 0.01. Figure 2. Pace change in 10-km race and marathon for women and men. Data showed as mean ± standard deviation. **—Significant differences at p < 0.01. Medicina 2021, 57, 389 7 of 11 Medicina 2021, 57, x FOR PEER REVIEW 7 of 11 Figure 3. Pace change in 10-km race and marathon for women’s (upper panel) and men’s (lower panel) age groups. Data showed as mean ± standard deviation. **—Significant differences at p< 0.01. Figure 3. Pace change in 10-km race and marathon for women’s (upper panel) and men’s (lower panel) age groups. Data showed as mean ± standard deviation. **—Significant differences at p < 0.01. Medicina 2021, 57, 389 8 of 11 4. Discussion The main aim of this study was to assess and compare pacing between 10-km runners and marathoners, in regards to their sex and age. The hypothesis, that 10-km runners would show more even pacing than marathoners, particularly women and middle age runners, was partially confirmed. Contrary to the previous results obtained on elite 10-km runners, where an even or negative pacing profile was observed [1,4], recreational runners (both men and women) showed a positive pacing profile (Figure 1). However, when compared to the marathoners, the pace slowing that occurred in the second half of the 10-km race was significantly less. It seems that running performance does not affect pacing in the 10-km, as it does in the marathon [11,26]. In our study, marathoners achieved a faster speed than the 10-km runners (Table 1), thus we can assume that the 10-km runners were mostly beginners in comparison to the marathoners. Similar results were previously obtained for recreational half-marathoners [10,22], where plummet in pace was less prominent in recreational half- marathoners than the marathoners. Therefore, we can assume that the pacing difference between 10-km race and marathon was not related to the runners’ performance level, but could be attributed to the increased fatigue when participating in the longer distance race [22]. 4.1. Pacing in 10-km and Marathon Finally, note that time (expressed as minutes and seconds) per kilometre was utilized as a unit of speed. Similar studies have often used meters per second or kilometres per hour [10,11] as more common units of speed, particularly in the field of sports science. However, practitioners (e.g., coaches, runners of all levels, as well as some sports scientists) regularly use time per kilometre as a very comprehensive practical tool when assessing and comparing someone’s running speed or pace maintenance (see Materials and Methods for additional information). 4.2. Pace Change in Men and Women in 10-km and Marathon When women and men were compared, no differences in pace change were obtained in the 10-km race (Figure 2). In contrast to the 10-km race, the marathon women had a signif- icantly lower pace change in comparison to the marathon men. When we relate this finding to the previous studies comparing pacing in recreational half-marathoners and marathon- ers [10,22], it appears that in long-distance races shorter than a marathon, the men’s pacing strategy is equally good as the women’s one, potentially event slightly better (Figure 2). That would confirm the previously observed hypothesis that women have an advantage in pacing over men only in distances equal to or longer than marathons [10]. The obtained findings can possibly diminish the previously reported psychological effect on men’s pac- ing (i.e., a riskier and faster start by men, due to greater competitiveness [14,15,27]). It can be assumed that less variation in pacing, as observed in recreational female marathoners, was due to a better fat utilization to obtain energy [28], rather than burning glycogen stored in muscles [12]. 4.3. Age Group Pace Change in 10-km and Marathon In all age groups (in both women and men), the 10-km runners showed a lower pace change in comparison to the marathon runners (Figure 3). It appears that both the youngest and the oldest marathoners change pace more than other age groups (Figure 3, lower panel), which is also confirmed in the 2017 Vienna marathon and half-marathon [29]. Younger, less experienced marathoners might encounter an inadequate control mechanism of pacing by altering pace often, which in turn might induce an excess of fatigue. The control mechanism of pacing depends on the information of the endpoint and race duration, an inherent clock setting scalar timing, and the knowledge of pacing from previously finished races [30]. Younger inexperienced runners could lack this pacing template in the brain, hence they cannot control pace during prolonged activities, such as a marathon. On the Medicina 2021, 57, 389 9 of 11 other hand, elderly men spend more time running (i.e., run slowly). As a consequence, fatigue-induced change in pace is more likely to occur, regardless of the well-developed pacing template [29]. Similar results are obtained in women, whereas the youngest and oldest marathoners change pace the most (Figure 3, upper panel). This phenomenon is, however, less pronounced compared to men due to previously explained sex differences (see preceding paragraph). 4.4. Limitations Several limitations of this study should be noted. First, it was impossible to include the half-marathon mid-race split in this study, since the organizers usually don’t provide a split at 10.550 km. Also, this split is not quite popular among runners (both elite and recreational). Second, additional splits in the 10-km race would provide better insight into the runners’ pacing strategies. Unfortunately, almost no mass participation events provide these splits in races shorter than half-marathon. Third, this study has assessed only one event in four consecutive years, thus limiting the potential generalization of the obtained findings. On the other hand, this allows a greater sample of participants. Finally, combining all runners older than 65 into the one age group (since there is a limited number of runners older than 65), limits our knowledge about pacing in older 10-km and marathon runners. 4.5. Practical Applications Based on these findings, strength and conditioning coaches (e.g., running coaches) should adopt different training programmes for marathoners, in comparison to the par- ticipants in the shorter long-distance events. Particular emphasis should be placed on individualized training plans for beginners, with the purpose of achieving an even or negative pacing profile (or at least to reduce plummet in the speed in the second half of the race). For example, recreational runners could run several 10-km races or half-marathons with the goal of achieving an even or negative pacing profile, before they attempt to run a marathon. This pacing strategy might aid athletes to have a faster race time, decrease the risk of musculoskeletal injuries, and enhance the enjoyment of endurance running. Finally, a particular novelty of this study was the use of minutes per kilometre as a unit of speed, as a very “practitioner friendly” and quite comprehensive tool when assessing and comparing someone’s running speed or pace maintenance (see Methods for additional information). Similar studies could utilize this measurement of speed more often, thus providing more comparable findings. 5. Conclusions In conclusion, both 10-km runners and marathoners showed positive pacing strate- gies. Moreover, women are less likely to slow in the marathon, however not in shorter long-distance events. Finally, pace changing is more prominent in youngest and oldest marathoners comparing to the other age groups. Author Contributions: Conceptualization, I.C. and P.T.N.; methodology, I.C. and P.T.N.; software, I.C., P.T.N. and E.V.; validation, I.C. and P.T.N.; formal analysis, I.C., P.T.N. and E.V.; investigation, I.C., P.T.N. and E.V.; resources, I.C., P.T.N. and E.V.; data curation, I.C., P.T.N. and E.V.; writing— original draft preparation, I.C., P.T.N. and B.K.; writing—review and editing, I.C., P.T.N. and B.K.; visualization, I.C. and P.T.N.; supervision, P.T.N. and B.K.; project administration, E.V. and B.K.; funding acquisition, E.V. and B.K. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: The present research was approved by the Institutional Review Board of Kanton St. Gallen, Switzerland, with a waiver of the requirement for informed consent of the participants as the research concerned the study of publicly available data (Ethical Medicina 2021, 57, 389 10 of 11 Committee St. Gallen 1 June 2010). This research was conducted in accordance with ethical standards derived from the Declaration of Helsinki adopted in 1964 and revised in 2013. Informed Consent Statement: Patient consent was waived due to as the research concerned the study of publicly available data. Data Availability Statement: BMW Oslo Maraton. Available online: https://oslomaraton.no/ (accessed on 25April 2019). Conflicts of Interest: The authors declare no conflict of interest. References 1. Thiel, C.; Foster, C.; Banzer, W.; de Koning, J. Pacing in Olympic track races: Competitive tactics versus best performance strategy. J. Sports Sci. 2012, 30, 1107–1115. [CrossRef] 2. Foster, C.; De Koning, J.J.; Hettinga, F.; Lampen, J.; La Clair, K.L.; Dodge, C.; Bobbert, M.; Porcari, J.P. Pattern of energy expenditure during simulated competition. Med. Sci. Sports Exerc. 2003, 35, 826–831. [CrossRef] [PubMed] 3. De Koning, J.J.; Foster, C.; Bakkum, A.; Kloppenburg, S.; Thiel, C.; Joseph, T.; Cohen, J.; Porcari, J.P. Regulation of pacing strategy during athletic competition. PLoS ONE 2011, 6, e15863. [CrossRef] 4. Tucker, R.; Lambert, M.I.; Noakes, T.D. An analysis of pacing strategies during men’s world-record performances in track athletics. Int. J. Sports Physiol. Perform. 2006, 1, 233–245. [CrossRef] [PubMed] 5. Ely, M.R.; Martin, D.E.; Cheuvront, S.N.; Montain, S.J. Effect of ambient temperature on marathon pacing is dependent on runner ability. Med. Sci. Sports Exerc. 2008, 40, 1675–1680. [CrossRef] [PubMed] 6. Abbiss, C.R.; Laursen, P.B. Describing and understanding pacing strategies during athletic competition. Sports Med. 2008, 38, 239–252. [CrossRef] 7. INEOS. INEOS Challenge. Available online: https://www.ineos159challenge.com (accessed on 24 April 2020). 8. Hanley, B. Pacing profiles and pack running at the IAAF World Half Marathon Championships. J. Sports Sci. 2015, 33, 1189–1195. [CrossRef] 9. Renfree, A.; Gibson, A.S.T. Influence of different performance levels on pacing strategy during the Women’s World Championship marathon race. Int. J. Sports Physiol. Perform. 2013, 8, 279–285. [CrossRef] 10. Cuk, I.; Nikolaidis, P.T.; Knechtle, B. Sex differences in pacing during half-marathon and marathon race. Res. Sports Med. 2020, 28, 111–120. [CrossRef] 11. Nikolaidis, P.T.; Knechtle, B. Effect of age and performance on pacing of marathon runners. Open Access J. Sports Med. 2017, 8, 171–180. [CrossRef] 12. March, D.S.; Vanderburgh, P.M.; Titlebaum, P.J.; Hoops, M.L. Age, sex, and finish time as determinants of pacing in the marathon. J. Strength Cond. Res. 2011, 25, 386–391. [CrossRef] 13. Nikolaidis, P.T.; Knechtle, B. Pacing in age group marathoners in the “New York City Marathon”. Res. Sports Med. 2018, 26, 86–99. [CrossRef] 14. Deaner, R.O.; Carter, R.E.; Joyner, M.J.; Hunter, S.K. Men are more likely than women to slow in the marathon. Med. Sci. Sports Exerc. 2014, 47, 607–616. [CrossRef] [PubMed] 15. Ogles, B.M.; Masters, K.S. A typology of marathon runners based on cluster analysis of motivations. J. Sport Behav. 2003, 26, 69–85. 16. Damasceno, M.V.; Lima-Silva, A.E.; Pasqua, L.A.; Tricoli, V.; Duarte, M.; Bishop, D.J.; Bertuzzi, R. Effects of resistance training on neuromuscular characteristics and pacing during 10-km running time trial. Eur. J. Appl. Physiol. 2015, 115, 1513–1522. [CrossRef] [PubMed] 17. Lima-Silva, A.E.; Bertuzzi, R.C.M.; Pires, F.O.; Barros, R.V.; Gagliardi, J.F.; Hammond, J.; Kiss, M.A.; Bishop, D.J. Effect of performance level on pacing strategy during a 10-km running race. Eur. J. Appl. Physiol. 2010, 108, 1045–1053. [CrossRef] 18. Hausswirth, C.; Le Meur, Y.; Bieuzen, F.; Brisswalter, J.; Bernard, T. Pacing strategy during the initial phase of the run in triathlon: Influence on overall performance. Eur. J. Appl. Physiol. 2010, 108, 1115–1123. [CrossRef] 19. Nikolaidis, P.T.; Cuk, I.; Clemente-suárez, V.J.; Villiger, E.; Knechtle, B. Number of finishers and performance of age group women and men in long-distance running: Comparison among 10km, half-marathon and marathon races in Oslo. Res. Sports Med. 2020. [CrossRef] 20. De Leeuw, A.W.; Meerhoff, L.A.; Knobbe, A. Effects of pacing properties on performance in long-distance running. Big Data 2018, 6, 248–261. [CrossRef] 21. Coquart, J.; Alberty, M.; Bosquet, L. Validity of a nomogram to predict long distance running performance. J. Strength Cond. Res. 2009, 23, 2119–2123. [CrossRef] 22. Nikolaidis, P.T.; ´Cuk, I.; Knechtle, B. Pacing of women and men in half-marathon and marathon races. Medicina 2019, 55, 14. [CrossRef] 23. BMW Oslo Maraton. Available online: https://oslomaraton.no/ (accessed on 25 April 2019). 24. Nikolaidis, P.T.; Cuk, I.; Rosemann, T.; Knechtle, B. Performance and Pacing of Age Groups in Half-Marathon and Marathon. Int. J. Environ. Res. Public Health 2019, 16, 1777. [CrossRef] Medicina 2021, 57, 389 11 of 11 25. Cohen, J. Statistical Power Analysis for the Behavioural Sciences; Lawrence Erlbaum Associates: Hillsdale, NJ, USA, 1988. 26. Breen, D.; Norris, M.; Healy, R.; Anderson, R. Marathon pace control in masters athletes. Int. J. Sports Physiol. Perform. 2018, 13, 332–338. [CrossRef] [PubMed] 27. Crofts, C.; Schofield, G.; Dickson, G. Women-only mass participation sporting events: Does participation facilitate changes in physical activity? Ann. Leis. Res. 2012, 15, 148–159. [CrossRef] 28. Tarnopolsky, M.A. Sex differences in exercise metabolism and the role of 17-beta estradiol. Med. Sci. Sports Exerc. 2008, 40, 648–654. [CrossRef] [PubMed] 29. Cuk, I.; Nikolaidis, P.T.; Markovic, S.; Knechtle, B. Age differences in pacing in endurance running: Comparison between marathon and half-marathon Men and Women. Medicina 2019, 55, 479. [CrossRef] [PubMed] 30. Foster, C.; Gibson, A.S.C.; Tucker, R.; Rauch, L.H.G.; Noakes, T.D.; Baden, D.A.; Lambert, E. V The Role of Information Processing Between the Brain and Peripheral Physiological Systems in Pacing and Perception of Effort. Sports Med. 2006, 36, 705–722. [CrossRef]

Pacing in Long-Distance Running: Sex and Age Differences in 10-km Race and Marathon.

04-17-2021

Cuk, Ivan,Nikolaidis, Pantelis T,Villiger, Elias,Knechtle, Beat

eng

PMC6651135

International Journal of Environmental Research and Public Health Article Women Reduce the Performance Difference to Men with Increasing Age in Ultra-Marathon Running Karin J. Waldvogel 1, Pantelis T. Nikolaidis 2,3 , Stefania Di Gangi 1, Thomas Rosemann 1 and Beat Knechtle 1,4,* 1 Institute of Primary Care, University of Zurich, 8091 Zurich, Switzerland 2 Exercise Physiology Laboratory, 18450 Nikaia, Greece 3 School of Health and Caring Sciences, University of West Attica, 12243 Athens, Greece 4 Medbase St. Gallen Am Vadianplatz, 9001 St. Gallen, Switzerland * Correspondence: beat.knechtle@hispeed.ch; Tel.: +41-(0)-71-226-93-00 Received: 8 June 2019; Accepted: 2 July 2019; Published: 4 July 2019



 Abstract: Age and sex are well-known factors influencing ultra-marathon race performance. The fact that women in older age groups are able to achieve a similar performance as men has been documented in swimming. In ultra-marathon running, knowledge is still limited. The aim of this study was to analyze sex-specific performance in ultra-marathon running according to age and distance. All ultra-marathon races documented in the online database of the German Society for Ultra-Marathon from 1964 to 2017 for 50-mile races (i.e., 231,980 records from 91,665 finishers) and from 1953 to 2017 for 100-mile races (i.e., 107,445 records from 39,870 finishers) were analyzed. In 50-mile races, race times were 11.74 ± 1.95 h for men and 12.31 ± 1.69 h for women. In 100-mile races, race times were 26.6 ± 3.49 h for men and 27.47 ± 3.6 h for women. The sex differences decreased with older age and were smaller in 100-mile (4.41%) than in 50-mile races (9.13%). The overall age of peak performance was 33 years for both distances. In summary, women reduced the performance difference to men with advancing age, the relative difference being smaller in 100-mile compared to 50-mile races. These findings might aid coaches and ultra-marathon runners set long-term training goals considering their sex and age. Keywords: age of peak performance; athlete; sex difference; ultra-endurance 1. Introduction The oldest entry in the collection of ultra-marathon running statistics provided by the “German Society for Ultra-Marathon” [1] was a 89 km run from London to Brighton taking place in 1837. Since then, the popularity of ultra-marathon running has substantially increased [2–5]. Ultra-marathon running competitions are mainly specified by duration in hours or days (e.g., six hours to ten days) or by distance in km or miles (e.g., 50 km, 100 km, 50 miles, and 100 miles). For a race to be considered as an ultra-marathon, the duration has to be at least 6 hours, or the distance has to be longer than 42.195 km (26.2 miles) [5,6]. Over the last decades, the number of ultra-marathon competitions [7] as well as the number of participants in these races has increased exponentially [8]. This increase appears to be mostly due to increasing numbers of athletes aged over 40 years (i.e., master athletes) [7], as well as women increasingly participating [3,8]. While very few women participated in the first ultra-marathon running competitions, their share has increased ever since [7,9,10]. Since 2004, approximately 20% of the runners have been women, but there are no records documenting women participating in the USA 161 km ultra-marathon distance in the 1970s [7]. Int. J. Environ. Res. Public Health 2019, 16, 2377; doi:10.3390/ijerph16132377 www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2019, 16, 2377 2 of 16 Multiple determinants of the ultra-marathon’s success have been identified. One very important factor is age [10–12]. Knowing the age of peak performance has been assessed as being indispensable for optimization of the training schedule and to plan a successful career as an ultra-runner [13]. Comparing marathon and ultra-marathon running, differences in the age of peak performance have been very recently reported. In marathon running, the best performances of women and men are achieved between 25 and 35 years of age [4,11,14–16]. In above-marathon distances, the age of peak performance is higher [4]. Several studies report that the best results are observed in men aged 30 to 49 years and in women aged 30 to 54 years in 100 km ultra-marathon races [13,17]. One explanation might be that most runners start their careers with marathons, only later in life enhancing the challenge with ultra-marathons [17]. Moreover, compared to marathon races, ultra-marathons require the necessary level of performance, and this depends even more on the critical factors of adequate training preparation with an appropriate nutrition plan and mental strength [6]. Another essential factor influencing race performance is the athlete’s sex. Even though the performance of women compared to men in endurance running was inferior in the past [9], the sex-related gap has decreased in the last couple of decades [18]. This observation led to speculation about whether and how women could reduce the difference in running times to a level where they might outperform men in long-distance races. Other authors have hypothesized that this might be more likely to happen with very long distances, such as in ultra-marathons [18–21]. In contrast to such expectations, some results seem to indicate a larger sex gap in ultra-marathons compared to marathons [20], although there might be a potential bias underlying these results. Most studies either did not consider all participants and only focused on the top athletes [15,22,23], or had a limited sample size of athletes, only investigating a small number of races and/or a limited period [11,16]. Comparing only the top ten world record performances carries the risk of the results being affected by athletes with the highest performance level. For example, Lepers et al. [22] restricted their analysis of triathletes to the top ten men of each age group in the Olympic triathlon and Ironman triathlon world championships of 2006 and 2007 (440 athletes in total) and found an age-related performance decline at the age of 50 years in swimming and at the age of 45 years in cycling and running. In contrast, Käch et al. [24], investigating 329,066 men and 81,815 women participating in Ironman triathlon competitions held between 2002 and 2015, found a performance decline at profoundly earlier ages (in swimming, at 25–29 years of age in women and men; in cycling and running, at 30–34 years of age in women and at 35–39 years of age in men). According to Käch et al. [24], the participants in Ironman triathlons are not only the top-performing athletes of each age group, but also recreational athletes, the latter typically not being included in an analysis of top-ten athletes. Top-performing athletes tend to have more experience, mental strength, training volume, and training intensity than recreational athletes and are thus more likely to be included in analyses of top-ten performers [24]. This could also explain that the performance of unselected athletes (i.e., investigation of performance of every participating athlete, as found by Käch et al. [24], tends to decline at an earlier age than performance in top-ten athletes (as found by Lepers et al. [22]). Could a similar mechanism also explain discrepant findings on the performance of men compared to women? Recent studies investigating master swimmers in pool and open-water swimming showed that women in older age groups (80 years and older) achieved a similar performance to men in an investigation of 65,584 freestyle pool swimmers (29,467 women and 36,117 men) competing in 50 to 800 m [25] races and when 7592 freestyle open-water swimmers (2829 women and 4768 men) competing in 3000 m [26] races in the FINA World championships from 1986–2014 and 1992–2014, respectively. In contrast, Senefeld et al. [27], who conducted a similar study except that they focused on the top ten swimmers in the years between 1986 and 2011 (6760 athletes in total, men and women), found that the performance of women in every age group was inferior and, contrary to Knechtle et al. [25,26], that the sex gap increased with age. The differences in the selection of performance levels could possibly explain these discrepant findings in comparisons of men versus women. In support of this interpretation, other studies found Int. J. Environ. Res. Public Health 2019, 16, 2377 3 of 16 a reduction in the sex gap in swimming performance with increasing age for different disciplines such as breaststroke [28], backstroke [29], butterfly [30], and in the individual medley event [31]. The commonality across these studies is that they included all participants in their investigation, rather than only the top ten performance participants. The results indicate that selection based on performance levels has an influence on the results regarding sex-specific performance differences, usually in favor of men. In contrast, studies performed with all athletes tend to display a lower or no sex gap. The fact that women in older age groups (i.e., older than 80 years) achieve a similar performance to men has only been reported for different swimming disciplines, but not for running. Knechtle et al. [32] reported results on ultra-marathon performance in men and women and found that with increasing age and race distance, the sex gap increased rather than decreased. However, our knowledge of whether women in older age groups would be able to achieve a similar performance for longer running distances is still limited. For instance, a recent study on road running records from 5 km to 6 days showed that men were faster than women, the sex gap decreased with increasing age, and it did not vary by race distance or duration [33]. We therefore investigated whether women in 50-mile and 100-mile ultra-marathon races would be able to reduce the gap to men in older age groups. In contrast to previous studies, we analyzed a much larger data set, containing all 50-mile ultra-marathon races held between 1964 and 2017 and all 100-mile races between 1953 and 2017, thus avoiding selection bias by not only focusing on the top participants. Based on previous findings for master swimmers, we hypothesized that the sex gap in performance in ultra-marathons would decrease with increasing age, and that this decrease would be independent from the race distance. 2. Materials and Methods 2.1. Ethical Approval This study was approved by the Institutional Review Board of the Kanton St. Gallen, Switzerland, with a waiver of the requirement for informed consent of the participants, as the study involved the analysis of publicly available data (1 June 2010). 2.2. Data Sampling The investigation comprised all ultra-marathon competitions with running distances documented in “miles” in the online database of the German Society for Ultra-Marathon (Deutsche Ultramarathon Vereinigung e.V.). A total of 7769 competitions with 456,167 men and women participating in the years from 1928 to 2017 [34] were extracted. The data set was retrieved in multiple steps. First, we used the Google Chrome browser (Version 66.0.3359.139) with the add-on “Web Scraper” (Version 0.3.7) to retrieve the Uniform Resource Locator (URL) of each ultra-marathon competition registered in the online database. Each URL was saved in Microsoft Excel 2013 (Version 15.0.4569.1504). Subsequently, the Microsoft Excel-integrated Visual Basic Application (VBA) was used to filter the database contents, excluding every URL of competitions that did not have a distance specified in miles. In a final step, also using Excel-VBA, the raw data of each competition was extracted and uniformly formatted. The resulting file was visually controlled for inconsistences, and these were corrected in accordance with the original data. For the purpose of the present study, we analyzed 339,425 records of athletes either finishing a 50-mile race from 1964 to 2017 or a 100-mile race from 1953 to 2017. Other race distances were excluded due to insufficient data. The following variables were extracted: year of race, race distance, name of race, country of race, race time (h), running speed (km/h), name of athlete, year of birth, nationality of athlete, and sex of athlete. Age was derived by subtracting the year of birth from 2017. Int. J. Environ. Res. Public Health 2019, 16, 2377 4 of 16 2.3. Statistical Analysis Descriptive statistics are presented as means ± standard deviations. Performance, or race time, was recorded in the format “hours:minutes:seconds” (h:min:s) and converted into hours, as a numerical variable. For 50- and 100-mile ultra-marathon races, t-tests were performed to compare the average performance between men and women by age group and by country. It was acknowledged that analyses of variance (ANOVAs) might have been easier to interpret; however, the mixed regression analysis was preferred since it was necessary to correct for clustered observations within runners who participate more than once. ANOVA would have not accounted for clustered observations. The age groups were 10–19, 20–29, 30–39, 40–49, 50–59, 60–74, and 75–95 years, and only observations with non-missing ages were considered in analyses involving age. Country groups were identified through participation prevalence by country: United States of America (USA), Canada (CAN), Great Britain (GBR), and Republic of South Africa (RSA). The other countries were grouped together. Age was considered as a continuous variable, in 1-year intervals, when defined as a predictor variable for ultra-marathon time. A non-linear regression mixed model with basis splines was performed to find the age of peak performance, which is the age at which the time record-fitted value has a minimum. The mixed model was used to correct for repeated measurements within runners (clusters) through the random effects of intercepts. Different regression model specifications were initially considered, with age–sex, age–country, and country–sex interaction terms and with different hypotheses about the age and time trend. Model selection was performed using both the Akaike information criterion (AIC) and the Bayes information criterion (BIC). In the final selected model, age, calendar year, sex, country, and a country–sex interaction term were considered as fixed effect predictors. The statistical model was specified as follows: Ultra − marathon time (Y) ∼ [ fixed ef fects (X) = BS(year, df = 3) + BS(age, df = 3) + sex + country + country ∗ sex] + [random ef fects of intercept = runners] where BS (year, df=3) and BS (age, df = 3) are 3 degrees of freedom (df) basis splines changing with calendar year and age, respectively; country*sex denotes the country–sex interaction term. Two different analyses were performed, one for 50-mile and one for 100-mile races. In the 50 miles analysis, South Africa was combined with other countries because of the low number of runners. Results of the regression models are presented as estimates and standard errors. In addition, sex differences (%) in performance were examined, defined as 100 × (women’s race time-men’s race time)/men’s race time. For all tests and regressions, statistical significance was defined as p < 0.05. All statistical analyses were carried out with R [35]. The packages ggplot2, lme4, and lmerTest were used, respectively, for data visualization and for the mixed model. 3. Results Between 1964 and 2017, a total of n = 231,980 records on 91,665 different finishers with information on age were retrieved from the database on 50-mile ultra-marathon races. For 100-mile races, a total of n = 107,445 records on 39,870 different finishers was available for the period between 1953 and 2017. Overall, the average number of observations per runner was 2.53 in 50-mile and 2.69 in 100-mile races. In 50-/100-mile races, the number of women was 23,548 (26%)/7789 (20%) with 55,540 (24% of the total observations)/20,154 (19% of the total observations) records, and the number of men was 68,107 (74%)/32,081 (80%) with 176,440 (76%)/87,291 (81%) records. The proportions of observations of finishers aged 50 years and above were 24.4% (men) and 17% (women) in 50-mile races and 24.9% (men) and 18.3% (women) in 100-mile races, indicating that finishing men tended to be slightly older than women. The vast majority of finishers participated in races in the USA (85.2%); 6.1%, 3.8%, and 0.1% of the sample participated in Great Britain, Canada, and South Africa, respectively, and 4.1% in races taking place in 43 other countries. Int. J. Environ. Res. Public Health 2019, 16, 2377 5 of 16 In Table 1, the number of observations and the average performance by sex, age group, and country are reported for 50- and 100-mile races. In both 50-/100-mile races, the shortest average race times were observed in the 20–29 years age group, both in men (10.30 h/26.07 h) and in women (11.18 h/27.14 h); the lowest average performances were observed in the 75–95 years age group, again both in men (14.20 h/29.73 h) and in women (13.40 h/29.00 h). In 50-mile races, the shortest average race times were observed in Canada (10.47 h in men and 11.35 h in women) and the longest in Great Britain (11.67 h in men and 12.87 h in women). In 100-mile races, the shortest average running times occurred in South Africa (21.82 h in men, 23.19 h in women) and the longest in the group of the 43 “other” countries (27.73 h in men and 28.11 h in women). Performance differences between sexes were significant (p < 0.001) for all age groups <75 years in the 50-mile races and for age groups between 20 and 60 years in the 100-mile races. In the ≥75 years of age group, better performances occurred in women compared to men, even though the difference failed to attain statistical significance due to the small sample size. The magnitude of the difference was, however, similar to that seen in younger age groups, where men are faster than women, and particularly in 5- mile races. With the largest performance sex gap in favor of men seen in the youngest age group (10–19 years), a clear performance trend over age is visible for both distances. Table 1. Mean ultra-marathon performance (50 and 100 miles) by sex, age group, and country (South Africa, due to a small sample size for 50-mile races, is combined with other countries). p-values of a t-test of mean performance between sexes are shown. 50 miles, n = 231,980 100 miles, n = 107,445 Age group Sex n Mean (hours) Sd (hours) p n Mean (hours) Sd (hours) p 10–19 Men 1312 10.9778 2.3152 <0.001 131 26.6158 5.1444 0.057 Women 177 12.0706 2.0483 12 30.9698 7.0097 20–29 Men 18,124 10.3022 2.1982 <0.001 5966 26.0697 5.4017 <0.001 Women 6409 11.1797 2.2465 1410 27.1409 4.9447 30–39 Men 53,553 10.3663 2.1998 <0.001 26,069 26.1852 5.6369 <0.001 Women 19,256 11.2043 2.2210 6778 27.3619 5.1514 40–49 Men 60,351 10.6421 2.1462 <0.001 33,387 26.9095 5.5481 <0.001 Women 20,234 11.5077 2.2047 8257 27.7625 5.0640 50–59 Men 33,857 11.1670 2.0841 <0.001 17,867 27.8913 5.2604 <0.001 Women 8210 12.1018 2.1911 3350 28.5333 5.0671 60–74 Men 9054 12.0289 2.1031 <0.001 3844 28.9254 4.9203 0.358 Women 1230 12.9829 2.4215 342 28.6806 4.6994 75–95 Men 189 14.1952 3.6604 0.199 27 29.7292 6.2894 0.571 Women 24 13.4018 2.6675 5 29.0034 0.8413 Country Sex n Mean (hours) Sd (hours) p n Mean (hours) Sd (hours) p Canada Men 6208 10.4748 2.2036 <0.001 2924 26.2718 4.4521 <0.001 Women 2563 11.3481 2.2938 1000 27.3393 4.6320 Great Britain Men 11,249 11.6678 3.0373 <0.001 4724 26.7545 6.1957 0.009 Women 2805 12.8717 3.3837 785 27.3641 6.0431 United States Men 149,514 10.6281 2.0732 <0.001 62,949 27.1163 4.9311 <0.001 Women 48,307 11.4174 2.1129 15,679 27.8919 4.5937 South Africa Men 3830 21.8187 3.3647 <0.001 Women 585 23.1866 3.9734 Other Men Women 9469 1865 10.8642 11.4223 2.6745 2.7839 <0.001 12,864 2105 27.7271 28.1111 7.6246 7.5810 0.031 (Note: Due to the small sample size for 50-mile races, South Africa was combined with other countries; p-values are from comparisons of mean performances between sexes). Int. J. Environ. Res. Public Health 2019, 16, 2377 6 of 16 Regarding country, in both distances and for all country groups, performance differences were significant (p < 0.001) between sexes due to a better performances in men, the largest differences being observed in South Africa. Table 2 describes, for both distances, the results of the statistical models, as described in the methods section (model selection statistics omitted). For 50 miles, race times were 11.74 (Sd = 1.95) h for men and 12.31 (Sd = 1.69) for women, with a sex difference of 9.13%. For 100 miles, race times were 26.6 (Sd = 3.49) h for men and 27.47 (Sd = 3.6) h for women, with a sex difference of 4.41%. Women were significantly slower than men (p < 0.001), the estimated sex differences being 0.74 (SE = 0.017) and 0.81 (SE = 0.075) hours in 50- and 100-mile races, respectively. For 50 miles, compared to the USA, finishers in Canada and in other countries were significantly (p < 0.001) faster by 0.19 (SE = 0.040) and 0.092 (SE = 0.028) hours, respectively. In contrast, finishers in GBR were significantly (p < 0.001) slower by 0.656 hours. For 100 miles, compared to the USA, finishers in GBR, CAN, and the RSA were significantly (p < 0.001) faster, with runners in the RSA being faster than in the USA by an estimated 3.938 hours. Other countries were slower by 0.893 hours, p < 0.001, compared to the USA. Table 2. Regression analysis (mixed model) of ultra-marathons (50 and 100 miles). Estimates and standard errors (SEs) of fixed effects are reported. P-value ranges are marked with asterisks (see note). Smoothing terms, basis splines, are denoted with BS(x) t, where x = year, age; t = 1,2,3. 50 miles Estimate (SE) 100 miles Estimate (SE) Intercept 12.462 *** 23.658 *** (0.152) (1.766) Year BS (year) 1 −4.009 *** 1.399 (0.236) (2.416) BS (year) 2 −1.218 *** 5.882 *** (0.126) (1.609) BS (year) 3 −0.313 * 6.651*** (0.148) (1.750) Age BS (age) 1 −2.427 *** −8.204*** (0.176) (1.055) BS (age) 2 −0.511*** 3.674 *** (0.090) (0.652) BS (age) 3 4.039 *** −0.637 (0.189) (1.523) Country (ref. United States) Canada −0.190 *** −1.001*** (0.040) (0.156) Great Britain 0.656 *** −0.322 ** (0.028) (0.109) Other −0.092 *** 0.893 *** (0.028) (0.070) South Africa −3.938 *** (0.128) Sex (ref. Men) Women (W) 0.740 *** 0.810 *** (0.017) (0.075) Int. J. Environ. Res. Public Health 2019, 16, 2377 7 of 16 Table 2. Cont. 50 miles Estimate (SE) 100 miles Estimate (SE) Country*Sex Canada*W 0.057 0.751* (0.074) (0.311) Great Britain*W 0.562 *** −0.133 (0.061) (0.273) Other countries*W −0.191 ** −0.339 (0.067) (0.180) South Africa*W 0.129 (0.331) Observations 231,980 107,445 Note: * p < 0.05; ** p < 0.01; *** p < 0.001. The Country*Sex interaction terms, for example the term Great Britain*W, estimates how much greater the effect of being a woman in a particular country (e.g., Great Britain) was on race time, compared to the USA. The interaction effects (Table 2) are visualized in Figure 1 (50 miles) and Figure 2 (100 miles). They were particularly pronounced for GBR in 50-mile races (0.562 hours) and for CAN in 100-mile races (0.751 hours), where the performances of women and men differed clearly more than in the USA. The distance between the fitted curves in men and women is largest for GBR in 50-mile races and for CAN in 100-mile races. Int. J. Environ. Res. Public Health 2019, 16, x FOR PEER REVIEW 8 of 17 (0.061) (0.273) Other countries*W −0.191 ** −0.339 (0.067) (0.180) South Africa*W 0.129 (0.331) Observations 231,980 107,445 Note: * p < 0.05; ** p < 0.01; *** p < 0.001. 234 The Country*Sex interaction terms, for example the term Great Britain*W, estimates how much 235 greater the effect of being a woman in a particular country (e.g., Great Britain) was on race time, 236 compared to the USA. The interaction effects (Table 2) are visualized in Figures 1 (50 miles) and 2 237 (100 miles). They were particularly pronounced for GBR in 50-mile races (0.562 hours) and for CAN 238 in 100-mile races (0.751 hours), where the performances of women and men differed clearly more 239 than in the USA. The distance between the fitted curves in men and women is largest for GBR in 240 50-mile races and for CAN in 100-mile races. 241 242 Figure 1. Ultra-marathon speed, 50 miles, by sex, age (in years), and country. Points are race-time averages. 243 Lines are fitted curves (mixed model). Vertical lines with numeric labels are the ages at peak performance. 244 USA = United States of America, CAN = Canada, GBR = Great Britain, W = women, M = men. 245 246 Figure 1. Ultra-marathon speed, 50 miles, by sex, age (in years), and country. Points are race-time averages. Lines are fitted curves (mixed model). Vertical lines with numeric labels are the ages at peak performance. USA = United States of America, CAN = Canada, GBR = Great Britain, W = women, M = men. Int. J. Environ. Res. Public Health 2019, 16, 2377 8 of 16 Int. J. Environ. Res. Public Health 2019, 16, x FOR PEER REVIEW 9 of 17 247 Figure 2. Ultra-marathon speed, 100 miles, by sex, age (in years), and country. Points are race-time averages. 248 Lines are fitted curves (mixed model). Vertical lines with numeric labels are the ages at peak performance. The 249 sample sizes decrease towards the minimum and maximum of the age axes, with some of the points reflecting 250 only individuals; for example, the five points corresponding with GBR men 75+ years of age reflect one 251 individual each, one of the three remarkable individuals (Geoffrey Oliver) accounting for three of the five 252 points [36]. USA = United States of America, CAN = Canada, GBR = Great Britain, RSA = Republic of South 253 Africa, W = women, M = men. 254 255 All the effects in Table 2, together with the age and year of peak performance, are shown 256 graphically in Figures 1–4. Both in 50-mile (Figure 1) and 100-mile (Figure 2) races, running times 257 decreased and, after reaching a minimum at 33 years (peak performance), increased with increasing 258 age. Regarding the calendar period, in 50-mile races, 1985 was the year of the best performance 259 (Figure 3), whereas in 100-mile races, performance worsened consistently over time (Figure 4). In 260 Figure 5, the estimated sex differences in performance by country are shown over age. For both 261 distances, the differences in favor of men increased up to about 33 years, and the increase was 262 subsequently followed by a decrease. In 100- but not in 50-mile races, the differences re-increased 263 slightly after about 80 years of age. For both distances, the estimated sex differences were smaller for 264 100- than for 50-mile races. Over calendar time, from 1953 to 2017, the sex difference in performance 265 decreased continuously in all countries in 100-mile races. For roughly the same period, the sex 266 difference in performance peaked at around 1985 in all countries for 50-mile races (Figure 6). 267 Figure 2. Ultra-marathon speed, 100 miles, by sex, age (in years), and country. Points are race-time averages. Lines are fitted curves (mixed model). Vertical lines with numeric labels are the ages at peak performance. The sample sizes decrease towards the minimum and maximum of the age axes, with some of the points reflecting only individuals; for example, the five points corresponding with GBR men 75+ years of age reflect one individual each, one of the three remarkable individuals (Geoffrey Oliver) accounting for three of the five points [36]. USA = United States of America, CAN = Canada, GBR = Great Britain, RSA = Republic of South Africa, W = women, M = men. All the effects in Table 2, together with the age and year of peak performance, are shown graphically in Figures 1–4. Both in 50-mile (Figure 1) and 100-mile (Figure 2) races, running times decreased and, after reaching a minimum at 33 years (peak performance), increased with increasing age. Regarding the calendar period, in 50-mile races, 1985 was the year of the best performance (Figure 3), whereas in 100-mile races, performance worsened consistently over time (Figure 4). In Figure 5, the estimated sex differences in performance by country are shown over age. For both distances, the differences in favor of men increased up to about 33 years, and the increase was subsequently followed by a decrease. In 100- but not in 50-mile races, the differences re-increased slightly after about 80 years of age. For both distances, the estimated sex differences were smaller for 100- than for 50-mile races. Over calendar time, from 1953 to 2017, the sex difference in performance decreased continuously in all countries in 100-mile races. For roughly the same period, the sex difference in performance peaked at around 1985 in all countries for 50-mile races (Figure 6). Int. J. Environ. Res. Public Health 2019, 16, 2377 9 of 16 Int. J. Environ. Res. Public Health 2019, 16, x FOR PEER REVIEW 10 of 17 268 269 Figure 3. Ultra-marathon speed, 50 miles, by sex, calendar year, and country. Points are race-time averages. 270 Lines are fitted curves (mixed model). Vertical lines with numeric labels are the ages at peak performance. 271 USA = United States of America, CAN = Canada, GBR = Great Britain, W = women, M = men. 272 273 4. Discussion 274 The aim of this study was to examine the sex gap in performance in ultra-marathons. We 275 hypothesized a decrease of the sex gap with increasing age and that this decrease would be 276 independent from race distance. The main findings were that the (i) sex difference in performance 277 was smaller in older than in younger athletes; (ii) the relative sex difference in performance was 278 smaller in 100- than in 50-mile races; (iii) the sex difference in performance approaches a historical 279 minimum; (iv) the peak performance age was 33 years; (v) the average performance worsened over 280 the last three decades. Minor findings were that (vi) men were slightly older than women; (vii) more 281 than two thirds (70%) of the finishers had participated in 50-mile races; (viii) three quarters (76%) of 282 all finishers were men; (ix) the proportion of men was higher in 100-mile races (80%) than in 50-mile 283 races (74%); (x) in South African races, men and women demonstrated the best 100-mile 284 performances. 285 286 Figure 3. Ultra-marathon speed, 50 miles, by sex, calendar year, and country. Points are race-time averages. Lines are fitted curves (mixed model). Vertical lines with numeric labels are the ages at peak performance. USA = United States of America, CAN = Canada, GBR = Great Britain, W = women, M = men. Int. J. Environ. Res. Public Health 2019, 16, x FOR PEER REVIEW 11 of 17 287 Figure 4. Ultra-marathon speed, 100 miles, by sex, calendar year, and country. Points are race-time averages. 288 Lines are fitted curves (mixed model). Vertical lines with numeric labels are the ages at peak performance. 289 USA = United States of America, CAN = Canada, GBR = Great Britain, RSA = Republic of South Africa, W = 290 women, M = men. 291 292 4.1. The Sex difference in performance was smaller in older than in younger athletes 293 In 50-mile races, the decline in the sex difference always decreasing up to the highest age. There 294 are multiple possible physiological mechanisms in men for the reduction in the performance sex gap 295 with increasing age, including lower levels of anabolic hormones [37], a decrease in neuromuscular 296 efficiency [38], and a reduced ability to synthesize protein [39] as well as body fat [40]. In addition, 297 the loss in skeletal muscle mass is more pronounced in men at the age of 60 years and above 298 Figure 4. Ultra-marathon speed, 100 miles, by sex, calendar year, and country. Points are race-time averages. Lines are fitted curves (mixed model). Vertical lines with numeric labels are the ages at peak performance. USA = United States of America, CAN = Canada, GBR = Great Britain, RSA = Republic of South Africa, W = women, M = men. Int. J. Environ. Res. Public Health 2019, 16, 2377 10 of 16 4. Discussion The aim of this study was to examine the sex gap in performance in ultra-marathons. We hypothesized a decrease of the sex gap with increasing age and that this decrease would be independent from race distance. The main findings were that the (i) sex difference in performance was smaller in older than in younger athletes; (ii) the relative sex difference in performance was smaller in 100- than in 50-mile races; (iii) the sex difference in performance approaches a historical minimum; (iv) the peak performance age was 33 years; (v) the average performance worsened over the last three decades. Minor findings were that (vi) men were slightly older than women; (vii) more than two thirds (70%) of the finishers had participated in 50-mile races; (viii) three quarters (76%) of all finishers were men; (ix) the proportion of men was higher in 100-mile races (80%) than in 50-mile races (74%); (x) in South African races, men and women demonstrated the best 100-mile performances. Int. J. Environ. Res. Public Health 2019, 16, x FOR PEER REVIEW 12 of 17 308 Figure 5. Sex differences by age (in years) and country in 50- and 100-mile ultra-marathons. Curves represent 309 fitted values. For 50-mile races, South Africa was combined with other countries. 310 USA = United States of America, CAN = Canada, GBR = Great Britain, RSA = Republic of South Africa, W = 311 women, M = men. Sex differences (%) in performance were defined as 100× (women’s race time–men’s race 312 time)/(men’s race time). 313 314 In contrast to 50-mile races, in 100-mile races, the age-related downward trend in the sex 315 difference reversed, the sex difference again increasing after about 80 years of age. It has to be noted, 316 however, that the number of athletes in the oldest age group was rather small, in particular in 317 100-mile races. Thus, the increase in the sex gap in 100-mile races could simply be due to chance. 318 Alternatively, however, the possibility of an increasing out-selection of relatively slow men at higher 319 ages, in particular in 100-mile races, cannot be excluded. This does not appear completely 320 implausible as physical performance is predictive of longevity at older ages [42,43], possibly 321 underlying a deficit in high-performing men. However, as the increase of the sex gap at very high 322 ages did not occur in 50-mile races, plain chance appears to be the more plausible explanation. 323 324 325 Figure 5. Sex differences by age (in years) and country in 50- and 100-mile ultra-marathons. Curves represent fitted values. For 50-mile races, South Africa was combined with other countries. USA = United States of America, CAN = Canada, GBR = Great Britain, RSA = Republic of South Africa, W = women, M = men. Sex differences (%) in performance were defined as 100× (women’s race time–men’s race time)/(men’s race time). 4.1. The Sex Difference in Performance Was Smaller in Older Than in Younger Athletes In 50-mile races, the decline in the sex difference always decreasing up to the highest age. There are multiple possible physiological mechanisms in men for the reduction in the performance sex gap with increasing age, including lower levels of anabolic hormones [37], a decrease in neuromuscular efficiency [38], and a reduced ability to synthesize protein [39] as well as body fat [40]. In addition, the loss in skeletal muscle mass is more pronounced in men at the age of 60 years and above compared to women of the same age, with sarcopenia present in ~53% of men compared to ~47% Int. J. Environ. Res. Public Health 2019, 16, 2377 11 of 16 of women [41]. Our finding of a sex gap reduction with increasing age in ultra-marathon running is consistent with recent findings of studies analyzing master swimmers competing in pool and open-water races [6,25,26,29–31]. The factor of sarcopenia was also suggested by Knechtle et al. [25], who investigated 65,584 freestyle master swimmers between 1986 and 2014. Sarcopenia might thus be an important factor in ultra-marathon running as well. Finally, compared to men, women tend to live longer and to be in better physical condition later in life [30]. A larger higher-age population of high-performing women as compared to men in 50-mile ultra-marathon races can thus be expected based on these considerations. Int. J. Environ. Res. Public Health 2019, 16, x FOR PEER REVIEW 13 of 17 326 Figure 6. Sex differences by calendar year and country in 50-mile and 100-mile ultra-marathons. 327 Curves represent fitted values. For 50-mile races, South Africa was combined with other countries. 328 USA = United States of America, CAN = Canada, GBR = Great Britain, RSA = Republic of South 329 Africa, W = women, M = men. Sex differences (%) in performance were defined as 100 × (women’s 330 race time–men’s race time) / men’s race time. 331 4.2. Relative sex difference in performance was smaller in 100- than in 50-Mile Races 332 Our second hypothesis of the decrease in the sex gap in performance with increasing age being 333 independent from race distance was not confirmed. For both race distances, a decline of the relative 334 sex difference is clearly visible, and the decline is more pronounced in 50-mile races than 100-mile 335 races. One explanation for this finding could be that in extremely long distances, like 100-mile races, 336 there might exist a sex-independent pace limit [44]. 337 This limit might constitute a performance maximum, outweighing sex differences at increasing 338 ages (“ceiling effect”). Nikolaidis et al. [11] also found a decrease in the sex gap with increasing race 339 distance from half-marathon to marathon and to 100-km ultra-marathon races. In contrast, Coast et 340 al. [20] found, more than 10 years earlier, an increasing sex gap with increasing running distances. 341 However, these authors had restricted their analysis to world-best running performances at 342 distances from 100 m to 200 km, and results might thereby have been biased by the selection of 343 mostly top athletes. Furthermore, the authors indicate that their results might have been confounded 344 by the reduced number of women in longer-distance events. The question of whether the sex gap 345 depends on race distance thus requires further research. 346 Figure 6. Sex differences by calendar year and country in 50-mile and 100-mile ultra-marathons. Curves represent fitted values. For 50-mile races, South Africa was combined with other countries. USA = United States of America, CAN = Canada, GBR = Great Britain, RSA = Republic of South Africa, W = women, M = men. Sex differences (%) in performance were defined as 100 × (women’s race time–men’s race time)/men’s race time. In contrast to 50-mile races, in 100-mile races, the age-related downward trend in the sex difference reversed, the sex difference again increasing after about 80 years of age. It has to be noted, however, that the number of athletes in the oldest age group was rather small, in particular in 100-mile races. Thus, the increase in the sex gap in 100-mile races could simply be due to chance. Alternatively, however, the possibility of an increasing out-selection of relatively slow men at higher ages, in particular in 100-mile races, cannot be excluded. This does not appear completely implausible as physical performance is predictive of longevity at older ages [42,43], possibly underlying a deficit in high-performing men. However, as the increase of the sex gap at very high ages did not occur in 50-mile races, plain chance appears to be the more plausible explanation. Int. J. Environ. Res. Public Health 2019, 16, 2377 12 of 16 4.2. Relative Sex Difference in Performance Was Smaller in 100- Than in 50-Mile Races Our second hypothesis of the decrease in the sex gap in performance with increasing age being independent from race distance was not confirmed. For both race distances, a decline of the relative sex difference is clearly visible, and the decline is more pronounced in 50-mile races than 100-mile races. One explanation for this finding could be that in extremely long distances, like 100-mile races, there might exist a sex-independent pace limit [44]. This limit might constitute a performance maximum, outweighing sex differences at increasing ages (“ceiling effect”). Nikolaidis et al. [11] also found a decrease in the sex gap with increasing race distance from half-marathon to marathon and to 100-km ultra-marathon races. In contrast, Coast et al. [20] found, more than 10 years earlier, an increasing sex gap with increasing running distances. However, these authors had restricted their analysis to world-best running performances at distances from 100 m to 200 km, and results might thereby have been biased by the selection of mostly top athletes. Furthermore, the authors indicate that their results might have been confounded by the reduced number of women in longer-distance events. The question of whether the sex gap depends on race distance thus requires further research. 4.3. Sex Difference in Performance Approaches a Historical Minimum Over the period of 1953 to 2017 (end of study period), the sex difference decreased across all countries in 100-mile races. In contrast, in 50-mile races, the decline was restricted, across all countries, to the period of about 1985 to 2017, whereas in the preceding period from 1964 on, the sex difference had increased across all countries. For either distance, the range of sex differences between countries exceeds that within countries and is larger for 50- than for 100-mile races, the ranking of the countries being different in 50- and 100-mile races. 4.4. Peak Performance Age Was 33 Years In spite of some outliers at certain ages and in certain countries, based on very few or even single individuals per year of age, as demonstrated in Figures 1 and 2, the model-based peak performance age was 33 years, for both race distances. It has to be noted that due to the age*sex and age*country interaction terms being dismissed during the course of the pre-specified stepwise model-building process, the performance peaks do not show variation across sex and country. The model-based estimate of the age of peak performance of 33 years is in line with the available literature. Most previous studies suggested that the age of peak performance in ultra-marathons lies between 30 to 49 years for men and between 30 to 54 years for women [3,4,13,17,45]. However, our result is more in the suggested range of marathon peak performance age of 25 to 35 years for men and women [11,14–16]. 4.5. Average Performance Worsened Over the Last Three Decades In 50-mile races, the average running speed improved in both sexes from 1964 (start of the study period) up to 1985, subsequently worsening until the end of the study period (2017). In 100-mile races, a decline of the average running speed occurred over the whole study period, i.e., from 1953 to 2017. The performance improvement in 50-mile races up to 1985 is essentially attributable to US races, with relatively low average running speeds (see Figure 3). As about 85% of the races in the data set were from the USA, these particular races have inevitably impacted the overall shape of the performance-over-calendar-year-curve, the model specification not allowing for a country-specific shape due to not considering a country*calendar year interaction term. Had these low-performance US races been omitted from fitting the specified model, this would have resulted in a continuous performance decrease over the whole study period for 50-mile races, as it did for 100-mile races. Historically, the USA was the first country where ultra-marathon running became a popular activity among 436 recreational athletes [46]. It is reasonable to assume that initially, these recreational athletes had engaged preferably in 50-mile rather than 100-mile races, lowering the average performance. Int. J. Environ. Res. Public Health 2019, 16, 2377 13 of 16 For example, 494,414 runners participated in 50-km ultra-marathon races between 1975 and 2016 [4], as compared to only 370,051 runners who participated in 100-km ultra-marathon races between 1959 and 2016 [13]. This is confirmed with the present data, where more than two thirds (70%) of the finishers had participated in 50-mile races. Despite this impact of the US-specific phenomenon in shaping the performance model curve, the general trend visible in the data is a performance decline over the study period. One factor for this general decrease in running speed across calendar years could be the popularity of ultra-marathon races gradually increasing worldwide and the races increasingly attracting recreational (i.e., master) athletes. As a consequence, the average performance would have gradually shifted to lower levels. It can be assumed that 100-mile races have always been less attractive to recreational athletes, as they require a more rigorous preparation than 50-mile races. An example for the more rigorous preparation with increasing race distances is provided by Rüst et al., who compared training characteristics between marathoners and 100-km ultra-marathoners and found that ultra-marathoners completed significantly more hours and kilometers during their training. It is therefore likely that the influx of recreational athletes into 100-mile races has been more gradual than in, for example, 50-mile races, and this trend leads to a non-linearity of the performance trend curve. 4.6. Limitations and Strength A limitation of the present study was that it considered specific ultra-running race distances (50 and 100 miles) and thus, caution would be needed to generalize the findings to ultra-running races of other distances or durations [33]. On the other hand, a strength was the large data set that was available for analysis, which was not restricted to top athletes and covered the whole range of performance levels. The depth of the dataset both temporally as well as geographically, and the number of race distances included resulted in the opportunity to provide a comprehensive historic coverage of both 50- and100-mile ultra-marathon results. The data was publicly available, and the collection of data was independent from the data analysis, and therefore the replication of the present analyses is possible. The study findings may aid coaches and ultra-marathon runners in setting long-term training goals based on an athlete’s age and sex. For example, knowledge of the peak age of performance (33 years in both race distances) may influence individuals seeking to race in these distances and recruitment from a coaching perspective. Furthermore, the variation in performance by sex might also influence the training stimulus to be more homogeneous because the sex difference is small (e.g., 100 miles or elder age groups). In addition to performance, the abovementioned practical applications were also relevant from a health perspective. The role of exercise in the prevention and treatment of diseases (e.g., coronary artery disease, stroke, hypertension, diabetes, arthritis, osteoporosis, dyslipidemia, obesity, depression, cancer, and chronic obstructive pulmonary disease) has been well recognized [47]. The findings of the present study can aid physicians prescribing endurance exercise considering sex and age [48]. 5. Conclusions In summary, as age increases, the performance difference between women and men decreases and also becomes lower with longer distances. Based on the model, the overall age of peak performance was 33 years. Future investigations should not just include races measured in miles but also those measured in kilometers, and also analyze time-limited races. Potentially relevant covariates should be considered and whenever possible, acquired prospectively through interviews when they cannot be accessed through administrative databases. If this data had been available, a more comprehensive analysis could have been conducted, including covariate-adjusted time-series analysis. Int. J. Environ. Res. Public Health 2019, 16, 2377 14 of 16 Author Contributions: Conceptualization, B.K., K.J.W., and P.T.N.; methodology, S.D.G.; software, S.D.G.; validation, S.D.G.; formal analysis, S.D.G.; investigation, B.K.; resources, B.K.; data curation, S.D.G.; writing—original draft preparation, K.J.W., B.K., and P.T.N.; writing—review and editing, S.D.G. and T.R.; visualization, S.D.G.; supervision, B.K.; project administration, B.K.; funding acquisition, T.R. Funding: This research received no external funding. Conflicts of Interest: The authors declare no conflict of interest. References 1. Ultra-Marathon, G.S.f. 1st London to Brighton Race (gbr). Available online: http://www.statistik.d-u-v.org/ getresultevent.php?event=46648 (accessed on 14 May 2018). 2. Jampen, S.C.; Knechtle, B.; Rüst, C.A.; Lepers, R.; Rosemann, T. Increase in finishers and improvement of performance of masters runners in the marathon des sables. Int. J. Gen. Med. 2013, 6, 427–438. [PubMed] 3. Knechtle, B.; Rüst, C.A.; Rosemann, T.; Lepers, R. Age-related changes in 100 km ultra-marathon running performance. Age 2012, 34, 1033–1045. [CrossRef] [PubMed] 4. Nikolaidis, P.T.; Knechtle, B. Age of peak performance in 50 km ultramarathoners—Is it older than in marathoners? Open Access J. Sports Med. 2018, 9, 37–45. [CrossRef] [PubMed] 5. Zaryski, C.; Smith, D.J. Training principles and issues for ultra-endurance athletes. Curr. Sports Med. Rep. 2005, 4, 165–170. [CrossRef] [PubMed] 6. Knechtle, B.; Valeri, F.; Zingg, M.A.; Rosemann, T.; Rüst, C.A. What is the age for the fastest ultra-marathon performance in time-limited races from 6 h to 10 days? Age 2014, 36, 9715. [CrossRef] [PubMed] 7. Hoffman, M.D.; Ong, J.C.; Wang, G. Historical analysis of participation in 161km ultramarathons in north america. Int. J. History Sport 2010, 27, 1877–1891. [CrossRef] [PubMed] 8. Cejka, N.; Rüst, C.A.; Lepers, R.; Onywera, V.; Rosemann, T.; Knechtle, B. Participation and performance trends in 100-km ultra-marathons worldwide. J. Sports Sci. 2014, 32, 354–366. [CrossRef] 9. Da Fonseca-Engelhardt, K.; Knechtle, B.; Rüst, C.A.; Knechtle, P.; Lepers, R.; Rosemann, T. Participation and performance trends in ultra-endurance running races under extreme conditions—‘Spartathlon’ versus ‘badwater’. Extrem. Physiol. Med. 2013, 2, 15. [CrossRef] 10. Knechtle, B.; Nikolaidis, P. Ultra-marathon running. Dan. Sportsmed. 2015, 19, 6–10. 11. Nikolaidis, P.T.; Onywera, V.O.; Knechtle, B. Running performance, nationality, sex, and age in the 10-km, half-marathon marathon, and the 100-km ultramarathon IAAF 1999–2015. J. Strength Cond. Res. 2017, 31, 2189–2207. [CrossRef] 12. Rüst, C.A.; Knechtle, B.; Knechtle, P.; Rosemann, T. Comparison of anthropometric and training characteristics between recreational male marathoners and 24-hour ultramarathoners. Open Access J. Sports Med. 2012, 3, 121–129. [PubMed] 13. Nikolaidis, P.T.; Knechtle, B. Performance in 100-km ultra-marathoners—At which age it reaches its peak? J. Strength Cond. Res. 2018. [CrossRef] [PubMed] 14. Knechtle, B.; Nikolaidis, P.T.; Zingg, M.A.; Rosemann, T.; Rüst, C.A. Differences in age of peak marathon performance between mountain and city marathon running—The ‘jungfrau marathon’ in Switzerland. Chin. J. Physiol. 2017, 60, 11–22. [CrossRef] [PubMed] 15. Lara, B.; Salinero, J.J.; Del Coso, J. The relationship between age and running time in elite marathoners is u-shaped. Age 2014, 36, 1003–1008. [CrossRef] [PubMed] 16. Zavorsky, G.S.; Tomko, K.A.; Smoliga, J.M. Declines in marathon performance: Sex differences in elite and recreational athletes. PLoS ONE 2017, 12, e0172121. [CrossRef] [PubMed] 17. Knechtle, B. Ultramarathon runners: Nature or nurture? Int. J. Sports Physiol. Perform. 2012, 7, 310–312. [CrossRef] 18. Peter, L.; Rüst, C.A.; Knechtle, B.; Rosemann, T.; Lepers, R. Sex differences in 24-hour ultra-marathon performance—A retrospective data analysis from 1977 to 2012. Clinics 2014, 69, 38–46. [CrossRef] 19. Bam, J.; Noakes, T.D.; Juritz, J.; Dennis, S.C. Could women outrun men in ultramarathon races? Med. Sci. Sports Exerc. 1997, 29, 244–247. [CrossRef] 20. Coast, J.R.; Blevins, J.S.; Wilson, B.A. Do gender differences in running performance disappear with distance? Can. J. Appl. Physiol. 2004, 29, 139–145. [CrossRef] Int. J. Environ. Res. Public Health 2019, 16, 2377 15 of 16 21. Hoffman, M.D. Ultramarathon trail running comparison of performance-matched men and women. Med. Sci. Sports Exerc. 2008, 40, 1681–1686. [CrossRef] 22. Lepers, R.; Sultana, F.; Bernard, T.; Hausswirth, C.; Brisswalter, J. Age-related changes in triathlon performances. Int. J. Sports Med. 2010, 31, 251–256. [CrossRef] [PubMed] 23. Rüst, C.A.; Knechtle, B.; Rosemann, T.; Lepers, R. Analysis of performance and age of the fastest 100-mile ultra-marathoners worldwide. Clinics 2013, 68, 605–611. [CrossRef] 24. Käch, I.W.; Rüst, C.A.; Nikolaidis, P.T.; Rosemann, T.; Knechtle, B. The age-related performance decline in ironman triathlon starts earlier in swimming than in cycling and running. J. Strength Cond. Res. 2018, 32, 379–395. [CrossRef] [PubMed] 25. Knechtle, B.; Nikolaidis, P.T.; König, S.; Rosemann, T.; Rüst, C.A. Performance trends in master freestyle swimmers aged 25–89 years at the fina world championships from 1986 to 2014. Age 2016, 38, 1–8. [CrossRef] [PubMed] 26. Knechtle, B.; Nikolaidis, P.T.; Rosemann, T.; Rüst, C.A. Performance trends in 3000 m open-water age group swimmers from 25 to 89 years competing in the fina world championships from 1992 to 2014. Res. Sports Med. 2017, 25, 67–77. [CrossRef] [PubMed] 27. Senefeld, J.; Joyner, M.J.; Stevens, A.; Hunter, S.K. Sex differences in elite swimming with advanced age are less than marathon running. Scand. J. Med. Sci. Sports 2016, 26, 17–28. [CrossRef] 28. Knechtle, B.; Nikolaidis, P.T.; Rosemann, T.; Rust, C.A. Performance trends in age group breaststroke swimmers in the fina world championships 1986–2014. Chin. J. Physiol. 2016, 59, 247–259. [CrossRef] 29. Unterweger, C.M.; Knechtle, B.; Nikolaidis, P.T.; Rosemann, T.; Rüst, C.A. Increased participation and improved performance in age group backstroke master swimmers from 25–29 to 100–104 years at the fina world masters championships from 1986 to 2014. SpringerPlus 2016, 5, 645. [CrossRef] 30. Knechtle, B.; Nikolaidis, P.T.; Rosemann, T.; Rüst, C.A. Performance trends in master butterfly swimmers competing in the fina world championships. J. Hum. Kinet. 2017, 57, 199–211. [CrossRef] 31. Nikolaidis, P.T.; Knechtle, B. Performance trends in individual medley events during fina world master championships from 1986 to 2014. J. Sports Med. Phys. Fitness 2018, 58, 690–698. 32. Knechtle, B.; Valeri, F.; Nikolaidis, P.T.; Zingg, M.A.; Rosemann, T.; Rüst, C.A. Do women reduce the gap to men in ultra-marathon running? SpringerPlus 2016, 5, 672. [CrossRef] 33. Sousa, C.V.; da Silva Aguiar, S.; Rosemann, T.; Nikolaidis, P.T.; Knechtle, B. American masters road running records—The performance gap between female and male age group runners from 5 km to 6 days running. Int. J. Environ. Res. Public Health 2019, 16, 2310. [CrossRef] [PubMed] 34. Ultra-Marathon, G.S.f. Available online: https://statistik.d-u-v.org/ (accessed on 21 May 2018). 35. R Development Core Team (2008). A language and environment for statistical computing. R Foundation for Statistical Computing V, Austria. Available online: http://www.R-project.org (accessed on 29 April 2019). 36. Francis, A. 85-Year-Old Sets New World Records in 24-Hour Race. Available online: https://runningmagazine. ca/sections/runs-races/85-year-old-sets-new-world-records-in-24-hour-race/ (accessed on 19 April 2019). 37. Harman, S.M.; Metter, E.J.; Tobin, J.D.; Pearson, J.; Blackman, M.R. Longitudinal effects of aging on serum total and free testosterone levels in healthy men. J. Clin. Endocrinol. Metab. 2001, 86, 724–731. [CrossRef] [PubMed] 38. Goodpaster, B.H.; Park, S.W.; Harris, T.B.; Kritchevsky, S.B.; Nevitt, M.; Schwartz, A.V.; Simonsick, E.M.; Tylavsky, F.A.; Visser, M.; Newman, A.B. The loss of skeletal muscle strength, mass, and quality in older adults: The health, aging and body composition study. J. Gerontol. A Biol. Sci. Med. Sci. 2006, 61, 1059–1064. [CrossRef] [PubMed] 39. Dhillon, R.J.S.; Hasni, S. Pathogenesis and management of sarcopenia. Clin. Geriatr. Med. 2017, 33, 17–26. [CrossRef] [PubMed] 40. Knechtle, B.; Rüst, C.A.; Knechtle, P.; Rosemann, T. Does muscle mass affect running times in male long-distance master runners? Asian J. Sports Med. 2012, 3, 247–256. [CrossRef] 41. Brown, J.C.; Harhay, M.O.; Harhay, M.N. Sarcopenia and mortality among a population-based sample of community-dwelling older adults. J. Cachexia Sarcopenia Muscle 2016, 7, 290–298. [CrossRef] 42. Cardiology A.C.O. Higher Fitness Level Can Determine Longer Lifespan After Age 70: For Older People, Fitness May be More Informative Than Traditional Cardiovascular Risk Factors. Available online: www. sciencedaily.com/releases/2019/03/190306081829.htm (accessed on 19 April 2019). Int. J. Environ. Res. Public Health 2019, 16, 2377 16 of 16 43. Mandsager, K.; Harb, S.; Cremer, P.; Phelan, D.; Nissen, S.E.; Jaber, W. Association of cardiorespiratory fitness with long-term mortality among adults undergoing exercise treadmill testing. JAMA Netw. Open 2018, 1, e183605. [CrossRef] 44. Hubble, C.; Zhao, J. Gender differences in marathon pacing and performance prediction. J. Sports Anal. 2016, 2, 19–36. [CrossRef] 45. Hoffman, M.D.; Wegelin, J.A. The western states 100-mile endurance run: Participation and performance trends. Med. Sci. Sports Exerc. 2009, 41, 2191–2198. [CrossRef] 46. Haberman, A. Escape and pursuit: Contrasting visions of the 1970s long-distance running boom in american popular culture. Sport Hist. Rev. 2017, 48, 1–17. [CrossRef] 47. Elrick, H. Exercise is medicine. Phys. Sportsmed. 1996, 24, 72–76. [CrossRef] [PubMed] 48. Crump, C.; Sundquist, K.; Sundquist, J.; Winkleby, M.A. Exercise is medicine: Primary care counseling on aerobic fitness and muscle strengthening. J. Am. Board Fam. Med. 2019, 32, 103–107. [CrossRef] [PubMed] © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Women Reduce the Performance Difference to Men with Increasing Age in Ultra-Marathon Running.

07-04-2019

Waldvogel, Karin J,Nikolaidis, Pantelis T,Di Gangi, Stefania,Rosemann, Thomas,Knechtle, Beat

eng

PMC5330462

0% 0% 0% 0% 0% 0% 4min 4min 4min 4min 4min 4min 25% 25% 25% 25% 25% 25% 33% VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min Swimmer 1 13,5 414,9 394,9 6,3 1,0 66,8 51,0 2366,6 1946,4 35,9 0,8 143,9 53,1 2260,6 1982,3 34,3 0,9 147,2 53,6 Swimmer 2 11,7 368,5 343,3 5,8 0,9 64,5 67,9 2692,6 2466,1 42,1 0,9 172,4 68,4 2665,1 2464,4 41,6 0,9 180,9 65,1 Swimmer 3 14,3 397,4 395,0 7,4 1,0 86,5 68,7 3025,4 2568,5 56,0 0,8 167,9 70,3 2852,3 2554,5 52,8 0,9 172,7 73,8 Swimmer 4 24,1 394,4 582,3 7,0 1,5 91,3 69,9 2792,3 2385,0 49,9 0,9 185,6 69,6 2788,8 2422,6 49,8 0,9 196,0 68,0 Swimmer 5 9,6 312,7 299,6 5,6 1,0 71,6 63,8 2718,5 2428,0 48,5 0,9 166,0 68,1 2817,2 2577,6 50,3 0,9 166,0 66,4 Swimmer 6 9,8 328,3 263,7 5,9 0,8 63,6 61,8 2330,0 2095,9 41,6 0,9 162,9 64,8 2309,1 2114,8 41,2 0,9 159,3 62,9 Swimmer 7 22,9 815,1 691,0 12,7 0,9 92,1 56,6 2653,0 2091,1 41,5 0,8 153,3 58,0 2589,3 2140,4 40,5 0,8 152,3 59,0 Swimmer 8 14,2 517,2 455,2 7,2 0,9 82,3 47,4 2281,8 1980,4 31,7 0,9 127,6 43,3 2237,2 1915,4 31,1 0,9 149,8 45,4 Swimmer 9 15,1 502,9 426,8 7,7 0,9 90,0 66,8 3044,8 2577,4 46,8 0,8 169,8 75,7 3152,1 2759,7 48,5 0,9 175,3 72,9 Swimmer 10 11,9 376,3 262,8 6,3 0,7 69,7 69,8 2931,6 2152,7 48,9 0,7 154,8 64,8 2654,7 2055,4 44,2 0,8 152,4 66,1 MEAN 14,7 442,8 411,5 7,2 0,94 77,8 62,4 2683,7 2269,2 44,3 0,8 160,4 63,6 2632,7 2298,7 43,4 0,87 165,2 63,3 DP 5,0 146,4 137,9 2,1 0,21 11,7 8,1 280,8 241,5 7,2 0,1 16,3 9,6 294,3 291,3 7,1 0,0 15,9 8,7 0% 0% 0% 0% 0% 0% 4min 4min 4min 4min 4min 4min 25% 25% 25% 25% 25% 25% 33% VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min Swimmer 1 13,79 370,58 429,88 5,61 1,16 75,67 72,13 3042,59 2688,52 46,10 0,88 167,33 68,19 2841,20 2584,59 43,05 0,91 169,58 66,71 Swimmer 2 13,83 377,43 357,56 5,90 0,94 61,13 62,04 2915,39 2420,11 45,55 0,83 164,58 64,55 2970,72 2546,70 46,42 0,86 172,42 63,15 Swimmer 3 15,27 444,79 430,23 8,24 0,97 89,00 73,64 2933,25 2724,70 54,32 0,93 174,83 78,97 2951,16 2794,82 54,65 0,95 182,58 81,70 Swimmer 4 9,69 291,91 285,48 5,21 0,98 87,00 77,52 3174,44 2659,72 56,69 0,84 185,50 76,54 3020,31 2606,85 53,93 0,86 190,67 76,28 Swimmer 5 10,38 299,00 328,25 5,34 1,11 88,29 66,05 2871,12 2620,54 51,27 0,91 171,17 73,61 3058,88 2791,46 54,62 0,91 177,33 70,25 Swimmer 6 10,31 365,65 283,20 6,53 0,77 73,67 57,87 2346,35 1970,56 41,90 0,84 168,17 60,74 2335,05 2032,72 41,70 0,87 167,33 61,31 Swimmer 7 12,41 395,16 316,19 6,17 0,80 66,58 62,40 2775,27 2241,20 43,36 0,81 157,08 62,40 2775,27 2241,20 43,36 0,81 157,08 64,71 Swimmer 8 11,49 376,70 337,57 5,23 0,90 63,45 61,64 2895,43 2467,70 40,21 0,85 163,58 62,78 2883,27 2540,75 40,05 0,88 169,55 55,62 Swimmer 9 13,05 413,18 380,95 6,36 0,93 63,22 84,00 3100,94 2855,07 47,71 0,92 171,75 85,25 3128,77 2975,14 48,13 0,95 181,75 80,19 Swimmer 10 11,07 292,81 239,71 4,88 0,83 61,57 68,26 2801,06 2270,28 46,68 0,81 165,00 74,16 2868,83 2448,80 47,81 0,85 165,00 70,57 MEAN 12,13 362,72 338,90 5,95 0,94 72,96 68,56 2885,58 2491,84 47,38 0,86 168,90 70,72 2883,35 2556,30 47,37 0,89 173,33 69,05 DP 1,84 52,47 62,50 0,97 0,12 11,51 8,21 228,04 270,59 5,30 0,05 7,66 8,23 220,18 274,12 5,49 0,04 9,84 8,43 0% 0% 0% 0% 0% 0% 4min 4min 4min 4min 4min 4min 25% 25% 25% 25% 25% 25% 33% VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min Swimmer 1 13,09 449,30 376,76 6,81 0,84 69,75 76,43 3315,86 2873,08 50,24 0,87 173,75 83,92 3311,52 2986,70 50,17 0,90 180,67 80,47 Swimmer 2 15,24 406,82 407,61 6,36 1,01 65,60 65,13 2980,44 2590,57 46,57 0,87 172,50 66,83 3024,01 2706,99 47,25 0,90 177,83 67,25 Swimmer 3 11,29 362,87 281,86 6,72 0,77 74,00 89,56 3154,63 2796,33 58,42 0,89 173,58 85,79 3078,74 2725,10 57,01 0,89 170,08 90,05 Swimmer 4 25,91 390,33 599,12 6,97 1,54 86,50 81,64 3234,78 2775,31 57,76 0,86 192,18 79,62 3135,62 2755,10 55,99 0,88 199,00 82,27 Swimmer 5 10,35 290,60 320,72 5,19 1,09 94,33 72,33 2981,41 2766,05 53,24 0,93 170,92 75,07 3029,07 2825,73 54,09 0,93 181,92 75,96 Swimmer 6 13,05 377,45 360,18 6,74 0,95 57,36 70,04 2541,92 2295,68 45,39 0,90 168,75 72,87 2566,93 2343,23 45,84 0,91 174,00 68,29 Swimmer 7 14,72 417,38 381,27 6,52 0,91 80,83 78,58 2990,85 2552,96 46,73 0,85 171,00 79,29 3034,43 2599,56 47,41 0,86 171,83 79,37 Swimmer 8 10,85 370,64 304,82 5,15 0,82 61,42 73,38 3113,18 2614,83 43,24 0,84 175,00 71,76 3079,26 2571,28 42,77 0,84 175,00 68,62 Swimmer 9 15,85 504,46 434,19 7,76 0,86 81,80 73,23 3155,90 2788,84 48,55 0,88 176,08 74,65 3156,15 2828,80 48,56 0,90 177,00 77,78 Swimmer 10 15,60 361,68 327,73 5,98 0,91 69,38 76,25 3062,68 2417,72 50,62 0,79 156,08 73,19 3124,15 2464,11 51,64 0,79 161,18 74,44 MEAN 14,60 393,15 379,42 6,42 0,97 74,10 75,66 3053,16 2647,14 50,08 0,87 172,98 76,30 3053,99 2680,66 50,07 0,88 176,85 76,45 DP 4,46 57,25 90,46 0,80 0,22 11,65 6,69 211,40 186,55 5,09 0,04 8,79 5,81 191,12 189,78 4,60 0,04 9,77 7,17 97,5% MLSS 100% MLSS 102.5% MLSS 33% 33% 33% 33% 33% 50% 50% 50% 50% 50% 50% 66% 66% 66% 66% 66% 66% 75% 75% 75% 75% VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg 2242,3 1985,4 34,0 0,9 147,4 51,8 2189,9 1911,6 33,2 0,9 146,3 51,4 2166,1 1877,4 32,8 0,9 145,3 54,7 2295,8 1984,6 34,8 2510,0 2341,2 39,2 0,9 180,9 59,4 2528,6 2289,3 39,5 0,9 184,6 64,4 2563,7 2355,8 40,1 0,9 187,3 63,0 2536,6 2342,3 39,6 2909,6 2641,3 53,9 0,9 175,8 71,4 2852,2 2533,6 52,8 0,9 174,8 71,6 2892,9 2561,3 53,6 0,9 175,5 73,3 2922,2 2567,6 54,1 2742,7 2410,8 49,0 0,9 194,3 68,9 2750,0 2411,7 49,1 0,9 196,1 66,3 2664,6 2334,6 47,6 0,9 196,0 69,9 2749,6 2420,5 49,1 2805,7 2551,3 50,1 0,9 167,0 64,5 2708,3 2469,3 48,4 0,9 172,9 66,4 2631,2 2435,9 47,0 0,9 162,4 69,4 2705,0 2536,2 48,3 2265,5 2094,3 40,5 0,9 165,2 65,0 2265,1 2112,8 40,4 0,9 166,7 65,3 2281,2 2098,1 40,7 0,9 168,0 63,7 2265,8 2055,8 40,5 2571,1 2130,9 40,2 0,8 154,1 58,5 2500,9 2054,4 39,1 0,8 152,0 59,4 2545,0 2094,9 39,8 0,8 152,9 60,0 2598,2 2122,9 40,6 2295,8 1993,4 31,9 0,9 151,8 46,3 2233,3 1991,7 31,0 0,9 152,0 48,6 2256,0 2034,1 31,3 0,9 150,2 49,2 2442,2 2121,6 33,9 3032,5 2594,0 46,7 0,9 174,6 69,5 2925,8 2486,8 45,0 0,8 172,3 76,2 3114,5 2659,5 47,9 0,9 176,1 78,0 3143,1 2668,4 48,4 2714,3 2040,1 45,2 0,8 154,8 74,1 2840,1 2177,3 47,3 0,8 166,1 77,1 2823,4 2218,2 47,1 0,8 170,7 80,4 2808,5 2240,5 46,8 2609,0 2278,3 43,1 0,9 166,6 62,9 2579,4 2243,9 42,6 0,9 168,4 64,7 2593,9 2267,0 42,8 0,9 168,4 66,2 2646,7 2306,0 43,6 278,7 259,6 7,1 0,1 14,9 8,9 276,3 225,0 7,1 0,0 15,4 9,4 301,4 246,8 7,1 0,0 16,2 10,0 276,3 236,8 6,7 33% 33% 33% 33% 33% 50% 50% 50% 50% 50% 50% 66% 66% 66% 66% 66% 66% 75% 75% 75% 75% VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg 2827,32 2531,27 42,84 0,90 171,08 70,12 2952,59 2634,14 44,74 0,89 173,83 72,13 3053,25 2729,82 46,26 0,89 176,33 71,96 3048,10 2712,34 46,18 2901,74 2497,48 45,34 0,86 173,58 61,99 2882,23 2439,64 45,03 0,85 173,08 64,26 2913,49 2506,80 45,52 0,86 173,50 62,34 2881,68 2458,01 45,03 2916,23 2759,68 54,00 0,95 184,42 80,49 2830,92 2656,11 52,42 0,94 184,50 80,77 2733,79 2567,16 50,63 0,94 183,58 80,27 2722,05 2554,29 50,41 2948,36 2630,34 52,65 0,89 194,33 71,52 2857,92 2509,36 51,03 0,88 191,75 73,06 3013,99 2767,21 53,82 0,92 194,33 70,29 2864,99 2645,72 51,16 3001,33 2678,50 53,60 0,89 179,17 66,70 2882,75 2558,43 51,48 0,89 176,92 69,73 2923,41 2651,43 52,20 0,91 181,75 67,52 2803,30 2547,33 50,06 2337,86 2056,19 41,75 0,88 168,67 64,50 2287,93 2030,48 40,86 0,89 167,00 67,65 2291,83 2052,55 40,93 0,90 164,50 58,16 2432,14 1961,31 43,43 2725,63 2331,59 42,59 0,86 160,92 64,43 2728,58 2283,50 42,63 0,84 160,33 66,33 2665,91 2264,53 41,65 0,85 161,25 65,34 2684,29 2230,57 41,94 2626,70 2226,30 36,48 0,85 158,82 55,94 2666,27 2336,44 37,03 0,88 164,83 59,03 2739,24 2380,83 38,04 0,87 169,33 60,46 2889,50 2428,55 40,13 3051,70 2886,70 46,95 0,95 181,67 75,43 2840,37 2635,80 43,70 0,93 172,92 81,84 2903,35 2788,16 44,67 0,96 177,83 82,47 2966,25 2797,18 45,63 2807,02 2385,71 46,78 0,85 165,00 72,35 2831,07 2387,76 47,18 0,84 165,00 79,77 2819,78 2460,82 47,00 0,87 165,00 75,89 2845,98 2376,35 47,43 2814,39 2498,38 46,30 0,89 173,77 68,35 2776,06 2447,17 45,61 0,88 173,02 71,46 2805,80 2516,93 46,07 0,90 174,74 69,47 2813,83 2471,17 46,14 210,19 253,88 5,75 0,04 11,19 7,10 189,52 196,27 4,97 0,03 9,59 7,58 219,36 236,47 5,08 0,04 10,20 8,26 170,97 244,14 3,70 33% 33% 33% 33% 33% 50% 50% 50% 50% 50% 50% 66% 66% 66% 66% 66% 66% 75% 75% 75% 75% VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg R HR VE VO2 VCO2 VO2/Kg ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg --- bpm l/min ml/min ml/min ml/min/Kg 3341,42 3049,45 50,63 0,91 183,17 79,72 3241,26 2841,47 49,11 0,88 182,50 88,40 3390,00 3025,39 51,36 0,89 186,42 87,28 3336,51 2976,60 50,55 2966,54 2700,27 46,35 0,91 182,75 64,41 2880,63 2480,72 45,01 0,86 183,92 66,15 2862,99 2559,54 44,73 0,89 184,00 66,40 2914,14 2638,48 45,53 3150,33 2791,58 58,34 0,89 175,00 98,06 3153,99 2841,98 58,41 0,90 178,25 96,75 3024,82 2645,98 56,02 0,87 181,92 99,41 3139,50 2688,20 58,14 3247,06 2903,55 57,98 0,89 202,80 82,47 3023,98 2639,39 54,00 0,87 202,25 80,56 3082,09 2687,66 55,04 0,87 202,33 83,27 3076,27 2708,17 54,93 3062,80 2841,17 54,69 0,93 177,92 72,78 3043,06 2734,27 54,34 0,90 181,64 73,50 2894,09 2661,58 51,68 0,92 181,67 71,91 2854,38 2614,29 50,97 2520,16 2271,71 45,00 0,90 174,83 72,94 2489,02 2299,62 44,45 0,92 180,00 77,87 2545,57 2345,17 45,46 0,92 182,58 78,72 2566,66 2301,03 45,83 3008,38 2585,44 47,01 0,86 178,75 79,14 2987,48 2434,69 46,68 0,81 175,33 78,73 2983,64 2491,55 46,62 0,83 178,67 84,52 3107,48 2616,32 48,55 3048,61 2472,87 42,34 0,81 175,00 72,59 3131,50 2568,81 43,49 0,82 175,00 76,23 3181,40 2618,63 44,19 0,82 175,00 72,99 3085,42 2539,00 42,85 3147,78 2884,41 48,43 0,92 175,67 88,40 3208,21 3026,99 49,36 0,94 181,08 85,39 3000,76 2775,28 46,17 0,93 179,25 84,11 3040,35 2808,09 46,77 3130,26 2504,56 51,74 0,80 162,33 73,02 2993,56 2446,53 49,48 0,82 163,45 74,63 3077,50 2499,22 50,87 0,82 165,42 76,75 3175,85 2576,76 52,49 3062,34 2700,50 50,25 0,88 178,82 78,35 3015,27 2631,45 49,43 0,87 180,34 79,82 3004,28 2631,00 49,21 0,88 181,73 80,54 3029,66 2646,69 49,66 220,58 238,56 5,44 0,04 10,20 9,58 215,56 227,35 4,86 0,05 9,68 8,59 219,84 183,94 4,33 0,04 9,31 9,35 210,15 175,88 4,69 75% 75% 100% 100% 100% 100% 100% 100% MÉDIA MÉDIA R HR VE VO2 VCO2 VO2/Kg R HR VO2max VO2 %VO2max HR --- bpm l/min ml/min ml/min ml/min/Kg --- bpm ml/kg/min bpm 0,9 149,0 55,4 2331,0 1998,7 35,3 0,9 150,1 54,93 34,31238 62,5% 147,0 0,9 186,1 64,7 2613,1 2435,2 40,8 0,9 185,3 52,00 40,42342 77,7% 182,5 0,9 176,8 74,1 2902,7 2518,9 53,8 0,9 177,8 61,42 53,85525 87,7% 174,5 0,9 197,4 68,0 2735,6 2393,9 48,9 0,9 197,4 68,06 49,03992 72,1% 194,7 0,9 165,1 61,6 2579,7 2323,8 46,1 0,9 169,6 60,69 48,3817 79,7% 167,0 0,9 168,1 66,0 2271,0 2064,6 40,6 0,9 168,3 50,22 40,78513 81,2% 165,5 0,8 153,6 59,1 2598,9 2102,9 40,6 0,8 152,5 52,57 40,30458 76,7% 153,0 0,9 160,0 46,8 2378,0 2023,6 33,0 0,9 159,8 44,45 31,99293 72,0% 150,2 0,8 177,1 78,8 3203,6 2650,9 49,3 0,8 179,2 52,52 47,50891 90,5% 174,9 0,8 170,6 70,4 2674,2 1983,9 44,6 0,7 165,9 52,20 46,30187 88,7% 162,2 0,87 170,4 64,5 2628,8 2249,7 43,3 0,86 170,6 54,9 43,3 78,9% 167,1 0,0 14,7 9,3 279,3 243,7 6,5 0,1 14,7 6,7 6,9 8,7% 15,0 75% 75% 100% 100% 100% 100% 100% 100% MÉDIA R HR VE VO2 VCO2 VO2/Kg R HR VO2max VO2 --- bpm l/min ml/min ml/min ml/min/Kg --- bpm ml/kg/min 0,89 176,67 71,12 3086,82 2693,14 46,77 0,87 179,58 54,93 45,1339 82,2% 173,5 0,85 174,08 61,60 2925,87 2462,07 45,72 0,84 177,25 52,00 45,51592 87,5% 172,6 0,94 183,92 79,94 2757,26 2490,42 51,06 0,90 184,17 61,42 52,49911 85,5% 182,6 0,92 197,08 68,78 2795,58 2469,81 49,92 0,88 194,50 68,06 52,74383 77,5% 192,6 0,91 179,75 69,20 2895,19 2655,08 51,70 0,92 184,00 60,69 52,1326 85,9% 178,6 0,81 167,83 58,40 2330,69 1968,72 41,62 0,85 171,67 50,22 41,73944 83,1% 167,9 0,83 157,67 65,58 2701,22 2187,86 42,21 0,81 158,75 52,57 42,53606 80,9% 159,0 0,84 170,58 57,10 2650,55 2241,88 36,81 0,85 168,67 44,45 38,39474 86,4% 166,5 0,94 177,67 83,74 3005,09 2805,42 46,23 0,93 179,33 52,52 46,14607 87,9% 177,6 0,83 165,00 87,70 2881,13 2474,72 48,02 0,86 165,00 52,20 47,2735 90,6% 165,0 0,88 175,03 70,32 2802,94 2444,91 46,01 0,87 176,3 54,9 46,41 84,7% 173,6 0,05 10,91 10,51 213,41 252,47 4,67 0,04 10,5 6,7 4,88 3,8% 9,7 75% 75% 100% 100% 100% 100% 100% 100% MÉDIA R HR VE VO2 VCO2 VO2/Kg R HR VO2max VO2 --- bpm l/min ml/min ml/min ml/min/Kg --- bpm ml/kg/min 0,89 185,67 88,20 3364,26 2997,77 50,97 0,89 188,75 54,93 50,43472 91,8% 183,0 0,91 187,42 66,58 2873,55 2603,78 44,90 0,91 186,42 52,00 45,76404 88,0% 182,1 0,86 184,70 97,72 3019,46 2588,33 55,92 0,86 183,55 61,42 57,46418 93,6% 178,2 0,88 200,00 95,59 3145,77 2845,19 56,17 0,90 203,17 68,06 55,98362 82,3% 200,2 0,92 184,33 78,63 2881,35 2703,83 51,45 0,94 186,50 60,69 52,9239 87,2% 180,7 0,90 183,75 84,03 2579,67 2307,31 46,07 0,89 185,33 50,22 45,43349 90,5% 178,5 0,84 181,17 84,72 3034,16 2530,18 47,41 0,83 180,92 52,57 47,20184 89,8% 176,8 0,82 175,00 76,50 3103,46 2519,60 43,10 0,81 175,00 44,45 43,14054 97,0% 175,0 0,92 171,33 81,25 3001,93 2786,15 46,18 0,93 163,00 52,52 47,71665 90,9% 174,8 0,81 167,55 76,00 3050,02 2555,47 50,41 0,84 169,33 52,20 51,03662 97,8% 163,6 0,88 182,09 82,92 3005,36 2643,76 49,26 0,88 182,20 54,9 49,71 90,9% 179,3 0,04 9,16 9,37 204,16 195,44 4,49 0,04 11,15 6,7 4,71 4,6% 9,2

Oxygen uptake kinetics and energy system's contribution around maximal lactate steady state swimming intensity.

02-28-2017

Pelarigo, Jailton Gregório,Machado, Leandro,Fernandes, Ricardo Jorge,Greco, Camila Coelho,Vilas-Boas, João Paulo

eng

PMC9876921

1 Vol.:(0123456789) Scientific Reports | (2023) 13:1415 | https://doi.org/10.1038/s41598-023-28398-2 www.nature.com/scientificreports A macro to micro analysis to understand performance in 100‑mile ultra‑marathons worldwide Mabliny Thuany 1, Katja Weiss 2,3, Elias Villiger 4, Volker Scheer 5, Nejmeddine Ouerghi 6,7, Thayse Natacha Gomes 8,9 & Beat Knechtle 2,3* The purposes of this study were (i) to describe differences in participation in 100‑mile ultra‑marathons by continent; (ii) to investigate differences in performance between continents; and (iii) to identify the fastest runners by continent and country. Data from 148,169 athletes (119,408 men), aged 18–81 years, and finishers in a 100‑miles ultra‑marathon during 1870–2020 were investigated. Information about age, gender, origin, performance level (top three, top 10, top 100) was obtained. Kruskal–Wallis tests and linear regressions were performed. Athletes were mostly from America and Europe. A macro‑analysis showed that the fastest men runners were from Africa, while the fastest women runners were from Europe and Africa. Women from Sweden, Hungary and Russia presented the best performances in the top three, top 10 and top 100. Men from Brazil, Russia and Lithuania were the fastest. The lowest performance and participation were observed for runners from Asia. In summary, in 100‑miles ultra‑marathon running, the majority of athletes were from America, but for both sexes and performance levels, the fastest runners were from Africa. On a country level, the fastest women were from Sweden, Hungary and Russia, while the fastest men were from Brazil, Russia and Lithuania. The athletes’ performance is influenced by both individual (e.g., genetic, morphological, training) and environ- mental factors (e.g., coach, family, social characteristics)1,2. Moving beyond the athlete-centered approach, recent studies were developed to understand the role of the environment in the athlete’s performance3–5. The ‘birthplace effect’ has been largely studied in team sports, such as soccer6, ice hockey7, basketball8, volleyball9, and handball10. Furthermore, among individual sports such as running, the interest in understanding the link between the environment and the athletes’ performance has increased in the last years11,12. These interests were associated with the increasing numbers of both runners and running events across the world13, especially after the 1970’s in North America and after the 1980’s in Europe14. There is ample evidence that the fastest long-distance runners, such as marathoners, originate from the Afri- can continent, particularly from Kenya and Ethiopia15,16. This representation is related to a plethora of factors, which include–but are not limited to–physiological characteristics, training, lifestyle behaviors, and motivational factors17,18. Considering Brazil, the Southeast region as the richest region of the country is the region with the highest number of elite long-distance runners in the country19. However, in the context of ultra-marathon running, little is known about where the fastest ultra-marathoners come from. One of the few studies found that Russian and Japanese were the fastest for the 100-km ultra- marathon race distance16,20. Similar results were found by Cejka et al.21, where most of the finishers in 100-km ultra-marathons were from Europe, but Japanese runners were the fastest. On the other hand, data covering OPEN 1Centre of Research, Education, Innovation and Intervention in Sport (CIFI2D), Faculty of Sport, University of Porto, Porto, Portugal. 2Medbase St. Gallen Am Vadianplatz, Vadianstrasse 26, 9001 St. Gallen, Switzerland. 3Institute of Primary Care, University of Zurich, Zurich, Switzerland. 4Klinik Für Innere Medizin, Kantonsspital St. Gallen, St. Gallen, Switzerland. 5Ultra Sports Science Foundation, 109 Boulevard de L’Europe, 69310 Pierre-Benite, France. 6University of Jendouba, High Institute of Sport and Physical Education of Kef, UR13JS01, 7100 Kef, Tunisia. 7Faculty of Medicine of Tunis, University of Tunis El Manar, Rabta Hospital, LR99ES11, 1007 Tunis, Tunisia. 8Post-Graduation Program of Physical Education, Department of Physical Education, Federal University of Sergipe, São Cristóvão Sergipe 49100-000, Brazil. 9Department of Physical Education and Sport Sciences, University of Limerick, Limerick V94T9PX, Ireland. *email: beat.knechtle@hispeed.ch 2 Vol:.(1234567890) Scientific Reports | (2023) 13:1415 | https://doi.org/10.1038/s41598-023-28398-2 www.nature.com/scientificreports/ 96,036 athletes (88,286 men and 7,750 women) finishing the oldest 100-km ultra-marathon in the world (‘100 km Lauf Biel’ in Switzerland), showed that Switzerland, Germany, and France were the countries with the highest number of participants throughout the history of the race22. These initial insights suggest that the place of the competition and the countries’ economic indicators are related to these results. To date, most studies regarding participation and performance trends have shown an increase in the number of finishers in the last decades’23,24, especially in events hosted in the USA, where most of the events were tak- ing place24,25. Notwithstanding the relevance of these studies, it is important to present a more generalized view by using a macro-level approach investigating the between-continents differences, followed by a micro-level analysis within-country. It has also been shown that athletes from specific countries improved their performance over years in specific races such as the ‘Spartathlon’26 whereas in other races such as the ‘100 km Lauf Biel’, the performance of athletes from specific countries decreased22. Generally, the fastest athletes were able to improve their performance across years in the world’s most famous ultra-marathon races such as the ‘Spartathlon’26 and the ‘Comrades’27. In this sense, the purposes of this study were (i) to verify the participation of athletes by continent and country in 100-mile ultra-marathons (161-km) performed between 1870 and 2020, (ii) to compare the athletes’ perfor- mance between continents and countries, (iii) to identify the fastest athletes by country and by continent and, (iv) to investigate the trend in performance over years of the athletes from the fastest countries. Previous studies showed that North America and Europe presented higher numbers of ultra-marathon events worldwide (431 and 455, respectively)25 compared to Asia (139), Africa (59), Australia (35), South America (18), as well as a higher participation in 161-km ultra-marathons28. For another way, Russian athletes were the fastest in long-distance running races such as the ‘Comrades Marathon’29, and in 100-km ultra-marathon running races20. Based on it, we hypothesized that (i) the highest number of athletes would be found on the American continent, especially in USA and Canada, while Russian runners would be the fastest and (ii) the fastest runners would be able to improve their performance over time. Methods Ethical approval. The study was performed following the Declaration of Helsinki, and the institutional review board of St Gallen, Switzerland, approved this study (EKSG 01/06/2010). Since the study involved the analysis of publicly available data, the requirement for informed consent was waived. Participants were not iden- tified during the data management and during all sections of the manuscript. Design and sample. Data used in the present study was obtained from the website of the ‘Deutsche Ultra- marathon-Vereinigung’ DUV (https:// stati stik.d- u-v. org/ getev entli st. php) and corresponded to the officially available results of the participants enrolled in 100-mile (161 km) ultra-marathon race events held during 1870– 2020 for both genders. The available information included the year of the event, race distance, year of birth, gen- der, general ranking, country, team, mean running speed, and race time. Based on this information, we clustered the athletes by country and then by continent (e.g., Africa, America, Asia, Oceania, and Europe). Performance levels were categorized based on the general classification considering the top three, top 10 and top 100 for both genders. Exclusion criteria were: athletes aged below 18 years, athletes clustered in countries with less than 10 participations (when comparison within a continent was made), mean running speed higher than 20 km/h, and missing information about the country of origin. Statistical analysis. Descriptive information was presented in mean (standard deviation), minimum (min), maximum (max) values, and frequency (%). Data normality was formally tested using the Kolmogorov– Smirnov test. Based on the athletes’ performance, runners were classified considering their ranking position (top 100, top 10, and top three), for both genders. Following, athletes were clustered based on their birthplace (e.g., country and continent). To compare the performance level according to the continent, we performed the Kruskal–Wallis test, followed by the post-hoc test adjusted by the number of comparisons. A simple linear regression was performed to verify the performance trends. Running speed (miles/h) was the outcome variable, while the time (year) was considered as a predictor. The regression was performed considering both, the total sample and the ranking position (top three, top 10, and top 100), as well as the three best countries over time for both genders. For the total sample regression analysis, we considered data from the 1970s, since a running boom in high and middle-income countries from this decade was verified 30. For the three best countries according to the performance level, we considered data from 2010. Data analysis was performed in the SPSS 26, and the significance level was set at 0.05. Results A total of 148,169 athletes aged 18–81 years and of both genders (women = 28,761; men = 119,408), finished at least one 100-mile ultra-marathon during 1870–2020. Athletes came from 113 countries from all five continents. Most of them were from the American continent (72.7%), followed by Europe (15.9%), Asia (5.6%), Africa (3.7%), and Oceania (2.1%). Running speed distribution between athletes showed a similar pattern for all continents. Most of the runners presented mean values of 5–7 miles/h, and a small number of athletes presented running speeds higher than 10 miles/h. Mean running speed was 6.12 miles/h for the total sample (women = 5.96 miles/h; men = 6.21 miles/h). Participation and performance by continent—a macro‑analysis. Figure 1 presents the distribu- tion of the athletes, according to their continent of origin into different performance groups (top three, top 10, top 100, and all athletes) for both genders. Most women in all the performance levels were from the American 3 Vol.:(0123456789) Scientific Reports | (2023) 13:1415 | https://doi.org/10.1038/s41598-023-28398-2 www.nature.com/scientificreports/ continent, followed by Europe. Visual differences are presented for the top three and top 10, where Oceania has a higher number of athletes compared to the African continent (e.g., top three and top 10). Considering athletes clustering by continent, significant differences in performance were shown (H(4) = 1766.22; p < 0.001). The descriptive analysis showed that African women achieved the highest mean running speed (6.89 ± 1.04 miles/h), followed by athletes from Europe (6.45 ± 1.38 miles/h), America (5.89 ± 1.01 miles/h), Oceania (5.85 ± 1.37 miles/h), and Asia (5.17 ± 1.49 miles/h). Significant differences were observed between most of the continents, except between America and Oceania (p-adjusted = 0.178). For men, for all the performance levels (top three, top 10, top 100 and all athletes), most of the athletes were from both the American and the European continent. The lowest percentages of participation were found for Oceania. Considering the performance by continent, athletes from Africa presented the highest mean values for running speed (7.42 ± 1.33 miles/h), followed by Europe (6.53 ± 1.53 miles/h), Oceania (6.39 ± 1.59 miles/h), America (6.13 ± 1.17 miles/h), and Asia (5.34 ± 1.47 miles/h). Differences in the performance were found between conti- nents (H(4) = 859.43; p < 0.001), with significant differences between all of them. Participation and performance by country—a micro‑analysis. For the top three athletes, a total of 36 and 65 countries were listed for both women and men, respectively. Most of the athletes were from the USA (62.1% and 56.9% for women and men, respectively). Similarly, the highest number of athletes from the USA were observed in the top 10 (62.6% and 58.4% for women and men, respectively) and top 100 (72.4% and 66% for women and men, respectively). Therefore, within the European continent, the majority of the athletes were from Germany and the Great-Britain for all performance levels. Table 1 presents the results for the comparison between performance levels for both genders. Significant dif- ferences were observed for both genders and all performance levels. For men athletes, the fastest runners were from the African continent. For the top three women, the fastest ones were from Europe, while for the top 10 and top 100 the fastest were Africans. Table 2 presents the descriptive results for women runners, considering athletes from the fastest country in each continent and the performance level (e.g., top three, top 10 and top 100). For the African continent, the fastest women were from South Africa, in the three performance groups. For the other continents, there was no verified pattern of association between the countries and performance levels, nonetheless, the highest running speeds were observed in the European continent when compared to the other continents. Among the top three, top 10 and top 100, women from Sweden, Hungary and Russia achieved the fastest running speeds. Men descriptive results for athletes from the fastest countries in each continent are presented in Table 3. For the top three, the fastest athletes were from America (Brazil: 9.54 ± 1.75 miles/h). In the top 10 and top 100, athletes from Europe (e.g., Russia and Lithuania) presented the highest mean running speed compared to the other continents. Figure 2 presents the performance trend results for both genders considering all the sample. For all perfor- mance groups and both genders, performance decreased over time (p < 0.001). Figure 1. Percentage of women and men participation according to performance level. Table 1. Descriptive results (mean ± SD for miles/h) for performance (speed mean, mile/h) according to gender and performance level. SD standard deviation. a difference for Asia; bdifference for America; cdifference for Oceania; ddifference for Europe. Men Women Top three Top 10 Top 100 Top three Top 10 Top 100 Africa 9.37 (1.7)abcd 8.64 (1.6)abcd 7.45 (1.3)abcd 7.52 (1.7) 7.39 (1.5)bc 6.91 (1.0)abc America 7.93 (1.5)a 7.28 (1.4)a 6.29 (1.2)ab 7.37 (1.7)d 6.80 (1.4)a 6.04 (1.1)ab Asia 7.10 (1.8) 6.59 (1.7) 5.76 (1.5) 7.29 (2.1)d 6.44 (1.8)a 5.71 (1.5) Europe 8.29 (1.9)abc 7.64 (1.8)abc 6.74 (1.5)ac 8.29 (1.8) 7.38 (1.7)abc 6.72 (1.4)abcd Oceania 8.05 (1.7)a 7.27 (1.6)a 6.45 (1.6)a 7.45 (1.5) 6.71 (1.4) 5.91 (1.4)a 4 Vol:.(1234567890) Scientific Reports | (2023) 13:1415 | https://doi.org/10.1038/s41598-023-28398-2 www.nature.com/scientificreports/ Figure 3 presents the linear regression results for performance across time considering athletes from the best countries in the top three, top 10, and top 100, respectively. For women, runners from Russia, Hungary, and Finland were the fastest, while runners from Brazil, Russia, and Lithuania were the fastest among men. The time frame considered was the last 10 years (from 2010), in which a significant performance decline was shown for women from Sweden (r2 = 0.10; p < 0.001; 95%CI = − 0.18– − 0.08). Based on the r2 values, the magnitude of the performance decline was about 0.10miles/h across the years. For men, performance decline was showed for Brazilian (r2 = 0.27; p < 0.001; 95%CI = − 0.47 – − 0.18) and Russian runners (r2 = 0.02; p = 0.017; 95%CI = − 0.18 – − 0.018), in which a decline of 0.27 miles/h and 0.02 miles/h was showed across the years, respectively. Athletes Table 2. Women descriptive results (min, mean ± SD, max) for the best country in each continent, based on performance level. Min–Minimum value; Max–Maximum value. SD standard deviation. We only considered the best country for each continent. Countries with total athletes below 10 were not considered. Country Total athletes Speed (miles/h) (Min–Max) Speed (miles/h) (Mean ± SD) Top three Africa South Africa 35 3.08–10.81 7.52 (1.72) America Canada 82 4.35–10.00 7.44 (1.23) Asia Philippines 13 5.14–6.36 5.71 (0.36) Europe Sweden 17 5.52–10.60 8.31 (1.47) Oceania Australia 33 4.46–10.44 7.55 (1.35) Top 10 Africa South Africa 150 2.41–10.81 7.39 (1.52) America USA 2658 2.80–19.42 6.82 (1.50) Asia Japan 26 4.08–11.59 8.13 (2.02) Europe Hungary 11 7.59–11.12 9.07 (1.06) Oceania New Zealand 48 3.41–9.64 6.87 (1.55) Top 100 Africa South Africa 713 2.41–10.81 6.92 (1.05) America El Salvador 11 5.47–8.06 6.65 (0.83) Asia Cyprus 67 3.38–10.37 7.21 (1.52) Europe Russia 55 5.19–10.29 8.05 (1.21) Oceania Australia 445 3.38–10.44 5.93 (1.36) Table 3. Men descriptive results (min, mean ± SD, max) for the best country in each continent, based on a performance level. Min–Minimum value; Max–Maximum value. SD standard deviation. We only considered the best country for each continent. Countries with total athletes below 10 were not considered. Country Total athletes Speed (miles/h) (Min–Max) Speed (miles/h) (Mean ± SD) Top three Africa South Africa 411 3.14–13.47 9.35 (1.73) America Brazil 16 4.97–11.65 9.54 (1.75) Asia Taiwan 62 3.36–10.24 8.25 (1.47) Europe Hungary 32 5.73–11.95 9.48 (1.60) Oceania Australia 374 4.19–13.21 8.07 (1.72) Top 10 Africa South Africa 1166 2.52–13.47 8.63 (1.67) America Brazil 28 4.97–11.65 8.77 (1.91) Asia Taiwan 214 3.36–10.49 7.97 (1.22) Europe Russia 100 5.03–14.03 9.36 (1.51) Oceania New Zealand 239 3.38–12.39 7.31 (1.60) Top 100 Africa Zimbabwe 12 4.61–11.88 8.01 (2.60) America Brazil 64 3.43–11.65 7.61 (2.00) Asia Israel 23 4.01–10.49 7.50 (1.78) Europe Lithuania 13 4.31–12.53 8.41 (2.05) Oceania Australia 1850 3.42–1.21 6.51 (1.60) 5 Vol.:(0123456789) Scientific Reports | (2023) 13:1415 | https://doi.org/10.1038/s41598-023-28398-2 www.nature.com/scientificreports/ from Lithuania presented a visual increase in running speed over the last 10 years, however, a non-significant effect was shown. Discussion The purpose of this study was to identify where the fastest runners in 100 miles ultra-marathons come from, considering the performance level. The main results showed that (i) for both genders and all performance levels, most of the athletes were from the American and European continents; (ii) a macro-analysis showed that the fastest men were from Africa, while the fastest women were from Europe and Africa; (iii) women from Sweden, Figure 2. Linear regression results, considering all sample (A) Women top three; (B) Women top 10; (C) Women top 100; (D) Men top three; (E) Men top 10; (F) Men Top 100). Figure 3. (A) Sweden women top three; (B) Hungary women top 10; (C) Russian women top 100; (D) Brazilian men top 3; (E) Russian men top 10; (F) Lithuanian men top 100. 6 Vol:.(1234567890) Scientific Reports | (2023) 13:1415 | https://doi.org/10.1038/s41598-023-28398-2 www.nature.com/scientificreports/ Hungary and Russia presented the best performance in top three, top 10 and top 100; (iv) men runners from Brazil, Russia and Lithuania were the fastest in top three, top 10 and top 100, respectively; and (v) the lowest performance and participation were achieved by athletes from Asia. Participation and performance by continent—a macro‑analysis. The first important finding was that most of the finishers were from America, but African runners were the fastest when analysis was performed by continent. These findings confirm our hypothesis. For both genders, the highest runner’s frequency was from the American continent, especially from the USA. Similar results were reported by Hoffman23 in 100 miles (161 km) ultra-marathon running competition in North America. The authors showed that from 1977 to 2008, the number of annual finish rates increased, but no improvements in performance were verified23. The results of the present study can be related to the American cultural and fitness revolutions of the 1960s and 1970s, which included a ‘running boom’13. Specifically for the ultra-marathon races, a historical perspective ˗ the USA was one of the main ultra-marathon birthplaces around the world31—can influence the highest number of race events and athletes from these countries. In addition, the increase in participation among women and older athletes can be associated with this result32. A previous report covering approximately 107 million race results from 1986 to 2018 showed that the USA was the country with the highest number of runners, but with the slowest athletes33. Accordingly, the highest proportions of women participants were from USA and Canada, while Switzerland and Italy were the countries with the lowest women participation33. The highest participation of these countries can be related to the highest number of ultra-marathon events performed in these countries25. Regarding continents, 443 ultra-marathon events were developed in America, where 431 are situated in North America. Following, Europe hosted 421 events. For data used in this study, more than half of the race events were performed in EUA (58.5%), with about 20% performed in Great Britain (5.6%), Australia (4.8%), South Africa (4.7%), Canada (4.6%), and Germany (3.4%). Besides the higher events performed in-locus, the athletes’ socioeconomic characteristics can also be related to running participation34,35. Athletes from a high-income country can present a better contextual indicator for traveling and participating in remote events36. In another way, the lowest performances showed for athletes from the USA can be related to changes in running motivation across the years. As shown in studies that include short to long-distance events, the psychological, social, and physical are the main reasons for running37–39, especially in non-professional athletes. The macro-analysis has shown that the African continent presented the best mean values for running speed. This is an interesting finding, considering that African athletes are the strongest in long-distance running such as half-marathon and marathon15,40. However, these runners are from Kenya and Ethiopia, different from the present results, where most of them are from South Africa. These results are similar to findings in a previous report covering 85% of ultra-running events worldwide during 1996–2018, including trail runs, mountain runs and road runs. South Africa was the country with the fastest athletes, with a running pace of 10:36 min/mile, followed by Sweden (11:56 min/mile), Germany (12:01 min/mile), Netherlands (12:41 min/mile), and Great- Britain (12:44 min/mile), while the slowest were from Argentina (15:20 min/mile), Mexico (15:30 min/mile) and Malaysia (15:55 min/mile). These results were also associated with findings that countries from Asia presented the poorest performance36. Participation and performance by country—a micro‑analysis. The micro-analysis showed that ath- letes from Sweden, Hungary, and Russia presented the best performance in the top three, top 10, and top 100 for women, and those from Brazil, Russia, and Lithuania were the fastest in the top three, top 10, and top 100 for men. We hypothesized that the fastest runners would originate from Russia, but the results partially disagree. These differences can be related to the methodological approach for the present study where we present the data for performance level (i.e., top three, top 10, top 100). For example, Nikolaidis et al.16, in a study including athletes ranked in World Athletics (i.e., IAAF) during 1999–2015, showed that among women, athletes from Russia were faster than athletes from France and Germany in ultra-marathon events. Similar results were shown in athletes who finished a 100-km ultra-marathon between 1959 and 201620, when considering the top 10 by nationality, runners from Russia and Hungary were the fastest. Men from Brazil in the top three are untypical considering previous studies16,20,41. Notwithstanding, regard- ing the increase in runner’s participants and race events across the country42,43, few studies were developed to understand the participation and performance in ultra-running events44,45. The country characteristics, which include variations in weather, altimetry, nutritional habits, cultural aspects, and lifestyle among the regions, should be investigated in future studies to understand the association with performance in ultramarathon events. Considering both, the total sample and the fastest countries, performance decreased over time for both gen- ders and performance levels. These results are similar to previous findings23,32. These results can be linked to the changes in the runner’s profile (e.g., intrapersonal motivation, training background, previous experience)46–48, and event characteristics (weather, altimetry). Differently, athletes from Lithuania showed an increase in performance over the last few years. This increase was not statistically significant, however, factors that explain these results can be related to the low number of athletes over the years, which can bias the results. The generalization of the present findings need to be considered carefully. In another way, the decrease in performance in other countries is according to previous findings, showing that countries have slowed down over the last 10 years and that those with have slowed down most are among the slowest in the rankings36. More and more is known about the factors that predispose to achieve outstanding results in ultramarathon running, but without pointing to the most important ones49. Gajda et al. considered the success in ultra-mara- thons as a complex multifactorial cause and called them the “mosaic theory”. Among the factors that guarantee success they mention genetic factors such as the presence of haplogroup H mtDNA (subgroup HV0a1, belong- ing to the HV cluster), characterizing athletes with the greatest endurance49. Normal resistance to pain is also 7 Vol.:(0123456789) Scientific Reports | (2023) 13:1415 | https://doi.org/10.1038/s41598-023-28398-2 www.nature.com/scientificreports/ important50. However, none of these factors isolated guarantees success for an individual athlete or a particular nation in ultra-marathons. Additional investigations considering the environment (natural, built, and social) are necessary. Limitations and strengths. Limitations of the study are related to the nature of the data used. The accu- racy of the data (e.g., the race distance in each event, the accuracy of the information in the first years, and the information about the birthplace of the athletes), as well as the missing data in specific time frames, and sam- ple size variability between countries need to be considered. These limitations are challenged to be solved. To reduce the bias, the countries with a total number of athletes below 10, as well as the regression analysis with data before the 1970s were not considered. In addition, information about individual (e.g., training volume and intensity, running experience, running strategies) and contextual factors (i.e., number of competitions, econom- ical support, cultural aspects and race course, different elevation changes, weather) are unavailable. Individual and contextual characteristics are helpful to deeply understand runners’ profile, as well as to deeply understand the impact of hosting events’ effect, and the characteristics of the race course for runners’ performance. The role of the individual characteristics for ultramarathon performance was previously investigated, however, little information is available about the role of social, economic, cultural, and geographical characteristics to increase the participation, as well as the role of the participation in performance outcomes. Future studies need to con- sider data triangulation, including the place in which competitions are performed, the participation and the performance outcomes, adopting different strategies regarding the performance level and sample size required within countries. Finally, we did not control for the migration or multiple events participation across the years, that is, an athlete can have moved to represent another country than his/her home country or take part in more than one event over the years. In another way, we presented a detailed analysis, considering both macro-and micro-level approaches. Since the highest number of ultramarathon events are performed in North America and Europe, even though considering mean values the fastest are from Africa and Europe, the practical application of the present study includes supports local sport police programs to increase the availability of events in countries which they are underrepresented. Athletes living in countries with a higher number of events present a higher participation rate, since the costs (travel, host) are lower51. Additionally, being familiar with local characteristics (language, cultural habits, and weather) is associated with performance improvement51. Conclusion For the 100-miles ultra-marathons, most of the athletes were from the American and European continent, despite the fastest being from Africa. A micro-analysis showed that European countries’ (Sweden, Hungary, and Russia) were the best for women, while for men, Brazil, Russia and Lithuania were the fastest in the top three, top 10, and top 100. Regarding the best countries, a decrease in performance was shown over time, except for athletes from Lithuania. These results can be used to public policies to provide the highest number of race events among countries, especially those in Asia and Oceania, which showed the lowest engagement. Data availability The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request. Received: 24 May 2022; Accepted: 18 January 2023 References 1. Tucker, R. & Collins, M. What makes champions? A review of the relative contribution of genes and training to sporting success. Br. J. Sports Med. 46(8), 555–561. https:// doi. org/ 10. 1136/ bjspo rts- 2011- 090548 (2012). 2. Yan, X., Papadimitriou, I., Lidor, R. & Eynon, N. Nature versus nurture in determining athletic ability. Med. Sport Sci. 61, 15–28. https:// doi. org/ 10. 1159/ 00044 5238 (2016). 3. Cote, J., Macdonald, D., Baker, J. & Abernethy, B. When, “where” is more important than “when”: Birthplace and birthdate effects on the achievement of sporting expertise. J. Sports Sci. 24(10), 1065–1073. https:// doi. org/ 10. 1080/ 02640 41050 04324 90 (2006). 4. Rees, T. et al. The great British medalists project: A review of current knowledge on the development of the world’s best sporting talent. Sports Med. 46(8), 1041–1058. https:// doi. org/ 10. 1007/ s40279- 016- 0476-2 (2016). 5. Rossing, N. N., Stentoft, D., Flattum, A., Côté, J. & Karbing, D. S. Influence of population size, density, and proximity to talent clubs on the likelihood of becoming elite youth athlete. Scand. J. Med. Sci. Sports 28(3), 1304–1313. https:// doi. org/ 10. 1111/ sms. 13009 (2018). 6. Smith, K. L. & Weir, P. L. Female youth soccer participation and continued engagement: Associations with community size, com- munity density, and relative age. Front. Sports Act Liv. 2, 552597. https:// doi. org/ 10. 3389/ fspor. 2020. 552597 (2020). 7. Wattie, N., Schorer, J. & Baker, J. Seeing the forest but not the trees: Heterogeneity in community size effects in Canadian ice hockey players. J. Sports Sci. 36(4), 436–44. https:// doi. org/ 10. 1080/ 02640 414. 2017. 13134 44 (2018). 8. Leite, N., Arede, J., Shang, X., Calleja-González, J. & Lorenzo, A. The influence of contextual aspects in talent development: Inter- action between relative age and birthplace effects in NBA-drafted players. Front Sports Act Living. 3, 642707. https:// doi. org/ 10. 3389/ fspor. 2021. 642707 (2021). 9. Hancock, D. J., Coutinho, P., Côté, J. & Mesquita, I. Influences of population size and density on birthplace effects. J Sports Sci. 36(1), 33–8. https:// doi. org/ 10. 1080/ 02640 414. 2016. 12766 14 (2018). 10. Rossing, N. N., Nielsen, A. B., Elbe, A. M. & Karbing, D. S. The role of community in the development of elite handball and football players in Denmark. Eur. J. Sport Sci. 16(2), 237–45. https:// doi. org/ 10. 1080/ 17461 391. 2015. 10094 92 (2016). 11. Santos, P. A. et al. Human development index and the frequency of nations in athletics world rankings. Sport Sci. Health. 15(2), 393–398. https:// doi. org/ 10. 1007/ s11332- 019- 00529-1 (2019). 12. Truyens, J., Bosscher, V. D., Heyndels, B. & Westerbeek, H. A resource-based perspective on countries’ competitive advantage in elite athletics. Int. J. Sport Policy Polit. 6(3), 459–489. https:// doi. org/ 10. 1080/ 19406 940. 2013. 839954 (2014). 8 Vol:.(1234567890) Scientific Reports | (2023) 13:1415 | https://doi.org/10.1038/s41598-023-28398-2 www.nature.com/scientificreports/ 13. Scheerder, J., Breedveld, K., Borgers, J. Running across Europe—The Rise and Size of One of the Largest Sport Markets. Scheerder J, Breedveld K, Borgers J (eds) (Palgrave Macmillan, New York, 2015. 14. van Mechelen, W. Running injuries. A review of the epidemiological literature. Sports Med. 14(5), 320–35. https:// doi. org/ 10. 2165/ 00007 256- 19921 4050- 00004 (1992). 15. Hamilton, B. East African running dominance: What is behind it?. Br. J. Sports Med. 34, 391–394 (2000). 16. Nikolaidis, P. T., Onywera, V. & Knechtle, B. Running performance, nationality, sex, and age in the 10-km, half-marathon, mara- thon, and the 100-km ultramarathon IAAF 1999–2015. J. Strength Cond. Res. 31(8), 2189–2207. https:// doi. org/ 10. 1519/ JSC. 00000 00000 001687 (2017). 17. Onywera, V., Scott, R. A., Boit, M. K. & Pitsiladis, Y. P. Demographic characteristics of elite Kenyan endurance runners. J. Sports Sci. 24(4), 415–22. https:// doi. org/ 10. 1080/ 02640 41050 01890 33 (2006). 18. Wilber, R. & Pitsiladis, Y. Kenyan and Ethiopian distance runners: What makes them so good?. Int. J. Sports Physiol. Perform. 7(2), 92–102 (2012). 19. Thuany, M., Gomes, T. N., Souza, R. F., Almeida, M. (2021) Onde estão os melhores corredores do Brasil? Revista Brasileira de Ciência e Movimento: In press. 20. Knechtle, B., Nikolaidis, P. T. & Valeri, F. Russians are the fastest 100-km ultra-marathoners in the world. PLoS ONE. 13(7), e0199701. https:// doi. org/ 10. 1371/ journ al. pone. 01997 01 (2018). 21. Cejka, N. et al. Participation and performance trends in 100-km ultra-marathons worldwide. J. Sports Sci. 32(4), 354–66. https:// doi. org/ 10. 1080/ 02640 414. 2013. 825729 (2014). 22. Knechtle, B., Scheer, V., Nikolaidis, P. T. & Sousa, C. V. Participation and performance trends in the oldest 100-km ultramarathon in the world. Int. J. Environ. Res. Public Health. 17, 1719. https:// doi. org/ 10. 3390/ ijerp h1705 1719 (2020). 23. Hoffman, M. Performance trends in 161-km ultramarathons. Int. J. Sports Med. 31, 31–37 (2010). 24. Hoffman, M., Ong, J. & Wang, G. Historical analysis of participation in 161 km ultramarathons in North America. Int. J. Hist. Sport. 27(11), 1877–1891 (2010). 25. Worlds’ marathon. (2021) Ultra marathon destinations by continent [cited 2022 26 February]. Available from: https:// world smara thons. com/c/ runni ng/ desti natio ns/ ultra_ marat hon. 26. Knechtle, B. et al. From athens to sparta—37 years of spartathlon. Int. J. Environ. Res. Public Health. 18(9), 4914. https:// doi. org/ 10. 3390/ ijerp h1809 4914 (2021). 27. Nikolaidis, P. T., Knechtle, B., Vancini, R., Gomes, M. & Sousa, C. Participation and performance in the oldest ultramarathon- Comrades marathon 1921–2019. Int. J. Sports Med. 42(7), 638–644. https:// doi. org/ 10. 1055/a- 1303- 4255 (2021). 28. Gerosa, D., Rüst, C. A., Rosemann, T. & Knechtle, B. Participation and performance trends in 161km ultra-marathons in terms of nationality: A retrospective data analysis of worldwide participation from 1998–2011. J. Human Sport Exerc. 9(2), 592–615. https:// doi. org/ 10. 14198/ jhse. 2014. 92. 01 (2014). 29. Nikolaidis, P. T. & Knechtle, B. Russians are the fastest and the youngest in the “Comrades marathon”. J. Sports Sci. 37(12), 1387–1392. https:// doi. org/ 10. 1080/ 02640 414. 2018. 15599 79 (2019). 30. Haberman, A. L. Thousands of solitary runners come together: Individualism and communitarianism in the 1970s running boom. J. Sport Hist. 44(1), 35–49. https:// doi. org/ 10. 5406/ jspor thist ory. 44.1. 0035 (2017). 31. Noakes, T. D. The limits of endurance exercise. Basic Res. Cardiol. 101(5), 408–417. https:// doi. org/ 10. 1007/ s00395- 006- 0607-2 (2006). 32. Hoffman, M. D. & Wegelin, J. A. The western states 100-mile endurance run: Participation and performance trends. Med. Sci. Sports Exerc. 41(12), 2191–2198. https:// doi. org/ 10. 1249/ MSS. 0b013 e3181 a8d553 (2009). 33. Andersen, J. J. (2019) The State of Running 2019: RunRepeat [17 August 2020]. Available from: https:// runre peat. com/ state- of- runni ng. 34. Breuer, C., Hallmann, K. & Wicker, P. Determinants of sport participation in different sports. Manag. Leis. 16(4), 269–286. https:// doi. org/ 10. 1080/ 13606 719. 2011. 613625 (2013). 35. Thuany, M., Gomes, T. N., Almeida, M. (2020) Running around the country: Uma análise do fenômeno out running entre corre- dores brasileiros. 36. RunRepeat. (2021) The State of Ultra Running 2020 [cited 2021 24 September 2021]. Available from: https:// runre peat. com/ state- of- ultra- runni ng. 37. Poczta, J. & Malchrowicz-Mośko, E. Running as a form of therapy socio-psychological functions of mass running events for men and women. Int. J. Environ. Res. Public Health. 15(10), 1–15. https:// doi. org/ 10. 3390/ ijerp h1510 2262 (2018). 38. Krouse, R., Ransdell, L., Lucas, S. & Pritchard, M. Motivation, goal orientation, coaching, and training habits of women ultrarun- ners. J. Strength Cond. Res. 1, 2835–2842. https:// doi. org/ 10. 1519/ JSC. 0b013 e3182 07e964 (2011). 39. Goodsell, T. L., Harris, B. D. & Bailey, B. W. Family status and motivations to run: A qualitative study of marathon runners. Leis. Sci. 35(4), 337–352. https:// doi. org/ 10. 1080/ 01490 400. 2013. 797326 (2013). 40. Noble, T. J. & Chapman, R. F. Marathon specialization in elites: A head start for africans. Int. J. Sports Physiol. Perform. 13(1), 102–106. https:// doi. org/ 10. 1123/ ijspp. 2017- 0069 (2020). 41. Nikolaidis, P. T. & Knechtle, B. Russians are the fastest and the youngest in the “Comrades marathon”. J. Sports Sci. 37(12), 1387–1392. https:// doi. org/ 10. 1080/ 02640 414. 2018. 15599 79 (2019). 42. Salgado, J. V. V. & Chacon-Mikahil, M. P. T. Corrida de rua: Análise do crescimento do número de provas e de praticantes. Conexões 4(1), 90–98 (2006). 43. de Atletismo, FP. (2019) Demonstrativo de Corridas de Rua nos Últimos Anos no Estado de São Paulo [cited 2019 Novembro]. Available from: www. atlet ismof pa. org. br/ source/ Demon strat ivo- de- Corri das- de- Rua- nos- Ultim os- Anos- no- Estado- de- Sao- Paulo- 2017. pdf. 44. Alves, D., Cruz, R., Lima-Silva, A., Domingos, P., Bertuzzi, R., Osiecki, R., et al. (2019) Are experienced and high-level race walking athletes able to match pre-programmed with executed pacing?. Braz. J. Med. Biol. Res. 52(6) https:// doi. org/ 10. 1590/ 1414- 431X2 01985 93. 45. Fonseca, F. S., Cavalcante, J. A. M., Almeida, L. S. C. & Fialho, J. V. A. P. Análise do perfil sociodemográfico, motivos de adesão, rotina de treinamento e acompanhamento profissional de praticantes de corrida de rua. Rev. Bras. Ciênc. Mov. 27, 189–198 (2019). 46. Waśkiewicz, Z., Nikolaidis, P., Chalabaev, A., Rosemann, T. & Knechtle, B. Motivation in ultra-marathon runners. Psychol. Res. Behav. Manag. 12, 31–37. https:// doi. org/ 10. 2147/ prbm. s1890 61 (2019). 47. Knechtle, B. Ultramarathon runners: Nature or nurture?. Int. J. Sports Physiol. Perform. 7(4), 310–312. https:// doi. org/ 10. 1123/ ijspp.7. 4. 310 (2012). 48. Knechtle, B. & Nikolaidis, P. Physiology and pathophysiology in ultra-marathon running. Front. Physiol. 9, 634. https:// doi. org/ 10. 3389/ fphys. 2018. 00634 (2018). 49. Gajda, R. et al. To be a champion of the 24-h ultramarathon race. If not the heart mosaic theory?. Int. J. Environ. Res. Public Health 18(5), 1–25 (2021). 50. Gajda, R., Walasek, P. & Jarmuszewski, M. Right knee—The weakest point of the best ultramarathon runners of the world? A case study. Int. J. Environ. Res. Public Health 17(16), 1–11. https:// doi. org/ 10. 3390/ ijerp h1716 5955 (2020). 51. Jürgens, D. et al. An analysis of participation and performance by nationality at “Ironman Switzerland” from 1995 to 2011. J. Sci. Cycl. 1(2), 10–20 (2012). 9 Vol.:(0123456789) Scientific Reports | (2023) 13:1415 | https://doi.org/10.1038/s41598-023-28398-2 www.nature.com/scientificreports/ Author contributions Conceptualization: M.T., B.K.; Data curation: B.K., E.V.; Formal analysis: M.T. Methodology: B.K.; Writing— original draft: M.T., K.W., V.S., N.O., T.N.G., B.K. Funding This research has not received any kind of financial support. All the work was done voluntarily for the prepara- tion of the manuscript. Competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to B.K. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2023

A macro to micro analysis to understand performance in 100-mile ultra-marathons worldwide.

01-25-2023

Thuany, Mabliny,Weiss, Katja,Villiger, Elias,Scheer, Volker,Ouerghi, Nejmeddine,Gomes, Thayse Natacha,Knechtle, Beat

eng

PMC6332098

Molecules 2015, 20, 19002-19013; doi:10.3390/molecules201019002 molecules ISSN 1420-3049 www.mdpi.com/journal/molecules Article The Occurrence of Propyl Lactate in Chinese Baijius (Chinese Liquors) Detected by Direct Injection Coupled with Gas Chromatography-Mass Spectrometry Jihong Wu 1,2, Yang Zheng 1,2, Baoguo Sun 1,2,3, Xiaotao Sun 1,2, Jiyuan Sun 1,2, Fuping Zheng 1,2 and Mingquan Huang 1,2,3,* 1 School of Food and Chemical Engineering, Beijing Technology and Business University, Beijing 100048, China; E-Mails: wujihong12@126.com (J.W.); an2zhengyang@126.com (Y.Z.); sunbg@btbu.edu.cn (B.S.); sunxiaotao@th.btbu.edu.cn (X.S.); sunjinyuan@th.btbu.edu.cn (J.S.); zhengfp@th.btbu.edu.cn (F.Z.) 2 Beijing Key Laboratory of Flavor Chemistry, Beijing Technology and Business University, Beijing 100048, China 3 Beijing Innovation Centre of Food Nutrition and Human Health, Beijing 100048, China * Author to whom correspondence should be addressed; E-Mail: huangmq@th.btbu.edu.cn; Tel./Fax: +86-10-6898-5382. Academic Editor: Luca Forti Received: 1 September 2015 / Accepted: 13 October 2015 / Published: 19 October 2015 Abstract: As one of the oldest distillates in the world, flavor compounds of Chinese Baijiu (Chinese liquor) were extremely complex. Propyl lactate was firstly detected by direct injection and gas chromatography-mass spectrometry (GC-MS) in 72 Chinese Baijius. The objectives were to detect the contents of propyl lactate and evaluate its contribution to the aroma of Chinese Baijiu based on odor activity values (OAVs). The levels of propyl lactate in these distillates were determined by internal standard method and selective ion monitoring (SIM), which ranged from 0.050 to 1.900 mg·L−1 under investigation. Its detection threshold was determined by Three-Alternative Forced-Choice (3-AFC) and curve fitting (CF), which was 0.740 mg·L−1 in 38% ethanol solution. The contribution of propyl lactate on the aroma of these distillate drinks was evaluated by their odor activity values (OAVs), which varied from 0.066 to 4.440. The OAVs of propyl lactate were found to exceed 1 in 13 Chinese Baijius, including 50° Jingzhi Guniang 5 years (4.440), 52° Jingzhi Guniang 10 years (3.024), Jingyanggang (2.568), Xianghe Ronghe Shaofang (2.313), and 1956 Laolang (1.431), which indicated that propyl lactate was one of odor-active components in these Chinese Baijius. OPEN ACCESS Molecules, 2015, 20 19003 Keywords: Chinese Baijiu; propyl lactate; gas chromatography-mass spectrometry (GC-MS); threshold; odor activity values (OAVs) 1. Introduction Chinese Baijiu (Chinese liquor) is one of the oldest distillates in the world, and it is the most popular spirits in China with the annual production of about 4 million metric tons. The general process for the production of Chinese Baijiu is as follows. At first, the grain raw materials, such as wheat, sorghum, corn, rice or glutinous rice, are cooked with steam. And then some saccharification and fermentation agents (“DaQu” or “XiaoQu”) are added into the cooked grain matrix. Finally, the liquors are distilled out with steam from the fermentation products after several months or years of fermentation [1]. The fresh distillates need to be aged for a long time in order to balance the flavors. The final commercialized products are blended with aged distillate drinks, fresh distillate drinks and water based on certain ratios according to different formulations of spirit drinks [2]. Flavor compounds of Chinese Baijiu are extremely complex due to different raw materials, various microorganisms and diverse procedures in different production regions, and a great number of compounds have been studied extensively [3–6], such as esters, alcohols, ketones, acids, and so on. Lactates have been reported to occur in Chinese liquors widely, such as methyl lactate [4], ethyl lactate [7], butyl lactate [8], hexyl lactate [9], isopropyl lactate [10], isobutyl lactate [4] and isoamyl lactate [4]. Lactates are recognized to be important flavor compounds in Chinese Baijiu [11]. Ethyl lactate and butyl lactate were detected in “Gujing” Baijiu during investigation carried out in our lab. In the meantime, lactic acid and propanol were also found in this distillate, which had been reported in other Chinese Baijius [12,13]. It is a little strange that propyl lactate has not been reported in Chinese Baijiu up to now. The same are other alcoholic beverages, except distilled Calvados [14] and Chinese rice wine [15]. Meanwhile, the odor properties of propyl lactate and its contribution to aroma of these spirits are also not known. We checked the newest edition (2011 edition) of NIST (National Institute of Standards and Technology) library and found that the mass spectrogram of propyl lactate was not included in this library. The occurrence of propyl lactate in Chinese Baijiu might be overlooked since the identification of volatile components was usually performed by searching NIST library. The objective of this work were (i) the detection of the occurrence of propyl lactate in 72 Chinese Baijius and (ii) the evaluation of its contribution to the aroma of Chinese Baijius based on odor activity values (OAVs). 2. Results and Discussion 2.1. Identification of Propyl Lactate in Distillate Samples The mass spectra of an unknown compound shown at 17.4 min in sample 33 and propyl lactate standard were presented as Figure 1, and the differential spectrum of them was located at the bottom, which indicated the unknown compound shown at 17.4 min being matched with propyl lactate. The two magnified TICs of the unknown compound in sample 33 (a TIC) and propyl lactate standard (b TIC) at 17.4 min were shown in Figure 2. These two peaks were completely overlapped. Molecules, 2015, 20 19004 Figure 1. TIC of propyl lactate in GC-MS. The mass spectra of an unknown compound shown at 17.4 min in sample 33 and propyl lactate standard are presented. The differential spectrum of them is located at the bottom. Figure 2. The magnifying TIC of sample 33 and propyl lactate at 17.4 min. The two magnified TICs are shown, one is the unknown compound in sample 33 (a TIC), and the other is propyl lactate standard (b TIC) at 17.4 min. 2.2. Quantitative Analysis of Propyl Lactate in Distillate Samples The concentrations of propyl lactate were determined in these spirit drinks by the internal standard method with the selected ions monitoring mode of GC-MS. The LOD and the LOQ of the method were 0.025 mg·L−1 and 0.050 mg·L−1, respectively. The internal standard curve equations included three equations in different concentration ranges, including y = 4.4587x + 0.0112 (0.050 < y < 0.60), y = 1.9987x + 0.2673 (0.60 < y < 1.00) and y = 2.1972x + 0.0569 (1.00 < y < 4.50). The corresponding correlation coefficients (R2) were 0.9994, 0.9917 and 0.9993, respectively. The quantitative results of propyl lactate were shown in Table 1. The content of propyl lactate in other 10 Chinese Baijiu, numbered 73 to 82, were also included in Table 1, which our lab had reported before [16]. Molecules, 2015, 20 19005 Table 1. The concentrations of propyl lactate in 82 Chinese Baijiu samples (Detection threshold 0.740 mg·L−1). No. Sample Name (Alcohol % by Volume) Manufacturer Concentrations of Propyl Lactate (mg·L−1) OAV of Propyl Lactate 1 Guizhou Yuanjiang Chenniang 18 Years (52°) Kweichow Moutai Co., Ltd. 0.064 ± 0.003 0.086 2 1956 Laolang (53°) Sichuan Langjiu Group Co., Ltd. 1.059 ± 0.008 1.431 3 Jiabin Lang (50°) Sichuan Langjiu Group Co., Ltd. 0.060 ± 0.002 0.081 4 Xiaohutuxian (52°) Guizhou Xingyi Yunfeng Co., Ltd. a tr - b 5 Tianzhilan (42°) Jiangsu Yanghe Distillery Co., Ltd tr - b 6 Xiangquan (54°) Jiugui Liquor Co., Ltd. 0.068 ± 0.001 0.092 7 Jinliufu Sixing (38°) Wuliangye Group 0.064 ± 0.001 0.086 8 Jinliufu Hongsejingdian (38°) Wuliangye Group tr - b 9 Shuanggou Daqu (38°) Shuanggou Distillery 0.091 ± 0.002 0.123 10 Niulanshan Bainian (38°) Shunxin Agriculture Ture tr - b 11 Jianzhuang Chenjiu (52°) Wuliangye Group tr - b 12 Laishigang Chenniang 3 Years (53°) Laishigang Group 0.062 ± 0.001 0.084 13 Zhijiang Zhixin 5 Years (52°) Zhijiang Group 0.081 ± 0.002 0.109 14 Jiannanchun (52°) Sichuan Jiannanchun Jituan Co., Ltd. tr - b 15 Liulingzui 3 (52°) Liulingzui Group 0.100 ± 0.002 0.135 16 Zhonghua Dukang K3 (50°) Luoyang Dukang Holdings Limited Official Website 0.052 ± 0.001 0.070 17 Yangshao Caitaofang Jiuliang Miaopin (52°) Yangshao Co., Ltd. 0.553 ± 0.006 0.747 18 Mianrou Dukang (50°) Luoyang Dukang Holdings Limited Official Website 0.065 ± 0.001 0.088 19 Shamochun Shengshi (42°) Neimenggu Dahekou Co., Ltd. 0.064 ± 0.001 0.086 20 Luzhou LaojiaoTouqu (52°) Luzhou Laojiao Co., Ltd. tr - b 21 Luzhou Laojiao Chentouqu 8 Years (52°) Luzhou Laojiao Co., Ltd. tr - b 22 Hetao Laojiao Jinzun (42°) Hetao Liquor tr - b 23 Yingjia K6 (38°) Yingjia Gongjiu Co., Ltd. 0.075 ± 0.003 0.101 24 Neimenggu Sorghum Blue Era (38°) Neimenggu Dahekou Co., Ltd. 0.068 ± 0.001 0.092 25 Rouhe Shuanggou (42°) Shuanggou Distillery 0.059 ± 0.001 0.080 26 Xinghuacun Baishun (45°) Fenjiu Group 0.094 ± 0.001 0.127 27 Jingjiu Jixing (36°) Wuliangye Group 0.065 ± 0.002 0.088 28 Guizhou Yuanjiang Zhenpin 9 Years (38°) Guizhou Maotai Distillery Group Technology Development Company 0.068 ± 0.002 0.092 Molecules, 2015, 20 19006 Table 1. Cont. No. Sample Name (Alcohol % by Volume) Manufacturer Concentrations of Propyl Lactate (mg·L−1) OAV of Propyl Lactate 29 Guizhou DongcangYuanjiu 30 Years (38°) Zhenpin Jiuye 0.302 ± 0.010 0.408 30 Xifeng Yucang (38°) Xifeng Co., Ltd. 0.056 ± 0.001 0.076 31 Furuiwang (38°) Furui Co., Ltd. 0.079 ± 0.001c 0.107 32 Laobaifen Fengtan 15 Years (38°) Fenjiu Group 0.112 ± 0.001 0.151 33 Honghuaci Erguotou (56°) Sanhe Fucheng Co., Ltd. 0.224 ± 0.042 0.303 34 Jingdu Yujiu (42°) Shuangqinghe Co., Ltd. tr - b 35 Zhoufuji Erguotou (42°) Zhoufuji Co., Ltd. tr - b 36 Mendaolv (62°) Ningheyuan Co., Ltd. tr - b 37 Weirenmin Fuwu (53°) Guizhou Maotai Distillery Group 0.130 ± 0.002 0.176 38 Beijing Erguotou Qinghuaci (52°) Jiuzhongjiu Co., Ltd. 0.064 ± 0.002 0.086 39 Zhougong Baisui (35°) Huangjia Jingdu Co., Ltd. 0.052 ± 0.003 0.070 40 Tianshan Laobing (38°) Tianshan Co., Ltd. tr - b 41 Beijing Erguotou I (56°) Tongquanyong Co., Ltd. tr - b 42 Xianghe Ronghe Shaofang (53°) Kweichow Moutai Co., Ltd. 1.712 ± 0.023 2.313 43 Dajinjiu (42°) Dajin Co., Ltd. tr - b 44 Jingdu Heitan (42°) Huangjia Jingdu Co., Ltd. 0.066 ± 0.003 0.089 45 Wuliang Yuanjiu (50°) Yuqiao Co., Ltd. 0.300 ± 0.054 0.405 46 Guocuijiu (52°) Luzhou Guocui Co., Ltd. 0.052 ± 0.001 0.070 47 Hongdu Tezhen 25 Years (50°) Hongdu Co., Ltd. tr - b 48 Beijing Erguotou II (56°) Shuangqinghe Co., Ltd. 0.075 ± 0.006 0.101 49 Beijing Fangzhuang Erguotou (52°) Fangzhuang Co., Ltd. tr - b 50 Laobeijing Erguotou (41°) Duxing Co., Ltd. tr - b 51 Sichuan Sorghumjiu I (40°) Yaoquan Laojiao Co., Ltd. tr - b 52 Guantoushan (40°) Guantoushan Co., Ltd. 0.093 ± 0.001 0.126 53 Mengguwang (44°) Mengguwang Co., Ltd. 0.090 ± 0.002 0.122 54 Caoyuan Andaqing (62.8°) Andaqing Co., Ltd. 0.100 ± 0.004 0.135 55 Luchun (52°) Luzhou Laojiao Co., Ltd. tr - b 56 Yujingfang Shaojiu (38°) Yujingfang Shaojiu Group 0.058 ± 0.002 0.078 Molecules, 2015, 20 19007 Table 1. Cont. No. Sample Name (Alcohol % by Volume) Manufacturer Concentrations of Propyl Lactate (mg·L−1) OAV of Propyl Lactate 57 Sichuan Sorghum II (42°) Culiangfang Co., Ltd. tr - b 58 Xianli Jianguo 60 Years (50°) MaotaiJiucheng Co., Ltd. tr - b 59 Longfeng (38°) Longfeng Co., Ltd. tr - b 60 Caoyuan Liema (62°) Menggudao Co., Ltd. tr - b 61 Shuijing Kongdong (52°) Liuhuchun Co., Ltd. 0.066 ± 0.001 0.089 62 Gubei Chunliang (42°) Gubeichun Co., Ltd. 0.050 ± 0.001 0.066 63 Banmasuo Chenniao 3 Years (65°) Muniu Co., Ltd. 0.074 ± 0.001 0.100 64 Heitudi (38°) Hecheng Co., Ltd. tr - b 65 Hengshui Laobaigan (39°) Hengshui Laobaigan Co., Ltd. 0.149 ± 0.005 0.201 66 Bancheng Laojiu (42°) Qianlongzui Co., Ltd. 0.075 ± 0.001 0.101 67 Jingyanggang (38°) Jingyanggang Co., Ltd. 1.900 ± 0.002 2.568 68 65667 Troops Tegong T99B (38°) Beimao Co., Ltd. 0.250 ± 0.002 0.338 69 Huanghelong Laoliangfang (52°) Huanghelong Group 0.100 ± 0.001 0.135 70 Guojiao 1573 (52°) Luzhou Laojiao Co., Ltd. tr - b 71 Mengzhilan M6 (40.8°) Jiangsu Yanghe Distillery Co., Ltd 0.176 ± 0.006 0.238 72 Kouzijiao Zhencang 20 Years (41°) Kouzi Yjiuye 0.052 ± 0.002 0.070 73 Moutai (53°) c Kweichow Moutai Co., Ltd. 0.851 ± 0.001 c 1.150 74 Xifeng (55°) c Xifeng Co., Ltd. 0.818 ± 0.011 c 1.105 75 Guojing 1# (65°) c Shandong Bandaojing Co., Ltd. 1.008 ± 0.018 c 1.362 76 Gujing Yuanjiang (65°) c Anhui Gujing Group Co., Ltd. 2.237 ± 0.022 c 3.023 77 Jinshiyuan Yuanjiang (59°) c Jiangsu King’s Luck Brewery Joint-Stock Co., Ltd. 0.932 ± 0.024 c 1.259 78 Jinshiyuan (53°) c Jiangsu King’s Luck Brewery Joint-Stock Co., Ltd. 0.810 ± 0.017 c 1.095 79 Jingzhi Guniang10 Years (52°) c Shandong Jingzhi Liquor Co., Ltd. 3.024 ± 0.025 c 4.086 80 Jingzhi Guniang 5 Years (50°) c Shandong Jingzhi Liquor Co., Ltd. 3.286 ± 0.060 c 4.440 81 Wuyue Duzun (52°) c Taishan Liuor Group Co., Ltd. 0.788 ± 0.006 c 1.065 82 Jiuchao Chenxiang (42°) c Shandong Lanling Meijiu Co., Ltd. 0.910 ± 0.014 c 1.230 a tr: The concentrations of propyl lactate were between 0.025 mg·L−1 and 0.050 mg·L−1; b No OAVs because of the concentrations of propyl lactate was much less than LOQ; c the concentrations of propyl lactate were from the reported article [16]. Molecules, 2015, 20 19008 As shown in Table 1, propyl lactate did occur in all Chinese Baijius under investigation, though the concentrations in some distillate drinks were between 0.050 mg·L−1 (LOQ) and 0.025 mg·L−1 (LOD). The top 5 distillate samples were Jingzhi Guniang 5 years (50°) (3.286 mg·L−1), Jingzhi Guniang 10 years (52°) (3.024 mg·L−1), Gujing Yuanjiang (65°) (2.237 mg·L−1), “Jingyanggang” (38°) (1.900 mg·L−1), and “Xianghe Ronghe Shaofang” (53°) (1.712 mg·L−1). 2.3. Detection Threshold of Propyl Lactate and OAV Analysis The determination of detection threshold of propyl lactate was conducted by the statistical analyses and the Curve Fitting, and the result was shown in Figure 3. Figure 3. Scatter diagram by CF for determination of propyl lactate detection threshold. The X-axis represents the concentration of propyl lactate to the base 10 logarithm (X = LogA) and Y-axis is the correct recognition ratio, which is the ratio of the correct recognition numbers in total recognition numbers ሺY ൌ ୒ሺୡ୭୰୰ୣୡ୲ሻ ୒ሺ୲୭୲ୟ୪୪ሻ ሻ. The calibration curve equation was y = 0.1623x + 0.6877, and the correlation coefficient R2 = 0.9663. The threshold was the corresponding X value (0.740 mg·L−1) when Y = 66.67% [17]. 2.4. Discussion The unknown compound shown at 17.4 min in sample 33 was identified to be propyl lactate based on the comparison of mass spectra and retention time with the standard. Propyl lactate was also discovered in all other samples under investigation by the same method. Propyl lactate was possibly formed through the esterification of lactic acid with propanol, both of which have been reported during the fermentation. There were a great number of lactobacilli during the fermentation process of Chinese spirit drinks, which could convert sugars to lactic acid [18,19]. Higher alcohols could be formed during the fermentation under aerobic condition from sugar or under anaerobic conditions from amino acids [20] since the raw materials, sorghum, rice, sticky rice, wheat and corn, were rich sources of amino acids. Propanol can be produced from threonine by yeast via the Ehrlich metabolic pathway [21]. Small amounts of propanol could also be formed by yeast through reduction of propanal. The esterification of lactic acid with propanol could be taken place directly or catalyzed by esterases during the fermentation and aging process of Chinese liquor production [20]. The esterases might be from yeasts, Molecules, 2015, 20 19009 molds, or bacteria, which existed in “Daqu” or “Xiaoqu” [21]. Fan [22] also reported that the “Daqu” had high esterase activities. The main formation process of propyl lactate was as Figure 4. Figure 4. The pathway of propyl lactate formation by esterase catalyzation. The odor activity values (OAV) equaled to the ratio of the concentration of propyl lactate and its detection threshold value. If a compound has an OAV > 1.0, then it would contribute to the flavor of a product [23]. The OAVs of propyl lactate in 82 Chinese Baijius were listed in Table 1. There were 13 distillate samples with the OAVs of propyl lactate higher than 1, including 50° Jingzhi Guniang 5 years (4.440), 52° Jingzhi Guniang 10 years (3.024), 38° Jingyanggang (2.568), 53° Xianghe Ronghe Shaofang (2.313), and so on, whereas other 69 liquor samples with the OAVs lower than 1. Propyl lactate had a grape-like fruity, milk and ester odor [17]. The results indicated that propyl lactate was one of odor-active components of these 13 Chinese Baijius. Whether propyl lactate was a key odor compound or not based on more experiments and proofs, which we would be to study next. 3. Experimental Section 3.1. Chemicals Chemicals and standards were GC grade with a high purity (>99.0%).The water was boiling for at least 0.5 h and redistilled twice before use. Methyl lactate (PubChem CID: 11040), used as internal standard (IS), and propyl lactate (PubChem CID: 92821) were obtained from Tokyo Chemical Industry CO., Ltd. (Shanghai, China). Absolute ethanol (PubChem CID: 702) was obtained from Merck (Darmstadt, Germany). 3.2. Spirit Drink Samples A total of 72 spirit drinks, shown as Table 1, were obtained from different factories in China, or supermarkets, such as Wal-Mart and Carrefour in Beijing, China. 3.3. Qualitative and Quantitative Analysis by GC-MS 1.0 µL of Chinese spirit drinks was injected and analyzed by GC-MS with the full scan mode, and the occurrence of propyl lactate was confirmed by comparing its retention time and mass spectrum with the standards. Starches amylase sugars lactic acid lactobacilli H3C H C COOH OH H3C H2 C CH2OH threonine yeast Ehrlic metabolic pathway CH H C COOH H3C HO NH2 Propanol H3C H C COOCH2C2H5 OH esterases Propyl lactate Molecules, 2015, 20 19010 The concentrations of propyl lactate in these distillates were determined by the internal standard method with the selected ions monitoring mode of GC-MS. At first, a series of the standard solutions, such as 5.000 mg·L−1, 2.500 mg·L−1, 1.250 mg·L−1, 0.625 mg·L−1, 0.313 mg·L−1, 0.156 mg·L−1, 0.078 mg·L−1, 0.039 mg·L−1 and 0.020 mg·L−1, were prepared with absolute ethanol and analyzed by GC-MS. Then 1.0 mL of each distillate drink sample with 10.0 µL of methyl lactate solution (100.000 mg·L−1) were placed in 72 tightly closed sample vials, numbered 1 to 72, for GC-MS analysis. Finally, the concentrations of propyl lactate were calculated by the software of GC-MS. The conditions of GC-MS (6890A-5975C, Agilent technologies Co., Ltd., Beijing, China) were as follows. GC conditions: DB-FFAP capillary column (30 m × 0.25 mm, 0.25 μm film thickness, Santa Clara, CA, USA); carrier gas, helium, 99.9995%; flow rate, 1.0 mL·min−1; The oven temperature was programmed at 50 °C for 2 min, then raised to 100 °C at 6 °C·min−1, then raised to 170 °C at 3 °C·min−1 for 2 min, and then raised to 200 °C at 10 °C·min−1 for 2 min, and finally raised to 230 °C at 15 °C·min−1 for 5 min; inlet temperature, 250 °C; transfer line temperature, 250 °C; injection volume, 1 µL; split ratio, 20:1. MS conditions: electron ionization source, 70 eV; ion source and quadruple temperatures, 230 °C and 150 °C, respectively; The monitored ions and other parameters of selected-ion-monitoring (SIM) mode were listed in Table 2. Table 2. The monitored ions and other parameters. Compound Mode Mass List or Range Methyl lactate Full Scan 50–500 SIM 45, 75, 89,105 Propyl lactate Full Scan 50–500 SIM 45, 75, 117 3.4. Determination of the Detection Threshold of Propyl Lactate 3-AFC test was recommended as the national standard method for determination of Chinese Baijiu flavors thresholds by GB/T 22366-2008 [16], which is a general guidance for measuring odor, flavor and taste detection threshold, because of its higher efficiency and accuracy. Meanwhile, CF method was adopted as threshold calculation method, which was recommended by ASTM E1432-2004 Standard Practice. So 3-AFC and CF were selected to determine the detection threshold of propyl lactate in Chinese spirit drinks. Most concentrations of alcohol in the distillate samples under investigation were 38 vol% or nearby, so 38% ethanol solution was used as the benchmark. A series of propyl lactate solutions were prepared for sensory evaluation with 38% ethanol solution, such as 0.137 mg·L−1 (A1), 0.412 mg·L−1 (A2), 1.235 mg·L−1 (A3), 3.704 mg·L−1 (A4), 11.111 mg·L−1 (A5) and 33.333 mg·L−1 (A6). There were eighteen samples for sensory evaluation, which were equally divided into six groups. All these samples were placed in 10 mL tulip-like glass wine cups. Two samples of each group were the control samples and the remaining one had different concentration of propyl lactate in it. Each sample was marked in a random four-digit number. A group of 30 untrained and normal olfaction assessors were invited to determine propyl lactate detection threshold. All samples were assessed at room temperature. For each test, the assessors needed to pick out the one which was “very different from the references”, and wrote down the number. Each test was replicated 3 times, so that 90 responses were obtained for each testing Molecules, 2015, 20 19011 concentration. Then the Curve Fitting (CF) was conducted by the statistical analyses, and the curve was drawn. When Y = 66.7%, the corresponding X value was the detection threshold value of propyl lactate. 4. Conclusions In summary, this work reported the occurrence of propyl lactate in 72 Chinese Baijius for the first time. The concentration of propyl lactate ranged from 0.050 to 1.900 mg·L−1 in 72 Chinese Baijius. The detection threshold of propyl lactate in 38% ethanol solution was 0.740 mg·L−1. Based on OAV analysis in this research, propyl lactate had much contribution to the aroma of 13 Chinese Baijius , including 50° Jingzhi Guniang 5 years (4.440), 52° Jingzhi Guniang 10 years (3.024), Jingyanggang (2.568), Xianghe Ronghe Shaofang (2.313), and 1956 Laolang (1.431). Acknowledgements The financial supports from National Natural Science Foundation of China (31471665 and 31301466) and National Natural Science Foundation of Beijing (KZ201410011015) are gratefully acknowledged. Author Contributions Conceived and designed the experiments: B.S., F.Z. and M.H. Performed the experiments: J.W. (mostly), Y.Z., X.S. and J.S. Analyzed the data: J.W. Wrote the paper: J.W. and M.H. Conflicts of Interest The authors declare no conflict of interest. The founding sponsors had no role in the design of the study, in the collection, analyses, or interpretation of data, in the writing of the manuscript, and in the decision to publish the results. References 1. Yu, Q.W. Chuantong Baijiu Niangzao Jishu; China Light Industry Press: Beijing, China, 2013; pp. 17–19. 2. Fan, W.L.; Shen, H.Y.; Xu, Y. Quantification of volatile compounds in Chinese soy sauce aroma type liquor by stir bar sorptive extraction and gas chromatography-mass spectrometry. J. Sci. Food Agric. 2011, 91, 1187–1198. 3. Zhu, S.K.; Lu, X.; Ji, K.L.; Guo, K.L.; Li, Y.L.; Wu, C.Y.; Xu, G.W. Characterization of flavor compounds in Chinese liquor Moutai by comprehensive two-dimensional gas chromatography/ time-of-flight mass spectrometry. Anal. Chim. Acta 2007, 597, 340–348. 4. Fan, W.L.; Xu, Y. Identification of volatile compounds of Fenjiu and Langjiu by liquid-liquid extraction coupled with normal phase liquid chromatography (part one). Liquor -Mak. Sci. Technol. 2013, 3, 17–27. Molecules, 2015, 20 19012 5. Liao, Y.H.; Zhao, S.; Zhang, Y.B.; Zhang, X.; Tong, R.N.; Xu, M. Analysis of flavor substances in Erguotou wine by LLE, SDE, SPME and GC-MS combined with Kovats retention indices. J. Chin. Inst. Food Sci. Technol. 2014, 14, 220–228. 6. Zheng, Y.; Zhao, J.W.; Zhang, F.G.; Huang, M.Q.; Sun, B.G.; Zheng, F.P.; Sun, J.Y. Analysis of volatile compounds of Bandaojing sesame-flavor liquor. Food Sci. 2014, 35, 60–65. 7. Peng, Q.; Tian, R.G.; Chen, F.R.; Li, B.B.; Gao, H.G. Discrimination of producing area of Chinese Tongshan kaoliang spirit using electronic nose sensing characteristics combined with the chemometrics methods. Food Chem. 2015, 178, 301–305. 8. Zhu, S.L.; Gao, C.Q.; Cui, G.Y. Analysis of trace compositions of Meilanchun sesame-flavor Liquor. Liquor -Mak. Sci. Technol. 2012, 6, 106–110. 9. Fan, W.L.; Xu, Y. Determination of volatile compounds of fermented-grains in the solid phase fermentation by HS-SPME coupled with GC-MS. Liquor Mak. 2008, 35, 94–98. 10. Wu, Z.Z.; Fan, Z.Y.; Zuo, G.Y.; Wang, H.C. GC-MS direct analysis of qualitative and quantitative sampling of liqour. Liquor Mak. 2009, 36, 88–90. 11. Fan, W.L.; Xu, Y. Flavor Chemistry of Alcoholic Beverage; China Light Industry Press: Beijing, China, 2014; pp.117–121. 12. Guo, W.J.; Lu, J.C. Research on characteristic flavoring components of Gujinggong liquor. Liquor -Mak. Sci. Technol. 2001, 5, 83–85. 13. Feng, J.Q.; Yan, Z.M.; Wei, X.J. Detection and analysis of the trace components in Shedian wine by gas chromatography-mass spectrometry. J. Henan Inst. Sci. Technol. 2011, 39, 23–26. 14. Ledauphin, J.; Saint-Clair, J.-F.; Lablanquie, O.; Hugues, G.; Nicole, F.; Elisabeth, G.; Daniel, B. Identification of trace volatile compounds in freshly distilled calvados and cognac using preparative separations coupled with gas chromatography-mass spectrometry. J. Agric. Food Chem. 2004, 52, 5124–5134. 15. Luan, J. Research on the flavor in rice wine. China Brew. 2002, 6, 21–24. 16. Wu, J.H.; Li, A.J.; Huang, M.H.; Zheng, F.P.; Sun, J.Y.; Sun, B.G. Research and discovery of propyl lactate in Chinese liquor. J. Chin. Inst. Food Sci. Technol. 2015, 15, 194–200. 17. National Standard of the People’s Republic of China. Sensory Analysis-Methodology-General Guidance for Measuring Odour, Flavour and Taste Detection Thresholds by a Three-Alternative Forced-Choice (3-AFC) Procedure; GB/T 22366-2008. China Standards Press: Beijing, China, 2008. 18. Li, W.Q. Relationship between Luzhou flavor liquor, lactic acid bacteria, lactic acid and ethyl lactate. Liquor Mak. 2010, 37, 90–94. 19. Cai, X.Y. Flavor compounds analysis of Chinese liquor by direct-injection technique using capillary GC: An overview. Liquor Mak. 2004, 3, 7–10. 20. Fan, W.L.; Qian, C.M. Identification of aroma compounds in Chinese “Yanghe Daqu” liquor by normal phase chromatography fractionation followed by gas chromatography/olfactometry. Flavour Fragr. J. 2006, 21, 333–342. 21. Liu, Y.P.; Huang, M.Q.; Zheng, F.P.; Chen, H.T.; Sun, B.G. Recent advances in extraction and analysis of volatile flavor compounds in Chinese liquor. Food Sci. 2010, 31, 437–441. 22. Wang, Y; Fan, W.L.; Xu, Y.; Diao, Y.Q.; Lu, H.Z. Study on the determination method about the esterifying enzymes from DaQu of Chinese strong flavour liquor making. Liquor Mak. 2003, 30, 18–21. Molecules, 2015, 20 19013 23. Antalick, G; Perello, M.C.; Revel, D.G. Development, validation and application of a specific method for the quantitative determination of wine esters by headspace-solid-phase microextraction-gas chromatography-mass spectrometry. Food Chem. 2010, 121, 1236–1245. Sample Availability: Samples of the compounds are available from the authors. © 2015 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/4.0/).

The Occurrence of Propyl Lactate in Chinese Baijius (Chinese Liquors) Detected by Direct Injection Coupled with Gas Chromatography-Mass Spectrometry.

10-19-2015

Wu, Jihong,Zheng, Yang,Sun, Baoguo,Sun, Xiaotao,Sun, Jiyuan,Zheng, Fuping,Huang, Mingquan

eng

PMC6939913

Reports © 2019 The Reviewers; Decision Letters © 2019 The Reviewers and Editors; Responses © 2019 The Reviewers, Editors and Authors. Published by the Royal Society under the terms of the Creative Commons Attribution License http://creativecommons.org/licenses/by/4.0/, which permits unrestricted use, provided the original author and source are credited Review History RSPB-2019-1961.R0 (Original submission) Review form: Reviewer 1 Recommendation Accept with minor revision (please list in comments) Scientific importance: Is the manuscript an original and important contribution to its field? Excellent General interest: Is the paper of sufficient general interest? Excellent Quality of the paper: Is the overall quality of the paper suitable? Excellent Is the length of the paper justified? Yes The force–length and force–velocity potential of the human soleus muscle is related to the energetic cost of running Sebastian Bohm, Falk Mersmann, Alessandro Santuz and Adamantios Arampatzis Article citation details Proc. R. Soc. B 286: 20192560. http://dx.doi.org/10.1098/rspb.2019.2560 Review timeline Original submission: 23 August 2019 1st revised submission: 5 November 2019 2nd revised submission: 26 November 2019 Final acceptance: 26 November 2019 Note: Reports are unedited and appear as submitted by the referee. The review history appears in chronological order. 2 Should the paper be seen by a specialist statistical reviewer? No Do you have any concerns about statistical analyses in this paper? If so, please specify them explicitly in your report. No It is a condition of publication that authors make their supporting data, code and materials available - either as supplementary material or hosted in an external repository. Please rate, if applicable, the supporting data on the following criteria.  Is it accessible?  N/A  Is it clear?  N/A  Is it adequate?  N/A Do you have any ethical concerns with this paper? No Comments to the Author This study tests the fascicle length and shortening velocity (imaged using ultrasound) of the soleus during running at a moderate speed in 19 participants. The force-length properties are also measured in a series of isometric contractions on a dynamometer, and the force-velocity properties of the muscle are estimated from the literature. The soleus fascicles are assessed by how close they are to optimal length (force length potential) and how slow they shorten (force velocity potential), and these are compared to the actual cost of running as measured using expired gas analysis from these same participants. It is shown that the fascicles have a high but constant force length potential, and a high force velocity potential: this force velocity potential varies between participants and was related to the running economy. The fascicle velocity was further related (via multi-regression analysis) to the tendon gearing, Achilles tendon moment arm, belly gearing and ROM in these participants. This is an interesting and timely question, and it is posed well. A thorough set of experimental data are collected for the analysis, and the data appear to have been collected very well. The data show convincing relations and support the hypothesis. The paper is written clearly, and the figures are appropriate. Major comments: 1. Choice of maximum shortening velocity. The major finding is that the running economy is sensitive to the force-velocity potential: the absolute values of these depend on the Vmax and curvature of the force-velocity relation. It is not clear where the value of Vmax = 11.72 s-1 comes from. I have followed the cited papers back through several layers of citations, and cannot find definitive justification for these values beyond a statement that Vmax is typically 10-12 s-1 for modelling studies: this in itself does not justify the value. Note that actual measurements of Vmax in humans are rare. Reports of Vmax of 2 and 6 s-1 have been made for isolated bundles of slow and fast human muscle fibres, respectively (Faulkner et al. 1986), and it has been argued that these values could be less than 8 s-1 and greater than 14 s-1, respectively (Epstein & Herzog 1998) from intact human experiments. Higher values are used for musculoskeletal simulations in part to overcome the tendency of simulations to under-predict muscle forces, but this is not a scientific justification for inflating these values. 3 This submitted manuscript would benefit from a clearer description and rationale for the force- velocity parameters chosen. However, I note (as also discussed on line 370) that the actual choice of value has little effect on the conclusions in the study. 2. Multiple regression. The variables used in the multiple regression all pertain to the length change of the soleus muscle fascicles. However, fascicle velocity is a function of change in length and the time taken for this change. Thus, the time taken for the muscle shortening should be considered. Was this constant across participants? Was this considered for the multiple regression (and removed later because it had no effect – this would be a good approach but should be reported)? 3. Experimental details. Additional details should be presented in the Methods section for how the EMG magnitude is quantified, and how fascicles are identified in the ultrasound images. Minor comment. 1. State in the text whether errors are standard deviations or standard errors of the mean. Review form: Reviewer 2 (Natalie Holt) Recommendation Major revision is needed (please make suggestions in comments) Scientific importance: Is the manuscript an original and important contribution to its field? Acceptable General interest: Is the paper of sufficient general interest? Acceptable Quality of the paper: Is the overall quality of the paper suitable? Good Is the length of the paper justified? Yes Should the paper be seen by a specialist statistical reviewer? No Do you have any concerns about statistical analyses in this paper? If so, please specify them explicitly in your report. No It is a condition of publication that authors make their supporting data, code and materials available - either as supplementary material or hosted in an external repository. Please rate, if applicable, the supporting data on the following criteria.  Is it accessible?  N/A  Is it clear?  N/A  Is it adequate?  N/A 4 Do you have any ethical concerns with this paper? No Comments to the Author See attached file. (See Appendix A) Decision letter (RSPB-2019-1961.R0) 23-Sep-2019 Dear Dr Bohm: I am writing to inform you that your manuscript RSPB-2019-1961 entitled "The force-length and force-velocity potential of the human soleus muscle is related to the energetic cost of running" has, in its current form, been rejected for publication in Proceedings B. This action has been taken on the advice of referees, who have recommended that substantial revisions are necessary. With this in mind we would be happy to consider a resubmission, provided the comments of the referees are fully addressed. However please note that this is not a provisional acceptance. While both reviewers see value to the study, both have concerns about some of the assumptions and other aspects of the methods/interpretation. They and the Associate Editor need to be won over more if this study is to be accepted. The resubmission will be treated as a new manuscript. However, we will approach the same reviewers if they are available and it is deemed appropriate to do so by the Editor. Please note that resubmissions must be submitted within six months of the date of this email. In exceptional circumstances, extensions may be possible if agreed with the Editorial Office. Manuscripts submitted after this date will be automatically rejected. Please find below the comments made by the referees, not including confidential reports to the Editor, which I hope you will find useful. If you do choose to resubmit your manuscript, please upload the following: 1) A ‘response to referees’ document including details of how you have responded to the comments, and the adjustments you have made. 2) A clean copy of the manuscript and one with 'tracked changes' indicating your 'response to referees' comments document. 3) Line numbers in your main document. To upload a resubmitted manuscript, log into http://mc.manuscriptcentral.com/prsb and enter your Author Centre, where you will find your manuscript title listed under "Manuscripts with Decisions." Under "Actions," click on "Create a Resubmission." Please be sure to indicate in your cover letter that it is a resubmission, and supply the previous reference number. In your revision process, please take a second look at how open your science is; our policy is that all data involved with the study should be made openly accessible-- see: https://royalsociety.org/journals/ethics-policies/data-sharing-mining/  5 Insufficient sharing of data can delay or even cause rejection of a paper. Sincerely, Professor John Hutchinson, Editor mailto: proceedingsb@royalsociety.org Associate Editor Board Member: 1 Comments to Author: Associate Editor: Douglas L Altshuler The authors have performed an integrative study of muscle physiology and energetics during running. The authors and I agree that the methods are sound and creative, and the results are clear and interesting. One of the referees expressed some concern that the study may be too narrow in scope for the general science audience of ProcB. I would encourage the authors to revise their manuscript, and it would be helpful to see how this concern about breadth could be addressed. Reviewer(s)' Comments to Author: Referee: 1 Comments to the Author(s) This study tests the fascicle length and shortening velocity (imaged using ultrasound) of the soleus during running at a moderate speed in 19 participants. The force-length properties are also measured in a series of isometric contractions on a dynamometer, and the force-velocity properties of the muscle are estimated from the literature. The soleus fascicles are assessed by how close they are to optimal length (force length potential) and how slow they shorten (force velocity potential), and these are compared to the actual cost of running as measured using expired gas analysis from these same participants. It is shown that the fascicles have a high but constant force length potential, and a high force velocity potential: this force velocity potential varies between participants and was related to the running economy. The fascicle velocity was further related (via multi-regression analysis) to the tendon gearing, Achilles tendon moment arm, belly gearing and ROM in these participants. This is an interesting and timely question, and it is posed well. A thorough set of experimental data are collected for the analysis, and the data appear to have been collected very well. The data show convincing relations and support the hypothesis. The paper is written clearly, and the figures are appropriate. Major comments: 1. Choice of maximum shortening velocity. The major finding is that the running economy is sensitive to the force-velocity potential: the absolute values of these depend on the Vmax and curvature of the force-velocity relation. It is not clear where the value of Vmax = 11.72 s-1 comes from. I have followed the cited papers back through several layers of citations, and cannot find definitive justification for these values beyond a statement that Vmax is typically 10-12 s-1 for modelling studies: this in itself does not justify the value. Note that actual measurements of Vmax in humans are rare. Reports of Vmax of 2 and 6 s-1 have been made for isolated bundles of slow and fast human muscle fibres, respectively (Faulkner et al. 1986), and it has been argued that these values could be less than 8 s-1 and greater than 14 s-1, respectively (Epstein & Herzog 1998) from intact human experiments. Higher values are used for 6 musculoskeletal simulations in part to overcome the tendency of simulations to under-predict muscle forces, but this is not a scientific justification for inflating these values. This submitted manuscript would benefit from a clearer description and rationale for the force- velocity parameters chosen. However, I note (as also discussed on line 370) that the actual choice of value has little effect on the conclusions in the study. 2. Multiple regression. The variables used in the multiple regression all pertain to the length change of the soleus muscle fascicles. However, fascicle velocity is a function of change in length and the time taken for this change. Thus, the time taken for the muscle shortening should be considered. Was this constant across participants? Was this considered for the multiple regression (and removed later because it had no effect – this would be a good approach but should be reported)? 3. Experimental details. Additional details should be presented in the Methods section for how the EMG magnitude is quantified, and how fascicles are identified in the ultrasound images. Minor comment. 1. State in the text whether errors are standard deviations or standard errors of the mean. Referee: 2 Comments to the Author(s) See attached file Author's Response to Decision Letter for (RSPB-2019-1961.R0) See Appendix B. RSPB-2019-2560.R0 Review form: Reviewer 1 Recommendation Accept with minor revision (please list in comments) Scientific importance: Is the manuscript an original and important contribution to its field? Excellent General interest: Is the paper of sufficient general interest? Excellent Quality of the paper: Is the overall quality of the paper suitable? Excellent Is the length of the paper justified? Yes 7 Should the paper be seen by a specialist statistical reviewer? No Do you have any concerns about statistical analyses in this paper? If so, please specify them explicitly in your report. No It is a condition of publication that authors make their supporting data, code and materials available - either as supplementary material or hosted in an external repository. Please rate, if applicable, the supporting data on the following criteria.  Is it accessible?  N/A  Is it clear?  N/A  Is it adequate?  N/A Do you have any ethical concerns with this paper? No Comments to the Author The authors have addressed all my previous concerns in a careful manner. There remains one further comment that they may choose to consider for the manuscript, and it still concerns the choice of Vmax. I appreciate the further analysis that the authors have attempted, to provide a value of Vmax for the Soleus. However, it should be noted that the running velocity of 2.5 m/s is not all that fast, and indeed the EMG averages less than 50%. As such, it is likely that the fastest muscle fibres will not have been recruited, and hence the weighted mean taken for Vmax may thus be an overestimate. Coupled to this, with more than half of the muscle inactive, the actual Vmax may be less than its constituent fibres (for additional reasons: Holt et al. Proc Roy Soc B 2014). If the Vmax for the Soleus were less than the estimated 6.77 L/s for this experimental situation, then it is likely that the actual spread of Force-velocity potentials would be larger than shown in Fig. 3. It is thus worth considering that you have actually resulted with a conservative evaluation of the importance of the force-velocity potential. Review form: Reviewer 2 (Natalie Holt) Recommendation Accept with minor revision (please list in comments) Scientific importance: Is the manuscript an original and important contribution to its field? Good General interest: Is the paper of sufficient general interest? Good 8 Quality of the paper: Is the overall quality of the paper suitable? Good Is the length of the paper justified? Yes Should the paper be seen by a specialist statistical reviewer? No Do you have any concerns about statistical analyses in this paper? If so, please specify them explicitly in your report. No It is a condition of publication that authors make their supporting data, code and materials available - either as supplementary material or hosted in an external repository. Please rate, if applicable, the supporting data on the following criteria.  Is it accessible?  N/A  Is it clear?  N/A  Is it adequate?  N/A Do you have any ethical concerns with this paper? No Comments to the Author See attached file. (See Appendix C) Decision letter (RSPB-2019-2560.R0) 20-Nov-2019 Dear Dr Bohm I am pleased to inform you that your manuscript RSPB-2019-2560 entitled "The force-length and force-velocity potential of the human soleus muscle is related to the energetic cost of running" has been accepted for publication in Proceedings B. Congratulations!! The referee(s) have recommended publication, but also suggest some minor revisions to your manuscript. Therefore, I invite you to respond to the referee(s)' comments and revise your manuscript. Because the schedule for publication is very tight, it is a condition of publication that you submit the revised version of your manuscript within 7 days. If you do not think you will be able to meet this date please let us know. To revise your manuscript, log into https://mc.manuscriptcentral.com/prsb and enter your Author Centre, where you will find your manuscript title listed under "Manuscripts with  9 Decisions." Under "Actions," click on "Create a Revision." Your manuscript number has been appended to denote a revision. You will be unable to make your revisions on the originally submitted version of the manuscript. Instead, revise your manuscript and upload a new version through your Author Centre. When submitting your revised manuscript, you will be able to respond to the comments made by the referee(s) and upload a file "Response to Referees". You can use this to document any changes you make to the original manuscript. We require a copy of the manuscript with revisions made since the previous version marked as ‘tracked changes’ to be included in the ‘response to referees’ document. Before uploading your revised files please make sure that you have: 1) A text file of the manuscript (doc, txt, rtf or tex), including the references, tables (including captions) and figure captions. Please remove any tracked changes from the text before submission. PDF files are not an accepted format for the "Main Document". 2) A separate electronic file of each figure (tiff, EPS or print-quality PDF preferred). The format should be produced directly from original creation package, or original software format. PowerPoint files are not accepted. 3) Electronic supplementary material: this should be contained in a separate file and where possible, all ESM should be combined into a single file. All supplementary materials accompanying an accepted article will be treated as in their final form. They will be published alongside the paper on the journal website and posted on the online figshare repository. Files on figshare will be made available approximately one week before the accompanying article so that the supplementary material can be attributed a unique DOI. Online supplementary material will also carry the title and description provided during submission, so please ensure these are accurate and informative. Note that the Royal Society will not edit or typeset supplementary material and it will be hosted as provided. Please ensure that the supplementary material includes the paper details (authors, title, journal name, article DOI). Your article DOI will be 10.1098/rspb.[paper ID in form xxxx.xxxx e.g. 10.1098/rspb.2016.0049]. 4) A media summary: a short non-technical summary (up to 100 words) of the key findings/importance of your manuscript. 5) Data accessibility section and data citation It is a condition of publication that data supporting your paper are made available either in the electronic supplementary material or through an appropriate repository. In order to ensure effective and robust dissemination and appropriate credit to authors the dataset(s) used should be fully cited. To ensure archived data are available to readers, authors should include a ‘data accessibility’ section immediately after the acknowledgements section. This should list the database and accession number for all data from the article that has been made publicly available, for instance: • DNA sequences: Genbank accessions F234391-F234402 • Phylogenetic data: TreeBASE accession number S9123 • Final DNA sequence assembly uploaded as online supplemental material • Climate data and MaxEnt input files: Dryad doi:10.5521/dryad.12311 NB. From April 1 2013, peer reviewed articles based on research funded wholly or partly by RCUK must include, if applicable, a statement on how the underlying research materials – such  10 as data, samples or models – can be accessed. This statement should be included in the data accessibility section. If you wish to submit your data to Dryad (http://datadryad.org/) and have not already done so you can submit your data via this link http://datadryad.org/submit?journalID=RSPB&manu=(Document not available) which will take you to your unique entry in the Dryad repository. If you have already submitted your data to dryad you can make any necessary revisions to your dataset by following the above link. Please see https://royalsociety.org/journals/ethics-policies/data-sharing-mining/ for more details. 6) For more information on our Licence to Publish, Open Access, Cover images and Media summaries, please visit https://royalsociety.org/journals/authors/author-guidelines/. Once again, thank you for submitting your manuscript to Proceedings B and I look forward to receiving your revision. If you have any questions at all, please do not hesitate to get in touch. Sincerely, Professor John Hutchinson, Editor mailto: proceedingsb@royalsociety.org Associate Editor Board Member Comments to Author: Associate Editor: Doug Altshuler The authors have done a good job addressing the reviewer concerns. A few issues remain, and I agree with the reviewers that consideration of these final points would further strengthen the manuscript. Reviewer(s)' Comments to Author: Referee: 2 Comments to the Author(s). See attached file Referee: 1 Comments to the Author(s). The authors have addressed all my previous concerns in a careful manner. There remains one further comment that they may choose to consider for the manuscript, and it still concerns the choice of Vmax. I appreciate the further analysis that the authors have attempted, to provide a value of Vmax for the Soleus. However, it should be noted that the running velocity of 2.5 m/s is not all that fast, and indeed the EMG averages less than 50%. As such, it is likely that the fastest muscle fibres will not have been recruited, and hence the weighted mean taken for Vmax may thus be an overestimate. Coupled to this, with more than half of the muscle inactive, the actual Vmax may be less than its constituent fibres (for additional reasons: Holt et al. Proc Roy Soc B 2014). If the Vmax for the Soleus were less than the estimated 6.77 L/s for this experimental situation, then it is likely that the actual spread of Force-velocity 11 potentials would be larger than shown in Fig. 3. It is thus worth considering that you have actually resulted with a conservative evaluation of the importance of the force-velocity potential. Author's Response to Decision Letter for (RSPB-2019-2560.R0) See Appendix D. Decision letter (RSPB-2019-2560.R1) 26-Nov-2019 Dear Dr Bohm I am pleased to inform you that your manuscript entitled "The force-length and force-velocity potential of the human soleus muscle is related to the energetic cost of running" has been accepted for publication in Proceedings B. You can expect to receive a proof of your article from our Production office in due course, please check your spam filter if you do not receive it. PLEASE NOTE: you will be given the exact page length of your paper which may be different from the estimation from Editorial and you may be asked to reduce your paper if it goes over the 10 page limit. If you are likely to be away from e-mail contact please let us know. Due to rapid publication and an extremely tight schedule, if comments are not received, we may publish the paper as it stands. If you have any queries regarding the production of your final article or the publication date please contact procb_proofs@royalsociety.org Your article has been estimated as being 10 pages long. Our Production Office will be able to confirm the exact length at proof stage. Open Access You are invited to opt for Open Access, making your freely available to all as soon as it is ready for publication under a CCBY licence. Our article processing charge for Open Access is £1700. Corresponding authors from member institutions (http://royalsocietypublishing.org/site/librarians/allmembers.xhtml) receive a 25% discount to these charges. For more information please visit http://royalsocietypublishing.org/open-access. Paper charges An e-mail request for payment of any related charges will be sent out shortly. The preferred payment method is by credit card; however, other payment options are available. Electronic supplementary material: All supplementary materials accompanying an accepted article will be treated as in their final form. They will be published alongside the paper on the journal website and posted on the online  12 figshare repository. Files on figshare will be made available approximately one week before the accompanying article so that the supplementary material can be attributed a unique DOI. You are allowed to post any version of your manuscript on a personal website, repository or preprint server. However, the work remains under media embargo and you should not discuss it with the press until the date of publication. Please visit https://royalsociety.org/journals/ethics- policies/media-embargo for more information. Thank you for your fine contribution. On behalf of the Editors of the Proceedings B, we look forward to your continued contributions to the Journal. Sincerely, Proceedings B mailto: proceedingsb@royalsociety.org In this study, the authors explore the effect of muscle force-length and force-velocity conditions on the energetic cost of running. In addition, they explore the determinants of of muscle fiber length change. I find this to be an extensive and well collected data set that uses in vivo determination of force-length and force-velocity relationships, and application of these to muscle function during running to address these questions. There appear to me to be a few major limitations of the study. These should be addressed throughout. 1) Organismal energy consumption is measured, but only the length and velocity profile of soleus. This prevents the authors from drawing the more interesting conclusion that muscle shortening velocity is a determinant of energy consumption. This is acknowledged and somewhat addressed in the discussion, but remains a major limitation to the study. 2) The strict adherence to force-length and force-velocity relationships as defining features of muscle performance seems somewhat outdated given a wealth of literature showing that these relationships do not hold under conditions relevant to locomotion (i.e. history dependence, activation-dependent changes). These advances do not negate this study, however, it would be a more accurate representation of the field to acknowledge that they exist, and present this study as a means to investigate the importance of these relationships. 3) This paper is fundamentally concerned with the effect of contractile conditions on muscle energy consumption. However, there is very little discussion of why length and velocity might affect energy consumption beyond required activation, despite a wealth of evidence on this i.e. how the cost per unit force varies across the force-length relationship in isolated muscle. In addition, it may be worth considering findings such as the effect of contractile history on cost (Joumaa et al., 2013), and the complexity of the cost of work (Holt et al., 2104; Curtin et al., 2019) in a more comprehensive discussion of in vivo muscle energetics. Specific comments Lines 50-51 – The ongoing debating between the cost of force and the cost of work as determinants of organismal cost should be acknowledged here. This could then also lead to a more nuanced discussion of factors dictating muscle energetics beyond simply level of activation. Lines 123-124 and 185-187 – It is relatively unclear to me how the force-velocity relationship was determined here. It appears as though force and velocity were determined as fibers shortened against the tendon? Can the authors make this clearer, better define where in the contraction force and velocity were determined, and comment on how this might affect findings compared to a more standard isotonic or isovelocity protocol. Line 197-198 – The meaning of this is unclear to me. This description of touchdown and toe-off should be reworded for clarification The results section is relatively dense. The authors may wish to consider moving some of the findings less critical to addressing their question to a table, to improve readability. Appendix A Line 299-300 – The assertion that the triceps surae consumes 40% of the cost during running is crucial to the argument of this paper. Yet it is not clear how this value is arrived at from the Fletcher and MacIntosh paper cited (the paper seems to give a large range of values for muscle energy consumption and not to relate this to organismal cost), and how reliable the output of their simple model is for this purpose. Could the authors give a little more detail on this (in the manuscript if of sufficient interest, or simply here). It may also be useful to combine this 40% estimate with the relative size of soleus to give a better representation of its likely contribution to energy consumption, considering fiber type as soleus is likely cheaper than gastrocs (Barclay, 1993). Lines 304-309 - The authors make a good case for why small changes in velocity would require an increase in activation and therefore cost. This effect should be seen in EMG recordings. It would seem that the argument could be strengthened by showing this as it would provide a more causal link between the change in muscle level function and organismal level cost. Line 344-345 – There seems to be some discrepancy regarding activation in here. The implication seems to be that muscle activation is higher in early stance to enable the tendon to be stretched, and then recoil to slow shortening velocity in the later part of stance. Yet a central claim of the paper is that cost is lower when shortening velocity is lower, due to a lower requirement for activation. It seems like the variation in required activation could balance out over the course of a stance phase? Could it be clarified as to why the early increase in activation to enable tendon stretch doesn’t seem to be costly in the way that the latter reduction is deemed to be cheap? Line 345 – typo “were”? Line 375 – The study doesn’t seem to show that energy consumption is related to the force-length- velocity potential, but rather just the force-velocity potential. Referee: 1 Comment: This study tests the fascicle length and shortening velocity (imaged using ultrasound) of the soleus during running at a moderate speed in 19 participants. The force-length properties are also measured in a series of isometric contractions on a dynamometer, and the force-velocity properties of the muscle are estimated from the literature. The soleus fascicles are assessed by how close they are to optimal length (force length potential) and how slow they shorten (force velocity potential), and these are compared to the actual cost of running as measured using expired gas analysis from these same participants. It is shown that the fascicles have a high but constant force length potential, and a high force velocity potential: this force velocity potential varies between participants and was related to the running economy. The fascicle velocity was further related (via multi-regression analysis) to the tendon gearing, Achilles tendon moment arm, belly gearing and ROM in these participants. This is an interesting and timely question, and it is posed well. A thorough set of experimental data are collected for the analysis, and the data appear to have been collected very well. The data show convincing relations and support the hypothesis. The paper is written clearly, and the figures are appropriate. Response: Thank you for your valuable comments. All changes are underlined in the revised version of the manuscript and the references cited in the responses can be found at the end of the document. Please note that some parts of the methods are now presented in the electronic supplementary material due to length restrictions of the journal. Major comments: Comment: 1. Choice of maximum shortening velocity. The major finding is that the running economy is sensitive to the force-velocity potential: the absolute values of these depend on the Vmax and curvature of the force-velocity relation. It is not clear where the value of Vmax = 11.72 s-1 comes from. I have followed the cited papers back through several layers of citations, and cannot find definitive justification for these values beyond a statement that Vmax is typically 10-12 s-1 for modelling studies: this in itself does not justify the value. Note that actual measurements of Vmax in humans are rare. Reports of Vmax of 2 and 6 s-1 have been made for isolated bundles of slow and fast human muscle fibres, respectively (Faulkner et al. 1986), and it has been argued that these values could be less than 8 s-1 and greater than 14 s-1, respectively (Epstein & Herzog 1998) from intact human experiments. Higher values are used for musculoskeletal simulations in part to overcome the tendency of simulations to under-predict muscle forces, but this is not a scientific justification for inflating these values. This submitted manuscript would benefit from a clearer description and rationale for the force- velocity parameters chosen. However, I note (as also discussed on line 370) that the actual choice of value has little effect on the conclusions in the study. Response: Thank you for this comment. Indeed, experimental assessed values of Vmax for human muscles in vivo are very rare and for the soleus not reported so far to our knowledge. This was the reason why we based our calculations on recommendations for modelling approaches [1,2]. When extending our sensitivity analysis about the effect of the magnitude of Vmax, on the correlation of the Appendix B force-velocity potential and energetic cost, we found that the correlation remained statistically significant (p<0.05) for Vmax values higher than 3.0 LO/s. Therefore, the observed association of force- velocity potential and energetic cost seems quite strong. Furthermore, we reported a direct correlation between the operating velocity and the energetic cost (r = 0.561 p = 0.012). Since this association is independent of the choice of Vmax, we can be confident about the general study findings. Biological support for the choice of Vmax could be derived indirectly when referring to in vitro studies on the human soleus muscle. Luden et al., (2008) showed Vmax values for MHC I type fibers of 0.77 LO/s and 2.91 LO/s for MHC IIA type fibers measured at 15°C [3]. Considering the temperature coefficient provided by Ranatunga et al., (1984) [4] it can be predicted that Vmax would increase to 4.4 LO/s for MHC I type fibers and to 16.8 LO/s for MHC IIA type fibers under physiological temperature conditions (37 °C). The fiber type distribution in the human soleus muscle can be estimated from literature reports, i.e. Johnson et al., 1973: type 1 fibers 87,7%, type 2 fibers (a and b) 12,3% (average of surface and deep fiber location) [5]; Larsson and Moss, 1993: type 1 89%, type 2A 11% [6]; Edgerton et al., 1975: slow twitch 70%, fast twitch 30% [7]; Luden et al., 2008: 74% MHC I, 20% MHC IIA (norm to 100%) [3]). Using an average of those reported distribution values (type 1: 81%, type 2: 19%), Vmax for soleus under physiological temperature can be calculated as 6.77 LO/s. The broad literature basis for the average fiber type distribution was also used to update arel to 0.175 (i.e. 0.1+0.4FT, where FT is the fast twitch fiber type percentage [8,9]) and accordingly brel to 1.182 (arel * Vmax [10]). We addressed the comment of the reviewer and recalculated the respective values (force-velocity potential and normalized velocities and their ranges) using the updated Vmax of 6.77 LO/s, arel and brel in the revised manuscript. Again, this did not change any of the statistical correlation outcomes. We added the following information to the revised manuscript (page: 4, line: 155): “Furthermore, we assessed the force-velocity relationship of soleus using the classical Hill equation [11], the muscle-specific maximum fascicle shortening velocity (Vmax) and constants of arel and brel. Vmax was derived from the study of Luden et al. (2008), which showed Vmax values for type 1 fibers of 0.77 LO/s and 2.91 LO/s for type 2 fibers of the human soleus muscle measured in vitro at 15°C [3]. Considering the temperature coefficient [4], Vmax can be predicted as 4.4 LO/s for type 1 fibers and 16.8 LO/s for type 2 fibers under physiological temperature conditions (37 °C). Using an average fiber type distribution (type 1 fibers: 81%, type 2: 19%) of the human soleus muscle reported in literature [3,5–7], Vmax can be calculated as 6.77 LO/s. arel was calculated as 0.1+0.4FT, where FT is the fast twitch fiber type percentage (see above), which then equals to 0.175 [8,9]. The product of arel and Vmax then gives brel as 1.182 [10]. After rearrangement of the Hill equation and extension to the eccentric component, the operating velocity normalized to Vmax can be used to calculate the individual force potential according to the force-velocity relationship.” Comment: 2. Multiple regression. The variables used in the multiple regression all pertain to the length change of the soleus muscle fascicles. However, fascicle velocity is a function of change in length and the time taken for this change. Thus, the time taken for the muscle shortening should be considered. Was this constant across participants? Was this considered for the multiple regression (and removed later because it had no effect – this would be a good approach but should be reported)? Response: Thanks for this comment. The time for the muscle shortening (i.e. stance time) showed some variability among the participants as to be expected, i.e. mean 304 ms, SD 23.1 ms, maximum 362 ms, minimum 270 ms. With respect to the four variables (tendon gearing, belly gearing, tendon lever arm and ankle angle range) and the effect of time, we reasoned that the tendon and belly gearing are ratios between velocities, and are thereby independent of time (i.e. tendon gearing: VMTU/VBelly and belly gearing: VBelly/VFascicle, where V is the stance phase-averaged velocity). The tendon lever arm is a quantity that is also independent from time, while only the ankle angle range is time-depended. We now calculated the stance phase-averaged absolute ankle angle velocity and rebuilt the regression model, which is now expressed by the updated equation: 𝐹𝑎𝑠𝑖𝑐𝑙𝑒 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦 = −9.788 (𝑡𝑒𝑛𝑑𝑜𝑛 𝑔𝑒𝑎𝑟𝑖𝑛𝑔) + 0.716 (𝑙𝑒𝑣𝑒𝑟 𝑎𝑟𝑚) − 42.097 (𝑏𝑒𝑙𝑙𝑦 𝑔𝑒𝑎𝑟𝑖𝑛𝑔) + 0.209 (𝑚𝑒𝑎𝑛 𝑎𝑛𝑘𝑙𝑒 𝑎𝑛𝑔𝑙𝑒 𝑣𝑒𝑙𝑜𝑐𝑖𝑡𝑦) + 51.341 The model remained significant (p < 0.001, R2 = 0.928, adjusted R2 = 0.907) and so did the four independent variables (p < 0.001 for tendon gearing, tendon lever arm and belly gearing, and p = 0.002 for ankle angle velocity). The standardized coefficients changed slightly to -1.006 for tendon gearing, 0.638 for lever arm, -0.367 for belly gearing and 0.310 for the ankle angle velocity. We changed this part in the revised manuscript ((page: 6, line: 262; page: 7, line: 315). Comment: 3. Experimental details. Additional details should be presented in the Methods section for how the EMG magnitude is quantified, and how fascicles are identified in the ultrasound images. Response: We amended the information on the EMG assessment during running in the revised manuscript as supplementary material (“EMG processing”). Note that the low-pass filter cut-off frequency was changed from 6 Hz to 20 Hz. “Raw EMG signals (running and MVC) were processed by a fourth-order high-pass Butterworth zero- phase filter with a 50 Hz cut-off frequency then a full-wave rectification and a low-pass zero-phase filter with a 20 Hz cut-off frequency for creating a linear envelope of the signal [12,13].” We added some more information on the fascicle identification in the revised manuscript as supplementary material (“Fascicle length determination from the ultrasound images”): “The procedure included an approximation of the deeper and upper aponeurosis by a best linear fit through three manually placed and frame-by-frame adjusted marks. By means of the bwtraceboundary function of the Matlab Image Processing toolbox the algorithm then identified the shape and orientation of image brightness features between both aponeuroses in each frame, which are indicative for the hyperechoic perimysial connective tissue parts aligned with the muscle fascicles (fig. 1A). The feature identification criteria were set to: minimal length of 23 pixels (i.e. 0.4 cm, from the bottom left to the top right), area to length ratio of 8.5, angle between feature and deeper aponeurosis between 10° and 70° and 80% of the pixels on a line between the start and end point of a feature had to be white [14]. Every frame was visually controlled for adequate feature placement and manually corrected if necessary. Based on the identified features, a linear averaged reference fascicle was calculated (fig. 1A). Reliability of the tracking approach was confirmed and reported in two previous studies [14,15].” Minor comment. Comment: 1. State in the text whether errors are standard deviations or standard errors of the mean. Response: We added the information that standard deviations are presented in the revised manuscript (page: 5, line: 221). Referee: 2 Comment: In this study, the authors explore the effect of muscle force-length and force-velocity conditions on the energetic cost of running. In addition, they explore the determinants of muscle fiber length change. I find this to be an extensive and well collected data set that uses in vivo determination of force-length and force-velocity relationships, and application of these to muscle function during running to address these questions. There appear to me to be a few major limitations of the study. These should be addressed throughout. Response: Thank you for your valuable comments. All changes are underlined in the revised version of the manuscript and the references cited in the responses can be found at the end of the document. Please note that some parts of the methods are now presented in the electronic supplementary material due to length restrictions of the journal. Comment: 1) Organismal energy consumption is measured, but only the length and velocity profile of soleus. This prevents the authors from drawing the more interesting conclusion that muscle shortening velocity is a determinant of energy consumption. This is acknowledged and somewhat addressed in the discussion, but remains a major limitation to the study. Response: We agree with the reviewers comment that although the soleus contributes to a great portion of the overall energetic costs during running [16] other limb muscles are involved that were not considered in the present study and this remains a limitation. However, the main energy source (positive work) is the ankle joint (41%) while the contribution of the knee and hip joint is comparably lower during running [17]. The soleus is the greatest muscle among the main plantar flexors with respect to physiological cross-sectional area and volume (soleus: 131 cm² and 477 cm³, gastrocnemius medialis: 51 cm² and 285 cm³, gastrocnemius lateralis: 24 cm ² and 146 cm³ [18]), giving this muscle a key role. Using a modeling approach, Hamner and Delp (2013) showed that the contribution of soleus to the vertical acceleration of the center of mass at the same velocity as in our study (i.e. 2.5 m/s) is remarkably higher than those of the other lower limb muscles (7.5x than gastroc., 9.5x than vasti, 24x than rect. fem., 9.5x than tib. ant., 12x than glut. max.; visual inspection of fig. 5). For the fore-aft acceleration a similar superior contribution of soleus was shown (3x higher than gastroc. and 8.8x higher than hamstrings [19]). The propulsive function of soleus during running is achieved by active shortening. Active shortening reduces the force-velocity potential as discussed extensively in the present manuscript. Consequently, a greater active muscle volume is required to achieve the required mechanical energy gain. In contrast, the quadriceps muscle group, the main contributor during early stance, decelerating and supporting body mass [19,20], features more economical fascicle dynamics. Recently we showed that the fascicles of the vastus lateralis muscle, as a representative of the quadriceps muscle group operates with a high force-length (i.e. 0.91) and force-velocity potential (i.e. 0.97) during the stance phase of running. Operating at high force potentials reduces the energetic cost of this energetically expensive (due to its long fascicle length) muscle by reducing active muscle volume. This may indicate that the mechanical energy by muscular work required for steady state running is generated by muscles that are metabolically less expensive, likely to compensate for the reduction of the force-velocity potential. We discussed the reviewers comment as a limitation in the revised manuscript (page: 8, line: 360): “Although the soleus likely contributes to a great portion of the overall energetic costs during running, other limb muscles that were not considered in the present study are involved. However, the main energy source (positive work) is the ankle joint (41%) [17] and the soleus is the greatest muscle among the main plantar flexors with respect to physiological cross-sectional area (soleus 63%, gastroc. med. 25%, gastroc. lat. 12%) and volume (53%, 31% and 16% [18]). The key role of soleus is further supported by the modeling study of Hamner and Delp (2013), which showed that the soleus is by far the biggest contributor to the vertical acceleration and fore-aft acceleration of the center of mass [19]. This function is achieved by active shortening, which reduces the force-velocity potential and consequently requires a greater active muscle volume. In contrast, the quadriceps muscle group, the main contributor during early stance, decelerating and transferring body mass [19,20], features more economical fascicle dynamics. Recently we showed that the fascicles of the vastus lateralis muscle as a representative of the quadriceps muscle group operates with a high force-length (i.e. 0.91) and force- velocity potential (i.e. 0.97) during the stance phase of running. Operating at high force potentials minimizes the cost of this muscle, which is energetically expensive due to its long fascicle length (i.e. LO = 94 mm [15]), by reducing active muscle volume. This may indicate that the mechanical energy by muscular work required for steady state running is generated by muscles that are metabolically less expensive, likely to compensate for the reduction of the force-velocity potential.” Comment: 2) The strict adherence to force-length and force-velocity relationships as defining features of muscle performance seems somewhat outdated given a wealth of literature showing that these relationships do not hold under conditions relevant to locomotion (i.e. history dependence, activation-dependent changes). These advances do not negate this study, however, it would be a more accurate representation of the field to acknowledge that they exist, and present this study as a means to investigate the importance of these relationships. Response: First note – History dependence: We agree with the reviewers comment that the phenomenon of history dependence of force production after active muscle lengthening or shortening may be present for the soleus during running and may affect the force production [21–23]. In the present study, the soleus fascicles shortened continuously during running when activated, which would indicate a condition of force depression. Force depression has been shown to increase with increasing shortening magnitude [24], with decreasing shortening velocity [25] and with increasing activation levels [26]. Since the soleus shortening magnitude was notable (25.9 ± 7.8 %LO), the shortening velocity moderate (0.118 ± 0.039 Vmax) and the activation submaximal (average during stance phase: 0.32 ± 0.19 EMGmax; maximum activation: 0.52 ± 0.18 EMGmax), an effect of force depression on the force production can theoretically be expected. Yet, the force-length and force-velocity relationships remain the basic mechanisms for muscle force production. Interestingly, force depression is likely to be reduced due to the tendon and belly gearing mechanisms because those reduce the shortening magnitude and activation. The observed main finding of a correlation of the operating velocity and force-velocity potential with the energetic cost, however, does not neglect the presence of force depression but indicates that shortening velocity and consequently the force-velocity potential has a direct effect on the muscle energetics. We added the following sentences in the introduction and discussion of the revised version of the manuscript (page: 2, line: 59; page: 7, line: 307): “Besides the operating length and velocity as the main determinants, the history dependence of force generation [23], i.e. increased force after active muscle lengthening [27] and decreased force after active shortening [22,25], may additionally influence the force potential.” “Furthermore, we showed that the soleus shortened continuously during the stance phase of running, which reflects a condition for force depression. Since a depression of force was shown to be accompanied by a decrease in the ATPase activity [28], force depression would have little or no effect on the energetic cost itself.” Second note – Shift in optimal length: Furthermore, it is correct that we assessed the force-length curve during maximal isometric contractions at different ankle joint angles and, using this relationship, we calculated the force-length potential of the soleus muscle during running at submaximal activation. There is evidence from early [29] and more recent [30,31] in vitro studies that the force-length curve depends on muscle activation, i.e. optimum length increases with submaximal activation. However, a recent study by Fontana and Herzog (2016) on the human vastus lateralis muscle showed that this holds not necessarily true for in vivo assessments [32]. In contrast to the in vitro studies, a rightward shift of optimal length was not observed when force was normalized to the maximum EMG signal (i.e. optimal length remained constant at different levels of activation). The authors suggested that the disagreement of the in vitro and in vivo studies might be an artefact related to the in vitro testing setup (e.g. non-physiological stimulation frequency range or calcium concentrations). Therefore, we can argue that mapping the submaximal fascicle operating length onto the force-length curve in the present in vivo study should not affect the findings. We added the following information in the discussion part of revised manuscript as follows (page: 9, line: 385): “Furthermore, we assessed the force-length curve during maximal isometric contractions and used it to calculate the force-length potential of the soleus muscle during running at submaximal activation. There is evidence from in vitro studies that the force-length curve depends on muscle activation [29– 31]. However, in a recent in vivo study by Fontana and Herzog (2016) on the human vastus lateralis muscle, a rightward shift of optimal length with submaximal activation was not observed when force was normalized to the maximum EMG signal [32]. The authors suggested that the shift in optimal length phenomenon might be related to the in vitro testing setup (e.g. non-physiological stimulation frequency range or Ca2+concentrations). Therefore, we can argue that mapping the submaximal fascicle operating length onto the force-length curve in the present in vivo study should not affect the findings.” Comment: 3) This paper is fundamentally concerned with the effect of contractile conditions on muscle energy consumption. However, there is very little discussion of why length and velocity might affect energy consumption beyond required activation, despite a wealth of evidence on this i.e. how the cost per unit force varies across the force-length relationship in isolated muscle. In addition, it may be worth considering findings such as the effect of contractile history on cost (Joumaa et al., 2013), and the complexity of the cost of work (Holt et al., 2104; Curtin et al., 2019) in a more comprehensive discussion of in vivo muscle energetics. Response: First note - Variation of cost per unit force across the force-length relationship: We agree with the reviewer that the energy turnover can differ across the force-length relationship in isolated animal muscle fibers tested in vitro [33]. During isometric contractions at sarcomere length shorter than optimal length, the force output is reduced but the ATPase rate seems not to greatly differ from the rate at optimal length, indicating a comparably higher cost of contraction at shorter length [34,35]. However, this effect seems to be more pronounced at very short lengths, which might not be covered during regular in vivo movements like locomotion, i.e. soleus operating range (0.75-1.01 LO). We added the following information in the revised version of the manuscript as follows (page: 7, line: 300): “Besides the favorable high force-length potential for economical force production, operating close to optimal length may additionally preserved from relatively higher energetic cost that can arise when contracting at shorter length. In vitro evidence showed that although force is reduced at shorter sarcomere length, the ATPase rate seems not to differ from the rate at optimal length, indicating comparably higher cost of contraction at shorter length [34,35]. However, this effect seems more pronounced at very short lengths, a portion of the force-length curve that is likely not covered by the soleus during running (operating range 0.75-1.01 LO).” Second note - Effect of contractile history on cost: We added the following paragraph to the discussion part of the revised manuscript (page: 7, line: 307): “Furthermore, we showed that the soleus shortened continuously during the stance phase of running, which reflects a condition for force depression. Since a depression of force was shown to be accompanied by a decrease in the ATPase activity [28], force depression would have little or no effect on the energetic cost itself.” Third note - complexity of the cost of work: Thanks for this comment. We agree with the reviewer on the ongoing debate on the cost of force and the cost of work. From our perspective, when a muscle contracts, force is generated and this consumes metabolic energy independently of the contraction type (i.e. isometric, eccentric, concentric). During concentric contractions (active shortening) positive mechanical work is generated and during eccentric contractions (active lengthening) the work is negative. In stretch-shortening conditions, the net work could be zero when positive and negative work cancel each other out. Under isometric contractions, no mechanical work is generated by definition, which would again indicate no mechanical energy production (Joule), although force is generated and metabolic energy expended. The energy index of work in the context of the explanation of metabolic energy, therefore, might not be very appropriate (metabolic energy is not zero when work is zero e.g. during isometric contractions). Instead, an index of force and metabolic energy might better reflect the organismal cost during locomotion. With our study, we cannot provide any new information on this discussion because work and force of soleus were not measured during running (which in our opinion is not possible at the moment). Therefore, we think that this topic is beyond of the scope of the present study and for this reason we would prefer not to go deeper in the discussion of cost of force and work but rather stay close to our experimental results. Specific comments Comment: Lines 50-51 – The ongoing debating between the cost of force and the cost of work as determinants of organismal cost should be acknowledged here. This could then also lead to a more nuanced discussion of factors dictating muscle energetics beyond simply level of activation. Response: As responded in more detail to the previous comment, we would not like to refer the manuscript to the discussion of cost of work and force because this is beyond the scope of the present study. By our study design (force and work not measured) and results we cannot provide any significant contribution to the mentioned ongoing discussion. Comment: Lines 123-124 and 185-187 – It is relatively unclear to me how the force-velocity relationship was determined here. It appears as though force and velocity were determined as fibers shortened against the tendon? Can the authors make this clearer, better define where in the contraction force and velocity were determined, and comment on how this might affect findings compared to a more standard isotonic or isovelocity protocol. Response: The force-velocity curve in the present study was not derived from experimentally measured force estimates and fascicle velocities. In the first version of the manuscript Vmax was calculated based on the soleus muscle-specific constants of arel and brel reported by literature [10] as 11.75 LO/s. According to a comment from the other reviewer, we now based our choice of Vmax on more biological evidence as follows. The in vitro study of Luden et al., (2008) on the human soleus muscle reported Vmax values for MHC I type fibers of 0.77 LO/s and 2.91 LO/s for MHC IIA type fibers measured at 15°C [3]. Considering the temperature coefficient provided by Ranatunga et al., (1984) [4] it can be predicted that Vmax would increase to 4.4 LO/s for MHC I type fibers and to 16.8 LO/s for MHC IIA type fibers under physiological temperature conditions (37 °C). The fiber type distribution in the human soleus muscle can be estimated from literature reports, i.e. Johnson et al., 1973: type 1 fibers 87,7%, type 2 fibers (a and b) 12,3% (average of surface and deep fiber location) [5]; Larsson and Moss, 1993: type 1 89%, type 2A 11% [6]; Edgerton et al., 1975: slow twitch 70%, fast twitch 30% [7]; Luden et al., 2008: 74% MHC I, 20% MHC IIA (norm to 100%) [3]). Using an average of this reported distribution values (type 1: 81%, type 2: 19%), Vmax for soleus under physiological temperature can be calculated as 6.77 LO/s. The broad literature basis for the average fiber type distribution was also used to update arel to 0.175 (i.e. 0.1+0.4FT, where FT is the fast twitch fiber type percentage [8,9]) and accordingly brel to 1.182 (arel * Vmax [10]). We then assessed the force-velocity curve by using the classical Hill formula, (i.e. (F+a)(v+b)=(Fmax+a)b), and the muscle-specific values of Vmax, arel and brel. We recalculated the respective values (force-velocity potential and normalized velocities and their ranges) using the updated Vmax of 6.77 LO/s, arel and brel in the revised manuscript. Note that this adjustment in the calculation did not changed any statistical result but only few numerical expressions (underlined in the revision). A revised and more detailed description of the calculation of the force- velocity potential is also now provided in the updated manuscript (see below). The reason why we did not measured Vmax experimentally is that precise measurements of Vmax in vivo in humans are extremely challenging, technically and methodologically (e.g. restricted high dynamometer velocities, limited ultrasound capture frequencies in high velocities, limited range of motion to reach maximum force in high velocities, consideration of antagonistic co-contraction, mechanical properties of the tendon, history dependence effects). We added the following information to the revised manuscript (page: 4, line: 155): “Furthermore, we assessed the force-velocity relationship of soleus using the classical Hill equation [11] and the muscle-specific maximum fascicle shortening velocity (Vmax) and constants of arel and brel. Vmax was derived from the study of Luden et al. (2008), which showed Vmax values for type 1 fibers of 0.77 LO/s and 2.91 LO/s for type 2 fibers of the human soleus muscle measured in vitro at 15°C [3]. Considering the temperature coefficient [4], Vmax can be predicted as 4.4 LO/s for type 1 fibers and 16.8 LO/s for type 2 fibers under physiological temperature conditions (37 °C). Using an average fiber type distribution (type 1 fibers: 81%, type 2: 19%) of the human soleus muscle reported in literature [3,5–7], Vmax can be calculated as 6.77 LO/s. arel was calculated as 0.1+0.4FT, where FT is the fast twitch fiber type percentage (see above), which then equals to 0.175 [8,9]. The product of arel and Vmax then gives brel as 1.182 [10]. After rearrangement of the Hill formula and extension to the eccentric component, the normalized operating velocity (to Vmax) can be used to calculate the individual force potential according to the force-velocity curve.” Comment: Line 197-198 – The meaning of this is unclear to me. This description of touchdown and toe-off should be reworded for clarification. Response: We changed the description to be more clear as follows (page: 4, line: 175): “The touchdown of the foot and toe off were defined by the kinematic data as the first and second peak in knee extension, respectively [36,37].“ Comment: The results section is relatively dense. The authors may wish to consider moving some of the findings less critical to addressing their question to a table, to improve readability. Response: Some of the results are now presented in the table to improve readability. Comment: Line 299-300 – The assertion that the triceps surae consumes 40% of the cost during running is crucial to the argument of this paper. Yet it is not clear how this value is arrived at from the Fletcher and MacIntosh paper cited (the paper seems to give a large range of values for muscle energy consumption and not to relate this to organismal cost), and how reliable the output of their simple model is for this purpose. Could the authors give a little more detail on this (in the manuscript if of sufficient interest, or simply here). It may also be useful to combine this 40% estimate with the relative size of soleus to give a better representation of its likely contribution to energy consumption, considering fiber type as soleus is likely cheaper than gastrocs (Barclay, 1993). Response: The statement that the triceps surae consumes 40% of the energy during running can be derived from the comparison of figure 4 and 5 in the paper of Fletcher and MacIntosh (2015) and is numerically presented by the authors themselves in several subsequent published manuscripts (e.g. [38,39]). We agree with the reviewer that the presented calculations on muscle energy consumption in the aforementioned study may only provide a rough estimate. We also do not persist on the fixed value of 40% but rather we would like to understand this value as an indication of the great contribution of the triceps surae to the overall energetic cost. Within the triceps surae the gastrocnemius medialis and lateralis contribute to the propulsion as well but the physiological cross-sectional area (PCSA) and volume of soleus are notably higher (soleus: 131 cm² and 477 cm³, gastrocnemius medialis: 51 cm² and 285 cm³, gastrocnemius lateralis: 24 cm ² and 146 cm³ [18]). Further calculations on the separate contribution of the single muscles of the triceps surae based on portions of force are very difficult if even possible because of strong underlying assumptions of the calculation. E.g., calculating the soleus muscle force using the PCSA relative to the other triceps muscles (gastroc. med and lat.) would premise that the force-potential due to the force length/velocity relationship and activation of all triceps surae muscles are equal. This assumption cannot be correct because the gastrocnemi are biarticular muscles. For this reason, we would not like to include this approach in our manuscript but rather stay on the more direct findings. We softened our formulation by deleting the 40% in revised manuscript (page: 7, line: 280). Comment: Lines 304-309 - The authors make a good case for why small changes in velocity would require an increase in activation and therefore cost. This effect should be seen in EMG recordings. It would seem that the argument could be strengthened by showing this as it would provide a more causal link between the change in muscle level function and organismal level cost. Response: Thanks for this comment. We did not go into any correlation analysis in the study because the parameter of surface EMG activation does not reflects active muscle volume adequately. However, a significant correlation can be found for the force-length-velocity potential (EMG mean: r = -0.504, p = 0.028; EMG max: r = -0.525, p = 0.021; EMG integral: : r = -0.504, p = 0.028). Please note that the processing of the EMG signal can affect the correlation coefficients but not the significance itself (p < 0.05). Here a 20 Hz low pass filter was used after rectification and preprocessing with a high pass filter of 50 Hz. Given the mentioned limitation, the observed correlation might provide a cautious indication that a decreased EMG activity is associated with a higher force-length-velocity potential of the soleus muscle during the stance phase of running and that may affect the metabolic cost. We added the association between EMG activity and force-length-velocity potential in the revised manuscript without an extended interpretation because, as we mentioned before, the active muscle volume cannot be assessed accurately from the EMG activity (page: 6, line: 255; page: 7, line: 291). Comment: Line 334-335 – There seems to be some discrepancy regarding activation in here. The implication seems to be that muscle activation is higher in early stance to enable the tendon to be stretched, and then recoil to slow shortening velocity in the later part of stance. Yet a central claim of the paper is that cost is lower when shortening velocity is lower, due to a lower requirement for activation. It seems like the variation in required activation could balance out over the course of a stance phase? Could it be clarified as to why the early increase in activation to enable tendon stretch doesn’t seem to be costly in the way that the latter reduction is deemed to be cheap? Response: Thanks for this comment. The rationale of this argumentation is that the observed activation pattern can be interpreted as appropriate for a coordinated MTU interaction during the running task with respect to economy. We changed the formulation in the respective section as follows (page: 7, line: 322): “The soleus produces mechanical work/energy for the lift and acceleration of the body throughout the entire stance phase. In the first half, where the MTU is elongated, the fascicles actively shorten. This means that a part of the mechanical energy of the human body is transferred to the tendon. Also, in this setting the muscle fascicles produce work under favorable conditions due to the force-length and force-velocity relationships (both potentials in this phase were very high) and save work as strain energy in the tendon. In the second half, the tendon strain energy is returned and at the same time the fascicles produce work by active shortening at a reduced force-velocity potential (fascicle shortening velocity is higher in this phase). The higher shortening velocity is associated with a reduction in the EMG activity and an increase in belly gearing. It has been suggested that increased gearing at fast shortening velocities and lower forces is a mechanism that allows particularly slower type fascicles to be more effective in generating forces [40]. This supports the idea that the observed activation pattern fostered an economical MTU interaction during running.” Comment: Line 345 – typo “were”? Response: We corrected the typo accordingly. Comment: Line 375 – The study doesn’t seem to show that energy consumption is related to the force-length-velocity potential, but rather just the force-velocity potential. Response: The force-length-velocity potential is the product of the force-length and force-velocity potential and was inversely associated with the energetic cost like the force-velocity potential. The force-length potential was consistently high among the participants and showed no significant association to the energetic cost. This indicates that the reason for the association of the force-length- velocity potential to the energetic cost was caused by the observed correlation of the force-velocity potential, i.e. variability in the force-length-velocity potential relied on the variability of the force- velocity potential that cohered the variability of the energetic cost. However, as we mentioned in the discussion, a high force-length potential is also important for economical muscle force generation. References used for the responses: 1. Epstein M. 1998 Theoretical models of skeletal muscle: biological and mathematical considerations. Chichester [u.a.]: Chichester ua : Wiley. 2. Zajac FE. 1989 Muscle and tendon: properties, models, scaling, and application to biomechanics and motor control. Crit. Rev. Biomed. Eng. 17, 359–411. 3. Luden N, Minchev K, Hayes E, Louis E, Trappe T, Trappe S. 2008 Human vastus lateralis and soleus muscles display divergent cellular contractile properties. Am. J. Physiol. - Regul. Integr. Comp. Physiol. 295, R1593– R1598. (doi:10.1152/ajpregu.90564.2008) 4. Ranatunga KW. 1984 The force-velocity relation of rat fast- and slow-twitch muscles examined at different temperatures. J. Physiol. 351, 517–529. 5. Johnson MA, Polgar J, Weightman D, Appleton D. 1973 Data on the distribution of fibre types in thirty-six human muscles. An autopsy study. J. Neurol. Sci. 18, 111–129. (doi:10.1016/0022-510x(73)90023-3) 6. Larsson L, Moss RL. 1993 Maximum velocity of shortening in relation to myosin isoform composition in single fibres from human skeletal muscles. J. Physiol. 472, 595–614. (doi:10.1113/jphysiol.1993.sp019964) 7. Edgerton VR, Smith JL, Simpson DR. 1975 Muscle fibre type populations of human leg muscles. Histochem. J. 7, 259–266. 8. Winters JM, Stark L. 1985 Analysis of Fundamental Human Movement Patterns Through the Use of In- Depth Antagonistic Muscle Models. IEEE Trans. Biomed. Eng. BME-32, 826–839. (doi:10.1109/TBME.1985.325498) 9. Winters JM, Stark L. 1988 Estimated mechanical properties of synergistic muscles involved in movements of a variety of human joints. J. Biomech. 21, 1027–1041. (doi:10.1016/0021-9290(88)90249-7) 10. Miller RH, Umberger BR, Caldwell GE. 2012 Sensitivity of maximum sprinting speed to characteristic parameters of the muscle force–velocity relationship. J. Biomech. 45, 1406–1413. (doi:10.1016/j.jbiomech.2012.02.024) 11. Hill Archibald Vivian. 1938 The heat of shortening and the dynamic constants of muscle. Proc. R. Soc. Lond. Ser. B - Biol. Sci. 126, 136–195. (doi:10.1098/rspb.1938.0050) 12. Nikolaidou ME, Marzilger R, Bohm S, Mersmann F, Arampatzis A. 2017 Operating length and velocity of human M. vastus lateralis fascicles during vertical jumping. R. Soc. Open Sci. 4, 170185. (doi:10.1098/rsos.170185) 13. Santuz A, Ekizos A, Janshen L, Baltzopoulos V, Arampatzis A. 2017 On the Methodological Implications of Extracting Muscle Synergies from Human Locomotion. Int. J. Neural Syst. 27, 1750007. (doi:10.1142/S0129065717500071) 14. Marzilger R, Legerlotz K, Panteli C, Bohm S, Arampatzis A. 2018 Reliability of a semi-automated algorithm for the vastus lateralis muscle architecture measurement based on ultrasound images. Eur. J. Appl. Physiol. 118, 291–301. (doi:10.1007/s00421-017-3769-8) 15. Bohm S, Marzilger R, Mersmann F, Santuz A, Arampatzis A. 2018 Operating length and velocity of human vastus lateralis muscle during walking and running. Sci. Rep. 8, 5066. (doi:10.1038/s41598-018-23376-5) 16. Fletcher JR, MacIntosh BR. 2015 Achilles tendon strain energy in distance running: consider the muscle energy cost. J. Appl. Physiol. 118, 193–199. (doi:10.1152/japplphysiol.00732.2014) 17. Novacheck TF. 1998 The biomechanics of running. Gait Posture 7, 77–95. (doi:10.1016/S0966- 6362(97)00038-6) 18. Albracht K, Arampatzis A, Baltzopoulos V. 2008 Assessment of muscle volume and physiological cross- sectional area of the human triceps surae muscle in vivo. J. Biomech. 41, 2211–2218. (doi:10.1016/j.jbiomech.2008.04.020) 19. Hamner SR, Delp SL. 2013 Muscle contributions to fore-aft and vertical body mass center accelerations over a range of running speeds. J. Biomech. 46, 780–787. (doi:10.1016/j.jbiomech.2012.11.024) 20. Dorn TW, Schache AG, Pandy MG. 2012 Muscular strategy shift in human running: dependence of running speed on hip and ankle muscle performance. J. Exp. Biol. 215, 1944–1956. (doi:10.1242/jeb.064527) 21. Josephson RK. 1999 Dissecting muscle power output. J. Exp. Biol. 202, 3369–3375. 22. Chen J, Hahn D, Power GA. 2019 Shortening-induced residual force depression in humans. J. Appl. Physiol. 126, 1066–1073. (doi:10.1152/japplphysiol.00931.2018) 23. Herzog W. 2004 History dependence of skeletal muscle force production: implications for movement control. Hum. Mov. Sci. 23, 591–604. (doi:10.1016/j.humov.2004.10.003) 24. Maréchal G, Plaghki L. 1979 The deficit of the isometric tetanic tension redeveloped after a release of frog muscle at a constant velocity. J. Gen. Physiol. 73, 453–467. 25. Abbott BC, Aubert XM. 1952 The force exerted by active striated muscle during and after change of length. J. Physiol. 117, 77–86. (doi:10.1113/jphysiol.1952.sp004733) 26. De Ruiter CJ, De Haan A, Jones DA, Sargeant AJ. 1998 Shortening-induced force depression in human adductor pollicis muscle. J. Physiol. 507, 583–591. (doi:10.1111/j.1469-7793.1998.583bt.x) 27. Edman KA, Elzinga G, Noble MI. 1982 Residual force enhancement after stretch of contracting frog single muscle fibers. J. Gen. Physiol. 80, 769–784. (doi:10.1085/jgp.80.5.769) 28. Joumaa V, Fitzowich A, Herzog W. 2017 Energy cost of isometric force production after active shortening in skinned muscle fibres. J. Exp. Biol. 220, 1509–1515. (doi:10.1242/jeb.117622) 29. Rack PMH, Westbury DR. 1969 The effects of length and stimulus rate on tension in the isometric cat soleus muscle. J. Physiol. 204, 443–460. (doi:10.1113/jphysiol.1969.sp008923) 30. Holt NC, Azizi E. 2014 What drives activation-dependent shifts in the force–length curve? Biol. Lett. 10, 20140651. (doi:10.1098/rsbl.2014.0651) 31. Holt NC, Azizi E. 2016 The effect of activation level on muscle function during locomotion: are optimal lengths and velocities always used? Proc R Soc B 283, 20152832. (doi:10.1098/rspb.2015.2832) 32. Fontana H de B, Herzog W. 2016 Vastus lateralis maximum force-generating potential occurs at optimal fascicle length regardless of activation level. Eur. J. Appl. Physiol. 116, 1267–1277. (doi:10.1007/s00421- 016-3381-3) 33. Barclay CJ. 2015 Energetics of contraction. Compr. Physiol. 5, 961–995. (doi:10.1002/cphy.c140038) 34. Stephenson DG, Stewart AW, Wilson GJ. 1989 Dissociation of force from myofibrillar MgATPase and stiffness at short sarcomere lengths in rat and toad skeletal muscle. J. Physiol. 410, 351–366. (doi:10.1113/jphysiol.1989.sp017537) 35. Hilber K, Sun Y-B, Irving M. 2001 Effects of sarcomere length and temperature on the rate of ATP utilisation by rabbit psoas muscle fibres. J. Physiol. 531, 771–780. (doi:10.1111/j.1469-7793.2001.0771h.x) 36. Smith L, Preece S, Mason D, Bramah C. 2015 A comparison of kinematic algorithms to estimate gait events during overground running. Gait Posture 41, 39–43. (doi:10.1016/j.gaitpost.2014.08.009) 37. Fellin RE, Rose WC, Royer TD, Davis IS. 2010 Comparison of methods for kinematic identification of footstrike and toe-off during overground and treadmill running. J. Sci. Med. Sport 13, 646–650. (doi:10.1016/j.jsams.2010.03.006) 38. Fletcher JR, MacIntosh BR. 2018 Theoretical considerations for muscle-energy savings during distance running. J. Biomech. 73, 73–79. (doi:10.1016/j.jbiomech.2018.03.023) 39. Fletcher JR, MacIntosh BR. 2017 Running Economy from a Muscle Energetics Perspective. Front. Physiol. 8, 433. (doi:10.3389/fphys.2017.00433) 40. Wakeling JM, Blake OM, Wong I, Rana M, Lee SSM. 2011 Movement mechanics as a determinate of muscle structure, recruitment and coordination. Philos. Trans. R. Soc. Lond. B Biol. Sci. 366, 1554–1564. (doi:10.1098/rstb.2010.0294) Referee: 2 I appreciate the authors addressing these comments and amending the manuscript accordingly. A few final responses to the authors’ responses are made in red below Comment: In this study, the authors explore the effect of muscle force-length and force-velocity conditions on the energetic cost of running. In addition, they explore the determinants of muscle fiber length change. I find this to be an extensive and well collected data set that uses in vivo determination of force-length and force-velocity relationships, and application of these to muscle function during running to address these questions. There appear to me to be a few major limitations of the study. These should be addressed throughout. Response: Thank you for your valuable comments. All changes are underlined in the revised version of the manuscript and the references cited in the responses can be found at the end of the document. Please note that some parts of the methods are now presented in the electronic supplementary material due to length restrictions of the journal. Comment: 1) Organismal energy consumption is measured, but only the length and velocity profile of soleus. This prevents the authors from drawing the more interesting conclusion that muscle shortening velocity is a determinant of energy consumption. This is acknowledged and somewhat addressed in the discussion, but remains a major limitation to the study. Response: We agree with the reviewers comment that although the soleus contributes to a great portion of the overall energetic costs during running [16] other limb muscles are involved that were not considered in the present study and this remains a limitation. However, the main energy source (positive work) is the ankle joint (41%) while the contribution of the knee and hip joint is comparably lower during running [17]. The soleus is the greatest muscle among the main plantar flexors with respect to physiological cross-sectional area and volume (soleus: 131 cm² and 477 cm³, gastrocnemius medialis: 51 cm² and 285 cm³, gastrocnemius lateralis: 24 cm ² and 146 cm³ [18]), giving this muscle a key role. Using a modeling approach, Hamner and Delp (2013) showed that the contribution of soleus to the vertical acceleration of the center of mass at the same velocity as in our study (i.e. 2.5 m/s) is remarkably higher than those of the other lower limb muscles (7.5x than gastroc., 9.5x than vasti, 24x than rect. fem., 9.5x than tib. ant., 12x than glut. max.; visual inspection of fig. 5). For the fore-aft acceleration a similar superior contribution of soleus was shown (3x higher than gastroc. and 8.8x higher than hamstrings [19]). The propulsive function of soleus during running is achieved by active shortening. Active shortening reduces the force-velocity potential as discussed extensively in the present manuscript. Consequently, a greater active muscle volume is required to achieve the required mechanical energy gain. In contrast, the quadriceps muscle group, the main contributor during early stance, decelerating and supporting body mass [19,20], features more economical fascicle dynamics. Recently we showed that the fascicles of the vastus lateralis muscle, as a representative of the quadriceps muscle group operates with a high force-length (i.e. 0.91) and force-velocity potential (i.e. 0.97) during the stance phase of running. Operating at high force potentials reduces the energetic cost of this energetically expensive (due to its long fascicle length) muscle by reducing active muscle volume. This may indicate that the Appendix C mechanical energy by muscular work required for steady state running is generated by muscles that are metabolically less expensive, likely to compensate for the reduction of the force-velocity potential. We discussed the reviewers comment as a limitation in the revised manuscript (page: 8, line: 360): “Although the soleus likely contributes to a great portion of the overall energetic costs during running, other limb muscles that were not considered in the present study are involved. However, the main energy source (positive work) is the ankle joint (41%) [17] and the soleus is the greatest muscle among the main plantar flexors with respect to physiological cross-sectional area (soleus 63%, gastroc. med. 25%, gastroc. lat. 12%) and volume (53%, 31% and 16% [18]). The key role of soleus is further supported by the modeling study of Hamner and Delp (2013), which showed that the soleus is by far the biggest contributor to the vertical acceleration and fore-aft acceleration of the center of mass [19]. This function is achieved by active shortening, which reduces the force-velocity potential and consequently requires a greater active muscle volume. In contrast, the quadriceps muscle group, the main contributor during early stance, decelerating and transferring body mass [19,20], features more economical fascicle dynamics. Recently we showed that the fascicles of the vastus lateralis muscle as a representative of the quadriceps muscle group operates with a high force-length (i.e. 0.91) and force- velocity potential (i.e. 0.97) during the stance phase of running. Operating at high force potentials minimizes the cost of this muscle, which is energetically expensive due to its long fascicle length (i.e. LO = 94 mm [15]), by reducing active muscle volume. This may indicate that the mechanical energy by muscular work required for steady state running is generated by muscles that are metabolically less expensive, likely to compensate for the reduction of the force-velocity potential.” Comment: 2) The strict adherence to force-length and force-velocity relationships as defining features of muscle performance seems somewhat outdated given a wealth of literature showing that these relationships do not hold under conditions relevant to locomotion (i.e. history dependence, activation-dependent changes). These advances do not negate this study, however, it would be a more accurate representation of the field to acknowledge that they exist, and present this study as a means to investigate the importance of these relationships. Response: First note – History dependence: We agree with the reviewers comment that the phenomenon of history dependence of force production after active muscle lengthening or shortening may be present for the soleus during running and may affect the force production [21–23]. In the present study, the soleus fascicles shortened continuously during running when activated, which would indicate a condition of force depression. Force depression has been shown to increase with increasing shortening magnitude [24], with decreasing shortening velocity [25] and with increasing activation levels [26]. Since the soleus shortening magnitude was notable (25.9 ± 7.8 %LO), the shortening velocity moderate (0.118 ± 0.039 Vmax) and the activation submaximal (average during stance phase: 0.32 ± 0.19 EMGmax; maximum activation: 0.52 ± 0.18 EMGmax), an effect of force depression on the force production can theoretically be expected. Yet, the force-length and force-velocity relationships remain the basic mechanisms for muscle force production. Interestingly, force depression is likely to be reduced due to the tendon and belly gearing mechanisms because those reduce the shortening magnitude and activation. The observed main finding of a correlation of the operating velocity and force-velocity potential with the energetic cost, however, does not neglect the presence of force depression but indicates that shortening velocity and consequently the force-velocity potential has a direct effect on the muscle energetics. We added the following sentences in the introduction and discussion of the revised version of the manuscript (page: 2, line: 59; page: 7, line: 307): “Besides the operating length and velocity as the main determinants, the history dependence of force generation [23], i.e. increased force after active muscle lengthening [27] and decreased force after active shortening [22,25], may additionally influence the force potential.” “Furthermore, we showed that the soleus shortened continuously during the stance phase of running, which reflects a condition for force depression. Since a depression of force was shown to be accompanied by a decrease in the ATPase activity [28], force depression would have little or no effect on the energetic cost itself.” Second note – Shift in optimal length: Furthermore, it is correct that we assessed the force-length curve during maximal isometric contractions at different ankle joint angles and, using this relationship, we calculated the force-length potential of the soleus muscle during running at submaximal activation. There is evidence from early [29] and more recent [30,31] in vitro studies that the force-length curve depends on muscle activation, i.e. optimum length increases with submaximal activation. However, a recent study by Fontana and Herzog (2016) on the human vastus lateralis muscle showed that this holds not necessarily true for in vivo assessments [32]. In contrast to the in vitro studies, a rightward shift of optimal length was not observed when force was normalized to the maximum EMG signal (i.e. optimal length remained constant at different levels of activation). The authors suggested that the disagreement of the in vitro and in vivo studies might be an artefact related to the in vitro testing setup (e.g. non-physiological stimulation frequency range or calcium concentrations). Therefore, we can argue that mapping the submaximal fascicle operating length onto the force-length curve in the present in vivo study should not affect the findings. We added the following information in the discussion part of revised manuscript as follows (page: 9, line: 385): “Furthermore, we assessed the force-length curve during maximal isometric contractions and used it to calculate the force-length potential of the soleus muscle during running at submaximal activation. There is evidence from in vitro studies that the force-length curve depends on muscle activation [29– 31]. However, in a recent in vivo study by Fontana and Herzog (2016) on the human vastus lateralis muscle, a rightward shift of optimal length with submaximal activation was not observed when force was normalized to the maximum EMG signal [32]. The authors suggested that the shift in optimal length phenomenon might be related to the in vitro testing setup (e.g. non-physiological stimulation frequency range or Ca2+concentrations). Therefore, we can argue that mapping the submaximal fascicle operating length onto the force-length curve in the present in vivo study should not affect the findings.” I find the rationale of this Herzog and Fontana paper quite difficult to follow, particular with regards to how they cite other studies. They attempt to distinguish between activation and force which, while may have some bearing on the mechanism responsible, does nothing to counter the finding that shifts in optimum length not predicted by the sliding filament theory occur with changing contractile conditions. For the purposes of this argument, what I take from this study is that if you change muscle force production, optimum length shifts (Fig. 3 Herzog and Fontana 2016). This is in line with other in vivo studies of this phenomenon (Ichinose et al., 1997; Kwah et al 2013), and makes it less obvious that this potential effect should be ignored in the present paper. This does not change the arguments of this paper, and I leave this to the authors discretion, but it is my feeling that it would be a stronger paper if it shifted its focus from the conviction that force-length and force-velocity potential of the muscle dictates in vivo performance (if this were true, Hill-type muscle models would do a better job (Lee et al., 2013; Dick et al., 2017)) to an argument that multiple factors influence muscle mechanical and energetic performance under dynamic conditions, and that this paper seeks to understand to what extent force-velocity effects dictate energetic performance. It’s a subtle shift that would require minor rewording throughout, but it’s my feeling that this would much better reflect that state of the field. Reference to the Herzog and Fontana paper is made again in the discussion, the authors may wish to consider how well it supports their argument and the contradictory findings of other in vivo papers. Comment: 3) This paper is fundamentally concerned with the effect of contractile conditions on muscle energy consumption. However, there is very little discussion of why length and velocity might affect energy consumption beyond required activation, despite a wealth of evidence on this i.e. how the cost per unit force varies across the force-length relationship in isolated muscle. In addition, it may be worth considering findings such as the effect of contractile history on cost (Joumaa et al., 2013), and the complexity of the cost of work (Holt et al., 2104; Curtin et al., 2019) in a more comprehensive discussion of in vivo muscle energetics. Response: First note - Variation of cost per unit force across the force-length relationship: We agree with the reviewer that the energy turnover can differ across the force-length relationship in isolated animal muscle fibers tested in vitro [33]. During isometric contractions at sarcomere length shorter than optimal length, the force output is reduced but the ATPase rate seems not to greatly differ from the rate at optimal length, indicating a comparably higher cost of contraction at shorter length [34,35]. However, this effect seems to be more pronounced at very short lengths, which might not be covered during regular in vivo movements like locomotion, i.e. soleus operating range (0.75-1.01 LO). We added the following information in the revised version of the manuscript as follows (page: 7, line: 300): “Besides the favorable high force-length potential for economical force production, operating close to optimal length may additionally preserved from relatively higher energetic cost that can arise when contracting at shorter length. In vitro evidence showed that although force is reduced at shorter sarcomere length, the ATPase rate seems not to differ from the rate at optimal length, indicating comparably higher cost of contraction at shorter length [34,35]. However, this effect seems more pronounced at very short lengths, a portion of the force-length curve that is likely not covered by the soleus during running (operating range 0.75-1.01 LO).”  Second note - Effect of contractile history on cost: We added the following paragraph to the discussion part of the revised manuscript (page: 7, line: 307): “Furthermore, we showed that the soleus shortened continuously during the stance phase of running, which reflects a condition for force depression. Since a depression of force was shown to be accompanied by a decrease in the ATPase activity [28], force depression would have little or no effect on the energetic cost itself.” Third note - complexity of the cost of work: Thanks for this comment. We agree with the reviewer on the ongoing debate on the cost of force and the cost of work. From our perspective, when a muscle contracts, force is generated and this consumes metabolic energy independently of the contraction type (i.e. isometric, eccentric, concentric). During concentric contractions (active shortening) positive mechanical work is generated and during eccentric contractions (active lengthening) the work is negative. In stretch-shortening conditions, the net work could be zero when positive and negative work cancel each other out. Under isometric contractions, no mechanical work is generated by definition, which would again indicate no mechanical energy production (Joule), although force is generated and metabolic energy expended. The energy index of work in the context of the explanation of metabolic energy, therefore, might not be very appropriate (metabolic energy is not zero when work is zero e.g. during isometric contractions). Instead, an index of force and metabolic energy might better reflect the organismal cost during locomotion. With our study, we cannot provide any new information on this discussion because work and force of soleus were not measured during running (which in our opinion is not possible at the moment). Therefore, we think that this topic is beyond of the scope of the present study and for this reason we would prefer not to go deeper in the discussion of cost of force and work but rather stay close to our experimental results. This cost of work argument could entirely be thought of as cost of muscle fiber shortening argument. Which is obviously very pertinent to this paper. It is my opinion that this paper would be strengthened by greater discussion of this complexity and what the data presented here do to advance our understanding -i.e. cheap work (shortening) may be possible in some cases (Holt et al., 2014; Curtin et al., 2019), but in this case, more rapid active muscle shortening does seem to incur energetic costs. But again, I leave this to the authors discretion. Specific comments Comment: Lines 50-51 – The ongoing debating between the cost of force and the cost of work as determinants of organismal cost should be acknowledged here. This could then also lead to a more nuanced discussion of factors dictating muscle energetics beyond simply level of activation. Response: As responded in more detail to the previous comment, we would not like to refer the manuscript to the discussion of cost of work and force because this is beyond the scope of the present study. By our study design (force and work not measured) and results we cannot provide any significant contribution to the mentioned ongoing discussion. Comment: Lines 123-124 and 185-187 – It is relatively unclear to me how the force-velocity relationship was determined here. It appears as though force and velocity were determined as fibers shortened against the tendon? Can the authors make this clearer, better define where in the contraction force and velocity were determined, and comment on how this might affect findings compared to a more standard isotonic or isovelocity protocol. Response: The force-velocity curve in the present study was not derived from experimentally measured force estimates and fascicle velocities. In the first version of the manuscript Vmax was calculated based on the soleus muscle-specific constants of arel and brel reported by literature [10] as 11.75 LO/s. According to a comment from the other reviewer, we now based our choice of Vmax on more biological evidence as follows. The in vitro study of Luden et al., (2008) on the human soleus muscle reported Vmax values for MHC I type fibers of 0.77 LO/s and 2.91 LO/s for MHC IIA type fibers measured at 15°C [3]. Considering the temperature coefficient provided by Ranatunga et al., (1984) [4] it can be predicted that Vmax would increase to 4.4 LO/s for MHC I type fibers and to 16.8 LO/s for MHC IIA type fibers under physiological temperature conditions (37 °C). The fiber type distribution in the human soleus muscle can be estimated from literature reports, i.e. Johnson et al., 1973: type 1 fibers 87,7%, type 2 fibers (a and b) 12,3% (average of surface and deep fiber location) [5]; Larsson and Moss, 1993: type 1 89%, type 2A 11% [6]; Edgerton et al., 1975: slow twitch 70%, fast twitch 30% [7]; Luden et al., 2008: 74% MHC I, 20% MHC IIA (norm to 100%) [3]). Using an average of this reported distribution values (type 1: 81%, type 2: 19%), Vmax for soleus under physiological temperature can be calculated as 6.77 LO/s. The broad literature basis for the average fiber type distribution was also used to update arel to 0.175 (i.e. 0.1+0.4FT, where FT is the fast twitch fiber type percentage [8,9]) and accordingly brel to 1.182 (arel * Vmax [10]). We then assessed the force-velocity curve by using the classical Hill formula, (i.e. (F+a)(v+b)=(Fmax+a)b), and the muscle-specific values of Vmax, arel and brel. We recalculated the respective values (force-velocity potential and normalized velocities and their ranges) using the updated Vmax of 6.77 LO/s, arel and brel in the revised manuscript. Note that this adjustment in the calculation did not changed any statistical result but only few numerical expressions (underlined in the revision). A revised and more detailed description of the calculation of the force- velocity potential is also now provided in the updated manuscript (see below). The reason why we did not measured Vmax experimentally is that precise measurements of Vmax in vivo in humans are extremely challenging, technically and methodologically (e.g. restricted high dynamometer velocities, limited ultrasound capture frequencies in high velocities, limited range of motion to reach maximum force in high velocities, consideration of antagonistic co-contraction, mechanical properties of the tendon, history dependence effects). We added the following information to the revised manuscript (page: 4, line: 155): “Furthermore, we assessed the force-velocity relationship of soleus using the classical Hill equation [11] and the muscle-specific maximum fascicle shortening velocity (Vmax) and constants of arel and brel. Vmax was derived from the study of Luden et al. (2008), which showed Vmax values for type 1 fibers of 0.77 LO/s and 2.91 LO/s for type 2 fibers of the human soleus muscle measured in vitro at 15°C [3]. Considering the temperature coefficient [4], Vmax can be predicted as 4.4 LO/s for type 1 fibers and 16.8 LO/s for type 2 fibers under physiological temperature conditions (37 °C). Using an average fiber type distribution (type 1 fibers: 81%, type 2: 19%) of the human soleus muscle reported in literature [3,5–7], Vmax can be calculated as 6.77 LO/s. arel was calculated as 0.1+0.4FT, where FT is the fast twitch fiber type percentage (see above), which then equals to 0.175 [8,9]. The product of arel and Vmax then gives brel as 1.182 [10]. After rearrangement of the Hill formula and extension to the eccentric component, the normalized operating velocity (to Vmax) can be used to calculate the individual force potential according to the force-velocity curve.” Line 104-103 – it therefore seems misleading to say ‘as a function of their experimentally assessed force-velocity relationships’ Line 126-127 – similar issue in that this seems to suggest experimental measurements of force-velocity relationships in this study Comment: Line 197-198 – The meaning of this is unclear to me. This description of touchdown and toe-off should be reworded for clarification. Response: We changed the description to be more clear as follows (page: 4, line: 175): “The touchdown of the foot and toe off were defined by the kinematic data as the first and second peak in knee extension, respectively [36,37].“ Comment: The results section is relatively dense. The authors may wish to consider moving some of the findings less critical to addressing their question to a table, to improve readability. Response: Some of the results are now presented in the table to improve readability. Comment: Line 299-300 – The assertion that the triceps surae consumes 40% of the cost during running is crucial to the argument of this paper. Yet it is not clear how this value is arrived at from the Fletcher and MacIntosh paper cited (the paper seems to give a large range of values for muscle energy consumption and not to relate this to organismal cost), and how reliable the output of their simple model is for this purpose. Could the authors give a little more detail on this (in the manuscript if of sufficient interest, or simply here). It may also be useful to combine this 40% estimate with the relative size of soleus to give a better representation of its likely contribution to energy consumption, considering fiber type as soleus is likely cheaper than gastrocs (Barclay, 1993). Response: The statement that the triceps surae consumes 40% of the energy during running can be derived from the comparison of figure 4 and 5 in the paper of Fletcher and MacIntosh (2015) and is numerically presented by the authors themselves in several subsequent published manuscripts (e.g. [38,39]). We agree with the reviewer that the presented calculations on muscle energy consumption in the aforementioned study may only provide a rough estimate. We also do not persist on the fixed value of 40% but rather we would like to understand this value as an indication of the great contribution of the triceps surae to the overall energetic cost. Within the triceps surae the gastrocnemius medialis and lateralis contribute to the propulsion as well but the physiological cross-sectional area (PCSA) and volume of soleus are notably higher (soleus: 131 cm² and 477 cm³, gastrocnemius medialis: 51 cm² and 285 cm³, gastrocnemius lateralis: 24 cm ² and 146 cm³ [18]). Further calculations on the separate contribution of the single muscles of the triceps surae based on portions of force are very difficult if even possible because of strong underlying assumptions of the calculation. E.g., calculating the soleus muscle force using the PCSA relative to the other triceps muscles (gastroc. med and lat.) would premise that the force-potential due to the force length/velocity relationship and activation of all triceps surae muscles are equal. This assumption cannot be correct because the gastrocnemi are biarticular muscles. For this reason, we would not like to include this approach in our manuscript but rather stay on the more direct findings. We softened our formulation by deleting the 40% in revised manuscript (page: 7, line: 280). Comment: Lines 304-309 - The authors make a good case for why small changes in velocity would require an increase in activation and therefore cost. This effect should be seen in EMG recordings. It would seem that the argument could be strengthened by showing this as it would provide a more causal link between the change in muscle level function and organismal level cost. Response: Thanks for this comment. We did not go into any correlation analysis in the study because the parameter of surface EMG activation does not reflects active muscle volume adequately. However, a significant correlation can be found for the force-length-velocity potential (EMG mean: r = -0.504, p = 0.028; EMG max: r = -0.525, p = 0.021; EMG integral: : r = -0.504, p = 0.028). Please note that the processing of the EMG signal can affect the correlation coefficients but not the significance itself (p < 0.05). Here a 20 Hz low pass filter was used after rectification and preprocessing with a high pass filter of 50 Hz. Given the mentioned limitation, the observed correlation might provide a cautious indication that a decreased EMG activity is associated with a higher force-length-velocity potential of the soleus muscle during the stance phase of running and that may affect the metabolic cost. We added the association between EMG activity and force-length-velocity potential in the revised manuscript without an extended interpretation because, as we mentioned before, the active muscle volume cannot be assessed accurately from the EMG activity (page: 6, line: 255; page: 7, line: 291). Comment: Line 334-335 – There seems to be some discrepancy regarding activation in here. The implication seems to be that muscle activation is higher in early stance to enable the tendon to be stretched, and then recoil to slow shortening velocity in the later part of stance. Yet a central claim of the paper is that cost is lower when shortening velocity is lower, due to a lower requirement for activation. It seems like the variation in required activation could balance out over the course of a stance phase? Could it be clarified as to why the early increase in activation to enable tendon stretch doesn’t seem to be costly in the way that the latter reduction is deemed to be cheap? Response: Thanks for this comment. The rationale of this argumentation is that the observed activation pattern can be interpreted as appropriate for a coordinated MTU interaction during the running task with respect to economy. We changed the formulation in the respective section as follows (page: 7, line: 322): “The soleus produces mechanical work/energy for the lift and acceleration of the body throughout the entire stance phase. In the first half, where the MTU is elongated, the fascicles actively shorten. This means that a part of the mechanical energy of the human body is transferred to the tendon. Also, in this setting the muscle fascicles produce work under favorable conditions due to the force-length and force-velocity relationships (both potentials in this phase were very high) and save work as strain energy in the tendon. In the second half, the tendon strain energy is returned and at the same time the fascicles produce work by active shortening at a reduced force-velocity potential (fascicle shortening velocity is higher in this phase). The higher shortening velocity is associated with a reduction in the EMG activity and an increase in belly gearing. It has been suggested that increased gearing at fast shortening velocities and lower forces is a mechanism that allows particularly slower type fascicles to be more effective in generating forces [40]. This supports the idea that the observed activation pattern fostered an economical MTU interaction during running.” Comment: Line 345 – typo “were”? Response: We corrected the typo accordingly. Comment: Line 375 – The study doesn’t seem to show that energy consumption is related to the force-length-velocity potential, but rather just the force-velocity potential. Response: The force-length-velocity potential is the product of the force-length and force-velocity potential and was inversely associated with the energetic cost like the force-velocity potential. The force-length potential was consistently high among the participants and showed no significant association to the energetic cost. This indicates that the reason for the association of the force-length- velocity potential to the energetic cost was caused by the observed correlation of the force-velocity potential, i.e. variability in the force-length-velocity potential relied on the variability of the force- velocity potential that cohered the variability of the energetic cost. However, as we mentioned in the discussion, a high force-length potential is also important for economical muscle force generation. Response to referees  Response to referee 1  Comment:  The authors have addressed all my previous concerns in a careful manner.  Response:   Once again thank you for your valuable review.   Comment:  There  remains  one  further  comment  that  they  may  choose  to  consider  for  the  manuscript, and it still concerns the choice of Vmax. I appreciate the further analysis that the authors  have attempted, to provide a value of Vmax for the Soleus. However, it should be noted that the  running velocity of 2.5 m/s is not all that fast, and indeed the EMG averages less than 50%. As such, it  is likely that the fastest muscle fibres will not have been recruited, and hence the weighted mean taken  for Vmax may thus be an overestimate. Coupled to this, with more than half of the muscle inactive,  the actual Vmax may be less than its constituent fibres (for additional reasons: Holt et al. Proc Roy Soc  B  2014).  If  the  Vmax  for  the  Soleus  were  less  than  the  estimated  6.77  L/s  for  this  experimental  situation, then it is likely that the actual spread of Force‐velocity potentials would be larger than shown  in Fig. 3. It is thus worth considering that you have actually resulted with a conservative evaluation of  the importance of the force‐velocity potential.  Response:   Thank you for this comment. We agree with the opinion of the reviewer that Vmax during  submaximal  running  in  vivo  might  be  influenced  by  factors  not  considered  in  our  calculation  (e.g.  selective slow fiber type recruitment, muscle resistance to shortening). Referring to the results of Holt  et al. (2014), Vmax could be in deed lower. We now acknowledge this aspect in the respective paragraph  of the revised manuscript (see below, page: 8, line: 378). We also added the aspect of more economical  selective slow fiber type recruitment for submaximal slow contractions as during running to a more  comprehensive paragraph in the discussion section of the revised manuscript (see response to reviewer  2, page: 9, line: 399).  “To assess the force‐velocity potential we used a biologically funded value of Vmax, based on in vitro  studies human soleus, i.e. 6.77 L0 s−1 (279.0 ± 34.9 mm s−1). However, during submaximal running in  vivo the lower activation level and selective slow fiber type recruitment may affect the actual force‐ velocity potential of the soleus muscle.   Response to referee 2  General:   Once again thank you for your valuable review.  Comment:   I find the rationale of this Herzog and Fontana paper quite difficult to follow, particular  with regards to how they cite other studies. They attempt to distinguish between activation and force  which, while may have some bearing on the mechanism responsible, does nothing to counter the  Appendix D finding that shifts in optimum length not predicted by the sliding filament theory occur with changing  contractile conditions. For the purposes of this argument, what I take from this study is that if you  change muscle force production, optimum length shifts (Fig. 3 Herzog and Fontana 2016). This is in line  with other in vivo studies of this phenomenon (Ichinose et al., 1997; Kwah et al 2013), and makes it  less obvious that this potential effect should be ignored in the present paper.   This does not change the arguments of this paper, and I leave this to the authors discretion, but it is  my feeling that it would be a stronger paper if it shifted its focus from the conviction that force‐length  and force‐velocity potential of the muscle dictates in vivo performance (if this were true, Hill‐type  muscle models would do a better job (Lee et al., 2013; Dick et al., 2017)) to an argument that multiple  factors influence muscle mechanical and energetic performance under dynamic conditions, and that  this paper seeks to understand to what extent force‐velocity effects dictate energetic performance.  It’s a subtle shift that would require minor rewording throughout, but it’s my feeling that this would  much better reflect that state of the field. Reference to the Herzog and Fontana paper is made again  in the discussion, the authors may wish to consider how well it supports their argument and the  contradictory findings of other in vivo papers.      Response:       First note: Fontana and Herzog (2016) paper and lack of shift in optimal length  As described by Fontana  and Herzog  (2016) – and  we agree on that – the  shift in optimal length  reported in the former human in vivo study of Ichinose et al., (1997) is constrained by the experimental  setup because the authors controlled the torque (i.e. used a percentage of the maximum force) and not  the muscle activation in each of the assessed knee joint angles. Due to the force‐depended elongation  of tendon and aponeurosis, the result of a shift in optimal fascicle length is to be expected and the  conclusion of a shift in optimal length is misleading. The experimental constrain from the Ichinose et  al., (1997) study was overcome in the Fontana and Herzog (2016) study by referring the fascicle length  to activation level, leading to the lack of shift in optimal length.     We added the following text in the revised manuscript (page: 9, line: 395):     “The discrepancy of the in vitro and in vivo evidence clearly warrants future investigation to elucidate  the shifting length phenomenon in the context of in vivo submaximal locomotion. Given the current  human in vivo evidence [1], we can argue that mapping the submaximal fascicle operating length onto  the force‐length curve in the present in vivo study should not affect the findings.”      Second note: Complexity of energetic cost and muscle contraction  We  agree  with  the  opinion  of  the  reviewer  that  multiple  factors  may  affect  energetic  cost  during  submaximal  human  running  and  further  that  simple  models  not  reflect  the  complexity  of  muscle  mechanical  and  energetic  performance  under  dynamic  conditions  appropriately.  To  address  the  reviewers general comment we added the following paragraph to the limitations section of the revised  manuscript (page: 9, line: 400):    “In the present study we focused on the understanding of the contribution of the force‐length and force‐ velocity potential to the energetic cost of running and we showed that the force‐velocity potential is  inversely related to the energetic cost, explaining about one third of its variance. We argue that an  increase of active muscle volume due to the decreased force‐velocity potential would increase the  energetic  cost  of  running.  However,  it  must  be  acknowledged  that  the  energetic  cost  of  muscle  contraction is complex and multifactorial. Independent of active muscle volume, in higher shortening  velocities the rate of cross‐bridges cycling is increased and as a consequence the consumed energy. In  our study, shortening velocities of the soleus muscle were in average 0.118 Vmax throughout the stance  phase, a range where the rate of ATP hydrolysis shows a steep increase [2]. Furthermore, in submaximal  intensity contractions as during our investigated running velocity selective slow fiber type activation  might decrease the energetic cost by reducing the contribution of energetically more expensive fast  twitch fibers.”      Comment:   This  cost  of  work  argument  could  entirely  be  thought  of  as  cost  of  muscle  fiber  shortening argument. Which is obviously very pertinent to this paper. It is my opinion that this paper  would be strengthened by greater discussion of this complexity and what the data presented here do  to advance our understanding ‐i.e. cheap work (shortening) may be possible in some cases (Holt et al.,  2014; Curtin et al., 2019), but in this case, more rapid active muscle shortening does seem to incur  energetic costs. But again, I leave this to the authors discretion.    Response:   Thanks for this comment. We added the aforementioned paragraph to the manuscript  to provide a broader discussion of this topic.        Comment:   Line  104‐103  –  it  therefore  seems  misleading  to  say  ‘as  a  function  of  their  experimentally assessed force‐velocity relationships’. Line 126‐127 – similar issue in that this seems to  suggest experimental measurements of force‐velocity relationships in this study.    Response:   To avoid any confusion we reworded the sentences as follows:    “In the present study, we investigated the operating length and velocity of the soleus muscle fascicles  (i.e. bundles of fibers) during running as a function of the experimentally determined force‐length and  assessed  force‐velocity  relationships  (i.e.  force‐length  and  force‐velocity  potential)  and  their  association to the energetic cost of running. “    “The derived optimal fascicle length for force production was further used to calculate the force‐velocity  relationship of the soleus fascicles.”        References  1.  Fontana H de B, Herzog W. 2016 Vastus lateralis maximum force‐generating potential occurs at optimal fascicle length  regardless of activation level. Eur. J. Appl. Physiol. 116, 1267–1277. (doi:10.1007/s00421‐016‐3381‐3)  2.  Barclay CJ. 2015 Energetics of contraction. Compr. Physiol. 5, 961–995. (doi:10.1002/cphy.c140038)           

The force-length-velocity potential of the human soleus muscle is related to the energetic cost of running.

12-18-2019

Bohm, Sebastian,Mersmann, Falk,Santuz, Alessandro,Arampatzis, Adamantios

eng

PMC7557501

Vol.:(0123456789) 1 3 European Journal of Applied Physiology (2020) 120:2495–2505 https://doi.org/10.1007/s00421-020-04472-9 ORIGINAL ARTICLE The influence of Achilles tendon mechanical behaviour on “apparent” efficiency during running at different speeds Andrea Monte1,2  · Constantinos Maganaris2  · Vasilios Baltzopoulos2  · Paola Zamparo1 Received: 8 April 2020 / Accepted: 10 August 2020 / Published online: 25 August 2020 © The Author(s) 2020 Abstract Purpose We investigated the role of elastic strain energy on the “apparent” efficiency of locomotion (AE), a parameter that is known to increase as a function of running speed (up to 0.5–0.7) well above the values of “pure” muscle efficiency (about 0.25–0.30). Methods In vivo ultrasound measurements of the gastrocnemius medialis (GM) muscle–tendon unit (MTU) were combined with kinematic, kinetic and metabolic measurements to investigate the possible influence of the Achilles tendon mechani- cal behaviour on the mechanics (total mechanical work, WTOT) and energetics (net energy cost, Cnet) of running at different speeds (10, 13 and 16 km h−1); AE was calculated as WTOT/Cnet. Results GM fascicles shortened during the entire stance phase, the more so the higher the speed, but the majority of the MTU displacement was accommodated by the Achilles tendon. Tendon strain and recoil increased as a function of running speed (P < 0.01 and P < 0.001, respectively). The contribution of elastic energy to the positive work generated by the MTU also increased with speed (from 0.09 to 0.16 J kg−1 m−1). Significant negative correlations (P < 0.01) were observed between tendon work and metabolic energy at each running speed (the higher the tendon work the lower the metabolic demand) and significant positive correlations were observed between tendon work and AE (P < 0.001) at each running speed (the higher the tendon work the higher the efficiency). Conclusion These results support the notion that the dynamic function of tendons is integral in reducing energy expenditure and increasing the “apparent” efficiency of running. Keywords Running efficiency · Tendon mechanics · Elastic energy · Gastrocnemius medialis Abbreviations AE Apparent efficiency BCoM Body centre of mass Cnet Net energy cost Ek Kinetic energy Ep Potential energy ET Total energy F–L Force–length relationship F–V Force–velocity relationship GM Gastrocnemius medialis GRF Ground reaction force MTU Muscle–tendon unit PCSA Physiological cross-sectional area V Velocity ̇VO2net Net oxygen uptake WEXT External mechanical work (at the whole-body level) Wfas Positive work done by the fascicles WINT Internal mechanical work (at the whole-body level) WMTU Positive work done by the MTU Wten Positive work done by the tendon WTOT Total mechanical work (at the whole body level) Communicated by Olivier Seynnes. * Paola Zamparo paola.zamparo@univr.it 1 Department of Neurosciences, Biomedicine and Movement Sciences, University of Verona, via Felice Casorati, 43, 37131 Verona, Italy 2 Research Institute for Sport and Exercise Sciences (RISES), Liverpool John Moores University, Liverpool, UK 2496 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 Introduction Human locomotion entails the motion of the body through an environment: air in terrestrial locomotion whilst in con- tact with the ground, and water in aquatic locomotion. The minimum required work that has to be done to maintain the motion of any object in its surrounding environment, is given by the product of the resistance offered by the environ- ment and the distance covered during the motion. The effi- ciency of the locomotor apparatus can thus be expressed as the ratio between the work necessary to maintain motion and the chemical energy transformed by the muscles. This “loco- motion” efficiency (i.e. the total mechanical work generated at whole-body level as a proportion of metabolic cost), has been investigated in several forms of terrestrial and aquatic locomotion, such as swimming (e.g. Zamparo et al. 2002), cycling (e.g. Minetti et al. 2001), walking and running (e.g. Cavagna and Kaneko 1977; Lejeune et al. 1988; Williams and Cavanagh 1987). In the above-mentioned studies, “locomotion” efficiency was, actually, calculated as the ratio between (total) mechan- ical work per unit distance (WTOT) and net energy cost (Cnet, the metabolic energy expended per unit distance); in turn, Cnet was calculated as the ratio between net oxygen uptake and locomotion velocity ( ̇VO2net/v) and WTOT was calculated as the sum of two components: WEXT (the work done to raise and accelerate the body centre of mass within the environ- ment) and WINT (the work associated with the acceleration of body segments with respect to the centre of mass). As calculated, “locomotion” efficiency approximates “pure” muscle efficiency values (about 0.25–0.30, as reported by Woledge et al. 1985) in the forms of locomotion where elas- tic recoil is negligible [e.g. swimming or cycling, as reported by Zamparo et al. (2002) and Minetti et al. (2001)] whereas in the case of running, the efficiency calculated in this man- ner can reach far larger values [e.g. up to 0.5–0.7, as reported by Cavagna and Kaneko (1977)]. “Locomotion” efficiency is, therefore, often referred to as “apparent” efficiency because an increase beyond pure muscle efficiency values does not indicate that the muscles work in a more efficient way (Ettema 2001). Rather, these increased efficiency values are an indication of the conver- sion of metabolic energy into mechanical work at whole- body level. As suggested by Alexander (1991), measuring “apparent” efficiency can thus help in understanding whether mechanical work is “recycled” via storage and release of elastic energy (an energy saving mechanism). As an example, when running at steady-state speed, tendons stretch and recoil; through this succession of stretch–shortening cycles, tendons could play an important role as energy savers allowing this form of locomotion to be particularly efficient (Roberts and Azizi 2011). In these conditions, indeed, “apparent” efficiency (AE = WTOT/Cnet) increases linearly with speed because WTOT increases whereas Cnet does not show appreciable changes when the velocity increases (e.g. Cavagna and Kaneko 1977). In other conditions (e.g. shuttle running or uphill running), AE is much lower and this could be attributed to the fact that, in these conditions, the tendon acts more as a power amplifier (Roberts and Azizi 2011). As an example, during shuttle running, AE is lowest over short shuttle distances covered at maximal speed (e.g. when the accelerations and decelerations are larger and the tendons’ capability to save metabolic energy is expected to be reduced) and highest over long shuttle distances (e.g. in conditions that approximate those of constant speed, linear, running where the metabolic energy saving mechanism is expected to be more prominent) (Zamparo et al. 2019). A further example is that of running on sand where AE is lower relatively to running on a hard surface and this could be attributable to a decrease in “mus- cle–tendon efficiency”, the sand acting as a damper which reduces the energy that can be recoiled from the stretched tendon (Lejeune et al. 1988). Taken together, these findings suggest a link between the capability to exploit the elastic energy mechanisms in tendons and the values of “apparent” efficiency in human running. The dynamic function of tendons is indeed integral to reduce the energy expenditure of steady-state running: ener- getic savings may occur by shifting the operating regions of the muscles on their force–length and force–velocity curves (Ramsey and Street 1940), by reducing muscle work (Biewener and Roberts 2000), or by reducing active muscle volume (Holt et al. 2016). As an example, an active muscle uses less metabolic energy and produces more force when operating under isometric conditions compared to shorten- ing (Fenn 1924). In addition, if the muscle operates close to optimal length (the length corresponding to optimum myofilament overlap) it will produce more force (Gordon et al. 1966) for a given activation level. Therefore, a quasi- isometric behaviour (i.e. slow shortening speed) around opti- mal muscle length enables the muscle to produce high forces more economically. During different forms of locomotion (e.g. walking and running), the elastic elements could accommodate the largest part of the MTU length changes, allowing the fascicles to work at a high force–length–veloc- ity potential (Fukunaga et al. 2001; Lichtwark et al. 2007; Bohm et al. 2019; Monte et al. 2020). Without tendons, the fascicle shortening velocity would be higher, increasing the cross-bridge turnover and the energy demand for muscle contraction (Woledge et al. 1985). Furthermore, since the force per cross-bridge decreases with increasing velocity (de Tombe and Ter Keurs 1990), a decrease in the muscle force potential would require an increased muscle activation to maintain the same level of force to support and accelerate the body’s centre of mass, thereby increasing the energy 2497 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 cost of locomotion (Fletcher and MacInthos 2017). Based on these theoretical considerations, if the whole muscle–ten- don unit length changes could be attributed to the tendon only, the muscle fibres would operate under isometric condi- tions, reducing the fascicle length changes (and, therefore, the mechanical work done by the muscles) and the level of muscle activation required for a given force (Fletcher and MacInthos 2017). These theoretical notions were sup- ported by recent studies of Bohm et al. (2019) and Monte et al. (2020), which revealed that the plantar flexor mus- cles operated quasi-isometrically at a high-force potential during running at different speeds. Furthermore, Bohm and co-workers found a negative significant correlation between the force–length–velocity potential of the soleus muscle and the energy cost of running at 10 km h−1, suggesting that the higher the force potential the lower the energy expended. Therefore, the ankle plantar flexors seem to play an impor- tant role in the mechanical and physiological demand of human running. The human ankle plantar flexors produce forces of up to 12 times body weight during running at increasing speed (Komi 1990) and are the main force producers amongst all the major lower limb muscle groups (Dorn et al. 2012). Due to their unique design (short muscle fibres connected to the heel via a long and compliant Achilles tendon), the ankle plantar flexors, have the capacity to generate high amounts of power with minimal energy expenditure. Thanks to their long tendon, the plantar flexors can store elastic energy up to about 60% of the MTU mechanical work during running (Monte et al. 2020; Farris and Sawicki 2012; Lai et al. 2014) and their contribution increases as a function of speed. This behaviour seems to be particularly relevant for determin- ing energy expenditure, mechanical work and, therefore, “apparent” efficiency; however, there are no studies that have investigated the role of plantar flexor tendons on “apparent” running efficiency. The aim of this study was to verify experimentally the theoretical link between human plantar flexor muscle–ten- don behaviour and “apparent” running efficiency. In particu- lar, we combined in vivo ultrasound, kinematic and kinetic measurements during running at different speeds and we calculated the relative contribution of GM muscle fascicles and Achilles tendon to the mechanical work done by the MTU to investigate the role of Achilles tendon behaviour on the mechanical power output at whole-body level (WTOT) and on the energy demands ( ̇VO2net and Cnet) during running at increasing speeds. Our main hypothesis was that the con- tribution of tendon work to the total work done by the MTU would increase with running speed and that this increase could, at least partially, explain the concurrent increase in “apparent” efficiency. Materials and methods Ethical approval All participants received written and oral information and instructions before the study and gave their written informed consent to the experimental procedure. The experimental protocol was approved by the Ethical Committee of Liv- erpool John Moores University (protocol number: 18/ SPS/028) and was performed in accordance with the Hel- sinki Declaration. Participants The experiments were performed on 15 male endurance athletes, as a part of a larger study (Monte et al. 2020). All participants (24 ± 2.4 years of age; 74 ± 2.8 kg of body mass; 1.77 ± 0.04 m of stature; 8.5 ± 2.2 years of training: 5 ± 1 workouts per week) received written and oral instructions before the study and gave their written informed consent to participate in the experimental procedures. The experi- mental protocol was approved by the Ethical Committee of Liverpool John Moores University (protocol number: 18/ SPS/028) and performed in accordance with the Helsinki Declaration. Experimental design During each running trial, the participants ran at steady-state speed using a self-selected cadence, step length and running technique. All participants used a forefoot running pattern. The trajectories of 50 reflective markers were recorded using 12 camera system (Vicon Vero 2.2, Oxford Metrics, United Kingdom), sampling at 250 Hz. The markers were placed at specific anatomical position on the subjects’ head, trunk, arms, pelvis, lower limbs and foots. This marker set was proposed by Lai et al. (2015) to investigate the ankle moment generation. Moreover, we added another 14 markers (five on the right shank/foot and one at the great trochanter and cheekbones, bilaterally) to measure the tendon lever arm during running as described by Rasske et al. (2017) and the internal work (see below) with the marker set proposed by Minetti et al. (1993). Ground reaction forces (GRFs) were recorded using an instrumented treadmill (M-GAIT, MOTEK) with two 3-axial (horizontal, vertical and mediolateral) force plates sampling at 1500 Hz (Lai et al. 2018). Resultant GRFs, centre of pres- sure and free moment vectors were measured and recorded by the treadmill’s software. A B-mode ultrasound apparatus (Telemed Echo Blaster 128) with a linear probe operated at a scanning depth and 2498 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 with of 6 cm (sample frequency 7 MHz) was used to record the GM fascicles at a sampling rate of 60 Hz. Ultrasound images were recorded from the right leg of each athlete dur- ing each running trial, with the probe placed in the sagittal plane at the mid-belly of the muscle. The position of the scanning probe was manipulated until the superficial and deep aponeuroses and the connective tissue that surrounds the muscle fascicles were clearly visible (Lichtwark et al. 2007; Cronin and Finni 2013). All signals were synchronized by a digital output gen- erated by the ultrasound scanner that triggered all instru- mentations (the Vicon cameras, ground reaction forces and ultrasound). A metabolic gas analysis was performed to measure oxy- gen consumption during each running trial ( ̇VO2) by means of a breath by breath metabolimeter (CORTEX Metalyzer 3B, CORTEX Biophysik, Germany). Six min of baseline data collection in a standing position was performed before these tests, the running trials were separated by 5 min of rest and data collected in the last minute of rest/exercise were averaged and used in further analyses. Net energy cost (Cnet) was calculated as ( ̇VO2net/v), by dividing net oxy- gen uptake ( ̇VO2net = ̇VO2 −  ̇VO2rest), in ml O2 kg−1 min−1, with the treadmill velocity (v, expressed in m min−1) and using an energy equivalent that takes into account the res- piratory exchange ratio (RER): ̇VO2net (4.94·RER + 16.04) J ml O2 −1 (Garby and Astrup 1987); Cnet is thus expressed in J kg−1 m−1. In previous studies quantifying “apparent” efficiency, net energy expenditure ( ̇VO2net instead of over- all ̇VO2) has invariably been utilized to calculate the cost of transport (e.g. Minetti et al. 1993, 2001, 2002; Saibene and Minetti 2003; Zamparo et al. 2002, 2016) because, in addition to the metabolic cost of locomotion, overall ̇VO2 also encompasses resting energy expenditure, which is not “utilized” to transport the body. Data analysis In the last minute of each running trial, kinematic, kinetic and ultrasound data were analysed during the stance phase of ten consecutive steps for each participant. This timing was chosen to coincide with the determination of oxygen uptake. Data of each instrumentation (except for the oxygen consumption data) were interpolated to 200 sample points. Kinetics and mechanical work Marker trajectories were filtered with a forward and reverse low-pass Butterworth filter (second order: cut-off 10 Hz), whereas GRF was filtered through a forward and reverse low pass, fourth-order Butterworth filter with a cut-off frequency of 30 Hz (consistent with the Nyquist theorem). Spectral analysis showed peaks of noise frequencies at 41, 47 and 100 Hz, which were speed and gait independent and conse- quently induced by the treadmill engine. Inverse kinematics was used to calculate the angular rota- tion for each body segment (Lai et al. 2015, 2018). The foot was modelled as a rigid segment and the ankle joint was represented as a universal joint with the centre of rotation at the midpoint between the medial and lateral malleoli mark- ers and was reconstructed relative to the shank line (Schache et al. 2011). A standard inverse dynamic approach was used to obtain ankle joint torque, while ankle joint power was calculated as the product of ankle joint moment and ankle joint angular velocity. To calculate the internal work, the body was considered to be composed of 11 body segments: head–trunk, thighs, shanks, feet, upper arms and forearms (Minetti et al. 1993). Based on the intrinsic characteristics of the limbs (mass of each segment and radius of gyration) determined according to Dempster inertial parameters (Winter 1979), and their 3D angular velocity and acceleration, the work necessary to rotate and accelerate the limbs with respect to body centre of mass (BCoM) (e.g. the internal work, WINT, J kg−1 m−1) was calculated (Cavagna and Kaneko 1977; Minetti 1988; Minetti et al. 1993; Pavei et al. 2017). The work done to raise and accelerate the body centre of mass with respect to the environment (WEXT, J kg−1 m−1) was calculated based on the summation of all increases in total mechanical energy (ET = EP + EK), where the time course of potential (EP) and kinetic (Ek = Ekx + Eky + Ekz) energy were calculated based on the BCoM trajectory. The BCoM position was calculated by a double integration of the GRF signal, according to Cavagna (1975), and using as integra- tion constant the treadmill speed (as described by Saibene and Minetti 2003). The sum of internal and external work represents the total mechanical work generated to move the body over a unit distance (WTOT, J kg−1 m−1). “Apparent” efficiency (AE) was then calculated from the ratio WTOT/Cnet (both expressed in J kg−1 m−1, see above). Total mechanical work (WTOT, as computed here) is, therefore, not the total work done by the muscles or by the MTUs but represents the mechanical work at whole-body level. Moreover, by means of this method, instead of the work of a force, the work done on the body is computed; this work, in turn, is calculated based on the work–energy principle, which states that the work done on an object is equal to the change in its (kinetic and potential) energy (e.g. Zatiorsky 2000). Muscle fascicle and series elastic element behaviour In vivo muscle fascicle length and pennation angle were measured from the ultrasound videos. Pennation angle was defined as the angle between the collagenous tissue and 2499 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 the deep aponeurosis (Lichtwark et al. 2007; Seynnes et al. 2015). A validated automatic tracking algorithm was used to quantify muscle fascicle length and pennation angle (Cronin and Finni 2013; Gillet et al. 2013). Each frame of the tracked muscle fascicle lengths and pennation angles was visually examined to check the algorithm’s accuracy. Whenever the muscle fascicle length or pennation angle was deemed inac- curate, the two points on the aponeuroses defining the mus- cle fascicles were manually repositioned. The instantaneous MTU length of GM in the stance phase was computed using instantaneous joint angles as proposed by Hawkins and Hull (1990); the instantaneous tendon length was calculated as the difference between the MTU length and the muscle belly length, taking into account the effect of pennation angle (Fukunaga et al. 2001). The behavior of the MTU, the fas- cicles and the Achilles tendon were investigated during the absorption phase (where net ankle joint power is negative) and the propulsive phase (where net ankle joint power is positive) during stance. The average MTU length, fascicle length and tendon length during the absorption and propul- sive phases are reported in Table 1 along with tendon strain (the maximum value during the stance phase) and tendon recoil (the maximum value during the propulsive phase). Fascicle and tendon length velocities were computed by dif- ferentiating the lengths of each component with respect to time in the stance phase (Lai et al. 2015, 2018). Muscle fascicle and tendon mechanical work The amounts of mechanical work done by the MTU, by the muscle fascicles and by the Achilles tendon were calculated by integrating the corresponding power curves over the entire stance phase (see Fig. 1). In turn, the power devel- oped by each component was obtained by multiplying the corresponding force and velocity values. Force production was determined as proposed by Farris and Sawicki (2012) whereas the velocity was calculated as the first derivative of the length changes. Briefly, tendon force was calculated as the net ankle torque divided by the tendon lever arm (esti- mated as suggested by Rasske et al. (2017). The force attrib- utable to GM was estimated by multiplying “overall” tendon force by the relative PCSA of this muscle which, according to the literature, amounts to ~ 16% of the PCSA of all the plantar flexors (Fukunaga et al. 1996). To estimate muscle fascicle force, tendon force was divided by the cosine of the pennation angle (e.g. Lichtwark and Wilson 2005). The positive work done by the MTU (WMTU) was cal- culated in the portion of stance where the MTU generates positive power (Fig. 1, upper panel). Positive muscle fibre work (Wfas) was calculated as the positive muscle fibre work done during the propulsion phase (Fig. 1, middle panel). From these data, positive tendon work (Wten) was finally calculated, which represents the mechanical energy that can be derived from tendon recoil during the propulsion phase (Fig. 1, lower panel). Statistics A one-way ANOVA for repeated measures was conducted to test for possible differences among running speeds for all the investigated variables. When significant main effects were found, a post hoc pairwise comparison using Fisher’s least significant difference was used to determine the effect of speed. To determine the relationships between Wten and Cnet, AE and WTOT, the Pearson’s correlation coefficient was used. Statistical analysis was performed with SPSS (v24.0). All data extracted for statistical analysis were normally dis- tributed (Shapiro–Wilk normality test, P > 0.05). Results Table 1 reports the average values of muscle–tendon unit length, fascicle length and tendon length during the absorp- tion and propulsive phases, as well as the values of tendon strain and recoil. All these parameters increased signifi- cantly as a function of speed (main effect: P < 0.001), apart from GM fascicle length, which decreased as a function of speed both during absorption and propulsion. Significant Table 1 Average muscle–tendon unit (MTU) length, fascicle length and tendon length during the absorption and propulsive phases of ground contact at 10, 13 and 16 km h−1 Tendon strain (the maximum value during the stance phase) and ten- don recoil (the maximum value during the propulsive phase) are also reported. Data are means ± SD and are expressed in cm Significant differences from 10  km  h−1 (*P < 0.05; **P < 0.01; ***P < 0.001); significant differences between 13 and 16  km  h−1 (#P < 0.05; ##P < 0.01; ###P < 0.001) 10 km h−1 13 km h−1 16 km h−1 MTU length  Absorption phase 47.41 ± 3.4 53.33 ± 3.9**# 58.87 ± 4.2***#  Propulsive phase 44.32 ± 3.2 42.88 ± 2.7*# 40.04 ± 2.8**# Fascicle length  Absorption phase 4.37 ± 1.01 4.28 ± 0.97*# 4.19 ± 0.83**#  Propulsive phase 3.99 ± 0.98 3.81 ± 0.98*# 3.62 ± 0.98**# Tendon length  Absorption phase 22.53 ± 2.2 23.78 ± 1.97*## 24.51 ± 1.89**##  Propulsive phase 21.12 ± 1.9 20.02 ± 2.01*# 19.48 ± 1.88**# Tendon strain  Absorption phase 1.10 ± 0.49 1.35 ± 0.52**## 1.62 ± 0.57***## Tendon recoil  Propulsive phase 0.89 ± 0.55 1.11 ± 0.48**## 1.35 ± 0.51***## 2500 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 differences were observed among speeds in all investigated muscle and tendon parameters. During the propulsive phase, the average ankle joint power was: 5.40 ± 0.88, 7.38 ± 0.96, 9.72 ± 0.91 W kg−1 at 10, 13 and 16 km h−1, respectively. Significant differences were observed among all running velocities (main effect: P < 0.001). In Fig. 1, the profile of negative (absorbed) and positive (generated) mechanical power of the MTU (upper panel), muscle fascicles (middle panel) and Achilles tendon (bot- tom panel) during the stance phase are reported in the panels on the left. The panels on the right depict the aver- age values of positive work during the propulsive phase. All variables increased as a function of speed (MTU: Fig. 1 Panels on the left: profile of mechanical power absorbed (negative) and generated (posi- tive) by the MTU (upper panel), muscle fascicle (middle panel) and Achilles tendon (lower panel) during the stance phase at the investigated running speeds (solid line: 10 km h−1; dotted line: 13 km h−1; dashed line: 16 km h−1). Note that the mechanical power absorbed by the Achilles tendon is always higher than that returned during its recoil. Panels on the right: positive mechanical work done by the MTU (upper panel), muscle fascicle (middle panel) and Achilles tendon (lower panel) during the stance phase at all the investigated run- ning speed. Positive work was calculated as the first integral of the positive mechanical power generated during the stance 2501 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 P < 0.001; fascicles: P < 0.05; tendon: P < 0.01) and sig- nificant differences were observed among running speeds in all the investigated parameters. The Achilles tendon contributed for the 65% of the MTU work. Table 2 reports mechanical and metabolic data at the three investigated running speeds. With running speed, WEXT decreased (main effect: P < 0.001), whereas ̇VO2net, AE, WINT and WTOT increased (main effect: P < 0.001, for all variables). The comparisons among speeds showed sig- nificant differences among running trials for each of these variables. No differences in Cnet were observed as a function of speed. Figure 2 shows the correlations between (positive) Achil- les tendon work and WTOT, Cnet and “apparent” efficiency (AE) at each running speed (10, 13 and 16 km h−1, from top to bottom). Significant correlations were observed for all the investigated parameters at all the investigated speeds. The subjects with the highest tendon work were those with the highest total mechanical work (upper panel, P < 0.01 in all cases) and with the lower energy cost (middle panel, P < 0.01 in all cases). Positive significant correlations were observed between tendon work (Wten) and AE at all the investigated speeds (lower panel, P < 0.001 in all cases): the subjects with the highest tendon work were those with the highest “apparent” efficiency. In Fig. 3, the mean values of AE are reported as a func- tion of the mean values of (positive) Achilles tendon work, at the three investigated speeds; this relationship is described by the following equation: AE = 1.56 Wten + 0.33. Discussion In this study, we investigated the role of elastic strain energy (e.g. the work produced by the Achilles tendon fascicles of the GM muscle tendon unit) on locomotion (“apparent”) efficiency at increasing running speeds Our results reveal that the work provided by the recoil of the Achilles ten- don at each speed is linked to: (1) a reduction in the energy cost of running, (2) an increase in the mechanical work at whole-body level and (3) an increase in the “apparent” effi- ciency. These novel in vivo experimental results support the notion of elastic energy reutilization impacting positively on the economy/efficiency of running. “Apparent” efficiency Although the contribution of the Achilles tendon to the economy/efficiency of running through the reutilization of elastic energy is a mechanism since long postulated, data reported in this paper constitute a novel observation for in vivo human running, allowing for a better interpretation of the changes in AE in different experimental conditions. In this study, AE was calculated based on values of (total) mechanical work (i.e. the sum of internal and external work). Although the concept of total mechanical work estimation has been debated (as also acknowledged for some aspects by the original authors, Willems et al. 1995), an increase in efficiency attributable to elastic energy reutilization was also observed in studies where mechanical work was calculated based on a different (joint power) approach. As an example, Farris and Sawicki (2011) calculated values of efficiency of about 0.45 during running at 11 km h−1 and Voigt et al. (1995) reported values of efficiency of about 0.65 during hopping (with a frequency of 2 Hz). This supports the idea that measuring “apparent” efficiency can help in under- standing whether mechanical work has been “recycled” via storage and release of elastic energy, thus indicating the presence of an energy saving mechanism (as suggested by Alexander 1991). Muscle and tendon contribution Regarding the underpinning mechanisms, a possible expla- nation for the increase in “apparent” efficiency with speed is that the plantar flexor muscles favour the use of tendon elastic strain energy over muscle fibre work (Lichtwark et al. 2007), and that this energy is enhanced when running speed advances towards maximum running velocity (Cavagna and Kaneko 1977). Our data are in line with these considerations Table 2 Mechanical and metabolic data (mean ± SD) during running at 10, 13 and 16 km⋅h−1 WEXT external mechanical work, WINT internal mechanical work, WTOT total mechanical work, ̇VO2net net oxygen uptake, Cnet energy cost of run- ning, AE “apparent” efficiency Significant difference from 10  km  h−1 (*P < 0.05; **P < 0.01; ***P < 0.001); significant difference between 13 and 16  km  h−1(#P < 0.05; ##P < 0.01; ###P < 0.001) WEXT (J kg−1 m−1) WINT (J kg−1 m−1) WTOT (J kg−1 m−1) ̇VO2net (ml kg−1 min−1) Cnet (J kg−1 m−1) AE 10 km h−1 1.60 ± 0.09 0.31 ± 0.05 1.91 ± 0.04 32.2 ± 5.4 3.98 ± 0.42 0.49 ± 0.03 13 km h−1 1.49 ± 0.08**# 0.67 ± 0.07***### 2.16 ± 0.03**## 42.9 ± 4.6***### 4.04 ± 0.38 0.53 ± 0.03**## 16 km h−1 1.33 ± 0.08***# 0.95 ± 0.08***### 2.28 ± 0.06***## 52.6 ± 04.2***### 4.08 ± 0.34 0.57 ± 0.05***## 2502 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 and are comparable to those of previous studies that have estimated the relative contribution of tendon elastic strain energy to the positive work done by the MTU for the ankle plantar flexors during running (Hof et al. 2002; Farris and Sawicki 2012; Lai et al. 2014). This mechanism may be a consequence of the plantar flexor muscle fibres remaining relatively isometric as run- ning speed increases, a behaviour that allows to generate large muscle forces and facilitates the storage and recovery of tendon elastic strain energy. This was recently verified by in vivo studies that analysed the muscle and tendon behav- iour of the plantar flexors during running at increasing speed (Lai et al. 2018; Werkhausen et al. 2019; Monte et al. 2020; Bohm et al. 2019). For instance, Monte et al. (2020) dem- onstrated that the gastrocnemius medialis muscle fascicles shorten during the entire stance phase, but that the series elastic components accommodate much of the displace- ment of the MTU, allowing the series elastic components to provide the larger amount of mechanical power of the MTU. This result suggests that fibres in distal limb muscles, such as the ankle plantar flexors, act like isometric struts to facilitate greater storage and recovery of tendon elastic strain energy at fast locomotion speeds (as indicated by Biewener and Roberts 2000). The strong relationship between tendon work and total mechanical work at whole-body level we observed (middle panel of Fig. 2) suggests that the elastic energy provided by the Achilles tendon recoil during the propulsive phase would affect the total mechanical work provided by the body. Indeed, as showed by Monte et al. (2020) with faster run- ning speed, the GM muscle fascicle operating range shifts towards smaller lengths (on the ascending limb of the F–L relationship) yet operating quasi-isometrically and at a high force potential (> 80% of the maximum isometric force) and this behaviour allows the muscle fascicles to reduce the Fig. 2 Correlations between (positive) tendon work and total mechan- ical work (at the whole-body level), net energy cost of running and “apparent” efficiency at the three investigated speeds (blue dots: 10 km h−1, red squares: 13 km h−1, green triangles: 16 km h−1). At each speed, the subjects with the higher tendon work are those with the larger WTOT, the lower Cnet and the larger AE. Upper panel: corre- lations between tendon work and total mechanical work at 10 km h−1 (WTOT = 3.34⋅Wten + 1.57, N = 15, R2 = 0.65, P < 0.01), 13  km  h−1 (WTOT = 2.82⋅Wten + 1.81, N = 15, R2 = 0.52, P < 0.05) and 16 km h−1 (WTOT = 11.40⋅Wten + 0.56, N = 15, R2 = 0.55, P < 0.05). Middle panel: correlations between tendon work and net energy cost of running at 10 km h−1 (Cnet = − 36.56⋅Wten + 7.68, N = 15, R2 = 0.72, P < 0.001), 13  km  h−1 (Cnet = −  49.43⋅Wten + 10.15, N = 15, R2 = 0.60, P < 0.01) and 16 km h−1 (Cnet = − 35.36⋅Wten + 9.49, N = 15, R2 = 52, P < 0.05). Lower panel: correlations between tendon work and “apparent” efficiency at 10  km  h−1 (AE = 5.61⋅Wten – 0.08, N = 15, R2 = 0.75, P < 0.001), 13  km  h−1 (AE = 7.43⋅Wten – 0.38, N = 15, R2 = 0.65, P < 0.01) and 16  km  h−1 (AE = 7.96⋅Wten – 0.65, N = 15, R2 = 54, P < 0.05) Fig. 3 Mean values of “apparent” efficiency as a function of the mean values of (positive) Achilles tendon work, at the three investi- gated speeds (blue dots: 10 km h−1, red dots: 13 km h−1, green dots: 16 km h−1); this relationship is described by the following equation: AE = 1.56 Wten + 0.33. The intercept with the “Y” axes (0.33) indi- cates the value of “apparent” efficiency that could be expected were the Achilles tendon not operating as energy saver 2503 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 amount of mechanical work performed. The series elastic components accommodate the largest part of the MTU dis- placement as they are stretched not only by the muscle force but also, and to a much larger degree, by the ground reaction forces. This allows the AT to perform the largest amount of mechanical work within the MTU. What is the benefit of a larger tendon work? If the ten- don makes a contribution to the whole MTU work during muscle contraction, the contribution of the active con- tractile components would be reduced, thus reducing the energy expended during the contraction (Roberts and Azizi 2011). In particular, as also reported in the “Introduction”, if the whole MTU length changes were attributable to the tendon alone, the muscle fibres would operate under “pure” isometric conditions and at a high force potential, requiring the lowest level of muscle activation (Fletcher and MacIntosh 2017). This phenomenon was recently veri- fied by Bohm et al. (2019), who observed that the subjects with the higher soleus muscle force potential were those with the lower energy cost of running, suggesting that, the lower the shortening velocity of the soleus muscle the lower the energy demand during running. Furthermore, these authors suggested that the main mechanism for the underlying reduction of the fascicle shortening velocity during the stance phase was a greater tendon gearing (i.e. larger tendon displacement with respect to the muscle fascicles). However, the utilization of tendon elasticity does not come entirely “free-of-charge”. Tendons operate in series with muscles and can only act as useful springs when mus- cles generate force. Force generation by muscles requires metabolic energy, and thus there is a cost to operate ten- don springs (e.g. Fletcher and MacIntosh 2017; Roberts and Azizi 2011; Roberts 2002). It has been proposed that the net metabolic benefit of tendon elasticity in running is best understood in the context of two properties of skel- etal muscle (Roberts 2002; Roberts and Scales 2002). The first is the ‘Fenn effect’, which states that active muscles use more energy when performing work than when generat- ing force isometrically (Fenn 1924). Thus, to the extent that tendons allow muscles to generate force without doing work (or while doing less work), they reduce the rate of energy consumption in the muscle. The second mechanism is the influence that tendon mechanics can have on the recruited muscle volume during running. Owing to the F–L and F–V properties of muscles, force can be produced with fewer active muscle fibres if the muscle operates at low or zero (i.e. isometric) shortening velocity (Fletcher and MacIntosh 2017). The fascicles length changes observed in this study are probably not sizeable enough to increase the metabolic cost of muscle contraction. Indeed, in vitro evidence showed that although muscle force is reduced at shorter sarcomere lengths and a greater muscle activation is needed to reach a given amount of force, the ATPase rate seems not to dif- fer from the rate at optimal length at least until 0.75 of L0; this suggests that if force potential is > 0.75 then muscle metabolic requirement is not affected, whereas when mus- cle length is < 0.75, a higher cost of contraction should be expected (Stephenson et al. 1989; Joumaa et al. 2017). As shown by Monte et al. (2020) and Bohm et al. (2019), the GM and soleus muscle fascicle do not operate below 75% of their force potential during running, so the effects of fascicle length changes on muscle energy demands could be consid- ered to be negligible. Methodological limitations and considerations In our study, the Achilles tendon length was not directly measured; instead, we used a geometric model to calculate the deformation of the entire Achilles tendon–aponeuroses complex from ankle and knee joint kinematics. However, this approach might overestimate the contribution of the mechanical work done by the tendon only, as shown by Zelik and Franz (2017). In addition, some recent studies (e.g. Kes- sler et al. 2020) have reported that using a rigid-body foot model (as done in this study) could lead to an overestimation of ankle joint power, thus affecting Achilles tendon mechani- cal work estimates. One other limitation is the estimation of GM forces based on reported values of relative PCSA, assuming consistent force contribution at all running speeds and a negligible inter-muscular force transmission between the individual plantar flexor muscles. The former assump- tion has often been used before (e.g. Kurokawa et al. 2001; Fukunaga et al. 1996; Farris and Sawicki 2012) to distrib- ute forces between synergist muscles, but the validity of the outcome forces needs to be confirmed, especially during dynamic conditions. In support of the latter assumption are the findings of Tijs et al. (2015), who have shown that non- myotendinous forces are likely to have a minimal effect on the overall function of muscles. Besides the Achilles tendon stretch–recoil, there are sev- eral other parameters that could affect energy expenditure as running speed increases and have not been considered in our analysis. For instance, higher activation of other agonist and antagonist muscles in the lower limbs, torso and upper limbs would contribute to the increase in metabolic energy expenditure with increasing speed (Arellano and Kram 2014). Also, elastic mechanisms other than Achilles ten- don recoil, such as the arch of the foot, could provide extra energy savings with increasing speed and make the running task less energy demanding. Furthermore, elastic energy could be stored also in the transversal plane of the MTU; indeed, biaxial loading of aponeuroses allows for variation in tendon stiffness and energy storage in a variety of locomo- tor behaviours (such as running, jumping and landing; e.g. Arellano et al. 2019). 2504 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 Conclusion In conclusion, a larger mechanical work provided by the tendons is expected to reduce the metabolic demands of run- ning and to increase locomotion efficiency. Our data support this notion. The relationship between AE and tendon work supports previous suggestions that a value of “apparent” effi- ciency close to muscle efficiency values (0.25–0.30) should be expected when no elastic energy can be stored in the tendon; the intercept of the relationship between AE and tendon work (0.33) indeed indicates the value that could be expected in the absence of Achilles tendon stretching–recoil- ing behaviour. As suggested by Alexander (1991), measur- ing “apparent” efficiency can help in understanding whether mechanical work is “recycled” via storage and release of elastic energy. Author contributions AM, CM, VB and PZ contributed to conception and design of the study. AM recorded and analyzed the data under the supervision of CM, VB and PZ. AM, CM, VB and PZ contrib- uted to the interpretation of the data, wrote and critically revised the manuscript. All authors approved the final version of the manuscript and agreed to be accountable for all aspects of the work. All persons included as an author qualify for authorship, and all those who qualify for authorship are listed. Funding The authors received no funding for this work. Open access funding provided by Università degli Studi di Verona within the CRUI- CARE Agreement. Compliance with ethical standards Conflict of interest The authors declare there are no competing inter- ests. Availability of data and material All relevant data are reported in the manuscript. Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. References Alexander RM (1991) Energy-saving mechanisms in walking and run- ning. J Exp Biol 160:55–69 Arellano CJ, Kram R (2014) Partitioning the metabolic cost of human running: a task-by-task approach. Integ Comp Biol 54:1084–1098 Arellano CJ, Konow N, Gidmark NJ, Roberts TJ (2019) Evidence of a tunable biological spring: elastic energy storage in aponeuroses varies with transverse strain in vivo. Proc Biol Sci 286:20182764 Biewener AA, Roberts TJ (2000) Muscle and tendon contributions to force, work, and elastic energy savings: a comparative perspective. Exerc Sport Sci Rev 28:99–107 Bohm S, Mersmann F, Santuz A, Arampatzis A (2019) The force– length–velocity potential of the human soleus muscle is related to the energetic cost of running. Proc Biol Sci 286(1917):20192560 Cavagna GA (1975) Force platforms as ergometers. J Appl Physiol 39:174–179 Cavagna GA, Kaneko M (1977) Mechanical work and efficiency in level walking and running. J Physiol (Lond) 268:467–481 Cronin NJ, Finni T (2013) Treadmill versus overground and barefoot versus shod comparisons of triceps surae fascicle behaviour in human walking and running. Gait Posture 38:528–533 de Tombe PP, Ter Keurs HE (1990) Force and velocity of sarcomere shortening in trabeculae from rat heart. Effects of temperature. Circ Res 66:1239. https ://doi.org/10.1161/01.RES.66.5.1239 Dorn TW, Schache AG, Pandy MG (2012) Muscular strategy shift in human running: dependence of running speed on hip and ankle muscle performance. J Exp Biol 215:1944–1956 Ettema GJC (2001) Muscle efficiency: the controversial role of elastic- ity and mechanical energy conversion in stretch-shortening cycles. Eur J Appl Physiol 85:457–465 Farris DJ, Sawicki GS (2011) The mechanical and energetics of human walking and running: a joint level perspective. J R Soc Interface 9:110–118 Farris DJ, Sawicki GS (2012) Human medial gastrocnemius force- velocity behaviour shifts with locomotion speed and gait. PNAS 109:977–982 Fenn WO (1924) The relation between the work performed and the energy liberated in muscular contraction. J Physiol 58:373–395 Fletcher JR, MacIntosh BR (2017) Running economy from a muscle energetics perspective. Front Physiol 8:433 Fukunaga T, Roy RR, Shellock FG, Hodgson JA, Edgerton VR (1996) Specific tension of human plantar flexors and dorsiflexors. J Appl Physiol 80:158–165 Fukunaga T, Kubo K, Kawakami Y, Fukashiro S, Kanehisa H, Maga- naris C (2001) In vivo behaviour of human muscle tendon during walking. Proc Biol Sci 268:229–233 Garby L, Astrup A (1987) The relationship between the respiratory quotient and the energy equivalent of oxygen during simultaneous glucose and lipid oxidation and lipogenesis. Acta Physiol Scand 129:443–447 Gillett JG, Barrett RS, Lichtwark GA (2013) Reliability and accuracy of an automated tracking algorithm to measure controlled pas- sive and active muscle fascicle length changes from ultrasound. Comput Methods Biomech Biomed Engin 16:678–687 Gordon AM, Huxley AF, Julian FJ (1966) The variation in isomet- ric tension with sarcomere length in vertebrate muscle fibres. J Physiol 184:170–192 Hawkins D, Hull ML (1990) A method for determining lower extrem- ity muscle-tendon lengths during flexion/extension movements. J Biomech 23:487–494 Hof AL, Van Zandwijk JP, Bobbert MF (2002) Mechanics of human triceps surae muscle in walking, running and jumping. Acta Phys- iol Scand 174:17–30 Holt NC, Danos N, Roberts TJ, Azizi E (2016) Stuck in gear: age- related loss of variable gearing in skeletal muscle. J Exp Biol 219:998–1003 Joumaa V, Fitzowich A, Herzog W (2017) Energy cost of isometric force production after active shortening in skinned muscle fibres. J Exp Biol 220:1509–1515 Kessler SE, Lichtwark GA, Welte LKM, Rainbow MJ, Kelly LA (2020) Regulation of foot and ankle quasi-stiffness during human 2505 European Journal of Applied Physiology (2020) 120:2495–2505 1 3 hopping across a range of frequencies. J Biomech. https ://doi. org/10.1016/j.jbiom ech.2020.10985 3 Komi PV (1990) Relevance of in vivo force measurements to human biomechanics. J Biomech 23(Suppl. 1):23–34 Kurokawa S, Fukunaga T, Fukashiro S (2001) Behavior of fascicles and tendinous structures of human gastrocnemius during vertical jumping. J Appl Physiol 90:1349–1358 Lai A, Schache AG, Lin YC, Pandy MG (2014) Tendon elastic strain energy in the human ankle plantar-flexors and its role with increased running speed. J Exp Biol 217:3159–3168 Lai A, Lichtwark GA, Schache AG, Lin YC, Brown NAT, Pandy MG (2015) In vivo behavior of the human soleus muscle with increas- ing walking and running speeds. J Appl Physiol 118:1266–1275 Lai A, Lichtwark GA, Schache AG, Pandy MG (2018) Differences in in vivo muscle fascicle and tendinous tissue behaviour between the ankle plantar flexors during running. Scand J Med Sci Sports 28(7):1828–1836 Lejeune TM, Willems PA, Heglund NC (1988) Mechanics and ener- getics of human locomotion on sand. J Exp Biol 201:2071–2080 Lichtwark GA, Wilson AM (2005) In vivo mechanical properties of the human Achilles tendon during one-legged hopping. J Exp Biol 208:4715–4725 Lichtwark GA, Bougoulias K, Wilson AM (2007) Muscle fascicle and series elastic element length changes along the length of the human gastrocnemius during walking and running. J Biomech 40:157–164 Minetti AE (1998) A model equation for the prediction of mechanical internal work of terrestrial locomotion. J Biomech 31:463–468 Minetti AE, Ardigò LP, Saibene F (1993) Mechanical determinants of gradient walking energetics in man. J Physiol 472:725–735 (Erratum in: J Physiol 475:548) Minetti AE, Pinkerton J, Zamparo P (2001) From bipedalism to bicy- clism: evolution in energetics and biomechanics of historic bicy- cles. Proc Biol Sci 268:1351–1360 Minetti AE, Moia C, Roi GS, Susta D, Ferretti G (2002) Energy cost of walking and running at extreme uphill and downhill slopes. J Appl Physiol 93:1039–1046 Monte A, Blatzopoulos V, Maganaris CN, Zamparo P (2020) Gas- trocnemius Medialis and Vastus Lateralis in vivo muscle-tendon behaviour during running at increasing speeds. Scand J Med Sci Sports. https ://doi.org/10.1111/sms.13662 Pavei G, Seminati E, Cazzola D, Minetti AE (2017) On the estimation accuracy of the 3D body centre of mass trajectory during human locomotion: inverse vs. forward dynamics. Front Physiol 8:129 Ramsey RW, Street SF (1940) The isometric length-tension diagram of isolated skeletal muscle fibers of the frog. J Cell Comp Physiol 15:11–34 Rasske K, Thelen DG, Franz JR (2017) Variation in the human Achilles tendon moment arm during walking. Comput Methods Biomech Biomed Engin 20(2):201–205 Roberts TJ (2002) The integrated function of muscles and tendons dur- ing locomotion. Comp Biochem Physiol 133:1087–1099 Roberts TJ, Azizi E (2011) Flexible mechanisms: the diverse roles of biological springs in vertebrate movement. J Exp Biol 214:353–361 Roberts TJ, Scales JA (2002) Mechanical power output during running accelerations in wild turkeys. J Exp Biol 205:1485–1494 Saibene F, Minetti AE (2003) Biomechanical and physiological aspects of legged locomotion in humans. Eur J Appl Physiol 88:297–316 Schache AG, Blanch PD, Dorn TW, Brown NA, Rosemond D, Pandy MG (2011) Effect of running speed on lower limb joint kinetics. Med Sci Sports Exerc 43:1260–1271 Seynnes OR, Bojsen-Moller J, Albracht K, Arndt A, Cronin NJ, Finni T, Magnusson SP (2015) Ultrasound based testing of ten- don mechanical properties: a critical evaluation. J Appl Physiol 118:133–141 Stephenson DG, Stewart AW, Wilson GJ (1989) Dissociation of force from myofibrillar MgATPase and stiffness at short sarcomere lengths in rat and toad skeletal muscle. J Physiol 410:351–366 Tijs C, van Dieën JH, Maas H (2015) No functionally relevant mechani- cal effects of epimuscular myofascial connections between rat ankle plantar flexors. J Exp Biol 218:2935–2941 Voigt M, Bojsen-Moller F, Simonsen EB, Dyhre-Poulsen P (1995) The influence of tendon Youngs modulus, dimensions and instantane- ous moment arms on the efficiency of human moment. J Biomech 28:281–291 Werkhausen A, Cronin NJ, Albracht K, Bojsen-Møller J, Seynnes OR (2019) Distinct muscle-tendon interaction during running at dif- ferent speeds and in different loading conditions. J Appl Physiol 127(1):246–253 Willems PA, Cavagna GA, Heglund NC (1995) External, internal and total work in human locomotion. J Exp Biol 198:379–393 Williams KR, Cavanagh PR (1987) Relationship between distance running mechanics, running economy, and performance. J Appl Physiol 63:1236–1245 Winter D (1979) A new definition of mechanical work done in human movement. J Appl Physiol 64:79–83 Woledge RC, Curtin NA, Homsher E (1985) Energetic aspects of mus- cle contraction. Academic Press, London Zamparo P, Pendergast DR, Termin B, Minetti AE (2002) How fins affect the economy and efficiency of human swimming. J Expl Biol 205:2665–2676 Zamparo P, Pavei G, Nardello F, Bartolini D, Monte A, Minetti AE (2016) Mechanical work and efficiency of 5+5 m shuttle running. Eur J Appl Physiol 116:1911–1919 Zamparo P, Pavei G, Monte A, Nardello F, Otsu T, Numazu N, Fujii N, Minetti AE (2019) Mechanical work in shuttle running as a func- tion of speed and distance: implications for power and efficiency. Hum Mov Sci 66:487–496 Zatsiorsky VM (2000) Kinetics of human motion. Leeds, Champaign Zelik KE, Franz JR (2017) It’s positive to be negative: Achilles tendon work loops during human locomotion. PLoS ONE 12:e0179976 Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

The influence of Achilles tendon mechanical behaviour on "apparent" efficiency during running at different speeds.

08-25-2020

Monte, Andrea,Maganaris, Constantinos,Baltzopoulos, Vasilios,Zamparo, Paola

eng

PMC8952301




 Citation: Benjamin, D.; Odof, S.; Abbès, B.; Fourchet, F.; Christiaen, B.; Taïar, R. Shock Response Spectrum Analysis of Fatigued Runners. Sensors 2022, 22, 2350. https:// doi.org/10.3390/s22062350 Academic Editor: Dragan Indjin Received: 18 February 2022 Accepted: 16 March 2022 Published: 18 March 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). sensors Article Shock Response Spectrum Analysis of Fatigued Runners Daniel Benjamin 1,2 , Serge Odof 3, Boussad Abbès 2 , François Fourchet 4,5,6, Benoit Christiaen 1 and Redha Taïar 2,* 1 Podiatry Medicine Department, Centre Luxembourg, 75005 Paris, France; d.benjamin@centre-luxembourg.com (D.B.); b.christiaen@centre-luxembourg.com (B.C.) 2 MATériaux et Ingénierie Mécanique (MATIM), Université de Reims Champagne Ardenne, 51100 Reims, France; boussad.abbes@univ-reims.fr 3 École Nationale Supérieure D’ingénieurs de Reims (ESIREIMS), Université de Reims Champagne Ardenne, 51100 Reims, France; serge.odof@univ-reims.fr 4 Physiotherapy Department, Hospital La Tour, 1217 Meyrin, Switzerland; francois.fourchet@gmail.com 5 French Society of Sports Physical Therapist (SFMKS Lab), 93380 Pierrefite sur Seine, France 6 Inter-University Laboratory of Human Movement Biology (LIBM), Savoie Mont-Blanc University, 73000 Chambery, France * Correspondence: redha.taiar@univ-reims.fr Abstract: The purpose of this study was to determine the effect of fatigue on impact shock wave attenuation and assess how human biomechanics relate to shock attenuation during running. In this paper, we propose a new methodology for the analysis of shock events occurring during the proposed experimental procedure. Our approach is based on the Shock Response Spectrum (SRS), which is a frequency-based function that is used to indicate the magnitude of vibration due to a shock or a transient event. Five high level CrossFit athletes who ran at least three times per week and who were free from musculoskeletal injury volunteered to take part in this study. Two Micromachined Microelectromechanical Systems (MEMS) accelerometers (RunScribe®, San Francisco, CA, USA) were used for this experiment. The two RunScribe pods were mounted on top of the foot in the shoelaces. All five athletes performed three maximum intensity runs: the 1st run was performed after a brief warmup with no prior exercise, then the 2nd and the 3rd run were performed in a fatigued state. Prior to the 2nd and the 3rd run, the athletes were asked to perform at maximum intensity for two minutes on an Assault AirBike to tire them. For all five athletes, there was a direct correlation between fatigue and an increase in the aggressiveness of the SRS. We noticed that for all five athletes for the 3rd run the average SRS peaks were significantly higher than for the 1st run and 2nd run (p < 0.01) at the same natural frequency of the athlete. This confirms our hypothesis that fatigue causes a decrease in the shock attenuation capacity of the musculoskeletal system thus potentially involving a higher risk of overuse injury. Keywords: Shock Response Spectrum; fatigue; injuries; gait analysis; Micromachined Microelec- tromechanical Systems (MEMS) accelerometer 1. Introduction Running is the exercise of choice for millions of people all over the world and across the age spectrum. One of the main reasons for its popularity stems from its simplicity. However, running also carries the risk of increased musculoskeletal injuries and there is a need to understand the etiology of injury in order to efficiently prevent it [1]. One of the important functions of the human musculoskeletal system is to attenuate and dissipate shock waves initiated with foot ground contact [2]. Those shock waves are initiated by most types of motion, such as walking and running. The demarcation between walking and running occurs when periods of double support during the stance phase of the gait cycle (both feet are simultaneously in contact with the ground) give way to two periods of double float at the beginning and the end of the swing phase of gait (neither foot is Sensors 2022, 22, 2350. https://doi.org/10.3390/s22062350 https://www.mdpi.com/journal/sensors Sensors 2022, 22, 2350 2 of 11 touching the ground) [3]. Generally, as speed increases further, initial contact changes from being on the hindfoot to the forefoot. Running involves repeated single-leg impacts between the foot and the surface. Such impacts are characterized by a transient peak in the ground reaction force (impact force), rapid deceleration of the lower extremity (impact shock), and the initiation of a wave of acceleration and deceleration (impact shock wave) that is propagated through the body [4]. The impact shock wave experienced by the body due to landings must be attenuated by several structures and mechanisms in the body including bone, synovial fluids, carti- lage, soft tissues, joint kinematics and muscular activity. Passively, shock attenuation is achieved by soft tissues and bone. Actively, shock attenuation is achieved through eccentric muscle action [5]. This active mechanism is thought to be far more significant than the passive mechanism in attenuating shock. Since muscles are thought to play a primary role in energy and shock absorption during landing, it has been hypothesized that reduced muscular function, through fatigue, decreases the shock absorbing capacity of the body and subsequently can lead to an increased chance of injury [6]. Fatigue has been defined as any reduction in the force generating capacity of the total neuromuscular system regardless of the force required in any given situation [7]. The loads produced by repeated impacts have been linked to degenerative joint diseases and athletic overuse injuries including, for example, stress fractures, shin splints, osteoarthritis and lower back pain. Although the exact mechanisms of impact related injury are relatively unknown and controversial evidence linking impact, fatigue and injuries are well documented [8–11]. In this paper, we propose a new methodology for the analysis of shock events occurring during the proposed experimental procedure. Our approach is based on the Shock Response Spectrum (SRS) [12], which is a frequency-based function that is used to indicate the magnitude of vibration due to a shock or a transient event [13]. The main aim is to analyze the ability of the human musculoskeletal system to attenuate the mechanical stresses resulting from the fatigue effect by Shock Responses Spectrum (SRS) of the foot strike– generated shock waves during running. Most previous studies focused on shocks/impacts, ground force reaction, or spectral or vertical impact load rate. Using SRS as a measurement in running gait analysis has never been studied as of today. This innovative approach could pave the way to a whole new way of assessing a runner’s gait pattern using smart connected shoes. The purpose of this study was to determine the effect of fatigue on impact shock wave attenuation and assess how human biomechanics relate to shock attenuation during running. It was hypothesized that fatigue would cause a decrease in the shock attenua- tion capacity of the musculoskeletal system, thus potentially involving a higher risk of overuse injury. 2. Materials and Methods 2.1. Procedures Five high level CrossFit athletes (four males, one female) who ran at least three times per week and who were free from musculoskeletal injury volunteered to take part in this study. The athletes had a mean age of 26.4 (±3.9) years, stature 182.3 (±5.7) cm, and body mass 81.7 (±8.5) kg, respectively. The athletes usually performed 10 km to 15 km runs twice a week and one sprint interval training of various lengths and intensities. The study was conducted in accordance with the Helsinki Declaration on human experimentation stated in compliance with the 1964 Helsinki Declaration and its later amendments. Every participant provided written consent after information was given on the aim, protocol, and methodology of the study. The original study was approved by the Medical and Ethical Board of the Centre Luxembourg (protocol code LUX_2021_0308_CLAB and date of approval of 3 August 2021). Two Micromachined Microelectromechanical Systems (MEMS) accelerometers (RunScribe®) were used for this experiment. The two RunScribe pods were mounted on top of the foot in the shoelaces (Figure 1). Sensors 2022, 22, 2350 3 of 11 Sensors 2022, 22, x FOR PEER REVIEW 3 of 11 (MEMS) accelerometers (RunScribe®) were used for this experiment. The two RunScribe pods were mounted on top of the foot in the shoelaces (Figure 1). Figure 1. MEMS accelerometers placement. The RunScribe pods encompass 9-Axis Motion Tracking which combines a 3-axis gy- roscope, 3-axis accelerometer, and 3-axis compass in the same device together with an onboard Digital Motion Processor. This enables us also to measure at a 500 Hz sampling rate: Efficiency (Stride Rate, Contact Time, Flight Ratio), Motion (Footstrike Type, Prona- tion, Pronation Velocity), Shock (Impact Gs, Braking Gs), Symmetry and Power. After a warmup, participants were asked to perform a first 800 m run at maximum intensity. Right after the first run, they jumped on an Assault AirBike (Rogue, Columbus, OH, USA) (Figure 2) where they were asked to perform at maximum intensity for 2 min (the power had to stay above 400 Watts for the 2 min). They dismounted the Assault Air- Bike and they were then asked again to perform a second 800 m run at maximum inten- sity. The same protocol was then repeated with another 2 min on the Assault AirBike then a third run at maximum intensity. The RunScribe pods were turned off during the Assault AirBike sessions and were only recording the three 800 m run intervals. Figure 2. Assault AirBike. The Assault AirBike, also known as “the Devil’s tricycle” was used to induce fatigue because they procure a unique and extremely challenging effort. It is considered among the crossfit community as the most dreaded but most effective tool for HIIT (High Inten- sity Interval Training) and metabolic conditioning. 2.2. Shock Response Spectrum Calculation Spectral analysis is commonly used to study the structure of composite waveforms, such as the impact shock waves. The primary tool of spectral analysis is the Fast Fourier Transformation (FFT) that enables us to determine the runner’s natural frequency [14] which corresponds to the peak of the Power Spectral Density (Figure 3): Figure 1. MEMS accelerometers placement. The RunScribe pods encompass 9-Axis Motion Tracking which combines a 3-axis gyroscope, 3-axis accelerometer, and 3-axis compass in the same device together with an onboard Digital Motion Processor. This enables us also to measure at a 500 Hz sampling rate: Efficiency (Stride Rate, Contact Time, Flight Ratio), Motion (Footstrike Type, Pronation, Pronation Velocity), Shock (Impact Gs, Braking Gs), Symmetry and Power. After a warmup, participants were asked to perform a first 800 m run at maximum intensity. Right after the first run, they jumped on an Assault AirBike (Rogue, Columbus, OH, USA) (Figure 2) where they were asked to perform at maximum intensity for 2 min (the power had to stay above 400 Watts for the 2 min). They dismounted the Assault AirBike and they were then asked again to perform a second 800 m run at maximum intensity. The same protocol was then repeated with another 2 min on the Assault AirBike then a third run at maximum intensity. The RunScribe pods were turned off during the Assault AirBike sessions and were only recording the three 800 m run intervals. Sensors 2022, 22, x FOR PEER REVIEW 3 of 12 (MEMS) accelerometers (RunScribe®) were used for this experiment. The two RunScribe pods were mounted on top of the foot in the shoelaces (Figure 1). Figure 1. MEMS accelerometers placement. The RunScribe pods encompass 9-Axis Motion Tracking which combines a 3-axis gy- roscope, 3-axis accelerometer, and 3-axis compass in the same device together with an onboard Digital Motion Processor. This enables us also to measure at a 500 Hz sampling rate: Efficiency (Stride Rate, Contact Time, Flight Ratio), Motion (Footstrike Type, Prona- tion, Pronation Velocity), Shock (Impact Gs, Braking Gs), Symmetry and Power. After a warmup, participants were asked to perform a first 800 m run at maximum intensity. Right after the first run, they jumped on an Assault AirBike (Rogue, Columbus, OH, USA) (Figure 2) where they were asked to perform at maximum intensity for 2 min (the power had to stay above 400 Watts for the 2 min). They dismounted the Assault Air- Bike and they were then asked again to perform a second 800 m run at maximum inten- sity. The same protocol was then repeated with another 2 min on the Assault AirBike then a third run at maximum intensity. The RunScribe pods were turned off during the Assault AirBike sessions and were only recording the three 800 m run intervals. Figure 2. Assault AirBike. The Assault AirBike, also known as “the Devil’s tricycle” was used to induce fatigue because they procure a unique and extremely challenging effort. It is considered among the crossfit community as the most dreaded but most effective tool for HIIT (High Intensity Interval Training) and metabolic conditioning. 2.2. Shock Response Spectrum Calculation Spectral analysis is commonly used to study the structure of composite waveforms, such as the impact shock waves. The primary tool of spectral analysis is the Fast Fourier Sensors 2022, 22, 2350 4 of 11 Transformation (FFT) that enables us to determine the runner’s natural frequency [14] which corresponds to the peak of the Power Spectral Density (Figure 3): PSD = 1 N  Z +∞ −∞ a(t)e−j2π f tdt  2 (1) where N is the number of points of the recording, a(t) is the acceleration modulus, f is the frequency and t is the time. Sensors 2022, 22, x FOR PEER REVIEW 4 of 11 𝑃𝑆𝐷 ൌ 1 𝑁 ቤන 𝑎ሺ𝑡ሻ𝑒ି௝ଶగ௙௧ ାஶ ିஶ 𝑑𝑡ቤ ଶ (1) where 𝑁 is the number of points of the recording, 𝑎ሺ𝑡ሻ is the acceleration modulus, 𝑓 is the frequency and 𝑡 is the time. Figure 3. SRS vs. PSD running analysis. Power Spectral Density (PSD) provides a convenient method for separating different frequency components in the impact shock wave, such as acceleration moments due to impact shock [15]. In this paper, we propose a new methodology for the analysis of shock events occur- ring during the proposed experimental procedure. Our approach is based on the Shock Response Spectrum (SRS), which is a frequency-based function that is used to indicate the magnitude of vibration due to a shock or a transient event. The following procedure, con- sisting of several steps, is adopted in the present study: • Step 1: The acceleration modulus 𝑎ሺ𝑡ሻ is extracted from the recording. Figure 4 illustrates the acceleration modulus results for one CrossFit athlete. Figure 4. Acceleration modulus 𝑎ሺ𝑡ሻ for one CrossFit athlete. • Step 2: The power spectral density (PSD) given in Equation (1) is then calculated using a Fast Fourier Transform (FFT). This calculation allows us to determine the fundamental fre- quency of the runner 𝑓଴ corresponding to the position of the largest peak of the PSD. The inverse of this frequency gives the time period of the runner’s step as: 𝑇 ൌ 1/𝑓଴. The pro- posed algorithm extracts automatically the “first” step from the entire signal, and thus defines the “pattern” of the runner as shown in Figure 5. Figure 3. SRS vs. PSD running analysis. Power Spectral Density (PSD) provides a convenient method for separating different frequency components in the impact shock wave, such as acceleration moments due to impact shock [15]. In this paper, we propose a new methodology for the analysis of shock events occurring during the proposed experimental procedure. Our approach is based on the Shock Response Spectrum (SRS), which is a frequency-based function that is used to indicate the magnitude of vibration due to a shock or a transient event. The following procedure, consisting of several steps, is adopted in the present study: • Step 1: The acceleration modulus a(t) is extracted from the recording. Figure 4 illustrates the acceleration modulus results for one CrossFit athlete. Sensors 2022, 22, x FOR PEER REVIEW 4 of 11 𝑃𝑆𝐷 ൌ 1 𝑁 ቤන 𝑎ሺ𝑡ሻ𝑒ି௝ଶగ௙௧ ାஶ ିஶ 𝑑𝑡ቤ ଶ (1) where 𝑁 is the number of points of the recording, 𝑎ሺ𝑡ሻ is the acceleration modulus, 𝑓 is the frequency and 𝑡 is the time. Figure 3. SRS vs. PSD running analysis. Power Spectral Density (PSD) provides a convenient method for separating different frequency components in the impact shock wave, such as acceleration moments due to impact shock [15]. In this paper, we propose a new methodology for the analysis of shock events occur- ring during the proposed experimental procedure. Our approach is based on the Shock Response Spectrum (SRS), which is a frequency-based function that is used to indicate the magnitude of vibration due to a shock or a transient event. The following procedure, con- sisting of several steps, is adopted in the present study: • Step 1: The acceleration modulus 𝑎ሺ𝑡ሻ is extracted from the recording. Figure 4 illustrates the acceleration modulus results for one CrossFit athlete. Figure 4. Acceleration modulus 𝑎ሺ𝑡ሻ for one CrossFit athlete. • Step 2: The power spectral density (PSD) given in Equation (1) is then calculated using a Fast Fourier Transform (FFT). This calculation allows us to determine the fundamental fre- quency of the runner 𝑓଴ corresponding to the position of the largest peak of the PSD. The inverse of this frequency gives the time period of the runner’s step as: 𝑇 ൌ 1/𝑓଴. The pro- posed algorithm extracts automatically the “first” step from the entire signal, and thus defines the “pattern” of the runner as shown in Figure 5. Figure 4. Acceleration modulus a(t) for one CrossFit athlete. • Step 2: The power spectral density (PSD) given in Equation (1) is then calculated using a Fast Fourier Transform (FFT). This calculation allows us to determine the fundamental frequency of the runner f0 corresponding to the position of the largest peak of the PSD. The inverse of this frequency gives the time period of the runner’s step as: T = 1/ f0. The proposed algorithm extracts automatically the “first” step from the entire signal, and thus defines the “pattern” of the runner as shown in Figure 5. Sensors 2022, 22, 2350 5 of 11 Sensors 2022, 22, x FOR PEER REVIEW 5 of 11 Figure 5. Pattern of one CrossFit athlete. • Step 3: We then carry out the cross-correlation 𝐶𝐶ሺ𝜏ሻ between the runner’s pattern and the recording’s duration 𝑎ሺ𝑡ሻ: 𝐶𝐶ሺ𝜏ሻ ൌ න 𝑎ሺ𝑡ሻ𝑝𝑎𝑡𝑡𝑒𝑟𝑛ሺ𝑡 ൅ 𝜏ሻ𝑑𝑡 ାஶ ିஶ (2) We observe that at each step, the convolution is maximum. For each maximum value of 𝐶𝐶ሺ𝜏ሻ we calculate the SRS of each step and of the entire signal as explained in the next step. • Step 4: The calculation of the SRS is based on the acceleration time history. It applies an ac- celeration time history as a common base excitation ሺ𝑦ሷሻ to an array of single-degree-of- freedom (SDOF) systems composed of spring ሺ𝑘௜ሻ, mass ሺ𝑚௜ሻ and damper ሺ𝑑௜ሻ, as de- picted in Figure 6. Figure 6. SRS model. 𝑥ሷ௜ is the absolute response of each system to the input 𝑦ሷ. This can be determined by applying Newton’s law to a free-body diagram of an individual system, as shown in Fig- ure 7. Figure 7. Free-body diagram of an individual system. The force balance yields the following governing differential equation of motion: 𝑚𝑥ሷ ൅ 𝑑𝑥ሶ ൅ 𝑘𝑥 ൌ 𝑑𝑦ሶ ൅ 𝑘𝑦 (3) Figure 5. Pattern of one CrossFit athlete. • Step 3: We then carry out the cross-correlation CC(τ) between the runner’s pattern and the recording’s duration a(t): CC(τ) = Z +∞ −∞ a(t)pattern(t + τ)dt (2) We observe that at each step, the convolution is maximum. For each maximum value of CC(τ) we calculate the SRS of each step and of the entire signal as explained in the next step. • Step 4: The calculation of the SRS is based on the acceleration time history. It applies an acceleration time history as a common base excitation Sensors 2022, 22, 2350 6 of 11 By defining the relative displacement z = x − y, Equation (3) can be rewritten as: ..z + 2ξω .z + ω2z = − ..y (4) where ω0 = k m is the natural frequency in radians per second and ξ = d (2ω0m) is the damping ratio. Moreover, ξ is usually represented by the amplification factor Q = 1 (2ξ). Since the base excitation ..y is an arbitrary function of time, Equation (4) does not have a closed-form solution. To calculate the SRS of each step and of the entire signal, we have used the algorithm for the calculation of the SRS proposed in [13]. SRS enables us to determine the maximum acceleration a system will undergo when one knows the natural frequency f0 and the quality factor Q for each possible natural frequency. In this study, a relative damping of 5% was used, resulting in Q = 10. SRS can also be calculated for the entire duration of a recording. We then observed the peaks at the fundamental and harmonic frequencies of the recorded signal [16]. In this context, SRS combines both the notion of transfer function and response to transient regimes. Intra comparison of the SRS offers a lot more finesse to the analysis since the frequency is also taken into account. The aggressiveness of a running step is not only due to the value of the maximum acceleration but also to the general shape of the movement, only the SRS allows this to be taken into account in the analysis. Figure 8 gives the general workflow for SRS determination. Sensors 2022, 22, x FOR PEER REVIEW 6 of 11 By defining the relative displacement 𝑧 ൌ 𝑥 െ 𝑦, Equation (3) can be rewritten as: 𝑧ሷ ൅ 2𝜉𝜔𝑧ሶ ൅ 𝜔ଶ𝑧 ൌ െ𝑦ሷ (4) where 𝜔଴ ൌ 𝑘 𝑚 ൗ is the natural frequency in radians per second and 𝜉 ൌ 𝑑 ሺ2𝜔଴𝑚ሻ ൗ is the damping ratio. Moreover, 𝜉 is usually represented by the amplification factor 𝑄 ൌ 1 ሺ2𝜉ሻ ൗ . Since the base excitation 𝑦ሷ is an arbitrary function of time, Equation (4) does not have a closed-form solution. To calculate the SRS of each step and of the entire signal, we have used the algorithm for the calculation of the SRS proposed in [13]. SRS enables us to determine the maximum acceleration a system will undergo when one knows the natural frequency 𝑓଴ and the quality factor 𝑄 for each possible natural frequency. In this study, a relative damping of 5% was used, resulting in 𝑄 ൌ 10. SRS can also be calculated for the entire duration of a recording. We then observed the peaks at the fundamental and har- monic frequencies of the recorded signal [16]. In this context, SRS combines both the no- tion of transfer function and response to transient regimes. Intra comparison of the SRS offers a lot more finesse to the analysis since the fre- quency is also taken into account. The aggressiveness of a running step is not only due to the value of the maximum acceleration but also to the general shape of the movement, only the SRS allows this to be taken into account in the analysis. Figure 8 gives the general workflow for SRS determination. Figure 8. Workflow for SRS determination. 3. Results A goal of the present study was to analyze the effect of fatigue through SRS on the ability of the human musculoskeletal system to attenuate foot strike–generated shock waves. The results of this study suggest that, for the analysis of impact shock during run- ning, the different components of the acceleration signal can be distinguished in the fre- quency domain by means of spectral analysis as shown in Figure 9. Figure 8. Workflow for SRS determination. 3. Results A goal of the present study was to analyze the effect of fatigue through SRS on the ability of the human musculoskeletal system to attenuate foot strike–generated shock waves. The results of this study suggest that, for the analysis of impact shock during running, the different components of the acceleration signal can be distinguished in the frequency domain by means of spectral analysis as shown in Figure 9. The main advantage of spectral analysis over time-domain analysis of the impact shock wave is the ability to separate spectral peaks from the rest of the data. Since the motion, impact, and resonant components of the acceleration signal have different fundamental frequencies: they produce peaks at different points in the power spectrum [12]. Sensors 2022, 22, 2350 7 of 11 Sensors 2022, 22, x FOR PEER REVIEW 7 of 11 Figure 9. Example of SRS results for one CrossFit athlete extracted for three runs on both feet. The main advantage of spectral analysis over time-domain analysis of the impact shock wave is the ability to separate spectral peaks from the rest of the data. Since the motion, impact, and resonant components of the acceleration signal have different funda- mental frequencies: they produce peaks at different points in the power spectrum [12]. The hypothesis is that fatigue hampers the ability of the human musculoskeletal sys- tem to protect itself from overloading due to foot strike–generated shock waves, loss of protection may manifest as an increased shock wave amplitude. For all five athletes, there was a direct correlation between fatigue and an increase in the aggressiveness of the SRS as shown in Figure 10. We noticed that for all five athletes for the 3rd run the average SRS peak was significantly higher than for the 1st run and 2nd run (p < 0.01) at the same natural frequency of the athlete. This confirms our hypothesis that fatigue causes a decrease in the shock attenuation capacity of the musculoskeletal system thus potentially involving a higher risk of overuse injury. Figure 10. Average SRS peaks for every athlete. 102 G 86 G 74 G Figure 9. Example of SRS results for one CrossFit athlete extracted for three runs on both feet. The hypothesis is that fatigue hampers the ability of the human musculoskeletal system to protect itself from overloading due to foot strike–generated shock waves, loss of protection may manifest as an increased shock wave amplitude. For all five athletes, there was a direct correlation between fatigue and an increase in the aggressiveness of the SRS as shown in Figure 10. We noticed that for all five athletes for the 3rd run the average SRS peak was significantly higher than for the 1st run and 2nd run (p < 0.01) at the same natural frequency of the athlete. This confirms our hypothesis that fatigue causes a decrease in the shock attenuation capacity of the musculoskeletal system thus potentially involving a higher risk of overuse injury. Sensors 2022, 22, x FOR PEER REVIEW 7 of 11 Figure 9. Example of SRS results for one CrossFit athlete extracted for three runs on both feet. The main advantage of spectral analysis over time-domain analysis of the impact shock wave is the ability to separate spectral peaks from the rest of the data. Since the motion, impact, and resonant components of the acceleration signal have different funda- mental frequencies: they produce peaks at different points in the power spectrum [12]. The hypothesis is that fatigue hampers the ability of the human musculoskeletal sys- tem to protect itself from overloading due to foot strike–generated shock waves, loss of protection may manifest as an increased shock wave amplitude. For all five athletes, there was a direct correlation between fatigue and an increase in the aggressiveness of the SRS as shown in Figure 10. We noticed that for all five athletes for the 3rd run the average SRS peak was significantly higher than for the 1st run and 2nd run (p < 0.01) at the same natural frequency of the athlete. This confirms our hypothesis that fatigue causes a decrease in the shock attenuation capacity of the musculoskeletal system thus potentially involving a higher risk of overuse injury. Figure 10. Average SRS peaks for every athlete. 102 G 86 G 74 G Figure 10. Average SRS peaks for every athlete. When fatigue begins, we could hypothesize that athletes will slow down as a protective means. The result could be moving away from the state of fatigue, in which case the acceleration data could have not increased. It was not the case in our study. Previous studies have shown that the loading rate of the lower limb is directly and highly correlated with running speed, and the vertical impact force increased with increas- ing running velocity [17]. Sensors 2022, 22, 2350 8 of 11 Muscle activation lowers the bending stress on bone and attenuates the peak dynamic loads that can damage musculoskeletal tissues. Previous studies have suggested that the fatigued muscles cannot support “optimal” running and they also suggested that fatigue of the runner may lead to modification of landing phase mechanics. It was also found that the transfer of mechanical energy between the eccentric and concentric phases is drastically reduced during muscle fatigue. Such changes may be involved in the development of injuries [18–20]. 4. Discussion According to the results presented in this study, for the acceleration data to increase, fatigue should be present. We may conclude that the musculoskeletal system becomes less capable of handling foot strike–induced shock waves when the muscles are significantly fatigued. One of the most common running overuse injuries are bone stress fractures (SF) [21,22]. In bones, microcracks are normally present and are thought to be fatigue-related cracks because their numbers increase following repetitive loading. Bone remodeling serves to repair fatigue microcracks. When a bone is loaded repeatedly, resulting in repetitive or cyclic strain, the subsequent accumulation of microdamage is believed to be the threshold of a pathological continuum that is clinically manifested as stress reactions and SF (29). Ultimately, if the activity is not ceased and the bone is not able to self-repair, a complete bone fracture might ensue. Notably, with increasing strain or greater strain rates, the number of loading cycles a bone–29 can withstand before a fatigue failure occurs is reduced [23]. Stress fractures are the clinical manifestation of the accumulation of fatigue damage in bones [24–26]. Although the effect of running and its mechanical strain in bone tissues is well documented, the evidence for SF etiology is less conclusive [24,27–30]. Nevertheless, several researchers reported clear relationships between bone stress related injuries and fatigue. For instance, it is known that the tensile strains on the tensile side of a bending bone are dampened by the contraction of adjacent muscles, aiming at protecting the bone from stress related injury [29–34]. It may then be hypothesized that muscles also play the role of shock absorbers and that consequently, muscle fatigue might decrease their absorption properties, resulting in a more aggressive loading rate or loading peak at the bones as fatigue increases [31–33]. The obtained results showed that acceleration amplitude steadily increased with the fatigue group and that there was a clear association between fatigue and shock waves (as revealed by the SRS). We may then confirm the conclusions of the aforementioned studies, that the human musculoskeletal system becomes less capable of single leg strike–induced shock waves absorption when the muscles are significantly fatigued. This condition may promote the development of injuries and the present results have a significant implication regarding the etiology of running injuries. Therefore, several recommendations may be effective towards runners’ community or coaches in order to reduce this stress related injury risk, notably as proposed by the multifactorial model of Brukner and Khan [35]. First, it may be advantageous to ensure that the majority of training and exercise is performed to avoid severe fatigue and in line with the load management theory. For instance, external parameters must be considered, such as progressive increment of training loads [36], training surfaces or footwear adaptations [35]. Understanding the influence of SRS on fatigue and on the magnitude of dynamic loading on the human musculoskeletal system will allow the development of proper training procedures and may participate in the reduction of damages to the musculoskeletal tissues. Secondly and directly in accordance with the present research purpose, lower limb muscles resistance to fatigue is a major component of stress related injury prevention in run- ners. The present outcomes are in line with former findings reporting that fatigue-related imbalance between the plantar flexors and dorsiflexors may compromise the protective action of these muscles on the lower leg bony structures [29]. Here, it is plausible that deteriorated properties of the calf muscles due to fatigue may affect the role of these soft tissues to protect the bone from stress injury risk. Sensors 2022, 22, 2350 9 of 11 Finally, injuries in running are also often provoked by fatigue and improper technique, which are both reflected in the runner’s kinematics [18,37,38]. A gait retraining approach has been proposed by several researchers through a modest increase in step rate or a transi- tion from rearfoot to forefoot strike and was considered as effective notably at reducing impact forces and vertical load rate and then at preventing running-related bone stress injuries [37,39]. An individualized approach is nevertheless of high interest and most likely available nowadays. Indeed, state-of-the-art research on kinematics in sports uses optical motion capture systems that are inaccessible to most athletes. With the recent development of Micromachined Microelectromechanical Systems (MEMS), inertial sensors have become widely used in the research of wearable running gait analysis [40] due to several factors, such as being easy-to-use and low-cost. Considering the fact that each individual has a unique way of running, inertial sensors can be applied to the problem of gait recognition where assessed gait can be interpreted as a biometric signature. Thus, inertial sensor-based gait recognition has a great potential to play an important role in many health-related applications. In this work, we demonstrated the potential of wearable technology for the assessment of kinematic parameters using the example of running. We concluded that wear- able technology opens possibilities for technique improvement and injury risk reduction to a wide spectrum of athletes. Since inertial sensors are included in smart devices that are nowadays present at every step, inertial sensor-based gait recognition has become a very attractive and emerging field of research that will provide many interesting discoveries. Although the small sample size is indeed a limitation to applying our findings to the general population, this study is a qualitative and prospective research study exploring a novel and unknown topic. Using SRS as a measurement in running gait analysis has never been studied as of today, leaving us with very little data similar to our study design to be able to calculate a traditional sample size. However, these results still provide valuable information regarding the use of SRS as a biomechanical risk factor in runners. Larger studies on this topic will further advance our understanding of injury risk in runners. It is acknowledged that the relatively low subject numbers used in this study limit the drawing of definitive conclusions, this is particularly true if the study findings conflict with those of previous investigations. The results of our research were in accordance with previous studies and our hypothesis. In the future, a study on a larger group of athletes will be carried out to confirm our previous findings. It is also planned to carry out the same type of studies on high level runners and compare results. Our obvious hypothesis is that elite runners will have a unique ability to dampen the SRS and/or sustain a much higher SRS threshold. Author Contributions: Conceptualization, D.B., S.O., B.A. and R.T.; methodology D.B., S.O., B.A. and R.T.; software, S.O. and B.A.; validation, S.O., B.A., F.F. and R.T.; formal analysis, S.O. and B.A.; investigation, D.B., B.C. and R.T.; resources, D.B. and B.C.; data curation, D.B. and R.T.; writing original draft preparation, D.B.; writing review and editing, R.T. and B.A.; visualization, B.C. and F.F.; supervision, R.T.; project administration, R.T. and D.B. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Ethical Review Board of the University of Reims Champagne-Ardenne and the medical board of the Centre Luxembourg. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: The data presented in this study are available on request from the corresponding author. The data are not publicly available due to ethical reasons. Conflicts of Interest: The authors declare no conflict of interest. Sensors 2022, 22, 2350 10 of 11 References 1. Francis, P.; Whatman, C.; Sheerin, K.; Hume, P.; Johnson, M.I. The Proportion of Lower Limb Running Injuries by Gender, Anatomical Location and Specific Pathology: A Systematic Review. J. Sports Sci. Med. 2018, 18, 21–31. 2. Davis, I.S.; Bowser, B.J.; Mullineaux, D.R. Greater vertical impact loading in female runners with medically diagnosed injuries: A prospective investigation. Br. J. Sports Med. 2016, 50, 887–892. [CrossRef] [PubMed] 3. Novacheck, T.F. The biomechanics of running. Gait Posture 1998, 7, 77–95. [CrossRef] 4. Lafortune, M.A.; Lake, M.J.; Hennig, E.M. Differential shock transmission response of the human body to impact severity and lower limb posture. J. Biomech. 1996, 29, 1531–1537. [CrossRef] 5. Gruber, A.H.; Boyer, K.A.; Derrick, T.R.; Hamill, J. Impact shock frequency components and attenuation in rearfoot and forefoot running. J. Sport Health Sci. 2014, 3, 113–121. [CrossRef] 6. Coventry, E.; O’Connor, K.M.; Hart, B.A.; Earl, J.E.; Ebersole, K.T. The effect of lower extremity fati gue on shock attenuation during single-leg landing. Clin. Biomech. 2006, 21, 1090–1097. [CrossRef] 7. Verbitsky, O.; Mizrahi, J.; Voloshin, A.; Treiger, J.; Isakov, E. Shock Transmission and Fatigue in Human Running. J. Appl. Biomech. 1998, 14, 300–311. [CrossRef] 8. Hoenig, T.; Rolvien, T.; Hollander, K. Footstrike patterns in runners: Concepts, classifications, techniques, and implications for running-related injuries. Ger. J. Sports Med. 2020, 71, 55–61. [CrossRef] 9. Mercer, J.A.; Bates, B.T.; Dufek, J.S.; Hreljac, A. Characteristics of shock attenuation during fatigued running. J. Sports Sci. 2003, 21, 911–919. [CrossRef] 10. Nigg, B.M. Impact forces in running. Curr. Opin. Orthop. 1997, 8, 43–47. [CrossRef] 11. Gallant, J.L.; Pierrynowski, M.R. A theoretical perspective on running-related injuries. J. Am. Podiatr. Med. Assoc. 2014, 104, 211–220. [CrossRef] [PubMed] 12. Benjamin, D.; Odof, S.; Abbes, B.; Nolot, J.B.; Erre, D.; Fourchet, F.; Taiar, R. Shock response spectrum analysis in running performance. Comput. Methods Biomech. Biomed. Eng. 2020, 23, s28–s30. [CrossRef] 13. Lalanne, C. Mechanical Vibration and Shock Analysis Mechanical Shock, 2nd ed.; Wiley-ISTE: London, UK, 2009; Volume 2, pp. 51–92. 14. Alexander, R.; Jayes, A.S. Fourier analysis of forces exerted in walking and running. J. Biomech. 1980, 13, 383–390. [CrossRef] 15. Johnson, G.R. The use of spectral analysis to assess the performance of shock absorbing footwear. Eng. Med. 1986, 15, 117–122. [CrossRef] [PubMed] 16. Benjamin, D.; Abbes, B.; Odof, S.; Nolot, J.B.; Fourchet, F.; Chiementin, X.; Taiar, R. Harmonic decomposition and analysis of running gait. Comput. Methods Biomech. Biomed. Eng. 2019, 22, s343–s344. [CrossRef] 17. Shorten, M.R.; Winslow, D.S. Spectral Analysis of Impact Shock during Running. Int. J. Sport Biomech. 1992, 8, 288–304. [CrossRef] 18. Milner, C.E.; Hamill, J.; Davis, I.S. Distinct hip and rearfoot kinematics in female runners with a history of tibial stress fracture. J. Orthop. Sports Phys. Ther. 2010, 40, 59–66. [CrossRef] 19. Mizrahi, J.; Verbitsky, O.; Isakov, E. Fatigue-induced changes in decline running. Clin. Biomech. 2001, 16, 207–212. [CrossRef] 20. Warden, S.J.; Hurst, J.A.; Sanders, M.S.; Turner, C.H.; Burr, D.B.; Li, J. Bone adaptation to a mechanical loading program significantly increases skeletal fatigue resistance. J. Bone Miner. Res. 2005, 20, 809–816. [CrossRef] 21. Clansey, A.C.; Hanlon, M.; Wallace, E.S.; Lake, M.J. Effects of fatigue on running mechanics associated with tibial stress fracture risk. Med. Sci. Sports Exerc. 2012, 44, 1917–1923. [CrossRef] 22. Matheson, G.O.; Clement, D.B.; Mckenzie, D.C.; Taunton, J.E.; Lloyd-Smith, D.R. Stress fractures in athletes: A study of 320 cases. Am. J. Sports Med. 1987, 15, 46–58. [CrossRef] [PubMed] 23. Edwards, W.B.; David, T.; Rudolphi, T.J.; Gillette, J.C.; Derrick, T.R. Effects of running speed on a probabilistic stress fracture model. Clin. Biomech. 2010, 25, 372–377. [CrossRef] [PubMed] 24. Chen, T.L.; An, W.W.; Chan, Z.Y.S.; Au, I.P.H.; Zhang, Z.H.; Cheung, R.T.H. Immediate effects of modified landing pattern on a probabilistic tibial stress fracture model in runners. Clin. Biomech. 2016, 33, 49–54. [CrossRef] [PubMed] 25. Hadid, A.; Epstein, Y.; Shabshin, N.; Gefen, A. Biomechanical Model for Stress Fracture-related Factors in Athletes and Soldiers. Med. Sci. Sports Exerc. 2018, 50, 1827–1836. [CrossRef] [PubMed] 26. Martin, B. Mathematical model for repair of fatigue damage and stress fracture in osteonal bone. J. Orthop. Res. 1995, 13, 309–316. [CrossRef] [PubMed] 27. Derrick, T.R.; Edwards, W.B.; Fellin, R.E.; Seay, J.F. An integrative modeling approach for the efficient estimation of cross-sectional tibial stresses during locomotion. J. Biomech. 2016, 49, 429–435. [CrossRef] [PubMed] 28. Bennell, K.L.; Malcolm, S.A.; Wark, J.D.; Brukner, P.D. Models for the pathogenesis of stress fractures in athletes. Br. J. Sports Med. 1996, 30, 200–204. [CrossRef] 29. Mizrahi, J.; Verbitsky, O.; Isakov, E. Fatigue-related loading imbalance on the shank in running: A possible factor in stress fractures. Ann. Biomed. Eng. 2000, 28, 463–469. [CrossRef] 30. Mizrahi, J.; Verbitsky, O.; Isakov, E. Shock accelerations and attenuation in downhill and level running. Clin. Biomech. 2000, 15, 15–20. [CrossRef] 31. Fyhrie, D.P.; Milgrom, C.; Hoshaw, S.J.; Simkin, A.; Dar, S.; Drumb, D.; Burr, D.B. Effect of fatiguing exercise on longitudinal bone strain as related to stress fracture in humans. Ann. Biomed. Eng. 1998, 26, 660–665. [CrossRef] 32. Schaffler, M.B.; Radin, E.L.; Burr, D.B. Long-term fatigue behavior of compact bone at low strain magnitude and rate. Bone 1990, 11, 321–326. [CrossRef] Sensors 2022, 22, 2350 11 of 11 33. Beck, B.R. Tibial stress injuries. An aetiological review for the purposes of guiding management. Sports Med. 1998, 26, 265–279. [CrossRef] [PubMed] 34. Nordin, M.; Franke, V. Biomechanics of bone. In Basic Biomechanics of the Musculoskeletal System; Nordin, M., Frankel, V., Eds.; Lea and Febiger: Philadelphia, PA, USA, 1989; pp. 3–29. 35. Brukner, P. Book Chapter 4—Sports injuries overuse. In Brukner & Khan’s Clinical Sports Medicine: Injuries, 5th ed.; Brukner, P., Clarsen, B., Cook, J., Cools, A., Crossley, K., Hutchinson, M., McCrory, P., Bahr, R., Khan, K., Eds.; McGraw Hill: New York, NY, USA, 2017; Volume 1, Available online: https://csm.mhmedical.com/content.aspx?bookid=1970&sectionid=168688996 (accessed on 31 January 2022). 36. Blanch, P.; Gabbett, T.J. Has the athlete trained enough to return to play safely? The acute:chronic workload ratio permits clinicians to quantify a player’s risk of subsequent injury. Br. J. Sports Med. 2016, 50, 471–475. [CrossRef] 37. Pohl, M.B.; Mullineaux, D.R.; Milner, C.E.; Hamill, J.; Davis, I.S. Biomechanical predictors of retrospective tibial stress fractures in runners. J. Biomech. 2008, 41, 1160–1165. [CrossRef] [PubMed] 38. Dixon, J.; Creaby, W.; Allsopp, J. Comparison of static and dynamic biomechanical measures in military recruits with and without a history of third metatarsal stress fracture. Clin. Biomech. 2006, 21, 412–419. [CrossRef] 39. Willy, W.; Buchenic, L.; Rogacki, K.; Ackerman, J.; Schmidt, A.; Willson, J.D. In-field gait retraining and mobile monitoring to address running biomechanics associated with tibial stress fracture. Scand. J. Med. Sci. Sports 2016, 26, 197–205. [CrossRef] 40. Sprager, S.; Juric, M.B. Inertial Sensor-Based Gait Recognition: A Review. Sensors 2015, 15, 22089–22127. [CrossRef]

Shock Response Spectrum Analysis of Fatigued Runners.

03-18-2022

Benjamin, Daniel,Odof, Serge,Abbès, Boussad,Fourchet, François,Christiaen, Benoit,Taïar, Redha

eng

PMC7184578

1 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports Greyhound racing ideal trajectory path generation for straight to bend based on jerk rate minimization Md. imam Hossain✉, David eager & paul D. Walker this paper presents methods for modelling and designing an ideal path trajectory between straight and bend track path segments for racing greyhounds. to do this, we numerically generate clothoid and algebraic curve segments for racing quadrupeds using a sequential vector transformation method as well as using a helper equation for approaching ideal clothoid segments that would respect greyhound kinematic parameters and boundary conditions of the track. further, we look into the limitations of using a clothoid curve for racing dog track path design and propose a smooth composite curve for track transition design which roughly maintains G3 curvature continuity for smooth jerk to overcome limitations of a clothoid transition. finally, we show results from race data modelling and past injury data, which provide a strong indication of clothoid curve segments improving the dynamics and safety of racing greyhounds while reducing injuries. In the greyhound racing sports industry, injuries to dogs are highly prevalent1. The sport has grown exponentially in recent years due to live wagering accessibility and various revenue sharing programs2. As a result, it has become evident that better track design is required to reduce the likelihood of racing greyhound injuries at the tracks. Observations3 confirmed that in greyhound racing congestion occurs at the entrance to the first bend. Also, researchers theorized that a smooth-running path is required for curved track design without which quadrupeds are more likely to lose coordination at specific transitions4. Similarly, it was shown that various track shapes have considerable effects on greyhound injury rates indicating track curvature influences5. When it comes to track shapes and smooth paths, transition curves are an essential part of path design in many areas such as road design and train track designs6. Similarly, transition curves help reduce disturbances in quadruped gait symmetry4. This is because, quadrupeds are subject to a centrifugal force which induces an outward pull on the curved track path, forcing quadrupeds to deviate from navigating the track path4. Theoretically, a transition curve would also assist navigation of the body around the curved path even if it is not sufficiently banked7. Clothoid transition curves are extensively found in road and rail track designs such as it was found from the analysis that the Tokaido Shinkansen high-speed rail uses a 600 m clothoid transition in one of the 2.1 km radius bends to achieve a maximum travelling speed of 270 km/hr with minimal track path camber. Clothoid curves are essential for generating continuous curvature paths with straight and perfect arc segments8. This is achieved by linking constant curvature segments with clothoid segments8. For example, a clothoid can join a line and a circle with G2 curvature continuity where both the tangent vector and curvature at the line-circle intersection are continuous9. The performance of clothoid and other transition curves trajectories can be effectively analyzed by looking into their curvature profiles. Curvature is an import factor in trajectory designs as it affects the maximum speed a vehicle can travel without skidding or whether the pilot of an aeroplane suffers blackout as a result of g-forces10. Also, a valid curve is one which respects upper bound curvature constraints set by kinematics properties of mov- ing bodies11. In this paper, we illustrate numerical methods to approach clothoid curves and other transition curves to model and generate smooth running paths for greyhound racing. We also show galloping greyhound trajectory performance, relating it to injury rates and track shapes. The paper is organized as follows. Sections one and two Faculty of Engineering and Information Technology, University of Technology Sydney, Broadway, 2007, NSW, Australia. ✉e-mail: MDImam.Hossain@uts.edu.au open 2 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ describe greyhound trajectory and trajectory dynamics, respectively. In sections three and four, clothoid transi- tion generation and approaching ideal clothoid transitions for racing greyhounds are presented. Ideal transition curves developed for galloping greyhounds are presented in section five. Finally, section six evaluates racing greyhound trajectory performance for existing tracks. trajectory of a racing greyhound. In the greyhound racing industry, the trajectory of a racing greyhound is oftentimes overlooked for track designs and injury prevention measures despite its significance in dynamic out- comes for the animal. One key parameter which determines the trajectory of a racing greyhound is the curvature of its running path. Curvature, κ(s), is the change of heading relative to distance travelled8. Also, the curvature can be thought of as the inverse of the radius of curvature, which denotes the turning radius at any point in the path12. Furthermore, a related variable, sharpness α, is the change of curvature for distance travelled which also forms the basis for constructing continuous curvature path trajectories8. While designing a path, curvature change must remain smooth throughout the trajectory of a moving object as the centrifugal acceleration expe- rienced is directly proportional to the path curvature12. As a result, in trajectory generation for motion planning the smoothness of a trajectory is directly related to the smoothness of its curvature profile13. Likewise, for the path to be feasible, it must conform to continuous position, heading, as well as curvature at all points8. Now, if the path of the trajectory is defined by a function y = f(x) then the radius of curvature ρ at any given point can be found from the following equation:14 ρ =    +  ( )  1, , (1) dy dx d y dx 2 3/2 2 2 Then, the curvature is, κ ρ = 1 (2) However, if the path of the trajectory cannot be translated into a continuous function, then any three adjacent data points lying on the path can be used to calculate the radius of curvature at any given point using the circum- radius of a triangle formula15. The circumradius formula (3) provides the radius of the circumcircle of a triangle which is inherently cyclic16,17. The triangle is defined by the adjacent data points lying on the path, as shown below in Fig. 1. ρ = R = abc 4A (3) Where, a, b, and c denote the three sides of a triangle defined by three adjacent data points on the path, and A is the area of the triangle. An ideal racing greyhound trajectory would involve looking into two major control factors, greyhound head- ing which deals with curvature and sharpness of the running path and greyhound kinetics which deals with the acceleration/deceleration of a greyhound. Racing greyhound trajectory dynamics. The trajectory of a racing greyhound induces dynamic grey- hound conditions such as centrifugal acceleration, centrifugal jerk, and greyhound heading yaw rate. It also influ- ences racing greyhound states such as leaning, braking forces as a result of ground reaction force, centripetal force, stride frequency, and stride length. A sharp discontinuity in any of the dynamic conditions would result in a significantly unpredictable dynamic imbalance for a racing greyhound. During racing, such a situation would put a greyhound in considerably uncontrollable situations where there are already racing situations such as con- gestion and tight bends with variable track cross-falls along the width of the tracks. To design a trajectory for Figure 1. Calculating an arbitrary path’s instantaneous radius of curvature using data points lying on the path. 3 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ racing greyhounds which would meet the specific track design goals, the trajectory performance can be evaluated by looking into two dynamic factors of racing greyhounds namely: centrifugal jerk and yaw rate. These two fac- tors are highly sensitive to the trajectory performance of an object in motion as both are related to the radius of curvature of the trajectory. Modelling centrifugal jerk. Jerk is the rate change of acceleration. Like centrifugal acceleration, the effect of jerk is also experienced in the body18. Essentially, jerk is the increasing or decreasing of the force in the body18. Eager, et al.18 explains the use of jerk as a measure of safety in various disciplines including mechanical engineer- ing and civil engineering as well as its application in greyhound racing. Lower jerk values are essential as they indicate that the change in centrifugal acceleration is minimal for a greyhound while it is navigating its trajectory. For, humans, there are derived maximum acceleration change and corresponding time duration for this change for roller coaster rides18. No such derivations exist for racing quadrupeds yet. As a result, modelling of the centrif- ugal jerk for racing animals becomes an essential part of optimum trajectory generation. The first step to centrifugal jerk analysis is finding the instantaneous radius of the trajectory or calculating the radius of curvature at all points in the trajectory path. For cars and trains, calculation of the instantaneous radius of curvature is found by using geometric primitives and splines and approximated using continuous functions as their respective heading change is continuous. However, for greyhounds, the heading change by the greyhound is not continuous and expected to occur at every stride. Furthermore, greyhounds are known to have a stride frequency greater than 3 Hz19. This implies a greyhound would change its heading if required more than three times a second where the magnitude of each heading change could vary from stride to stride. Therefore, we can gather all the location coordinate data for strides of a single racing greyhound and calculate the instantaneous radius of curvature ρ of the racing greyhound using either the circumradius formula (3) or the perpendicular bisectors method. Then, we can calculate the racing greyhound’s instantaneous centrifugal acceleration from the instantaneous speed and radius of curvature. Finally, the instantaneous jerk is derived from the rate of change in the centrifugal acceleration. Modelling yaw rate. The yaw rate is the rate change of heading or turning. It relates a racing greyhound’s angular displacement to its forward speed. It also provides an indication of the stability of the path a racing grey- hound is taking. For example, it was shown from the race kinematic simulation and race data that racing grey- hounds’ yaw rate is not smooth immediately after jumping out from the starting boxes20. For a constant radius curve path, the yaw rate is simply the radius of curvature over speed (4) which is used for calculating a vehicle’s momentary radius of turn. For a racing greyhound trajectory, the yaw rate can be directly related to the sum of the lateral forces. A lower yaw rate would indicate lower lateral forces such as centrifugal force and frictional force acting on a greyhound. To maintain a smooth trajectory, a racing greyhound needs to maintain a smooth yaw rate. However, since the speed of the racing greyhound varies over time as well as the lateral frictional forces from the traction ground, maintaining a smooth yaw rate would also require careful balancing of these two factors while designing tracks to facilitate a smooth trajectory for a racing greyhound. ψ ρ  = s (4) clothoid track segments for deriving natural racing greyhound trajectory. The clothoid segment is a curve known for its curvature being proportional to its length21. This property of the clothoid is useful as it allows the gradual development of centrifugal acceleration or can act as centrifugal acceleration easement, which significantly reduces the risk of accidents occurring12. Recent research shows that there are different types of curves already developed, which can be used as centrifugal acceleration easement curves12. For example, Quintic polynomial and B-splines functions are computationally less expensive and also able to provide curvature con- tinuity for curve design13. However, the drawbacks of these functions are complex curvature profiles which are hard to follow as they are not necessarily smooth13. This is where clothoids are useful as their curvature profile is a straight line making them easy to follow13. Furthermore, clothoids are characterized by a linear curvature, allow minimizing of curvature variation where piecewise clothoids exhibit excellent smoothness properties22. For these fundamental reasons, currently clothoids are extensively found in road design and robot path planning to achieve smooth transitions in the trajectories22. We found that clothoids are essential at the race track not only for developing smooth path trajectories but also for reducing the likelihood of certain types of race dynamics hazards. From the race videos, it was noted that a greyhound is more likely to change lanes to a higher radius upon entering the first bend. This could be due to the track bend lacking adequate transition to accommodate for greyhound natural instantaneous yaw rate change and leaning rate change limits. As a result, the prospect of the greyhound bumping into another nearby greyhound increases significantly. This specific race dynamic outcome can be reduced or nearly eliminated if the track path has clothoid segments which match natural greyhound heading turning rate change limits. Generating clothoid segments for track path design. There are many methods available for comput- ing the clothoid. Most methods involve approximations to the clothoid21. For example, it can be approximated by high degree polynomial curves23, such as by an S-power series24 as well as by an arc spline9. Also, continued fractions and rational functions are commonly used for approximations9. A more recent development in the spline primitives found in much computer-aided design software makes it easy to approximate a clothoid while respecting boundary conditions such as curvature and tangent continuity. Also, spline primitives are known for good and fast controllability with positional and tangential constraints making them ideal for various applica- tions22. Each of the methods available results in different degrees of accuracy and may not be suitable for efficient 4 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ greyhound track path design purposes. This is mainly due to less controllability in generating a clothoid accord- ing to greyhound kinematics. Moreover, to accommodate the clothoid segment into the path design, a coordinate respecting system must be incorporated or derived from the existing clothoid methods which respects different design boundary conditions. computing clothoid curves using existing methods. The most common method of computing a clothoid can be found in its definition in terms of Fresnel integrals24 where it is computed using the Fresnel sine and cosine functions as shown in Eqs. (5a) and (5b) and using some forms of Taylor series expansions on the functions which converge for an independent variable8. Series expansion functions are extensively used because the clothoid defining formulas are transcendental functions21. The parametric plot of Fresnel sine and cosine functions provides coordinate values of the clothoid curve. However, this does not respect any form of unit scal- ing or boundary conditions as well as not allowing computing the clothoid for a specific rate of change of cur- vature, sharpness or smoothing applications. Similarly, Eqs. (6a) and (6b) give an approximation of Fresnel sine and cosine functions which converge for all independent variables x. Another common method involves utilizing auxiliary functions8, as shown in Eqs. (7a) and (7b). ( ) = ∫ S x t dt sin( ) (5a) x 0 2 ( ) = ∫ C x t dt cos( ) (5b) x 0 2 ∫ ∑       = = − + + = ∞ + S x t dt x n n sin( ) ( 1) (2 1)!(4 3) (6a) x n n n 0 2 0 4 3 ∫ ∑       = = − + = ∞ + C x t dt x n n cos( ) ( 1) (2 )!(4 1) (6b) x n n n 0 2 0 4 1 Equations (5a) and (5b) then can be written in the auxiliary function form, as shown below:8 π π = +         −         C x f x x g x x ( ) 1 2 sin 2 cos 2 (7a) 2 2 π π = −         −         S x f x x g x x ( ) 1 2 cos 2 sin 2 (7b) 2 2 Where auxiliary functions f and g are defined as: π π =    −              −    −              f x S x x C x x ( ) 1 2 cos 2 1 2 sin 2 (8a) 2 2 π =    −              +    −           π    g x C x x S x n x ( ) 1 2 cos 2 1 2 si 2 (8b) 2 2 Likewise, for auxiliary function definition of the clothoid a good rational approximation to compute the clothoid is using the following auxiliary functions8. = + . + . + . f x x x x ( ) 1 0 926 2 1 792 3 104 (9a) 2 = + . + . + . g x x x x ( ) 1 2 4 142 3 492 6 670 (9b) 2 3 Moreover, recently, researchers developed more efficient numerical methods where one such method is using arc length parameterisation12. While analytical methods lack parameterisation for different application case sce- narios researchers are becoming more reliant on developing numerical techniques for computing the clothoids. A numerical approach for generating the clothoid curve transitions for racing greyhounds and other quadrupeds. It is evident that existing methods lack greyhound kinematic parameterisation for racing greyhound transition design purposes. A numerical method is generally preferred as a first approach for incorporating different parametrisation into the clothoid curves. To develop a numerical technique for the clothoid which incorporates greyhound kinematics variables, we looked into the characteristics of the mathemat- ical model of the clothoid curve. A clothoid curve transition accomplishes a gradual transition from the straight to the circular curve of the constant radius where the curvature changes from zero to a finite value. As a result, the tangent vector ti, which lies on the clothoid curve, also gradually rotates from zero to a finite angle Fig. 2. Furthermore, let us assume a greyhound changes its heading with every stride as noted from the race data and 5 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ galloping gait of a greyhound. With these two crucial pieces of information relating to the clothoid curve tangent vector and the greyhound heading step-change length, we can apply vector transformation to generate a clothoid curve positional vector Pi Fig. 2. Now, we define the clothoid tangent vector as a function of greyhound stride length constant as denoted by transition segment length and a variable denoted by transition deflection angle. The transition deflection angle ai defines the local rotation of the clothoid curve tangent vector at a specific tran- sition segment location i relative to the horizontal axis. Moreover, as a clothoid curve transition would gradually increase its curvature with constant curvature acceleration, the transition deflection angle ai is a function of the transition deflection angle acceleration constant. The transition deflection angle acceleration d defines the rate change of curvature per transition segment length of the clothoid curve, which essentially tells us how quickly the clothoid tangent vector rotation is accelerating. Finally, once the transition deflection angle is calculated for local ith transition segment, the clothoid curve positional vector can be calculated as shown in Fig. 2 and Eq. (11). To generate the entire clothoid curve for the specified number of transition segments by the constant n the process of translating and then rotating the clothoid tangent vector is iterated to get the clothoid positional vectors for all the transition segments. For example, Fig. 3 shows a clothoid curve generated using this method when transition segment length s equals 1 m, the number of transition segments n equals 250 and transition deflection angle acceleration d is 0.02 degrees. d = transition deflection angle acceleration ai = transition deflection angle relative to horizontal axis s = transition segment length n = number of transition segments ti = transition tangent vector i = transition segment number =       =    × ×    t f s a cos a s sin a s , ( ) ( ) (10) i i i i = = + − − P f t P P t ( , ) (11) i i i i i 1 1 Where, = ∑ × = × + × + × + … + × = a d i d d d i d 1 2 3 i k i 1 And, κ × ∝ d i Figure 2. Racing greyhound clothoid path generation using numerical method parameterization. Figure 3. A clothoid curve with curvature combs containing 250 single meter segments and with a turning acceleration of 0.02 degrees per segment. 6 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ Where κ denotes the curvature of the clothoid curve. Now, for instance, using the numerical method explained above to generate a clothoid curve transition for racing greyhounds with a transition exit radius of approximately 52 m and a total transition length of 45 m, we would have to consider the d constant to be 0.69 degrees per transition segment, the s constant to be 5 m (assum- ing that average stride length of a greyhound is 5 m) and the n constant to be 9. The curvature and jerk results of this clothoid transition curve for racing greyhounds are shown in Fig. 4. The numerical calculation of ai and Pi is shown in Table 1. Using this numerical method approach, we showed how an optimized clothoid curve transition could be determined numerically by tweaking curve generating factors. The controlling of initial values as set by d, s, and n allows generating any combination of clothoid curves as required for different kinematic path design goals. Figure 4. A clothoid curve transition for racing greyhounds with a total 45 m transition length having a an approximately 52 m turning radius at the end of the transition. i ai (deg. per segment) Pi X coordinate (m) Pi Y coordinate (m) 1 0.69 5.00 0.00 2 2.07 10.00 0.06 3 4.14 15.00 0.24 4 6.9 19.98 0.60 5 10.35 24.95 1.20 6 14.49 29.87 2.10 7 19.32 34.71 3.35 8 24.84 39.43 5.01 9 31.05 43.96 7.11 Table 1. Numerically calculated values of ai and Pi variables for a clothoid curve. 7 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ Designing ideal clothoid segments for racing greyhounds and other quadrupeds. When design- ing clothoid segments, it is essential that greyhound heading is not changing at the maximum performable rate since such a heading would put a greyhound into a limit state turning while maintaining a high speed. An ideal clothoid segment would have continuous curvature to allow a greyhound to navigate the path with a minimal amount of veering effort. In the next section, we derive a helper equation which can be used for specifying ideal clothoid transitions as well as for modelling dynamics for racing greyhounds at the tracks. Deriving an equation for exact clothoid requirements for racing greyhounds and other quadru- peds. Equation (3) produced a relationship between greyhound kinematics such as heading turning angle accel- eration and turning radius at the end of a natural clothoid transition. First, let’s assume for a clothoid transition a racing greyhound would pass ns number of strides with a constant s meter stride length. Now, if the total clothoid transition length is T meters, then the number of greyhound strides ns in a transition is given by Eq. (13). Again, since the length of the greyhound’s strides remains unchanged in the clothoid transition, the greyhound’s turning angle a in the last stride of the transition can be defined by Eq. (14) if the greyhound heading turning angle is accelerating with d degrees per stride. Now, to calculate a greyhound’s heading radius of turn R near the end of the clothoid transition using Eq. (3), we use Heron’s formula (17) to calculate the area of the triangle A (17) formed by last two greyhound strides s1 and s2. Furthermore, using the cosine rule we calculate the unknown side s of the triangle formed by the last two greyhound strides s1 and s2. Finally, by plugging in values for R and simplifying the equation, we reach a final equation form (18) which defines a racing greyhound’s turning radius R at the end of the clothoid transition in terms of transition length T, greyhound heading turning acceleration a and greyhound con- stant stride length s. Consequently, as Eq. (18) relates greyhound heading turning parameters to clothoid transition parameters, which is useful for modelling and designing ideal clothoid transitions for racing greyhounds. In the next section, we show some of the design and modelling of the clothoid transitions using Eq. (18). d = transition deflection angle acceleration (per stride) a = deflection angle of greyhound heading for last greyhound stride ns = total number of greyhound strides in the transition s = length of a single stride R = transition last stride turn radius T = transition length ns = T s (13) = × − a d (ns 1) (14) = + − − s s s 2s s Cos a ( 180) (15) 1 2 2 2 1 2 Where s1 and s2 are a racing greyhound’s last two strides in the transition. = + + p s s s 2 (16) 1 2 Where p is semi-perimeter of the inscribed triangle (Fig. 1) in the circle formed by a racing greyhound’s last two strides s1 and s2. = − − − A p p s p s p s ( )( )( ) (17) 1 2 =       + −          −    π π − − ( ) ( ) R s s s s 2 2 cos 2 2 cos 1 (18) d d 2 2 1 180 2 4 1 90 T s T s clothoid design for constant radius bend. Every track has a bend radius requirement as calculated from the physical infrastructure and design goals. If a track requires a 52 m radius bend at the end of the transition, then using Eq. (18), we find the following expected greyhound kinematics and transition design possibilities as shown in Table 2. It should be noted that there could be a large number of design outcomes for a single parameter design such as a design for a specific bend radius. The greyhound yaw rate at the entrance is simply the greyhound angular displacement rate change per stride times greyhound stride frequency. Also, in generating the folllowing results racing greyhound speed was assumed to be 19.5 m/s and stride frequency to be 3.5 Hz. As can be seen from Table 2, each of the clothoid transition possibilities can be applied at different locations at the track based on the race requirements. For instance, the clothoid transition Design No. 3 can be applied at the home turn bend exit since the greyhound speed and stride length would be much lower making it possible for a greyhound to adopt to higher yaw rate and angular displacement acceleration path navigation. 8 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ clothoid design for greyhound angular displacement rate change limits. For racing greyhounds known to have certain angular displacement rate change limits based on greyhound training and health back- ground histories, using Eq. (18) we can enumerate possible clothoid designs options. For example, if the expected racing greyhounds have a maximum angular displacement rate change limit of 0.5 deg/stride2, then we can con- sider the following clothoid transition design options as shown in Table 3. As can be seen from Table 3, using greyhound angular displacement rate change as a design constraint exhibits more diverse clothoid transitions in terms of transition length and transition exit bend radius. Design No. 1 shows that it is possible to have a short transition for a larger radius bend. Likewise, design no. 4 portrays a long transi- tion for smaller radius bend. As a result, the angular displacement rate based design approach provides excellent freedom in choosing clothoid transitions based on track requirements. Modelling of racing greyhound jerk dynamics. It is possible to calculate jerk exhibited by clothoid transitions using Eq. (18). Since a clothoid has uniform curvature acceleration, the jerk produced by a clothoid remains the same for the entire length of the transition. So, we can find jerk value at any arbitrary location in a clothoid transition to find overall jerk for the transition. For example, if we are interested in the jerk at the end of a clothoid transition, first we would calculate radius value R for both T and T-s for the transition. Then, we would calculate corresponding centrifugal acceleration values. Finally, since the jerk is the change in centrifugal acceleration over time, we simply divide the difference of centrifugal accelerations by the time taken by one stride. Table 4 presents some example calculations of racing greyhound jerk values for various clothoid transition designs considering instantaneous greyhound speed to be 19.5 m/s: An approach to designing ideal transitions for racing greyhounds. As can be seen from Fig. 4, it was found that racing greyhound clothoid transition curves have a significant flaw. Although the development of the curvature is gradual as can be seen from the curvature plot of Fig. 4, the jerk profile is not smooth and almost jumps instantaneously from zero to a higher value (Fig. 4). This is important, as such a dramatic change of jerk would impose a high energy release in a short time resulting in considerably unstable conditions for greyhounds navigating in and out of the transitions. Furthermore, the clothoid curve generation for racing greyhounds using the numerical method above showed that regardless of transition curve length jerk goes through a step change within one transition segment or one racing greyhound stride. Consequently, a clothoid curve transition was deemed not to be an ideal fit for racing greyhound track path designs. The clothoid transition curve does not maintain a smooth jerk initiation for a racing greyhound. Hence the curve can only be considered G2 continuous with matching curvature at the entrance and exit of the transition curve. This imposes several disadvantages in racing greyhound race dynamics at the tracks. For example, we can Design No. Clothoid transition length, T (m) *Greyhound yaw rate at the transition entrance (deg/s) Greyhound angular displacement rate change per stride, d (deg/stride2) Greyhound expected constant stride length, s (m) 1 75 1.2969 0.393 5.0 2 60 1.6533 0.501 5.0 3 40 2.3825 0.722 4.8 4 60 1.7952 0.544 5.2 Table 2. Clothoid transition options for 52 m radius bend. Design No. Clothoid transition length, T (m) Radius of constant bent at the transition end (m) Greyhound expected constant stride length, s (m) 1 45 71.6 5.0 2 50 70.5 5.25 3 60 52.0 5.0 4 70 53.7 5.5 Table 3. Clothoid transition options for racing greyhound accelerating with a maximum angular heading turning of 0.5 degrees per stride2. Design No. Clothoid transition length, T (m) Radius of constant bend at the transition end (m) Greyhound expected constant stride length, s (m) Greyhound angular displacement rate change per stride, d (deg/stride2) Absolute jerk (m/s3) 1 45 71.6 5.0 0.5 2.59 2 50 70.5 5.3 0.5 2.35 3 60 52.0 5.0 0.5 2.59 4 70 53.7 5.5 0.5 2.14 Table 4. Clothoid transitions racing greyhound’s jerk modelling using Eq. (18). 9 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ break down the disadvantages into two main categories, namely clustering related problems and path smoothing, where each is entangled with the other. The clustering of racing greyhounds is a common issue during races. This happens mainly due to single lure convergence as a result of the number of following galloping greyhounds. A tight convergence of the racing greyhound pack is noticeable at race tracks in the locations where track path cur- vature change is sudden and abrupt. As clustering is a precursor to various dynamics unstable conditions such as bumping of one greyhound by another, maintaining a smooth path profile such as G3 curvature continuity where the clustering occurs, becomes vital. As greyhounds follow the racing lure, they occupy different lanes such that they have different path radii and tend to cut corners forming various individual transitions into the bend which are all unique. A G2 curvature continuity as found in the clothoid transitions where the rate of change of the jerk is not smooth would induce all the racing greyhounds following the lure to follow one unique transition into the bend to keep instantaneous jerk to the minimum. This is not feasible. To overcome the limitations of clothoid transitions, we applied the numerical method of generating clothoid curves discussed in the previous section to develop moderate G3 curvature continuity transition curves for racing greyhounds. Also, two different transition curve configurations were selected for generating the curves as these configurations best match the many current tracks found in Australia in terms of real estate requirements. The configurations are a 45 m transition with transition end radius of 52 m and a 75 m transition with transition end radius of 70 m. First, we assume ai = X and plot for different X expressions to derive different curves where the curvature results for the curves are shown in Figs. 5 and 6. The X expression defines the nature of curvature function as the curve length increases from the origin. As seen from the plots (Figs. 5 and 6), when the X expression is linear it is a clothoid transition where the jerk is initiated immediately within one transition segment for both 45 m and 75 m transition configurations. To get G3 curvature continuity curves, we tried X0.6, X1.5, X2, and ((1.2)X−1) expres- sions. As can be seen from the plots, all the curves except the clothoid curve X and X0.6 curve maintain a moderate G3 curvature continuity with a smooth jerk profile. However, as X expressions are in power and logarithmic func- tion form for X0.6, X1.5, X2, and ((1.2)X−1) these curves result in higher jerk in the second half of the transition. This suggested that X1.5, X2, and ((1.2)X−1) curves could be used to develop a G3 curvature continuity transition curve for racing greyhounds if the jerk could be maintained in the second half of the transition. Thus, we decided to use these curves as auxiliary curves which would provide smooth jerk initiation for the transition. However, compared to other curves, the overall jerk and smoothness performance of the X1.5 is optimum. Here, we generate composite transition curves with various degrees of G3 curvature continuity for racing greyhound ideal path design. Each composite transition curve generated combines the X1.5 curve as an auxiliary curve and a clothoid curve as the main curve. So, the overall transition curve generating function can be consid- ered as a piecewise function shown in Eq. (19) where the auxiliary curve function g is applicable until q transition segment is reached.       =    < f x g x z x x q ( ) ( ) if otherwise (19) Where, x = y d i ( , ) Figure 7 shows curvature and jerk results for four different composite curves as ideal transitions for racing greyhounds, plotted using the numerical method explained in the earlier section, the configurations for these composite curves are given in Table 5. As can be seen from Fig. 7, composite transition curves have strong advantages over pure clothoid transition in terms of curvature and jerk continuities and excellent moderate G3 continuity for the first half of the transition. The overall instantaneous jerk is significantly lower in the composite curve transitions compared to clothoid tran- sitions. This is because the window of jerk initiation is much longer in composite curve transitions because of the gradual development of jerk and on average it is four transition segments or four greyhound strides compared to just one stride in the clothoid transitions. Greyhound racing data results A racing greyhound getting injured at the tracks provides an indication of its overall racing trajectory perfor- mance. Also, we can analyze the trajectory of a racing greyhound at the tracks to measure track path performance. Below, we present two such case scenarios by analyzing racing greyhound track data and injury rates. Race injury data results for track path renovation. In the greyhound racing track path design, it was found that only circular arcs (constant curvature) and lines (zero curvature) were used extensively despite non-continuous curvature resulting at the segment intersections25. A discontinuity in the curvature implies that a greyhound must change its heading instantaneously and abruptly resulting in a path which is not feasible8. Also, track survey data from Australia shows that a brief transition is applied, made of an arc spline consisting of one or more circular arcs joined with continuous tangent vectors. This particular design practice also leads to multiple discontinuities in track path curvature. We looked into one particular greyhound racing track (Track A) located in Australia and its two years of racing history. In the first year, it had a track path design with G1 continuity constituting half-circle bends and straights (Fig. 8). In the second racing year, the track was renovated with clothoid curve transitions into and out of the constant radius bends (Fig. 9). A 40 m clothoid transition was adjoined between a straight and a constant bend section for four bend and straight intersections. The outcome of this clothoid transition incorporation into 10 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ the original track path design eases centrifugal acceleration effect on the greyhounds where the centrifugal force is raised gradually from zero to an approximate nominal 240 N (Fig. 9). The renovation at Track A definitely would have changed centrifugal jerk performance significantly as the clothoid curve joining straights and bends would maintain G2 curvature continuity for the track path. To see whether this resulted in a significant decrease in racing injury rates, injury data for a two-year period were analyzed containing one year injury data for before and after renovation. By assuming differences in other contributing factors to injury rates such as variations in weather, track maintaining conditions, different greyhound breeds and training patterns, race operating condi- tions between the years were minimal the injury rates should show general trends due to track path renovation changes. We found that before the clothoid intervention at Track A the normalized catastrophic and major injury Figure 5. Different smooth curves curvature and jerk results as 45 m transition curves for greyhound racing. 11 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ rate per 1000 race starts was about 4.58 whereas after the clothoid intervention it was reduced to 4.22, a 7.9% reduction in this category of injury rates. Typically, this category of injury results from significant damage to greyhound physics. However, when we took into account all types of injuries at Track A for before and after the renovation, the normalized injury rates per 1000 race starts reduced to 26.71 from 44.68 injuries, a 40.2% reduction in overall injuries due to clothoid implementation at the track. Furthermore, under all injury types the most commonly occurring injury is happening in the greyhound forelegs responsible for turning assist for dog’s navigation. Metacarpal fractures and tibial fractures due to torsional stress occurring in the forelegs indicate navigational work stress on the greyhounds. Figure 6. Different smooth curves curvature and jerk results as 75 m transition curves for greyhound racing. 12 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ curvature of racing greyhound trajectory. Like any path following object, a racing greyhound has lim- itations of the radius of curvature or extrema of curvature for its running path. Also, when a racing greyhound runs following a track path which has curvature discontinuity or non-optimal transitions, a deviation in the greyhound’s position occurs from the projected track path trajectory. This phenomenon was observed in the greyhound location data in the races. Furthermore, numerical racing greyhound simulations confirmed that Figure 7. Four different straight to bend curvature graphs and jerk results for ideal racing greyhound transition curves. Composite curve configuration No. Transition deflection angle acceleration for auxiliary curve (deg.) Transition deflection angle acceleration for main clothoid curve (deg.) Total transition length (m) Transition exit radius (m) 1 0.3900 0.50 45 71.6 2 0.3900 0.52 45 68.9 3 0.1825 0.27 75 75.8 4 0.2500 0.32 75 64.0 Table 5. Kinematic and shape properties for four straight to bend composite curve transitions. 13 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ when a greyhound is following the line of sight of a lure, its yaw rate gradually builds up for the bend for a track shape which is less circular5. To see if there is any difference between racing greyhound path trajectory to track path, we analyzed racing greyhound location tracking data for a track which has non-optimal transition length to reduce the jerk magnitude. From the racing greyhound location tracking data and track survey data, we generated curvature results for both racing greyhound trajectory and track path (Fig. 10). The greyhound location data for all greyhounds starts were averaged to plot the results where ten races or eighty starts were considered to plot the results below. As can be seen from the curvature plot, there is a significant difference between racing greyhound trajectory and track path. This indicates racing greyhounds deviating from the track path to accommodate a more natural trajectory according to their physics. Also, it was observed from the analysis that transitions occurring in racing greyhound trajectory is relatively gradual and longer as indicated by the green dashed marker compared to the black dashed marker for track path. conclusions This paper presents a numerical method for generating racing greyhound clothoid transitions for track path designs along with an equation for modelling any kind of clothoid curves. The numerical technique is robust and can be algorithmically controlled to achieve defined goals compared to existing approaches for designing racing greyhound clothoid transitions. Moreover, it can be extended to function as a generator of other curves rather than just clothoid curves. By looking into jerk modelling data, an ideal transition curve is presented suitable for racing greyhound track path designs which overcomes limitations set by clothoid transitions. The effect of clothoid transitions in an existing track was verified by measuring injury rates over a two-year period. The trajectory of racing greyhounds in an existing track with inadequate transitions was analyzed to show non-optimum track path conditions. Finally, this paper showed evidence through modelling and injury data that clothoid and other composite curves improve racing dynamics safety for racing greyhounds. Furthermore, the methods presented here can also be used in designing and modelling trajectories for other moving bodies, including but not limited to horses, vehicles and trains. Figure 8. Track path curvature as shown by curvature combs for Track A with G1 curvature continuity for bends. Figure 9. Track path curvature as shown by curvature combs for Track A with G2 curvature continuity for bends. Figure 10. Track path and greyhounds trajectory curvature comparison. 14 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ Received: 1 December 2019; Accepted: 2 April 2020; Published: xx xx xxxx References 1. Sicard, G. K., Short, K. & Manley, P. A. A survey of injuries at five greyhound racing tracks. The Journal of small animal practice 40, 428–432, https://doi.org/10.1111/j.1748-5827.1999.tb03117.x (1999). 2. Beer, L. M. A study of injuries in Victorian racing greyhounds 2006–2011. (2014). 3. Hayati, H., Eager, D., Stevenson, R., Brown, T. & E., A. The impact of track related parameters on catastrophic injury rate of racing greyhounds in 9th Australian Congress on Applied Mechanics ACAM9 27–29 (Engineers Australia, Sydney, Australia 2017). 4. Fredricson, I., Dalin, G., Drevemo, S., Hjertén, G. & Alm, L. O. A Biotechnical Approach to the Geometric Design of Racetracks. Equine Veterinary Journal 7, 91–96, https://doi.org/10.1111/j.2042-3306.1975.tb03240.x (1975). 5. Mahdavi, F., Hossain, M. I., Hayati, H., Eager, D. & Kennedy, P. Track Shape, Resulting Dynamics and Injury Rates of Greyhounds in ASME 2018 International Mechanical Engineering Congress and Exposition. 6. Mathew, T. V. & Rao, K. K. Introduction to Transportation engineering. (2006). 7. Stubbs, A. K. Racetrack Design and Performance. (RIRDC, 2004). 8. Wilde, D. K. Computing clothoid segments for trajectory generation in 2009 IEEE/RSJ International Conference on Intelligent Robots and Systems. 2440–2445. 9. Meek, D. S. & Walton, D. J. An arc spline approximation to a clothoid. Journal of Computational and Applied Mathematics 170, 59–77, https://doi.org/10.1016/j.cam.2003.12.038 (2004). 10. Jia, Y. B. Curvature. 7 <http://web.cs.iastate.edu/~cs577/handouts/curvature.pdf> (2019). 11. Chen, Y., Cai, Y. & Thalmann, D. Efficient, Accurate and Robust Approximation of Clothoids for Path Smoothing. 12. Vázquez-Méndez, M. E. & Casal, G. The Clothoid Computation: A Simple and Efficient Numerical Algorithm in Journal of Surveying Engineering. (American Society of Civil Engineers). 13. Delingette, H., Hebert, M. & Ikeuchi, K. Trajectory generation with curvature constraint based on energy minimization in Proceedings IROS ‘91:IEEE/RSJ International Workshop on Intelligent Robots and Systems ‘91. 206–211 vol.201. 14. Hibbeler, R. C. Engineering Mechanics: Dynamics (13th Edition). (Prentice Hall, 2012). 15. Hossain, M. I., Hayati, H. & Eager, D. A Comparison of the Track Shape of Wentworth Park and Proposed Murray Bridge, University of Technology Sydney, Australia (2016). 16. Maslanka, D. J. Circumcircles and Incircles of Triangles. <http://mypages.iit.edu/~maslanka/C&I_Tri.pdf> (2018). 17. Kimberling, C. In The Mathematical Gazette Vol. 85 (1998). 18. Eager, D., Pendrill, A. M. & Reistad, N. Beyond velocity and acceleration: jerk, snap and higher derivatives. European Journal of Physics 37 (2016). 19. Hayati, H., Eager, D., Jusufi, A. & Brown, T. A Study of Rapid Tetrapod Running and Turning Dynamics Utilizing Inertial Measurement Units in Greyhound Sprinting in ASME 2017 International Design Engineering Technical Conferences and Computers and Information in Engineering Conference. 20. Hossain, M. I., Eager, D. & Walker, P. Simulation of Racing Greyhound Kinematics in Proceedings of the 9th International Conference on Simulation and Modeling Methodologies, Technologies and Applications - Volume 1: SIMULTECH Vol. 1 47-56 (SciTePress, Prague, Czech Republic, 2019). 21. Meek, D. S. & Walton, D. J. A note on finding clothoids. Journal of Computational and Applied Mathematics 170, 433–453, https:// doi.org/10.1016/j.cam.2003.12.047 (2004). 22. Bertails-Descoubes, F. Super-Clothoids. Computer Graphics Forum 31, 509–518 (2012). 23. Wang, L. Z., Miura, K. T., Nakamae, E., Yamamoto, T. & Wang, T. J. An approximation approach of the clothoid curve defined in the interval [0, π/2] and its offset by free-form29 curves. Computer-Aided Design 33, 1049–1058, https://doi.org/10.1016/S0010- 4485(00)00142-1 (2001). 24. Sánchez-Reyes, J. & Chacón, J. M. Polynomial approximation to clothoids via s-power series. Computer-Aided Design 35, 1305–1313, https://doi.org/10.1016/S0010-4485(03)00045-9 (2003). 25. Hongo, T., Arakawa, H., Sugimoto, G., Tange, K. & Yamamoto, Y. An Automatic Guidance System of a Self-Controlled Vehicle. IEEE Transactions on Industrial Electronics IE-34, 5–10, doi:10.1109/TIE.1987.350916 (1987). Acknowledgements This work was sponsored by Greyhound Racing NSW, Australia and Faculty of Engineering and Information Technology at the University of Technology Sydney, Australia. The authors would also like to acknowledge the support of Greyhound Racing Victoria for providing the greyhound location tracking and track survey data. The research was funded by Greyhound Racing NSW research grant “Identifying optimal greyhound race track design for canine safety and welfare Phase II”. Author contributions MIH conceived of the presented idea, developed the theoretical framework, performed the analytic calculations and performed the numerical modelling and simulations to derive the results. DE conceived of the presented idea, supervised the project and reviewed the manuscript. PW reviewed the manuscript. competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to M.I.H. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 15 Scientific RepoRtS | (2020) 10:7088 | https://doi.org/10.1038/s41598-020-63678-1 www.nature.com/scientificreports www.nature.com/scientificreports/ Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2020

Greyhound racing ideal trajectory path generation for straight to bend based on jerk rate minimization.

04-27-2020

Hossain, Md Imam,Eager, David,Walker, Paul D

eng

PMC7551623

nutrients Article Sex-Dependent Wheel Running Effects on High Fat Diet Preference, Metabolic Outcomes, and Performance on the Barnes Maze in Rats Tiffany Y. Yang 1, Zijun Gao 1 and Nu-Chu Liang 1,2,3,* 1 Department of Psychology, College of Liberal Arts and Sciences, University of Illinois—Urbana-Champaign, Champaign, IL 61820, USA; tyang42@illinois.edu (T.Y.Y.); zijung2@illinois.edu (Z.G.) 2 Division of Nutritional Sciences, College of Agricultural, Consumer and Environmental Sciences, University of Illinois—Urbana-Champaign, Urbana, IL 61801, USA 3 Neuroscience Program, College of Liberal Arts and Sciences, University of Illinois—Urbana-Champaign, Urbana, IL 61801, USA * Correspondence: ncliang8@illinois.edu; Tel.: +1-(217)-244-7873 Received: 22 July 2020; Accepted: 30 August 2020; Published: 5 September 2020



 Abstract: Excessive and prolonged intake of highly palatable, high fat (HF) foods contributes to the pathogenesis of obesity, metabolic syndrome, and cognitive impairment. Exercise can restore energy homeostasis and suppress HF diet preference in rats. However, it is unclear if exercise confers similar protection against the detrimental outcomes associated with a chronic HF diet preference and feeding in both sexes. We used our wheel running (WR) and two-diet choice (chow vs. HF) paradigm to investigate the efficacy of exercise in reversing HF diet-associated metabolic and cognitive dysregulation in rats, hypothesizing that beneficial effects of exercise would be more pronounced in males. All WR rats showed HF diet avoidance upon running initiation, and males, but not females, had a prolonged reduction in HF diet preference. Moreover, exercise only improved glucose tolerance and insulin profile in males. Compared to sedentary controls, all WR rats improved learning to escape on the Barnes maze. Only WR females increased errors made during subsequent reversal learning trials, indicating a sex-dependent effect of exercise on behavioral flexibility. Taken together, our results suggest that exercise is more effective at attenuating HF-associated metabolic deficits in males, and highlights the importance of developing sex-specific treatment interventions for obesity and cognitive dysfunction. Keywords: high fat diet; wheel running; oral glucose tolerance test; Barnes maze 1. Introduction In the United States, ~65% of adults are either overweight or obese presenting with chronic illnesses (e.g., cardiovascular disease, type 2 diabetes, hypertension, and cancer) that can be partially attributed to diet composition [1,2]. The shift in the types of food consumed and their nutritional qualities are associated with the change in environment (e.g., industrialization), including the development of agriculture, food processing, and animal husbandry [3]. Consumption of refined sugars (e.g., high-fructose corn syrup) and saturated fats has steadily increased [4–6] and has been termed the “Western diet,” which contains ~40% calories from both carbohydrates and fats [7]. Moreover, the modern environment favors a sedentary lifestyle and facilitates easy access to these highly processed, palatable, energy dense foods which tend to have higher glycemic loads [8] than unrefined foods and pose a threat to metabolic [9–12] and cognitive health [13]. The overconsumption of high fat (HF) food is associated with weight gain and increased abdominal adiposity, which contributes to development of peripheral metabolic dysregulation and cognitive Nutrients 2020, 12, 2721; doi:10.3390/nu12092721 www.mdpi.com/journal/nutrients Nutrients 2020, 12, 2721 2 of 22 deficits [13–15]. Exercise appears to be more effective than diet control at improving metabolic function [16]. People who are successful at maintaining long-term weight loss report high physical activity and a diet low in fat composition [17,18]. Rodent studies show that males consistently decrease food intake in response to exercise [19–22] whereas results are more variable in females [21–23]. In contrast, the majority of human studies focus on changes in food intake following acute rather than long-term exercise, and the response to exercise is highly variable in both sexes [24–29]. Although women may lose less weight than men during long-term exercise [30–33], there has been limited research to elucidate sex differences in exercise-related changes in energy intake and expenditure [34,35]. Therefore, it is unclear how closely sex differences in energy intake and macronutrient preference during exercise in humans parallel the rodent literature. Furthermore, compared with caloric restriction-mediated weight loss, exercise training-mediated weight loss led to greater decreases in visceral fat and hepatic insulin resistance in obese men and women [36]. Exercise can produce a modest improvement on peripheral glucose metabolism even without significant changes in body mass and composition. For example, short-term running exercise was able to reduce fasting glucose and portal vein free fatty acids in sucrose-fed rats without a concomitant reduction in adiposity [37]. Increasing the duration of running from four to 12 weeks in rats resulted in decreased mesenteric and subcutaneous fat in addition to increased insulin sensitivity and greater improvement in their overall metabolic profile [37]. Thus, while exercise can attenuate obesity-related insulin resistance, there may be an additive effect of exercise and adiposity loss on significantly improving peripheral insulin resistance. Diets high in fat composition have been implicated in affecting cognitive performance in both humans [38] and rodents [39]. More specifically, chronic HF feeding has been shown to impair hippocampal-dependent spatial learning and memory in rodent models [39–44]. Given that exercise has been shown to enhance cognition [45–47], it may also be able to attenuate HF-induced performance deficits. Indeed, studies have found that exercise, [48–52] but not dietary supplementation [52], reverses HF-mediated cognitive impairment. Most studies focus on diet-induced deficits in hippocampal dysfunction. However, beyond learning and memory, HF diet can also alter prefrontal cortex (PFC)-mediated executive function, including behavioral flexibility [15,42,53,54]. Deficits in behavioral flexibility can manifest prior to the development of HF-induced insulin resistance [55]. These deficits in behavioral flexibility may result in the inability to appropriately adapt dietary choices to external environmental and internal visceral cues and consequently, promote rigid diet choices in a viscous cycle that facilitates the development of obesity [56,57]. The few studies that examined PFC-dependent cognition found that HF feeding led to deficits in behavioral flexibility in rats [15,42,53,54]. Importantly, decreased behavioral flexibility was correlated with decreased insulin sensitivity but not body weight or plasma glucose level [54]. Dietary [58,59], but not drug [60], interventions are able to attenuate these HF diet-induced cognitive impairments. Taken together, these studies suggest that HF diet and insulin resistance may interact to promote the development and maintenance of cognitive rigidity. Sex differences in HF-induced deficits in metabolic and cognitive function may differentially affect the cognitive control of feeding behavior in males and females. The metabolic [61–64] and cognitive [65–67] outcomes of HF feeding appear to be more deleterious in males than females in both humans and rodents. However, obesity-related deficits, specifically in the domain of cognitive flexibility, are greater in women than men [68]. In contrast, HF-mediated deficits in behavioral flexibility are more pronounced in male rats compared to females [65–67]. Notably, both the human [69–73] and rodent [74–76] literature suggests that males are more responsive to the beneficial effects of exercise. Thus, exercise may more readily counteract the detrimental effects of HF-feeding in males than females. While exercise has some efficacy at counteracting HF-mediated cognitive decline in rodents [16,36–39,48,50,52], studies primarily focus on behaviors mediated by the hippocampus [48,50] rather than the PFC [49] and rarely include both sexes within the same experiment for the direct assessment of sex differences. Thus, whether exercise is able to attenuate functional dysregulation of the PFC to the same extent in both male and female rats is unclear. Nutrients 2020, 12, 2721 3 of 22 To our knowledge, no study has investigated the efficacy of exercise at reversing HF-induced deficits in metabolic and executive function in rats of both sexes, which could provide insight into the development of sex-specific prevention and treatment options. Furthermore, most rodent models lack a dietary choice component so the relationships between diet preference, exercise, and metabolic function are rarely explored. To address the gap in knowledge, rats underwent our established WR and two-diet choice protocol for six weeks, a period of time which is sufficient to induce diet-induced obesity [77]. With this long-term HF diet exposure, we examined if higher HF diet preference increases susceptibility to the adverse effects of HF diet, and whether exercise can lead to similar HF-diet associated alterations in metabolic profile and cognitive performance in male and female rats. We hypothesized that chronic HF feeding would impair peripheral metabolic function and PFC-mediated behavioral flexibility in a Barnes maze [58,78] to a greater extent in female than male rats that have a preference for HF diet. Furthermore, we hypothesized that exercise would have a greater efficacy at attenuating the adverse effects HF diet exposure in males compared to females. 2. Materials and Methods 2.1. Subjects The subjects included 24 male (250–275 g) and 24 female (150–175 g) Sprague-Dawley rats (Envigo, Indianapolis, IN, USA) that were ~7–8 weeks old upon arrival. Rats were group housed on a standard 12:12 light-dark cycle (lights on at 0700 h). During habituation, rats had ad lib access to a standard chow diet (chow; Teklad global 2018, Teklad Diets, Madison, WI, USA) and tap water. See Table 1 for details about macronutrient sources. During the experimental period, rats had diet choice between the standard, high carbohydrate chow and a novel 45% HF diet (HF; D12451, Research Diets, New Brunswick, NJ, USA). All groups were fed ad libitum. Table 1. Diet composition. Macronutrient Description Unit Teklad 2018 (3.1 kcal/g) 45% HF Diet (4.74 kcal/g) Protein Total % kcal 24 20 Carbohydrate Sucrose % kcal - 17 Other carbohydrates % kcal 58 18 Fat Total % kcal 18 45 Saturated fats % of total fat (wt) 0.9 31.4 Monounsaturated fats % of total fat (wt) 1.3 35.5 Polyunsaturated fats % of total fat (wt) 3.4 33.1 All experimental procedures were approved by the Institutional Animal Care and Use Committee at the University of Illinois, Urbana-Champaign (Protocol #16178) and are in accordance with the Guide for the Care and Use of Laboratory Animals [79]. 2.2. Procedures 2.2.1. Wheel Running and Two-Diet Choice Daily recording of body weight, food, and water intake occurred at 0800 h. Running activity was recordedonacomputerandprocesseddailyimmediatelyfollowingdailycare(VitalView,StarrLifeSciences, Oakmont, PA, USA). After habituation, sedentary (Sed) rats were moved to standard individual housing cages that included cotton nesting materials as enrichment, while wheel running (WR) rats were transferred to running wheel cages (13” diameter wheel; Mini Mitter, Starr Life Sciences, Oakmont, PA, USA) with the wheel locked for a 4-day acclimatization period. After this 4-day period, a novel 45% HF diet was introduced to all rats 2 h before dark onset (1700 h) during which running wheels were simultaneously unlocked for the WR rats. The sample size, n = 10–11 for Sed groups and n = 13–14 for WR groups, Nutrients 2020, 12, 2721 4 of 22 was determined based on our previous studies using the same two-diet choice and wheel running paradigm [80]. The observed power with such a sample size was ≥0.80, which is the typical cutoff used for power analyses. The wheel running and two-diet choice procedures continued for ~6 weeks after which the rats were sacrificed (Figure 1). Retroperitoneal, mesenteric, and gonadal fat pads were dissected and weighed after sacrifice. Nutrients 2020, 12, x FOR PEER REVIEW 4 of 22 same two-diet choice and wheel running paradigm [80]. The observed power with such a sample size was ≥0.80, which is the typical cutoff used for power analyses. The wheel running and two-diet choice procedures continued for ~6 weeks after which the rats were sacrificed (Figure 1). Retroperitoneal, mesenteric, and gonadal fat pads were dissected and weighed after sacrifice. Figure 1. Experimental timeline. Sed: sedentary, WR: wheel running; HF: high fat; OGTT: oral glucose tolerance test. 2.2.2. Oral Glucose Tolerance Test (OGTT) Two OGTTs were performed for this experiment, one at baseline and one after HF diet exposure (post-HF) to examine within-group effects of HF diet on glucose tolerance. Baseline OGTT was performed the day before the wheels were unlocked (day 0), and post-HF OGTT was performed the day of sacrifice (day 44), which was two days after the end of the Barnes maze. The day before both OGTTs, food was removed ~3 h after dark onset (2200 h) where rats would have eaten ≥60% of their daily food intake. Rats were only moderately fasted because complete overnight fasting enhances insulin-stimulated glucose utilization, and we were interested in assessing insulin action in a more physiological context [81,82]. After the baseline/fasting blood glucose (0 min) measurement and tail blood collection, rats underwent a glucose challenge where they were orally gavaged with 2 g/kg of 20% glucose dissolved in distilled water. Tail blood was collected from the same tail nick made during baseline sampling at 15, 30, 60, and 120 min time points from the time they were gavaged. Blood glucose levels were also measured at these time points using a handheld glucometer (AlphaTRAK2, Abbott, Abbott Park, IL, USA). Tail blood was centrifuged at 870 × g for 15 min at 4 °C. For each sampling time point, ~25 µL of plasma was collected and stored at −80 °C until the samples were processed for plasma insulin concentrations using the Rat Ultrasensitive Insulin ELISA (ALPCO, Salem, NH, USA) according to manufacturers’ protocol. Blood glucose and plasma insulin data from the baseline and post-HF OGTT was used to assess insulin sensitivity [83] with the following two methods: (1) Hepatic insulin resistance was calculated using the homeostasis model assessment method for insulin resistance (HOMA-IR) model [84] and (2) peripheral insulin resistance was calculated using Gutt’s index of insulin sensitivity (ISI0,120) [85]. HOMA − IR = fasting insulin (μU ml) × fasting glucose (mmol L ) 22.5 𝐈𝐒𝐈𝟎,𝟏𝟐𝟎 = glucose load (mg) + (glucose0 min − glucose120 min (mg L )) × 0.19 × body weight (kg) 120 × log ( insulin0 min + insulin120 min (mU L ) 2 ) + ( glucose0 min + glucose120 min (mmol L ) 2 ) (1) 2.2.3. Barnes Maze Figure 1. Experimental timeline. Sed: sedentary, WR: wheel running; HF: high fat; OGTT: oral glucose tolerance test. 2.2.2. Oral Glucose Tolerance Test (OGTT) Two OGTTs were performed for this experiment, one at baseline and one after HF diet exposure (post-HF) to examine within-group effects of HF diet on glucose tolerance. Baseline OGTT was performed the day before the wheels were unlocked (day 0), and post-HF OGTT was performed the day of sacrifice (day 44), which was two days after the end of the Barnes maze. The day before both OGTTs, food was removed ~3 h after dark onset (2200 h) where rats would have eaten ≥60% of their daily food intake. Rats were only moderately fasted because complete overnight fasting enhances insulin-stimulated glucose utilization, and we were interested in assessing insulin action in a more physiological context [81,82]. After the baseline/fasting blood glucose (0 min) measurement and tail blood collection, rats underwent a glucose challenge where they were orally gavaged with 2 g/kg of 20% glucose dissolved in distilled water. Tail blood was collected from the same tail nick made during baseline sampling at 15, 30, 60, and 120 min time points from the time they were gavaged. Blood glucose levels were also measured at these time points using a handheld glucometer (AlphaTRAK2, Abbott, Abbott Park, IL, USA). Tail blood was centrifuged at 870× g for 15 min at 4 ◦C. For each sampling time point, ~25 µL of plasma was collected and stored at −80 ◦C until the samples were processed for plasma insulin concentrations using the Rat Ultrasensitive Insulin ELISA (ALPCO, Salem, NH, USA) according to manufacturers’ protocol. Blood glucose and plasma insulin data from the baseline and post-HF OGTT was used to assess insulin sensitivity [83] with the following two methods: (1) Hepatic insulin resistance was calculated using the homeostasis model assessment method for insulin resistance (HOMA-IR) model [84] and (2) peripheral insulin resistance was calculated using Gutt’s index of insulin sensitivity (ISI0,120) [85]. HOMA − IR = fasting insulin  µU ml  × fasting glucose  mmol L  22.5 ISI0,120 = glucose load (mg)+(glucose0 min−glucose120 min ( mg L ))×0.19×body weight (kg) 120×log   insulin0 min+insulin120 min( mU L ) 2  +   glucose0 min+glucose120 min( mmol L ) 2   (1) Nutrients 2020, 12, 2721 5 of 22 2.2.3. Barnes Maze During the last week of the WR and two-diet choice period, all rats were trained on the Barnes maze starting 2.5 h after light onset (0930 h). A concealed overhead-mounted camera aimed directly at the center of the maze was used to film each trial, which was operated using a computer from the adjacent room. The Barnes maze was 99 cm high and 122 cm in diameter, with 20 evenly spaced holes that were 10 cm in diameter and 2 cm away from the edge. The apparatus was mounted on a rotatable wooden support system that allowed the maze to be rotated 360◦. The escape box (30 × 12.5 × 14.5 cm) was mounted underneath one hole with a 20◦ incline ramp. The location of five visuospatial cues was held constant during training and reversal learning. For both training and reversal learning trials, a trail ended when the rat entered the escape box or after the allotted time had elapsed. The maze and escape box were cleaned using a non-alcohol-based coverage spray between each rat to eliminate odor cues. After daily care, rats were single caged in standard tubs and moved to the room adjacent to the testing room for at least 1 h of habituation prior to testing. There were 4 trials/day during training (total of 16 trials) with 4 different starting locations (Figure 2). Rats were placed on the edge of the maze facing the wall/away from the center of the maze. All rats finished a trial at the first starting location with an inter-trial interval of 30 min before being tested at the second starting location. In other words, all rats completed a trial from the same starting location before the next round of trials began. The order of the starting locations remained the same for each training day. Rats were given 90 s to find the escape box, and if a rat failed to find the escape box within the allotted time, they were gently guided into the box, which was then covered. Rats were allowed to remain in the escape box for 15 s before being returned to their home cage. Nutrients 2020, 12, x FOR PEER REVIEW 5 of 22 During the last week of the WR and two-diet choice period, all rats were trained on the Barnes maze starting 2.5 h after light onset (0930 h). A concealed overhead-mounted camera aimed directly at the center of the maze was used to film each trial, which was operated using a computer from the adjacent room. The Barnes maze was 99 cm high and 122 cm in diameter, with 20 evenly spaced holes that were 10 cm in diameter and 2 cm away from the edge. The apparatus was mounted on a rotatable wooden support system that allowed the maze to be rotated 360°. The escape box (30 × 12.5 × 14.5 cm) was mounted underneath one hole with a 20° incline ramp. The location of five visuospatial cues was held constant during training and reversal learning. For both training and reversal learning trials, a trail ended when the rat entered the escape box or after the allotted time had elapsed. The maze and escape box were cleaned using a non-alcohol-based coverage spray between each rat to eliminate odor cues. After daily care, rats were single caged in standard tubs and moved to the room adjacent to the testing room for at least 1 h of habituation prior to testing. There were 4 trials/day during training (total of 16 trials) with 4 different starting locations (Figure 2). Rats were placed on the edge of the maze facing the wall/away from the center of the maze. All rats finished a trial at the first starting location with an inter-trial interval of 30 min before being tested at the second starting location. In other words, all rats completed a trial from the same starting location before the next round of trials began. The order of the starting locations remained the same for each training day. Rats were given 90 s to find the escape box, and if a rat failed to find the escape box within the allotted time, they were gently guided into the box, which was then covered. Rats were allowed to remain in the escape box for 15 s before being returned to their home cage. On the fifth day of the Barnes maze, rats were placed at the center of the maze facing away from the escape box for a probe trial to ensure they learned the task. The procedures were the same from testing. After the probe trial, the escape box was rotated 180°, but none of the visuospatial cues were moved. For the 3 reversal learning trials with 30 min inter-trial intervals, rats were given 150 s to locate the new location of the escape box and were gently guided in if they failed to find the escape box and allowed to remain in the box for 15 s. Figure 2. Barnes maze. Starting locations on the Barnes maze for training and reversal learning trials. (A) Rats underwent 4 trails/day during training. (B) On the testing day, rats went through a probe trial to assess task acquisition and 3 reversal learning trials to assess behavioral flexibility. 2.3. Statistical Analysis Figure 2. Barnes maze. Starting locations on the Barnes maze for training and reversal learning trials. (A) Rats underwent 4 trails/day during training. (B) On the testing day, rats went through a probe trial to assess task acquisition and 3 reversal learning trials to assess behavioral flexibility. On the fifth day of the Barnes maze, rats were placed at the center of the maze facing away from the escape box for a probe trial to ensure they learned the task. The procedures were the same from testing. After the probe trial, the escape box was rotated 180◦, but none of the visuospatial cues were moved. For the 3 reversal learning trials with 30 min inter-trial intervals, rats were given 150 s to locate Nutrients 2020, 12, 2721 6 of 22 the new location of the escape box and were gently guided in if they failed to find the escape box and allowed to remain in the box for 15 s. 2.3. Statistical Analysis Statistical analyses were performed using Statistica 13.3 (TIBCO, Palo Alto, CA, USA). Data are presented as the mean ± standard error of the mean (SEM). Post hoc Fisher’s LSD tests were performed when significant main effects or interactions were identified. For the WR and two-diet choice portion, raw data included weekly averages of body weight, energy intake, and running activity. These measures were analyzed separately using 3-way mixed model ANOVAs with sex (male vs. female) and exercise (Sed vs. WR) as between-subject factors and time (6 weekly averages) as the within-subject factor. Diet choice was analyzed using a 4-way mixed model ANOVA with sex and exercise as the between-subject factors and diet (chow vs. HF) and time (6 weekly averages) as the within-subject factors. Raw data from OGTT included plasma glucose and insulin measurements. Baseline OGTT and post-HF OGTT results were analyzed separately using a 3-way mixed model ANOVAs with sex and exercise as between-subject factors, and time (0, 15, 30, 60, and 120 min) as the within-subjects factor. Glucose and insulin area under curve (AUC) results were analyzed separately using a 2-way mixed model ANOVA with sex and exercise as the between-subject factors and time (baseline vs. post-HF) as the within-subjects factor. Two measures of insulin sensitivity were first calculated using HOMA-IR and ISI0,120 and then analyzed separately using a 2-way mixed model ANOVA with sex and exercise as the between-subject factors and time (baseline vs. post-HF) as the within-subject factor. A 2-way ANOVA with sex and exercise as the between-subject factors was performed to analyze trunk plasma insulin levels at the termination of the experiment. In addition, separate correlation analyses were performed to determine if there was an association between glucose AUC during OGTT, insulin AUC during OGTT, and trunk plasma insulin levels with average HF diet preference ratio. For the Barnes maze, the training days were manually scored for latency to enter the escape box and errors. The test day was manually scored for the same measures. An error was counted each time the rat checked a hole other than the one leading to the escape box by poking its nose into the hole. At least two individuals video scored the training and testing portion of the Barnes maze for all rats. Training and testing data were analyzed using a 3-way mixed model ANOVA with sex and exercise as between-subject factors and trial (daily average) as the within-subject factor. 3. Results 3.1. Wheel Running and Two-Diet Choice Sedentary rats decreased HF diet intake and increased chow intake across time whereas WR rats expressed the opposite diet choice pattern (time × diet × exercise F (5,220) = 37.35, p < 0.001; Figure 3A–D). Upon initial access to the two-diet choice, Sed rats showed extreme preference for the HF diet whereas WR rats avoided it. Subsequently, these opposite diet choice patterns were reflected as Sed rats decreased and WR rats increased HF diet preference over time (time × exercise F (5,220) = 42.28, p < 0.001; Figure 3E,F). Furthermore, HF diet preference did not appear to be influenced by sex, i.e., all WR females and 12 out of 14 WR males reversed HF diet avoidance (sex F (1,44) = 1.63, p > 0.20). However, when examining the average ratios of HF diet preference across the duration of two-diet choice, nine out of 14 WR males had a HF diet preference ratio < 0.5, which indicates that they preferred HF to chow diet for less than half of the six weeks choice period. In addition, WR females reversed HF diet avoidance earlier than males and expressed greater preference for HF diet (time × sex × exercise F (5,220) = 3.92, p < 0.05). Nutrients 2020, 12, 2721 7 of 22 Nutrients 2020, 12, x FOR PEER REVIEW 7 of 22 Figure 3. Diet choice and HF diet preference ratios in males (M, left) and females (F, right). WR rats expressed opposite diet choice patterns from their Sed controls. The vertical and horizontal dashed red lines denote the start of the two-diet choice and WR period and preference for HF diet where any value greater than 0.5 indicates a preference for HF diet, respectively. (A,B) Sedentary male rats maintained higher intake of HF than chow diet throughout the experiment whereas there was a two- week period in which Sed females did not show a preference for either diet. * chow vs. HF, p < 0.05. (C,D) Both male and female WR rats increased HF diet intake across time. The reversal of HF diet avoidance occurred earlier in females than males. * chow vs. HF, p < 0.05. (E,F) HF diet preference went in opposite directions among Sed and WR rats in both sexes. * Sed vs. WR, p < 0.05. Figure 3. Diet choice and HF diet preference ratios in males (M, left) and females (F, right). WR rats expressed opposite diet choice patterns from their Sed controls. The vertical and horizontal dashed red lines denote the start of the two-diet choice and WR period and preference for HF diet where any value greater than 0.5 indicates a preference for HF diet, respectively. (A,B) Sedentary male rats maintained higher intake of HF than chow diet throughout the experiment whereas there was a two-week period in which Sed females did not show a preference for either diet. * chow vs. HF, p < 0.05. (C,D) Both male and female WR rats increased HF diet intake across time. The reversal of HF diet avoidance occurred earlier in females than males. * chow vs. HF, p < 0.05. (E,F) HF diet preference went in opposite directions among Sed and WR rats in both sexes. * Sed vs. WR, p < 0.05. Nutrients 2020, 12, 2721 8 of 22 3.2. Running Activity and Energy Intake Females ran more than males, and both sexes showed an inverted-U trend in running activity where running activity peaked and then decreased to baseline levels (time × sex F (5,125) = 9.73, p < 0.001; Figure 4A). There were sex-specific adaptations in energy intake to exercise (sex × exercise F (1,440) = 28.26, p < 0.001) across time (time × sex × exercise F (5,220) = 4.97, p < 0.001; Figure 4B). WR led to an initial decrease in total energy intake in males after which they increased food intake, but total energy intake was not different among Sed and WR males (post hoc p > 0.15). Conversely, female WR rats increased their total energy intake earlier than males and had significantly higher energy intake than their Sed counterparts (post hoc p < 0.001). Nutrients 2020, 12, x FOR PEER REVIEW 8 of 22 3.2. Running Activity and Energy Intake Females ran more than males, and both sexes showed an inverted-U trend in running activity where running activity peaked and then decreased to baseline levels (time × sex F (5,125) = 9.73, p < 0.001; Figure 4A). There were sex-specific adaptations in energy intake to exercise (sex × exercise F (1,440) = 28.26, p < 0.001) across time (time × sex × exercise F (5,220) = 4.97, p < 0.001; Figure 4B). WR led to an initial decrease in total energy intake in males after which they increased food intake, but total energy intake was not different among Sed and WR males (post hoc p > 0.15). Conversely, female WR rats increased their total energy intake earlier than males and had significantly higher energy intake than their Sed counterparts (post hoc p < 0.001). Figure 4. Running activity and total energy intake. (A) Female rats ran more than males, and both sexes showed and inverted-U trend in running activity. * Male vs. female, p < 0.05. (B) Female, but not male, running rats had higher energy intake than their Sed counterparts. * Sed vs. WR, p < 0.05. 3.3. Body Weight and Adiposity Exercise-mediated changes in total daily energy intake resulted in suppressed body weight gain in both males and females (exercise F (1,44) = 29.16, p < 0.001). Although exercise suppressed weight gain in females (Figure 5A), the difference in percent weight gain between Sed and WR rats at the end of the experiment was 10% in males and only 2% in females. The difference in body weight was reflected in fat composition. Exercise led to decreased retroperitoneal and mesenteric fat (exercise F (1,44) = 11.62 and 10.47, respectively, both p < 0.01) in both sexes (sex x exercise F (1,44) = 1.26 and 3.68, respectively, both p > 0.06; Figure 5B). Although the sex x exercise interaction did not reach statistical significance in the mesenteric fat pad (p = 0.061), post hoc tests indicate that the effect of exercise was driven by males. For the gonadal fat pad, there was a sex x exercise interaction where exercise resulted in a loss of gonadal fat only in males (F (1,44) = 8.16, p < 0.01; post hoc male Sed vs. WR p < 0.01 and female Sed vs. WR p > 0.49). Figure 4. Running activity and total energy intake. (A) Female rats ran more than males, and both sexes showed and inverted-U trend in running activity. * Male vs. female, p < 0.05. (B) Female, but not male, running rats had higher energy intake than their Sed counterparts. * Sed vs. WR, p < 0.05. 3.3. Body Weight and Adiposity Exercise-mediated changes in total daily energy intake resulted in suppressed body weight gain in both males and females (exercise F (1,44) = 29.16, p < 0.001). Although exercise suppressed weight gain in females (Figure 5A), the difference in percent weight gain between Sed and WR rats at the end of the experiment was 10% in males and only 2% in females. The difference in body weight was reflected in fat composition. Exercise led to decreased retroperitoneal and mesenteric fat (exercise F (1,44) = 11.62 and 10.47, respectively, both p < 0.01) in both sexes (sex x exercise F (1,44) = 1.26 and 3.68, respectively, both p > 0.06; Figure 5B). Although the sex x exercise interaction did not reach statistical significance in the mesenteric fat pad (p = 0.061), post hoc tests indicate that the effect of exercise was driven by males. For the gonadal fat pad, there was a sex x exercise interaction where exercise resulted in a loss of gonadal fat only in males (F (1,44) = 8.16, p < 0.01; post hoc male Sed vs. WR p < 0.01 and female Sed vs. WR p > 0.49). Nutrients 2020, 12, 2721 9 of 22 Nutrients 2020, 12, x FOR PEER REVIEW 9 of 22 Figure 5. Body weight and fat composition. (A) Exercise suppressed body weight in both sexes; however, this effect was more pronounced in males. * Sed vs. WR, p < 0.05. (B) Exercise led to decreased retroperitoneal, mesenteric, and gonadal fat in males and decreased only retroperitoneal adiposity in females. * Sed vs. WR, p < 0.05. 3.4. Oral Glucose Tolerance Test (OGTT) At chow only baseline (BL), there were no group differences in glucose clearance following an oral glucose challenge (time × sex × exercise F (4,176) = 1.78, p > 0.32; Figure 6A). Following six weeks of chronic HF feeding, females had higher fasting blood glucose levels than males (time × sex F (4,172) = 5.26, p < 0.001; post hoc 0 min males vs. females 103.33 vs. 116.61 mg/dL, p < 0.05; Figure 6B). However, blood glucose levels returned to pre-glucose challenge levels in females, but not males (post hoc 0 min vs. 120 min female p > 0.32 and male p < 0.001). Analysis of area under curve (AUC) of blood glucose during the baseline and post-HF diet OGTT indicated that exercise decreased glucose AUC in males, but not females (time × sex × exercise F (1,43) = 6.81, p < 0.05; post hoc male WR BL vs. HF p < 0.01 and female WR BL vs. HF p > 0.12; Figure 6C). Prior to the WR and two-diet choice experimental period, there were no baseline differences in insulin levels during an OGTT (time × sex × exercise F (4,156) = 0.85, p > 0.49; Figure 6D). During the post-HF diet OGTT, there was a trend for exercise to decrease plasma insulin that appeared to be driven by males (sex × exercise F (1,41) = 2.87, p = 0.09; Figure 6E). A one-way ANOVA revealed a group difference at 0 min, such that exercise resulted in lower plasma insulin levels only in males following long-term HF diet exposure (group F (1,42) = 4.05, p < 0.05; post hoc male Sed vs. WR p < 0.01 and female Sed vs. WR p > 0.13). Following an oral glucose challenge, there was no exercise effect in plasma insulin levels across different time points (time × exercise F (4,164) = 1.61, p > 0.17). On average, males had higher insulin AUC levels than females (sex F (1,35) = 4.41, p < 0.05; Figure 6F). There was also a time × sex × exercise effect (F (1,35) = 5.29, p < 0.05) where exercise decreased insulin AUC in males but not females. Chronic HF feeding resulted in higher insulin AUC than baseline levels in Sed males (post hoc M Sed BL vs. HF p < 0.001), and exercise suppressed this increase (post hoc M WR BL vs. HF p > 0.06). Male WR rats had lower insulin AUC post-HF diet exposure than their Sed counterparts (post hoc HF M Sed vs. M WR p < 0.001). In females, however, exercise did not suppress the amount of plasma insulin needed to clear the same dose of glucose (post hoc F Sed and F WR BL vs. HF both p < 0.01 and HF F Sed vs. F WR p > 0.98). Figure 5. Body weight and fat composition. (A) Exercise suppressed body weight in both sexes; however, this effect was more pronounced in males. * Sed vs. WR, p < 0.05. (B) Exercise led to decreased retroperitoneal, mesenteric, and gonadal fat in males and decreased only retroperitoneal adiposity in females. * Sed vs. WR, p < 0.05. 3.4. Oral Glucose Tolerance Test (OGTT) At chow only baseline (BL), there were no group differences in glucose clearance following an oral glucose challenge (time × sex × exercise F (4,176) = 1.78, p > 0.32; Figure 6A). Following six weeks of chronic HF feeding, females had higher fasting blood glucose levels than males (time × sex F (4,172) = 5.26, p < 0.001; post hoc 0 min males vs. females 103.33 vs. 116.61 mg/dL, p < 0.05; Figure 6B). However, blood glucose levels returned to pre-glucose challenge levels in females, but not males (post hoc 0 min vs. 120 min female p > 0.32 and male p < 0.001). Analysis of area under curve (AUC) of blood glucose during the baseline and post-HF diet OGTT indicated that exercise decreased glucose AUC in males, but not females (time × sex × exercise F (1,43) = 6.81, p < 0.05; post hoc male WR BL vs. HF p < 0.01 and female WR BL vs. HF p > 0.12; Figure 6C). Prior to the WR and two-diet choice experimental period, there were no baseline differences in insulin levels during an OGTT (time × sex × exercise F (4,156) = 0.85, p > 0.49; Figure 6D). During the post-HF diet OGTT, there was a trend for exercise to decrease plasma insulin that appeared to be driven by males (sex × exercise F (1,41) = 2.87, p = 0.09; Figure 6E). A one-way ANOVA revealed a group difference at 0 min, such that exercise resulted in lower plasma insulin levels only in males following long-term HF diet exposure (group F (1,42) = 4.05, p < 0.05; post hoc male Sed vs. WR p < 0.01 and female Sed vs. WR p > 0.13). Following an oral glucose challenge, there was no exercise effect in plasma insulin levels across different time points (time × exercise F (4,164) = 1.61, p > 0.17). On average, males had higher insulin AUC levels than females (sex F (1,35) = 4.41, p < 0.05; Figure 6F). There was also a time × sex × exercise effect (F (1,35) = 5.29, p < 0.05) where exercise decreased insulin AUC in males but not females. Chronic HF feeding resulted in higher insulin AUC than baseline levels in Sed males (post hoc M Sed BL vs. HF p < 0.001), and exercise suppressed this increase (post hoc M WR BL vs. HF p > 0.06). Male WR rats had lower insulin AUC post-HF diet exposure than their Sed counterparts (post hoc HF M Sed vs. M WR p < 0.001). In females, however, exercise did not suppress the amount of plasma insulin needed to clear the same dose of glucose (post hoc F Sed and F WR BL vs. HF both p < 0.01 and HF F Sed vs. F WR p > 0.98). Nutrients 2020, 12, 2721 10 of 22 Nutrients 2020, 12, x FOR PEER REVIEW 10 of 22 Figure 6. Blood glucose (top row) and plasma insulin (bottom row) results from an OGTT at baseline (BL) and post-HF diet exposure. (A) There were no group differences in blood glucose at BL. (B) Blood glucose levels following an oral glucose challenge returned to fasting levels faster in females than males. (C) Decreased glucose clearance, indicated by AUC, from BL occurred in male WR and female Sed rats. * BL vs. HF, p < 0.05. (D) There were no group differences in plasma insulin at BL. (E) Following HF diet exposure, exercise decreased fasting plasma insulin levels in males but not females. @ M Sed vs. WR, p < 0.05. (F) Male WR rats had lower insulin AUC than their Sed counterparts. In contrast, both Sed and WR females had higher insulin AUC post-HF exposure than at chow BL. * BL vs. HF, p < 0.05, @ HF M Sed vs. WR, p < 0.05. Male Sed rats had impaired hepatic insulin sensitivity following HF feeding (time × sex × exercise F(1,40) = 7.73, p < 0.01; post hoc M Sed BL vs. post-HF p < 0.0001) as evidenced by increased HOMA-IR index (Table 2). Exercise protected against this detrimental metabolic effect of HF diet in male WR rats by attenuating increases in insulin resistance (post hoc WR BL vs. post-HF p > 0.18 and post-HF Sed vs. WR, p < 0.01). In females, both Sed and WR rats had evidence of insulin resistance after long-term HF diet preference (post hoc Sed and WR BL vs. HF, both p < 0.05) and exercise did not have the same protective effect as seen in males (post-hoc post-HF F Sed vs. F WR p > 0.08). Peripheral insulin sensitivity was analyzed using ISI0,120 and sex differences in the protective effect of exercise failed to reach statistical significance (time × sex × exercise F(1,43) = 2.29, p > 0.13). However, a priori t-tests revealed that post-HF feeding, male Sed rats had lower ISI0,120 than their WR counterparts, indicating reduced insulin sensitivity in the Sed but not WR group (t(10,14) = −2.17, p < 0.05). This difference between Sed and WR groups was absent in females (t(10,13) = −0.21, p > 0.83). Table 2. Hepatic and peripheral indices of insulin sensitivity. Exercise protected against the development of insulin resistance by HF diet to a greater extent in males than females. HOMA-IR: Homeostatic assessment of insulin resistance; ISI0,120: Gutt’s insulin sensitivity index (mg × L2 × mmol−1 × mU−1); *: Baseline vs. Post-HF, p < 0.05; ^: Sed vs. WR, p < 0.05. Data are represented as the mean ± SEM. Insulin Sensitivity Group HOMA-IR ISI0,120 Baseline Post-HF Baseline Post-HF Male Sed 1.22 ± 0.31 * 5.01 ± 0.99 ^ 0.79 ± 0.04 * 0.63 ± 0.02 ^ Male WR 1.34 ± 0.30 2.15 ± 0.45 0.74 ± 0.03 0.71 ± 0.03 Figure 6. Blood glucose (top row) and plasma insulin (bottom row) results from an OGTT at baseline (BL) and post-HF diet exposure. (A) There were no group differences in blood glucose at BL. (B) Blood glucose levels following an oral glucose challenge returned to fasting levels faster in females than males. (C) Decreased glucose clearance, indicated by AUC (area under curve), from BL occurred in male WR and female Sed rats. * BL vs. HF, p < 0.05. (D) There were no group differences in plasma insulin at BL. (E) Following HF diet exposure, exercise decreased fasting plasma insulin levels in males but not females. @ M Sed vs. WR, p < 0.05. (F) Male WR rats had lower insulin AUC than their Sed counterparts. In contrast, both Sed and WR females had higher insulin AUC post-HF exposure than at chow BL. * BL vs. HF, p < 0.05, @ HF M Sed vs. WR, p < 0.05. Male Sed rats had impaired hepatic insulin sensitivity following HF feeding (time × sex × exercise F (1,40) = 7.73, p < 0.01; post hoc M Sed BL vs. post-HF p < 0.0001) as evidenced by increased HOMA-IR index (Table 2). Exercise protected against this detrimental metabolic effect of HF diet in male WR rats by attenuating increases in insulin resistance (post hoc WR BL vs. post-HF p > 0.18 and post-HF Sed vs. WR, p < 0.01). In females, both Sed and WR rats had evidence of insulin resistance after long-term HF diet preference (post hoc Sed and WR BL vs. HF, both p < 0.05) and exercise did not have the same protective effect as seen in males (post-hoc post-HF F Sed vs. F WR p > 0.08). Peripheral insulin sensitivity was analyzed using ISI0,120 and sex differences in the protective effect of exercise failed to reach statistical significance (time × sex × exercise F (1,43) = 2.29, p > 0.13). However, a priori t-tests revealed that post-HF feeding, male Sed rats had lower ISI0,120 than their WR counterparts, indicating reduced insulin sensitivity in the Sed but not WR group (t(10,14) = −2.17, p < 0.05). This difference between Sed and WR groups was absent in females (t(10,13) = −0.21, p > 0.83). Analysis of trunk plasma insulin after 6 weeks of HF feeding revealed that WR females had higher plasma insulin than their Sed counterparts (M Sed 1.04 ± 0.10, M WR 0.54 ± 0.05, F Sed 0.44 ± 0.06, and F WR 0.69 ± 0.06 ng/mL; sex × exercise F (1,44) = 31.40, p < 0.001; post hoc female Sed vs. WR p < 0.05) whereas the opposite pattern was observed in Sed and WR males (post hoc p < 0.001). Moreover, a regression analysis revealed a positive correlation between average ratios of HF diet preference and plasma insulin levels at sacrifice in males (F (1,22) = 7.72, R = 0.51, p < 0.01; Figure 7A) but not females (F (1,22) = 0.95, R = 0.01 p > 0.94; Figure 7B). No such correlation was found between average ratios of HF diet preference and either glucose or insulin AUC. Nutrients 2020, 12, 2721 11 of 22 Table 2. Hepatic and peripheral indices of insulin sensitivity. Exercise protected against the development of insulin resistance by HF diet to a greater extent in males than females. Group Insulin Sensitivity HOMA-IR ISI0,120 Baseline Post-HF Baseline Post-HF Male Sed 1.22 ± 0.31 * 5.01 ± 0.99 ˆ 0.79 ± 0.04 * 0.63 ± 0.02 ˆ Male WR 1.34 ± 0.30 2.15 ± 0.45 0.74 ± 0.03 0.71 ± 0.03 Female Sed 1.43 ± 0.38 * 3.15 ± 0.47 0.71 ±0.03 0.64 ± 0.02 Female WR 2.25 ± 0.49 * 4.63 ± 0.81 0.70 ± 0.03 0.65 ± 0.02 HOMA-IR: Homeostatic assessment of insulin resistance; ISI0,120: Gutt’s insulin sensitivity index (mg × L2 × mmol−1 × mU−1); *: Baseline vs. Post-HF, p < 0.05; ˆ: Sed vs. WR, p < 0.05. Data are represented as the mean ± SEM. Nutrients 2020, 12, x FOR PEER REVIEW 11 of 22 Female Sed 1.43 ± 0.38 * 3.15 ± 0.47 0.71 ±0.03 0.64 ± 0.02 Female WR 2.25 ± 0.49 * 4.63 ± 0.81 0.70 ± 0.03 0.65 ± 0.02 Analysis of trunk plasma insulin after 6 weeks of HF feeding revealed that WR females had higher plasma insulin than their Sed counterparts (M Sed 1.04 ± 0.10, M WR 0.54 ± 0.05, F Sed 0.44 ± 0.06, and F WR 0.69 ± 0.06 ng/mL; sex × exercise F (1,44) = 31.40, p < 0.001; post hoc female Sed vs. WR p < 0.05) whereas the opposite pattern was observed in Sed and WR males (post hoc p < 0.001). Moreover, a regression analysis revealed a positive correlation between average ratios of HF diet preference and plasma insulin levels at sacrifice in males (F (1,22) = 7.72, R = 0.51, p < 0.01; Figure 7A) but not females (F (1,22) = 0.95, R = 0.01 p > 0.94; Figure 7B). No such correlation was found between average ratios of HF diet preference and either glucose or insulin AUC. Figure 7. Correlation between trunk plasma insulin levels at the end of the experiment and the average ratios of HF diet preference. (A) There was a moderate, positive correlation between HF diet preference and trunk plasma insulin in males. (B) There was no relationship between HF diet preference and trunk plasma insulin levels in females. 3.5. Barnes Maze During training, there was a sex difference in latency whereby males were slower to locate the escape box than females (sex and trial × sex F (1,43) = 66.58 and F (3,129) = 7.59, respectively, both p < 0.001; Figure 8A,C). In rats of both sexes, exercise led to decreased latency to locate the escape box (trial × exercise and trial × sex × exercise F (3,129) = 4.23 and 2.00, p < 0.01 and p > 0.11, respectively). Although an effect of trial by exercise interaction on errors committed across training days reached statistical significance (F (3,129) = 3.37, p < 0.05), post hoc tests revealed no specific group differences on any given day. Thus, exercise slightly reduced the numbers of errors made during learning on the Barnes maze in both sexes (trial × sex × exercise F (3,129) = 0.65, p > 0.58; Figure 8B,D). On average, male rats made more errors than females during training (sex F (1,43) = 5.25, p < 0.05). A factorial ANOVA on the probe trial revealed that there was no effect of sex or exercise on task acquisition in regards to latency (sex × exercise F (1,43) = 1.80, p > 0.18) and errors made (F (1,43) = 1.67, p > 0.20). There was also no effect of sex or exercise on errors made or latency to locate the escape box during reversal learning (sex × exercise F (1,43) = 0.79 and 0.15, respectively, both p > 0.37). However, when the percent increase in errors made between the probe and reversal trials was analyzed using a factorial ANOVA, there was a sex × exercise effect (F (1,34) = 6.48, p < 0.05). Post hoc analyses revealed that female WR rats increased more errors than their Sed counterparts (post hoc p < 0.05) whereas this effect was not seen in male rats (post hoc male Sed vs. WR p > 0.10). Figure 7. Correlation between trunk plasma insulin levels at the end of the experiment and the average ratios of HF diet preference. (A) There was a moderate, positive correlation between HF diet preference and trunk plasma insulin in males. (B) There was no relationship between HF diet preference and trunk plasma insulin levels in females. 3.5. Barnes Maze During training, there was a sex difference in latency whereby males were slower to locate the escape box than females (sex and trial × sex F (1,43) = 66.58 and F (3,129) = 7.59, respectively, both p < 0.001; Figure 8A,C). In rats of both sexes, exercise led to decreased latency to locate the escape box (trial × exercise and trial × sex × exercise F (3,129) = 4.23 and 2.00, p < 0.01 and p > 0.11, respectively). Although an effect of trial by exercise interaction on errors committed across training days reached statistical significance (F (3,129) = 3.37, p < 0.05), post hoc tests revealed no specific group differences on any given day. Thus, exercise slightly reduced the numbers of errors made during learning on the Barnes maze in both sexes (trial × sex × exercise F (3,129) = 0.65, p > 0.58; Figure 8B,D). On average, male rats made more errors than females during training (sex F (1,43) = 5.25, p < 0.05). A factorial ANOVA on the probe trial revealed that there was no effect of sex or exercise on task acquisition in regards to latency (sex × exercise F (1,43) = 1.80, p > 0.18) and errors made (F (1,43) = 1.67, p > 0.20). There was also no effect of sex or exercise on errors made or latency to locate the escape box during reversal learning (sex × exercise F (1,43) = 0.79 and 0.15, respectively, both p > 0.37). However, when the percent increase in errors made between the probe and reversal trials was analyzed using a factorial ANOVA, there was a sex × exercise effect (F (1,34) = 6.48, p < 0.05). Post hoc analyses revealed Nutrients 2020, 12, 2721 12 of 22 that female WR rats increased more errors than their Sed counterparts (post hoc p < 0.05) whereas this effect was not seen in male rats (post hoc male Sed vs. WR p > 0.10). Both male and female rats used a non-spatial serial search strategy (Video S1), e.g., a clockwise or counterclockwise sequential search, rather than a direct search strategy where the rats utilize spatial cues, e.g., signs in the testing room to find the escape box. Despite not using the visual cues, the decreased latency and errors indicated that all rats learned the task (Video S2). While Sprague-Dawley rats have poor visual acuity biasing them towards using a serial search strategy [78], visual acuity has not been correlated with deficits in learning and memory issues in mice tested on the Barnes maze [86]. Moreover, our results for latency and errors made are comparable to what has been reported in the literature [78,87]. Nutrients 2020, 12, x FOR PEER REVIEW 12 of 22 Both male and female rats used a non-spatial serial search strategy (Video S1), e.g., a clockwise or counterclockwise sequential search, rather than a direct search strategy where the rats utilize spatial cues, e.g., signs in the testing room to find the escape box. Despite not using the visual cues, the decreased latency and errors indicated that all rats learned the task (Video S2). While Sprague- Dawley rats have poor visual acuity biasing them towards using a serial search strategy [78], visual acuity has not been correlated with deficits in learning and memory issues in mice tested on the Barnes maze [86]. Moreover, our results for latency and errors made are comparable to what has been reported in the literature [78,87]. Figure 8. Barnes maze results for males (M, top) and females (F, bottom). (A,C) Both male and female rats decreased latency to locate the escape box during training. On average, females had shorter latencies to find the escape box than males across training days. (B,D) All groups committed fewer errors when searching for the escape box across training days. 4. Discussion Currently, it is unclear whether exercise has a similar efficacy at reversing the adverse metabolic and cognitive effects of HF preference and intake in rats of both sexes. To address this, we used a long-term two-diet choice and WR model to examine the relationship between preference for HF diet and the detrimental metabolic and cognitive outcomes associated with chronic HF feeding, and whether exercise has the ability to attenuate these negative effects. We found that both male and female WR rats recovered from their initial running-induced HF diet avoidance and increased both HF diet intake and preference across time (Figure 3C,D). Exercise had a protective effect in males, but Figure 8. Barnes maze results for males (M, top) and females (F, bottom). (A,C) Both male and female rats decreased latency to locate the escape box during training. On average, females had shorter latencies to find the escape box than males across training days. (B,D) All groups committed fewer errors when searching for the escape box across training days. 4. Discussion Currently, it is unclear whether exercise has a similar efficacy at reversing the adverse metabolic and cognitive effects of HF preference and intake in rats of both sexes. To address this, we used a long-term two-diet choice and WR model to examine the relationship between preference for HF diet and the detrimental metabolic and cognitive outcomes associated with chronic HF feeding, Nutrients 2020, 12, 2721 13 of 22 and whether exercise has the ability to attenuate these negative effects. We found that both male and female WR rats recovered from their initial running-induced HF diet avoidance and increased both HF diet intake and preference across time (Figure 3C,D). Exercise had a protective effect in males, but not females, on HF-mediated weight gain and adiposity (Figure 5) and metabolic dysfunction (Figure 6C,F). Moreover, the positive association between preference for HF diet and trunk plasma insulin was only seen in males (Figure 7A). Wheel running rats had significantly faster escape latencies and made slightly fewer errors than their sedentary counterparts in both sexes across training days, suggesting improved learning and memory through exercise (Figure 8). Although our interpretation of behavioral flexibility was limited due to the utilization of a non-spatial search strategy by the rats (Video S1), female WR rats were the only group to increase the number of errors made between the probe and reversal learning trials compared to their Sed controls. Taken together, our results suggest that exercise-mediated changes in HF diet preference lead to sex-specific effects in regards to the protective effect of exercise on both peripheral metabolic function and cognitive performance. The opposite diet choice patterns observed among Sed and WR rats (Figure 3A–D) may be due to the increased metabolic requirement from exercise to maintain energy balance where the HF diet is a more efficient fuel source than the standard chow diet, which is higher in carbohydrates than fats [88,89]. Indeed, human studies have shown that there is a crossover effect during which the ratio of lipolysis to carbohydrate oxidation during submaximal and endurance exercise increases [90–92]. While this crossover effect is influenced by exercise duration and intensity, it may partially contribute to differences in macronutrient preference among sedentary and physically active individuals. Our previous short-term studies with the same paradigm of wheel running and two-diet choice revealed that the majority of male rats express persistent HF diet avoidance whereas the majority of females reverse HF diet avoidance [80,93–95]. These results are consistent with the report that estradiol enhances lipid metabolism during exercise in rats [96]. Furthermore, results with respiratory exchange ratio as a measure of substrate utilization from human studies using indirect calorimetry suggest that compared to men, women utilize more fat as the fuel source as a result of long-term exercise [90,91,97,98]. A direct assessment of fuel oxidation will be necessary to support our hypothesis that sex differences in substrate utilization may contribute to differences in running-associated macronutrient preference. When the choice duration was extended, both male and female WR groups preferred the HF diet by the end of the six-week period (Figure 3E,F). It is unclear whether this increase in fat preference would occur in humans if the behavior can be examined without the influence of the cognitive component of making healthier food choices in subjects who incorporate regular exercise as a lifestyle. Nevertheless, carbohydrate metabolism is positively correlated with exercise intensity in humans [99] whereas fat oxidation is more likely to occur during low intensity exercise, especially when prolonged [100]. Thus, this shift in fat preference may be a compensatory result of increased energy requirement from long-term aerobic exercise where carbohydrates are no longer the most efficient fuel substrate. The addition of groups undergoing different types of exercise (e.g., strength, treadmill, swimming, etc.) for different lengths of time would provide additional evidence for the effect of exercise on energy intake and macronutrient preference if results are consistent. Although both male and female WR rats reversed their initial avoidance for HF diet, there was a sex difference in which the reversal of HF diet avoidance occurred with females reversing earlier than males. One potential explanation for this sex difference is that females are more prone to hedonic and binge eating than males [101,102] and their feeding behavior appears to be driven by palatability rather than physiological hunger or metabolic state [103,104]. Both of these factors could act together and exacerbate the development of obesity [56,57], which is more prevalent in females [105]. In addition, females have higher reward sensitivity than males [106,107] which may predict decreased restraint of fat intake [108]. HF diet is highly palatable and can stimulate eating in the absence of hunger by acting on the reward system [104], potentially leading to overeating. Although WR is naturally rewarding for rodents [109], it may not be a sufficient substitute for the reinforcing effects of the palatable HF diet for females [110]. In support of this, studies have shown that male rats are more Nutrients 2020, 12, 2721 14 of 22 responsive to the reinforcing effects of voluntary WR to attenuate seeking of drugs of abuse [111,112]. The high sensitivity to WR reinforcement could facilitate males’ ability to maintain lower preference for HF diet for a longer duration than females that are more sensitive to the reinforcing effects of diet palatability [110]. Sex differences in adapting to the increased energy requirement of exercise led to differential efficacies for exercise to attenuate HF-mediated insults on metabolic function in male and female rats. Consistent with the literature [21,23,113], females ran more than males (Figure 4A) and compensated for the increased energy expenditure by increasing total energy intake earlier than males [22,114]. This may have led to the limited effect of exercise on suppressing body weight and adiposity [19–22] and attenuating HF mediated metabolic dysregulation in females (Figure 6). Here, we report that exercise suppressed HF diet preference, total energy intake, body weight, and adiposity, and improved glucose metabolism to a greater degree in males than females. This is consistent with the consensus in the literature stating that males are more responsive to the beneficial effects of exercise, resulting in improved glucose tolerance and insulin sensitivity [74–76]. Thus, it appears that exercise has a protective effect on insulin sensitivity in males despite increased HF diet intake [69,70,73,115–120]. We also found a positive association between HF diet preference and insulin levels in males but not females (Figure 7). The greater protective effect of exercise on peripheral metabolic function in males may be mediated by two effects acting in concert: (1) a slower compensatory response to the increased energy expenditure from exercise [114] and (2) maintenance of a lower preference for HF diet for a longer duration of time compared to females. Our results suggest that that without a concurrent decrease in body weight, adiposity, and HF diet preference, exercise has a limited effect on significantly improving peripheral insulin resistance during chronic access to HF diet [16,36,37]. Female rats had increased HOMA-IR (Table 2) above the 2.60 cutoff [121] for evidence of hepatic insulin resistance after long-term HF feeding regardless of the opportunity to exercise. In contrast, exercise appeared to protect against the development of insulin resistance in males. The results of ISI0,120 also suggest that male WR rats were the only group that maintained insulin sensitivity after six weeks of exposure to HF diet. Although higher HF diet and total energy intake in WR females may play a role in these observed sex differences and additional pair-fed groups will be needed to assess such possibility in future studies, our results reveal that exercise produces more protective effects against insulin resistance in males than females by reducing HF diet preference and consumption. Consequently, an optimal treatment for weight loss and improving insulin sensitivity for females would be a combination of diet and exercise [122]. The improved performance during training on the Barnes maze in WR rats of both sexes relative to their Sed counterparts suggests that exercise can be protective against insults to cognitive behavior from chronic HF diet consumption independent of sex (Figure 8). This aligns with rodent literature linking HF feeding to deficits in cognitive behaviors [15,42,53,54], which exercise can reverse [44,48–52]. In contrast to reports that male and female rats performed similarly on the Barnes maze [123,124], we found that females had faster latencies than males to locate the escape box. Rather than enhanced learning, this effect could be a result of hormonal/estrous status or higher general locomotor activity exhibited by females [125,126]. Although our results conflict with previous literature reporting either a male advantage or no difference in performance between males and females, sex differences have not been consistently reported in regards to learning visuospatial tasks [86,127–129]. Moreover, task performance is influenced by a variety of factors including task design, species, strain, hormonal status, stress, and age [130]. To our knowledge, there have been a limited number of studies that investigate sex differences in cognitive behavior using the Barnes maze in rats [127,128,131] with the Morris water maze being the more popular task [130]. More standardized research is necessary to draw firm conclusions on subtle sex differences in cognitive and spatial ability given how the Barnes maze is sensitive to a variety of conditions that may influence task performance and subsequent interpretation of the data. Nutrients 2020, 12, 2721 15 of 22 Despite similar search strategies and learning outcomes, there was evidence of sex-specific effects of running on reversal learning on the Barnes maze. The non-spatial serial search strategy (Video S1) is not uncommon in Sprague-Dawley rats [128] that have lower visual acuity than other strains (e.g., Long-Evans and Wistar) [128,132] where using a serial search strategy may be more efficient. Notably, the increase in errors made from the probe to reversal learning trials was higher in WR females compared to their Sed counterparts, whereas there was no such group difference in males. This sex-specific effect may be interpreted as WR females having the worst behavioral flexibility among all groups, which conflicts with the finding that PFC-mediated deficits in behavioral flexibility are more pronounced in males than females following HF feeding [65–67]. However, our interpretation of the data is limited such that (1) rats did not appear to use spatial cues to locate the escape box and (2) learning without utilizing spatial cues may render weak task acquisition and as such, poses a confound for the analysis of potentially increased errors during the reversal learning trials. Thus, an alternate interpretation to our result that only female WR rats significantly increased their errors during reversal learning may be that this was the only group that acquired the original task and could exhibit deficits in behavioral flexibility. Nevertheless, while adjustments can be made to promote the use of spatial cues (e.g., moving cues closer to the maze, adding an aversive stimulus/bright light, etc.), the Barnes maze may not be the most optimal behavioral task to assess subtle deficits in cognitive behavior. In humans, obesity generally results in mild rather than severe cognitive impairment [133,134]. Therefore, a more sensitive behavioral task may allow us to uncover cognitive deficits more readily than the Barnes maze. Cognition is a complex and multi-faceted construct. Thus, while we found evidence for differences in cognitive performance, future studies should include a battery of behavioral tests to tap into different aspects of cognitive behavior that may be adversely influenced by HF diet. Importantly, prolonged HF diet intake and preference may impair specific, rather than global, domains of cognitive function in a sex-specific fashion. Impairments in one domain may be more influential in the regulation of feeding behavior and lead to worse outcomes depending on sex. Caution should be taken when interpreting the results due to inherent limitations resulting from the complexity of the study design. Different fat sources and compositions (e.g., polyunsaturated, monounsaturated) may differentially affect aspects of metabolism and cognition. Future studies should utilize diets matched in macronutrient sources as much as possible to limit the confounding effect of differences in raw materials. To our knowledge, no study has assessed changes in insulin sensitivity following HF diet and voluntary exercise in rats of both sexes. The addition of a naïve chow-fed control group is necessary to make between-group comparisons to strengthen the argument for the protective effect of exercise against HF diet. Exercise is more likely to have a greater beneficial effect in males, but the addition of a WR group maintained on only HF diet is necessary to make this conclusion given the shift in HF diet preference across time and the highly variable HF diet intake between subjects and sexes. Here, we focused on voluntary exercise; however, the effect of exercise may shift depending on exercise conditions (forced vs. voluntary, strength vs. endurance, acute vs. chronic, etc.). An extensive investigation is necessary before results can be directly translated to humans. Nevertheless, our results lend support to our hypotheses regarding the protective effect of exercise against the detrimental outcomes of HF feeding and inform the development of more optimized designs for future studies. 5. Conclusions We examined sex differences in exercise-mediated changes in diet choice and the degree to which exercise can reverse the metabolic dysregulation and improve cognitive performance associated with long-term HF feeding. The protective effect of exercise on suppressing HF diet preference and HF-mediated insults to peripheral metabolism was specific to males whereas exercise similarly enhanced learning on the Barnes maze in both males and females. Intriguingly, despite less improvement on their metabolic profile, female WR rats still benefited from the exercise and showed improved performance on the Barnes maze. This finding suggests that cardio-based exercise can potentially exert differential effects on metabolic and cognitive function. Taken together, these results suggest that the adverse Nutrients 2020, 12, 2721 16 of 22 metabolic effects of chronic HF feeding and preference are especially detrimental to females, but exercise remains a good intervention option for both males and females to prevent cognitive decline resulting from poor dietary choices. Supplementary Materials: The following are available online at http://www.mdpi.com/2072-6643/12/9/2721/s1, Video S1: Serial search training trial, Video S2: Serial search probe trial. Author Contributions: Conceptualization, N.-C.L. and T.Y.Y.; methodology, N.-C.L. and T.Y.Y.; formal analysis, T.Y.Y. and N.-C.L.; data curation, T.Y.Y. and Z.G.; writing—original draft preparation, T.Y.Y.; writing—review and editing, N.-C.L.; visualization, T.Y.Y.; funding acquisition, N.-C.L. All authors have read and agreed to the published version of the manuscript. Funding: This study was supported by the UIUC Psychology Department startup funds (to N.-C.L.). Acknowledgments: We thank Fatima Najera and Gabi Petrus for their help with videoscoring and animal care. Conflicts of Interest: The authors declare no conflict of interest. References 1. Hedley, A.A.; Ogden, C.L.; Johnson, C.L.; Carroll, M.D.; Curtin, L.R.; Flegal, K.M. Prevalence of overweight and obesity among US children, adolescents, and adults, 1999–2002. JAMA 2004, 291, 2847–2850. [CrossRef] [PubMed] 2. Crittenden, A.N.; Schnorr, S.L. Current views on hunter-gatherer nutrition and the evolution of the human diet. Am. J. Phys. Anthropol. 2017, 162 (Suppl. 63), 84–109. [CrossRef] 3. Cordain, L.; Eaton, S.B.; Sebastian, A.; Mann, N.; Lindeberg, S.; Watkins, B.A.; O’Keefe, J.H.; Brand-Miller, J. Origins and evolution of the Western diet: Health implications for the 21st century. Am. J. Clin. Nutr. 2005, 81, 341–354. [CrossRef] [PubMed] 4. Ziegler, E. Secular changes in the stature of adults and the secular trend of the modern sugar consumption. Z. Kinderheilkd 1967, 99, 146–166. [CrossRef] [PubMed] 5. Shan, Z.; Rehm, C.D.; Rogers, G.; Ruan, M.; Wang, D.D.; Hu, F.B.; Mozaffarian, D.; Zhang, F.F.; Bhupathiraju, S.N. Trends in Dietary Carbohydrate, Protein, and Fat Intake and Diet Quality Among US Adults, 1999–2016. JAMA 2019, 322, 1178–1187. [CrossRef] 6. Vadiveloo, M.; Scott, M.; Quatromoni, P.; Jacques, P.; Parekh, N. Trends in dietary fat and high-fat food intakes from 1991 to 2008 in the Framingham Heart Study participants. Br. J. Nutr. 2014, 111, 724–734. [CrossRef] 7. Okreglicka, K. Health effects of changes in the structure of dietary macronutrients intake in western societies. Rocz. Panstw. Zakl. Hig. 2015, 66, 97–105. 8. Foster-Powell, K.; Holt, S.H.; Brand-Miller, J.C. International table of glycemic index and glycemic load values: 2002. Am. J. Clin. Nutr. 2002, 76, 5–56. [CrossRef] 9. Steenbergen, L.; Colzato, L.S. Overweight and Cognitive Performance: High Body Mass Index Is Associated with Impairment in Reactive Control during Task Switching. Front. Nutr. 2017, 4, 51. [CrossRef] 10. Yang, Y.; Shields, G.S.; Guo, C.; Liu, Y. Executive function performance in obesity and overweight individuals: A meta-analysis and review. Neurosci. BioBehav. Rev. 2018, 84, 225–244. [CrossRef] 11. Smith, E.; Hay, P.; Campbell, L.; Trollor, J.N. A review of the association between obesity and cognitive function across the lifespan: Implications for novel approaches to prevention and treatment. Obes. Rev. 2011, 12, 740–755. [CrossRef] 12. Chen, F.T.; Chen, S.R.; Chu, I.H.; Liu, J.H.; Chang, Y.K. Multicomponent Exercise Intervention and Metacognition in Obese Preadolescents: A Randomized Controlled Study. J. Sport Exerc. Psychol. 2017, 39, 302–312. [CrossRef] [PubMed] 13. Martin, B.; Pearson, M.; Kebejian, L.; Golden, E.; Keselman, A.; Bender, M.; Carlson, O.; Egan, J.; Ladenheim, B.; Cadet, J.L.; et al. Sex-dependent metabolic, neuroendocrine, and cognitive responses to dietary energy restriction and excess. Endocrinology 2007, 148, 4318–4333. [CrossRef] [PubMed] 14. Winocur, G.; Greenwood, C.E.; Piroli, G.G.; Grillo, C.A.; Reznikov, L.R.; Reagan, L.P.; McEwen, B.S. Memory impairment in obese Zucker rats: An investigation of cognitive function in an animal model of insulin resistance and obesity. Behav. Neurosci. 2005, 119, 1389–1395. [CrossRef] Nutrients 2020, 12, 2721 17 of 22 15. Kanoski, S.E.; Meisel, R.L.; Mullins, A.J.; Davidson, T.L. The effects of energy-rich diets on discrimination reversal learning and on BDNF in the hippocampus and prefrontal cortex of the rat. Behav. Brain Res. 2007, 182, 57–66. [CrossRef] [PubMed] 16. Maesako, M.; Uemura, K.; Kubota, M.; Kuzuya, A.; Sasaki, K.; Hayashida, N.; Asada-Utsugi, M.; Watanabe, K.; Uemura, M.; Kihara, T.; et al. Exercise is more effective than diet control in preventing high fat diet-induced beta-amyloid deposition and memory deficit in amyloid precursor protein transgenic mice. J. Biol. Chem. 2012, 287, 23024–23033. [CrossRef] [PubMed] 17. Klem, M.L.; Wing, R.R.; McGuire, M.T.; Seagle, H.M.; Hill, J.O. A descriptive study of individuals successful at long-term maintenance of substantial weight loss. Am. J. Clin. Nutr. 1997, 66, 239–246. [CrossRef] [PubMed] 18. Shick, S.M.; Wing, R.R.; Klem, M.L.; McGuire, M.T.; Hill, J.O.; Seagle, H. Persons successful at long-term weight loss and maintenance continue to consume a low-energy, low-fat diet. J. Am. Diet. Assoc. 1998, 98, 408–413. [CrossRef] 19. Looy, H.; Eikelboom, R. Wheel running, food intake, and body weight in male rats. Physiol. Behav. 1989, 45, 403–405. [CrossRef] 20. Kawaguchi, M.; Scott, K.A.; Moran, T.H.; Bi, S. Dorsomedial hypothalamic corticotropin-releasing factor mediation of exercise-induced anorexia. Am. J. Physiol. Regul. Integr. Comp. Physiol. 2005, 288, R1800–R1805. [CrossRef] 21. Tokuyama, K.; Saito, M.; Okuda, H. Effects of wheel running on food intake and weight gain of male and female rats. Physiol. Behav. 1982, 28, 899–903. [CrossRef] 22. Carrera, O.; Cerrato, M.; Vazquez, R.; Sineiro, C.; Gutierrez, E. Gender dimorphic effects of voluntary running in laboratory rats depends on maturational status. Q. J. Exp. Psychol. (Hove) 2011, 64, 823–832. [CrossRef] [PubMed] 23. Eckel, L.A.; Moore, S.R. Diet-induced hyperphagia in the rat is influenced by sex and exercise. Am. J. Physiol. Regul. Integr. Comp. Physiol. 2004, 287, R1080–R1085. [CrossRef] [PubMed] 24. Blundell, J.E.; King, N.A. Physical activity and regulation of food intake: Current evidence. Med. Sci. Sports Exerc. 1999, 31, S573–S583. [CrossRef] [PubMed] 25. Donnelly, J.E.; Herrmann, S.D.; Lambourne, K.; Szabo, A.N.; Honas, J.J.; Washburn, R.A. Does increased exercise or physical activity alter ad-libitum daily energy intake or macronutrient composition in healthy adults? A systematic review. PLoS ONE 2014, 9, e83498. [CrossRef] 26. Ebrahimi, M.; Rahmani-Nia, F.; Damirchi, A.; Mirzaie, B.; Asghar Pur, S. Effect of Short-term Exercise on Appetite, Energy Intake and Energy-regulating Hormones. Iran. J. Basic Med. Sci. 2013, 16, 829–834. 27. King, N.A.; Hopkins, M.; Caudwell, P.; Stubbs, R.J.; Blundell, J.E. Beneficial effects of exercise: Shifting the focus from body weight to other markers of health. Br. J. Sports Med. 2009, 43, 924–927. [CrossRef] 28. Schubert, M.M.; Desbrow, B.; Sabapathy, S.; Leveritt, M. Acute exercise and subsequent energy intake. A meta-analysis. Appetite 2013, 63, 92–104. [CrossRef] 29. Castro, E.A.; Carraca, E.V.; Cupeiro, R.; Lopez-Plaza, B.; Teixeira, P.J.; Gonzalez-Lamuno, D.; Peinado, A.B. The Effects of the Type of Exercise and Physical Activity on Eating Behavior and Body Composition in Overweight and Obese Subjects. Nutrients 2020, 12, 557. [CrossRef] 30. Donnelly, J.E.; Smith, B.K. Is exercise effective for weight loss with ad libitum diet? Energy balance, compensation, and gender differences. Exerc. Sport Sci. Rev. 2005, 33, 169–174. [CrossRef] 31. Anderson, J.W.; Konz, E.C.; Frederich, R.C.; Wood, C.L. Long-term weight-loss maintenance: A meta-analysis of US studies. Am. J. Clin. Nutr. 2001, 74, 579–584. [CrossRef] [PubMed] 32. Despres, J.P.; Bouchard, C.; Savard, R.; Tremblay, A.; Marcotte, M.; Theriault, G. The effect of a 20-week endurance training program on adipose-tissue morphology and lipolysis in men and women. Metabolism 1984, 33, 235–239. [CrossRef] 33. Williams, R.L.; Wood, L.G.; Collins, C.E.; Callister, R. Effectiveness of weight loss interventions–is there a difference between men and women: A systematic review. Obes. Rev. 2015, 16, 171–186. [CrossRef] [PubMed] 34. Washburn, R.A.; Honas, J.J.; Ptomey, L.T.; Mayo, M.S.; Lee, J.; Sullivan, D.K.; Lambourne, K.; Willis, E.A.; Donnelly, J.E. Energy and Macronutrient Intake in the Midwest Exercise Trial 2 (MET-2). Med. Sci. Sports Exerc. 2015, 47, 1941–1949. [CrossRef] [PubMed] Nutrients 2020, 12, 2721 18 of 22 35. Westerterp, K.R.; Meijer, G.A.; Janssen, E.M.; Saris, W.H.; Ten Hoor, F. Long-term effect of physical activity on energy balance and body composition. Br. J. Nutr. 1992, 68, 21–30. [CrossRef] [PubMed] 36. Coker, R.H.; Williams, R.H.; Yeo, S.E.; Kortebein, P.M.; Bodenner, D.L.; Kern, P.A.; Evans, W.J. The impact of exercise training compared to caloric restriction on hepatic and peripheral insulin resistance in obesity. J. Clin. Endocrinol. Metab. 2009, 94, 4258–4266. [CrossRef] [PubMed] 37. Nara, M.; Takahashi, M.; Kanda, T.; Shimomura, Y.; Kobayashi, I. Running exercise improves metabolic abnormalities and fat accumulation in sucrose-induced insulin-resistant rats. Obes. Res. 1997, 5, 348–353. [CrossRef] 38. Greenwood, C.E.; Winocur, G. High-fat diets, insulin resistance and declining cognitive function. Neurobiol. Aging 2005, 26 (Suppl. 1), 42–45. [CrossRef] 39. Winocur, G.; Greenwood, C.E. Studies of the effects of high fat diets on cognitive function in a rat model. Neurobiol. Aging 2005, 26 (Suppl. 1), 46–49. [CrossRef] 40. Greenwood, C.E.; Winocur, G. Cognitive impairment in rats fed high-fat diets: A specific effect of saturated fatty-acid intake. Behav. Neurosci. 1996, 110, 451–459. [CrossRef] 41. Winocur, G.; Greenwood, C.E. The effects of high fat diets and environmental influences on cognitive performance in rats. Behav. Brain Res. 1999, 101, 153–161. [CrossRef] 42. Stranahan, A.M.; Norman, E.D.; Lee, K.; Cutler, R.G.; Telljohann, R.S.; Egan, J.M.; Mattson, M.P. Diet-induced insulin resistance impairs hippocampal synaptic plasticity and cognition in middle-aged rats. Hippocampus 2008, 18, 1085–1088. [CrossRef] [PubMed] 43. Valladolid-Acebes, I.; Stucchi, P.; Cano, V.; Fernandez-Alfonso, M.S.; Merino, B.; Gil-Ortega, M.; Fole, A.; Morales, L.; Ruiz-Gayo, M.; Del Olmo, N. High-fat diets impair spatial learning in the radial-arm maze in mice. Neurobiol. Learn. Mem. 2011, 95, 80–85. [CrossRef] [PubMed] 44. Kanoski, S.E.; Davidson, T.L. Different patterns of memory impairments accompany short- and longer-term maintenance on a high-energy diet. J. Exp. Psychol. Anim. Behav. Process. 2010, 36, 313–319. [CrossRef] [PubMed] 45. Brockett, A.T.; LaMarca, E.A.; Gould, E. Physical exercise enhances cognitive flexibility as well as astrocytic and synaptic markers in the medial prefrontal cortex. PLoS ONE 2015, 10, e0124859. [CrossRef] [PubMed] 46. Fordyce, D.E.; Farrar, R.P. Enhancement of spatial learning in F344 rats by physical activity and related learning-associated alterations in hippocampal and cortical cholinergic functioning. Behav. Brain Res. 1991, 46, 123–133. [CrossRef] 47. Voss, M.W.; Vivar, C.; Kramer, A.F.; van Praag, H. Bridging animal and human models of exercise-induced brain plasticity. Trends Cogn. Sci. 2013, 17, 525–544. [CrossRef] 48. Molteni, R.; Wu, A.; Vaynman, S.; Ying, Z.; Barnard, R.J.; Gomez-Pinilla, F. Exercise reverses the harmful effects of consumption of a high-fat diet on synaptic and behavioral plasticity associated to the action of brain-derived neurotrophic factor. Neuroscience 2004, 123, 429–440. [CrossRef] 49. Klein, C.; Jonas, W.; Iggena, D.; Empl, L.; Rivalan, M.; Wiedmer, P.; Spranger, J.; Hellweg, R.; Winter, Y.; Steiner, B. Exercise prevents high-fat diet-induced impairment of flexible memory expression in the water maze and modulates adult hippocampal neurogenesis in mice. Neurobiol. Learn. Mem. 2016, 131, 26–35. [CrossRef] 50. Noble, E.E.; Mavanji, V.; Little, M.R.; Billington, C.J.; Kotz, C.M.; Wang, C. Exercise reduces diet-induced cognitive decline and increases hippocampal brain-derived neurotrophic factor in CA3 neurons. Neurobiol. Learn. Mem. 2014, 114, 40–50. [CrossRef] 51. Woo, J.; Shin, K.O.; Park, S.Y.; Jang, K.S.; Kang, S. Effects of exercise and diet change on cognition function and synaptic plasticity in high fat diet induced obese rats. Lipids Health Dis. 2013, 12, 144. [CrossRef] [PubMed] 52. Gibbons, T.E.; Pence, B.D.; Petr, G.; Ossyra, J.M.; Mach, H.C.; Bhattacharya, T.K.; Perez, S.; Martin, S.A.; McCusker, R.H.; Kelley, K.W.; et al. Voluntary wheel running, but not a diet containing (-)-epigallocatechin-3-gallate and beta-alanine, improves learning, memory and hippocampal neurogenesis in aged mice. Behav. Brain Res. 2014, 272, 131–140. [CrossRef] [PubMed] 53. Aslani, S.; Vieira, N.; Marques, F.; Costa, P.S.; Sousa, N.; Palha, J.A. The effect of high-fat diet on rat’s mood, feeding behavior and response to stress. Transl. Psychiatry 2015, 5, e684. [CrossRef] [PubMed] 54. McNeilly, A.D.; Williamson, R.; Sutherland, C.; Balfour, D.J.; Stewart, C.A. High fat feeding promotes simultaneous decline in insulin sensitivity and cognitive performance in a delayed matching and non-matching to position task. Behav. Brain Res. 2011, 217, 134–141. [CrossRef] [PubMed] Nutrients 2020, 12, 2721 19 of 22 55. Bocarsly, M.E.; Fasolino, M.; Kane, G.A.; LaMarca, E.A.; Kirschen, G.W.; Karatsoreos, I.N.; McEwen, B.S.; Gould, E. Obesity diminishes synaptic markers, alters microglial morphology, and impairs cognitive function. Proc. Natl. Acad. Sci. USA 2015, 112, 15731–15736. [CrossRef] 56. Edwards, C.G.; Walk, A.M.; Thompson, S.V.; Mullen, S.P.; Holscher, H.D.; Khan, N.A. Disordered Eating Attitudes and Behavioral and Neuroelectric Indices of Cognitive Flexibility in Individuals with Overweight and Obesity. Nutrients 2018, 10, 1902. [CrossRef] 57. Perpina, C.; Segura, M.; Sanchez-Reales, S. Cognitive flexibility and decision-making in eating disorders and obesity. Eat. Weight Disord. 2017, 22, 435–444. [CrossRef] 58. McNeilly, A.D.; Gao, A.; Hill, A.Y.; Gomersall, T.; Balfour, D.J.K.; Sutherland, C.; Stewart, C.A. The effect of dietary intervention on the metabolic and behavioural impairments generated by short term high fat feeding in the rat. Physiol. Behav. 2016, 167, 100–109. [CrossRef] 59. Boitard, C.; Parkes, S.L.; Cavaroc, A.; Tantot, F.; Castanon, N.; Laye, S.; Tronel, S.; Pacheco-Lopez, G.; Coutureau, E.; Ferreira, G. Switching Adolescent High-Fat Diet to Adult Control Diet Restores Neurocognitive Alterations. Front. Behav. Neurosci. 2016, 10, 225. [CrossRef] 60. McNeilly, A.D.; Williamson, R.; Balfour, D.J.; Stewart, C.A.; Sutherland, C. A high-fat-diet-induced cognitive deficit in rats that is not prevented by improving insulin sensitivity with metformin. Diabetologia 2012, 55, 3061–3070. [CrossRef] 61. Amengual-Cladera, E.; Llado, I.; Gianotti, M.; Proenza, A.M. Sex differences in the effect of high-fat diet feeding on rat white adipose tissue mitochondrial function and insulin sensitivity. Metabolism 2012, 61, 1108–1117. [CrossRef] [PubMed] 62. Barron, A.M.; Rosario, E.R.; Elteriefi, R.; Pike, C.J. Sex-specific effects of high fat diet on indices of metabolic syndrome in 3xTg-AD mice: Implications for Alzheimer’s disease. PLoS ONE 2013, 8, e78554. [CrossRef] [PubMed] 63. Elias, M.F.; Elias, P.K.; Sullivan, L.M.; Wolf, P.A.; D’Agostino, R.B. Lower cognitive function in the presence of obesity and hypertension: The Framingham heart study. Int. J. Obes. Relat. Metab. Disord. 2003, 27, 260–268. [CrossRef] [PubMed] 64. Underwood, E.L.; Thompson, L.T. High-fat diet impairs spatial memory and hippocampal intrinsic excitability and sex-dependently alters circulating insulin and hippocampal insulin sensitivity. Biol. Sex Differ. 2016, 7, 9. [CrossRef] 65. Shields, G.S.; Trainor, B.C.; Lam, J.C.; Yonelinas, A.P. Acute stress impairs cognitive flexibility in men, not women. Stress 2016, 19, 542–546. [CrossRef] 66. Laredo, S.A.; Steinman, M.Q.; Robles, C.F.; Ferrer, E.; Ragen, B.J.; Trainor, B.C. Effects of defeat stress on behavioral flexibility in males and females: Modulation by the mu-opioid receptor. Eur. J. Neurosci. 2015, 41, 434–441. [CrossRef] 67. Hwang, L.L.; Wang, C.H.; Li, T.L.; Chang, S.D.; Lin, L.C.; Chen, C.P.; Chen, C.T.; Liang, K.C.; Ho, I.K.; Yang, W.S.; et al. Sex differences in high-fat diet-induced obesity, metabolic alterations and learning, and synaptic plasticity deficits in mice. Obesity (Silver Spring) 2010, 18, 463–469. [CrossRef] 68. Li, W.; Qiu, Q.; Sun, L.; Yue, L.; Wang, T.; Li, X.; Xiao, S. Sex differences in obesity and cognitive function in a cognitively normal aging Chinese Han population. Neuropsychiatr. Dis. Treat. 2017, 13, 2405–2410. [CrossRef] 69. Gollisch, K.S.; Brandauer, J.; Jessen, N.; Toyoda, T.; Nayer, A.; Hirshman, M.F.; Goodyear, L.J. Effects of exercise training on subcutaneous and visceral adipose tissue in normal- and high-fat diet-fed rats. Am. J. Physiol. Endocrinol. Metab. 2009, 297, E495–E504. [CrossRef] 70. Tokuyama, K.; Suzuki, M. Intravenous glucose tolerance test-derived glucose effectiveness in endurance-trained rats. Metabolism 1998, 47, 190–194. [CrossRef] 71. Goodyear, L.J.; Hirshman, M.F.; Knutson, S.M.; Horton, E.D.; Horton, E.S. Effect of exercise training on glucose homeostasis in normal and insulin-deficient diabetic rats. J. Appl. Physiol. (1985) 1988, 65, 844–851. [CrossRef] [PubMed] 72. Beaudry, J.L.; Dunford, E.C.; Leclair, E.; Mandel, E.R.; Peckett, A.J.; Haas, T.L.; Riddell, M.C. Voluntary exercise improves metabolic profile in high-fat fed glucocorticoid-treated rats. J. Appl. Physiol. (1985) 2015, 118, 1331–1343. [CrossRef] [PubMed] 73. Bongbele, J.; Gutierrez, A.; Cardin, S.; Lavoie, J.M. Effect of physical training on insulin response to intravenous glucose in male peripubertal rats. J. Appl. Physiol. (1985) 1992, 73, 1227–1231. [CrossRef] [PubMed] Nutrients 2020, 12, 2721 20 of 22 74. Caudwell, P.; Gibbons, C.; Finlayson, G.; Naslund, E.; Blundell, J. Exercise and weight loss: No sex differences in body weight response to exercise. Exerc. Sport Sci. Rev. 2014, 42, 92–101. [CrossRef] 75. Donnelly, J.E.; Smith, B.; Jacobsen, D.J.; Kirk, E.; Dubose, K.; Hyder, M.; Bailey, B.; Washburn, R. The role of exercise for weight loss and maintenance. Best Pract. Res. Clin. Gastroenterol. 2004, 18, 1009–1029. [CrossRef] 76. Hill, J.O.; Thiel, J.; Heller, P.A.; Markon, C.; Fletcher, G.; DiGirolamo, M. Differences in effects of aerobic exercise training on blood lipids in men and women. Am. J. Cardiol. 1989, 63, 254–256. [CrossRef] 77. Buettner, R.; Scholmerich, J.; Bollheimer, L.C. High-fat diets: Modeling the metabolic disorders of human obesity in rodents. Obesity (Silver Spring) 2007, 15, 798–808. [CrossRef] 78. Gawel, K.; Gibula, E.; Marszalek-Grabska, M.; Filarowska, J.; Kotlinska, J.H. Assessment of spatial learning and memory in the Barnes maze task in rodents-methodological consideration. Naunyn. Schmiedebergs Arch. Pharmacol. 2019, 392, 1–18. [CrossRef] 79. National Research Council. Guide for the Care and Use of Laboratory Animals, 8th ed.; National Research Council: Washington, DC, USA, 2011. [CrossRef] 80. Yang, T.Y.; Liang, N.C. Ovarian hormones mediate running-induced changes in high fat diet choice patterns in female rats. Horm. Behav. 2018, 100, 81–93. [CrossRef] 81. Bowe, J.E.; Franklin, Z.J.; Hauge-Evans, A.C.; King, A.J.; Persaud, S.J.; Jones, P.M. Metabolic phenotyping guidelines: Assessing glucose homeostasis in rodent models. J. Endocrinol. 2014, 222, G13–G25. [CrossRef] 82. Ayala, J.E.; Bracy, D.P.; McGuinness, O.P.; Wasserman, D.H. Considerations in the design of hyperinsulinemic-euglycemic clamps in the conscious mouse. Diabetes 2006, 55, 390–397. [CrossRef] [PubMed] 83. Tran, T.T.; Gupta, N.; Goh, T.; Naigamwalla, D.; Chia, M.C.; Koohestani, N.; Mehrotra, S.; McKeown-Eyssen, G.; Giacca, A.; Bruce, W.R. Direct measure of insulin sensitivity with the hyperinsulinemic-euglycemic clamp and surrogate measures of insulin sensitivity with the oral glucose tolerance test: Correlations with aberrant crypt foci promotion in rats. Cancer Epidemiol. Biomarkers Prev. 2003, 12, 47–56. [PubMed] 84. Matthews, D.R.; Hosker, J.P.; Rudenski, A.S.; Naylor, B.A.; Treacher, D.F.; Turner, R.C. Homeostasis model assessment: Insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985, 28, 412–419. [CrossRef] [PubMed] 85. Gutt, M.; Davis, C.L.; Spitzer, S.B.; Llabre, M.M.; Kumar, M.; Czarnecki, E.M.; Schneiderman, N.; Skyler, J.S.; Marks, J.B. Validation of the insulin sensitivity index (ISI(0,120)): Comparison with other measures. Diabetes Res. Clin. Pract. 2000, 47, 177–184. [CrossRef] 86. O’Leary, T.P.; Savoie, V.; Brown, R.E. Learning, memory and search strategies of inbred mouse strains with different visual abilities in the Barnes maze. Behav. Brain Res. 2011, 216, 531–542. [CrossRef] 87. Rosenfeld, C.S.; Ferguson, S.A. Barnes maze testing strategies with small and large rodent models. J. Vis. Exp. 2014, e51194. [CrossRef] 88. Brooks, G.A. Importance of the ’crossover’ concept in exercise metabolism. Clin. Exp. Pharmacol. Physiol. 1997, 24, 889–895. [CrossRef] 89. Horowitz, J.F. Fatty acid mobilization from adipose tissue during exercise. Trends Endocrinol. Metab. 2003, 14, 386–392. [CrossRef] 90. Ruby, B.C.; Robergs, R.A. Gender differences in substrate utilisation during exercise. Sports Med. 1994, 17, 393–410. [CrossRef] 91. Horton, T.J.; Pagliassotti, M.J.; Hobbs, K.; Hill, J.O. Fuel metabolism in men and women during and after long-duration exercise. J. Appl. Physiol. (1985) 1998, 85, 1823–1832. [CrossRef] 92. Brooks, G.A.; Mercier, J. Balance of carbohydrate and lipid utilization during exercise: The “crossover” concept. J. Appl. Physiol. (1985) 1994, 76, 2253–2261. [CrossRef] [PubMed] 93. Liang, N.C.; Bello, N.T.; Moran, T.H. Wheel running reduces high-fat diet intake, preference and mu-opioid agonist stimulated intake. Behav. Brain Res. 2015, 284, 1–10. [CrossRef] [PubMed] 94. Moody, L.; Liang, J.; Choi, P.P.; Moran, T.H.; Liang, N.C. Wheel running decreases palatable diet preference in Sprague-Dawley rats. Physiol. Behav. 2015, 150, 53–63. [CrossRef] [PubMed] 95. Yang, T.Y.; Gardner, J.C.; Gao, Z.; Pan, Y.X.; Liang, N.C. Role of glucocorticoid signaling in exercise-associated changes in high-fat diet preference in rats. Am. J. Physiol. Regul. Integr. Comp. Physiol. 2020, 318, R515–R528. [CrossRef] 96. Ellis, G.S.; Lanza-Jacoby, S.; Gow, A.; Kendrick, Z.V. Effects of estradiol on lipoprotein lipase activity and lipid availability in exercised male rats. J. Appl. Physiol (1985) 1994, 77, 209–215. [CrossRef] Nutrients 2020, 12, 2721 21 of 22 97. Roepstorff, C.; Steffensen, C.H.; Madsen, M.; Stallknecht, B.; Kanstrup, I.L.; Richter, E.A.; Kiens, B. Gender differences in substrate utilization during submaximal exercise in endurance-trained subjects. Am. J. Physiol. Endocrinol. Metab. 2002, 282, E435–E447. [CrossRef] 98. Carter, S.L.; Rennie, C.; Tarnopolsky, M.A. Substrate utilization during endurance exercise in men and women after endurance training. Am. J. Physiol. Endocrinol. Metab. 2001, 280, E898–E907. [CrossRef] 99. Romijn, J.A.; Coyle, E.F.; Sidossis, L.S.; Gastaldelli, A.; Horowitz, J.F.; Endert, E.; Wolfe, R.R. Regulation of endogenous fat and carbohydrate metabolism in relation to exercise intensity and duration. Am. J. Physiol. 1993, 265, E380–E391. [CrossRef] 100. Hawley, J.A.; Hopkins, W.G. Aerobic glycolytic and aerobic lipolytic power systems. A new paradigm with implications for endurance and ultraendurance events. Sports Med. 1995, 19, 240–250. [CrossRef] 101. Hudson, J.I.; Hiripi, E.; Pope, H.G., Jr.; Kessler, R.C. The prevalence and correlates of eating disorders in the National Comorbidity Survey Replication. Biol. Psychiatry 2007, 61, 348–358. [CrossRef] 102. Klump, K.L.; Racine, S.; Hildebrandt, B.; Sisk, C.L. Sex differences in binge eating patterns in male and female adult rats. Int. J. Eat. Disord. 2013, 46, 729–736. [CrossRef] [PubMed] 103. Tapia, M.A.; Lee, J.R.; Weise, V.N.; Tamasi, A.M.; Will, M.J. Sex differences in hedonic and homeostatic aspects of palatable food motivation. Behav. Brain Res. 2019, 359, 396–400. [CrossRef] [PubMed] 104. Buczek, L.; Migliaccio, J.; Petrovich, G.D. Hedonic Eating: Sex Differences and Characterization of Orexin Activation and Signaling. Neuroscience 2020, 436, 34–45. [CrossRef] [PubMed] 105. Ng, M.; Fleming, T.; Robinson, M.; Thomson, B.; Graetz, N.; Margono, C.; Mullany, E.C.; Biryukov, S.; Abbafati, C.; Abera, S.F.; et al. Global, regional, and national prevalence of overweight and obesity in children and adults during 1980-2013: A systematic analysis for the Global Burden of Disease Study 2013. Lancet 2014, 384, 766–781. [CrossRef] 106. Lynch, W.J. Sex differences in vulnerability to drug self-administration. Exp. Clin. Psychopharmacol. 2006, 14, 34–41. [CrossRef] 107. Carroll, M.E.; Anker, J.J. Sex differences and ovarian hormones in animal models of drug dependence. Horm. Behav. 2010, 58, 44–56. [CrossRef] 108. Day, C.J.; McHale, S.; Francis, J. Individual differences and preference for dietary fat using the Fat Preference Questionnaire((c)) in a UK sample. Appetite 2012, 58, 679–686. [CrossRef] 109. Belke, T.W.; Wagner, J.P. The reinforcing property and the rewarding aftereffect of wheel running in rats: A combination of two paradigms. Behav. Process. 2005, 68, 165–172. [CrossRef] 110. Spierling, S.R.; Kreisler, A.D.; Williams, C.A.; Fang, S.Y.; Pucci, S.N.; Kines, K.T.; Zorrilla, E.P. Intermittent, extended access to preferred food leads to escalated food reinforcement and cyclic whole-body metabolism in rats: Sex differences and individual vulnerability. Physiol. Behav. 2018, 192, 3–16. [CrossRef] 111. Peterson, A.B.; Hivick, D.P.; Lynch, W.J. Dose-dependent effectiveness of wheel running to attenuate cocaine-seeking: Impact of sex and estrous cycle in rats. Psychopharmacology (Berl) 2014, 231, 2661–2670. [CrossRef] 112. Zhou, Y.; Zhao, M.; Zhou, C.; Li, R. Sex differences in drug addiction and response to exercise intervention: From human to animal studies. Front. Neuroendocrinol. 2016, 40, 24–41. [CrossRef] 113. Pitts, G.C. Body composition in the rat: Interactions of exercise, age, sex, and diet. Am. J. Physiol. 1984, 246, R495–R501. [CrossRef] [PubMed] 114. Foright, R.M.; Johnson, G.C.; Kahn, D.; Charleston, C.A.; Presby, D.M.; Bouchet, C.A.; Wellberg, E.A.; Sherk, V.D.; Jackman, M.R.; Greenwood, B.N.; et al. Compensatory eating behaviors in male and female rats in response to exercise training. Am. J. Physiol. Regul. Integr. Comp. Physiol. 2020. [CrossRef] [PubMed] 115. Latour, M.G.; Shinoda, M.; Lavoie, J.M. Metabolic effects of physical training in ovariectomized and hyperestrogenic rats. J. Appl. Physiol. (1985) 2001, 90, 235–241. [CrossRef] [PubMed] 116. Lamontagne, J.; Jalbert-Arsenault, E.; Pepin, E.; Peyot, M.L.; Ruderman, N.B.; Nolan, C.J.; Joly, E.; Madiraju, S.R.; Poitout, V.; Prentki, M. Pioglitazone acutely reduces energy metabolism and insulin secretion in rats. Diabetes 2013, 62, 2122–2129. [CrossRef] [PubMed] 117. Vallerand, A.L.; Lupien, J.; Deshaies, Y.; Bukowiecki, L.J. Intensive exercise training does not improve intravenous glucose tolerance in severely diabetic rats. Horm. Metab. Res. 1986, 18, 79–81. [CrossRef] 118. Goodyear, L.J.; Kahn, B.B. Exercise, glucose transport, and insulin sensitivity. Annu. Rev. Med. 1998, 49, 235–261. [CrossRef] Nutrients 2020, 12, 2721 22 of 22 119. Goodyear, L.J.; Hirshman, M.F.; Horton, E.D.; Knutson, S.M.; Wardzala, L.J.; Horton, E.S. Exercise training normalizes glucose metabolism in a rat model of impaired glucose tolerance. Metabolism 1991, 40, 455–464. [CrossRef] 120. LaPier, T.L.; Swislocki, A.L.; Clark, R.J.; Rodnick, K.J. Voluntary running improves glucose tolerance and insulin resistance in female spontaneously hypertensive rats. Am. J. Hypertens 2001, 14, 708–715. [CrossRef] 121. Ascaso, J.F.; Pardo, S.; Real, J.T.; Lorente, R.I.; Priego, A.; Carmena, R. Diagnosing insulin resistance by simple quantitative methods in subjects with normal glucose metabolism. Diabetes Care 2003, 26, 3320–3325. [CrossRef] 122. Swift, D.L.; McGee, J.E.; Earnest, C.P.; Carlisle, E.; Nygard, M.; Johannsen, N.M. The Effects of Exercise and Physical Activity on Weight Loss and Maintenance. Prog. Cardiovasc. Dis. 2018, 61, 206–213. [CrossRef] [PubMed] 123. Villanueva Espino, L.A.; Silva Gomez, A.B.; Bravo Duran, D.A. Cognitive training increases dendritic arborization in the dorsal hippocampal CA1 and CA3 neurons of female and male Long-Evans rats. Synapse 2020, 74, e22140. [CrossRef] [PubMed] 124. Betancourt, E.; Wachtel, J.; Michaelos, M.; Haggerty, M.; Conforti, J.; Kritzer, M.F. The impact of biological sex and sex hormones on cognition in a rat model of early, pre-motor Parkinson’s disease. Neuroscience 2017, 345, 297–314. [CrossRef] [PubMed] 125. Beatty, W.W. Gonadal hormones and sex differences in nonreproductive behaviors in rodents: Organizational and activational influences. Horm. Behav. 1979, 12, 112–163. [CrossRef] 126. Tropp, J.; Markus, E.J. Sex differences in the dynamics of cue utilization and exploratory behavior. Behav. Brain Res. 2001, 119, 143–154. [CrossRef] 127. Jonasson, Z. Meta-analysis of sex differences in rodent models of learning and memory: A review of behavioral and biological data. Neurosci. BioBehav. Rev. 2005, 28, 811–825. [CrossRef] 128. Locklear, M.N.; Kritzer, M.F. Assessment of the effects of sex and sex hormones on spatial cognition in adult rats using the Barnes maze. Horm. Behav. 2014, 66, 298–308. [CrossRef] 129. Bucci, D.J.; Chiba, A.A.; Gallagher, M. Spatial learning in male and female Long-Evans rats. Behav. Neurosci. 1995, 109, 180–183. [CrossRef] 130. Koss, W.A.; Frick, K.M. Sex differences in hippocampal function. J. Neurosci. Res. 2017, 95, 539–562. [CrossRef] 131. Barrett, G.L.; Bennie, A.; Trieu, J.; Ping, S.; Tsafoulis, C. The chronology of age-related spatial learning impairment in two rat strains, as tested by the Barnes maze. Behav. Neurosci. 2009, 123, 533–538. [CrossRef] 132. Sadeghian, A.; Fathollahi, Y.; Javan, M.; Shojaei, A.; Kosarmadar, N.; Rezaei, M.; Mirnajafi-Zadeh, J. Spatial Learning and Memory in Barnes Maze Test and Synaptic Potentiation in Schaffer Collateral-CA1 Synapses of Dorsal Hippocampus in Freely Moving Rts. Basic Clin. Neurosci. 2019, 10, 461–468. [CrossRef] [PubMed] 133. Lin, K.A.; Choudhury, K.R.; Rathakrishnan, B.G.; Marks, D.M.; Petrella, J.R.; Doraiswamy, P.M.; Alzheimer’s Disease Neuroimaging, I. Marked gender differences in progression of mild cognitive impairment over 8 years. Alzheimers Dement. (NY) 2015, 1, 103–110. [CrossRef] [PubMed] 134. Seshadri, S.; Wolf, P.A.; Beiser, A.; Au, R.; McNulty, K.; White, R.; D’Agostino, R.B. Lifetime risk of dementia and Alzheimer’s disease. The impact of mortality on risk estimates in the Framingham Study. Neurology 1997, 49, 1498–1504. [CrossRef] [PubMed] © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Sex-Dependent Wheel Running Effects on High Fat Diet Preference, Metabolic Outcomes, and Performance on the Barnes Maze in Rats.

09-05-2020

Yang, Tiffany Y,Gao, Zijun,Liang, Nu-Chu

eng

PMC6572263

International Journal of Environmental Research and Public Health Article The Effect of Ballistic Exercise as Pre-Activation for 100 m Sprints Maria H. Gil 1,2, Henrique P. Neiva 1,2 , Nuno D. Garrido 2,3, Felipe J. Aidar 4,5,6,7, Maria S. Cirilo-Sousa 8,9, Mário C. Marques 1,2 and Daniel A. Marinho 1,2,* 1 Department of Sport Sciences, University of Beira Interior, 6201-001 Covilhã, Portugal; maria.helena.gil@hotmail.com (M.H.G.); henriquepn@gmail.com (H.P.N.); mariomarques@mariomarques.com (M.C.M.) 2 Research Center in Sports Sciences, Health Sciences and Human Development, CIDESD, 6200-001 Covilhã, Portugal; ndgarrido@gmail.com 3 Department of Sports, Exercise and Health Sciences, University of Trás-os-Montes e Alto Douro, 5001-801 Vila Real, Portugal 4 Department of Physical Education, Federal University of Sergipe - UFS, São Cristovão, SE 49100-000, Brazil; fjaidar@gmail.com 5 Post Graduate Program in Master’s level in Physical Education, Federal University of Sergipe-UFS, São Cristovão, SE 49100-000, Brazil 6 Post Graduate Program in Doctorade and Master’s level in Physiological Sciences, Federal University of Sergipe - UFS, São Cristovão, SE 49100-000, Brazil 7 Group of Studies and Research of Performance, Sport, Health and Paralympic Sports - GEPEPS, the Federal University of Sergipe - UFS, São Cristovão, SE 49100-000, Brazil 8 Associate Graduate Program in Physical, Department of Physical Education, Federal University of Paraíba, João Pessoa, PB 58051-900, Brazil; helpcirilo@yahoo.com.br 9 Department of Physical Education, Regional University of Cariri, Crato, CE 63105-010, Brazil * Correspondence: marinho.d@gmail.com; Tel.: +35-1-275329153 Received: 23 April 2019; Accepted: 21 May 2019; Published: 24 May 2019



 Abstract: The benefits of warm-up in sports performance has received a special interest in the current literature. However, there is a large gap of knowledge about the tasks to be performed, specifically in the real competitive environment. The purpose of the study was to verify the acute effects of a warm-up including ballistic exercises in 100 m running performance. In addition, a second 100 m trial was assessed to better understand the warm-up effects in training and competition. Eleven men (25.4 ± 6.2 years of age, 1.76 ± 0.08 m of height, 78.2 ± 8.6 kg of body mass) were submitted to three different protocols, in a randomized order: no warm-up (NWU), typical warm-up (WU) and WU complemented with ballistic exercises (PAP). Biomechanical, physiological and psychophysiological variables were assessed. Differences were found between the three conditions assessed in the first 100 m sprint with 7.4% and 7.6% faster performances after the WU and PAP, compared to NWU. Stride length was higher in the second part of the 100 m after PAP compared with WU. These results highlight the positive effects of warm-up for sprinting performance. The inclusion of ballistic exercises, besides being used to improve sprint performance, can increase stride length in the final of the 100 m race. Keywords: warm-up; performance; repeated-sprint; physiology; biomechanics 1. Introduction Warm-up practices have been used to prepare the athlete for training and/or competition [1]. It is believed that a well-designed warm-up causes physiological changes and helps the athlete to Int. J. Environ. Res. Public Health 2019, 16, 1850; doi:10.3390/ijerph16101850 www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2019, 16, 1850 2 of 12 increase the mental focus for the next task, allowing them to optimize the performance [2]. The main effects of warming up derived from increased body temperature and from the muscle movement, both contributing to decreased joint and muscle stiffness, improved nerve conduction rate, efficient metabolic reactions, increased blood flow to the active muscles, increased oxygen uptake and to potentiate post-activation mechanisms [3,4]. There has been an increase in interest in the warm-up issue, evidenced by the number of recent studies and most reporting high benefits for performance in different sports and activities [5–7]. Specifically, in running, it is now well stablished that warm-up improves sprint performance [1,6,8]. Usually, warm-up included a brief period of low intensity aerobic (e.g., light to submaximal running) and stretching exercises, followed by specific exercises related with the following activity and/or sport [1,9]. During this specific phase of warm-up, the coaches and the athletes have been using several exercises for the same purpose, such as dynamic and static stretching [10], agility exercises and ballistics [11]. The last of these, the ballistic exercises, are a recent trend of specific warm-up and are believed to cause a post-activation potentiation phenomenon, thus enhancing the performance [12,13]. Researchers have looked at the post-activation potentiation phenomenon, suggesting that it might improve muscle power manifestations [14,15]. This increase in force production usually happens after a maximum or near maximal muscle stimulation [10]. Post-activation potentiation seems to augment muscle force generating capacity as a result of the previous contractile history of the muscle cells involved in the previous contraction [14]. There is an acute effect that increases the speed of conduction of the nerve impulse to the muscle, increases the number of recruited motor units and improves the interaction mechanism of contractile filaments [16]. The main mechanisms responsible for this are not totally clear, but studies attributed improvements to the increased phosphorylation of the myosin regulatory light chain [17–19]. Post-activation potentiation seems to cause neuromuscular changes and improves type II muscle fiber activity, thus favoring performance in short-term maximal efforts [13]. An improvement of 3% was found in 40 m sprints after performing back squats at 85% of 1 repetition maximum (1RM) [20]. Improvements of 2% and 3% were also found in 10 and 30 m sprints after 10 repetitions of half back squat exercise at 90% 1RM [21]. Nevertheless, Kilduff et al. [22] found that one set of three repetitions of a squat exercise at 87% 1RM did not improve 15 m swimming performance, compared to a traditional in-water warm-up. Previous research mainly focused on high external loads of strength exercise during warm-up and it is known that it cannot be applied in a real competition context [13,20,23]. There is a real need for understanding the effects of the post-activation potentiation using some usual tasks that can be reproduced in a real competition venue. The first studies on this revealed that including depth jumping in the warm-up protocol increased both maximal strength [24] and vertical jump [25,26]. Byrne, Kenny and O’Rourke [27] concluded that the addition of three depth jumps resulted in a 5% improvement of 20 m running compared to a traditional warm-up. However, little is known regarding when these ballistic exercises are used before Olympic racing distances, such as the 100 m. Moreover, little is known about the effects of using post-activation potentiation strategies on the biomechanical variables during running. Running performance depends on the stride parameters and, for instance, the optimal ratio between stride length (SL) and stride frequency (SF) enable maximal sprinting velocity and efficiency [28]. This relationship is conditioned by the neuromuscular regulation of movement, morphological characteristics, motor abilities and energy substrates [29,30], all of which can be influenced by warm-up tasks [1,7,13]. Therefore, it was hypothesized that a warm-up that included ballistic exercises would improve 100 m running performance, by changing the stride parameters (SL and SF) and physiological response. So, the primary aim of the current study was to verify the acute effects of a warm-up including ballistic exercises inducing a post-activation potentiation, easy to apply in a real competition context, in 100 m running performance. In addition, a second 100 m trial was assessed to better understand the warm-up effects during competition and training. To the best of our knowledge, no previous investigation has used a second repetition, and this is important to understand the neuromuscular and metabolic responses, helping to develop optimized training strategies. Repeated efforts have been used as Int. J. Environ. Res. Public Health 2019, 16, 1850 3 of 12 determinants for success in a wide range of sports and may be associated with neuromuscular and metabolic factors that influence performance [31,32]. The primary outcomes for our study were the 100 m running performance (time) and biomechanical variables (SL and SF). Secondary outcomes included physiological (lactate concentration and heart rate) and psychophysiological (ratings of perceived effort) variables. 2. Materials and Methods 2.1. Participants Eleven men aged 20–36 years (mean ± SD: 25.4 ± 6.2 years of age, 1.76 ± 0.08 m of height, 78.21 ± 8.59 kg of body mass) volunteered to participate in this study. Participants were physically active sport science students. Each individual was asked to report any previous illness, injury or other physical issue that would hinder their performance. Participants were included on the basis that they were healthy, injury free, and engaged in physical activity regularly with an experience of running and testing for the last two years, although they were not competitive sprinters. Criteria of exclusion from the study was the evidence of any medical or orthopedic problem, a self-reported fitness classification below moderately active, or any other self-reported issue that would endanger their own health (assessed via questionnaire). After local ethics board approval, ensuring compliance with the Declaration of Helsinki, the subjects were informed about the study procedures, and a written informed consent was signed. 2.2. Design The purpose of the present study was to evaluate the effects of typical warm-up procedures (WU), the inclusion of post-activation potentiation exercises (PAP) and no warm-up (NWU) on 100 m running performance, analyzing biomechanical, physiological and psychophysiological variables. Each participant completed two 100 m time-trials after each warm-up condition, in a randomized order, separated by 48 h. The WU design was based on literature recommendations [8,32] and included a low intensity aerobic component followed by specific running tasks. The PAP protocol included lower body ballistic exercises according to previous suggestions [33] after completing WU. During the NWU condition, the subjects were asked not to perform any type of action or movement prior to the 100 m sprint, remaining seated for 5 min. This design was able to test whether the inclusion of PAP strategies during warm-ups affected running performance. 2.3. Experimental Procedures All the procedures took place at the same time of the day (8:00–12:00 a.m.) for each participant under the same environmental conditions (~22 ◦C air temperature and ~60% of humidity) in an athletics track facility. The participants were familiarized with the warm-up procedures 72 h before the experiments, and they were reminded to maintain the same routines during the assessment days, avoiding strenuous exercise, and abstaining from consuming caffeine 48 h before testing. After arriving, each participant remained seated for 5 min and baseline measurements of heart rate (HR; Vantage NV; Polar, Kempele, Finland) and blood lactate concentrations([La−]; Lactate Pro LT 1710; Arkray Inc., Kyoto, Japan) were then assessed. Each volunteer was then randomly assigned to a warm-up protocol (Figure 1). Int. J. Environ. Res. Public Health 2019, 16, 1850 4 of 12 Figure 1. Schematic representation of the study design and testing procedures used. HR = heart rate; [La−] = blood lactate concentration; SF = stride frequency; SL = stride length; RPE = ratings of perceived exertion. 2.4. Warm-Up Protocols The warm-ups were designed based on research [1,8,32] and with the help of an experienced coach. The main difference between the warm-ups were the inclusion of lower-body ballistic exercises to stimulate post-activation potentiation. WU comprised 5 min of easy run (lower than 65% of estimated maximal HR), eight exercise drills (20 m repetitions with 10 s of recovery between them), such as rhythmic jumps from foot to foot, ankle drills, skipping drills, high-knee running. Then, these technical exercises were followed by 2 × 40 m running at gradually increasing intensity. In the PAP condition, the participants performed the WU followed by 2 sets of 5 depth jumps from a box of 70 cm height (3 min recovery) as suggested by Maloney et al. [33]. Each jump was performed by stepping off a box with one foot, landing with bent knees, then immediately jumping with maximal effort. The subjects were instructed to jump as quick and high as possible and to keep their hands on their hips to eliminate any contribution of arm swing [9,27]. 2.5. Time-Trial Performance Once the participants finished warming-up, they remained seated for 5 min before performing the 100 m time-trials. The subjects started from a standing position with the trunk bent forward and the lower limbs apart and slightly bent, positioned behind the starting line. After official commands, each participant started maximal running using a standing start with the lead-off foot placed 1 m behind the first timing gate. Times were measured by photocell timing gates (Brower photocells, Wireless Sprint System, USA) placed at 0, 50, and 100 m so that the times needed to cover 0–50 m (T0–50), 50–100 m (T50–100) and 0–100 m (T100) could be determined. After 10 min rest, the subjects performed a second 100 m sprint. 2.6. Kinematics All the procedures were recorded by two video cameras (Casio Exilim Ex-F1, f = 30 Hz) placed perpendicular to the running track. This enabled the acquisition of basic kinematic data such as the number of strides performed by each subject, the average stride length (SL) and average stride frequency (SF) calculations, between 0 and 50 m and between 50 and 100 m, using an open-source software (Kinovea, version 0.8.15). In running, a stride is defined as the time between two consecutive specific discrete events, normally defined as two consecutive foot strikes on the same foot. SL is defined as the distance traveled during a stride and SF is defined as the rate of strides per min. SF was converted to International System Units (Hz) for further analysis. Knowing the time performed and thus the running velocity, SL was determined from the division of running velocity by SF [34,35]. Int. J. Environ. Res. Public Health 2019, 16, 1850 5 of 12 2.7. Physiological and Psychophysiological Variables Capillary blood samples for [La−] assessment were collected from the fingertips before and 5 min after warm-ups, 3 and 6 min after each 100 m sprint to obtain the highest value ([La−]peak) [36], and after 15 min of recovery. HR was assessed before and after each warm-up (5 min), immediately after each time-trial (1 min) and after 15 min of recovery. Additionally, the rating of perceived exertion (RPE) was recorded using a 10-point modified Borg scale (Borg [37], modified by Foster et al. [38]) after warm-ups and after the time-trials. 2.8. Statistical Analysis Standard statistical methods were used for the calculation of mean ± SD, and 95% confidence intervals for all variables. The normality of all distributions was verified using Shapiro–Wilk tests. Data for all variables analyzed were homogeneous and normally distributed. The effect of the warm-up procedures was analyzed by an ANOVA for repeated measures, with sphericity checked using Mauchly’s test. When the assumption of sphericity was not met, the significance of F-ratios was adjusted according to the Greenhouse–Geisser procedure. Bonferroni post-hoc analysis were performed to further investigate the effect of each condition. All these statistical procedures were performed using IBM SPSS Statistics for Windows®, version 22.0 (Armonk, NY, USA: IBM Corp.) and the level of statistical significance was set at p ≤ 0.05. In addition, the effect size was calculated to estimate variance between conditions (partial eta squared: ηp2) and Hedges’ g (ES) for within-subjects’ comparisons using the Excel spreadsheet by Lakens [39]. ES values of 0.20, 0.60, 1.20 and 2.00 were considered small, moderate, large and very large magnitudes, respectively [40]. For ηp2, cut-off values were interpreted as 0.01 for small, 0.09 for moderate and 0.25 for large. 3. Results Before warm-up, the physiological variables were not different between conditions. Baseline measurements of HR (70 ± 7 bpm vs. 69 ± 7 bpm vs. 70 ± 7 bpm; F = 0.35, p = 0.71, ηp2 = 0.04) and [La−] (2.5 ± 0.6 mmol·L−1 vs. 2.5 ± 0.6 mmol·L−1 vs. 2.5 ± 0.6 mmol·L−1; F = 0.41, p = 0.67, ηp2 = 0.04) were similar between the three conditions. Table 1 presents a comparison between the HR and the [La−] immediately after the warm-ups. It was possible to verify significant differences in HR (F = 19.80, p < 0.001, ηp2 = 0.69) and [La−] (F = 35.29, p < 0.00, ηp2 = 0.80), with higher values for either warm-ups compared with no warm-up condition. No differences were found in perceived exertion between warm-ups performed (WU: 4.27 ± 1.27 vs. PAP: 3.80 ± 1.40; p = 0.34, ES = 0.32). Table 1. Mean ± SD values (95% confidence interval) of physiological responses to no-warm-up (NWU), typical warm-up (WU) and post-activation potentiation warm-up (PAP) (n = 11). p-values and effect sizes (ES) are also presented. NWU vs. WU NWU vs. PAP WU vs. PAP NWU WU PAP p-Value ES p-Value ES p-Value ES HR (bpm) 72 ± 6 (68, 76) 99 ± 13 (89, 108) 91 ± 9 (86, 97) <0.001 ** 2.72 <0.001 ** 2.43 0.42 0.70 [La−] (mmol·L−1) 2.5 ± 0.6 (2.0, 2.9) 4.7 ± 1.1 (3.9, 5.5) 4.4 ± 1.0 (3.7, 5.1) <0.001 ** 2.48 <0.001 ** 2.27 0.63 0.27 Mean ± SD values (95% confidence limits). [La−] = blood lactate concentration. HR = heart rate. ** p ≤ 0.01. Int. J. Environ. Res. Public Health 2019, 16, 1850 6 of 12 Table 2 presents the results recorded in the first 100 m sprint after NWU, WU and PAP. Large differences were found between the three conditions assessed (F = 12.52, p = 0.005, ηp2 = 0.58) in the 100 m sprint. The participants were 7.44% and 7.57% faster after the WU and PAP, compared to NWU, respectively. Moreover, four of them were faster after WU and seven were faster after PAP. Table 2. Mean ± SD values of the first 100 m time trial, biomechanical and psychophysiological variables assessed during experimental protocols: no-warm-up (NWU), typical warm-up (WU) and with post-activation potentiation (PAP) (n = 11). NWU vs. WU NWU vs. PAP WU vs. PAP NWU WU PAP p-Value ES p-Value ES p-Value ES T0–50 (s) 7.30 ± 0.68 (6.85, 7.64) 7.01 ± 0.58 (6.68, 7.34) 7.00 ± 0.62 (6.61, 7.39) 0.34 0.44 0.39 0.44 1.00 0.02 T50–100 (s) 8.69 ± 0.69 (8.20, 9.18) 7.66 ± 0.73 (7.13, 8.18) 7.66 ± 0.91 (7.01, 8.31) 0.03 * 1.39 0.04 * 1.23 1.00 0.00 T100 (s) 15.99 ± 0.96 (15.30, 16.68) 14.67 ± 1.29 (13.75, 15.60) 14.66 ± 1.52 (13.58, 15.74) 0.01 *** 1.12 0.02 * 1.03 1.00 0.01 T0–50 SF (Hz) 1.97 ± 0.19 (1.84, 2.11) 2.08 ± 0.14 (1.97, 2.18) 2.04 ± 0.13 (1.95, 2.13) 0.12 0.64 0.55 0.42 0.15 0.28 T50–100 SF (Hz) 1.72 ± 0.21 (1.57, 1.87) 1.89 ± 0.11 (1.81, 1.96) 1.91 ± 0.10 (1.84, 1.98) 0.05 * 1.02 0.03 ** 1.17 0.77 0.18 T0–50 SL (m) 3.51 ± 0.32 (3.28, 3.75) 3.47 ± 0.34 (3.23, 3.71) 3.54 ± 0.36 (3.3, 3.86) 0.74 0.12 1.00 0.08 0.01 ** 0.19 T50–100 SL (m) 3.40 ± 0.33 (3.16, 3.63) 3.51 ± 0.42 (3.21, 3.81) 3.47 ± 0.42 (3.17, 3.77) 0.10 0.28 0.42 0.18 0.48 0.09 HR (bpm) 148 ± 24 (131, 165) 156 ± 22 (140, 172) 162 ± 18 (149, 175) 1.00 0.33 0.43 0.64 0.84 0.29 [La−]peak (mmol·L−1) 7.6 ± 1.8 (6.3, 8.8) 8.5 ± 1.3 (7.5, 9.4) 8.9 ± 1.5 (7.8, 10.1) 0.38 0.56 0.32 0.75 1.00 0.27 RPE 6 ± 2 (5, 7) 7 ± 1 (6, 8) 7 ± 1 (6, 7) 1.00 0.35 0.76 0.30 1.00 0.08 Mean ± SD values (95% confidence limits). HR = heart rate. [La−] = blood lactate concentration. RPE = ratings of perceived exertion. ** p ≤ 0.01 and * p ≤ 0.05. Warm-ups assessed resulted also in large effects in the SF during the first 50 m (F = 3.81, p = 0.07, ηp2 = 0.30) and the second 50 m (F = 9.29, p = 0.01, ηp2 = 0.51) of the time-trial. The SL showed to be clearly different only in the second 50 m of the time-trial (F = 4.14, p = 0.03, ηp2 = 0.32). After trial, no significant differences were found in [La−] values (F = 2.16, p = 0.14, ηp2 = 0.19), HR (F = 1.20, p = 0.32, ηp2 = 0.12) and RPE values (F = 0.18, p = 0.73, ηp2 = 0.02). In the second 100 m sprint (Table 3), no differences were found between warm-ups condition (F = 0.58, p = 0.50, ηp2 = 0.06). Nevertheless, we verified that there was a 6.12% improvement from the first to the second sprint of 100 m in the NWU condition, while the same did not occur in the other conditions. The different responses to each warm-up condition in the 100 m time trials can be easily confirmed in Figure 2. Int. J. Environ. Res. Public Health 2019, 16, 1850 7 of 12 Table 3. Mean ± SD values of the second 100 m time-trial, biomechanical and psychophysiological variables assessed during experimental protocols: no-warm-up (NWU), typical warm-up (WU) and with post-activation potentiation (PAP) (n = 11). NWU vs. WU NWU vs. PAP WU vs. PAP NWU WU PAP p-Value ES p-Value ES p-Value ES T0–50 (s) 7.16 ± 0.59 (6.73, 7.58) 7.03 ± 0.56 (6.63, 7.43) 6.97 ± 0.59 (6.55, 7.39) 0.80 0.22 0.32 0.31 1.00 0.10 T50–100 (s) 7.76 ± 0.61 (7.33, 8.20) 7.70 ± 0.82 (7.11, 8.29) 7.78 ± 0.98 (7.08, 8.48) 1.00 0.08 1.00 0.02 1.00 0.09 T100 (s) 14.92 ± 1.16 (14.09, 15.75) 14.73 ± 1.36 (13.76, 15.70) 14.75 ± 1.52 (13.67, 15.84) 1.00 0.14 1.00 0.12 1.00 0.01 T0–50 SF (Hz) 1.98 ± 0.16 (1.87, 2.10) 2.04 ± 0.09 (1.97, 2.11) 2.02 ± 0.12 (1.94, 2.10) 0.42 0.46 0.82 0.27 1.00 0.18 T50–100 SF (Hz) 1.88 ± 0.14 (1.77, 1.98) 1.89 ± 0.12 (1.80, 1.97) 1.86 ± 0.13 (1.77, 1.95) 1.00 0.07 1.00 0.14 1.00 0.23 T0–50 SL (m) 3.56 ± 0.32 (3.33, 3.79) 3.51 ± 0.32 (3.28, 3.74) 3.59 ± 0.38 (3.32, 3.86) 0.74 0.15 1.00 0.08 0.18 0.22 T50–100 SL (m) 3.47 ± 0.39 (3.19, 3.75) 3.49 ± 0.38 (3.22, 3.75) 3.52 ± 0.47 (3.18, 3.85) 1.00 0.05 1.00 0.11 1.00 0.07 HR (bpm) 164 ± 10 (157, 171) 161 ± 29 (140, 182) 172 ± 20 (158, 186) 1.00 0.15 0.25 0.51 0.63 0.43 [La−]peak [mmol·L−1] 10.6 ± 1.6 (9.5, 11.7) 11.7 ± 1.6 (10.6, 12.8) 11.7 ± 1.9 (10.4, 13.0) 0.16 0.66 0.43 0.60 1.00 0.00 RPE 7 ± 2 (6, 8) 7 ± 1 (6, 8) 7 ± 1 (7, 8) 1.00 0.06 1.00 0.14 0.84 0.23 Mean ± SD values (95% confidence limits). HR = heart rate. [La−] = blood lactate concentration. RPE = ratings of perceived exertion. Figure 2. Mean changes (±90% CI) verified between conditions, specifically without warm-up (NWU), after typical warm-up (WU) and after WU complemented with ballistic exercises (PAP) in each 100 m time-trial. No significant differences were found in running kinematics during the second sprint, specifically regarding the SF in the first (F = 1.89, p = 0.18, ηp2 = 0.17) and second 50 m (F = 0.33, p = 0.72, ηp2 = 0.04), and regarding the SL in the first (F = 2.19, p = 0.14, ηp2 = 0.19) and second 50 m (F = 0.68, p = 0.52, ηp2 = 0.07). No significant differences were found in [La−] values (F = 2.83, p = 0.09, ηp2 = 0.24), HR (F = 1.21, p = 0.32, ηp2 = 0.12) and RPE values (F = 0.18, p = 0.73, ηp2 = 0.02) after the second time trial. Int. J. Environ. Res. Public Health 2019, 16, 1850 8 of 12 No differences were found after 15 min of recovery in the HR (NWU: 94 ± 8 bpm vs. WU: 97 ± 18 bpm vs. PAP: 96 ± 6 bpm; F = 0.17, p = 0.84, ηp2 = 0.02) and in the [La−] values (7.8 ± 1.3 mmol·L−1 vs. 7.9 ± 1.3 mmol·L−1 vs. 7.7 ± 1.0 mmol·L−1; F = 0.37, p = 0.70, ηp2 = 0.04). 4. Discussion The main purpose of the current study was to verify the acute effects of a warm-up including ballistic exercises, easy to apply on a real competition context, in 100 m running performance. It was intended to benefit from some post-activation potentiation, and thus optimize sprint running performance. This hypothesis was partially confirmed by the increased performance verified in the first sprint compared with the non-existence of warm-up. Nevertheless, by including some post-activation potentiation strategies such as the ballistic exercises, there were no additional effects in performance compared to the typical warm-up procedures. These results are in accordance with previous scientific evidence that reported optimized sprint performances after a typical warm-up or a post-activation potentiation warm-up (e.g., [23,41]) but failed to evidence additional improvement in performances after the use of ballistic exercises, as expected (e.g., [42,43]). Both warm-ups resulted in higher SF in the second part of the first time-trial compared with no warm-up. Interestingly, SL was higher in the second part of the 100 m after PAP compared with WU. This suggest that there are some specific technical adaptations that occur in response to different warm-up stimulations. The warm-up is intended to optimize the athletes’ preparedness, by increasing temperature, blood flow and muscle and metabolic efficiency to produce faster responses which are determinant to performance [1,7]. The ability of the muscle to produce force can be acutely modified by warm-up by including some conditioning muscle contractions [22]. The post-activation potentiation elicits transient improvements in performance and has been investigated as a strategy to include during warm-up for increasing performance [23,41]. The common exercises related to potentiation post activation phenomenon have used heavy-load (75–95% 1RM) resistance exercise [23]. However, ballistic exercises can be used as alternative since these are usually related with type II motor unit recruitment [44]. In fact, ballistic activities are more practical and feasible before competition compared to exercises requiring high-intensity external loads. That was the main reason for the assessment of ballistic exercises during warm-up in the current studies. Recent studies found some benefits by using depth jumping during the warm-up protocol to both maximal strength [24], sprint performance [42] and vertical jump [26,42]. However, to the best of our knowledge, no studies evaluated this warm-up strategy when applied to official running distances such as the 100 m race and tried to understand the biomechanical responses during the race. The results showed that the 100 m running performance was positively influenced by the warm-up. All the participants performed better after either warm-ups and, despite no statistically significant differences found between WU and PAP (p = 1.00, ES = 0.01), seven athletes recorded their best times after PAP. This could mean that there might be an individual response to PAP stimulation, as already highlighted by Till and Cook [42]. These authors found no differences in 20 m running performance by adding different post-activation potentiation strategies to usual warm-up, such as deadlift (5 repetitions at 5 repetitions maximum), or tuck jump (5 repetitions), or isometric maximum voluntary knee extensions (3 repetitions for 3 s) [42]. Nevertheless, others found positive effects on the use of ballistic exercises in running performance. Byrne et al. [27] verified that a brief warm-up of 5 min of running, dynamic stretches and three vertical jumps resulted in 5% better performance in 20 m sprint compared to the warm-up without the jumps. Accordingly, Lima et al. [45] found that 2 × 5 jumps from a height of 0.75 m caused 2% faster 50 m sprint performance. More recently, Turner et al. [44] found that the utilization of alternate-leg plyometric bounding provided an effective strategy for acutely improving sprint acceleration performance (10 and 20 m). Thus, it would be expected that there would be greater differences between the warm-ups performed, since the use of ballistic exercises during warm-up have been suggested as potentiating performance in explosive and short-term efforts [14,15]. In fact, most studies looked at race distances markedly lower than that used in the current study. This Int. J. Environ. Res. Public Health 2019, 16, 1850 9 of 12 longer distance might have caused the potentiation effects to disappear among other determinants of performance [46]. Maximal running performance results from an optimal ratio between SF and SL [28]. Some studies claimed SL to be the most influencing variable for maximal running velocity [47] while others suggested the SF [28]. Nevertheless, it is a fact that the runners adjust the SL and the SF to run most efficiently, optimizing velocity according to their own characteristics [48]. In the current study, better sprint times after warm-up could be caused by the ability to maintain a higher SF on final 50 m of the 100 m sprint without compromising the SL values. This situation did not occur in the NWU condition. Our results corroborated with previous research that suggested that there is a biomechanical adaptation in response to different warm-up procedures [2,49]. Interestingly, the running kinematics showed to be different in response to WU or PAP. In the PAP condition, the participants showed greater SL in the beginning of the race, contrarily to the SF that showed to be lower, compared with the WU condition. The PAP seemed to acutely stimulate the force required for an increased SL and perhaps improving the efficiency of the movement, that remained higher in the beginning of the second sprint. The effects of warm-up on acute motor learning and on sensorimotor responses could lead to different biomechanical movement patterns after different warm-ups [49]. Our WU ended with some specific running exercises and the PAP ended with jumps. It is a fact that the running exercises and running acceleration exercises could have prepared the participants to perform higher SF, while the jumps generated a greater capacity to exert muscular power, hence more effective force in less time and thereupon greater SL. So, this different biomechanical running adaptation might be partially explained by the specificity of the preload stimulus, since the vertical jump is biomechanically different from horizontal running. The physiological variables showed an increased response to warm-up, with higher HR and [La−] values after warm-up, and within the range of values that some authors suggested to be adequate for a proper warm-up [49,50]. This perhaps explain the better response in the first sprint after either warm-up procedures. Nevertheless, those differences disappeared after the first time-trial, which may be seen as a specific warm-up stimulus that in some way places the participant at a similar preparation level. The first sprint enhanced the neuromotor excitability that resulted in performance optimization in a second 100 m sprint [6,31]. The non-existence of differences between HR and [La−] values might suggest that PAP stimulation by the ballistic exercises used were not enough to induce physiological stress. Once again, this could be caused by the lack of specificity of the jumps and/or an insufficient load to stimulate some higher responses in PAP. We should be aware of a possible individualized effect of PAP stimulation, that was already documented before [15,51]. Moreover, we could speculate that the interval after PAP was not adequate for each runner [52,53]. It is known that the PAP effect may last for 5 to 10 min [53] and within this period, there are different moments of maximal potentiation for each individual [23]. However, our results were reliable and enlightening about the use of both warm-up procedures and that PAP could be used as an alternative to traditional warm-up. Some limitations, however, should be addressed. In fact, our results could not entirely be extrapolated to performance of higher skilled sprints during official events since the participants were not sprint specialists/athletes and it is known that post-activation potentiation could be influenced by training levels [19]. Also, further studies should include a larger number of participants and include females to clarify some of the analyzed findings. However, we took several steps to strengthen our statistical analysis as described in the statistical section. Future research should investigate different post-activation potentiation strategies (e.g., combining different jumps or short-term sprints) and different recovery times between the warm-up and the race. Moreover, other evaluation methods could be used to complement our measures and to deepen our findings, such as body temperature and other biomechanical variables (e.g., contact time and horizonal forces production). Considering our limitations, readers should interpret our results with discernment. Even so, the current findings are still relevant for coaches and researchers for increased knowledge on warm-up and the effects on performance. Int. J. Environ. Res. Public Health 2019, 16, 1850 10 of 12 5. Conclusions The results suggested that 100 m running performance was positively influenced by warm-up procedures, evidenced by the best results after the WU and the PAP compared to the NWU condition. Moreover, our results suggested that 100 m is equally optimized after WU or PAP, but with different running kinematics. Thus, in support of our original hypotheses, we have demonstrated that warming up benefits the 100 m running performance and that ballistic exercises, easy to perform by using body mass, can be used as an alternative to typical warm-up procedures. Some practical applications can be drawn. It seems clear that 100 m sprinters should warm-up for better competitive and training performances. When no warm-up is possible, a single 100 m trial can be enough to stimulate and prepare the athlete for that unusual situation. Yet, it is usually possible to warm-up before the race or training session and in this case, the PAP could be included in the warm-up to potentiate some individual benefits. Moreover, if the individual 100 m race strategy depends on having a higher SF, a typical warm-up should be used, whereas if higher SL is needed, the warm-up including ballistic exercises should be used. The current results alert coaches and researchers the need for tailored and customized warm-up designs and specifically post-activation potentiation strategies during warm-up. The current study took a novel approach to warm-up research by examining the effects of including post-activation potentiation exercises (i.e., ballistic exercises) in running performance and in running stride kinematics. Author Contributions: Conceptualization, H.P.N., D.A.M. and M.C.M.; methodology, H.P.N., D.A.M.; software, M.H.G.; validation, H.P.N., D.A.M. and M.C.M.; formal analysis, H.P.N. and D.A.M.; investigation, M.H.G., D.A.M., N.D.G., M.S.C.-S. and F.J.A.; resources, N.D.G., F.J.A. and M.S.C.-S.; data curation, M.H.G. and H.P.N.; writing—original draft preparation, H.P.N. and M.H.G.; writing—review and editing, H.P.N., N.D.G., F.J.A. and D.A.M.; visualization, H.P.N., D.A.M. and M.C.M.; supervision, M.C.M. and D.A.M. Funding: This research received no external funding. Acknowledgments: This work was supported by national funding through the Portuguese Foundation for Science and Technology, I.P., under project UID/DTP/04045/2019 and the European Fund for Regional Development (FEDER) allocated by European Union through the COMPETE 2020 Programme (POCI-01-0145-FEDER-006969)–competitiveness and internationalization (POCI). Conflicts of Interest: The authors declare no conflict of interest. References 1. Silva, L.M.; Neiva, H.P.; Marques, M.C.; Izquierdo, M.; Marinho, D.A. Effects of warm-up, post-warm-up, and re-warm-up strategies on explosive efforts in team sports: A systematic review. Sports Med. 2018, 48, 2285–2299. [CrossRef] [PubMed] 2. Neiva, H.P.; Marques, M.C.; Barbosa, T.M.; Izquierdo, M.; Viana, J.L.; Teixeira, A.M.; Marinho, D.A. The effects of different warm-up volumes on the 100 m swimming performance: A randomized crossover study. J. Strength Cond. Res. 2015, 29, 3026–3036. [CrossRef] [PubMed] 3. Kilduff, L.P.; West, D.J.; Williams, N.; Cook, C.J. The influence of passive heat maintenance on lower body power output and repeated sprint performance in professional rugby league players. J. Sci. Med. Sport 2013, 16, 482–486. [CrossRef] [PubMed] 4. Swanson, J. A functional approach to warm-up and flexibility. Strength Cond. J. 2006, 28, 30–36. [CrossRef] 5. Neiva, H.P.; Marques, M.C.; Fernandes, R.J.; Viana, J.L.; Barbosa, T.M.; Marinho, D.A. Does warm-up have a beneficial effect on 100 m freestyle? Int. J. Sports Physiol. Perform. 2014, 9, 145–150. [CrossRef] 6. Marinho, D.A.; Gil, M.H.; Marques, M.C.; Barbosa, T.M.; Neiva, H.P. Complementing warm-up with stretching routines: Effects in sprint performance. Sports Med. Inter. Open 2017, 1, E101–E106. [CrossRef] [PubMed] 7. Neiva, H.P.; Marques, M.C.; Barbosa, T.M.; Izquierdo, M.; Marinho, D.A. Warm-up and performance in competitive swimming. Sports Med. 2014, 44, 319–330. [CrossRef] 8. Zois, J.; Bishop, D.J.; Ball, K.; Aughey, R.J. High-intensity warm-ups elicit superior performance to current soccer warm-up routine. J. Sci. Med. Sport 2011, 14, 522–528. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2019, 16, 1850 11 of 12 9. Andrade, D.C.; Henriquez-Olguín, C.; Beltrán, A.R.; Ramírez, M.A.; Labarca, C.; Cornejo, M.; Álvarez, C.; Ramírez-Campillo, R. Effects of general, specific and combined warm-up on explosive muscular performance. Biol. Sport 2015, 32, 123–128. [CrossRef] [PubMed] 10. Kallerud, H.; Gleeson, N. Effects of stretching on performances involving stretch-shortening cycles. Sports Med. 2013, 43, 733–750. [CrossRef] 11. Perrier, E.T.; Pavol, M.J.; Hoffman, M.A. The acute effects of a warm-up including static or dynamic stretching on countermovement jump height, reaction time, and flexibility. J. Strength Cond. Res. 2011, 25, 1925–1931. [CrossRef] 12. Blagrove, R.C.; Holding, K.M.; Patterson, S.D.; Howatson, G.; Hayes, P.R. Efficacy of depth jumps to elicit a post-activation performance enhancement in junior Endurance runners. J. Sci. Med. Sport 2019, 22, 239–244. [CrossRef] [PubMed] 13. Gil, M.H.; Neiva, H.P.; Sousa, A.C.; Marques, M.C.; Marinho, D.A. Current approaches on warming up for sports performance: A critical review. Strength Cond. J. 2019. [CrossRef] 14. Hodgson, M.; Docherty, D.; Robbins, D. Post-activation potentiation: Underlying physiology and implications for motor performance. Sports Med. 2005, 35, 585–595. [CrossRef] 15. Seitz, L.B.; de Villarreal, E.S.; Haff, G.G. The temporal profile of postactivation potentiation is related to strength level. J. Strength Cond. Res. 2014, 28, 706–715. [CrossRef] [PubMed] 16. Saez Saez de Villarreal, E.; González-Badillo, J.J.; Izquierdo, M. Optimal warm-up stimuli of muscle activation to enhance short and long-term acute jumping performance. Eur. J. Appl. Physiol. 2007, 100, 393–401. [CrossRef] 17. MacIntosh, B.R.; Robillard, M.E.; Tomaras, E.K. Should postactivation potentiation be the goal of your warm up? J. Appl. Physiol. Nutr. Metabol. 2012, 37, 546–550. [CrossRef] [PubMed] 18. Tillin, N.A.; Bishop, D. Factors modulating post-activation potentiation and its effect on performance of subsequent explosive activities. Sports Med. 2009, 39, 147–166. [CrossRef] 19. Xenofondos, A.; Laparidis, K.; Kyranoudis, A.; Galazoulas, C.; Bassa, E.; Kotzamanidis, C. Post-activation potentiation: Factors affecting it and the effect on performance. J. Phys. Educ. Sport 2010, 28, 32–38. 20. Rahimi, R. The acute effects of heavy versus light-load squats on sprint performance. Ser. Phys. Educ. Sport. 2007, 5, 163–169. 21. Chatzopoulos, D.E.; Michailidis, C.J.; Giannakos, A.K.; Alexiou, K.C.; Patikas, D.A.; Antonopoulos, C.B.; Kotzamanidis, C.M. Post-activation potentiating effects after heavy resistance exercise on running speed. J. Strength Cond. Res. 2007, 21, 1278–1281. [CrossRef] 22. Kilduff, L.P.; Cunningham, D.J.; Owen, N.J.; West, D.J.; Bracken, R.M.; Cook, C.J. Effect of postactivation potentiation on swimming starts in international sprint swimmers. J. Strength Cond. Res. 2011, 25, 2418–2423. [CrossRef] [PubMed] 23. Kilduff, L.P.; Owen, N.; Bevan, H.; Bennett, M.; Kingsley, M.I.; Cunningham, D. Influence of recovery time on post-activation potentiation in professional rugby players. J. Sport Sci. 2008, 26, 795–802. [CrossRef] [PubMed] 24. Masamoto, N.; Larsen, R.; Gates, T.; Faigenbaum, A. Acute effects of plyometric exercise on maximum squat performance in athletes. J. Strength Cond. Res. 2003, 17, 68–71. [PubMed] 25. Hilfiker, R.; Hubner, K.; Lorenz, T.; Marti, B. Effects of drop jumps added to the warm-up of elite sport athletes with a high capacity for explosive force development. J. Strength Cond. Res. 2007, 21, 550–555. [CrossRef] 26. Stieg, J.L.; Faulkinbury, K.J.; Tran, T.T.; Brown, L.E.; Coburn, J.W.; Judelson, D.A. Acute effects of depth jump volume on vertical jump performance in collegiate women soccer players. Kinesiology 2011, 43, 25–30. 27. Byrne, P.J.; Kenny, J.; O’Rourke, B. Acute potentiating effect of depth jumps on sprint performance. J. Strength Cond. Res. 2014, 28, 610–615. [CrossRef] [PubMed] 28. Krzysztof, M.; Mero, A. A kinematics analysis of three best 100 m performances ever. J. Hum. Kinet. 2013, 36, 149–160. [CrossRef] 29. Coh, M.; Milanovic, D.; Kampmiller, T. Morphologic and kinematic characteristics of elite sprinters. Coll. Antropol. 2001, 25, 605–610. 30. Prampero, P.; Fusi, S.; Sepulcri, J.; Morin, B.; Belli, A.; Antonutto, G. Sprint running: A new energetic approach. J. Exp. Biol. 2005, 208, 2809–2816. [CrossRef] 31. Spencer, M.; Bishop, D.; Dawson, B.; Goodman, C. Physiological and metabolic responses of repeated-sprint activities: Specific to field-based team sports. Sports Med. 2005, 35, 1025–1044. [CrossRef] [PubMed] 32. Taylor, J.M.; Weston, M.; Portas, M.D. The effect of a short practical warm-up protocol on repeated sprint performance. J. Strength Cond. Res. 2013, 27, 2034–2038. [CrossRef] Int. J. Environ. Res. Public Health 2019, 16, 1850 12 of 12 33. Maloney, S.J.; Turner, A.N.; Fletcher, I.M. Ballistic exercise as a pre-activation stimulus: A review of the literature and practical applications. Sports Med. 2014, 44, 1347–1359. [CrossRef] 34. Hamill, J.; Knutzen, K.M. Biomechanical Basis of Human Movement; Lippincott Williams & Wilkins: Philadelphia, PA, USA, 2009. 35. Hunter, J.P.; Marshall, R.N.; McNair, P.J. Interaction of step length and step rate during sprint running. Med. Sci. Sports Exerc. 2004, 36, 261–271. [CrossRef] 36. Goodwin, M.L.; Harris, J.E.; Hernández, A.; Gladden, L.B. Blood Lactate Measurements and Analysis during Exercise: A Guide for Clinicians. J. Diabetes Sci. Technol. 2007, 1, 558–569. [CrossRef] [PubMed] 37. Borg, G. Borg’s Perceived Exertion and Pain Scales; Human Kinetics Publishers: Champaign, IL, USA, 1998. 38. Foster, C.; Florhaug, J.A.; Franklin, J.; Gottschall, L.; Hrovatin, L.A.; Parker, S.; Doleshal, P.; Dodge, C. A new approach to monitoring exercise training. J. Strength Cond. Res. 2001, 15, 109–115. [PubMed] 39. Lakens, D. Calculating and reporting effect sizes to facilitate cumulative science: A practical primer for t-tests and ANOVAs. Front. Psychol. 2013, 4, 1–12. [CrossRef] 40. Hopkins, W.G.; Marshall, S.W.; Batterham, A.M.; Hanin, J. Progressive statistics for studies in sports medicine and exercise science. Med. Sci. Sports Exerc. 2009, 41, 3–13. [CrossRef] 41. West, D.; Cunningham, D.; Bevan, H.; Crewther, B.; Cook, C.; Kilduff, L. Influence of active recovery on professional rugby union player’s ability to harness postactivation potentiation. J. Sports Med. Phys. Fitness 2013, 53, 203–208. 42. Till, K.A.; Cooke, C. The effects of postactivation potentiation on sprint and jump performance of male academy soccer players. J. Strength Cond. Res. 2009, 23, 1960–1967. [CrossRef] 43. Johnson, M.; Baudin, P.; Ley, A.L.; Collins, D.F. A Warm-Up Routine That Incorporates a Plyometric Protocol Potentiates the Force-Generating Capacity of the Quadriceps Muscles. J. Strength Cond. Res. 2019, 33, 380–389. [CrossRef] 44. Turner, A.P.; Bellhouse, S.; Kilduff, L.P.; Russell, M. Postactivation potentiation of sprint acceleration performance using plyometric exercise. J. Strength Cond. Res. 2015, 29, 343–350. [CrossRef] 45. Lima, B.J.; Marin, D.; Barquilha, G.; da Silva, L.; Puggina, E.; Pithon-Curi, T.; Hirabara, S.M. Acute effects of drop jump potentiation protocol on sprint and countermovement vertical jump performance. Hum. Mov. Sci. 2011, 12, 324–330. [CrossRef] 46. Gastin, P.B. Energy system interaction and relative contribution during maximal exercise. Sports Med. 2001, 31, 725–741. [CrossRef] 47. Mackala, K. Optimisation of performance through kinematic analysis of the different phases of the 100 meters. New Stud. Athlet. 2007, 22, 7–16. 48. Salo, A.I.; Bezodis, I.N.; Batterham, A.M.; Kerwin, D.G. Elite sprinting: Are athletes individually step-frequency or step-length reliant? Med. Sci. Sports Exerc. 2011, 43, 1055–1062. [CrossRef] 49. Neiva, H.P.; Marques, M.C.; Barbosa, T.M.; Izquierdo, M.; Viana, V.L.; Teixeira, A.M.; Marinho, D.A. Warm-up for sprint swimming: Race-pace or aerobic stimulation? A randomized study. J. Strength Cond. Res. 2017, 31, 2423–2431. [CrossRef] 50. Raccuglia, M.; Lloyd, A.; Filingeri, D.; Faulkner, S.H.; Hodder, S.; Havenith, G. Post-warm-up muscle temperature maintenance: Blood flow contribution and external heating optimization. Eur. J. Appl. Physiol. 2016, 116, 395–404. [CrossRef] [PubMed] 51. Wilson, J.M.; Duncan, N.M.; Marin, P.J.; Brown, L.E.; Loenneke, J.P.; Wilson, S.M.; Jo, E.; Lowery, R.P.; Ugrinowitsch, C. Meta-analysis of postactivation potentation and power: Effects of conditioning activity, volume, gender, rest periods and training status. J. Strength Cond. Res. 2013, 27, 854–859. [CrossRef] [PubMed] 52. Baudry, S.; Duchateau, J. Postactivation potentiation in a human muscle: Effect on the rate of torque development of tetanic and voluntary isometric contractions. J. Appl. Physiol. 2007, 102, 1394–1401. [CrossRef] [PubMed] 53. Sue, A.S.; Adams, K.J.; DeBeliso, M. Optimal timing for post-activation potentiation in women collegiate volleyball players. Sports 2016, 4, 27. [CrossRef] [PubMed] © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

The Effect of Ballistic Exercise as Pre-Activation for 100 m Sprints.

05-24-2019

Gil, Maria H,Neiva, Henrique P,Garrido, Nuno D,Aidar, Felipe J,Cirilo-Sousa, Maria S,Marques, Mário C,Marinho, Daniel A

eng

PMC5862462

RESEARCH ARTICLE Pilates training improves 5-km run performance by changing metabolic cost and muscle activity in trained runners Paula Finatto, Edson Soares Da Silva, Alexandre B. Okamura, Bruna P. Almada, Henrique B. Oliveira, Leonardo A. Peyre´-Tartaruga* Exercise Research Laboratory, Escola de Educac¸ão Fı´sica, Fisioterapia e Danc¸a, Universidade Federal do Rio Grande do Sul, Porto Alegre, Rio Grande do Sul, Brazil * leonardo.tartaruga@ufrgs.br Abstract Purpose Strength training improves distance running economy and performance. This finding is based predominantly on maximal and explosive strength programmes applied to locomotor muscles, particularly on the lower limbs. It is not certain whether a minimization of metabolic cost (Cmet) and an improvement in running performance is feasible with strength training of the postural and trunk muscles. Methods Using kinematic, neuromuscular and metabolic measurements of running at two different speeds before and after a 12-week Pilates training programme, we tested the hypothesis that core training might improve the running Cmet and performance of trained runners. Thirty-two individuals were randomly assigned to the control group (CG, n = 16) or the Pila- tes group (PG, n = 16). Results Confirming our hypothesis, a significant improvement (p<0.05) was observed for running performance in the PG (pre: 25.65±0.4 min; post: 23.23±0.4 min) compared to the CG (pre: 25.33±0.58 min; post: 24.61±0.52 min). Similarly, the PG (4.33±0.07 J.kg-1.m-1) had better responses than the CG (4.71±0.11 J.kg-1.m-1) during post-training for Cmet. These findings were accompanied by decreased electromyographic activity of the postural muscles at sub- maximal running intensities in the PG. Conclusions Overall, these results provide a rationale for selecting strength training strategies that target adaptations on specific postural and locomotor muscles for trained distance runners. PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 1 / 19 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Finatto P, Silva ESD, Okamura AB, Almada BP, Oliveira HB, Peyre´-Tartaruga LA (2018) Pilates training improves 5-km run performance by changing metabolic cost and muscle activity in trained runners. PLoS ONE 13(3): e0194057. https://doi.org/10.1371/journal.pone.0194057 Editor: Yury P. Ivanenko, Fondazione Santa Lucia Istituto di Ricovero e Cura a Carattere Scientifico, ITALY Received: November 28, 2017 Accepted: February 25, 2018 Published: March 21, 2018 Copyright: © 2018 Finatto et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: This work was supported by Coordenac¸ão de Aperfeic¸oamento de Pessoal de Nı´vel Superior (CAPES) and Conselho Nacional de Desenvolvimento Cientı´fico e Tecnolo´gico (CNPq). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Introduction From the cardiorespiratory perspective, running performance, particularly at long distances, depends on the interaction of different factors [1], including high maximum oxygen consump- tion (VO2max), the ability to sustain a high fraction of VO2max for long periods, and the ability to move economically [2]. The latter parameter is designated as metabolic cost (Cmet) and cor- responds to the oxygen consumption spent to move a certain distance by running at a submax- imal intensity. Considering a group of runners with a similar body mass, an individual with a low Cmet would spend less energy and consequently would have lower oxygen consumption (VO2) than a runner with a high Cmet at a certain running speed [3,4]. A lower Cmet may be achieved via aerobic endurance training programmes, aerobic endur- ance combined with strength training, and plyometric training [5,6]. Another aspect that may be related to Cmet is muscle activation, particularly that of muscles of the trunk and lower limbs. Behm et al.[7] observed that a greater activation of the obliquus externus abdominis muscle and erector muscles of the upper and sacral spine is required during running for the control of movements and that the activation pattern of these muscles may be associated with better per- formance. For this reason, a specific training programme can promote greater stability, which would decrease necessary muscle recruitment and consequently positively affect Cmet [8,9]. Pilates training (PT) has been widely used to strengthen trunk muscles. PT is based on six key principles: concentration, control, precision, flow, breathing, and centre of force [10]. The centre of force was originally designated powerhouse and refers to the extensor muscles of the spine and hip, the flexor muscles of the spine and hip, and the muscles of the pelvic floor [10]. The centre of force is strengthened to promote further stabilization of the hip and trunk and favour the integrity of the spine [11]. To the best of our knowledge, no previous studies have specifically addressed the effects of PT on running. However, training programmes for the stability of the core muscles, which correspond to the flexor and extensor muscles of the trunk, along with the deeper muscles that stabilize the trunk, have shown conflicting results when performed for six weeks. Stanton et al. [12] found significant improvements in core stability in team sport athletes after core training using Swiss balls; however, they found no significant differences in the activation of abdominal and extensor muscles of the spine, VO2max or Cmet. By contrast, Sato and Mokha [9] found no significant improvement in the dynamic stability of trained runners after a core-training pro- gramme but found a significant decrease in the time of completion for a 5-km run. Core training and PT aim to strengthen the muscles of the trunk and lower limbs. However, the principles inherent to PT are not used in core training, and these principles distinguish these two training modalities and can influence the results of PT. Specific training programmes can result in a better pattern of activation of the trunk muscles, which would provide more stable joints and reduce the need for co-contractions for stabilization. Consequently, these programmes could lead to decreased Cmet and, in turn, improved running performance. We hypothesized that metabolic cost and trunk muscle activation will be reduced and, consequently, running performance may be improved. Therefore, it is essential to study the effects of strengthening the muscles of the centre of force by PT on Cmet and on the muscle activation patterns and biomechanical parameters that could be indicative of improved Cmet because this strategy can consequently increase running performance. Materials and methods Experimental design To investigate the effects of mat PT in recreational runners, cardiorespiratory and neuromus- cular adaptations were compared between a group that underwent running training combined Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 2 / 19 Competing interests: The authors have declared that no competing interests exist. Abbreviations: Anova, Analysis of variance; BF, biceps femoralis; Cmet, metabolic cos; Cmet10, metabolic cost at 10 km.h-1; Cmet12, metabolic cost at 12 km.h-1; CG, control group; E, Easy; EMG, Electromyographic; GM, gluteus medius; HRVT2, heart rate at second ventilatory threshold; LD, latissimus dors; LO, Longissimus; I, Interval; M, moderate; MVIC, maximal voluntary isometric contraction; PG, Pilates group; OI, obliquus internus abdominis; OE, obliquus externus abdominis; PT, Pilates training; RMS, root mean square; SE, Standard error; T, threshold; VL, vastus lateralis; VO2, oxygen consumption; VO2max, maximum oxygen consumption; VT2, second ventilatory threshold; W, Watts. with PT and a control group that underwent only running training. Both groups (Pilates and control) were trained for 12 weeks and were evaluated before and after the training period. The post-training evaluations were performed 72 hours after the last training session, and the subjects completed the evaluations within 10 days with at least 48 hours between the test ses- sions. The same assessors who were blinded to the training groups conducted the test sessions, and the same equipment was used in all of the sessions. The subjects were instructed to main- tain their eating habits during the study period. Participants Fifty-eight subjects were interviewed after placement of an advertisement about the study in a major newspaper in Porto Alegre, Brazil. The 32 enrolled volunteers were randomly assigned into two groups by electronic randomization: control group (CG; n = 16) (mean±SE, age: 18.44 ± 0.52 years; body mass: 73.64 ± 10.79 kg; height: 176.66 ± 9.89 cm, percent fat: 10.81 ± 2.49%) and Pilates group (PG; n = 16) (mean ± SE, age: 18.42 ± 0.51 years; body mass: 70.71±10.90 kg; height: 175.07 ± 8.06 cm, percent fat: 9.34±1.98%). During data collection, the CG lost three subjects. During the training period, one subject from the PG was excluded because his rate of absence from training was higher than 20%. Therefore, 15 subjects in the CG and 13 subjects in the PG completed all phases of the study. The inclusion criteria were as follows: male, practice of running for at least six months before the study, with experience in 5-km running races, age between 18 and 28 years, and absence of medical restrictions. The exclusion criteria were as follows: subjects with experience in Pilates and subjects with hor- monal, metabolic, neuromuscular, and/or cardiac disorders. All the participants had a running experience of at most 9 months prior to the start of the study, with the main frequency of 2 times a week. Each individual signed a free and informed consent form. This study was con- ducted according to the Helsinki Declaration and was approved by the Ethics Committee of the Federal University of Rio Grande do Sul, Brazil under registration no. 965734. Procedures Running training. Subjects from the CG and PG participated in a 12-week racetrack training programme (see the training program in the supplementary material, S1 Table). Two sessions per week were performed. The periodization of running training was based on the second ventilatory threshold (VT2) obtained in a maximal effort test on a treadmill with maxi- mum oxygen consumption (VO2max) in a first session of data collection. Accordingly, the training periodization was based on the heart rate at VT2 (HRVT2) according to the intensity zones proposed by Daniels [13]: easy (E), 71–86%; moderate (M), 82–98%; threshold (T), 96– 100%; and interval (I), 107–109% of HRVT2 (S1 Table). The training sessions were held at the three racetracks of the School of Physical Education of the Federal University of Rio Grande do Sul (Escola de Educac¸ão Fı´sica, Fisioterapia e Danc¸a da Universidade Federal do Rio Grande do Sul). Classical mat Pilates training. The classic mat PT programme lasted 12 weeks. The sub- jects from the PG underwent the running training described above in addition to two one- hour weekly sessions of PT performed on days alternate to the days of the running training. The organization of the session and the intensity and volume of training were in accordance with the Manual of the Pilates Method Alliance (California, USA). The sessions were organized into an initial section (execution of PT fundamentals), a main section (execution of PT exer- cises), and a final section (relaxation). During the initial section, the fundamentals of PT were performed, and the exercises were selected according to the training period. Classic mat PT Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 3 / 19 consisted of three series, and the number of repetitions and sequences were defined as shown in Table 1. Maximum amplitude of the electromyographic signal during MVIC In a second session, for the evaluation of the maximum isometric amplitude of the electromyo- graphic (EMG) signal of the aforementioned muscles, the procedures began with electrode placement and skin preparation on the muscle surfaces of interest [14]. Surface electrodes with a 15-mm total diameter (MeditraceTM, Mainsfield, Canada) were used in a bipolar configura- tion, whose inter-electrode distance of 2 cm [15]. After this procedure, the locations of electrode placement for the longissimus (LO), gluteus medius (GM), vastus lateralis (VL), biceps femoralis (BF) and latissimus dorsi (LD) were deter- mined according to the recommendations of the SENIAM project (Surface ElectroMyoGraphy for the Non-Invasive Assessment of Muscles; [16]). In the obliquus internus abdominis (OI) muscle, the electrodes were placed two cm medially and inferiorly to the anterior superior iliac spine [17]. In the obliquus externus abdominis (OE) muscle, the electrodes were placed at mid-distance between the lower part of the rib cage and the anterior superior iliac spine [17]. The reference electrode was placed at the tuberosity of the tibia of the right leg. A level of resis- tance between the electrodes of up to 3000 O was accepted. After placement of the electrodes, the subjects were instructed to perform a warm-up by walking for 5 min on a treadmill. All of the subjects were instructed and encouraged to exert the maximum force in each isometric test Table 1. 12-week periodization of Pilates training. Week 1 Weeks 2 to 6 Weeks 6 to 12 Initial section Fundamentals 1 to 7 Fundamentals 5 to 12 Fundamentals 13 to 17 Main section Pre-Pilates Basic Mat Pilates Intermediate Mat Pilates Final section Relaxation Relaxation Relaxation Exercises that composed the various levels Fundamentals Pre-Pilates Basic Mat Pilates Intermediate Mat Pilates 1. Breathing 1. The Hundred 1.The Hundred 1. The Hundred 2. Imprinting 2. Roll Down 2. The Roll Up 2. The Roll Up 3. Pelvic Bowl 3. Roll Up 3. Single Leg Circles 3. Leg Circles 4. Knee Sway 4. Single Leg Circles 4. Rolling Like a Ball 4. Rolling Like a Ball 5. Knee Folds/Stirs 5. Rolling Like a Ball 5. Single Leg Stretch 5. Single Leg Stretch 6. Leg Slides 6. Single Leg Stretch 6. Double Leg Stretch 6. Double Leg Stretch 7. Spinal Bridging 7. Double Leg Stretch 7. Legs Up and Down 7. Single Straight Leg 8. Prone Hip Extension 8. Spine Stretch Forward 8. Spine Stretch Forward 8. Double Straight Leg 9. Cervical Nod 9. Saw 9. Criss-Cross 10. Nose Circles 10. Single Leg Kicks 10. Spine Stretch Forward 11. Head Float 11. Beats 11. Open Leg Rocker 12. Ribcage/Angel Arms 12. Double Leg Kicks 12. Corkscrew 13. Rotating Arms 13. Saw 14. Torso Twist 14. Neck Pull 15. Flight 15. Single Leg Kicks 16. Cat 16. Double Leg Kicks 17. Bowing 17. Neck Pull 18. Side Kicks Series 19. Teaser 20. Seal https://doi.org/10.1371/journal.pone.0194057.t001 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 4 / 19 against the mechanical strength of the Velcro strips to produce this force as fast as possible. Three measurements of the maximal voluntary isometric contraction (MVIC) were obtained in each muscle, with a duration of 5 s and an interval of 3 min between each measurement pre- and post-training and used to normalize the EMG activation during running in each eval- uation period. To obtain muscle activation, the EMG signal was captured by two electromyo- graphs (Miotool 400, Miotec, Porto Alegre, Brazil), with a sampling frequency of 2000 Hz in each channel, using Miograph software (Miotec, Porto Alegre, Brazil) for later analysis in SAD32 software (UFRGS, Porto Alegre, Brazil). The signal was filtered using a fifth-order But- terworth band-pass filter with cut-off frequencies between 20 and 500 Hz. After filtering, the plateau period of isometric activation for 1-s intervals was identified. The root mean square (RMS) value was obtained via Hamming windowing in 1-s intervals. The measurement with the highest RMS value was considered valid. EMG activation during running Fifteen minutes after the completion of the MVIC measurements, the running protocol was initiated with two runs at speeds of 10 and 12 km.h-1 performed in random order for 7 minutes each. The EMG signal was recorded in the last minute of the protocol on a treadmill at each speed evaluated using the data acquisition software Miograph (Miotec, Porto Alegre, Brazil). The kinematic data used for evaluation of the EMG signal at the different stride phases were also obtained in the last minute of each speed evaluated by recording the run with a Casio (EXLIM-ZR1000) video camera at a sampling rate of 120 Hz. These data were aligned with the EMG data using a light signal that generates a spike in the EMG signal in the channel specified for the alignment. Subsequently, the files with data on running and MVICs were exported for analysis in SAD32 software (UFRGS, Porto Alegre, Brazil). For the analysis of EMG activation during the runs, the same signal filtering procedure as that used for MVIC was applied. The RMS curve was obtained via Hamming windowing in 0.1-second intervals. Subsequently, the RMS signal was shifted and clipped from the signal emitted by the alignment system, in agree- ment with the video analysis. The times corresponding to the three stride phases (pre-activa- tion: 100 ms before the contact of the heel with the floor [18]; b) contact phase (contact of the heel with the ground until detachment of the heel from the ground); and c) swing phase (detachment of the heel from the ground until contact of the heel to the ground)) were then identified in the videos for clipping of the EMG signal obtained from five main strides. Subse- quently, the mean time was calculated from these clippings to obtain the mean RMS value for each subject and for each stride phase. RMS values representative of EMG activation in each of the three phases were expressed as a percentage of MVIC. The pattern of EMG activation of the seven muscles analysed during the runs pre-and post-training in the CG and PG was ana- lysed via a temporal analysis of the EMG signal in relation to time, considering that the nor- malized x-axis varied between 0 and 100% of the stride. For this purpose, the raw EMG signal was shifted according to the alignment system and was rectified and filtered using a fifth-order Butterworth low band-pass filter with a cut-off frequency of 10 Hz [19]. After filtering, the EMG signals from five main strides were obtained to calculate the mean curve. The mean curve of each subject was resampled in 200 points and exported to Excel (Microsoft, Redmond, USA) for the calculation of the mean curve between the subjects. Metabolic cost The treadmill tests at both speeds were performed concurrently with the collection of EMG data. For this purpose, the subjects remained at rest for 15 min in the sitting position and at rest for 5 min in the orthostatic position to determine the at rest heart hate and VO2 for Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 5 / 19 confirmation of the initiation of the test. The respiratory exchange ratio should be below 0.85 to ensure that the individual started from the same resting state in each phase of the test. Subse- quently, a 5-min warm-up was performed on a treadmill and was immediately followed by two additional 7-min stages, at running speeds of 10 and 12 km.h-1. These speeds were randomized, respecting a 5-min interval between the runs or until the heart hate returned to resting levels. The VO2 values were collected in the last 4 min of each run, and the last 3 min were included in the analysis. Data were collected using a gas analyser model VO2000 (Medgraphics, Ann Arbor, USA). The Wmet was considered the difference between the VO2 measured during exercise and the VO2 at rest, in relation to time. Because the unit of measure used was watts (W), this differ- ence was multiplied by the energy coefficient (20.9 J.mL-1) and divided by the time in seconds (60 s). The metabolic cost values relative to the speeds of 10 km.h-1 (Cmet10) and 12 km.h-1 (Cmet12) were calculated by dividing Wmet by the speed in m.s-1. Time of completion of the 5-km run In a third session, a 5-km run was performed by all of the subjects in a single test to determine the total time of completion of the test. The 5-km time was controlled using timers and was con- firmed by filming. The race was always held at the same time and with similar temperature and relative humidity conditions. All data can be seen in the supplementary material (S2 Table). Statistics The comparisons of run performance variables, metabolic variables, muscle activation, and sample characteristics between the groups and time factors were performed using the general- ized estimating equations model. Bonferroni’s complementary test was used to identify signifi- cant differences. The significance level was set at α<0.05, and the statistical package used was SPSS version 18.0 (IBM, Armonk, USA). The sample size computation was based on data (Cmet and performance) from Sato & Mokha [9] and Stanton et al. [12]. The software used was GPOWER version 3.1 (Power as 1-beta error probability: 95%; Effect size: 0.90; Error assumed as alpha: 0.05). After calculation, 26 subjects were indicated for allocation equally for each group, 13 subjects in CG group and 13 in PG group. We decided insert more subjects in each group, due to a possible sample loss. Therefore, the present study was initiated with 32 individuals. Results Participant baseline characteristics The sample characterization data are shown in Table 2. No significant differences in this sec- tion were observed between the groups in the pre-training period. Table 2. Mean (standard deviation) age, height, body mass, body fat, lean mass, maximal oxygenuptake (VO2max), and speed at the second ventilatory threshold (VT2) in the pre-training period. Variable Group Control group (n = 16) Pilates group (n = 15) p-value Age (years) 18.44 (0.52) 18.42 (0.51) 0.996 Height (cm) 176.66 (9.89) 175.07 (8.06) 0.404 Body mass (kg) 73.64 (10.79) 70.71 (10.90) 0.391 Body fat (%) 10.81 (2.49) 9.34 (1.98) 0.205 Lean mass (%) 49.82 (2.26) 50.54 (2.40) 0.583 Speed at VT2 (km.h-1) 14.44 (1.33) 14.21 (1.05) 0.837 VO2max (mL.kg-1.min-1) 51.26 (5.43) 51.75 (7.55) 0.926 https://doi.org/10.1371/journal.pone.0194057.t002 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 6 / 19 Running performance and respiratory variables The variables running time, VO2max, Cmet10, and Cmet12 were not significantly different between the groups in the pre-training period. In the post-training period, the PG had signifi- cantly higher VO2max values (p<0.001), a significantly shorter 5-km running time (p<0.001), and a significantly lower Cmet12 (p = 0.019). For the time factor, significant differences were found in both groups for all of the variables evaluated (Table 3). Electromyographic variables Maximal voluntary isometric contraction. The comparisons between the training peri- ods indicated that the MVIC of the OE, OI, LO, BF, and GM muscles increased significantly in only the PG whereas the activation of the VL muscle increased significantly in both the CG and PG between pre- and post-training (Table 4). In addition, no significant differences in MVIC were found between the CG and PG in pre-training. In post-training, the MVIC of the OE, OI, LO, and GM muscles was significantly higher in the PG than in the CG. Muscle activity during running. The data on muscle activation during the stride phases, presented as a percentage of the MVIC, indicated a distinct behaviour in relation to the remaining variables analysed. In the pre-training period, significant differences in the level of activation of the OE and BF muscles were found in the swing phase at 10 km.h-1 (p = 0.018 and 0.048, respectively) and for the VL (p = 0.024) and BF (p = 0.26) muscles at 10 km.h-1 in the pre-activation phase. Obliquus externus abdominis. A significant increase in the level of activation of the OE muscle was found in the pre-activation phase between the training periods at 10 km.h-1 (p = 0.022) (Fig 1). However, no differences in the level of activation of this muscle were found between groups (p = 0.983). In addition, at 10 km.h-1, the percentage of muscle activation between the training periods decreased only in the swing phase in both groups (p = 0.002). The percentage of muscle activation significantly decreased in both groups in the support (p<0.001) and swing (p<0.001) phases at 12 km.h-1. Moreover, in the swing phase post-train- ing, the level of activation was lower in the PG compared to the CG (p = 0.009). Obliquus internus abdominis. At 10 km.h-1, the level of activation of the OI muscle increased significantly in the pre-activation stage between pre- and post-training in both Table 3. Effect of running training and running training combined with Pilates on performance and respiratory variables. Data Represent the Mean Values (Stan- dard Error) for 5-km Running Time, Maximum Oxygen Consumption (VO2max), Metabolic Cost at 10 km.h-1 (Cmet10), Metabolic Cost at 12 km.h-1 (Cmet12), Speed at the Second Ventilatory Threshold (VT2), and Oxygen Consumption at the Second Ventilatory Threshold (VO2 VT2). Variable Group Period Effect of time Effect of group Interaction group x time Pre-training Post-training p-value p-value p-value 5-km running time CG 25.33 (0.58) 24.61 (0.52) <0.001 0.441 <0.001 (min) PG 25.65 (0.44) 23.23 (0.40)a VO2max CG 51.32 (1.20) 53.72 (1.58) <0.001 0.204 <0.001 (mL.kg-1.min-1) PG 51.8 (1.73) 58.53 (1.59)a Cmet10 CG 4.27 (0.09) 3.85 (0.13) <0.001 0.868 0.923 (J.kg-1.m-1) PG 4.26(0.09) 3.82 (0.08) Cmet12 CG 5.22 (0.08) 4.71 (0.11) <0.001 0.014 0.019 (J.kg-1.m-1) PG 5.00 (0.10) 4.33 (0.07)a Significant difference between pre- and post-training a significant difference between the groups in post-training https://doi.org/10.1371/journal.pone.0194057.t003 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 7 / 19 groups (p = 0.009) (Fig 2). At the speed of 12 km.h-1, the level of activation increased only in the PG in the pre-activation phase between pre- and post-training (p = 0.01); however, the level of muscle activation in the PG was lower than that in the CG (p = 0.003). In the support and swing phases, the percentage of muscle activation decreased significantly between pre- and post-training at both speeds in both groups (p<0.001). At 12 km.h-1 in post- training, the level of activation in the PG was significantly lower than that in the CG (p = 0.01). Vastus lateralis. The results for the VA muscle are shown in Fig 3. In the pre-activation phase (group factor, p = 0.273; time factor, p = 0.260) and swing phase (group factor, p = 0.551; time factor, p = 0.565), there were no significant differences in muscle activation in the conditions analysed. However, in the support phase, there was a significant decrease in the level of activation between the two training periods in all of the cases analysed (p<0.001). For the VA muscle in particular, there were no differences between the groups in any stride phases. Longissimus. There were no significant differences in the level of activation of the LO muscle in the pre-activation phase in the conditions analysed. In the support phase at 10 km.h-1, the level of activation decreased in both groups (p = 0.001) (Fig 4). At 12 km.h-1, the level of activation decreased only in the PG between pre- and post-training (p = 0.003). Fur- thermore, in post-training, muscle activation in the PG was significantly lower than that in the CG (p = 0.002). In the swing phase at 10 km.h-1, there were no significant differences in the level of activation of the LO between groups (p = 0.630) or between training periods (p = 0.364). At 12 km.h-1 in post-training, the level of activation of this muscle in the PG was significantly lower than in pre- training (p<0.001) and was significantly lower than in the CG (p = 0.005). Biceps femoris. In the pre-activation phase at 10 km.h-1, there were no significant differ- ences in the level of activation of the BF muscle between pre- and post-training (p = 0.498) Table 4. Effects of running training (CG) and running training combined with Pilates (PG) on maximal voluntary isometric contraction (MVIC) in millivolts (mV) of the obliquus externus abdominis (OE), obliquus internus abdominis (OI), vastus lateralis (VL), longissimus (LO), biceps femoris (BF), gluteus medius (GM), and latissimus dorsi (LD) muscles. Variable Group Period Effect of time Effect of group Interaction group x time Pre-training Post-training p-value p-value p-value OE MVIC CG 233.00 (28.41) 229.36 (41.75) 0.005 0.048 0.047 (mV) PG 249.87 (22.12) 316.95 (26.70)a OI MVIC CG 527.14 (60.5) 510.52 (71.19) 0.03 0.037 0.006 (mV) PG 550.69 (60.85) 685.48 (73.46)a VL MVIC CG 425.53 (26.82) 483.16 (36.73) 0.032 0.193 0.138 (mV) PG 432.39 (40.79) 561.93 (46.61) LO MVIC CG 284.31 (17.51) 285.79 (15.25) 0.012 0.027 0.016 (mV) PG 299.39 (21.31) 371.22 (21.49)a BF MVIC CG 379.40 (29.5) 452.78 (37.11) <0.001 0.559 0.034 (mV) PG 370.47 (30.56) 510.20 (27.59) GM MVIC CG 471.15 (39.47) 529.03 (53.10) 0.006 0.007 0.040 (mV) PG 450.12 (33.60) 587.68 (45.61) a LD MVIC CG 348.14 (28.00) 376.24 (32.59) 0.502 0.660 0.745 (mV) PG 376.05 (32.86) 385.81 (47.95) Significant difference between pre- and post-training a significant difference between the groups in post-training https://doi.org/10.1371/journal.pone.0194057.t004 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 8 / 19 (Fig 5). However, in the pre-activation phase at 12 km.h-1, the level of activation of this muscle decreased significantly in both groups (p<0.001). In the support phase at 10 km.h-1 (p<0.001) and 12 km.h-1 (p<0.001), the level of activation of the BF decreased significantly between pre- and post-training regardless of the group evalu- ated. In the swing phase, the level of activation decreased significantly at the two speeds and in both groups. Gluteus medius. In the pre-activation phase at 10 km.h-1 and 12 km.h-1, no significant differences in the activation of the GM muscle were observed (group factor, p = 0.841, time factor, p = 0.083; group factor, p = 0.686, time factor, p = 0.081, respectively). In the support phase at 10 km.h-1 (p = 0.003) and 12 km.h-1 (p<0.001), the percentage of muscle activation decreased in both groups and at both speeds between the two training periods. At 12 km.h-1 in post-training, the percentage of muscle activation in the PG was significantly lower than that in the CG (p = 0.005). In the swing phase at 10 km.h-1, there were no significant differences in the percentage of muscle activation considering the time factor (p = 0.968) and group factor (p = 0.712). However, at 12 km.h-1, the level of activation of this muscle decreased in the CG Fig 1. Upper panels, mean pattern of activation of the obliquus externus abdominis muscle (mV). Lower panels, mean (± standard error) muscle activation in the three stride phases, presented as a percentage of the MVIC. Red lines represent 12 km.h-1, and blue lines represent 10 km.h-1. Dotted lines represent the control group (CG), and solid lines represent the Pilates group (PG). The vertical dotted lines indicate the end of the contact phase.  Significant difference between pre- and post- training; a significant difference between groups (p<0.005). https://doi.org/10.1371/journal.pone.0194057.g001 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 9 / 19 and PG considering the time factor (p<0.001), and muscle activation in the PG was signifi- cantly lower than in the CG (p = 0.005) in post-training (Fig 6). Latissimus dorsi. No significant differences were observed in the level of activation of the LD muscle in any of the running stride phases evaluated (Fig 7). Discussion The results support our hypotheses that distance running performance is enhanced after a 12-weeks Pilates training programme. The improvements in performance are accompanied by a critical reduction on Cmet and trunk muscle activation. This suggests that distance runners are able to transfer effective gains from a slow-type core strength training method to the run- ning movement. There is great interest in the mechanisms capable of minimizing energy expenditure during running because these mechanisms play an essential role in the search for strategies to improve performance. From the mechanical point of view, the "mass-spring” model reflects the Fig 2. Upper panels, mean activation pattern of the obliquus internus abdominis muscle (mV). Lower panels, mean (± standard error) muscle activation in the three stride phases, presented as a percentage of the MVIC. Red lines represent 12 km.h-1, and blue lines represent 10 km.h-1. Dotted lines represent the control group (CG), and solid lines represent the Pilates group (PG). The vertical dotted lines indicate the end of the contact phase.  Significant difference between pre- and post-training; a significant difference between groups (p<0.005). https://doi.org/10.1371/journal.pone.0194057.g002 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 10 / 19 occurrence of storage and release of elastic energy during running, which helps minimize the expenditure of metabolic energy [20]. Therefore, changes in this mechanism could affect per- formance and improve running economy (RE) [4]. Hoff et al.[21] concluded that shorter contact time with the ground is accompanied by a longer time at a lower level of muscle activation, indicating lower metabolic demand at the same submaximal speed. Therefore, lower metabolic demand in the muscles is dependent on a number of factors, including the activation level of the task. According to a recent model by Miller et al.[22] on energy minimization during running, the decrease in muscle activity is the primary strategy to generate greater energy economy during running. This model was built using the speed of 3.76 m.s-1, which is similar to the highest speed evaluated in this study (12 km.h-1) and supports the results found herein. In addition to the decrease in Cmet at both speeds in both study groups as a result of the training applied, we found an overall decrease in the percentage of muscle activation at the same speed in post-training. Moreover, the PG displayed a significantly greater decrease in Cmet12 and in the 5-km run performance compared to the CG. This decrease was accompanied by a higher VO2max and a further decrease in the level of activation of the OE (Δ6.77% in the Fig 3. Upper panels, mean activation pattern of the vastus lateralis muscle (mV). Lower panels, mean (± standard error) muscle activation in the three stride phases, presented as a percentage of the MVIC. Red lines represent 12 km.h-1, and blue lines represent 10 km.h-1. Dotted lines represent the control group (CG), and solid lines represent the Pilates group (PG). The vertical dotted lines indicate the end of the contact phase.  Significant difference between pre- and post-training; a significant difference between groups (p<0.005). https://doi.org/10.1371/journal.pone.0194057.g003 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 11 / 19 swing phase,), OI (Δ5.84% in the support phase and Δ9.73% in the swing phase), LO (Δ8.00% in the support phase and Δ16.03% in the swing phase), and GM (Δ9.81% in the support phase and Δ5.05% in the swing phase) muscles compared to the CG at 12 km.h-1, in accordance with the model of Miller et al.[22]. Therefore, if better performance in the 5-km run can be determined by a higher VO2max, the ability to sustain a higher fraction of VO2max, and a better RE, and because the decrease in the percentage of muscle activation optimizes energy minimization during running [21,22], our findings suggest the presence of a correlation between a 12-week training programme of classic PT and the mechanisms capable of minimizing energy during running, thus contribut- ing to improved performance (Fig 8). When analysed in isolation, the decreased percentage of muscle activation found during running in both groups at 10 and 12 km.h-1 can be explained by the so-called "neuromuscular economy" [23], which is defined as the decrease in the number of motor units recruited when considering a situation involving a similar submaximal task. This mechanism explains in part Fig 4. Upper panels, mean activation pattern of the longissimus muscle (mV). Lower panels, mean (± standard error) muscle activation in the three stride phases, presented as a percentage of the MVIC. Red lines represent 12 km.h-1, and blue lines represent 10 km.h-1. Dotted lines represent the control group (CG), and solid lines represent the Pilates group (PG). The vertical dotted lines indicate the end of the contact phase.  Significant difference between pre- and post-training; a significant difference between groups (p<0.005). https://doi.org/10.1371/journal.pone.0194057.g004 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 12 / 19 the results of the present study, in which the same running speeds were analysed in pre- and post-training. Besides, the running training performed by the CG decreased the percentage of muscle acti- vation at both speeds between pre- and post-training, and the decrease in the PG was signifi- cantly stronger. These results indicate the possible presence of neuromuscular economy, particularly when analysed together with the results for the MVIC. The maximum amplitude of the EMG signal in the VAS muscle was significantly higher post-training in both groups. However, the level of activation of the OA, LO, OI, and GM muscles increased significantly only in the PG, who underwent special training for these muscles. The increase in the maximum amplitude of the MVIC along with the higher VO2max found in post-training may decrease the relative loads, which correspond to the speeds in pre-train- ing and justify the decreased recruitment of motor units during the performance of the same task in the post-training period. This hypothesis would explain the findings in the PG, who showed a stronger decrease in the percentage of muscle activation during running and shorter 5-km run completion time compared to the CG. Fig 5. Upper panels, mean activation pattern of the biceps femoris muscle (mV). Lower panels, mean (± standard error) muscle activation in the three stride phases, presented as a percentage of the MVIC. Red lines represent 12 km.h-1, and blue lines represent 10 km.h-1. Dotted lines represent the control group (CG), and solid lines represent the Pilates group (PG). The vertical dotted lines indicate the end of the contact phase.  Significant difference between pre- and post-training; a significant difference between groups (p<0.005). https://doi.org/10.1371/journal.pone.0194057.g005 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 13 / 19 The reduction in the time of completion of this run, the improvement in Cmet12, and the greater decrease in the percentage of EMG activation in the PG during running appear to be associated with the control and stabilization of the lumbopelvic region. The correlation between EMG activity and stability is well established in the literature. In fact, findings empiri- cally show that the neural system responds to changes in spinal stability [24] and gives support to the adaptive process model on motor learning [25]. The control of the trunk is an important factor for Cmet, and leg movements are closely associated with lumbopelvic movements; thus, the latter depend on the stiffness of the abdominal muscles [3,25]. Therefore, the increase in running speed would cause more lumbopelvic movements and consequently greater instability, which would require greater neuromuscular control to achieve stability during cyclical movements such as running [8,26]. In turn, this increased neu- romuscular demand for stabilization of the lumbopelvic region appears to be associated with a greater contribution of concentric activations—which are more energy-consuming than eccentric and isometric activations—and reinforces the fact that an unstable system is also less economical [25,27]. Fig 6. Upper panels, mean activation pattern of the gluteus medius muscle (mV). Lower panels, mean (± standard error) muscle activation in the three phases, presented as a percentage of the MVIC. Red lines represent 12 km.h-1, whereas blue lines represent 10 km.h-1. Dotted lines represent the control group (CG), and solid lines represent the Pilates group (PG). The vertical dotted lines indicate the end of the contact phase.  Significant difference between pre- and post-training; a significant difference between groups (p<0.005). https://doi.org/10.1371/journal.pone.0194057.g006 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 14 / 19 From this perspective, lumbopelvic stabilization is one of the aims of Pilates. Among its guiding principles, breathing [28] and centring [29] have been shown to stimulate the deep abdominal muscles responsible for stabilization, including the rectus abdominis muscles, OI, OE, and transversus abdominis [8]. In this respect, Phrompaet et al.[30] evaluated the effects of PT in the control of lumbopelvic movements. At the end of eight weeks of training, the authors found that 65% of the subjects in the Pilates group passed the lumbopelvic stability test after four weeks of training, and 85% passed the test after eight weeks of training, whereas none of the subjects in the control group passed the test. The authors indicate that the improved recruitment of abdominal muscles during PT appears to help develop the strength of these muscles, leading to improved stability. However, EMG activity was not evaluated in that study. Sato and Mokha [9] evaluated a six-week core-training programme and found improve- ment in a 5-km run completion time. The run completion time decreased significantly in the experimental group (from 29.29±2.38 to 28.42±2.23 min) but did not decrease in the control group. In addition, Stanton et al. [12] evaluated participants after six weeks of core training Fig 7. Upper panel, mean activation pattern of the latissimus dorsi muscle (mV). Lower panel, mean (± standard error) muscle activation in the three stride phases, presented as a percentage of the MVIC. Red lines represent 12 km.h-1, and blue lines represent 10 km.h-1. Dotted lines represent the control group (CG), and solid lines represent the Pilates group (PG). The vertical dotted lines indicate the end of the contact phase.  Significant difference between pre- and post-training; a significant difference between groups (p<0.005). https://doi.org/10.1371/journal.pone.0194057.g007 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 15 / 19 with a Swiss ball and found a significant improvement in lumbopelvic stability; however, no differences were found in running economy, VO2max, or in the EMG activity of the trunk muscles. By contrast, in the present study, the run completion time decreased from 25.33±0.58 to 24.61±0.52 min in the CG and from 25.65±0.44 to 23.23±0.40 min in the PG. In addition, the PG had an improvement in VO2max (from 51.8±1.73 mL.kg-1.min-1 in pre-training to 58.53 ±1.59 mL.kg-1.min-1 in post-training, p<0.001) and Cmet12 (from 5.0±0.10 J.kg-1.m-1 in pre- training to 4.33±0.07 J.kg-1.m-1 in post-training, p<0.001) and a decrease in the percentage of EMG activation of the trunk muscles. However, despite the conflicting results with the literature with regard to core training, the present study is distinguished by the duration of the training period. In the cited studies, only Fig 8. Schematic drawing of the performance model proposed in this study [2,3,20,22]. https://doi.org/10.1371/journal.pone.0194057.g008 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 16 / 19 a six-week training programme was conducted whereas the present study utilized a 12-week training programme in both groups, and this longer training may have contributed to the results obtained. Moreover, unlike core training, PT should be performed considering its prin- ciples, which can help increase muscle activation to higher levels [28,29]. In conclusion, PT significantly improved the 5-km run performance. This improved perfor- mance is associated with the optimization of mechanisms capable of minimizing energy expenditure. That is, a lower percentage of EMG activation of the trunk muscles during run- ning as a result of strength gain. Therefore, the greater running economy seems to be positively influenced the 5-km run performance in recreational runners. Conclusions The results of this study indicate that PT can be incorporated into the training programmes of recreational runners to improve running performance and VO2max and to strengthen trunk muscles. In addition, in situations in which the development of aerobic power is limited by cardiac or pulmonary capacity, PT may improve performance at a lower metabolic cost by decreasing muscle demand during unnecessary pelvic movements and improve other health- related aspects, including a lower risk of injury. However, little is known about the effects of PT on mechanisms that minimize energy expenditure, mechanical parameters, and their cor- relation with running performance. Therefore, further studies are necessary to elucidate these relationships. Supporting information S1 Table. 12-week periodization of running training using the following intensity scores: E, Easy; M, Moderate; T, Threshold; and I, Interval, as a function of the heart rate at VT2. (DOCX) S2 Table. General dataset. (XLSX) Acknowledgments Coordenac¸ão de Aperfeic¸oamento de Pessoal de Nı´vel Superior (CAPES) and Conselho Nacio- nal de Desenvolvimento Cientı´fico e Tecnolo´gico (CNPq). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. We are grateful to the Locomotion Group of the Federal University of Rio Grande do Sul for dis- cussions and comments. L.A. Peyre´-Tartaruga is an established investigator of the Brazilian Research Council (CNPq), Brası´lia, Brazil. Author Contributions Conceptualization: Paula Finatto, Leonardo A. Peyre´-Tartaruga. Data curation: Paula Finatto, Edson Soares Da Silva, Alexandre B. Okamura, Bruna P. Almada, Henrique B. Oliveira, Leonardo A. Peyre´-Tartaruga. Formal analysis: Paula Finatto, Edson Soares Da Silva, Alexandre B. Okamura, Bruna P. Almada, Henrique B. Oliveira, Leonardo A. Peyre´-Tartaruga. Funding acquisition: Leonardo A. Peyre´-Tartaruga. Investigation: Paula Finatto, Edson Soares Da Silva, Alexandre B. Okamura, Bruna P. Almada, Henrique B. Oliveira, Leonardo A. Peyre´-Tartaruga. Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 17 / 19 Methodology: Paula Finatto, Edson Soares Da Silva, Alexandre B. Okamura, Bruna P. Almada, Henrique B. Oliveira, Leonardo A. Peyre´-Tartaruga. Project administration: Leonardo A. Peyre´-Tartaruga. Resources: Paula Finatto, Leonardo A. Peyre´-Tartaruga. Software: Paula Finatto, Edson Soares Da Silva, Henrique B. Oliveira, Leonardo A. Peyre´- Tartaruga. Supervision: Leonardo A. Peyre´-Tartaruga. Visualization: Paula Finatto, Alexandre B. Okamura, Leonardo A. Peyre´-Tartaruga. Writing – original draft: Paula Finatto. Writing – review & editing: Paula Finatto, Edson Soares Da Silva, Alexandre B. Okamura, Bruna P. Almada, Henrique B. Oliveira, Leonardo A. Peyre´-Tartaruga. References 1. Joyner MJ. Modeling: optimal marathon performance on the basis of physiological factors. J Appl Phy- siol. 1991; 70: 683–687. https://doi.org/10.1152/jappl.1991.70.2.683 PMID: 2022559 2. di Prampero PE, Atchou G, Bruckner JC, Moia C. The energetics of endurance running. Eur J Appl Phy- siol Occup Physiol. 1986; 55: 259–266. PMID: 3732253 3. Saunders PU, Pyne DB, Telford RD, Hawley JA. Factors affecting running economy in trained distance runners. Sports Med. 2004; 34: 465–485. PMID: 15233599 4. Foster C, Lucia A. Running economy: the forgotten factor in elite performance. Sports Med. 2007; 37: 316–319. PMID: 17465597 5. Millet GP, Jaouen B, Borrani F, Candau R. Effects of concurrent endurance and strength training on running economy and.VO(2) kinetics. Med Sci Sports Exerc. 2002; 34: 1351–1359. PMID: 12165692 6. Storen O, Helgerud J, Stoa EM, Hoff J. Maximal strength training improves running economy in distance runners. Med Sci Sports Exerc. 2008; 40: 1087–1092. https://doi.org/10.1249/MSS.0b013e318168da2f PMID: 18460997 7. Behm DG, Cappa D, Power GA. Trunk muscle activation during moderate- and high-intensity running. Appl Physiol Nutr Metab. 2009; 34: 1008–1016. https://doi.org/10.1139/H09-102 PMID: 20029508 8. Saunders SW, Schache A, Rath D, Hodges PW. Changes in three dimensional lumbo-pelvic kinematics and trunk muscle activity with speed and mode of locomotion. Clin Biomech. 2005; 20: 784–793. 9. Sato K, Mokha M. Does core strength training influence running kinetics, lower-extremity stability, and 5000-M performance in runners? J Strength Cond Res. 2009; 23: 133–140. https://doi.org/10.1519/ JSC.0b013e31818eb0c5 PMID: 19077735 10. Muscolino JE, Cipriani S. Pilates and the “powerhouse”—I. J Bodyw Mov Ther. 2004; 8: 15–24. 11. Culligan PJ, Scherer J, Dyer K, Priestley JL, Guingon-White G, Delvecchio D, et al. A randomized clini- cal trial comparing pelvic floor muscle training to a Pilates exercise program for improving pelvic muscle strength. Int Urogynecol J. 2010; 21: 401–408. https://doi.org/10.1007/s00192-009-1046-z PMID: 20094704 12. Stanton R, Reaburn PR, Humphries B. The effect of short-term Swiss ball training on core stability and running economy. J Strength Cond Res. 2004; 18: 522–528. https://doi.org/10.1519/1533-4287(2004) 18<522:TEOSSB>2.0.CO;2 PMID: 15320664 13. Daniels J. Daniels’ running formula. Champaign, IL: Human Kinetics; 2005. 14. De Luca CJ. The use of surface electromyography in biomechanics. J Appl Biomech. 1997; 13: 135– 163. 15. Beck TW, Housh TJ, Johnson GO, Weir JP, Cramer JT, Coburn JW, et al. The effects of interelectrode distance on electromyographic amplitude and mean power frequency during isokinetic and isometric muscle actions of the biceps brachii. J Electromyogr Kinesiol. 2005; 15: 482–495. https://doi.org/10. 1016/j.jelekin.2004.12.001 PMID: 15935960 16. Hermens HJ, Freriks B, Merletti R, Stegeman D, Blok J, Rau G, et al. European recommendations for surface electromyography. Roessingh Research and Development. 1999; 8: 13–54. Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 18 / 19 17. Marques NR, Morcelli MH, Hallal CZ, Gonc¸alves M. EMG activity of trunk stabilizer muscles during cen- tering principle of Pilates method. J Bodyw Mov Ther. 2013; 17: 185–191. https://doi.org/10.1016/j.jbmt. 2012.06.002 PMID: 23561865 18. Kyrolainen H, Avela J, Komi PV. Changes in muscle activity with increasing running speed. J Sports Sci. 2005; 23: 1101–1109. https://doi.org/10.1080/02640410400021575 PMID: 16194986 19. Cappellini G, Ivanenko YP, Poppele RE, Lacquaniti F. Motor patterns in human walking and running. J Neurophysiol. 2006; 95: 3426–3437. https://doi.org/10.1152/jn.00081.2006 PMID: 16554517 20. Saibene F, Minetti AE. Biomechanical and physiological aspects of legged locomotion in humans. Eur J Appl Physiol. 2003; 88: 297–316. https://doi.org/10.1007/s00421-002-0654-9 PMID: 12527959 21. Hoff J, Gran A, Helgerud J. Maximal strength training improves aerobic endurance performance. Med Sci Sports. 2002; 12: 288–295. 22. Miller RH, Umberger BR, Hamill J, Caldwell GE. Evaluation of the minimum energy hypothesis and other potential optimality criteria for human running. Proc Biol Sci. 2012; 279: 1498–1505. https://doi. org/10.1098/rspb.2011.2015 PMID: 22072601 23. Cadore EL, Pinto RS, Alberton CL, Pinto SS, Lhullier FL, Tartaruga MP, et al. Neuromuscular economy, strength, and endurance in healthy elderly men. J Strength Cond Res. 2011; 25: 997–1003. https://doi. org/10.1519/JSC.0b013e3181d650ba PMID: 20881506 24. Santos SP, Benda RN, Couto CR, Campos CE, Andrade AGP, Lage GM, et al. The level of perfor- mance stabilization influences motor adaptation on isometric force control task. PLoS ONE. 2017; 12: e0185939 https://doi.org/10.1371/journal.pone.0185939 PMID: 29073273 25. Gardner-Morse MG, Stokes IA. The effects of abdominal muscle coactivation on lumbar spine stability. Spine. 1998; 23: 86–91; discussion 91–82. PMID: 9460158 26. England SA, Granata KP. The influence of gait speed on local dynamic stability of. Gait Posture. 2007; 25: 172–178. https://doi.org/10.1016/j.gaitpost.2006.03.003 PMID: 16621565 27. Granata KP, Orishimo KF. Response of trunk muscle coactivation to changes in spinal stability. J Bio- mech. 2001; 34: 1117–1123. PMID: 11506782 28. Barbosa AWC, Guedes CA, Bonifa´cio DN, de Fa´tima Silva A, Martins FLM, Barbosa MCSA. The Pilates breathing technique increases the electromyographic amplitude level of the deep abdominal muscles in untrained people. J Bodyw Mov Ther. 2015; 19: 57–61. https://doi.org/10.1016/j.jbmt.2014.05.011 PMID: 25603743 29. Barbosa AW, Martins FL, Vitorino DF, Barbosa MC. Immediate electromyographic changes of the biceps brachii and upper rectus abdominis muscles due to the Pilates centring technique. J Bodyw Mov Ther. 2013; 17: 385–390. https://doi.org/10.1016/j.jbmt.2013.01.003 PMID: 23768286 30. Phrompaet S, Paungmali A, Pirunsan U, Sitilertpisan P. Effects of Pilates training on lumbo-pelvic sta- bility and flexibility. Asian J Sports Med. 2011; 2: 16–22. PMID: 22375213 Pilates training improves 5-km run performance PLOS ONE | https://doi.org/10.1371/journal.pone.0194057 March 21, 2018 19 / 19

Pilates training improves 5-km run performance by changing metabolic cost and muscle activity in trained runners.

03-21-2018

Finatto, Paula,Silva, Edson Soares Da,Okamura, Alexandre B,Almada, Bruna P,Storniolo, Jorge L L,Oliveira, Henrique B,Peyré-Tartaruga, Leonardo A

eng

PMC6651650

J Exerc Nutrition Biochem. 2018;22(2):007-011, http://dx.doi.org/10.20463/jenb.2018.0010 45 J Exerc Nutrition Biochem. 2019;23(2):045-050, http://dx.doi.org/10.20463/jenb.2019.0016 45 Effect of interval exercise versus continuous exercise on excess post- exercise oxygen consumption during energy-homogenized exercise on a cycle ergometer Won-Sang Jung1 / Hyejung Hwang1 / Jisu Kim1 / Hun-Young Park1 / Kiwon Lim1,2* 1. Physical Activity and Performance Institute (PAPI), Konkuk University, Seoul, Republic of Korea 2. Department of Physical Education, Konkuk University, Seoul, Republic of Korea [Purpose] The purpose of this study was to confirm that the difference in excess post-exercise oxygen consumption (EPOC) during exercise of the spending the same calories between the continuous and interval exercise. [Methods] Thirty-four healthy college students who did not regularly exercise volunteered to participate in our study. Continuous exercise was performed on an ergometer for 30 min at 60% of maximal oxygen con- sumption (VO2 max). Interval exercise was performed on a cycle ergometer at 80% VO2 max for 2 min initially, followed by 40% VO2 max for 1 min, and 80% VO2 max for 3 min. This was repeated six times for a total of 26 min. [Results] The major findings were as follows: (1) en- ergy consumption during exercise was not significantly different between continuous exercise and interval exercise groups; (2) EPOC was higher in interval exercise than in continuous exercise for all dependent variables (i.e., total oxygen consumption, total calories, summation of heart rate); and (3) there were no signifi- cant differences in the lipid profile between continuous and interval groups. [Conclusions] Our study confirmed that after equal- izing energy expenditure for continuous and interval exercise on a cycle ergometer in subjects in their twenties, interval exercise results in higher EPOC than continuous exercise. These data suggest that interval exercise may be more effective than continuous exer- cise in reducing body fat, for a given amount of energy expenditure. [Key words] continuous exercise, interval exercise, excess post-exercise oxygen consumption (EPOC), energy expenditure. Received: 2019/06/03, Revised: 2019/06/28, Accepted: 2019/06/28, Published: 2019/06/30 ©2019 Won-Sang Jung et al.; License Journal of Exercise Nutrition and Biochemistry. This is an open access article distributed under the terms of the creative commons attri- bution license (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduc- tion in any medium, provided the orginal work is properly cited. *Corresponding author : Kiwon Lim, Ph.D. Laboratory of Exercise Nutrition, Department of Physical Education, Konkuk University, 120 Neungdong-ro, Gwang- jin-gu, Seoul 05029, Republic of Korea Tel: +82-2-450-3827 / Fax: +82-452-6027 E-mail: exercise@konkuk.ac.kr ©2019 The Korean Society for Exercise Nutrition OPEN ACCESS http://dx.doi.org/10.20463/jenb.2019.0016 J Exerc Nutrition Biochem. 2019;23(2):045-050 INTRODUCTION Many studies have demonstrated the importance and effectiveness of exercise in managing health and losing weight. The American Col- lege of Sports Medicine (ACSM) recommends an exercise intensity of 40–85% heart rate reserve (HRR) or oxygen uptake reserve (VO2R), a target energy expenditure of 150–400 kcal (or 20–60 min), and over three per weeks, 30 min of continuous exercise1. The reported positive effects of continuous exercise include relative- ly high increases in blood levels of epinephrine, norepinephrine, and growth hormone, the increased use of fat as an energy source, and the secretion of insulin and cortisol2,3,4. Continuous exercise has also been reported to be effective for weight loss by increasing daily energy con- sumption5,6. However, despite these positive effects, not many people are able to maintain these habits due to time constraints, exercise intol- erance, and monotony7. Interval training has been recommended as a new exercise method that can eliminate these shortcomings8,9. Interval training is a form of exercise in which short periods of intense exercise are alternated with less-intense recovery periods. It is good for improv- ing both aerobic and anaerobic energy systems, and is very effective at increasing an individual’s VO2 max and anaerobic threshold10. Thus, it is one of the most effective ways to improve cardiopulmonary functions, metabolic functions, health, and weight loss in the general population and in athletes8. During vigorous and high intensity interval exercise, metabolic rates can increase exponentially, and the intensity and duration of exercise can greatly affect metabolic reactions, both during and after exercise6. In particular, during recovery, excess post-exercise oxygen consump- tion (EPOC) is used to restore the body to a resting state, and to adapt it to the exercise just performed. Several mechanisms are attributed to EPOC, such as replenishment of oxygen stores in muscle and blood, increased circulation and lactate removal, resynthesis of adenosine tri- phosphate (ATP) and creatine phosphate (CrP), increased triglyceride/ fatty acid cycling, and increased heart rate (HR), ventilation, and body J Exerc Nutrition Biochem. 2019;23(2):045-050, http://dx.doi.org/10.20463/jenb.2019.0016 46 Effect of interval exercise versus continuous exercise on EPOC Journal of Exercise Nutrition & Biochemistry temperature11,12,13. Several studies have stated the importance of EPOC to continuous exercise and interval exercise, and EPOC size have suggested more continual of exercise at higher inten- sity6,8,14. However, in previous studies, most of the calories homogenized between continuous exercise and interval exercise were mainly compared to the same exercise time and different exercise intensity, or the amount of exercise estimated using a calculation formula. The was difference in actual calorie when the calories were homogenized with absolute intensity and time during exercise and there was no difference in EPOC16. Therefore, identifying the effect of EPOC on the homogenization of energy consumption between continuous exercise and interval exercise will be important in determining the optimal exercise routine to promote health and weight loss in the future. For accu- rate homogenization of kinetic energy consumption, it is important to eliminate extrinsic variables that affect the accuracy of VO2 max measurements and EPOC measrue- ments14. EPOC has been shown to be influenced by train- ing status, exercise intensity and duration, and the thermic effect of food6,15. Thus, to make a direct comparison be- tween recovery oxygen consumption after continuous and interval exercise, it is important to ensure that exercise variables such as total work and duration are as similar as possible and to avoid confounding factors such as food intake before and after exercise14,16. Therefore, in this study we measured EPOC in con- tinuous and interval exercise during or after exercise. Participants were provided with the same pre-food intake, and we homogenized the energy expenditure of during ex- ercise between exercise types, in order to minimize extra variables, and accurately measure EPOC. In addition, the purpose of to provide was prescription of exercise by ver- ifying the effect of EPOC of the continuous and interval exercise in subjects with twenties without exercise experi- ence. The purpose of the present study was to confirm that the difference of EPOC during exercise of the spending the same amount of calories between the continuous and interval exercise. METHODS Participants Thirty-four healthy college students in their twenties (mean age = 23.65 ± 2.17 years; n =18 men 16 women)’ who did not exercise regularly volunteered to participate in the study. Subjects who met one or more of the follow- ing exclusion criteria were deemed not eligible and were excluded from the study: unstable angina, having had a cardiac infarction within the previous four weeks, uncom- pensated heart failure, severe valvular illness, pulmonary disease, uncontrolled hypertension, kidney failure, ortho- pedic/neurological limitations, cardiomyopathy, planned surgery during the research period, reluctance to sign the consent form, drug or alcohol abuse, or involvement in another study. All subjects were fully acquainted with the nature of the study and were informed of the experi- mental risks before signing a written consent form. It was explicitly stated to the subjects that they could withdraw from the study at any point. All subjects had their pre-test research fully explained to them and provided voluntary consent. All procedures of the study was approved by the Institutional Review Board of Konkuk University (7001355-201903-HR-305) in Korea and was conducted according to the Declaration of Helsinki. Experimental design To test EPOC and energy expenditure during and after continuous and interval exercise, we used a balanced re- peated measures crossover design. This approach required gathering data on the subjects’ completion of two training sessions on separate test days, in a randomized order. Each participant visited the laboratory three times. On the first visit we performed body composition tests (In- Body 770, Biospace Ltd, Seoul, Korea), and a maximal cardiopulmonary exercise test (Quark CPET, Cosmed, Italy) to determine the maximal values of VO2 (VO2 max). On the second and third visits, at 72 h after performing the maximal CPETs, respectively, individuals performed con- tinuous cycle ergometer exercise at 60% of VO2 max, and interval cycle ergometer exercise at 40% or 80% of VO2 max. As soon as the exercise ended, subjects came down from the cycle ergometer, sat on a chair, and measured EPOC for 60 min. Pre-testing measurements All subjects performed a maximal aerobic exercise test using a cycle ergometer (Aerobike, Combi 75 XL, Tokyo, Japan) in order to determine their VO2 max. The work rate at 50 rpm was 50 W for men and 25 W for women for the first 2 min, and was increased by 25 W for men and 12.5 W for women every 2 min. This continued either until exhaustion or until subjects were unable to maintain 50 rpm. The criteria for having reached the true VO2 max was showing a plateau in VO2 uptake, despite increased inten- sity of exercise and a respiratory exchange ratio (RER) above 1.15. HR was monitored using a Polar 800 device (Polar Electro, Kempele, Finland). Exercise training protocol Participants were transported to the laboratory at 8 am after a 12-h fast and 48-h abstention from vigorous physi- cal activity. They were given a standardized breakfast of 2 pieces of bread (200 kcal), 1 boiled egg (80 kcal), 1 cup of orange juice (120 kcal), and 1 cup of water. Subjects rest- ed in a comfortable posture after breakfast and participat- ed in the experiment 2 h later. Ambient room temperature was maintained at 23 ± 1 °C. After 10 min of quiet sitting as a habituation period, we measured VO2, ventilation, and RER for 5 min. The average was used as the base- line (BASE). The subjects then performed continuous or interval exercise on a cycle ergometer (Aerobike, Combi 75 XL, Tokyo, Japan). Speed was adjusted on an individ- ual basis, according to each subject’s fitness level. The J Exerc Nutrition Biochem. 2019;23(2):045-050, http://dx.doi.org/10.20463/jenb.2019.0016 47 Effect of interval exercise versus continuous exercise on EPOC Journal of Exercise Nutrition & Biochemistry continuous exercise training was performed for 30 min at 60% of VO2 max, and the interval exercise training was performed first for 2 min at 80% of VO2 max, followed by 1 min at 40% of VO2 max, and finally for 3 min at 80% of VO2 max. This was repeated six times for a total of 26 min. The calories expended between the continuous exercise (Con Ex) and interval exercise (Inter Ex) groups were not statistically different (212.24 ± 68.47 vs 214.85 ± 66.32, p=0.503). EPOC measurement Immediately after exercise, participants were seated in a chair and relative VO2, absolute VO2, Kcal, HR, and duration were monitored for 60 min. EPOC values were determined at the time when VO2, HR, and RER values returned to the resting baseline. Collection and analysis of lipid samples was done before exercise, immediately after exercise, after 30 min and after 60 min. Total cholesterol (TC), triglyceride (TG), high-density lipoprotein (HDL) cholesterol, and low-density lipoprotein (LDL) cholesterol were measured using a portable digital lipid analyzer (SD LipidoCare, SD Biosensor, Inc., Seoul, Korea). Statistics All statistical analyses were completed using IBM SPSS Statistics 23 (SPSS Inc., Chicago, IL, USA). Data normality was verified using the Shapiro-Wilk test, and descriptive data are presented as mean ± standard devia- tion. A paired t-test was used to compare the differences between the two protocols. The effects of condition on EPOC were analyzed using a mixed procedure. Where main effects were statistically significant, post-hoc pair- wise comparisons with Sidak-adjusted p-values were per- formed. Model-fitting was evaluated using Hurvich and Tsai’s criteria. All statistical assumptions were checked us- ing standard graphical procedures. Statistical significance was accepted for p<0.05. RESULTS Figure 1 shows that the amount of calories expended during exercise was not significantly different between continuous and interval exercise(p=0.503). Based on the EPOC results shown in Table 2, the EPOC duration was longer for interval exercise than in continuous exercise (31.24 ± 15.09 vs 45.90 ± 12.37, p < Variable Men (n=18) Women (n=16) Total (n=34) Age (years) 24.28±2.49 22.94±1.53 23.65±2.17 Height (cm) 177.43±7.78 159.48±4.30 168.98±11.06 Weight (kg) 75.38±9.98 53.88±6.10 65.26±13.67 BMI (kg/m2) 23.86±2.04 21.19±2.28 22.61±2.52 Lean body mass (kg) 61.11±8.14 37.17±2.84 49.84±13.60 Fat mass (kg) 14.27±5.30 16.71±4.42 15.41±4.99 % fat mss (%) 18.74±5.71 30.01±6.06 24.05±8.13 VO2max (mL/min/kg) 36.84±6.16 41.08±4.49 32.08±3.86 Note: SD = standard deviation, BMI = body mass index. Table 1. Participant characteristics. Data represent the mean ± SD Variables EPOC O2 Deficit VO2_total (mL/min) VO2/kg_total (mL/min/kg) Kcal_total (kcal/min) HR_sum VO2_total (mL/min) Kcal_total (kcal/min) HR_sum Con Ex 11992.4 ±6481.05 185.42 ±98.94 58.14 ±31.42 2931.64 ±1560.92 594.11 ±242.10 3.39 ±1.35 28.98 ±8.17 Inter Ex 17425.24 ±6329.98 266.81 ±79.62 82.72 ±28.69 4557.1 ±1419.05 721.9 ±347.90 3.88 ±1.9 29.63 ±10.81 Δ% 45.3 43.89 42.28 55.45 21.51 14.45 2.24 Sig (p) .000*** .000*** .000*** .000*** .009** 0.066 0.747 Men Con Ex 14980.78 ±6529.74 204.83 ±103.40 72.8 ±31.82 3026.21 ±1346.65 729.8 ±194.71 4.19 ±1.06 29.00 ±8.74 Inter Ex 21410.32 ±5411.21 289.68 ±84.35 100.96 ±23.79 4630.17 ±1330.58 916.99 ±347.18 4.94 ±1.96 30.73 ±13.36 Δ% 42.92 41.42 38.68 53 25.65 17.9 3.82 Sig (p) .001** .001** .004** .000*** .006** 0.066 0.758 Women Con Ex 8630.48 ±4616.70 163.59 ±91.98 41.65 ±21.82 2825.24 ±1811.68 441.71 ±197.76 2.48 ±1.04 28.28 ±7.74 Inter Ex 12942.03 ±3803.88 241.08 ±67.46 62.21 ±17.92 4474.89 ±1552.43 502.42 ±180.62 2.69 ±0.89 28.4 ±7.19 Δ% 49.96 47.37 49.36 58.39 13.74 8.47 0.42 Sig (p) .001** . 001** .001** .001** 0.394 0.557 0.938 Note: SD = standard deviation, Con Ex = continuous exercises, Inter Ex = Interval exercise, EPOC = excess post-exercise oxygen consump- tion, O2 = Oxygen , VO2 = oxygen consumption, HR = heart rate , Sum = summation, * p<.05, ** p<.01, *** p<.001. Table 2. Comparison of EPOC in Con EX vs Inter EX, ± SD J Exerc Nutrition Biochem. 2019;23(2):045-050, http://dx.doi.org/10.20463/jenb.2019.0016 48 Effect of interval exercise versus continuous exercise on EPOC Journal of Exercise Nutrition & Biochemistry .001), which showed higher Inter Ex than in the Con Ex in all variables including VO2 total (11992.40 ± 6481.05 vs 17425.24 ± 6329.98, p<0.001; men: 14980.78 ± 6529.74 vs 21410.32 ± 5411.21, p=0.001; women: 8630.48 ± 4616.70 vs 12942.03 ± 3803.88, p<0.001), VO2/kg total (185.42 ± 98.94 vs 266.81 ± 79.62, p<0.001; men: 204.83 ± 103.40 vs 289.68 ± 84.35, p=0.001; women: 163.59 ± 91.98 vs 241.08 ± 67.46, p=0.001), Kcal total (58.14 ± 31.42 vs 82.72±28.69, p<0.001; men: 72.80 ± 31.82 vs 100.96 ± 23.79, p=0.004; women: 41.65 ± 21.82 vs 62.21 ± 17.92, p = 0.001) and HR sum (2931.64 ± 1560.92 vs 4557.10 ± 1419.05, p<0.001; men: 3026.21 ± 1346.65 vs 4630.17 ± 1330.58, p<0.001; women: 2825.24 ± 1811.68 vs 4474.89 ± 1552.43, p=0.001). When the results of oxy- gen-deficient were examined, VO2 total (594.11 ± 242.10 vs 721.90 ± 347.90 p=0.009; men: 729.80 ± 194.71 vs 916.99 ± 347.18, p=0.006; women: 441.71 ± 197.76 vs 502.42 ± 180.62, p=0.394) showed a greater value than con Ex in inter Ex, and after separating the results for men and women, significant differences were only found in men. There was no significant difference in HR sum levels. Figure 2 is a comparison of lipid profiles on EPOC in continuous and interval exercise. There were no signifi- cant differences in total cholesterol, triglyceride, HDL- cholesterol, or LDL- cholesterol in all variables (p>0.05). DISCUSSION The purpose of this study was to confirm that there is a difference in excess post-exercise oxygen consumption (EPOC) between continuous and interval exercise, when expending the same number of calories. The major find- ings were: (1) energy consumption during the exercise was not significantly different between continuous exer- cise and interval exercise, (2) EPOC was higher in inter- val exercise than in continuous exercise for all dependent variables (e.g. total oxygen consumption, total calorie, and summation of heart rate), and (3) there was no signif- icant differences in lipid profiles. In previous studies that did not homogenize energy consumption during exercise, EPOC was higher in the interval exercise compared to continuous exercise and interval exercise17-20. In addition, Williams et al.21 com- pared the EPOC of 20 min of high intensity interval ex- ercise and 60 min of continuous exercise, and found that the EPOC 30 min after exercise was higher in interval exercise, but the total EPOC after exercise was higher in continuous exercise. Larsen et al.11 reported that with increasing intensity, EPOC and EPOC duration increase, but if interval times are shorter, EPOC is reduced to sim- ilar levels as seen in continuous exercise. Tucker et al.18 showed that in high-intensity interval exercise oxygen consumption was low, but EPOC was high. However, summation of oxygen consumption during exercise and EPOC was higher in continuous exercise. As such, when did not homogenize energy consumption of exercise in EPOC results show that the interval exercise is more effective, but it is difficult to suggest that the effect of the interval exercise is effective when the total exercise energy consumption is not significantly different. By dif- ference energy consumption of exercise in the continuous and interval exercise resulted in higher initial EPOC in the interval exercise but higher total energy consump- tion in the continuous exercise, so ensuring equivalence between the exercise is considered important. Thus, it our data suggest that the equalization of calories during exercise is an important factor in determining EPOC and is an important factor to consider in presenting the effects of exercise. In this study, we consider EPOC to have been significantly increased, because caloric expenditure was well-controlled during food-intake and exercise. In a study that homogenized energy consumption be- tween continuous and spaced movements, McGarvey et al.16 reported no significant differences in EPOC between 31 min of continuous exercise at 65% of VO2 max, and an interval exercise pattern of 90% VO2 max for 2 min fol- lowed by 30% VO2 max for 3 min, repeated 7 times for a total of 35 min. This may reflect differences in the EPOC measurement method. Most of the increase in oxygen consumption after exercise occurs in the early stages of recovery. As recovery continues, oxygen consumption de- creases drastically, and the size increase with increasing standardized-duration decreases. Therefore, it is necessary to end when VO2, HR, and RER return to the baseline. In Figure 1. Comparison of oxygen consumption during exercise Figure 2. Comparison of lipid profile on EPOC in Con EX vs Inter EX J Exerc Nutrition Biochem. 2019;23(2):045-050, http://dx.doi.org/10.20463/jenb.2019.0016 49 Effect of interval exercise versus continuous exercise on EPOC Journal of Exercise Nutrition & Biochemistry addition, the method to homogenize energy consumption during exercise and EPOC was performed well. In our study, interval exercise resulted in post-exercise VO2, kcal, HR and EPOC measures 40% higher than for continuous cycle ergometer exercise. The results of this study support the hypothesis that the magnitude of EPOC and its duration is primarily dependent on exercise inten- sity6,14. In relation to the increase in EPOC, the ‘Oxygen Debt’ theory may explain this finding. For example, it could be explained by the energy cost to resynthesize glycogen from lactate, the exercise-induced increase in core temperature, the resynthesis of ATP/CP stores, and changes in cytokine release20,22. Consequently, greater exercise intensity may further increase the oxygen deficit at the onset of exercise, thereby affecting the body's ho- meostatic nature and resulting in a larger post-exercise O2 intake. Mechanisms responsible for this could extend to increases in VO26,23. As shown in Table 2 of our study, the increase in oxygen deficit increased by more than 20% for interval exercise, as compared with continuous exer- cise. These results are therefore consistent with previous studies that show increased oxygen consumption during recovery after high intensity interval exercise, because of increased oxidative metabolism that supplements energy expenditure after exercise24-26. In conclusion, our study confirmed that after homoge- nizing the energy expenditure of continuous and interval exercise on a cycle ergometer, EPOC is higher in interval exercise than continuous exercise in subjects who are in their twenties. This observation is important as it may help us understand why interval exercise has a greater propensity to induce weight loss than continuous exer- cise. Furthermore, these data provides a metabolic basis for enhanced fat loss during interval training that will be useful in establishing public health guidelines on exercise recommendations and weight management practices to reduce body fat. This should be qualified as only appro- priate for young and healthy older populations who can perform such exercises. These exercise recommendations may promote weight loss and health, and result in better health outcomes in "time poor" modern lifestyles. Conse- quently, we suggest that interval exercise may be a more effective strategy in reducing body fat for energy expen- diture increase than continuous exercise. ACKNOWLEDGMENTS This work was supported by the Ministry of Educa- tion of the Republic of Korea and the National Research Foundation of Korea (NRF-2016S1A5B8914314). REFERENCES 1. American College of Sports Medicine. ACSM's guidelines for exercise testing and prescription, 10thEd. Lippincott Williams & Wilkins. 2018. 2. Ramos JS, Dalleck LC, Tjonna AE, Beetham KS, Coombes JS. The impact of high-intensity interval training versus moderate-intensity continuous training on vascular func- tion: a systematic review and meta-analysis. Sports Med. 2015;45:679-92. 3. Peake JM, Tan SJ, Markworth JF, Broadbent JA, Skinner TL, Cameron-Smith D. Metabolic and hormonal responses to isoenergetic high-intensity interval exercise and continuous moderate-intensity exercise. Am J Physiol Endocrinol Metab. 2014;307:E539-52. 4. Daly W, Seegers CA, Rubin DA, Dobridge JD, Hackney AC. Relationship between stress hormones and testosterone with prolonged endurance exercise. Eur J Appl Physiol. 2005;93:375-80. 5. Martins C, Stensvold D, Finlayson G, Holst J, Wisloff U, Kulseng B, Morgan L, King NA. Effect of moderate-and high-intensity acute exercise on appetite in obese individuals. Med Sci Sports Exerc. 2015;47:40-8. 6. Børsheim E, Bahr R. Effect of exercise intensity, duration and mode on post-exercise oxygen consumption. Sports Med. 2003;33:1037-60. 7. Cunha FA, Midgley AW, McNaughton LR, Farinatti PT. Effect of continuous and intermittent bouts of isocaloric cycling and running exercise on excess postexercise oxygen consump- tion. J Sci Med Sport. 2016;19:187-92. 8. Weston KS, Wisløff U, Coombes JS. High-intensity interval training in patients with lifestyle-induced cardiometabolic disease: a systematic review and meta-analysis. Br J Sports Med. 2014;48:1227-34. 9. Laursen PB, Jenkins DG. The scientific basis for high-inten- sity interval training: optimising training programmes and maximising performance in highly trained endurance athletes. Sports med. 2002;32:53-73. 10. Guiraud T, Nigam A, Gremeaux V, Meyer P, Juneau M, Bosquet L. High-intensity interval training in cardiac rehabili- tation. Sports med. 2012;42:587-605. 11. Larsen I, Welde B, Martins C, Tjønna AE. High-and moder- ate-intensity aerobic exercise and excess post-exercise oxy- gen consumption in men with metabolic syndrome. Scand J Med Sci Sports. 2014;24:e174-9. 12. Mooren FC. Excess Postexercise Oxygen Consumption. Encyclopedia of Exercise Medicine in Health and Disease. SpringerLink. 2012. 13. Short KR, West JM, Sedlock DA. The effect of upper body exercise intensity and duration on post-exercise oxygen con- sumption. Int J Sports Med. 1996;17:559-63. 14. LaForgia J, Withers RT, Gore CJ. Effects of exercise intensity and duration on the excess post-exercise oxygen consump- tion. J Sports Sci. 2006;24:1247-64. 15. Schaun GZ, Alberton CL, Ribeiro DO, Pinto SS. Acute effects of high-intensity interval training and moderate-intensity con- tinuous training sessions on cardiorespiratory parameters in healthy young men. Eur J Appl Physiol. 2017;117:1437-44. 16. McGarvey W, Jones R, Petersen S. Excess post-exercise oxygen consumption following continuous and interval cycling exercise. Int J Sport Nutr Exerc Metab. 2005;15:28-37. 17. Schaun GZ, Pinto SS, Praia ABC, Alberton CL. Energy ex- penditure and EPOC between water-based high-intensity J Exerc Nutrition Biochem. 2019;23(2):045-050, http://dx.doi.org/10.20463/jenb.2019.0016 50 Effect of interval exercise versus continuous exercise on EPOC Journal of Exercise Nutrition & Biochemistry interval training and moderate-intensity continuous training sessions in healthy women. J Sports Sci. 2018;36:2053-60. 18. Tucker WJ, Angadi SS, Gaesser GA. Excess postexercise oxygen consumption after high-intensity and sprint interval exercise, and continuous steady-state exercise. J Strength Cond Res. 2016;30:3090-7. 19. Gerber T, Borg ML, Hayes A, Stathis CG. High-intensity in- termittent cycling increases purine loss compared with work- load-matched continuous moderate intensity cycling. Eur J Appl Physiol. 2014;114:1513-20. 20. Townsend JR, Stout JR, Morton AB, Jajtner AR, Gonzalez AM, Wells AJ, Mangine GT, McCormack WP, Emerson NS, Robinson EH, Hoffman JR, Fragala MS, Cosio-Lima L. Ex- cess post exercise oxygen consumption (EPOC) following multiple effort sprint and moderate aerobic exercise. Kinesiol- ogy, 2013;45:16-21. 21. Williams CB, Zelt JG, Castellani LN, Little JP, Jung ME, Wright DC, Tschakovsky ME, Gurd BJ. Changes in mecha- nisms proposed to mediate fat loss following an acute bout of high-intensity interval and endurance exercise. Appl Physiol Nutr Metab. 2013;38:1236-44. 22. Sedlock DA. Post-exercise energy expenditure after cycle ergometer and treadmill exercise. J Strength Cond Res. 1992;6:19-23. 23. Noordhof DA, de Koning JJ, Foster C. The maximal accumu- lated oxygen deficit method: a valid and reliable measure of anaerobic capacity? Sports Med. 2010;40:285-302. 24. Sedlock DA, Lee MG, Flynn MG, Park KS, Kamimori GH. Ex- cess postexercise oxygen consumption after aerobic exercise training. Int J Sport Nutr Exerc Metab. 2010;20:336-49. 25. Paoli A, Moro T, Marcolin G, Neri M, Bianco A, Palma A, Grimaldi K. High-Intensity Interval Resistance Training (HIRT) influences resting energy expenditure and respiratory ratio in non-dieting individuals. J Transl Med. 2012;10:237. 26. Hagberg JM, Mullin JP, Nagle FJ. Effect of work intensity and duration on recovery O2. J Appl Physiol Respir Environ Exerc Physiol. 1980;48:540–4.

Effect of interval exercise versus continuous exercise on excess post-exercise oxygen consumption during energy-homogenized exercise on a cycle ergometer.

[]

Jung, Won-Sang,Hwang, Hyejung,Kim, Jisu,Park, Hun-Young,Lim, Kiwon

eng

PMC8128237

RESEARCH ARTICLE A new approach to quantify angles and time of changes-of-direction during soccer matches Tomohiro KaiID1,2, Shin HiraiID3, Yuhei Anbe1, Yohei TakaiID1* 1 National Institute of Fitness and Sports in Kanoya, Kanoya, Japan, 2 Kagoshima United FC, Kagoshima, Japan, 3 Mizuno Corporation, Suminoe-ku, Japan * y-takai@nifs-k.ac.jp Abstract Background and aims Soccer players frequently perform change-of-directions (CODs) at various speeds during matches. However, tracking systems have shown limitations to measure these efforts. Therefore, the aim of the present study was to propose a new approach to measure CODs using a local positioning system (LPS), and clarify position-related difference in profile of CODs by using the approach. Methods The x- and y-coordinate data for each soccer player were measured with a local positioning system. Speed, acceleration, jerk, and direction of speed were derived from the coordinate data. Based on accelerations of above 2 m/s2, the onsets and ends of CODs derived from jerk were identified (COD duration). Changes of direction of speed (θCOD) were determined for the corresponding period. Six collegiate male soccer players performed CODs according to 13 set angles (0–180˚; every 15˚) so that differences between θCOD and set angle could be determined (Exp. 1). Relative frequency distributions of θCOD and number of CODs were determined in 79 collegiate and amateur male soccer players during 9 soccer matches (Exp. 2). Results In Exp. 1, θCOD was positively related to set angle (r = 0.99). Each θCOD was smaller than the corresponding set angle, and the difference became greater with increasing COD angle. In Exp. 2, The number of CODs in a match was 183 ± 39 across all positions. There were no significant position-related differences in the number of CODs. The duration of a COD was 0.89 ± 0.49 s across all positions. The relative frequency distribution of θCOD revealed that the number of CODs at 0–15˚ and 105–135˚ tended to be higher than those at other angles during soccer matches. Further, θCOD was affected by the speed at the onset of COD during soccer matches (Exp. 2). PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 1 / 12 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Kai T, Hirai S, Anbe Y, Takai Y (2021) A new approach to quantify angles and time of changes-of-direction during soccer matches. PLoS ONE 16(5): e0251292. https://doi.org/10.1371/ journal.pone.0251292 Editor: Filipe Manuel Clemente, Instituto Politecnico de Viana do Castelo, PORTUGAL Received: February 9, 2021 Accepted: April 26, 2021 Published: May 17, 2021 Peer Review History: PLOS recognizes the benefits of transparency in the peer review process; therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. The editorial history of this article is available here: https://doi.org/10.1371/journal.pone.0251292 Copyright: © 2021 Kai et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Funding: This study was supported by a NIFS project for TASS. The funders had no role in study Conclusions The current findings demonstrate that θCOD derived from direction of speed and jerk may be a new indicator for evaluating COD during soccer matches. Introduction Recently, coaches and sports scientists of ball sports team use tracking technologies for design- ing training program and players’ condition, because it has become easy to measure players’ locations in ball sports using global positioning system (GPS) and local positioning system (LPS) [1, 2]. Many earlier studies have demonstrated the validity of position, speed and accel- eration data obtained from such a tracking technology, compared to a 3D motion capture sys- tem, radar/laser guns and timing gates [3–9]. Further, LPS is superior to GPS, whereas each system has a certain validity [10]. These systems are capable of measuring players’ coordinates, and for quantifying players’ acceleration, distance and numbers of actions in relation to speed derived from time-motion analysis. These parameters are predictors of match outcome and periodization on daily training [11, 12]. Besides time-motion analysis, the parameters obtained from tracking technologies are applied to quantify CODs locomotion in sport-specific course and small-sided games [3–6, 9, 10]. In the earlier studies, various type of courses which angles of CODs are predetermined are set [3, 5, 6, 9, 10], and small court is used [4]. Although the earlier findings demonstrate the magnitude of speed and acceleration, the experimental design is not real soccer matches. Dur- ing soccer matches, players perform changes-of-direction (CODs) during locomotion [13]. Notational analysis has revealed that 30% of all actions during English FA Premier League play were CODs (e.g., forward, lateral and backward running) [14]. However, quantifying relevant data, such as the number and type of actions, is a lengthy process [15], and notational analysis may be arbitrary [16]. Therefore, a convenient analytic method to quantify CODs during soc- cer matches is needed. Fitzpatrick et al. [2] demonstrated that, in the English U-18 Premier League, the direction of players’ locomotion at a speed of 6.67 m/s or more ranged from 0˚ to 30˚ by using GPS. This suggests that during matches, youth soccer players move close to a straight line at relatively high speeds. Further, an earlier study of soccer matches revealed that greater distances are covered at moderate speeds of 3.89 to 5.28 m/s than at high-intensity speeds of 5.28 to 6.39 m/s [17]. Although soccer players frequently perform CODs at various speeds during matches, to the best of our knowledge, little information is available concerning COD profiles during soccer matches in relation to speed by using LPS. Force is theoretically the product of mass and acceleration. Acceleration can be useful in describing a player’s physical load during soccer matches. Dalen et al. [1] demonstrated that position-related differences in the number of accelerations (>2 m/s2) was found in the first division of the Norwegian league. When a player changes direction of locomotion, he exerts a certain force against the ground. At the same time, a certain level of acceleration is produced, and then the direction of the player’s locomotion changes, and the speed and/or the direction of speed changes [3]. Jerk, which is derived by differentiating acceleration by time, is used to detect the onset of human joint movement and the magnitude of the movement [18]. There- fore, jerk should be useful in identifying the onset and end of a COD for a given acceleration, and the change in direction of speed should correspond to the direction of the COD. During professional soccer matches, position-related differences in acceleration are found, indicating that side midfielders and defenders accelerate more often than other positions [1]. PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 2 / 12 design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. Further, midfielders, relative to other positions, performed fewer CODs of 90 degrees or less [14]. Considering these findings, profile of CODs would differ among positions. This study thus proposes a new approach that uses direction of speed and jerk in quantifying CODs of soccer players during soccer matches, and to clarify position-related difference in profile of CODs by using the approach. Materials and methods Experimental design This investigation was conducted according to the Declaration of Helsinki and approved by the local Ethics Committee for human experimentation (#11–101). All participants were informed of the experimental procedures and possible risks of the measurements beforehand. Oral informed consent was obtained from each subject before each match. This study consisted of two experiments for quantifying CODs during soccer matches in relation to various speeds. In both experiments, players’ locations were measured with LPS. In the first experiment 1 (Exp. 1), male soccer players performed CODs in directions determined by 13 set angles (every 15˚, ranging from 0˚ to 180˚) while jogging at speeds of approximately 1.0 and 3.0 m/s. After the player turned at the determined location, he ran through a gate set at a distance of 2 m from the corresponding location. The participants were asked to perform CODs as fast and quickly as possible when they turned in a given direction. S1 Fig presents typical trajectory data and kinematic data (to be discussed below) in each angle for one player. In the second experiment 2 (Exp. 2), data were collected from 9 official soccer matches in Division 1 of a regional collegiate male soccer league and the Division 5 of a regional amateur soccer league for collegiate and amateur soccer players. Data were analyzed for the players who played for 90 min. Participants In the Exp. 1, six collegiate male soccer players (age, 21.0 ± 1.5 years, height, 172.8 ± 6.1 cm, body mass, 66.8 ± 9.2 kg; means ± SDs) participated in Exp. 1. They were field players, and belonged in the same team competing in a national university league in Japan, and had experi- ence of competitive soccer training for >9 years. They had participated in regular soccer-spe- cific training for more than five days (>1.5 hours/day) per week. Seventy-nine collegiate and amateur male soccer players (23.0 ± 4.1 years, 173.9 ± 5.1 cm, 67.5 ± 6.2 kg) involved in Exp. 2, and got in the official soccer matches in in Division 1 of a regional collegiate male soccer league or the Division 5 of a regional amateur soccer league. The number of players in each position was as follows; 23 players for central backs (CB, 23.4 ± 4.7 years, 177.4 ± 5.0 cm, 71.2 ± 5.7 kg), 16 players for side backs (SB, 22.6 ± 3.4 years, 171.7 ± 3.9 cm, 65.2 ± 4.7 kg), 15 players for central midfielders (CMF, 22.2 ± 2.9 years, 172.3 ± 5.0 cm, 64.8 ± 6.3 kg), 14 players for side midfielders (SMF, 23.7 ± 5.0 years, 172.3 ± 4.2 cm, 65.9 ± 4.9 kg), and 11 players for forwards (FW, 21.8 ± 3.7 years, 174.3 ± 4.5 cm, 69.8 ± 5.9 kg), respectively. Goalkeepers were excluded from data analysis. All participants involved in Exp. 1 and 2 were free of cardiovascular, metabolic, and immu- nologic disorders and/or orthopedic abnormalities, and were not using any medications that affected their muscular function and size. All study participants provided informed consent, and the study design was approved by an ethics review board (the Ethics Committee in National Institute of Fitness and Sports in Kanoya for human experimentation (#11–101)). PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 3 / 12 Players’ coordinate data X- and y-coordinate data for each player were measured with LPS (ZXY Sports Tracking, Chyronhego, New York, USA) at a sampling frequency of 20 Hz. A belt with a sensor (approx. 20 g) under their uniform were attached for each player. In Exp.2, three examiners helped to wear the sensor for starters before the start of soccer matches. Players were instructed to take off the belt if they felt uncomfortable during matches. The data obtained were processed as described below using Matlab (Mathworks ver. 2018b, New York, USA). 1 Kinematic data and filtering process. To obtain the smoothed time-series data for jerk, the time-series data of x- and y-coordinates were processed by a second-order Butterworth low-pass filter employing a zero phase lag before analysis. To identify the appropriate cutoff frequency for the low-pass filter, we repeated the filtering process at every 1 Hz from 1 Hz to 6 Hz. The time-series data for jerk with and without the filtering process are presented in S2 Fig. As shown in S2 Fig, the use of cutoff frequencies of 3 Hz to 6 Hz resulted in noise in the smoothed time-series data, while cutoff frequencies of <2 Hz produced less noise. Thus a 2 Hz cutoff frequency was adopted in this study. 2 Kinematic data for each player. To determine players’ velocity, displacement from (t- 1) to (t+1) was defined as (x(t+1)-x(t-1), y(t+1)-y(t-1)) of the smoothed coordinate data for each player. Player speed of players (|V(t)|) in m/s was calculated by differentiating the dis- placement by time. Player’s acceleration (|A(t)|) in m/s2 was derived by differentiating |V(t)| by time. Finally, jerk (j(t)) in m/s3 was calculated by differentiating |A(t)| by time. jVðtÞj ¼ ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðxðt þ 1Þ calculated for each position. The number of CODs and analytical durations were also deter- mined. Further, we examined the relationship between the speed at the onset of a COD and θCOD during soccer matches. Statistical analysis Descriptive data are expressed as means and standard deviations (SDs). In Exp. 1, a one-way analysis of variance (ANOVA) was used to test differences in θCOD for all combinations of set angles. Pearson’s product-moment correlation coefficients (r) were estimated for the relation- ships between θCOD and set angles. In Exp. 2, skewness and kurtosis were used to test whether the relative frequency distributions of θCOD were normally distributed, according to the method of Yokoyama [19]. A one-way ANOVA was used to test position-related differences in number of CODs and duration of the analytical period. All statistical procedures were con- ducted with SPSS statistical software (SPSS 25.0, IBM, New York, USA). The significance level was set at 0.05. Results Differences between θCOD and the set angles (Exp. 1) Table 1 presents descriptive data for θCOD at each set angle and the differences between set angles and θCOD. θCOD increased with increasing set angles. Error values between the set angles and θCOD also increased with increasing set angles. Duration of the analytical period was 1.04 ± 0.62 s. θCOD was positively related to set angles (r = 0.99), as shown in Fig 2. Fig 1. Analysis method for θCOD. Dotted line in line plot of acceleration indicates a threshold of acceleration (2 m/s2). https://doi.org/10.1371/journal.pone.0251292.g001 PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 5 / 12 CODs profiles during soccer matches (Exp. 2) Table 2 presents descriptive data on the number of CODs during the soccer matches. The number of CODs in a match was 183 ± 39 across all positions. There were no significant posi- tion-related differences in the number of CODs. The duration of a COD was 0.89 ± 0.49 s across all positions. Fig 3 shows the relative frequency distributions of θCOD during the soccer matches. The val- ues of skewness indicate that the distributions of θCOD were symmetric for all positions, except for SB. The skewness of SB was negative, indicating that the distribution was skewed to the left. The kurtosis was platykurtic for positions other than FW. The distribution in FW revealed lep- tokurtic. As seen in Fig 3, the relative frequency of CODs at 0–15˚ and 105–135˚ tended to be higher than that of CODs at other angles. Table 3 provides descriptive data on the number of CODs per match in each bin. The speed at the onset of a COD was 1.36 ± 0.96 m/s, ranging from 0.01 m/s to 6.99 m/s. As seen in Fig 4, players executed CODs in different directions at relatively lower speeds of <5 m/ s, but CODs in limited directions (-30˚-30˚) occurred at higher speeds of >5 m/s. Discussion This study aims to propose a new approach that uses direction of speed and jerk in quantifying CODs of soccer players during soccer matches, and to clarify position-related difference in profile of CODs by using the approach. As the results, change in direction of speed, θCOD, which was derived from direction of speed and jerk, increased with increasing set angles of the predetermined course (Exp. 1). Further, the relative frequency of θCOD during soccer matches revealed a platykurtic distribution in positions other than FW, but, in that of FW, the distribu- tion was leptokurtic (Exp. 2). Therefore, the θCOD proposed in this study may be an index of Table 1. Descriptive data on θCOD and the difference between each θCOD and set angle. Angle of CODs θCOD a Dif 0˚ -3.2 ± 3.2 3.2 ± 3.2 b 15˚ 12.2 ± 2.7 2.7 ± 2.7 b 30˚ 23.4 ± 1.6 6.6 ± 1.6 c 45˚ 35.4 ± 2.8 9.4 ± 2.9 d 60˚ 45.9 ± 4.1 14.1 ± 4.1 e 75˚ 57.2 ± 5.6 17.6 ± 5.4 f 90˚ 68.1 ± 8.7 21.5 ± 8.5 g 105˚ 77.2 ± 5.1 27.8 ± 5.1 h 120˚ 89.2 ± 5.6 30.8 ± 5.6 135˚ 102.5 ± 3.9 31.8 ± 4.6 150˚ 112.0 ± 4.1 37.8 ± 4.0 i 165˚ 131.4 ± 4.9 33.4 ± 5.2 i 180˚ 155.0 ± 7.5 24.6 ± 7.7 Values are means and SDs. COD, change of direction; a, significant difference with all combinations; b, significant difference between the corresponding angle and set angles above 60˚, c, significant difference between corresponding angle and set angles above 75˚; d, significant difference between corresponding angle and set angles above 90˚, e, significant difference between corresponding angle and set angles above 105˚; f, significant difference between corresponding angle and set angles above 105˚, except for 180˚; g, significant difference between corresponding angle and set angles above 120˚, except for 180˚; h, significant difference between corresponding angle and 150˚; i, significant difference between corresponding angle and 180˚. https://doi.org/10.1371/journal.pone.0251292.t001 PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 6 / 12 angle of CODs, and profile of CODs during soccer matches obtained from θCOD may be posi- tion-specific. In Exp.1, each θCOD was smaller than the set angle, and the difference between each set angle and θCOD became larger as the angle increased. This may have been due to a difference in the set course and the trajectory of center of mass (COM) of a player’s body. For example, in the COD at 90 deg, a player moved in an arc, rather than at a right angle, as seen in S1 Fig. Fig 2. Relationship between θCOD and set angles. Black dotted plot indicates the mean value of each set angle. Grey dotted plots indicate individual’s data within each set angle. Dotted line indicates approximate line. https://doi.org/10.1371/journal.pone.0251292.g002 PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 7 / 12 The direction of COM during change-of-direction running differed from the set angle, and the angle for COM was smaller than the set angle [20–22]. Suzuki et al. [20] revealed that the difference was 2˚ at 30˚, 8˚ at 60˚, and 12˚ at 90˚, and the difference increased with COD angles. In this study, the corresponding values were 6.6˚, 14.1˚, and 21.5˚, respectively, also becoming greater as COD angle increased. The mean values of analytical duration were 1.04 s for Exp. 1, and 0.89 s for Exp. 2. Gran- ero-Gil et al. [23] have defined change-of-direction locomotion as curvilinear locomotion that lasts more than 0.8 s, and they attempted to detect CODs during soccer matches by using an inertial sensor. The analytical durations in the present study were close to this definition, pro- viding support for the threshold reported by Granero-Gil et al. [23]. On the other hand, the Table 2. Descriptive data on number of CODs, analytical duration, skewness and kurtosis of frequency distribution. Number per match (times) Duration per COD (s) Skewness Kurtosis All 183 ± 39 0.89 ± 0.49 -1.50 -31.50 b CB 175 ± 38 0.88 ± 0.49 a 0.34 -17.91 b SB 183 ± 43 0.90 ± 0.50 -2.20 b -13.87 b CMF 196 ± 38 0.90 ± 0.48 -0.11 -14.59 b SMF 195 ± 36 0.89 ± 0.50 a -1.00 -9.54 b FW 173 ± 39 0.93 ± 0.50 -0.56 13.22 b Values are means and SDs. All, all positions; CB, center backs; SB, side backs; CMF, central midfielders; SMF, side midfielders; FW, forwards a, Significant difference in the measured variable between FW and other positions b, Significant different from normal distribution https://doi.org/10.1371/journal.pone.0251292.t002 Fig 3. Relative frequency distribution of θCOD per match. Each bin is set at every 15˚. A: all positions, B: center backs (CB), C: side backs (SB), D: central midfielders (CMF), E: side midfielders (SMF), F: forwards (FW). https://doi.org/10.1371/journal.pone.0251292.g003 PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 8 / 12 number of CODs in the present matches was approx. 183, lower than the corresponding value (471 times) reported by Granero-Gil et al. [23]. This discrepancy may be due to differences in the method of analysis. In this study, θCOD was estimated based on acceleration above 2 m/s2 [1], while the earlier study used the definition above to identify changing-of-direction locomo- tion [23]. Another reason may be due to be the technical error of acceleration measured by Table 3. Descriptive data on number of CODs per match in each bin range. CB SB CMF SMF FW 0–15˚ 16.7 ± 5.5 21.3 ± 7 17.2 ± 5.1 21.8 ± 7.5 22.4 ± 5.9 15–30˚ 13.6 ± 4.9 15.5 ± 5.5 15.5 ± 4.5 18.1 ± 5.4 15.6 ± 4.5 30–45˚ 13.1 ± 4.0 12.4 ± 4.6 13.7 ± 4.0 12.8 ± 4.9 15.8 ± 4.5 45–60˚ 12.2 ± 3.8 12.9 ± 4.6 15.6 ± 4.5 12.7 ± 2.9 13.4 ± 4.5 60–75˚ 13.5 ± 4.5 11.8 ± 4.7 16.1 ± 4.1 12.9 ± 3.8 10.6 ± 2.8 75–90˚ 14.1 ± 4.9 14.8 ± 5.0 16.9 ± 6.2 14.9 ± 4.4 15.9 ± 6 90–105˚ 16.3 ± 6.4 15.9 ± 4.9 18.9 ± 4.3 18.0 ± 4.4 14.2 ± 6.2 105–120˚ 18.6 ± 5.9 18.4 ± 6.4 19.9 ± 7.1 21.9 ± 6.3 17.4 ± 6.4 120–135˚ 17.7 ± 6.1 20.8 ± 6.7 20.6 ± 7.3 19.1 ± 6 15.2 ± 5.1 135–150˚ 16.4 ± 5.7 15.6 ± 5.4 18.2 ± 6.6 18.8 ± 6.8 13.7 ± 5.0 150–165˚ 11.8 ± 4.5 14.2 ± 5.9 13.2 ± 3.6 13.1 ± 4.6 8.8 ± 4.5 165–180˚ 8.0 ± 3.3 7.3 ± 3.6 7.2 ± 3.5 7.2 ± 3.5 6.6 ± 2.8 180˚ 2.7 ± 1.9 2.9 ± 2.1 3.1 ± 1.9 3.5 ± 2.2 3.3 ± 2.2 Values are means and SDs. All, all positions; CB, center backs; SB, side backs; CMF, central midfielders; SMF, side midfielders; FW, forwards. https://doi.org/10.1371/journal.pone.0251292.t003 Fig 4. Relationships between θCOD and speed at onset of changing of direction (COD). Black filled circle, the speed of more than 5 m/s; Grey filled circle, the speed below 5 m/s. A: all positions, B: center backs (CB), C: side backs (SB), D: central midfielders (CMF), E: side midfielders (SMF), F: forwards (FW). https://doi.org/10.1371/journal.pone.0251292.g004 PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 9 / 12 LPS. In fact, Linke et al. [4] have demonstrated that root mean square error of acceleration in sport-specific course ranged from 0.49 m/s2 to 1.34 m/s2. The degree of skewness demonstrated that θCOD during soccer matches was distributed symmetrically, except for players in the side back position. This indicates that no laterality of angles of CODs may be found for amateur soccer players. On the other hand, the asymmetric distribution of θCOD for SB may have been due to inter-individual differences in the distribu- tion of θCOD among SBs. In fact, the distribution of θCOD was asymmetric distribution for only one SB player (skewness: -2.85) but was symmetric for the remaining SB players (skewness: -1.91 to 0.65). To this point, further investigation of a larger sample size is needed. In positions other than FW, the kurtosis analysis revealed a platykurtic distribution of θCOD during soccer matches, indicating that soccer players changed direction at various angles during soccer matches. As seen in Fig 3, relative frequency of COD angles around 120 deg was more, regard- less of left or right. Bloomfield et al. found that the number of CODs within 0˚-90˚ accounted for more than 80% of the total number of CODs [14]. The corresponding value in this study was approx. 50%. The discrepancy may be attributable to different analysis methods. For FW, on the other hand, the distribution of θCOD during soccer matches was leptokurtic, indicating that FW may perform CODs with narrower angle than other positions. During soccer matches, θCOD ranged from 0˚ to 30˚ when speed at onset of a COD was rela- tively high (>5 m/s) (Fig 4). Kai et al. [24] revealed that the trajectory of players above 5 m/s was similar to liner locomotion. Fitzpatrick et al. [2] also demonstrated that direction of loco- motion at speeds of above 6.7 m/s ranged from 0˚ to 30˚. Propulsive force decreases with increasing running speed [25], implying that there is less space to accelerate the player’s body at a given high speed. Taken together, this evidence suggests that soccer players perform straight runs or CODs with a narrow direction angle (<30˚) at relatively high speeds. There are some limitations in this study. Firstly, LPS are limited in high intensity effort such as high speed straight running and CODs [3, 4, 6, 9]. During sport-specific course and small- sided games, the root mean square errors of instant speed over 4.17 m/s range from 0.34 m/s to 0.39 m/s [4]. However, the values may not be enough to change relationships between instant speed at onset of CODs and angles of CODs (Fig 4). Secondly, the approach used in this study cannot be used to determine direction of a body. For example, if a player runs backward in the opposite direction immediately after he moves in a straight run, the locomotion is estimated as a COD with a 180˚ turn. Further investigation of this point is needed. Thirdly, parameters derived from LPS may be influenced by measurement condition and experimental protocol [26], although the validity of tracking systems is shown in earlier studies abovementioned [3–9]. Unfortunately, we have the relevant data, and further investigation is needed in this point. In practical application, this study demonstrated that the relative distribution of θCOD was position-specific, and θCOD was affected by speed at the onset of the COD during soccer matches. To design regular soccer training that meet physical demands for each position, coaches and strength conditioners for soccer players have to know what kind of locomotion is taking place during soccer matches. Considering the current findings, the players of positions other than FW need to perform CODs toward various direction at relatively low speed (< 5 m/s). For FW, however, it’s better to perform CODs toward narrow angle (< 30 deg) at rela- tively high speed (> 5 m/s). Thus, the current findings may be useful information to achieve the principle of training specificity for soccer. Conclusion This study proposed a new approach to quantifying angle of CODs (θCOD) during soccer matches by using direction of speed and jerk. As the results, θCOD increases with increasing PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 10 / 12 the predetermined set angle, although θCOD was smaller than the predetermined set angle. Fur- ther, the relative frequency of θCOD derived from the proposed approach revealed position- specific, and θCOD was affected by speed at the onset of the COD during soccer matches. The current findings suggest that the approach proposed in this study may be useful to quantify angle of CODs during soccer matches. Supporting information S1 Fig. An example of typical trajectory data and kinematic data (speed, acceleration, jerk and direction of speed) in each angle when one player performed CODs in directions determined by 13 set angles (every 15˚, ranging from 0˚ to 180˚). A: 0˚, B: 15˚, C: 30˚, D: 45˚, E: 60˚, F: 75˚, G: 90˚, H: 105˚, I: 120˚, J: 135˚, K: 150˚, L: 165˚, M: 180˚. A bold line over- lapped in line plot indicates an analytical period. (ZIP) S2 Fig. The time-series data for jerk with and without the filtering process are presented. The use of cutoff frequencies of 3–6 Hz resulted in noise in the smoothed time-series data, while cutoff frequencies <2 Hz produced less noise. Thus, a 2 Hz cutoff frequency was adopted. (TIF) S1 Data. (XLSX) Author Contributions Conceptualization: Yohei Takai. Data curation: Tomohiro Kai, Yuhei Anbe. Formal analysis: Tomohiro Kai. Investigation: Yuhei Anbe. Methodology: Shin Hirai, Yohei Takai. Project administration: Yohei Takai. Writing – original draft: Tomohiro Kai. Writing – review & editing: Shin Hirai, Yohei Takai. References 1. Dalen T, Ingebrigtsen J, Ettema G, Hjelde GH, Wisløff U. Player Load, Acceleration, and Deceleration During Forty-Five Competitive Matches of Elite Soccer. J Strength Cond Res. 2016; 30(2):351–359. https://doi.org/10.1519/JSC.0000000000001063 PMID: 26057190 2. Fitzpatrick J, Linsley A, Musham C. Running the curve: a preliminary investigation into curved sprinting during football match-play. SPSR. 2019; 55:1–3. 3. Frencken WG, Lemmink KA, Delleman NJ. Soccer-specific accuracy and validity of the local position measurement (LPM) system. J Sci Med Sport. 2010; 13(6):641–645. https://doi.org/10.1016/j.jsams. 2010.04.003 PMID: 20594910 4. Linke D, Link D, Lames M. Validation of electronic performance and tracking systems EPTS under field conditions. PLoS One. 2018; 13(7):e0199519. https://doi.org/10.1371/journal.pone.0199519 PMID: 30036364 5. Luteberget LS, Spencer M, Gilgien M. Validity of the Catapult ClearSky T6 Local Positioning System for Team Sports Specific Drills, in Indoor Conditions. Front Physiol. 2018; 9:115. https://doi.org/10.3389/ fphys.2018.00115 PMID: 29670530 PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 11 / 12 6. Ogris G, Leser R, Horsak B, Kornfeind P, Heller M, Baca A. Accuracy of the LPM tracking system con- sidering dynamic position changes. J Sports Sci. 2012; 30(14):1503–1511. https://doi.org/10.1080/ 02640414.2012.712712 PMID: 22906154 7. Rico-Gonza´lez M, Arcos A, Clemente FM, Rojas-Valverde D, Pino-Ortega J. Accuracy and reliability of local positioning systems for measuring sport movement patterns in stadium-scale: A systematic review. Appl Sci. 2020; 10(17):5994. https://doi.org/10.3390/app10175994. 8. Sathyan T, Shuttleworth R, Hedley M, Davids K. Validity and reliability of a radio positioning system for tracking athletes in indoor and outdoor team sports. Behav Res Methods. 2012; 44(4):1108–1114. https://doi.org/10.3758/s13428-012-0192-2 PMID: 22477436 9. Stevens TG, de Ruiter CJ, van Niel C, van de Rhee R, Beek PJ, Savelsbergh GJ. Measuring acceleration and deceleration in soccer-specific movements using a local position measurement (LPM) system. Int J Sports Physiol Perform. 2014; 9(3):446–456. https://doi.org/10.1123/ijspp.2013-0340 PMID: 24509777 10. Hoppe MW, Baumgart C, Polglaze T, Freiwald J. Validity and reliability of GPS and LPS for measuring distances covered and sprint mechanical properties in team sports. PLoS One. 2018; 13(2):e0192708. https://doi.org/10.1371/journal.pone.0192708 PMID: 29420620 11. Buchheit M, Allen A, Poon TK, Modonutti M, Gregson W, Di Salvo V. Integrating different tracking sys- tems in football: multiple camera semi-automatic system, local position measurement and GPS technol- ogies. J Sports Sci. 2014; 32(20):1844–1857. https://doi.org/10.1080/02640414.2014.942687 PMID: 25093242 12. Oliva-Lozano JM, Rojas-Valverde D, Go´mez-Carmona CD, Fortes V, Pino-Ortega J. Impact of contex- tual variables on the representative external load profile of Spanish professional soccer match-play: A full season study. Eur J Sport Sci. 2020:1–10. https://doi.org/10.1080/17461391.2020.1751305 PMID: 32233969 13. Bangsbo J. The physiology of intermittent activity in football. In: Reilly T, Bangsbo J, Hughes M, London E, Spon FN, editors. Science and Football III 1997. p. 43–53. 14. Bloomfield J, Polman R, O’Donoghue P. Physical Demands of Different Positions in FA Premier League Soccer. J Sports Sci Med. 2007; 6(1):63–70. PMID: 24149226 15. Stein M, Seebacher D, Marcelino R, Schreck T, Grossniklaus M, Keim DA, et al. Where to go: Computa- tional and visual what-if analyses in soccer. J Sports Sci. 2019; 37(24):2774–2782. https://doi.org/10. 1080/02640414.2019.1652541 PMID: 31402759 16. Franks IM, Miller G. Eyewitness testimony in sport. J Sport Behav. 1986. 17. Barros RM, Misuta MS, Menezes RP, Figueroa PJ, Moura FA, Cunha SA, et al. Analysis of the dis- tances covered by first division brazilian soccer players obtained with an automatic tracking method. 2007; 6(2):233–242. 18. Flash T, Hogan N. The coordination of arm movements: an experimentally confirmed mathematical model. J Neurosci. 1985; 5(7):1688–1703. https://doi.org/10.1523/JNEUROSCI.05-07-01688.1985 PMID: 4020415 19. Yokoyama Y. A study on normality of the standing height in the Japanese youth. J Anthrop Soc Nippon. 1978; 86(4):313–320. https://doi.org/10.1537/ase1911.86.313. 20. Suzuki Y, Ae M, Enomoto Y. A kinematic analysis of cutting motion with side-step and cross-step tech- niques. Japan J Phys Educ Health and Sport Sci. 2010; 55:81–95. https://doi.org/10.5432/jjpehss.09038. 21. Suzuki Y, Ae M, Takenaka S, Fujii N. Comparison of support leg kinetics between side-step and cross- step cutting techniques. Sports Biomech. 2014; 13(2):144–153. https://doi.org/10.1080/14763141. 2014.910264 PMID: 25122999 22. Vanrenterghem J, Venables E, Pataky T, Robinson MA. The effect of running speed on knee mechani- cal loading in females during side cutting. J Biomech. 2012; 45(14):2444–2449. https://doi.org/10.1016/ j.jbiomech.2012.06.029 PMID: 22835648 23. Granero-Gil P, Go´mez-Carmona CD, Bastida-Castillo A, Rojas-Valverde D, de la Cruz E, Pino-Ortega J. Influence of playing position and laterality in centripetal force and changes of direction in elite soccer players. PLoS One. 2020; 15(4):e0232123. https://doi.org/10.1371/journal.pone.0232123 PMID: 32324801 24. Kai T, Horio K, Aoki T, Takai Y. High-intensity running is one of the determinats for achieving score- box possession during soccer matches. Football Sci. 2018; 15:61–69. 25. Cavagna GA, Komarek L, Mazzoleni S. The mechanics of sprint running. J Physiol. 1971; 217(3):709– 721. https://doi.org/10.1113/jphysiol.1971.sp009595 PMID: 5098087 26. Rico-Gonza´lez M, Los Arcos A, Rojas-Valverde D, Clemente FM, Pino-Ortega J. A Survey to Assess the Quality of the Data Obtained by Radio-Frequency Technologies and Microelectromechanical Sys- tems to Measure External Workload and Collective Behavior Variables in Team Sports. Sensors. 2020; 20(8). https://doi.org/10.3390/s20082271 PMID: 32316325 PLOS ONE Quantifying angle and time of changes-of-direction PLOS ONE | https://doi.org/10.1371/journal.pone.0251292 May 17, 2021 12 / 12

A new approach to quantify angles and time of changes-of-direction during soccer matches.

05-17-2021

Kai, Tomohiro,Hirai, Shin,Anbe, Yuhei,Takai, Yohei

eng

PMC7560936

Vol.:(0123456789) 1 3 European Journal of Applied Physiology (2020) 120:2397–2405 https://doi.org/10.1007/s00421-020-04463-w ORIGINAL ARTICLE High‑intensity decreasing interval training (HIDIT) increases time above 90% ̇VO2peak Filippo Vaccari1,2  · N. Giovanelli1,2 · S. Lazzer1,2 Received: 16 July 2019 / Accepted: 4 August 2020 / Published online: 11 August 2020 © The Author(s) 2020 Abstract Purpose Training near ̇VO2max is considered to be the most effective way to enhance ̇VO2max. High-intensity interval training (HIIT) is a well-known time-efficient training method for improving cardiorespiratory and metabolic function and ̇VO2max. While long HIIT bouts allow ̇VO2max to be achieved quickly, short HIIT bouts improve time to exhaustion (Tlim). The aim of this study was to evaluate the time spent above 90% ̇VO2peak (T > 90% ̇VO2peak) during three different HIIT protocols. Methods Twelve cyclists performed three HIIT sessions. Each protocol had the same work and recovery power and ratio of work·recovery−1. The protocols consisted of long-interval HIIT (LIHIIT, 3 min work—2 min recovery), short-interval HIIT (SIHIIT, 30 s work—20 s recovery), and high-intensity decreasing interval training (HIDIT, work from 3 min to 30 s and recovery from 2 min to 20 s). T > 90% ̇VO2peak, Tlim, blood lactate [La], and rate of perceived exertion (RPE) were measured at Tlim. Results T > 90% ̇VO2peak was greater in HIDIT (312 ± 207 s) than in SIHIIT (182 ± 225 s; P = 0.036) or LIHIIT (179 ± 145 s; P = 0.027). Tlim was not significantly different (P > 0.05) between HIDIT (798 ± 185 s), SIHIIT (714 ± 265 s), and LIHIIT (664 ± 282). At Tlim, no differences in [La] and RPE were found between protocols (P > 0.05). Conclusion HIDIT showed the highest T > 90% ̇VO2peak, suggesting that it may be a good strategy to increase time close to ̇VO2peak, despite similar Tlim, [La], and RPE at Tlim. Keywords ̇VO2max · ̇VO2max training · Time at ̇VO2max · HIIT Abbreviations %CP-Load Peak Percentage of critical power relative to load peak % ̇VO2peak Oxygen consumption in percentage relative to its peak %HRpeak Heart rate in percentage relative to its peak [La] Blood (capillary) lactate concentration ANOVA Analysis of variance CP Critical power CR10 Scale Validated scale of perceived exertion ES Effect size HIDIT Decreasing intervals HIIT (combining high phosphocreatine intensity from 3′ to 30″ and low intensity from 2′ to 20″) HIIT High-intensity interval training ICP Intermittent critical power LIHIIT Long intervals HIIT (3′ high—2′ low-intensity) [Pcr] Muscular concentration of phosphocreatine QR Gas-exchange ratio RPE Rate of perceived exertion SIHIIT Short intervals HIIT (30″ high—20″ low-intensity) Tlim (Time to exhaustion) T > 90% ̇VO2peak Time spent above 90% ̇VO2peak ̇VCO2 CO2 output ̇VO2 Pulmonary O2 uptake ̇VO2max Maximal theoretical aerobic power Communicated by Håkan Westerblad. * Filippo Vaccari filippo.vaccari@live.com 1 Department of Medicine, University of Udine, P.le Kolbe 4, 33100 Udine, Italy 2 School of Sport Sciences, University of Udine, Udine, Italy 2398 European Journal of Applied Physiology (2020) 120:2397–2405 1 3 ̇VO2peak Maximal ̇VO2 achieved during incre- mental exercise W′ Amount of work that can be done dur- ing exercise above CP Introduction Maximal oxygen uptake ( ̇VO2max) refers to the oxygen con- sumption attained during a maximal exercise. It is reached when the ̇VO2 does not increase any further despite further increases in workload, and it defines the limits of the car- diorespiratory system (Hill and Lupton 1923). ̇VO2max is a relevant parameter of cardiorespiratory capacity, which is important for both endurance athletes (di Prampero 2003) and patients (Poole et al. 2012). It has been shown that, to improve ̇VO2max, a training protocol should prolong the time at which the oxygen uptake remains close to the maxi- mum (within 5–10% of ̇VO2max) (Wenger and Bell 1986; Midgley and Mc Naughton 2006). High-intensity interval training (HIIT) is very effective at maintaining the metabolic rate near ̇VO2max (Buchheit and Laursen 2013a), better than continuous steady-state training (Midgley and Mc Naughton 2006), and can be comprised of either short or long bouts of high intensity (work) alternated with recovery periods (recovery) at low intensity (or rest) (Buchheit and Laursen 2013a). The minimum intensity that allows one to reach ̇VO2max during a steady-state exercise is called critical power (CP). Theoretically, it is possible to maintain a metabolic steady state and prolong effort up to the CP threshold indefinitely. In contrast, above the CP, even if the external power out- put remains constant, ̇VO2 increases up to ̇VO2max, leading to exhaustion within a few minutes (Jones and Vanhatalo 2017). HIIT can be set based on CP, setting the work intervals above CP and the recovery intervals below CP (Morton and Billat 2004). The CP is mathematically defined as the power asymptote of the hyperbolic relationship between power output and time to exhaustion (Jones et al. 2010). Physiologically, CP represents the boundary between steady- state and non-steady-state exercise intensity domains (Jones et al. 2010; Jones and Vanhatalo 2017). Exercise above CP leads to reduced muscle phosphocreatine concentration [Pcr] and pH (Meyer 1988; Chidnok et al. 2013; Jones and Van- hatalo 2017), making it difficult to prolong exercise (i.e., W′: amount of work that can be done during exercise above CP) (Ferguson et al. 2010; Skiba et al. 2012, 2014, 2015). Since muscle ̇VO2 is related to muscle reduction [Pcr] (di Prampero and Margaria 1968; Meyer 1988), the faster [Pcr] is depleted, the faster the ̇VO2 increases. Conversely, during the recovery phase (below CP), [Pcr] resynthesis and W′ recovery follow an exponential trend (Meyer 1988; Ferguson et al. 2010; Skiba et al. 2012, 2014; Jones and Vanhatalo 2017; Vinetti et al. 2017). Indeed, when exercise generates a large depletion of [Pcr], the resynthesis rate is faster in the beginning of the recovery and it slows when approaching complete restoration. Thus, an HIIT protocol that aims to stimulate ̇VO2max should start with long work intervals (2–4 min) to quickly increase ̇VO2. Subsequently, when the subject approaches exhaustion, short intervals can help to prolong the exer- cise for longer: in this situation, the recovery ratio is fast and requires only few seconds to ensure sufficient recovery while simultaneously preventing the ̇VO2 from decreasing too much. Therefore, the aim of this study was to compare the time above 90% of ̇VO2peak (T > 90% ̇VO2peak) in three different HIIT protocols. The proposed HIIT protocols had the same intensity and work/recovery ratio and were structured as follows: (1) constant long intervals (LIHIIT); (2) decreasing interval duration (high-intensity decreasing interval training, HIDIT), and (3) constant short intervals (SIHIIT). It has been hypothesized that the T > 90% ̇VO2peak should be longer in HIDIT. We hypothesized that the protocol with longer intervals followed by shorter intervals would elicit longer time above 90%. Materials and methods Subjects Twelve middle-age amateur cyclists, all non-smokers, were enrolled in the study (41 ± 11 years; 76 ± 10 kg; ̇VO2peak 4.32 ± 0.47 L min−1), Table 1. They reported at least three training sessions per week in the previous 6 months. None Table 1 Descriptive characteristics of the participants (n = 12) All values are mean and standard deviation (SD) HR heart rate, ̇VO2peak peak oxygen consumption, CP critical power, W′ total work sustainable above critical power, High and Low inten- sity the average intensity sustained during HIIT tests Mean ± SD Min–Max Age (year) 41 ± 11 29–62 Body mass (kg) 76 ± 10 66–95 HRpeak (b min−1) 174 ± 10 155–193 ̇VO2peak (L min−1) 4.32 ± 0.47 3.66–5.10 Load peak (W) 356 ± 40 295–436 CP (W) 254 ± 30 212–320 W’ (kJ) 12.8 ± 4.1 8.5–22.7 High intensity (W) 297 ± 35 249–364 Low intensity (W) 212 ± 30 172–275 2399 European Journal of Applied Physiology (2020) 120:2397–2405 1 3 of the subjects had evidence of significant diseases or took regular medications. Study protocol The Ethics Committee of the Friuli-Venezia-Giulia approved the study (protocol number 9626). During the first visit to the laboratory, an operator explained the purposes and objectives of the study to each subject and obtained writ- ten informed consent. Then, participants underwent medical examinations and performed a maximal ramp-incremental exercise test on a cycle ergometer to measure the ̇VO2peak. Although the objectives were explained to all subjects, the study hypothesis was not revealed so as not to influence the results. After the first visit, the participants were examined three or four times to determine the critical power, and they performed the SIHIIT, HIDIT, and LIHIIT tests three times. Every visit was separated from the previous one by 2 days. Participants were instructed to avoid the consumption of caffeinated beverages for at least 8 h before each test and to abstain from vigorous physical activity in the 24 h pre- ceding each testing session. Every subject concluded the entire protocol within 4 weeks from the first visit. The criti- cal power parameters were used to program the HIIT tests. Subsequently, during the three HIIT tests, time to exhaustion (Tlim), T > 90% ̇VO2peak, blood lactate concentration [La], rate of perceived exertion using the Borg CR10 Scale (Borg et al. 2010), and ̇VO2 were measured at the 3rd minute and at the end of exercise. Incremental exercise The incremental exercise was performed under medical supervision, and standard safety procedures were followed. During the first visit, an operator instructed the subjects to correctly report the rate of perceived exertion on the CR10 scale (Borg et al. 2010). The incremental exercise, critical power trials, and HIIT test protocols were performed uti- lizing a cycle ergometer (CE) (Monark Ergomedic 839E). Every test was preceded by the same warm-up procedure: 10 min cycling at 100 W followed by 2-min resting. During the first warm-up, subjects chose their preferred pedaling cadence (~ 90 rpm). The incremental exercise was a con- stant incremental ramp test started at 100 W and gradually increased by 1 W every 2.4 s (25 W min−1) throughout the test until voluntary exhaustion. The exhaustion (during the incremental test and the HIITs) was defined as the inability to maintain the assigned cadence within 10 rpm longer than 5 s despite strong encouragement from the operator. ̇VO2 and ̇VCO2 were measured breath-by-breath using a metabolic unit (Quark CPET, Cosmed, Italy). The ventila- tion was measured by a turbine calibrated before each test with a 3-L syringe at three different flow rates. Calibration of O2 and CO2 analysers was performed before each test by utilizing calibration gas mixtures of known composition (16.00% O2; 4.00% CO2). ̇VO2peak corresponded to the highest mean ̇VO2 obtained in 30 s at the end of the incre- mental exercise. Power–duration relationship The same warm-up and cadence from the incremental test were also used for the critical power (CP) test. CP and the amount of work that could be done during exercise above CP (W′) (Jones and Vanhatalo 2017; Burnley and Jones 2018) were estimated from three to four high-intensity trials at exhaustion from 80 to 100% of the peak power detected dur- ing the incremental test and adopted to result in ‘exhaustion’ in a minimum of ~ 2 min and a maximum of ~ 15 min (Jones and Vanhatalo 2017). The work done in each of the separate exercise bouts has been plotted against Tlim. The follow- ing work (W) − time (t) linear regression was then used to find CP and W′ (Moritani et al. 1981; Hill 1993; Jones and Vanhatalo 2017): According to the equation, CP is given by the slope of the regression, and the W′ is the y-intercept. HIIT tests After the incremental test and the critical power trials, sub- jects performed three HIIT tests in a randomized order. The power during the work and recovery bouts and the work/ recovery duration ratio were the same in each trial, although the duration of the intervals was changed (see Table 1 for mean values). The ratio work/recovery time was set at 3/2 for all the training tests. The power used for the high-intensity bouts was customized for each subject and corresponded to the power that was supposed to lead to exhaustion in 5 min (300 s) according to the following equation (Jones et al. 2010): and it corresponded to approximately 117% of CP. The power used for the low-intensity bout was mirrored below CP (approximately 83% of CP). Thus, the CP threshold was exactly in the middle between the high and low intensities. The three tests were structured as follows (Fig. 1): Short intervals (SIHIIT): 30 s at high intensity and 20 s at low intensity repeated until volitional exhaustion of the subject. (1) W = CPt + W. (2) Power = W t = 300 s + CP, 2400 European Journal of Applied Physiology (2020) 120:2397–2405 1 3 High-intensity decremental interval training (HIDIT): 3 min at high intensity and 2 min at low intensity; 2 min at high intensity and 1 min and 20 s at low intensity; 1 min at high intensity and 40 s at low intensity; 45 s at high intensity and 30 s at low intensity; and finally 30 s at high intensity and 20 s at low intensity, repeated until volitional exhaustion of the subject. The high–low ratio intensity duration was always 3/2. Long intervals (LIHIIT): 3 min at high intensity and 2 min at low intensity repeated until volitional exhaustion of the subject. Throughout the HIIT protocols, the ventilatory param- eters were measured using a breath-by-breath metabolic unit (CPET, Cosmed, Italy) and then averaged every 5 s. Before, after 3 min and at the end of exercise, ̇VO2, HR, [La], and RPE were measured, and the respiratory quotient (RQ) was calculated. An operator collected a capillary blood sample from the earlobe to measure the [La] with a dedicated device (Lactate Pro 2, Arkaray Inc., Japan), while the subjects reported RPE consulting the CR10 scale positioned in front of them. Finally, the total time spent above 90% of ̇VO2peak was determined as the sum of each averaged 5-s when the ̇V O2 was equal to or higher than 90% of ̇VO2peak. Statistical analyses Statistical analysis was performed using SPSS 20.0 software (IBM, Chicago, USA) with significance set at P < 0.05. All results were expressed as the means and standard deviations (SD). The differences between HIIT training protocols in Tlim; T > 90% ̇VO2peak; T > 90% ̇VO2peak—Tlim−1; work above CP (calculated as the total time in seconds above CP multiply by the difference between the high-intensity power and CP, in Watts); average ̇VO2; and, finally, the values at the third minute and at Tlim ( ̇VO2, HR, [La], CR10-scale and RQ) were investigated. All parameters were analyzed by one-way repeated-measures analysis of variance (ANOVA). Where the analysis found a significant difference, planned contrast between HIDIT and SIHIIT and between HIDIT and LIHIIT were used with Bonferroni correction to deter- mine the origin of such effects. The confidence intervals (CIs) of the differences and the effect size (ES) were calcu- lated using Cohen’s d (0 < d < 0.20, small; 0.20 < d < 0.50, medium; d > 0.50, large) (Cohen 1988). The precision of Cp and W′ estimation was calculated comparing the param- eter estimates with the work-time model and with the time−1 model through a t test. For our purposes, a sample size of 12 subjects was calculated to have a statistical power of 80% to refute the null hypothesis and to obtain an ES of 0.88 with an alpha error of 0.05 and a beta error of 0.20 using a one-way ANOVA with Bonferroni correction, according to a previous Fig. 1 HIIT protocols for a representative subject. SIHIIT: short-inter- val HIIT (30″ high—20″ low-intensity); HIDIT: decreasing intervals HIIT (combining high intensity from 3′ to 30″ and low intensity from 2′ to 20″); LIHIIT: long-interval HIIT (3′ high—2′ low-intensity); the dotted lines represent the breath-by-breath ̇VO2 data averaged every 5 s; the dashed lines represent the threshold of 90% of ̇VO2peak; the solid lines represent the actual power 2401 European Journal of Applied Physiology (2020) 120:2397–2405 1 3 study (De Aguiar et al. 2013) that implemented a procedure similar to that of our study. Results Incremental test and CP trials Peak values attained during the incremental test, CP, total work above CP (W′), and the power imposed for the high- and low-intensity bouts are shown in Table 1. Although the attainment of ̇VO2peak was not set as a priori criteria for the constant work rate tests of the power–duration relationship, it was always reached by the subjects. The parameter esti- mates through the “work-time model” used for our purposes have been compared with the parameter estimates through the “1·time−1” model, and the results were comparable, as shown in Table 2. HIIT tests The power corresponding to high-intensity intervals was 117 ± 6% of CP, and the low-intensity power was 83 ± 6% of the CP (Table 3). T > 90% ̇VO2peak was significantly longer in HIDIT compared with SIHIIT (P = 0.036; ES: 0.62) and LIHIIT (P = 0.027; ES: 0.64) (Table 3, Fig. 2), and the ratio T > 90% ̇VO2peak—Tlim−1 tended to be higher in HIDIT than in SIHIIT and LIHIIT (Table 3). However, there were no differences in Tlim and in work > CP (P = 0.136) between the three protocols (Table 3). Finally, the average ̇VO2 maintained during the HIDIT test was significantly higher than in LIHIIT (P = 0.022; ES: 0.17) but not significantly different than in SIHIIT (P = 0.106; ES: 0.10). % ̇VO2peak after 3 min was similar between HIDIT and LIHIIT (P = 0.339; ES: 0.18), but it was significantly higher in HIDIT than SIHIIT (P = 0.006; ES: 0.83) (Table 3). Addi- tionally, %HRpeak after 3 min was similar between HIDIT and LIHIIT (P = 0.160; ES: 0.37), but it was significantly higher in HIDIT compared with SIHIIT (P = 0.019; ES: 0.61). Similarly, the CR10-scale after 3 min was similar in HIDIT and LIHIIT (P = 0.824; ES: 0.05) but significantly higher than SIHIIT (P = 0.031; ES: 0.55). Finally, RQ after 3 min was not significantly different in HIDIT and LIHIIT (P = 0.410; ES: 0.05), but it was significantly higher than in SIHIIT (P = 0.031; ES: 0.25) (Table 3). There was no significant difference in [La] at rest before the three tests (SIHIIT, HIDIT, and LIHIIT) (1.13 ± 0.20; 1.19 ± 0.26; and 1.17 ± 0.27  mmol  L−1, respectively; P > 0.05), and after 3 min, [La] was similar in HIDIT and LIHIIT (P = 0.007; ES: 0.05), but lower in SIHIIT (P = 0.003; ES: 0.78) (Table. 3). At Tlim, neither [La] nor ̇VO2, HR nor RPE were significantly different between the three tests (see Table 3). Table 2 Comparison of the power–duration relationship derived from 1/time model CP and work-time model CP R2 coefficient of determination of the linear regression, CP critical power, W′ total work sustainable above the critical power Student paired t test: no significant differences between the parameters of the power–duration relationship derived from the two different CP models were observed Subject Critical power estimates W’ estimates R2 1/Time model CP (W) Work-time model CP (W) 1/Time model W′ (kJ) Work-time model W′ (kJ) 1/Time model Work-time model 1 212 217 11.9 11.2 0.966 0.997 2 259 262 9.9 9.5 0.999 0.994 3 221 225 8.5 7.8 0.999 0.942 4 254 252 12.8 13.3 1.000 0.997 5 278 278 9.9 9.9 1.000 1.000 6 248 240 12.5 14.2 0.996 0.956 7 256 255 8.0 8.1 0.999 1.000 8 320 317 13.3 14.0 0.999 0.993 9 258 258 13.7 13.8 1.000 1.000 10 280 275 22.7 24.9 0.999 0.981 11 223 223 18.1 18.2 0.999 1.000 12 243 240 12.2 13.0 0.997 0.984 Mean 254 254 12.8 13.2 0.996 0.987 Standard deviation 30 28 4.1 4.7 0.010 0.019 t test 0.456 0.178 0.183 2402 European Journal of Applied Physiology (2020) 120:2397–2405 1 3 Discussion The results of the present study show that a new HIDIT protocol maintains the ̇VO2 above 90% of ̇VO2peak for a longer period compared with two classical HIIT protocols with short and long intervals. Nevertheless, the Tlim, [La], HR, RPE, and ̇VO2 were similar among the protocols. This is the first study that has demonstrated that it is possible to increase the time close to ̇VO2peak solely through decreasing the duration of the intervals and, therefore, avoiding reduc- ing the power/intensity as previously shown (De Aguiar et al. 2013; Lisbôa et al. 2015; Rønnestad and Hansen 2016). In HIDIT (and LIHIIT), the protocol begins with 3 min at high intensity, as opposed to just 30 s in SIHIIT, and this resulted in a greater ̇VO2, HR, [La], CR10 scale, and RQ after 3 min of exercise. This is consistent with the studies by Millet et al. (2003) and Turner et al. (2006), in which during long-interval HIIT, a faster metabolic stimulation occurred at the beginning of the cycling exercise. However, in our study, there were no differences at Tlim in any of the parameters mentioned above, suggesting that the participants reached their personal maximal performances, regardless of Table 3 Main results of the HIIT tests and selected physiological variable at 3rd minute and at the end of the tests All values are mean and standard deviation (SD) SIHIIT short-interval HIIT, HIDIT high-intensity decremental intervals training, LIHIIT long-interval HIIT, Tlim time to exhaustion, T > 90% ̇VO2peak time spent above 90% ̇VO2peak, % ̇VO2peak oxygen uptake in percentage relative to its peak, mean% ̇VO2peak mean % ̇VO2peak maintained during HIIT tests, %HRpeak heart rate in percentage relative to its peak, [La] blood lactate concentration, CR10-scale perceived exer- tion, RQ respiratory quotient Significance by one-way repeated-measure ANOVA. When P < 0.05, planned contrasts with Bonferroni correction a P < 0.05 in post hoc HIDIT vs SIHIIT b P < 0.05 in post hoc HIDIT vs LIHIIT SIHIIT HIDIT LIHIIT P Tlim (s) 714 ± 265 798 ± 185 664 ± 282 0.144 T > 90% ̇VO2peak (s) 183 ± 225 312 ± 207a,b 179 ± 145 0.029 T > 90% ̇VO2peak × Tlim−1 0.25 ± 0.29 0.39 ± 0.24 0.26 ± 0.21 0.070 Work > CP (KJ) 18.74 ± 8.95 22.01 ± 10.40 19.28 ± 11.06 0.136 Mean % ̇VO2peak 81.50 ± 6.61 84.16 ± 4.00b 79.58 ± 7.08 0.044 Values at 3rd minute  % ̇VO2peak 85.33 ± 7.11 90.75 ± 5.94a 89.58 ± 6.52 0.004  %HRpeak 89.00 ± 4.00 91.00 ± 3.91a 92.60 ± 3.60 0.003  [La] (mmol L−1) 5.69 ± 1.62 8.03 ± 2.69a 7.85 ± 3.01 0.007  CR10-scale 5.29 ± 1.57 6.67 ± 2.12a 6.52 ± 2.03 0.008  RQ 1.04 ± 0.06 1.10 ± 0.09a 1.11 ± 0.08 > 0.001 Tlim  % ̇VO2peak 99.75 ± 8.62 100.17 ± 5.27 99.83 ± 8.36 0.981  %HRpeak 97.80 ± 3.99 97.40 ± 2.99 97.50 ± 3.98 0.802  [La] (mmol L−1) 10.75 ± 2.04 10.71 ± 4.72 10.83 ± 3.58 0.991  CR10-scale 9.48 ± 0.70 9.25 ± 1.78 9.56 ± 1.08 0.701  RQ 0.97 ± 0.05 0.95 ± 0.05 1.00 ± 0.10 0.113 Fig. 2 Time above 90% of ̇VO2 peak in seconds. *Significance by one-way repeated-measures ANOVA and planned contrast with Bon- ferroni correction between HIDIT and SIHIIT and between HIDIT and LIHIIT were used post hoc comparison, P < 0.05 2403 European Journal of Applied Physiology (2020) 120:2397–2405 1 3 the protocol adopted. Indeed, ̇VO2 and HR were close to the peak values (100% and 97%, respectively), while Borg scale was near 10 and [La] was above 10 mmol L−1. It is worth noting that HIDIT led to longer T > 90% ̇VO2peak despite the same RPE at the end of the exercise. In other words, HIDIT has potentially better training benefits, despite the same perceived effort. On the other hand, even though Tlim in HIDIT (798 s) was longer than in LIHIIT (664 s) and, similar to SIHIIT, (714 s), the ANOVA did not show any significant difference (P = 0.144). Our results seem to contradict results from the previous studies (Millet et al. 2003; Turner et al. 2006; Rønnestad and Hansen 2016). Millet et al. (2003) showed that when comparing some matched work HIIT pro- tocols, those with shorter intervals elicited lower ̇VO2, HR, and RPE at the end of the exercise, suggesting that the dura- tion may be longer when shorter intervals are used. Simi- larly, Turner et al. (2006) compared four HIIT protocols with the same intensity (work and recovery) and work/recovery ratio, reporting that in HIIT with shorter intervals, the [La] was lower after 30 min of exercise compared with longer intervals. In particular, in the HIIT protocol with shorter intervals (work 10 s/recovery 20 s), the [La] reached steady state after 30 min of exercise, whereas the one with longer intervals (work 90 s/recovery 180 s), the subjects lasted less than 10 min before exhaustion. Surprisingly, there are a few studies in which the authors analyze the effects of interval duration at a fixed work/recov- ery ratio and a fixed intensity (Millet et al. 2003; Turner et al. 2006; Rønnestad and Hansen 2016). It is known that increasing work interval durations prolongs the time close to ̇VO2max (Rozenek et al. 2007; Wakefield and Glaister 2009). Conversely, longer recovery interval duration decreases the time close to ̇VO2max (Smilios et al. 2017). However, to our knowledge, the only study that measured the time close to ̇VO2max and Tlim in HIIT matching work rate and work/ recovery ratio and isolating the interval duration variable was performed by Rønnestad and Hansen (Rønnestad and Hansen 2016). They compared three cycling HIIT protocols in which the intensity of the work bouts was set at maximal aerobic power ( ̇VO2max power), the recovery at 50% of the ̇VO2max power, and the work/recovery ratio was 2/1. They concluded that HIIT with shorter interval durations (30 s) led to a longer Tlim (~ 1400 s), a longer Time > 90% ̇VO2peak (~ 680 s) and a higher ratio of Time > 90% ̇VO2peak·Tlim−1 (0.55) (Rønnestad and Hansen 2016). Tlim, Time > 90% ̇VO2peak, and their ratio were lower in our study. This discrepancy may be attributed to the different protocols used and to the higher fitness level of the participants ( ̇V O2peak = 66 mL kg−1 min−1 compared to 57 mL kg−1 min−1) (Rønnestad and Hansen 2016). Another possible explana- tion might be the relative intensity at which our protocol was set (on average ~ 83% of load peak). This relative inten- sity refers to the load peak attained during a ramp protocol, which is reported to be 10–15% higher than the load peak reached with a step modality (Revill et al. 2002; Bentley and McNaughton 2003; Zuniga et al. 2012). Therefore, it can be assumed that the relative power would have been above 90% of the load peak if the incremental test was performed using steps. Nevertheless, the incremental ramp test was used alone in the present study only to determine ̇VO2peak, while the intensity of HIIT was set exclusively considering CP, as described above. In an attempt to benefit from faster ̇VO2 kinetics at the beginning of exercise, we imposed long first intervals. Alter- nately, other authors proposed a fast start strategy (De Agu- iar et al. 2013; Lisbôa et al. 2015; Rønnestad et al. 2019). Fast start strategy HIIT protocol (starting from 125% of the intermittent critical power, ICP, and decreasing it until 105%) enhanced the time above 95% of ̇VO2max compared to other protocols with a constant work rate at 125% ICP and a constant work rate at 105% ICP (De Aguiar et al. 2013). Nevertheless, the protocol that used lower intensity (105% ICP) increased Tlim, and the protocol that adopted higher intensity bouts (125% ICP) showed a greater ratio of Tlim/ time above 95% of ̇VO2max−1. Lisbôa et al (2015) decreased the intensity within every single interval, but attained simi- lar results. In addition, the recent work of Rønnestad et al. (2019) confirmed that the fast start pacing strategy can be a good strategy to increase the average ̇VO2, but the time close to ̇VO2max was not longer compared to traditional HIIT. Therefore, the fast start strategy is a useful tool to improve time near/at ̇VO2max and could be successfully applied to HIIT, although it impairs Tlim in comparison with protocols with the same final exercise work rate and the ratio T > 90% ̇VO2peak − Tlim−1 in comparison with protocols with the same initial intensity (De Aguiar et al. 2013). Compared to fast start protocols, HIDIT has the advantage of quickly stimulating oxygen uptake at the beginning without affecting Tlim. Moreover, fast start strategy HIIT reduces the ratio T > 90% ̇VO2peak—Tlim−1, while HIDIT tends to increase it (not significantly). Therefore, the HIDIT protocol that this study proposed combines the advantages of different previ- ously studied protocols and can be used during training ses- sions that aim to accumulate time close to ̇VO2max. Nonetheless, it is interesting that several participants were able to drastically increase the T > 90%VO2peak in the HIDIT protocol, whereas others performed much worse. In addition, as discussed above, the ANOVA failed to find differences in Tlim between the three HIIT proto- cols, which could be due to the heterogeneity of the sub- jects, despite our efforts to minimize differences by set- ting up HIIT reliant on CP and W′. In fact, high intensity was set as the percentage of CP that allowed each subject to last for 5 min before exhaustion according to equation [2]. While the intensity of HIIT is often set relying on % ̇VO2max, relying exclusively on ̇VO2max does not take 2404 European Journal of Applied Physiology (2020) 120:2397–2405 1 3 into account the anaerobic characteristics of the subjects, which are very important in HIIT. For instance, whether two athletes present a similar ̇VO2max intensity but differ- ent W′ (and CP) when exercising with similar % ̇VO2max intensity during HIIT, the exercise will actually involve a different proportion of their W′, which results in a dif- ferent exercise tolerance (Blondel et al. 2001). Therefore, expressing intensity as a percentage of CP for high-inten- sity exercises allows individual differences in W′ to be taken into account and eased as much as possible. Indeed, W′ was not correlated with Tlim of any HIIT test, since it has been used to adjust the intensity with equation [2]. Furthermore, there was no correlation among age/HRpeak, the ̇VO2 kinetics during the first 3 min of HIDIT and LIHIIT (unpublished), and the other main outcomes. Additionally, there were no relationships between ̇VO2peak or CP and the main outcomes as well. The lack of relationship among age and other variables suggests that age did not influ- ence our main results. In fact, our data may even support the idea that HIDIT could be applied in well-trained male adults over a wide range of age. Another major physiologi- cal determinant that may explain the variability between subjects in Tlim during interval and continuous exercises is the differences between lactate threshold intensity and ̇VO2max intensity (Midgley et al. 2007). Midgley et al. suggested that athletes with larger differences will replete their anaerobic capacity to a greater extent during each relief interval, increasing the time to exhaustion. Similarly, the relationship between the CP-load peak difference and Tlim during HIIT has been verified in this study to deter- mine whether it can affect the Tlim of HIIT. As a result, only 59% of the variance in Tlim in SIHIIT was explained by the difference between CP and load peak in percentage, while in the other two protocols, there were no relation- ships. Therefore, future research that aims to investigate Tlim in HIIT may benefit by selecting subjects with homo- geneous difference %CP-load peak, although Tlim in HIIT with longer intervals does not seem to correlate with it. It is, therefore, tempting to suggest that individuals with a wide gap between the CP and the load peak could benefit more from short-interval HIIT to prolong Tlim. Further research is needed to verify whether T > 90% ̇V O2peak may be enhanced with HIDIT in different HIIT pro- tocols (i.e., at different intensities) and in different popu- lations. However, HIDIT might be useful in sport training when the aim is to maintain a high ̇VO2max and/or maintain a specific power or velocity as long as possible, such as in training for track cycling races. If the aim is to allow the athlete to finish the race at a given time, the most specific training is to ride at that velocity for that race time for a distance as near as possible to the distance of the race. After the recovery, repeat for a shorter distance and so on. Starting with short intervals would not be sufficiently specific, and continuing with the first interval distance would not be pos- sible for the fatigued athlete. Furthermore, HIDIT could be useful for patients or for wellness purposes, setting a lower percentage of ̇VO2max or other physiological parameters. For example, if an exercise is intended to avoid exceeding a given [La] cut-off, it can start with a longer interval to save time and then decrease the length of the interval to avoid exceeding the [La] cut-off. However, we suggest adopting this protocol in athletes and patients who aim to train and improve their ̇VO2max. Conclusions In conclusion, HIDIT applied to cycling exercise in well- trained amateur cyclists can enhance T > 90% ̇VO2peak with- out reducing Tlim, the ratio of T > 90% ̇VO2peak and Tlim−1, or the average ̇VO2. In fact, the average ̇VO2 was even higher in HIDIT than in LIHIIT. Finally, despite the higher stimu- lation of ̇VO2, the rate of perceived exertion and the other physiological parameters at the end of the exercise were not different compared with long- or short-interval HIIT, suggesting that HIDIT was not more demanding. In light of the favorable or similar physiological and/or perceptual responses to HIDIT compared to the other protocols and given the improved capability to prolong the time close to ̇VO2peak, it could be used as a preferable method to elicit similar or greater physiological adaptations. Acknowledgements Open access funding provided by Università degli Studi di Udine within the CRUI-CARE Agreement. We would like to thank the participants in the study for their time and effort to ensure the success of the project, in particular the “Pedale Gemonese” associa- tion (Gemona del Friuli, Udine, Italia). The study was supported by Fondazione Pietro Pittini (Italy). Author contributions All authors conceived and designed the research. FV and NG conducted experiments. FV analyzed the data. FV wrote the manuscript, NG and SL the manuscript. All authors read and approved the manuscript. Compliance with ethical standards Conflict of interest The authors report no conflict of interest. Open Access This article is licensed under a Creative Commons Attri- bution 4.0 International License, which permits use, sharing, adapta- tion, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/. 2405 European Journal of Applied Physiology (2020) 120:2397–2405 1 3 References Blondel N, Berthoin S, Billat V, Lensel G (2001) Relationship between run times to exhaustion at 90, 100, 120, and 140% of V̇O2max and velocity expressed relatively to critical velocity and maximal velocity. Int J Sports Med 22:27–33. https ://doi. org/10.1055/s-2001-11357 Borg E, Borg G, Larsson K et al (2010) An index for breathlessness and leg fatigue. Scand J Med Sci Sport 20:644–650 Buchheit M, Laursen PB (2013) High-intensity interval training, solu- tions to the programming puzzle: part I: cardiopulmonary empha- sis. Sport Med 43:313–338 Burnley M, Jones AM (2018) Power–duration relationship: physiology, fatigue, and the limits of human performance. Eur J Sport Sci 18:1–12. https ://doi.org/10.1080/17461 391.2016.12495 24 Chidnok W, DiMenna FJ, Fulford J et al (2013) Muscle metabolic responses during high-intensity intermittent exercise measured by 31P-MRS: relationship to the critical power concept. Am J Physiol Regul Integr Comp Physiol 305:R1085–R1092 De Aguiar RA, Turnes T, De Oliveira Cruz RS, Caputo F (2013) Fast- start strategy increases the time spent above 95 %VO2max during severe-intensity intermittent running exercise. Eur J Appl Physiol 113:941–949 di Prampero PE (2003) Factors limiting maximal performance in humans. Eur J Appl Physiol 90:420–429 di Prampero PE, Margaria R (1968) Relationship between O2 con- sumption, high energy phosphates and the kinetics of the O2 debt in exercise. Pflugers Arch 304:11–19 Ferguson C, Rossiter HB, Whipp BJ et al (2010) Effect of recovery duration from prior exhaustive exercise on the parameters of the power-duration relationship. J Appl Physiol 108:866–874 Hill DW (1993) The critical power concept. A review. Sport Med 16:237–254 Hill AV, Lupton H (1923) Muscular exercise, lactic acid, and the supply and utilization of oxygen. QJM 16:135–171. https ://doi. org/10.1093/qjmed /os-16.62.135 Jones AM, Vanhatalo A (2017) The “Critical Power” concept: appli- cations to sports performance with a focus on intermittent high- intensity exercise. Sport Med 47:65–78 Jones AM, Vanhatalo A, Burnley M et al (2010) Critical power: impli- cations for determination of V’O2max and exercise tolerance. Med Sci Sports Exerc 42:1876–1890 Lisbôa FD, Salvador AF, Raimundo JAG et al (2015) Decreasing power output increases aerobic controbution during low-volume severe- intensity intermittent exercise. J Strength Cond Res 29:2434–2440 Meyer RA (1988) A linear model of muscle respiration explains mono- exponential phosphocreatine changes. Am J Physiol 254:548–553 Midgley A, Mc Naughton L (2006) Time at or near VO2max dur- ing continuous and intermittent running. J Sports Med Phys Fit 46:1–14 Midgley AW, McNaughton LR, Carroll S (2007) Physiological deter- minants of time to exhaustion during intermittent treadmill run- ning at vV’O2max. Int J Sports Med 28:273–280. https ://doi. org/10.1055/s-2006-92433 6 Millet GP, Candau R, Fattori P et al (2003) Responses to different intermittent runs at velocity associated with. Can J Appl Physiol 28:410–423 Moritani T, Nagata A, Devries HA, Muro M (1981) Critical power as a measure of physical work capacity and anaerobic threshold. Ergonomics 24:339–350 Morton RH, Billat VL (2004) The critical power model for intermittent exercise. Eur J Appl Physiol 91:303–307 Poole DC, Hirai DM, Copp SW, Musch TI (2012) Muscle oxygen trans- port and utilization in heart failure: implications for exercise (in) tolerance. AJP Hear Circ Physiol 302:H1050–H1063 Rønnestad BR, Hansen J (2016) Optimizing interval training at power output associated with peak oxygen uptake in well-trained cyclists. J Strength Cond Res 30:999–1006 Rønnestad BR, Rømer T, Hansen J (2019) Increasing Oxygen uptake in well-trained cross-country skiers during work intervals with a fast start. Int J Sports Physiol Perform. https ://doi.org/10.1123/ ijspp .2018-0360 Rozenek R, Funato K, Kubo J et al (2007) Physiological responses to interval training sessions at velocities associated with V’O2max. J Strength Cond Res 21:188–192 Skiba PF, Chidnok W, Vanhatalo A, Jones AM (2012) Modeling the expenditure and reconstitution of work capacity above critical power. Med Sci Sports Exerc 44:1526–1532 Skiba PF, Clarke D, Vanhatalo A, Jones AM (2014) Validation of a novel intermittent W′ model for cycling using field data. Orig Investig Int J Sport Physiol Perform 9:900–904. https ://www. IJSPP -Journ al.com Skiba PF, Fulford J, Clarke DC et al (2015) Intramuscular determinants of the ability to recover work capacity above critical power. Eur J Appl Physiol 115:703–713 Smilios I, Myrkos A, Zafeiridis A et al (2017) The effects of recov- ery duration during high-intensity interval exercise on time spent at high rates of oxygen consumption, oxygen kinetics and blood lactate. J Strength Cond Res. https ://doi.org/10.1519/JSC.00000 00000 00190 4 Turner AP, Cathcart AJ, Parker ME et al (2006) Oxygen uptake and muscle desaturation kinetics during intermittent cycling. Med Sci Sports Exerc 38:492–503 Vinetti G, Fagoni N, Taboni A et al (2017) Effects of recovery interval duration on the parameters of the critical power model for incre- mental exercise. Eur J Appl Physiol 117:1859–1867 Wakefield BR, Glaister M (2009) Influence of work-interval intensity and duration on time spent at a high percentage of V’O2 max during intermittent supramaximal exercise. J Strength Cond Res 23:2548–2554 Wenger HA, Bell GJ (1986) The interactions of intensity, frequency and duration of. Sport Med 3:346–356 Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

High-intensity decreasing interval training (HIDIT) increases time above 90% [Formula: see text]O₂peak.

08-11-2020

Vaccari, Filippo,Giovanelli, N,Lazzer, S

eng

PMC5968933

sports Article A Description and Comparison of Cardiorespiratory Fitness Measures in Relation to Pitching Performance Among Professional Baseball Pitchers Javair S. Gillett 1, J. Jay Dawes 2, Frank J. Spaniol 3,*, Matthew R. Rhea 4, Joe P. Rogowski 5, Mitchel A. Magrini 6, Roberto Simao 7 and Derek J. Bunker 4 1 Athletic Performance, Houston Rockets, Houston, TX 77002, USA; jgillett@rocketball.com 2 University of Colorado, Colorado Springs, CO 80918, USA; jdawes@uccs.edu 3 Texas A&M University-Corpus Christi, Corpus Christi, TX 78412, USA 4 A.T. Still University, Kirksville, MO 63501, USA; mrhea@atsu.edu (M.R.R.); Bunkjames7@hotmail.com (D.J.B.) 5 Athletic Heart Research Institute, Orlando, FL, 32801, USA; jrogowski@athletic-heart.com 6 Oklahoma State University, Stillwater, OK 74074, USA; mmagrini@uccs.edu 7 Federal University of Rio de Janeiro, Rio de Janeiro 21941-901, Brazil; rsimaoj@terra.com.br * Correspondence: frank.spaniol@tamucc.edu; Tel.: +1-361-549-7533 Academic Editor: Eling Douwe de Bruin Received: 18 November 2015; Accepted: 15 February 2016; Published: 25 February 2016 Abstract: The purpose of this study is to provide descriptive and comparative information regarding the cardiorespiratory fitness of professional baseball pitchers. Twenty-four (n = 24) major league (ML) baseball pitchers (starters n = 14; relievers n = 10) over seven seasons (2007–2013) were evaluated. A modified Bruce protocol and the CardioCoach™ CO2 metabolic analyzer were used to estimate VO2 max and anaerobic threshold (AT) at the beginning of each season. Performance data from each season was utilized to draw inference about pitching performance. One-way Analysis of Variance (ANOVA) was used to compare Starting (S) and Relief (R) pitchers above/below the group mean for VO2 max and AT. Pearson product moment correlations were also used to examine relationships between cardiorespiratory fitness and performance. Significant differences in performance were discovered between S pitchers above/below the overall group mean for VO2 max. (p ď 0.05) and for AT in Walks plus Hits per Inning Pitched (WHIP) (p ď 0.05) and Earned Run Average (ERA) (p ď 0.05). Significant relationships between VO2 max and Walks per 9 Innings (BB/9) (p ď 0.05), Home Runs per 9 innings (HR/9) (p ď 0.05), Wins (W) (p ď 0.05), Fielding Independent Pitching (FIP) (p ď 0.01), Strikeouts (K) (p ď 0.01), Hits per 9 innings (H/9) (p ď 0.01), Strikeouts per 9 innings (K/9) (p ď 0.01), ERA (p ď 0.01), and WHIP (p ď 0.01). Low, but significant, correlations were discovered between AT and WHIP (p ď 0.05) and ERA (ď0.05). CONCLUSION: Higher aerobic capacity appears to be more influential for S than R pitchers. Strength and conditioning practitioners should ensure that pitchers, especially S pitchers at the ML level, perform sufficient and appropriate endurance training to support pitching performance. Keywords: conditioning; endurance; pitching; performance; VO2 max 1. Introduction Baseball is a sport that requires short, explosive bursts of intense effort. While the duration of each play is relatively short, a typical professional baseball game takes approximately three hours to complete [1]. During a game, only two of the nine players on the field are involved in every single Sports 2016, 4, 14; doi:10.3390/sports4010014 www.mdpi.com/journal/sports Sports 2016, 4, 14 2 of 8 play, the pitcher and catcher. While catchers perform many of their skills at submaximal intensities, pitchers are expected to deliver every pitch at maximum, or near maximum effort [2]. A starting pitcher typically delivers 80–100 pitches or more per game, whereas a relief pitcher is required to throw significantly fewer pitches per performance (40 or less). Whereas a starting pitcher is typically allowed four days between performances, a reliever might be required to throw on consecutive days. Pitching relies heavily on the ATP-PCr system during the delivery of the pitch, followed by brief bouts (approximately 20–30 s) of aerobic recovery between pitches [3]. Subsequently, pitching appears to place a relatively low demand on the aerobic energy system [4]. In fact, Potteiger et al. [4] found that mean oxygen consumption ranged between 14.8–20.6 mL¨ kg ´1 min ´1 for pitchers pitching in a simulated game. The researchers noted that this would correspond to a continuous exercise intensity at approximately 45% of the participants mean VO2 max. However, since this study was performed in a laboratory setting where the pitchers did not face game competition, one may speculate that the aerobic demands may differ during actual competition. Stockholm and Morris [5] conducted a study in which a freshman collegiate baseball pitcher’s heart rate (HR) was monitored and recorded during competition (3 h and 10 min, 9 inning game) via the use of telemetry. It was discovered that the mean heart rate during the performance was approximately 87% of the player’s age-predicted heart rate max (HR max), with peak HR reaching 95% of the player’s age-predicted HR max. This is significantly greater than HRs achieved during a laboratory study [4] and during bullpen practice sessions prior to an intra-squad game [6]. This suggests that arousal and anxiety levels may impact the physiological demands of pitching. Furthermore, being that the bulk of competitions in professional baseball occur during the summer months, greater cardiorespiratory fitness may help players better accommodate the physiological challenges and delay the onset of fatigue when playing in hot/humid environments [5–13]. Very few studies have investigated the relationship between aerobic fitness and pitching performance [4]. This may in part be due to the observations that baseball is predominantly an anaerobic sport. Ebben et al. [14] found that the majority of major league strength coaches do not test anaerobic capacity or aerobic endurance, which may be due to the belief that possessing a high level of aerobic fitness does not appear to limit performance amongst professional or collegiate pitchers. Furthermore, studies that have investigated aerobic fitness and pitching performance have quantified performance in terms of maintenance of ball velocity, rather than the player’s actual game statistics [4]. While one may assume that the maintenance of ball velocity is an inherent predictor of performance, the art of pitching is a multifaceted and complex skill and success should not be limited to a singular variable, particularly at higher playing levels. There are other performance indicators and statistics that are more indicative of individual pitching performance and less dependent on team performance, which should be taken into consideration when seeking to evaluate or predict pitching effectiveness. Currently, there is very little data available regarding the cardiorespiratory fitness of baseball players, especially at the professional level. Thus, the primary purpose of this investigation was to provide descriptive information regarding the cardiorespiratory fitness profiles of Major League (ML) pitchers, and compare cardiorespiratory fitness levels between starting (S) and relief (R) pitchers. A secondary purpose was to examine the relationship between cardiorespiratory fitness and selected measures of pitching performance. 2. Method, Results, Discussion 2.1. Methods 2.1.1. Experimental Approach to the Problem The data used for this study was archival and approved by an Institutional Review Board for research with human subjects prior to data analysis. In order to compare performance between pitchers by position and fitness level and identify the relationships between cardiorespiratory fitness and pitching performance, physiological and performance statistics over seven seasons (2007–2013) Sports 2016, 4, 14 3 of 8 for 24 ML baseball pitchers were gathered and analyzed. A correlational analysis was also performed to examine possible relationships between these variables. In addition, statistical analyses were conducted to examine the differences between S and R pitchers. 2.1.2. Subjects Data for 24 professional pitchers, within the selected ML baseball organization, was utilized in this investigation. Only pitchers remaining within the organization throughout the entire season were included in this analysis. Since some of the subjects in this study completed multiple seasons with the organization, a total of 40 eligible cases (n = 27 S; n = 13 R) were available for evaluation. For the purpose of analysis, each case (year) was treated separately as both physiological and performance statistics are subsequent to change each season and therefore were evaluated individually. Only pitchers with 50 or more innings pitched per season were used in this study. 2.1.3. Procedures Physiological testing was performed within the first week of spring training camp each year. All physiological tests were performed after on-field baseball practice on days in which the pitcher did not throw a bullpen session. Pitchers completed a graded, submaximal exercise test on a treadmill (Life Fitness TI, Model 97Ti/Model CLST and Woodway, Model R-DESMO, Franklin Park, IL, USA). The CardioCoach™ CO2 metabolic analyzer and software (version 3.04.72, KORR Medical Technologies Inc., Salt Lake City, UT, USA) was used for gas analysis. This device has been found to be a valid method of assessing VO2 at submaximal and maximal levels [15]. Subjects performed a modified Bruce protocol and were asked to complete as many stages as possible. The test was terminated when the subject requested to stop or reached volitional fatigue. Utilizing the software provided via CardioCoach™ a linear regression equation was used to predict VO2peak. CardioCoach™ software was also used to identify oxygen consumption at anaerobic threshold (AT). In addition to cardiorespiratory fitness, the following anthropometric and physiological data was collected for each of the subjects: height (cm), weight (kg), percent body fat (%BF), and lean body mass (LBM). Body fat was estimated using a six-site skinfold test [16]. Performance data were accessed and gathered immediately after the season via reliable online databases [17–19]. These databases are open access sources for pitching statistics. Multiple databases were used for the purpose of comparison to ensure accuracy. The key pitching performance statistics used in this analysis included: Fielding Independent Pitching (FIP), Walks plus Hits per Inning Pitched (WHIP), and Strikeout to Walk Ratio (K/BB). These statistics were chosen because they tend to be the least influenced by uncontrollable variables (i.e., team performance, defense skill, batter skill, situational hitting, stadium environment, etc.) and more dependent on the pitcher’s overall pitching performance. Keeping this in mind additional performance statistics analyzed in this study included: Earned Run Average (ERA), Hits per 9 innings (H/9), Homeruns per 9 innings (HR/9), Strikeouts per 9 innings (K/9), and Walks per 9 innings (BB/9), Wins (W), Win/Loss Percentage (W/L%) and Strikeouts (K). Data were entered into a spreadsheet matching performance data with the physiological measures for each player for each year. 2.1.4. Statistical Analyses A descriptive data analysis was conducted for all pitchers in the sample and by their individual positions (S or R) Comparison between fitness measures among S and R pitchers on selected fitness and performance measures was conducted via an independent samples T-test based on position. Additionally, all players were then separated into either a higher or lower cardiorespiratory fitness group within their position based on the average VO2peak and AT. A one-way Analysis of Variance (ANOVA) was then utilized to examine differences between these groups. Pearson product moment correlations were then used to examine the relationship between physiological measures and performance data as a whole as well as divided by position. Data were evaluated using SPSS Statistics Sports 2016, 4, 14 4 of 8 version 22 (SPSS, Armonk, NY, USA). Statistical significance was set at p < 0.05. Data are presented as means and standard deviations. Correlations were considered high (0.80–1.00), moderately high (0.60–0.79), moderate (0.40–0.59) or low (0.20–0.39) [18]. 2.2. Results Descriptive data for the entire sample is presented in Table 1. For all pitchers, average VO2peak was 48.13 ˘ 5.30 (mL/kg/Min). For all pitchers VO2peak was found to have a low-moderate relationship (p < 0.05) between WHIP (r = ´0.484, p ď 0.01), K (r = 0.572, p ď 0.01) BB/9 (r = ´0.358, p ď 0.05) and W (r = 0.550, p ď 0.01). A strong relationship (r = 0.602, p ď 0.01) was found between VO2peak and K/BB. Additionally, low, but significant correlations were also found between AT and K (r = ´0.328, p ď 0.05), K/BB (r = ´327, p ď 0.05) and W (r = 0.358, p ď 0.05). Table 1. Descriptive data. Anthropometric and Fitness Variables All Pitchers (n = 40) Mean ˘ SD S Pitchers (n = 27) Mean ˘ SD R Pitchers (n = 13) Mean ˘ SD Age (YEARS) 28.03 ˘ 5.17 27.33 ˘ 5.46 29.46 ˘ 4.31 Weight (KG) 100.06 ˘ 6.80 99.27 ˘ 5.75 101.75 ˘ 8.59 Height (CM) 191.19 ˘ 5.11 192.28 ˘ 5.00 188.92 ˘ 4.80 Estimated Percent Body fat (%BF) 15.28 ˘ 3.45 14.21 ˘ 2.45 17.5 ˘ 4.2 Lean BODY mass (lBM) (KG) 84.63 ˘ 4.54 85.07 ˘ 3.85 83.72 ˘ 5.79 Estimated Max. Heart Rate (HR) 191.98 ˘ 5.16 192.66 ˘ 5.46 190.53 ˘ 4.31 VO2 max (mL¨ kg´1¨ min.) 48.13 ˘ 5.30 49.49 ˘ 4.59 45.28 ˘ 5.71 Anaerobic threshold (AT) 37.31 ˘ 7.87 38.63 ˘ 7.03 34.57 ˘ 9.06 HR at VO2 max 184.28 ˘ 9.70 182.33 ˘ 9.06 188.31 ˘ 10.09 HR at Anaerobic Threshold 162.72 ˘ 13.83 161.11 ˘ 11.97 166.08 ˘ 17.13 When comparing S vs. R pitchers there was a significant effect for position, t = 2.50, p ď 0.01, with S pitchers demonstrating higher VO2 max values (49.49 ˘ 4.59) when compared to relievers (45.28 ˘ 5.72) as presented in Figure 1. However, our analysis revealed no significant differences in AT between S and R. Sports 2016, 4, 14 4 of 8 as means and standard deviations. Correlations were considered high (0.80–1.00), moderately high (0.60–0.79), moderate (0.40–0.59) or low (0.20–0.39) [18]. 2.2. Results Descriptive data for the entire sample is presented in Table 1. For all pitchers, average VO2peak was 48.13 ± 5.30 (mL/kg/Min). For all pitchers VO2peak was found to have a low-moderate relationship (p < 0.05) between WHIP (r = −0.484, p ≤ 0.01), K (r = 0.572, p ≤ 0.01) BB/9 (r = −0.358, p ≤ 0.05) and W (r = 0.550, p ≤ 0.01). A strong relationship (r = 0.602, p ≤ 0.01) was found between VO2peak and K/BB. Additionally, low, but significant correlations were also found between AT and K (r = −0.328, p ≤ 0.05), K/BB (r = −327, p ≤ 0.05) and W (r = 0.358, p ≤ 0.05). Table 1. Descriptive data. Anthropometric and Fitness Variables All Pitchers (n = 40) Mean ± SD S Pitchers (n = 27) Mean ± SD R Pitchers (n = 13) Mean ± SD Age (YEARS) 28.03 ± 5.17 27.33 ± 5.46 29.46 ± 4.31 Weight (KG) 100.06 ± 6.80 99.27 ± 5.75 101.75 ± 8.59 Height (CM) 191.19 ± 5.11 192.28 ± 5.00 188.92 ± 4.80 Estimated Percent Body fat (%BF) 15.28 ± 3.45 14.21 ± 2.45 17.5 ± 4.2 Lean BODY mass (lBM) (KG) 84.63 ± 4.54 85.07 ± 3.85 83.72 ± 5.79 Estimated Max. Heart Rate (HR) 191.98 ± 5.16 192.66 ± 5.46 190.53 ± 4.31 VO2 max (mL·kg−1·min.) 48.13 ± 5.30 49.49 ± 4.59 45.28 ± 5.71 Anaerobic threshold (AT) 37.31 ± 7.87 38.63 ± 7.03 34.57 ± 9.06 HR at VO2 max 184.28 ± 9.70 182.33 ± 9.06 188.31 ± 10.09 HR at Anaerobic Threshold 162.72 ± 13.83 161.11 ± 11.97 166.08 ± 17.13 When comparing S vs. R pitchers there was a significant effect for position, t = 2.50, p ≤ 0.01, with S pitchers demonstrating higher VO2 max values (49.49 ± 4.59) when compared to relievers (45.28 ± 5.72) as presented in Figure 1. However, our analysis revealed no significant differences in AT between S and R. Figure 1. Differences in VO2 max between Starters and Relievers. A one-way ANOVA with pairwise comparisons revealed significantly better performance statistics among S pitchers with a VO2peak above the overall group mean in FIP (F(3,36) = 3.87, p ≤ 0.01, P), WHIP (F(3,36) = −4.60, p ≤ 0.01 ), K/BB (F(3,36) = 7.25, p ≤ 0.01), and ERA (F(3,36) = −4.58 p ≤ 0.01 ). Additionally, it was discovered that S pitchers with an AT above the overall group mean demonstrated significantly better performance as measured by WHIP (F(3,36) = 2.54 ,p ≤ 0.05), presented in Figure 2, and ERA (F(3,36) = 2.52, p ≤ 0.05), presented in Figure 3. When comparing R pitchers above and below VO2peak the only significant difference discovered in measures of pitching performance was in HR/9 (F(3,36) = 5.06, p ≤ 0.05). 49.49 45.28 43 44 45 46 47 48 49 50 VO2 max (mL·kg−1·min.) Starters Releivers Figure 1. Differences in VO2 max between Starters and Relievers. A one-way ANOVA with pairwise comparisons revealed significantly better performance statistics among S pitchers with a VO2peak above the overall group mean in FIP (F(3,36) = 3.87, p ď 0.01, P), WHIP (F(3,36) = ´4.60, p ď 0.01), K/BB (F(3,36) = 7.25, p ď 0.01), and ERA (F(3,36) = ´4.58 p ď 0.01). Additionally, it was discovered that S pitchers with an AT above the overall group mean demonstrated significantly better performance as measured by WHIP (F(3,36) = 2.54, p ď 0.05), presented in Figure 2, and ERA (F(3,36) = 2.52, p ď 0.05), presented in Figure 3. When comparing R pitchers above and below Sports 2016, 4, 14 5 of 8 VO2peak the only significant difference discovered in measures of pitching performance was in HR/9 (F(3,36) = 5.06, p ď 0.05). Sports 2016, 4, 14 5 of 8 Figure 2. Differences in walks and hits per inning pitched (WHIP) amongst S pitchers with higher and lower VO2 max. Figure 3. Differences in earned run average (ERA) amongst S pitchers with higher and lower VO2 max. Among starting pitchers, Pearson product-moment correlations revealed low- high relationships between VO2 max and BB/9 (r = −0.416), HR/9 (r = −0.478, p ≤ 0.05), W (r = 0.548), FIP (r = −0.567, p ≤ 0.01), K’s (r = 0.572, p ≤ 0.01), H/9 (r = −0.589, p ≤ 0.01), K/9 (r = 0.614, p ≤ 0.01), ERA (r = −678, p ≤ 0.01), and WHIP (r = −0.685, p ≤ 0.01). Low correlations were also found between AT and WHIP (r = −431, p ≤ 0.05) and ERA (r = −431, p ≤ 0.05). When separately analyzing R pitchers the only significant relationships (r = 0.592, p ≤ 0.05) was found between FIP and VO2 max. This relationship indicated that R pitchers with a higher VO2 max had a higher FIP. No other significant relationships between VO2 max or AT and performance were discovered among R pitchers. 2.3. Discussion The results of this study indicate that ML S pitchers are more aerobically fit than R pitchers and that several pitching performance variables appear to be related to greater VO2 max. For S pitchers only, there was a strong, significant correlation between VO2 max and FIP, WHIP, and ERA. There was also a moderate but significant relationship between AT and both WHIP and ERA for S pitchers only. For R pitchers no significant relationships were discovered between either VO2 max and AT for any of the selected measures of used in our analysis. In contrast, the sole correlation among R pitchers in this study was found when this group was subcategorized into high/low VO2 max. It was discovered that R pitchers with a higher VO2 max actually pitched worse than those with a lower VO2 max based on FIP. After analysis of R pitchers data it can be concluded that cardiorespiratory fitness may not Figure 2. Differences in walks and hits per inning pitched (WHIP) amongst S pitchers with higher and lower VO2 max. Sports 2016, 4, 14 5 of 8 Figure 2. Differences in walks and hits per inning pitched (WHIP) amongst S pitchers with higher and lower VO2 max. Figure 3. Differences in earned run average (ERA) amongst S pitchers with higher and lower VO2 max. Among starting pitchers, Pearson product-moment correlations revealed low- high relationships between VO2 max and BB/9 (r = −0.416), HR/9 (r = −0.478, p ≤ 0.05), W (r = 0.548), FIP (r = −0.567, p ≤ 0.01), K’s (r = 0.572, p ≤ 0.01), H/9 (r = −0.589, p ≤ 0.01), K/9 (r = 0.614, p ≤ 0.01), ERA (r = −678, p ≤ 0.01), and WHIP (r = −0.685, p ≤ 0.01). Low correlations were also found between AT and WHIP (r = −431, p ≤ 0.05) and ERA (r = −431, p ≤ 0.05). When separately analyzing R pitchers the only significant relationships (r = 0.592, p ≤ 0.05) was found between FIP and VO2 max. This relationship indicated that R pitchers with a higher VO2 max had a higher FIP. No other significant relationships between VO2 max or AT and performance were discovered among R pitchers. 2.3. Discussion The results of this study indicate that ML S pitchers are more aerobically fit than R pitchers and that several pitching performance variables appear to be related to greater VO2 max. For S pitchers only, there was a strong, significant correlation between VO2 max and FIP, WHIP, and ERA. There was also a moderate but significant relationship between AT and both WHIP and ERA for S pitchers only. For R pitchers no significant relationships were discovered between either VO2 max and AT for any of the selected measures of used in our analysis. In contrast, the sole correlation among R pitchers in this study was found when this group was subcategorized into high/low VO2 max. It was discovered that R pitchers with a higher VO2 max actually pitched worse than those with a lower VO2 max based on FIP. After analysis of R pitchers data it can be concluded that cardiorespiratory fitness may not Figure 3. Differences in earned run average (ERA) amongst S pitchers with higher and lower VO2 max. Among starting pitchers, Pearson product-moment correlations revealed low- high relationships between VO2 max and BB/9 (r = ´0.416), HR/9 (r = ´0.478, p ď 0.05), W (r = 0.548), FIP (r = ´0.567, p ď 0.01), K’s (r = 0.572, p ď 0.01), H/9 (r = ´0.589, p ď 0.01), K/9 (r = 0.614, p ď 0.01), ERA (r = ´678, p ď 0.01), and WHIP (r = ´0.685, p ď 0.01). Low correlations were also found between AT and WHIP (r = ´431, p ď 0.05) and ERA (r = ´431, p ď 0.05). When separately analyzing R pitchers the only significant relationships (r = 0.592, p ď 0.05) was found between FIP and VO2 max. This relationship indicated that R pitchers with a higher VO2 max had a higher FIP. No other significant relationships between VO2 max or AT and performance were discovered among R pitchers. 2.3. Discussion The results of this study indicate that ML S pitchers are more aerobically fit than R pitchers and that several pitching performance variables appear to be related to greater VO2 max. For S pitchers only, there was a strong, significant correlation between VO2 max and FIP, WHIP, and ERA. There was also a moderate but significant relationship between AT and both WHIP and ERA for S pitchers only. For R pitchers no significant relationships were discovered between either VO2 max and AT for any of the selected measures of used in our analysis. In contrast, the sole correlation among R pitchers in this study was found when this group was subcategorized into high/low VO2 max. It was discovered that R pitchers with a higher VO2 max actually pitched worse than those with a lower VO2 max based on FIP. After analysis of R pitchers data it can be concluded that cardiorespiratory fitness may not necessarily have a positive impact on successful pitching performance among R pitchers. While the Sports 2016, 4, 14 6 of 8 results may simply be a product of on-field demands, or due to different training and conditioning methods performed between S and R pitchers, it may also suggest that aerobic fitness has a larger impact on pitching performance for the S pitcher at the ML level. In an attempt to identify key fitness benchmarks related to performance, further evaluations were performed on S pitchers only. The S sample was divided into higher and lower VO2 max groups (greater than or less than 48.13 (mL/kg/min)) to examine differences in performance in FIP, WHIP, and ERA. The higher fitness group (n = 15, VO2 = 50.19 ˘ 4.23) had an average WHIP of 1.25 ˘ 0.18, FIP of 3.61 ˘ 0.66, and average ERA of 3.77 ˘ 1.02. The WHIP and ERA results were significantly better (p < 0.05) than the lower fitness group (n = 12, VO2 = 48.84 ˘ 4.97) which were found to have an average WHIP of 1.48 ˘ 0.17 and ERA of 5.04 ˘ 0.97. The higher fitness group also had a better K/9 (p < 0.05). In summary, S pitchers exhibiting above average VO2 max, pitched considerably better than S pitchers with below average VO2 max. The physiological demands on S pitchers are greater than R pitchers due to the higher workload and duration of time they are expected to perform. This may result in a greater reliance on overall cardiorespiratory fitness and endurance to sustain performance throughout the duration of a game. The apparent lack of importance related to both VO2 max and AT for the R pitcher is not surprising and provides additional support for constructing position or role-specific exercise programs. It makes sense that the relationship between VO2 and AT and pitching performance measures may simply indicate the need for higher levels of endurance for S pitchers and not specific requirements for successful pitching performances for R pitchers. The current findings support the need for personalized in-season conditioning programs dependent on the specific role a pitcher holds on their team. The S pitcher is usually on a 5 day rotation involving four days in between each pitching outing. When the major emphasis of conditioning programs is placed on minimizing reductions in power and strength over the course of the season, the use of slow, long distance runs could be counter-productive. The authors postulate that one high intensity interval training session between each pitching outing may be sufficient to help maintain aerobic fitness and maximize recovery. This approach may also serve to minimize reductions in power and strength and interference during power development and strength training sessions performed during the four days between pitching outings. Conditioning programs for R pitchers should not mimic S pitchers, even though some R pitchers are expected to pitch multiple innings and sustain higher pitch counts. Practitioners must determine whether or not this type of R pitcher should fall under a modified conditioning program similar to a S pitcher. In this case, the R pitcher would be allowed more days to rest following an outing making it more conducive to higher intensity conditioning sessions. The day following a long outing where a R pitcher has the day off might be best to incorporate more intense conditioning sessions. It should also be noted that more extensive modifications in conditioning programs might be required to meet anthropometric needs/goals. In summary, this is the first known study that examines aerobic capacity among ML pitchers and its potential impact on pitching performance. When interpreting the current findings it is important to realize that there are many uncontrollable variables that become more evident in field experiments. First, successful pitching performance at the professional level is certainly not identified by one single performance measure. There are many variables that lead to a positive pitching performance, making it difficult to come to precise conclusions on how much of an impact physical fitness really has on performance statistics in this analysis. In addition, pitching mechanics play an important role in successful pitching at any level. Therefore, the evaluation of the statistical findings should consider the complexity of pitching performance in general. Furthermore, it is important to note that subjects exercised to voluntary exhaustion. Subsequently, it is possible that a true VO2 max may not have been reached due to lack of motivation. All of these factors should be considered when evaluating the outcomes of this study. Sports 2016, 4, 14 7 of 8 3. Conclusions Practical Applications The current data suggests that metabolic training among ML baseball starting pitchers should focus on improving both aerobic capacity and anaerobic threshold. At the higher levels of play adequate training time should be devoted to maximizing these physiological parameters to improve chances of successful performance. Concurrent training strategies should focus on maximizing both physiological variables while attempting to minimize any potential interference in maximizing power. High intensity interval training seems to be the most efficient mode of training to improve a starting pitcher’s aerobic capacity. Neglecting important physiological components can have a detrimental effect on performance but athletes need evidence-based guidance to ensure productivity and a return on their training efforts. Strength and conditioning professionals should work closely with pitchers, coaches, and organizational management to design and implement appropriate training strategies for pitchers at various levels. The reliance on research such as the current analysis ensures that training programs are evidence-based and effective. For specific examples of proposed conditioning programs for pitchers, further examination of published works are suggested [7,8,20,21]. Author Contributions: Javair S. Gillett conceived the experiment, oversaw the testing procedures, analyzed the data, and contributed in the writing and review of the manuscript; J. Jay Dawes analyzed the data and contributed in the writing and review of the manuscript; Frank J. Spaniol contributed in the writing and review of the manuscript and served as the corresponding author; Mitchel A. Magrini contributed in the writing and review of the manuscript; Matthew R. Rhea analyzed the data and contributed in the writing and review of the manuscript; Joe P. Rogowski oversaw the testing procedures and contributed in the review of the manuscript; Robert Simao contributed in the writing and review of the manuscript; and Derek J. Bunker contributed in the writing and review of the manuscript. Conflicts of Interest: The authors declare no conflict of interest. References 1. Baseball Prospectus. Available online: http://www.baseballprospectus.com/sortable/index.php?cid=1667158 (accessed on 6 May 2015). 2. Bradbury, J.C.; Forman, S.L. The impact of pitch counts and days of rest on performance among major-league baseball pitchers. J. Strength Cond. Res. 2012, 26, 1181–1187. [CrossRef] [PubMed] 3. Szymanski, D.J.; Fredrick, G.A. College baseball/softball periodized torso program. Strength Cond. J. 1999, 22, 42–47. [CrossRef] 4. Potteiger, J.; Blessing, D.; Wilson, G.D. The physiological responses to a single game of baseball pitching. J. Strength Cond. Res. 1992, 6, 11–18. 5. Stockholm, A.; Morris, H. A pitcher’s heart rate during actual competition. Res. Q. Am. Assoc. Health Phys. Educ. Recreat. 1969, 40, 645–649. 6. Szymanski, D.J.; Myres, R.L. Heart rate responses of collegiate baseball pitchers while pitching and conditioning. J. Strength Cond. Res. 2007, 21, e28. 7. Syzmanski, D. Collegiate baseball in-season training. Strength Cond. J. 2007, 29, 68–80. [CrossRef] 8. Szymanski, D. Physiology of baseball pitching dictates specific exercise intensity for conditioning. Strength Cond. J. 2009, 31, 41–47. [CrossRef] 9. Tomlin, D.L.; Wenger, H. The relationship between aerobic fitness and recovery from high intensity intermittent exercise. Sports Med. 2001, 31, 1–11. [CrossRef] [PubMed] 10. Tripp, B.; Boswell, L.; Gansneder, B.; Shultz, S. Functional fatigue decreases 3-dimensional multi-joint position reproduction acuity in the overhead-throwing athlete. J. Athl. Train. 2004, 39, 316–320. [PubMed] 11. Tripp, B.L.; Yochem, E.M.; Uhl, T.L. Functional fatigue and upper extremity sensorimotor system acuity in baseball athletes. J. Athl. Train. 2007, 42, 90–98. [PubMed] 12. Van Wessel, T.; de Haan, A.; Van der Laarse, W.; Jaspers, R. The muscle fiber type–fiber size paradox: Hypertrophy or oxidative metabolism? Eur. J. Appl. Physiol. 2010, 110, 665–694. [CrossRef] [PubMed] Sports 2016, 4, 14 8 of 8 13. Yang, S.W. Assessment of professional baseball players aerobic exercise performance depending on their positions. J. Strength Cond. Res. 2014, 28, 3289–3292. [CrossRef] [PubMed] 14. Ebben, W.; Hintz, M.; Simenz, C. Strength and conditioning practices of major league baseball strength and conditioning coaches. J. Strength Cond. Res. 2005, 19, 538–546. [PubMed] 15. Dieli-Conwright, C.M.; Jensky, N.E.; Battaglia, G.; McCauley, S.; Schroeder, E.T. Validation of the CardioCoachCO2 for submaximal and maximal metabolic exercise testing. J. Strength Cond. Res. 2009, 23, 1316–1320. [CrossRef] [PubMed] 16. Yuhasz, M.S. Physical Fitness Manual; University of Western Ontario: London, Canada, 1974. 17. Sports Reference, LLC, 2013. Available online: www.baseball-reference.com (accessed on 6 May 2015). 18. ESPN MLB. Available online: http://espn.go.com/mlb/ (accessed on 6 May 2015). 19. FanGraphs. Available online: www.fangraphs.com or http://www.espn.go.com/mlb/stats/pitching/_/ league/nl/sort/pitches/type/expanded-2 (accessed on 6 May 2015). 20. Rhea, M.; Bunker, D. Baseball-specific conditioning. Int. J. Sports Physiol. Perform. 2009, 4, 402–407. [PubMed] 21. Rhea, M.; Oliverson, J.; Kenn, J.; Naclerio, F. Non-compatibility of power and endurance training among college baseball players. J Strength Cond. Res. 2008, 22, 230–234. [CrossRef] [PubMed] © 2016 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons by Attribution (CC-BY) license (http://creativecommons.org/licenses/by/4.0/).

A Description and Comparison of Cardiorespiratory Fitness Measures in Relation to Pitching Performance Among Professional Baseball Pitchers.

02-25-2016

Gillett, Javair S,Dawes, J Jay,Spaniol, Frank J,Rhea, Matthew R,Rogowski, Joe P,Magrini, Mitchel A,Simao, Roberto,Bunker, Derek J

eng

PMC10471793

Physiological Reports. 2023;11:e15806. | 1 of 11 https://doi.org/10.14814/phy2.15806 wileyonlinelibrary.com/journal/phy2 1 | INTRODUCTION Human activity is drastically constrained at extreme alti- tudes due to low environmental oxygen availability. Few humans can reach the highest elevation on earth, the summit of Mt. Everest (~8850 m), without supplemental oxygen, and those who do are limited to such a degree that a slow uphill walk approaches the maximum capacity for oxygen uptake and utilization (V̇O2max). Over millennia, several animal species have adapted to life at extreme altitudes, in part due to a high hemoglobin- oxygen affinity (Storz, 2007; Storz et al., 2010). The most common metric of hemoglobin- oxygen affinity is P50, the oxygen tension at which 50% of hemoglobin is saturated. Received: 10 August 2023 | Accepted: 11 August 2023 DOI: 10.14814/phy2.15806 O R I G I N A L A R T I C L E The dependence of maximum oxygen uptake and utilization (V̇O2max) on hemoglobin- oxygen affinity and altitude Kevin L. Webb1,2 | Michael J. Joyner1 | Chad C. Wiggins1 | Timothy W. Secomb3 | Tuhin K. Roy1,2 This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. © 2023 The Authors. Physiological Reports published by Wiley Periodicals LLC on behalf of The Physiological Society and the American Physiological Society. 1Department of Anesthesiology and Perioperative Medicine, Mayo Clinic, Rochester, Minnesota, USA 2Department of Physiology and Biomedical Engineering, Mayo Clinic, Rochester, Minnesota, USA 3Department of Physiology, University of Arizona, Tucson, Arizona, USA Correspondence Kevin L. Webb, Department of Anesthesiology and Perioperative Medicine, Mayo Clinic, Rochester, MN, USA. Email: webb.kevin@mayo.edu Funding information National Institutes of Health, Grant/ Award Number: R35 HL139854 Abstract Oxygen transport from the lungs to peripheral tissue is dependent on the affin- ity of hemoglobin for oxygen. Recent experimental data have suggested that the maximum human capacity for oxygen uptake and utilization (V̇O2max) at sea level and altitude (~3000 m) is sensitive to alterations in hemoglobin- oxygen af- finity. However, the effect of such alterations on V̇O2max at extreme altitudes remains largely unknown due to the rarity of mutations affecting hemoglobin- oxygen affinity. This work uses a mathematical model that couples pulmonary oxygen uptake with systemic oxygen utilization under conditions of high meta- bolic demand to investigate the effect of hemoglobin- oxygen affinity on V̇O2max as a function of altitude. The model includes the effects of both diffusive and con- vective limitations on oxygen transport. Pulmonary oxygen uptake is calculated using a spatially- distributed model that accounts for the effects of hematocrit and hemoglobin- oxygen affinity. Systemic oxygen utilization is calculated assuming Michaelis– Menten kinetics. The pulmonary and systemic model components are solved iteratively to compute predicted arterial and venous oxygen levels. Values of V̇O2max are predicted for several values of hemoglobin- oxygen affinity and hemoglobin concentration based on data from humans with hemoglobin muta- tions. The model predicts that increased hemoglobin- oxygen affinity leads to in- creased V̇O2max at altitudes above ~4500 m. 2 of 11 | WEBB et al. A low P50 corresponds to a high hemoglobin- oxygen binding affinity and vice- versa. Recent investigation has highlighted a rare human population with hemoglo- bin mutations causing high hemoglobin- oxygen affinity (low P50) (Charache et al., 1966; Thom et al., 2013). This human population has shown remarkable maintenance of exercise tolerance during normobaric hypoxia and at terrestrial high altitude (~3000 m) (Dominelli et al., 2020; Hebbel et al., 1978; Webb, Dominelli, et al., 2022a), per- haps in part due to the elevated hemoglobin values ob- served. Furthermore, there has been growing interest regarding the effects of pharmacologically altering the P50 in healthy individuals with typical hemoglobin (Stewart et al.,  2021, 2020). Despite hemoglobin function being central to oxygen transport and utilization, much remains unknown regarding the effects of an altered P50. Changes in P50 can markedly influence pulmonary ox- ygen loading and peripheral offloading. For instance, a de- crease in P50 may enhance pulmonary oxygen loading at the expense of blunted peripheral offloading. At sea level, where arterial blood is well- oxygenated, a low P50 likely hinders peripheral offloading leading to several compen- satory adaptations (i.e., enhanced red blood cell produc- tion and increased oxygen carrying capacity), as observed in humans with hemoglobin mutations (Webb, Domi- nelli, et al., 2022a). Yet at high and extreme altitudes, a low P50 likely improves arterial blood oxygenation with subsequent preservation of V̇O2max. However, experi- mental data regarding the influence of an altered P50 are largely limited to examination at high altitude conditions (~3000 m or 15% O2 in the laboratory setting). The ques- tion remains: how would humans with a low P50 tolerate extreme altitudes (Bencowitz et al., 1982)? Because hemoglobin mutations are exceedingly rare and research sojourns to extreme altitudes are challenging to achieve, we present a theoretical model of oxygen trans- port to investigate the dependence of V̇O2max on P50 and altitude. We hypothesized that relative to a normal P50, a low P50 would result in a greater V̇O2max at high altitude and that this relationship would be further potentiated at extreme altitudes. 2 | METHODS The V̇O2max achieved in any given situation reflects limi- tations in both pulmonary oxygen uptake and systemic oxygen utilization. We therefore simulated whole- body oxygen uptake and utilization to investigate the varia- tion in V̇O2max as a function of altitude in humans with normal and altered hemoglobin- oxygen affinity. A math- ematical model for oxygen uptake in the lung was com- bined with a model for oxygen utilization in the systemic circulation, as indicated in Figure 1, to predict arterial and venous oxygen tensions and oxygen consumption rate. To represent conditions of maximal exercise, the tis- sue oxygen demand and cardiac output were selected in accordance with observed sea- level V̇O2max values (see “Systemic oxygen utilization model” below). Arterial and venous oxygen tensions were then predicted at pro- gressively lower atmospheric pressures (i.e., increasing altitude). The resulting arteriovenous oxygen content dif- ferences were used to estimate oxygen consumption as a function of altitude and as a measure of functional capacity. To investigate the effects of hemoglobin- oxygen affinity on V̇O2max, calculations were performed for three cases of P50 (low, normal, and high) with a range of potential changes in hemoglobin concentration as a function of altitude (Table 1). Further details of the mathematical model are given below, and parameter values are provided in Table 2. 2.1 | Pulmonary oxygen uptake model The model for pulmonary oxygen uptake yields estimates of arterial oxygen tension (Pa ) for given values of alveo- lar oxygen tension (PA ) and venous oxygen tension (Pv ). Oxygen uptake is calculated using a single- compartment model that accounts for the effects of capillary diameter and hematocrit based on simplified capillary and erythro- cyte geometry (Roy & Secomb, 2019). The assumed value FIGURE 1 Schematic of modeling configuration, combining pulmonary oxygen uptake and systemic oxygen utilization. The dashed arrows indicate diffusive oxygen transport, and the solid arrows indicate convective oxygen transport within the blood. PA, alveolar oxygen tension; Pa, arterial oxygen tension; Pv, venous oxygen tension.  | 3 of 11 WEBB et al. for the lung diffusing capacity DLO2 (Table 2) was selected to account for the heterogeneity of perfusion present in the pulmonary circulation (Roy & Secomb, 2014). The dif- fusion of oxygen from alveoli into the blood during pul- monary capillary perfusion is represented by: where Pb is blood oxygen tension, C0 is oxygen content of fully saturated blood (calculated as 1.34 × [Hb]), Q is mean capillary flow rate, x is distance, S(Pb ) describes the oxy- hemoglobin dissociation curve, and Ltot is the total length of pulmonary capillaries. Setting t = x ∕L, where L is mean capillary length and Qtot is the total blood flow (i.e., cardiac output), yields: This equation was integrated from t = 0 to t = 1 with the initial condition Pb(0) = Pv, to obtain Pa = Pb(1). The cardiac output Qtot (L/min) was estimated based on a published correlation for measurements made over a range of ̇V O2 values (L/min) by averaging the results from two different experimental techniques (Calbet & Boushel, 2015): To facilitate the calculation of PA via the alveolar gas equation, data for subjects at altitude were obtained from measurements performed on climbers summiting Mt. Ev- erest, with an altitude of ~8850 m (Grocott et al., 2009). Values of hemoglobin concentration, PO2, and PCO2 were obtained by digitizing Figure 2 of Grocott et al. (Grocott et al., 2009) and a linear regression was used to fit these data as a function of altitude. The value of P50 was assumed to be 26.3 mmHg at sea level (see Table 1). To obtain an es- timate of P50 at the summit, values of PO2 and saturation as reported in table 2 of Grocott et al. (2009) were fit using a nonlinear regression and an assumed value of 2.7 for the Hill coefficient n: Values of barometric pressure were calculated from West et al. (1999). where a is the altitude in kilometers. The simplified form of the alveolar gas equation was then used to calculate alveolar PO2 as a function of altitude: where Pw represents water vapor pressure and R represents the respiratory quotient (see Table 2). The values of PCO2 used in Equation 8, from Ref. (Grocott et al., 2009), were assumed to approximate the values under conditions of V̇O2max. 2.2 | Systemic oxygen utilization model The systemic model yields estimates of Pv for given values of Pa and tissue oxygen demand. According to the model, (1) QC0 dS(Pb ) dx = DLO2 Ltot (PA − Pb ) (2) dPb dt = DLO2 (PA − Pb ) QtotC0S(Pb ) (3) Qtot = 4.37 + 5.33 ̇V O2 (4) Qtot = 4.43 + 5.22 ̇V O2 (5) S(P) = Pn Pn + P50 n (6) PB(a) = exp( − 0.00149a2 − 0.1112a + 6.63268) (7) PIO2 = FIO2 (PB(a) − Pw ) (8) PA(a) = PIO2 − PCO2(a) R TABLE 1 Cases considered for modeling V̇O2max as a function of altitude. Data depict cases considered for variation in hemoglobin- oxygen affinity (P50) and hemoglobin concentration in humans during sojourn to extreme altitudes. Normal hemoglobin- oxygen affinity High hemoglobin- oxygen affinity Low hemoglobin- oxygen affinity P50 (mmHg) Hemoglobin concentration (g/dL) P50 (mmHg) Hemoglobin concentration (g/dL) P50 (mmHg) Hemoglobin concentration (g/dL) Case 1 26.3, Ref. (20) 14.8, Ref. (17) Case 1 15.6, Ref. (5) 14.8 Case 1 37 14.8 Case 2 26.3 14.8 → 19.9, Ref. (17) Case 2 15.6 18.7, Ref. (21) Case 2 37 12.0 Case 3 26.3 → 24.8, Ref. (17) 14.8 Case 3 15.6 18.7 → 19.9* Case 3 37 12 → 16.2* Case 4 26.3 → 24.8 14.8 → 19.9 Case 4 15.6 → 14.7* 18.7 Case 4 37 → 34.9* 12.0 Case 5 15.6 → 14.7* 18.7 → 19.9* Case 5 37 → 34.9* 12 → 16.2* Note: Data were taken from experimental results when available. Arrows indicate changes in values from sea- level to an altitude of ~8400 m Ref (17). *Indicates data for case of high or low hemoglobin- oxygen affinity was assumed to change proportionally to alterations observed during extreme altitude sojourn among humans with normal hemoglobin- oxygen affinity. 4 of 11 | WEBB et al. as altitude increases, levels of capillary PO2 are reduced, limiting the pressure gradient for diffusive transport to tissue, eventually resulting in tissue hypoxia. Under these conditions, oxygen consumption rate falls short of oxygen demand. The local rate of oxygen consumption is generally assumed to depend on tissue PO2 with Michaelis– Menten kinetics (Popel,  1989). Estimating distributions of tissue PO2 levels would require several additional assumptions re- garding capillary density and oxygen transport properties. Thus, we employed a simplified approach, based on the as- sumption that Pv can be used as an approximation for tissue PO2. In this approach, oxygen consumption is assumed to be a function of Pv with Michaelis– Menten kinetics: where M is oxygen demand is oxygen demand, which is cal- culated such that predicted ̇V O2 at sea level corresponded to a typical observed value of ̇V O2max = 2750 mL O2/min in healthy young adults (van der Steeg & Takken, 2021). The oxygen demand M represents mitochondrial oxygen con- sumption capacity under conditions of unlimited oxygen supply. The model for systemic oxygen utilization uses a sim- plified approach, based on the assumption that Pv can be used as an approximation for tissue PO2. In reality, steep gradients in in tissue oxygen tensions around capillaries are present at V̇O2max, such that tissue PO2 is less than local capillary PO2. Also, intravascular PO2 declines in the axial direction along capillaries, such that venous PO2 represents a lower bound on capillary PO2. From these considerations, it follows that Pv represents an interme- diate value within the range of tissue PO2 levels and can be used as an approximate estimate of tissue PO2. The advantage of this approach, termed Fick– Michaelis– Menten (FickMM), is that it provides an estimate that is independent of capillary density and geometric arrange- ment, which are highly variable and for which data are not generally available for human subjects. According to the Fick principle, oxygen consumption rate must also satisfy: where S(P) describes the oxyhemoglobin dissociation curve and other quantities are defined in the main text. For any given set of conditions, the predicted values of ̇V O2 and Pv correspond to the simultaneous solution of Equations 9 and 10, as shown graphically in Figure 2. To evaluate the validity of approximating tissue PO2 by venous PO2, comparisons with a model using a conven- tional Krogh geometry and Michaelis– Menten kinetics were performed using the oxygen transport and geometric parameters described in Table  3. The calculations were performed by solving the radial diffusion equation in suc- cessive slices of a cylinder surrounding a central capillary, as described elsewhere (McGuire & Secomb, 2001). These simulations take into account the variations in tissue PO2 (9) ̇V O2 = M Pv Pv + P0 (10) ̇V O2 = QtotC0 (S(Pa ) − S(Pv )) Parameter Value Units Citation Water vapor pressure Pw 47 mmHg – Respiratory quotient R 0.8 – – Michaelis constant for oxygen consumption P0 10.5 mmHg (Golub & Pittman, 2012) Lung diffusing capacity DLO2 74 cm3 O2 min−1 mmHg−1 (Roy & Secomb, 2019) Capillary length Ltot 0.5 mm – Hill coefficient n 2.7 - (Hsia, 1998) TABLE 2 Parameter values used for oxygen transport calculations. FIGURE 2 Example of predicted maximal oxygen utilization rate using the systemic model. Conditions correspond to atmospheric pressure at sea level with typical hemoglobin- oxygen affinity and a hemoglobin concentration of 14.778 g/dL. Oxygen demand (M) was set as 4530 mL O2/min. The dashed line represents ̇VO2 calculated from Michaelis– Menten kinetics. The solid line represents ̇VO2 calculated from the Fick principle. Resulting predicted values at the intersection of the curves are ̇VO2 = 2750 mL O2/min and a venous partial pressure of oxygen (Pv ) of 16.2 mmHg. Pv, venous partial pressure of oxygen.  | 5 of 11 WEBB et al. with axial and radial position in the Krogh cylinder, when calculating the overall rate of oxygen consumption. From the results, the venous oxygen tension exiting the muscle compartment was estimated for a range of Pa, and for sev- eral capillary densities (McGuire & Secomb, 2001). The entire cardiac output was assumed to be directed to the muscle compartment. The results of these simulations (Figure 3) showed that the results for venous PO2 were similar to FickMM for capillary densities in the range of 1100– 1468 mm−2. While lower values of capillary density have been reported (Klau- sen et al., 1981; Qu et al., 1997; Richardson, 1995), prior calculations by McGuire and Secomb (2003) demonstrate that higher values of capillary density (1100– 1468 mm−2) are consistent with measured oxygen uptake and utiliza- tion rates, suggesting that histologically measured capil- lary densities may underestimate functional values in vivo. For lower capillary densities, the FickMM model would exaggerate the V̇O2 levels that could be achieved. Similar results were seen for simulations performed with high and low affinity hemoglobin variants (P50 = 15.6 and 37 mmHg). Corresponding calculations with a lower capillary den- sity and not all cardiac output going to the muscle would require a higher oxygen demand to match the values of V̇O2max assumed at sea level. If the Krogh model were used with a lower capillary density, then the predicted values of V̇O2max would be lower than those obtained with the FickMM model, but would show similar trends with altitude. The value of oxygen demand assumed for FickMM was calculated on the basis of the entire cardiac output being directed to the muscle compartment. Includ- ing effects of flow distribution to other organs would re- sult in lower predicted V̇O2max values. For these reasons, the estimates of V̇O2max reported in the paper can be con- sidered upper bounds. The determinants of V̇O2max may be represented by plotting convective and diffusive limitations of O2 delivery as a function of venous PO2, in a graph referred to as a “Wagner diagram” (Poole et al., 2012; Wagner, 1996). The diffusive limitation on oxygen transport was estimated using the Krogh cylinder model, using assumed values of capillary density, capillary diameter, intracapillary diffu- sion resistance, and blood and plasma oxygen diffusivity and solubility: where rc is the capillary radius, K and Kpl are Krogh dif- fusion coefficients in the tissue and the plasma, and Sh is the Sherwood number representing intravascular diffusion resistance. The tissue cylinder radius rt is computed based on an estimated capillary density obtained from Figure 3, such that the tissue PO2 matches the value obtained by the FickMM model. Convective oxygen delivery was calculated by the Fick principle as in Equation 10. Calculations were performed assuming that hemoglobin values did not change with altitude (Case 2) and that cardiac output was the same in all cases. (11) Pv = ̇V O2 [ rt 2 − rc 2 Kpl ∙ Sh − rt2 − rc2 − 2rt2ln(rt∕rc ) 4K ] TABLE 3 Assumed parameters and values used in the mathematical model. Description Parameter Value Units Source Sherwood number Sh 2.5 (Hellums et al., 1996) Plasma oxygen diffusivity Dpl 2.18E- 5 cm2 s−1 (Hellums et al., 1996) Plasma oxygen solubility αpl 2.82E- 5 mL O2 cm−3 mmHg−1 (Christoforides et al., 1969) Tissue capillary radius rc 2.5 μm (Roy & Secomb, 2014) Tissue oxygen diffusivity Dt 2.41E- 5 cm2 s−1 (Bentley et al., 1993) Tissue oxygen solubility αt 3.89E- 5 mL O2 cm−3 mmHg−1 (Bentley et al., 1993) FIGURE 3 Estimates of venous oxygen tension (Pv) obtained using the Fick- Michaelis Menten (FickMM) model as compared to using a Krogh model with Michaelis– Menten kinetics for muscle oxygen utilization. Results are presented as a function of arterial oxygen tension (Pa) for various capillary densities (1468, 1100, and 700 mm−2) depicted in the figure legend. Computations were performed using the oxygen transport parameters in Table 3. 6 of 11 | WEBB et al. 2.3 | Cases considered Cases of low P50, normal P50, and high P50 were investi- gated. The variation of hemoglobin parameters with alti- tude is not well established, particularly among humans with hemoglobin mutations. To encompass the likely range of variations in hemoglobin concentration and P50 with altitude, various cases indicated in Table 1 were con- sidered for each hemoglobin variant. These cases include ones in which the P50 was assumed to remain constant as a function of altitude, and others in which the P50 was as- sumed to decrease with altitude, as was observed Grocott et al. (2009). Corresponding values of hemoglobin concen- tration were assumed either to be constant or to vary with altitude according to the ratio observed experimentally, subject to the maximum value (Grocott et al., 2009). 3 | RESULTS Model predictions of blood oxygenation and V̇O2max as a function of altitude are presented for cases of low, normal, and high P50, and for several different assumptions about the variations of P50 and hemoglobin concentration with altitude, as indicated in Table 1. Profound blood gas alterations occur during human so- journ to extreme altitudes. Figure 4 displays predicted ox- ygen transport parameters for the three cases of P50 (low, normal, and high) as a function of altitude. The variation of arterial oxygen tension with increasing altitude is sim- ilar for all cases considered. However, arterial oxygen sat- uration and oxygen content are substantially increased at high and extreme altitudes for cases of low P50 compared to predicted values for cases of normal P50 and high P50. The predicted V̇O2max as a function of altitude is de- picted in Figure 5 for the cases of low, normal, and high P50. As expected, V̇O2max decreases with increasing altitude in all cases. However, the variation of V̇O2max with altitude is markedly dependent on hemoglobin- oxygen affinity. In the case of high P50, the predicted V̇O2max at sea level is greater than predicted values for normal P50, but markedly lower at altitudes above ~2500 m. Conversely, in the case of low P50, V̇O2max is lower than values predicted for normal P50 at sea level, but greater at altitudes above ~4500 m. FIGURE 4 Predicted oxygen transport parameters as a function altitude for cases of low, normal, and high hemoglobin- oxygen affinity. Data are depicted for several cases of hemoglobin- oxygen affinity (low P50 in blue, normal P50 in red, and high P50 in yellow) with variable hemoglobin concentrations to account for potential differences in the hematological response to extreme altitude sojourn indicated in Table 1. These parameters are derived for a given tissue oxygen demand that corresponds with sea- level maximum oxygen uptake and utilization (V̇O2max). The elevation associated with the summit of Everest is depicted by the dashed vertical line. Parameters corresponding to altitudes above 8400 m are derived from the extrapolation of oxygen transport parameters in Grocott et al. (2009). P50, oxygen tension at which 50% of hemoglobin is saturated with oxygen.  | 7 of 11 WEBB et al. Predictions of V̇O2max from the present model are pre- sented in Figure 6 together with lines and curves repre- senting limitations on oxygen utilization according to the Wagner diagram. At sea level, V̇O2max shows small varia- tions with P50, with a slight advantage at normal P50. At an altitude of ~8850 m, convective oxygen delivery is greatly reduced and shows a strong inverse dependence on P50. The higher rates of convective oxygen delivery at low P50 result from two factors: higher arterial oxygen saturation (~45%, vs. ~30% for high P50) and higher hemoglobin val- ues (~50% greater than for high P50). 4 | DISCUSSION 4.1 | Physiological implications V̇O2max is determined by convective and diffusive oxygen transport, both of which are influenced by alterations in hemoglobin- oxygen affinity (Hebbel et al.,  1977; Webb, Elshaer, et al.,  2022b). Specifically, high hemoglobin- oxygen affinity tends to enhance pulmonary oxygen up- take, particularly when alveolar oxygen tension is low, increasing convective oxygen transport. On the contrary, high hemoglobin- oxygen affinity implies a lower blood oxygen tension for a given level of oxygen saturation, such that the driving force for oxygen diffusion from blood to tissue is reduced. The relative influences of these two competing effects of high hemoglobin- oxygen affinity (low P50) on V̇O2max cannot easily be discerned by quali- tative arguments. Therefore, we investigated this rela- tionship using a mathematical model of oxygen transport that includes both pulmonary and systemic circulation and considers the effects of both convective and diffusive oxygen transport. The model also considers the influence of high altitude on oxygen availability and uptake in the lungs. Previous analyses of effects of P50 on oxygen transport at extreme altitude suggested that V̇O2max is insensitive to P50 over a considerable range (Bencowitz et al., 1982; Wag- ner, 1997). The present model differs from those analyses in two significant respects. First, it includes effects of vari- ations in hemoglobin levels in individuals with altered P50, which results in increased convective oxygen delivery in the case of low P50. Second, it takes into account the non- linear Michaelis– Menten kinetics of oxygen utilization as a function of tissue PO2, representing the finite rate of mi- tochondrial oxygen consumption when oxygen is not rate- limiting. Both of these effects result in increased predictions FIGURE 5 The dependence of predicted maximum oxygen uptake and utilization (V̇O2max) on hemoglobin- oxygen affinity (P50) and altitude. Data are depicted for several cases of hemoglobin- oxygen affinity (low P50 in blue, normal P50 in red, and high P50 in yellow) with variable hemoglobin concentrations to account for potential differences in the hematological response to extreme altitude sojourn indicated in Table 1. These parameters are derived for a given tissue oxygen demand that corresponds with sea- level V̇O2max. The elevation associated with the summit of Everest is depicted by the dashed vertical line. Parameters corresponding to altitudes above 8400 m are derived from the extrapolation of oxygen transport parameters in Grocott et al. (2009). P50, oxygen tension at which 50% of hemoglobin is saturated with oxygen. FIGURE 6 Predicted oxygen uptake and utilization (V̇O2) presented as a Wagner diagram for cases of low, normal, and high P50. Data for each group are presented at sea- level (solid lines) and an altitude of ~8850 m (summit of Mt. Everest, represented by dashed lines). The V̇O2max value is depicted by the intersection between convective oxygen transport (curved lines obtained via Fick principle) and diffusive oxygen transport (dotted line passing through the origin). Open circles plotted on curved lines denote the predicted V̇O2max as determined using Michaelis– Menten kinetics. Pa, arterial oxygen tension; V̇O2, oxygen uptake and utilization, P50; oxygen tension at which 50% of hemoglobin is saturated with oxygen. Values of Pa and P50 are in units of mmHg. 8 of 11 | WEBB et al. of V̇O2max at extreme altitude for reduced P50, and account for the apparent discrepancy with the earlier work. The diffusive limitation of oxygen transport, Equa- tion  11, is computed assuming a uniform rate of oxy- gen consumption throughout the tissue. In contrast, the FickMM model allows for variations of oxygen levels and oxygen consumption rates in the tissue, including possi- ble hypoxic regions. The resulting estimates of V̇O2max at altitude are slightly higher than those obtained from the intersections of the diffusive limitation line and the con- vective delivery curves, as shown in Figure 6. This differ- ence is most evident in the case of low P50. The diffusive limitation line shown in Figure 6 is based on the assump- tion that PO2 values approach zero only at the point in the tissue furthest from the distal end of the supplying capillary. If oxygen demand is further increased, overall oxygen consumption can increase beyond the value im- plied by diffusive limitation, even if some regions of tis- sue are hypoxic (McGuire & Secomb, 2001). Humans at high altitude experience a range of acute (dehydration, alkalosis, hypocapnia) and chronic (train- ing, acclimatization) effects, both of which may be associated with variations in P50 and hemoglobin con- centration (Mairbaurl & Weber, 2012; Monge & Leon- Velarde, 1991; Windsor & Rodway, 2007). Because these effects are not widely characterized among humans with hemoglobin mutations, we considered a range of cases for each case of hemoglobin- oxygen affinity (low, normal, and high P50). Thus, the predicted V̇O2max is provided with a range of values for each altitude and case examined, providing an indication of the sensitivity of the model to these potential variations in P50 and he- moglobin concentration. Individual variations in phys- iological parameters such as capillary density and lung diffusing capacity, as well as potential alterations of these parameters during extreme altitude sojourn, may be substantial and would obviously affect the predictions of this model. However, the primary trends in V̇O2max as a function of altitude are likely to remain similar even if baseline values are notably different. Other parame- ters of oxygen transport, such as lung diffusing capacity and the respiratory quotient, may vary with altitude but are assumed to be constant in the model. Additionally, the effects of non- muscle blood flow on overall oxy- gen transport are not considered, as the entire cardiac output is assumed to be directed to the skeletal muscle during maximal exercise. Our results revealed that at low altitudes, where at- mospheric pressure is more than sufficient to cause nearly complete saturation of hemoglobin, a low P50 does not confer an advantage in terms of oxygen utilization since convective transport is sufficient to supply skeletal muscle. At high altitudes, however, a low P50 increases convective oxygen delivery due to higher oxygen satu- ration values, despite the diffusion limitation resulting from lower blood oxygen tension. This improved oxy- gen delivery allows for better preservation of V̇O2max at high altitudes. In summary, a low P50 leads to a reduced driving force for oxygen diffusion from blood to tissue at low altitudes yet increased convective oxygen delivery at high altitudes. These two competing tendencies ap- proximately cancel at an altitude of ~4500 m such that high hemoglobin- oxygen affinity confers an advantage at higher altitudes. In the results presented here, the effects of capillary density are not explicitly considered under the approx- imation that venous oxygen tension is representative of tissue oxygen tension. More detailed calculations show that high capillary densities can lead to greater tissue oxygen tensions values than assumed here. A high cap- illary density may facilitate the advantage conferred by a low P50 due to decreased diffusion limitation. Con- versely, a low capillary density may negate the advan- tage of a low P50 at high and extreme altitude because oxygen delivery would then be limited by reduced mus- cle diffusing capacity. Practical applications Studies in comparative physiology show a wide range of adaptations to altitude, some of which have sup- ported that an increase in hemoglobin- oxygen affinity is likely beneficial for species adapted to high and ex- treme altitude (Natarajan et al., 2018; Storz, 2007; Storz et al., 2010). Across species, multiple factors including evolutionary pressures may influence the observed ad- aptations in hemoglobin- oxygen affinity. Further de- tailed investigation of this topic in terms of convective versus diffusive oxygen transport limitations would be appropriate, and the theoretical approach developed here may be applicable to such studies. Pharmacological agents have been developed that can alter P50 in healthy individuals (Henry et al., 2021; Safo & Kato, 2014; Woyke et al., 2021). Although these agents are mainly investigated for treatment of sickle- cell disease, they have also been used in healthy individuals (Stewart et al., 2021, 2020). According to the present re- sults, decreasing the P50 has significant effects on blood oxygenation and V̇O2max at altitude, some of which may prove beneficial depending on the environmental context. For instance, pharmacologically decreasing the P50 may have an ergogenic effect at high and extreme altitudes by increasing arterial blood saturation and im- proving convective oxygen delivery. This raises the pos- sibility that such agents could be used for “blood doping” in competitive sports. In military operations at high and extreme altitudes, environmental conditions may limit  | 9 of 11 WEBB et al. physical performance and cognitive function (McLaugh- lin et al., 2017). Pharmacological reduction in P50 may in- crease hypoxia tolerance (Dufu et al., 2021) and prevent decrements in physical performance (Stewart et al., 2021). However, further work is needed to examine the advan- tages or disadvantages of pharmacologically altering P50 in healthy individuals in various contexts. The present model may be useful for predicting the change in P50 at a given altitude that maximizes the ergogenic effect. 4.3 | Limitations A major simplification of this model is the use of venous PO2 as a measure of tissue PO2 for the purpose of cal- culating oxygen consumption according to Michaelis– Menten kinetics. The rationale for this assumption is that venous PO2 typically lies within the range of the minimum and maximum tissue PO2. As shown in Fig- ure 3, oxygen consumption rates calculated under this assumption show reasonable agreement with more de- tailed calculations using a Krogh cylinder model. This approach avoids the need to specify the geometry of the capillary network, since such detailed information is generally not available. However, the limitation of this approach is that it does not include the effects of capil- lary network geometry. Previous studies have indicated that the Bohr effect (pH dependent change in the P50) may play a notable role in the determination of V̇O2max (Severinghaus, 1994). How- ever, this effect was not considered in the present model. Because the magnitude of the Bohr effect at extreme alti- tudes is not known, it was excluded to facilitate comparisons across altitudes. Given that the Bohr effect is generally pre- served among humans with hemoglobin mutations (Boyer et al., 1972), its effect on V̇O2max values would likely be uni- directional and comparable between the groups examined. Additionally, past investigations have described aberrations in metabolic processes during exercise among humans with low P50 and suggested that skeletal muscle and mitochon- drial adaptations may compensate for the blunted oxygen offloading (Wranne et al., 1983). At extreme altitude, how- ever, V̇O2max is severely limited by the reduced oxygen availability, such that changes in maximal mitochondrial oxygen consumption are unlikely to affect V̇O2max. There- fore, the present model assumes similar mitochondrial function between groups with normal and altered P50. 5 | CONCLUSION The presented analyses leverage experimental data among humans with hemoglobin mutations to predict blood oxygenation and V̇O2max as a function of hemoglobin- oxygen affinity and altitude. We posit that high hemoglobin- oxygen affinity leads to improved blood oxygenation and better preserved V̇O2max values at extreme altitudes compared to values associated with normal hemoglobin- oxygen affinity. Additionally, we provide theoretical estimates for V̇O2max as a function of altitude among humans with mutations causing low hemoglobin- oxygen affinity, which has yet to be exam- ined experimentally. AUTHOR CONTRIBUTIONS Kevin L. Webb and Tuhin K. Roy conceived the presented idea. All author contributed to the methodological design of this work. Kevin L. Webb, Tuhin K. Roy, and Timothy W. Secomb contributed to model development, analyses, and data visualization. Kevin L. Webb and Tuhin K. Roy constructed the initial manuscript draft. All authors con- tributed to manuscript revising and have approved the final submission. ACKNOWLEDGMENTS The authors would like to thank the members of the Human and Integrative Physiology and Clinical Pharma- cology Laboratory at the Mayo Clinic for intellectual dis- cussion and feedback. FUNDING INFORMATION This work was funded by the National Institutes of Health grant R35- HL139854 (M.J.J.). CONFLICT OF INTEREST STATEMENT The authors have no conflict of interest to declare. ETHICS STATEMENT The presented study was exempt from obtaining IRB approval and does not present novel data pertaining to human nor animal subjects. DATA AVAILABILITY STATEMENT All pertinent data are presented within the manuscript. All code used to perform calculations will be shared upon reasonable request. ORCID Kevin L. Webb  https://orcid.org/0000-0003-3015-6076 Michael J. Joyner  https://orcid.org/0000-0002-7135-7643 Chad C. Wiggins  https://orcid.org/0000-0002-6458-0142 Timothy W. Secomb  https://orcid. org/0000-0002-0176-5502 Tuhin K. Roy  https://orcid.org/0000-0002-8182-7629 10 of 11 | WEBB et al. REFERENCES Bencowitz, H. Z., Wagner, P. D., & West, J. B. (1982). Effect of change in P50 on exercise tolerance at high altitude: A theoret- ical study. Journal of Applied Physiology (1985), 53, 1487– 1495. https://doi.org/10.1152/jappl.1982.53.6.1487 Bentley, T. B., Meng, H., & Pittman, R. N. (1993). Temperature dependence of oxygen diffusion and consumption in mam- malian striated muscle. American Journal of Physiology, 264, H1825– H1830. Boyer, S. H., Charache, S., Fairbanks, V. F., Maldonado, J. E., Noyes, A., & Gayle, E. E. (1972). Hemoglobin Malmö β- 97 (FG- 4) his- tidine→glutamine: A cause of polycythemia. The Journal of Clinical Investigation, 51, 666– 676. Calbet, J. A. L., & Boushel, R. (2015). Assessment of cardiac output with transpulmonary thermodilution during exercise in hu- mans. Journal of Applied Physiology (1985), 118, 1– 10. https:// doi.org/10.1152/jappl physi ol.00686.2014 Charache, S., Weatherall, D. J., & Clegg, J. B. (1966). Polycythemia associated with a hemoglobinopathy. The Journal of Clinical Investigation, 45, 813– 822. https://doi. org/10.1172/JCI10 5397 Christoforides, C., Laasberg, L. H., & Hedley- Whyte, J. (1969). Effect of temperature on solubility of O2 in human plasma. Journal of Applied Physiology, 26, 56– 60. Dominelli, P. B., Wiggins, C. C., Baker, S. E., Shepherd, J. R. A., Roberts, S. K., Roy, T. K., Curry, T. B., Hoyer, J. D., Oliveira, J. L., & Joyner, M. J. (2020). Influence of high affinity haemo- globin on the response to normoxic and hypoxic exercise. The Journal of Physiology, 598, 1475– 1490. https://doi.org/10.1113/ JP279161 Dufu, K., Williams, A. T., Muller, C. R., Walser, C. M., Lucas, A., Eaker, A. M., Alt, C., Cathers, B. E., Oksenberg, D., & Cabrales, P. (2021). Increased hemoglobin affinity for oxygen with GBT1118 improves hypoxia tolerance in sickle cell mice. American Journal of Physiology. Heart and Circulatory Physiology, 321, H400– H411. https://doi.org/10.1152/ajphe art.00048.2021 Golub, A. S., & Pittman, R. N. (2012). Oxygen dependence of respi- ration in rat spinotrapezius muscle in situ. American Journal of Physiology. Heart and Circulatory Physiology, 303, H47– H56. https://doi.org/10.1152/ajphe art.00131.2012 Grocott, M. P. W., Levett, D. Z. H., & Windsor, J. (2009). Arterial blood gases and oxygen content in climbers on Mount Everest. The New England Journal of Medicine, 360, 140– 149. https:// doi.org/10.1056/NEJMo a0801581 Hebbel, R. P., Eaton, J. W., Kronenberg, R. S., Zanjani, E. D., Moore, L. G., & Berger, E. M. (1978). Human llamas: Adaptation to altitude in subjects with high hemoglobin oxygen affinity. The Journal of Clinical Investigation, 62, 593– 600. https://doi. org/10.1172/JCI10 9165 Hebbel, R. P., Kronenberg, R. S., & Eaton, J. W. (1977). Hypoxic ventilatory response in subjects with normal and high oxygen affinity hemoglobins. The Journal of Clinical Investigation, 60, 1211– 1215. https://doi.org/10.1172/JCI10 8874 Hellums, J. D., Nair, P. K., Huang, N. S., & Ohshima, N. (1996). Simulation of intraluminal gas transport processes in the mi- crocirculation. Annals of Biomedical Engineering, 24, 1– 24. https://doi.org/10.1172/JCI10 8874 Henry, E. R., Metaferia, B., Li, Q., Harper, J., Best, R. B., Glass, K. E., Cellmer, T., Dunkelberger, E. B., Conrey, A., Thein, S. L., Bunn, H. F., & Eaton, W. A. (2021). Treatment of sickle cell disease by increasing oxygen affinity of hemoglobin. Blood, 138, 1172– 1181. https://doi.org/10.1182/blood.20210 12070 Hsia, C. C. W. (1998). Respiratory function of hemoglobin. New England Journal of Medicine, 338, 239– 248. https://doi. org/10.1056/NEJM1 99801 22338 0407 Klausen, K., Andersen, L. B., & Pelle, I. (1981). Adaptive changes in work capacity, skeletal muscle capillarization and enzyme levels during training and detraining. Acta Physiologica Scandinavica, 113, 9– 16. https://doi.org/10.1111/j.1748- 1716.1981.tb068 54.x Mairbaurl, H., & Weber, R. E. (2012). Oxygen transport by hemo- globin. Comprehensive Physiology, 2, 1463– 1489. https://doi. org/10.1002/cphy.c080113 McGuire, B. J., & Secomb, T. W. (2001). A theoretical model for oxygen transport in skeletal muscle under conditions of high oxygen demand. Journal of Applied Physiology, 91, 2255– 2265. https://doi.org/10.1152/jappl.2001.91.5.2255 McGuire, B. J., & Secomb, T. W. (2003). Estimation of capillary density in human skeletal muscle based on maximal oxygen consumption rates. American Journal of Physiology. Heart and Circulatory Physiology, 285, H2382– H2391. https://doi. org/10.1152/ajphe art.00559.2003 McLaughlin, C. W., Skabelund, A. J., & George, A. D. (2017). Impact of high altitude on military operations. Current Pulmonology Reports, 6, 146– 154. https://doi.org/10.1007/s1366 5- 017- 0181- 0 Monge, C., & Leon- Velarde, F. (1991). Physiological adaptation to high altitude: Oxygen transport in mammals and birds. Physiological Reviews, 71, 1135– 1172. https://doi.org/10.1152/ physr ev.1991.71.4.1135 Natarajan, C., Jendroszek, A., Kumar, A., Weber, R. E., Tame, J. R. H., Fago, A., & Storz, J. F. (2018). Molecular basis of hemoglobin adaptation in the high- flying bar- headed goose. PLoS Genetics, 14, e1007331. https://doi.org/10.1371/journ al.pgen.1007331 Poole, D. C., Hirai, D. M., Copp, S. W., & Musch, T. I. (2012). Muscle oxygen transport and utilization in heart failure: Implications for exercise (in)tolerance. American Journal of Physiology. Heart and Circulatory Physiology, 302, H1050– H1063. https:// doi.org/10.1152/ajphe art.00943.2011 Popel, A. S. (1989). Theory of oxygen transport to tissue. Critical Reviews in Biomedical Engineering, 17, 257– 321. Qu, Z., Andersen, J. L., & Zhou, S. (1997). Visualisation of capillaries in human skeletal muscle. Histochemistry and Cell Biology, 107, 169– 174. https://doi.org/10.1007/s0041 80050101 Richardson, R. (1995). Oxygen transport to tissue XVI (1st ed.). Springer. Roy, T. K., & Secomb, T. W. (2014). Theoretical analysis of the deter- minants of lung oxygen diffusing capacity. Journal of Theoretical Biology, 351, 1– 8. https://doi.org/10.1016/j.jtbi.2014.02.009 Roy, T. K., & Secomb, T. W. (2019). Effects of pulmonary flow heterogeneity on oxygen transport parameters in exercise. Respiratory Physiology & Neurobiology, 261, 75– 79. https://doi. org/10.1016/j.resp.2018.10.004 Safo, M. K., & Kato, G. J. (2014). Therapeutic strategies to Alter oxy- gen affinity of sickle hemoglobin. Hematology/Oncology Clinics of North America, 28, 217– 231. https://doi.org/10.1016/j. hoc.2013.11.001 Severinghaus, J. W. (1994). Exercise O2 transport model assum- ing zero cytochrome PO2 at VO2max. Journal of Applied Physiology (1985), 77, 671– 678, 678. https://doi.org/10.1152/ jappl.1994.77.2.671  | 11 of 11 WEBB et al. Stewart, G. M., Chase, S., Cross, T. J., Wheatley- Guy, C. M., Joyner, M., Curry, T., Lehrer- Graiwer, J., Dufu, K., Vlahakis, N. E., & Johnson, B. D. (2020). Effects of an allosteric hemoglobin af- finity modulator on arterial blood gases and cardiopulmonary responses during normoxic and hypoxic low- intensity exercise. Journal of Applied Physiology (1985), 128, 1467– 1476. https:// doi.org/10.1152/jappl physi ol.00185.2019 Stewart, G. M., Cross, T. J., Joyner, M. J., Chase, S. C., Curry, T., Lehrer- Graiwer, J., Dufu, K., Vlahakis, N. E., & Johnson, B. D. (2021). Impact of pharmacologically left shifting the oxygen– hemoglobin dissociation curve on arterial blood gases and pulmonary gas exchange during maximal exercise in hypoxia. High Altitude Medicine & Biology, 22(3), 249– 262. Storz, J. F. (2007). Hemoglobin function and physiological adaptation to hypoxia in high- altitude mammals. Journal of Mammalogy, 88, 24– 31. https://doi.org/10.1644/06- MAMM- S- 199R1.1 Storz, J. F., Scott, G. R., & Cheviron, Z. A. (2010). Phenotypic plas- ticity and genetic adaptation to high- altitude hypoxia in ver- tebrates. The Journal of Experimental Biology, 213, 4125– 4136. https://doi.org/10.1242/jeb.048181 Thom, C. S., Dickson, C. F., Gell, D. A., & Weiss, M. J. (2013). Hemoglobin variants: Biochemical properties and clinical cor- relates. Cold Spring Harbor Perspectives in Medicine, 3, a011858. https://doi.org/10.1101/cshpe rspect.a011858 van der Steeg, G. E., & Takken, T. (2021). Reference values for maximum oxygen uptake relative to body mass in Dutch/ Flemish subjects aged 6– 65 years: The LowLands fitness reg- istry. European Journal of Applied Physiology, 121, 1189– 1196. https://doi.org/10.1007/s0042 1- 021- 04596 - 6 Wagner, P. D. (1996). Determinants of maximal oxygen transport and utilization. Annual Review of Physiology, 58, 21– 50. https:// doi.org/10.1146/annur ev.ph.58.030196.000321 Wagner, P. D. (1997). Insensitivity of VO2max to hemoglobin- P50 at sea level and altitude. Respiration Physiology, 107, 205– 212. https://doi.org/10.1016/s0034 - 5687(96)02512 - 1 Webb, K. L., Dominelli, P. B., Baker, S. E., Klassen, S. A., Joyner, M. J., Senefeld, J. W., & Wiggins, C. C. (2022a). Influence of high hemoglobin- oxygen affinity on humans during hypoxia. Frontiers in Physiology, 12, 763– 933. https://doi.org/10.3389/ fphys.2021.763933 Webb, K. L., Elshaer, A. N., Dominelli, P. B., Senefeld, J. W., Hammer, S. M., Baker, S. E., Shepherd, J. R. A., Roy, T. K., Joyner, M. J., & Wiggins, C. C. (2022b). Muscle oxygenation during normoxic and hypoxic cycling exercise in humans with high- affinity hae- moglobin. Experimental Physiology, 107, 854– 863. https://doi. org/10.1113/EP090308 West, J. B. (1999). Barometric pressures on Mt. Everest: New data and physiological significance. Journal of Applied Physiology (1985), 86, 1062– 1066. https://doi.org/10.1152/jappl.1999.86.3.1062 Windsor, J. S., & Rodway, G. W. (2007). Heights and haematology: The story of haemoglobin at altitude. Postgraduate Medical Journal, 83, 148– 151. https://doi.org/10.1136/pgmj.2006.049734 Woyke, S., Mair, N., Ortner, A., Haller, T., Ronzani, M., Rugg, C., Ströhle, M., Wintersteiger, R., & Gatterer, H. (2021). Dose- and sex- dependent changes in hemoglobin oxygen affinity by the micronutrient 5- Hydroxymethylfurfural and α- ketoglutaric acid. Nutrients, 13, 3448. https://doi.org/10.3390/nu131 03448 Wranne, B., Berlin, G., Jorfeldt, L., & Lund, N. (1983). Tissue ox- ygenation and muscular substrate turnover in two subjects with high hemoglobin oxygen affinity. The Journal of Clinical Investigation, 72, 1376– 1384. https://doi.org/10.1172/JCI11 1094 How to cite this article: Webb, K. L., Joyner, M. J., Wiggins, C. C., Secomb, T. W., & Roy, T. K. (2023). The dependence of maximum oxygen uptake and utilization (V̇O2max) on hemoglobin- oxygen affinity and altitude. Physiological Reports, 11, e15806. https://doi.org/10.14814/phy2.15806

The dependence of maximum oxygen uptake and utilization (V̇O₂ max) on hemoglobin-oxygen affinity and altitude.

[]

Webb, Kevin L,Joyner, Michael J,Wiggins, Chad C,Secomb, Timothy W,Roy, Tuhin K

eng

PMC7696724

sensors Review Mechanical Power in Endurance Running: A Scoping Review on Sensors for Power Output Estimation during Running Diego Jaén-Carrillo 1 , Luis E. Roche-Seruendo 1 , Antonio Cartón-Llorente 1 , Rodrigo Ramírez-Campillo 2 and Felipe García-Pinillos 3,4,* 1 Department of Physiotherapy, Universidad San Jorge, Villanueva de Gállego, 30580 Zaragoza, Spain; djaen@usj.es (D.J.-C.); leroche@usj.es (L.E.R.-S.); acarton@usj.es (A.C.-L.) 2 Department of Physical Activity Sciences, Universidad de Los Lagos, 5290000 Osorno, Chile; r.ramirez@ulagos.cl 3 Department of Physical Education and Sport, University of Granada, 18071 Granada, Spain 4 Department of Physical Education, Sports and Recreation, Universidad de La Frontera, 4811000 Temuco, Chile * Correspondence: fgpinillos@ugr.es; Tel.: +34-660062066 Received: 10 September 2020; Accepted: 10 November 2020; Published: 13 November 2020



 Abstract: Mechanical power may act as a key indicator for physiological and mechanical changes during running. In this scoping review, we examine the current evidences about the use of power output (PW) during endurance running and the different commercially available wearable sensors to assess PW. The Boolean phrases endurance OR submaximal NOT sprint AND running OR runner AND power OR power meter, were searched in PubMed, MEDLINE, and SCOPUS. Nineteen studies were finally selected for analysis. The current evidence about critical power and both power-time and power-duration relationships in running allow to provide coaches and practitioners a new promising setting for PW quantification with the use of wearable sensors. Some studies have assessed the validity and reliability of different available wearables for both kinematics parameters and PW when running but running power meters need further research before a definitive conclusion regarding its validity and reliability. Keywords: biomechanics; endurance runners; long-distance athletes; wearable device 1. Introduction Endurance running events are on the apex of a performance revolution, with the sub-2-h marathon barrier just broken (i.e., Vienna in 2019). In the same way the power meter changed training and racing in cycling [1] by providing a fair tool to assess performance with accurate replication, it might also change the way runners compete and train. Power, a term originated in classical physics, is defined as the product of force and velocity [2]. Despite training delivers stress on the body, the way runners measure this level of stress has been very limited. The faster a runner goes, the higher the stress for a certain level of fitness. Training intensity is the true marker to fitness (i.e., capacity to deal with a particular amount of stress) [3]. The application of mechanical load (i.e., external training load factors) and psychological and physiological efforts (i.e., internal training load factors) are affected by training stress [4]. In running, some external load factors including volume and pace are widely used, while physiological internal load factors consider perceived exertion scales, heart rate, or blood lactate level [4]. On multiple training days, running distance alone could overshadow the accumulated training stress and, eventually, misinterpret the overall training stress [4]. Pace might be as clear as volume but, indeed, it is not easy to assess as Sensors 2020, 20, 6482; doi:10.3390/s20226482 www.mdpi.com/journal/sensors Sensors 2020, 20, 6482 2 of 20 the running settings (i.e., surface; slope gradient) as well as weather conditions (i.e., wind velocity) or individual internal factors (i.e., stress, sleep, illness) may affect pace considerably and, therefore, challenge pace intensity quantification. None of these variables provides a fair and repeatable method to measure training intensity and, when training stress is measured imprecisely, injury risk may be increased and performance negatively altered. Given that new wearable devices allow to measure external load metrics apart from both volume and pace, there should be a growing focus on a combination of both biomechanical external (i.e., power output (PW)) and internal load metrics in the future of athletes monitoring [4]. Running, as cycling, is cyclical in nature. When running, three dimensional movements are needed. Normally, the body describes a forward movement, vertical oscillation, and a bilateral rotation over the running cycle. For such movements, mechanical work is required accounting vertical and forward movements for most of it. Throughout such movements, a runner acquires both kinetic energy and potential energy changes. The applied work runners develop over the loading phase and the subsequent take-off push to lift their body at every stride to work against environmental factors (i.e., ground reaction force, gravity force, and surface) refers to the external mechanical work. Then, the foot absorbs energy when colliding with the ground and produces power when pushing off. During running, expensive equipment such as specific instrumented treadmills [5] have been utilised to acquire force data. Despite their proved accuracy, most coaches and practitioners are forced to avoid their use due to economic issues. Over the last years, inertial measurement units (IMUs) emerged, allowing the quantification of performance, providing coaches and athletes an easy-to-use tool to monitor PW during running (e.g., Runscribe (Scribe Lab. Inc., Half Moon Bay, CA, USA), Stryd (Stryd Inc. Boulder, CO, USA) or Myotest (Myotest SA, Sion, Switzerland)). Previous works have demonstrated the direct relationship between anthropometric measures (e.g., body mass) and spatiotemporal parameters and kinetics and kinematics [6–8]. Samozino and colleagues [9] attempted to supply an affordable method to assess force-velocity and power-velocity profiles, using anthropometric and spatiotemporal data along over-ground sprint acceleration. However, Samozino’s approach is inapplicable to submaximal velocities. Currently, an increasing number of systems allow the assessment of running power (new heart rate monitors by Polar (Polar Electro Ltd., Kempele, Finland) and Garmin (Garmin Ltd., Olathe, KS, USA)). Nevertheless, there is a lack of scientific evidence testing either its validity or reliability, as well as limited insights on the use and interpretation of power in endurance runners, being this reduced to a few books [3,10], and further information provided by the devices’ manufacturers (e.g., Stryd, https://blog.stryd.com/tag/validation-white-papers/; Myotest, https://www.myotest.com/technology; RunScribe, https://runscribe.com/blog/; Stryd, https://blog.stryd.com; Polar: https://www.polar.com/es/ smart-coaching/running-power). Although the validity and reliability of a wide array of wearable sensors have been shown for running spatiotemporal parameters measurement and they seem to be related with PW estimation [11–15], a deeper knowledge on PW in endurance running and a proper understanding on the use of power meters to quantify workload would be an outstanding step forward towards a new boundary within running training and performance. There is a need to measure training intensity with precision and wearable sensors might help monitor the training-induced stress and, although previous review articles have been focused on power data while running [16,17], none of those concentrated on validity and reliability of such wearables for running PW analysis. Advances in the knowledge of endurance running PW would allow the assessment and monitor of power not only in laboratory settings, but in the field as well. Therefore, the aim of this scoping review was to critically examine the available running power meters and the current evidences about their use and application to endurance running performance. Sensors 2020, 20, 6482 3 of 20 2. Materials and Methods A review of the literature was conducted following the guidelines of the Cochrane Collaboration and taking into consideration the guidance provided by previous studies focused on scoping reviews [18,19]. This design (i.e., scoping review) was selected in order to have a broader approach with the aim of mapping literature characterized by a variety of study designs. Additionally, findings were reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) for scoping reviews [20]. 2.1. Eligibility Criteria Despite the limited evidence on this topic, some a priori inclusion criteria were considered for this scoping review: (i) only peer-reviewed articles were included; (ii) studies that were not published in English were not explored; (iii) no restrictions for age or sex of participants were applied. Additionally, no limitations regarding the study design were established. All manuscripts related to running with power or power meters were considered, regardless the study design, except literature reviews (e.g., systematic reviews or metanalysis). 2.2. Information sources A systematic search was conducted in the electronic databases PubMed, MEDLINE and SCOPUS for relevant studies until 1 June 2020. Keywords were collected through experts’ opinion, a systematic literature review, and controlled vocabulary (e.g., Medical Subject Headings: MeSH). Boolean search syntax using the operators “AND” and “OR” was applied. The words “endurance”, “running”, “runner”, “power”, and “power meter” were used. Following is an example of a PubMed search: ((((((endurance) OR submaximal) NOT sprint) AND running) OR runner) AND power) OR power meter; Filters: Publication date from 1 January 2000; Humans; English. After an initial search, accounts were created in the respective databases. Through these accounts, the lead investigator received automatically generated emails for updates regarding the search terms used. These updates were received on a daily basis (if available), and studies were eligible for inclusion until the initiation of manuscript preparation on 5 June 2020. Following the formal systematic searches, additional hand-searches were conducted. Grey literature sources (e.g., conference proceedings) were also considered if a full-text version was available. In addition, the reference lists of included studies and previous reviews and meta-analyses were examined to detect studies potentially eligible for inclusion. 2.3. Study Selection In selecting studies for inclusion, the three-step method was followed [21]. The first step, according to this procedure, was an initial restricted search of the appropriate database collection, followed by an analysis of the text words included in the title and abstract, and the index terms used to characterize the document. A second search using all known keywords and index terms was performed through all included databases. Finally, the reference list of all the selected studies and reports has been checked for additional studies. The authors included the aforementioned filters (i.e., the language and the publication date limitations). 2.4. Methodological Quality in Individual Studies To analyse the methodological quality in studies, the recommendations by Cochrane Review Groups were taken into consideration [22]. Since all the studies examined show a cross-sectional design, quality was assessed using the modified version of the Quality Index developed by Downs and Black [23]. The original scale was reported to have good test–retest (r = 0.88) and inter-rater (r = 0.75) reliability and high internal consistency (Kuder–Richardson Formula 20 (KR-20) = 0.89). The modified version of the Downs and Black Quality Index is scored from 1 to 14, with higher scores indicating Sensors 2020, 20, 6482 4 of 20 higher-quality studies. Two independent reviewers (DJC-FGP) performed this process and, in the event of a disagreement about the methodological quality, a third reviewer (LERS) checked the data and took the final decision on it. Agreement between reviewers was assessed using a Kappa correlation for methodological quality. The agreement rate between reviewers was k = 0.93 which can be interpreted as almost perfect [24]. It is worth noting that the study by Snyder and colleagues [25] was excluded as it is a letter to the editor in response to Aubry and colleagues’ [26] work. 3. Results 3.1. Study Selection Figure 1 provides a graphical schematization of the study selection process. A total of 1281 studies were initially identified: 640 from PubMed, 378 from SCOPUS, and 263 from MEDLINE. Additionally, 6 studies were identified through other resources. From these 1287 studies, 674 after duplicates removed. The 613 studies excluded after titles and abstracts revisions were essentially based on a lack of relationship with the research interests of this review. After full-text revision, only 19 studies which included either validity or reliability of running wearable sensors suppling running PW and/or the specific discussion of such wearable sensors were considered for the current work. Sensors 2020, 20, x FOR PEER REVIEW 4 of 21 3. Results 3.1. Study Selection Figure 1 provides a graphical schematization of the study selection process. A total of 1281 studies were initially identified: 640 from PubMed, 378 from SCOPUS, and 263 from MEDLINE. Additionally, 6 studies were identified through other resources. From these 1287 studies, 674 after duplicates removed. The 613 studies excluded after titles and abstracts revisions were essentially based on a lack of relationship with the research interests of this review. After full-text revision, only 19 studies which included either validity or reliability of running wearable sensors suppling running PW and/or the specific discussion of such wearable sensors were considered for the current work. Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram. 3.2. Study Characteristics The main characteristics of the studies included in this review (n = 19) are presented in the Tables 1 and 2. Table 1 shows a summary of 12 studies using wearable sensors with the capacity of measuring power during different running exercises. Whereas three of those studies [11,27,28] examine the PW kinetics during different running protocols, the other four studies [15,25,26,29] investigate the relationship between PW and physiological parameters such as oxygen consumption (VO2) at different intensities. Additionally, two further works [30,31] analyse the application of mathematical models, based on power laws, to predict running performance, whereas a recent study [32] assesses the agreement level between two mathematical models and five power meter devices Figure 1. Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram. Sensors 2020, 20, 6482 5 of 20 3.2. Study Characteristics The main characteristics of the studies included in this review (n = 19) are presented in the Tables 1 and 2. Table 1 shows a summary of 12 studies using wearable sensors with the capacity of measuring power during different running exercises. Whereas three of those studies [11,27,28] examine the PW kinetics during different running protocols, the other four studies [15,25,26,29] investigate the relationship between PW and physiological parameters such as oxygen consumption (VO2) at different intensities. Additionally, two further works [30,31] analyse the application of mathematical models, based on power laws, to predict running performance, whereas a recent study [32] assesses the agreement level between two mathematical models and five power meter devices through different running conditions. Other studies examined some parameters provided by the RunScribe power meter to describe the effects of the fatigue induced over a marathon [33,34] and the influence of different types of ankle treatments on running biomechanics [35]. Sensors 2020, 20, 6482 6 of 20 Table 1. Studies (n = 12) involving the use of wearable sensors with the capacity of measuring power during running protocols. References Subject Description Aim System Used Protocol Outcome Measures Results Dobrijevic et al. (2017) [15] 30 physical education students (15 men and 15 women) To explore the properties of the F-V relationship of leg muscles exerting the maximum pulling F at a wide range of V on a standard motorized treadmill Motorized treadmill using externally fixed strain gauge dynamometer (CZL301, ALL4GYM, Serbia) connected to the subject wearing a wide and hard weightlifting belt Walking and running on a treadmill at different velocities (1.4−3.3 m.s−1), and maximum pulling F exerted horizontally were recorded Leg muscle capacities for producing maximum F, V, and power The F-V relationship of leg muscles tested through a wide range of treadmill V could be strong, linear, and reliable. Moreover, the two-velocity method could provide reliable and ecologically valid indices of F, V, and P producing capacities of leg muscles. García-Pinillos et al. (2019) [17] 49 endurance runners To examine how the PW changes while running at a continuous comfortable velocity on a motorized treadmill by comparing running power averaged during different time intervals Stryd system (foot pod) Runners performed a 3 min running protocol at comfortable velocity and P was examined over six recording intervals within the 3-min recording period: 0−10 s, 0−20 s, 0−30 s, 0−60 s, 0−120 s and 0−180 s Running PW P during running is a stable metric with negligible differences, in practical terms, between shorter (i.e., 10, 20, 30, 60 or 120 s) and longer recording intervals (i.e., 180 s) Aubry et al. (2018) [14] 24 male runners (13 recreational, 11 elite) To investigate the applicability of running power (and its individually calculated run mechanics) to be a useful surrogate of metabolic demand (Vo2), across different running surfaces, within different caliber runners. - Stryd system (chest strap) - Gas exchange measures (Cosmed Quark CPET and Cosmed K5 systems) 2 different test at 3 different paces, while wearing a Stryd on both an indoor and an outdoor test: -Treadmill vO2 test: running at 3 speeds for 2 min each -Outdoor vO2 test (on track): identical speeds for 4 min (1 min rest) - Spatiotemporal parameters - Running PW - vO2 Running power (with Stryd) is not a great reflection of the metabolic demand of running in a mixed ability population of runners Snyder et al. (2017) [13] Manuscript clarification: Request for clarification to Aubry et al. (2018) Some major methodological flaws in the mentioned paper are detected. The authors concluded that data analysis and, thereby, data interpretation are misleading Austin et al. (2018) [18] 17 well-trained distance runners To measure the correlations between running economy and P and form power at LT pace. - Stryd system (foot pod) - Gas exchange measures (Parvo Medics TrueOne 2400) Participants ran two 4 min trials: one with a self-selected cadence, and one with a target cadence lowered by 10% - Gas exchange measures - RPE - Power - Form power - SF RE is positively correlated with Stryd’s power and form power measures yet the footpod may not be sufficiently accurate to estimate differences in the running economy of runners García-Pinillos et al. (2019) [36] 18 recreationally-trained male endurance runners To determine if the P-V relationship in endurance runners fits a linear model when running at submaximal velocities, as well as to examine the feasibility of the “two-point method” for estimating P at different velocities Stryd system (foot pod) Incremental running protocol on a treadmill. Initial speed was set at 8 km.h−1, and speed increased by 1 km.h−1 every 3 min until exhaustion PW (W) The two-point method based on distant velocities was able to provide P with the same accuracy than the multiple-point method. Sensors 2020, 20, 6482 7 of 20 Table 1. Cont. References Subject Description Aim System Used Protocol Outcome Measures Results Vandewalle et al. (2018) [21] Data from 6 elite endurance runners - To apply the P-law and logarithmic models and four asymptotic models to the individual performances of the elite runners. - To compare the accuracy of these models. - To compare the predictions of MAS by interpolation and the prediction of maximal running speeds for long distances by extrapolation - The empirical models were compared from the performance of 6 elite endurance runners who participated in international competitions over a large range of distances Mathematical models to predict running performance The predictions of long-distance performances (maximal running speeds for 30, 60 min and marathon) by extrapolation of the logarithmic and power-law models were more accurate than the predictions by extrapolation in all the asymptotic models. Mulligan et al. (2018) [20] Data from various records for a range of distances To develop a novel, minimal and universal model for human running performance that employs a relative metabolic P scale - European and world records performances for eight distances, from 1 km to the marathon, were analyzed Mathematical models to predict running performance The model presented provides a quantitative method for extracting characteristic parameters from race performances of runners. This is the to date most accurate theoretical description of running performances that does not require any a priori fixing of physiological constants Gregory et al. (2019) [25] 12 young adults with history of ankle sprain RunScribe system (foot pod, on the heel) To evaluate the effects of ankle taping, bracing, and fibular reposition taping (FRT) on running biomechanics Four 400 m runs at self-selected pace on an outdoor track. Each run was performed in a different condition (control, taped, braced, FRT) - Spatiotemporal (CT, CycleT, SL) - Kinematic (PR, PRveloc) - Kinetic (impact G, braking G) Ankle taping and bracing were shown to be comparable in decreasing ankle kinematics and kinetics, while FRT caused minimal changes in running biomechanics Leuchanka et al. (2019a) [23] 15 endurance runners To examine the changes in spatiotemporal variables during a marathon race RunScribe system (foot pod, on the lace shoe) Monitoring spatiotemporal variables over a marathon race by comparing 3 points (km 5, 26 and 37) - Spatiotemporal (Pace, CT, SL and cadence) Significant differences were found in pace, SL, and CT when compared across 3 race points Leuchanka et al. (2019b) [24] 15 endurance runners To measure the kinematic asymmetry during a marathon race RunScribe system (foot pod, on the lace shoe) Monitoring kinematic variables over a marathon race by comparing 3 points (km 5, 26 and 37) - Kinematic variables for right and left foot (pace, strike index, PR, PRveloc) Changes in asymmetry were not found to be statistically significant over the marathon. Cerezuela-Espejo et al. (2020) [22] 10 endurance runners To analyse agreement level between power estimated PW by five commercial wearable systems and two theoretical models in different environments and conditions 5 systems: - Stryd App - Stryd Watch - RunScribe (foot pod) - Garmin Running P (watch and chest strap) - Polar Vantage (watch) Three submaximal running protocols on a treadmill (indoor) and an athletic track (outdoor), with changes in speed, body weight, and slope. Running PW derived from the 5 systems and theoretical PW from two mathematical models (TPw1 and TPw2). The closest agreement of the Stryd and PolarV technologies with the TPW1 and TPW2 models suggest these tools as the most sensitive, among those analysed, for PW measurement when changing environments and running conditions CP: critical power; LT: blood lactate thresholds; Vo2max: maximal oxygen uptake; tlim: exhausting time at a given intensity; W´: residual performance capacity; F: force; V: velocity; P: power; Dlim: exhaustion distance; MTT: Montreal Track Test; MAS: maximal aerobic speed; CT: ground contact time; SL: step length; PR: pronation excursion; PRveloc: pronation velocity; TPw1: Mathematical model for power output (PW) estimation 1; TPw2: Mathematical model for PW estimation 2; PW: power output. Sensors 2020, 20, 6482 8 of 20 The closest agreement of the Stryd and PolarV technologies with the TPW1 and TPW2 models suggest these tools as the most sensitive, among those analysed, for PW measurement when changing environments and running conditions Table 2 summarises the studies (n = 7) focused on the validity and reliability analysis of kinetic and kinematic parameters for different wearable sensors with the capacity to measure power. Of note, no studies have examined the concurrent validity of PW during running estimated from any power meter, finding only two studies [12,15] which examined the reliability of PW during running. The remaining 5 studies tested the validity and reliability of spatiotemporal parameters [11,14], kinematic parameters [37,38], or both variables [13]. Table 3 shows the methodological quality of the studies examined. Once the review studies and the letter to editors were excluded, 18 studies were assessed with this purpose. Out of a total score of 14 points, all studies reported from 11 to 14 points. Of note, 16 out of 17 studies reported 0 in the item 12 (i.e., participants prepared to participate representative of entire population) and 14 out of 17 studies reported 0 in the item 23 (i.e., randomised). Sensors 2020, 20, 6482 9 of 20 Table 2. Studies (n = 7) examining the reliability and validity of different wearable sensors with the capacity to measure power during running. References Subject Description Tested System Reference System Protocol Outcome Measures Results García-Pinillos et al. (2018) [16] 18 trained endurance runners Stryd system (foot pod) OptoGait system Incremental running test (8−20 km·h−1 with 3-min stages) on a treadmill - Spatiotemporal parameters (CT, FT, SL, SF) Stryd is reliable for measuring spatiotemporal parameters. It provides accurate SL and SF measures but underestimates CT (0.5−8%) and overestimates FT (3−67%) Koldenhoven et al. (2018) [32] 12 recreational runners RunScribe wearable sensor 3D motion capture system (Vicon system) 2.4 km running protocol on treadmill, at self-selected speed - PR, PRveloc, and CycleT RunScribe showed good to excellent concurrent validity for the outcome measures Brayne et al. (2018) [31] 13 runners Wireless accelerometer (RunScribe): skin mounted Uniaxial piezoresistive accelerometer (model 352C22, PCB Piezotronics): skin mounted Participants ran at 3 different speeds on a treadmill (2.5, 3.5, 4.5 m.s−1) for a total of 40 s (10 s to regulate running gait and 30 s data collection) - Peak tibial acceleration (g) RunScribe accelerometer accurately measures peak tibial accelerations when compared to a research accelerometer, at a range of speeds Hollis et al. (2019) [33] 15 recreational runners RunScribe system (foot pod, on the heel) Intra-system comparison (in different experimental conditions) Two 1600 m runs (slow: 3−4; fast: 5−6 on a 0−10 RPE scale) on two surfaces (track, grass). Randomized order. - Spatiotemporal (CT, CycleT, SL) - Kinematic (PR, PRveloc) - Kinetic (impact G, braking G) RunScribe sensor is valid to identify changes in the outcome measures when participants ran in different conditions. Navalta et al. (2019) [29] 20 young, healthy individuals Stryd system (foot pod) Intra-system reliability Two 5 min self-paced walks along a trail, and two 5 min trail runs (5 min rest period) - Pace and distance - Power: average elapsed power, maximal power, average elapsed form power - Stiffness: average elapsed leg spring - Spatiotemporal: CT - Vertical oscillation Trail running task returns moderate to excellent reliability across all measures García-Pinillos et al. (2019) [30] 49 amateur endurance runners RunScribe system (foot pod) on 2 locations: - Heel shoe - Lace shoe High-speed video analysis at 1000 Hz Treadmill running for 3 min at self-selected comfortable velocity - Spatiotemporal gait parameters (CT, FT, SL, SF) RunScribe is a valid system to measure spatiotemporal parameters during running on a treadmill. The location of the RunScribe plays an important role on the accuracy of spatiotemporal parameters. The lace shoe placement showed smaller errors for CT, FT and SL, whereas the heel shoe was more accurate for SF Cerezuela-Espejo et al. (2020) [19] 12 endurance-trained male athletes 5 systems: - Stryd App - Stryd Watch - RunScribe (foot pod) - Garmin Running P (watch and chest strap) - Polar Vantage (watch) - Metabolic cart (VO2) Participants were initially familiarized with the protocol and then, two protocols were performed in two different settings (outdoor vs. indoor): - Testing 1: Submaximal protocol with incremental speed - Testing 2: Submaximal protocol with incremental body weight A 3rd testing condition was performed only indoor, with increasing slope at submaximal velocity - P output during running The Stryd system is the most repeatable technology, among the five analyzed, for P estimation. The concurrent validity analysis indicated that PW estimated by the Stryd device showed the closest relationship with the VO2 directly measured by the metabolic cart. CT: ground contact time; CycleT: cycle time; SL: step length; PR: pronation excursion; PRveloc: pronation velocity; RPE: rate of perceived exertion; FT: flight time; SF: step frequency; VO2: oxygen uptake; RE: running economy; PW: power output. Sensors 2020, 20, 6482 10 of 20 Table 3. Modified Downs and Black scale [23]. Study Item 1 Item 2 Item 3 Item 6 Item 7 Item 10 Item 12 Item 15 Item 16 Item 18 Item 20 Item 22 Item 23 Item 25 Total (out of 14) Dobrijevic et al. (2017) [15] 1 1 1 1 1 1 1 1 1 1 1 1 0 1 13 Aubry et al. (2018) [14] 1 1 1 1 1 1 0 1 1 1 1 1 0 1 12 Austin et al. (2018) [18] 1 1 1 1 1 1 0 1 1 1 1 1 0 1 12 García-Pinillos et al. (2019) [36] 1 1 1 1 1 1 0 1 1 1 1 1 0 1 12 García-Pinillos et al. (2019) [17] 1 1 1 1 1 1 0 1 1 1 1 1 0 1 12 Vandewalle et al. (2018) [21] 1 1 1 1 1 1 0 U 1 1 1 0 0 1 11 Mulligan et al. (2018) [20] 1 1 1 1 1 1 0 U 1 1 1 0 0 1 11 Gregory et al. (2019) [25] 1 1 1 1 1 1 0 1 1 1 1 1 1 1 13 Leuchanka et al. (2019a) [23] 1 1 1 1 1 1 0 1 1 1 1 1 0 0 11 Leuchanka et al. (2019b) [24] 1 1 1 1 1 1 0 1 1 1 1 1 0 0 11 García-Pinillos et al. (2018) [16] 1 1 1 1 1 1 0 1 1 1 1 1 0 1 12 Koldenhoven et al. (2018) [32] 1 1 1 1 1 1 0 1 1 1 1 1 0 1 12 Brayne et al. (2018) [31] 1 1 1 1 1 1 0 1 1 1 1 1 1 1 13 Hollis et al. (2019) [33] 1 1 1 1 1 1 0 1 1 1 1 1 0 1 12 Navalta et al. (2019) [29] 1 1 1 1 1 1 0 1 1 1 1 1 0 1 12 García-Pinillos et al. (2019) [30] 1 1 1 1 1 1 0 1 1 1 1 1 0 1 12 Cerezuela-Espejo et al. (2020) [19] 1 1 1 1 1 1 0 1 1 1 1 1 1 1 13 Cerezuela-Espejo et al. (2020) [22] 1 1 1 1 1 1 0 1 1 1 1 1 1 1 13 Key: 0 = no; 1 = yes; U = unable to determine. Item 1: clear aim/hypothesis; Item 2: outcome measures clearly described; Item 3: patient characteristics clearly described; Item 6: main findings clearly described; Item 7: measures of random variability provided; Item 10: actual probability values reported; Item 12: participants prepared to participate representative of entire population; Item 15: Blinding of outcome measures; Item 16: analysis completed was planned; Item 18: appropriate statistics; Item 20: valid and reliable outcome measures; Item 22: participants recruited over same period; Item 23: Randomised; Item 25: adjustment made for confounding variables. Sensors 2020, 20, 6482 11 of 20 4. Discussion This review provides a critical assessment on the existing scientific literature regarding PW quantification in endurance running as well as the different current accessible devices for its estimation. After the meticulous analysis described above, a few studies aiming at assessing running power in relation to physiological parameters and power-duration relationship at several running intensities were found. Eighteen studies included in this review were assessed in order to determine the methodological quality and high scores were reported according to the modified Downs and Black scale [23] (i.e., all studies reported more than 11 points out of a total score of 14). Although no studies attempting to assess concurrent validity of PW estimation in running using power meters, their reliability for such estimation was analysed. The controversy surrounding power estimation in running is rooted in the question of whether it is indeed power which is being estimated. Unlike cycling, running entails negligible external mechanical work. It involves positive and negative work; the former, pushing off with each stride and the latter, braking on landing [39]. Moreover, elastic energy stored in the Achilles tendon and other tissues makes a significant contribution as up to fifty percent of power required for each step is released as these tissues stretch upon landing and subsequently recoil to aid pushing off. The issue when estimating power in running is that even perfect estimates do not closely correlate to effort required [39]. During cycling, the relationship between mechanical power and total metabolic energy consumption remains constant when conditions are altered, but this is not so when running [39,40]. Readers need to be aware that given the recent application of power meters to endurance running, the increasing need for PW quantification, and the consequent novelty of this research interest, the limited information available might make the discussion of the current study difficult. However, the subsequent sections seek to provide some insight into how running power quantification can help enhance running performance and its quality. 4.1. Current Evidence on PW during Running While in cycling PW is measured in reference to both direction and quantity of the force applied to the crank, as well as its angular velocity, power needs to be calculated in a different way while running. Since forward and vertical movements of the body account for most of the mechanical work, an accurate calculation of both horizontal and vertical power over the propulsion phase (i.e., a function of forward force and vertical force, respectively) is required to measure running power effectively. Mechanical power on flat terrain might be estimated in mechanical terms just as function of runner anthropometry (height, mass), spatiotemporal parameters (speed, step rate, ground contact time) and wind speed employing model proposed recently by Jenny and Jenny [41]. In steady running on flat surface, mechanical power and the rate of mechanical energy dissipated into heat should match. Considering this assumption and following the mathematical approach mentioned above [41], mechanical energy in steady flat running compiles the energy dissipated by aerodynamic drag, dissipation due to both vertical oscillation and braking. The aerodynamic contribution may be estimated based on air and runner density and running and wind velocity. However, when running on a treadmill wind speed can be considered zero reducing, thus, the importance of this variable. On the one hand, dissipation in vertical oscillation can be estimated regarding step rate, ground contact time, running velocity and a potential energy recovery factor. This factor is variable between subjects and that might be the main concern with this assumption. The lack of considering this factor could lead to overestimation in this part of the mechanical power. On the other hand, dissipation due to braking ground reaction force could be modelled by using the runner’s centre of mass excursion and spring-mass model assumptions. In that context, the power generated in a horizontal direction to maintain running velocity could be estimated by anthropometrics, running speed, and the aforementioned energy recovery factor. The most controversial part of such a model [41] might be the energy recovery factor. Nevertheless, when measuring mechanical power calculations employing ‘gold standard’ methods different Sensors 2020, 20, 6482 12 of 20 assumptions are, done making the assessment of mechanical PW a challenging measure even in the best testing conditions. The critical power (CP) in tasks such as swimming, cycling, and running and its relationship with VO2, blood lactate threshold, and work-exhaustion time was critically reviewed by Vandewalle and colleagues [42]. Theoretically, CP supposes the existence of a particular work-rate that can be held before exhaustion [43]. In this review [42], it is determined that CP matches a steady state during heavy submaximal exercises (i.e., between 6 and 30 min). On the contrary, CP is not a reliable predictor of exhaustion time considering the hyperbolic nature of power-exhaustion time relation [42]. Another review focused on the existing models for residual performance capacity estimation and its application for pacing [16]. The authors examined the quantity of work than can be executed in exercises above CP. Although the review by Vandewalle and colleagues found CP to be a poor predictor of exhaustion time given the power-exhaustion time relation, Jones and Vanhatalo determined that within a range of various exercise intensities (e.g., endurance running), this relationship gives a fundamental basis to proper understand the physiological bases of fatigue development, what may result in an outstanding effect for monitoring both training and athletic performance [16]. The power-duration relationship was also described over a wide range of power intensities [17]. Three different exercise intensities were identified. First, exercise intensity below aerobic threshold (i.e., fatigue appears slowly and it mainly has a central origin) was defined as moderate intensity. Then, intensity over lactate threshold but under CP was referred as heavy intensity (i.e., there is a depletion of muscle glycogen due to central and peripheral fatigue). Finally, severe intensity was identified referring to an intensity above the CP, which relates to gradual muscle metabolic homeostasis alterations and subsequent peripheral fatigue [17]. Literature shows different calculation methods for power-duration relationship such as power law [44,45] and hyperbolic models [46–48], and exponential decay operations [49,50]. Seemingly, hyperbolic calculations of power-duration relation suit best for both reasonable physiological estimations and a proper option to the fundamental data [17] but, the truth is that all these calculations are operationally weak for coaches and extremely time-consuming. In order to counteract the models mentioned above and to provide in-field application for running biomechanics monitoring and training loads tracking to clinicians, coaches and practitioners, wearable technologies were upgraded considerably and made economically affordable. A review study on wearable devices and their provided metrics (i.e., kinetic and kinematic parameters) in the evaluation and treatment of runners identified best practices, applications and potential limitations of such systems [51]. The author stated that clinicians should assure that the use of wearable sensors should be based on evidence aiming at running-related injuries prevention and performance enhancement, and the guidelines given by each sensor’s manufacturer must be followed [51]. Regarding evidence-based use of wearable sensors, the relationship between VO2 as metabolic demand and running PW measured by five commercially available technologies was recently assessed [15]. Twelve endurance-trained male athletes completed 10 submaximal multistage running tests wearing a portable metabolic computer. On two occasions (test-retest), the athletes performed three submaximal treadmill running protocols with manipulations in speed, body weight and slope, and the same protocol was repeated in an athletic track. The Stryd system showed the higher concurrent validity to the VO2 (r ≥ 0.911) between the five wearables, and it was also found as the more repeatable and sensitive in all the conditions studied. Furthermore, the level of agreement between these 5 wearable systems was also analysed against two physics theoretical models for PW estimation [10,52] in different running conditions [32], showing that the Stryd and Polar Vantage systems are the most sensitive tools for PW estimation in running given their close agreement with both theoretical models (r > 0.93). The Stryd power meter estimates power production while running separating this metric into two parts: power and form power. Apparently, power reflects the PW associated with changes in the athlete’s horizontal movement, while form power represents the power production originated by the combination of the oscillatory up and down movements of the centre of mass and lateral power as the athlete moves forward. This system utilises mathematical calculations to estimate these two Sensors 2020, 20, 6482 13 of 20 parameters from kinematic data collected from the described movements executed by the runner’s foot [29]. Form power apparently represents the power production originated by the combination of the oscillatory up and down movements of the centre of mass and lateral power as the athlete moves forward. On the other hand, the power-VO2 relationship in elite and recreational runners had been previously assessed by Aubry and colleagues [26]. To this aim, 13 amateur and 11 elite runners executed a two-setting protocol (i.e., indoor and outdoor). Indoors, participants developed 3 sequential paces (i.e., elite: 14, 16, and 18 km·h−1; amateur: 11−16 km·h−1) 2 min each, where VO2 was analysed via gases expiration system. Outdoors (no precipitations and minimal wind), participants were asked to run at the same pace that they ran indoors. Participants ran for 4 min each pace while measuring VO2 using a portable metabolic computer. Additionally, Stryd was used to calculate running power in both settings. Regarding the relationship metabolic demand-running power, the authors found a significant but weak correlation between VO2 and running power (r = 0.29, p = 0.02). Comparing both settings, metabolic demands were found to be significantly higher (i.e., greater VO2) outdoors (i.e., outdoor track) than when treadmill running. When speed increased, the difference in VO2 values become higher amongst treadmill and outdoor running [26]. Then, after assessing metabolic demand-running mechanics relationship, the authors found moderate strength associations for metabolic demand and ground contact time, vertical oscillation, and step frequency at treadmill running in recreational runners [26]. The authors of the aforementioned study concluded that the use of Stryd power meter should be avoided when assessing running economy as it is unable to distinguish the metabolic demands of an athlete when running on different settings (i.e., outdoors vs. indoor). Of note, the version used during the study is not mentioned (the latest version is even able to consider air resistance) limiting, therefore, their findings. Controversially, Snyder and colleagues clarified several important methodological mistakes made by Aubry and colleagues [26] which led to confusing conclusions [25]. Regarding surface, VO2 was measured long before steady state for treadmill tests (latest VO2 test started at 1:30 min), but much later over ground (latest VO2 test started at 3:30). It is well-known, as stated by Snyder and colleagues, that VO2 needs more than 1:30 min to reach steady state causing, therefore, great differences between VO2 when measured at 1:30 and 3:30 min, and, even greater at faster speeds [25]. The authors claimed that these methodological flaws exclude precise correlation analysis between VO2 and power measured with Stryd on different surfaces [25]. Considering speed, a speed-normalised power to speed-normalised VO2 correlation was reported in the article [26], therefore denying VO2 change because of speed [53]. Snyder and colleagues [25] suggested the use of the accepted physiological term ‘cost of transport’ instead of ‘metabolic demand’, which was used by the authors and leads to confusion in the readers and it does not vary over speed [54]. The actual power-VO2 correlation is proposed to address this error [25]. With respect to subjects, Snyder and colleagues [25] criticise the individual assessment of training metric as they [26] collect data by subject prior executing the correlation analysis when within-subject correlation between VO2 and further variables is appropriate for training and racing [55]. For such study [26], data collection should be developed over different within-subject measurements [25]. Furthermore, the Stryd reliability for PW during treadmill running at a self-selected constant speed with a slope gradient at 0% was proved to be a stable data between short and long intervals (i.e., 10–120 s and 180 s, respectively) [28]. No significant differences were found in the amount of power production between the different spans of times acquired (p = 0.276, partial ETA2 = 0.155) and an almost perfect association in the previously mentioned amount of power production recorded over the intervals (ICC ≥ 0.999). As the authors mentioned, the conditions in which the study was performed may influence the stability of running power over time and these findings should not be taken for granted when transferred to over-ground running [28]. The findings reported here seem to be very advantageous for clinicians and practitioners since, if compared to other physiological parameters such as heart rate or VO2, PW tend to stabilise over time earlier than others traditionally used. However, PW is a mechanical parameter which considers work per time. That work exhibits a muscular and Sensors 2020, 20, 6482 14 of 20 tendinous component. While muscle work needs oxygen consumption to produce work, tendons store and release energy without consuming any oxygen. Therefore, work produced while running requires different quantities of oxygen depending upon the amount of work is done by muscles or tendons. Thus, PW may not be directly related to running metabolic cost. Following the evidence-based use of wearable sensors, it has been found a linear power-velocity relationship(r = 0.999) at submaximal speed, and, the consequent used of the two-point method to predict PW in running at different speeds using the Stryd power meter [36]. The authors executed an incremental run-to-exhaustion protocol on a motorized treadmill at 0% slope gradient. The power-velocity relationship determined from three two-point methods at proximal (10 and 12 km·h−1), intermediate (10 and 14 km·h−1), and distal (10 and 17 km·h−1) speeds showed the same precision than the multiple-point method (used also by the authors to compare PWs through the study) to provide PW estimated by the Stryd power meter. As stated by the authors of the aforementioned study, since the two-point method can be developed faster and without developing fatigue in the athletes, it should be used when assessing PW to acquire accurate power estimations over a range of submaximal running speeds [36]. This might be an outstanding contribution to the strength and conditioning scene as the power-velocity relationship could be frequently updated influencing, therefore, on the quality of both running training and performance. The lack of evidence regarding the power-biomechanics (i.e., contact time, flight time, step frequency, step length, surface) relationship as well as the effect of fatigue on PW when running expose the need of further research on how the running gait parameters and environmental factors affect PW estimation. Bridging the gap between research and practical use of power in running would bring the stunning potential of such parameter to light. The insights provided here into the validity and reliability of the different commercially available wearable sensors for spatiotemporal parameters show the emerging potential of such devices for running PW measurement given their narrow association considering theoretical approaches previously proposed [6–9]. 4.2. Commercially Available Systems to Measure PW during Running Despite the application of IMUs for estimating PW during running being recent, different commercially systems are available. Two of the most widely used wearable sensors for such purposes are Stryd and Runscribe. Stryd system is a pioneer in manufacturing wearable power meters for running. Stryd estimates running power in watts. This power meter, a foot pod reinforced with carbon fibre (weight: 9.1 g) and based on an IMU of 6 different axis (i.e., 3-axis accelerometer and 3-axis gyroscope) and with a sampling rate of 1000 Hz, attaches to the runner’s shoe to estimate metrics for performance quantification (i.e., pace and distance, average elapsed power, maximal power, average elapsed form power, average elapsed leg spring, and average elapsed ground time). Some studies have analysed the reliability of this sensor for both spatiotemporal and PW parameters [11,12,15]. Of note, the latest version of Stryd is capable of estimating the energy expenditure of working against air resistance by measuring the air resistance one faces while running in regards with a white paper located at the manufacturer’s website and where the trials performed to assess the Stryd’s ability to determine wind speed are meticulously described (https://storage.googleapis.com/stryd_static_assets/white_papers/wind-white-paper-8-17. pdf). This sensor employs both kinematic and environmental microelectromechanical sensors together with user-supplied biometrics and proprietary physical and data-driven algorithms to calculate air resistance force as follows: FA = 1 2ρCdAv2 (1) where ρ stands for air density, Cd for drag coefficient, A for the cross-sectional area that encounters the air resistance, and v for the vector of the runner’s relative velocity with local air mass surrounding them. According to the aforementioned white paper, the Stryd system should be centrally located on the laces and towards the toe of the shoe as this placement reported the lowest error regarding wind measurement accuracy (i.e., wind technology is able to correctly report relative air speed under Sensors 2020, 20, 6482 15 of 20 4 km·h−1). However, no peer-reviewed research has been performed to assess the level of accuracy of such device when accounting for air resistance arising therefore the need to evaluate it in the near future. The use of the Runscribe wearable sensor attached to either the lace or heel of the shoes, based on a nine-axis (three-axis magnetometer, accelerometer, and gyroscope, respectively) IMU with an accuracy of 0.002 seconds (sampling rate: 500 Hz), is also widespread around the running world. The way Runscribe estimates power is based on GOVSS model [52] and various assumptions. GOVSS model estimates power using the runner’s speed, step rate, weight, and height, as well as slope gradient and wind velocity based on linear regression models [52]. Several studies attempted to determine the reliability and validity of such foot pods for either kinetic or kinematic parameters [13–15,37,38]. Despite the common use of the Stryd and Runscribe wearable sensors, there are other options for running power estimation commercially available. Cerezuela-Espejo and colleagues [15] also analysed Garmin Running Power (v1.6, Olathe, KS, USA) and Polar Vantage V (firmware 3.1.7, Polar, OY, Kempele, Finland). The Garmin device estimates PW data derived from the combination of a Garmin sport watch and one of the sensors recommended by the manufacturer (i.e., HRM-Run or HRM-Tri heart rate monitor and Running Dynamics Pod on the waist belt). Polar Vantage V estimates power production with no need of an extra sensor (e.g., foot pods). This multisport watch is capable of calculate indirectly several metrics such as average power, maximum power and laps power using the built-in barometer and GPS sensors. Although a positive relation with VO2 was found for both devices (r ≤ 0.841), they exhibited limited test-retest reliability, particularly Garmin Running Power in laboratory settings and Polar Vantage V outdoors. Myotest device, usually fixed onto a belt and fastened and placed level with the navel’s runner (according to manufacturer’s guidelines), provides, amongst others (i.e., cadence, runner’s centre of mass vertical movement, contact time, flight time, step length, stiffness, pace, distance), running PW. Unfortunately, the way Polar, Garmin, and Myotest estimate PW remains unrevealed. Every wearable sensor that provides power metrics employs some form of running power model combined with different assumptions. Therefore, there exist conditions in which such models do not concur until all the different wearable sensors standardise and implement the same model for running PW estimation. 4.3. How Valid and Reliable is PW during Running When Measured by These Devices? Despite the lack of a concurrent validity study where any of the commercially available power meters are compared with the ‘Gold Standard’ to measure running power (i.e., force-plate-instrumented treadmill or a long force platform system), the accuracy of the PW when running provided by these wearable devices might be limited. The variety of available technologies for running gait analysis (e.g., accelerometers, gyroscopes, force plates, pressure plates, and photoelectric cells) implies a variety of devices should exist for analysing stride characteristics. However, some of these devices have not been validated yet. The validity and reliability of a gait analysis system are essential to determine whether results are due to changes in gait pattern or are simply systematic measurement errors. As already mentioned, white (non-peer-reviewed) papers provided by manufacturers to promote the likely potential of their devices, attribute the different values of running power obtained by the different devices to differences in estimating power. Indeed, Myotest attempted to demonstrate validity and repeatability of Myotest App on an Apple watch for PW analysis in comparison with Garmin-Garmin Pod, Polar Vantage V, Stryd (White paper provided by the manufacturer, https://www.myotest.com/technology). A sample of 7 runners executed a 2000 m run protocol with an elevation gain of 22.8 m where 500 m were run on flat ground, 500 m uphill at a constant slope, 500 m of constant-slope downhill, and 500 m on flat ground at a self-selected speed over the entire protocol. It was reported that given the outputs shape and the existence of similar peaks, a correlation between the analysed systems is seemingly demonstrated considering that the different systems are sensitive to elevation changes (i.e., lower power at uphill/downhill shift and higher power at uphill running). Sensors 2020, 20, 6482 16 of 20 Mean-normalised power signal was used to remove the constant shift in the signals, and it was shown that PW measured with Myotest is closer to power measured with Garmin and Stryd. These findings must be taken cautiously as it is well-known that white papers lack the peer-review process. Concerning the reliability of such wearables, a recent study analysed the repeatability different devices (Stryd, Runscribe, Garmin Running Power, and Polar Vantage V) show when measuring power when running as well as their concurrent validity against VO2 [15]. For such a purpose, 12 highly-trained endurance runners executed a submaximal incremental running speed test and a submaximal incremental body weight test in two different settings (i.e., outdoor and indoor). An additional increasing slope gradient at submaximal speed test was executed only indoor. After completion, the authors found Stryd to be the most repeatable device for power estimation. Additionally, Stryd concurrent validity assessment for power estimation was found to show the closest relationship with the VO2max measured directly by metabolic cart [15]. Of note, the authors of this study distinguish between Stryd App and Stryd Watch. Although the Stryd sensor is found to be the same using both app and watch, the variations reported by the authors between these systems is not justified. It might be arguable that the normalisation applied by each system (i.e., Stryd app and watch) differs from one other, but this is not mentioned by the authors. Nevertheless, the findings reported by Cerezuela-Espejo and colleagues [15] constitute a huge contribution providing clinicians, coaches, and practitioners a reliable wearable sensor to quantify running power in training, retraining, and competition. Some of these devices have been used previously for measuring running kinetics (i.e., PW amongst others) and kinematics parameters (i.e., running spatiotemporal gait characteristics). The aforementioned GOVSS model [52] and Jenny’s model [41] for estimating mechanical power rely mainly on runners anthropometry, environmental factors (i.e., air density and wind speed) and running spatiotemporal parameters (i.e., speed, step rate and ground contact time). With this in mind the measurement of spatiotemporal parameters is essential for an accurate power estimation. Regarding this, some studies have shown good reliability of wearable sensors when measuring such parameters [11–15]. García-Pinillos and colleagues [11], over a speed incremental running protocol on a treadmill, tested the reliability of Stryd for running spatiotemporal parameters (i.e., contact time, flight time, step length, and step frequency) against a proved reliable photoelectric cell system for such purpose (i.e., Optogait system) [56]. The authors found that Stryd measures accurately step length and step frequency but underrates slightly contact time overrates flight time in comparison with such system. Likewise, the intra-Stryd reliability has also been analysed [12] over two different 5-min tasks (i.e., two self-paced walks along a trail a and two trail runs separated by a 5-min rest period) with 20 healthy individuals (it was not mentioned whether the participants had any running experience). The authors assessed all the data provided by the Stryd power meter. Regarding trail running, all variables were found to have relative test-retest reliability, meeting the set the intraclass correlation coefficient (ICC) threshold. When considering an interval of confidence equals to 95%, pace, average elapsed power, average elapsed form power, average elapsed leg spring, and vertical oscillation were deemed to have good to excellent reliability; maximal power, average elapsed ground time, and distance were reported to exhibit moderate to excellent reliability [12]. The intra-validity analysis of the Runscribe sensor has also been examined [13,14]. This sensor was used to measure spatiotemporal (i.e., contact time, step length, and cycle time), kinematic (i.e., foot pronation excursion and pronation velocity), and kinetic parameters (i.e., impact ground force and braking ground force) on two different surfaces (i.e., track and grass) at two different running speeds (comfortable self-selected speed and an increased speed) [13]. Over two 1600-m runs, first at a slow pace and then fast on two randomised-ordered surfaces (i.e., track and grass), Runscribe foot pod sensors were found to be valid to determine variations in the aforementioned spatiotemporal, kinetic, and kinematic parameters in different conditions (i.e., different surfaces) [13]. Furthermore, validity measurements regarding the Runscribe placement on the running shoes have also been examined [14]. In this study, the location of the Runscribe on the running shoes (i.e., heel or shoelace) was assessed against a reference technology (i.e., high-speed video camera at 1000 Hz). Sensors 2020, 20, 6482 17 of 20 The authors found Runscribe to be a valid system to examine spatiotemporal variables in treadmill running. Additionally, the location of the Runscribe needs to be considered as it was found to be sensitive to metrics accuracy. When analysing contact time, flight time, and step length, the shoelace placement is recommended as smaller errors were found when comparing to the Runscribe attached to the heel. In contrast, the heel showed higher accuracy when analysing step frequency [14]. In a recent study [15] where test-retest reliability of several wearable sensors was tested, Runscribe was found to be the second most repeatable sensor for speed, slope gradient, and body weight (standard error of measurement ≥ 30.1 W, coefficient variation [CV] ≥ 7.4%, ICC ≤ 0.709), only after the Stryd power meter, for indoor settings. However, when employed in outdoor, Runscribe exhibits both the highest errors and poorest repeatability (SEM ≥ 59.3 W, CV ≥ 14.8%, ICC ≤ 0.563) [15]. When its concurrent validity between PW estimation and VO2 consumption examined over an increasing speed test by a metabolic cart, Runscribe exhibited values of r ≥ 0.582 and standard error of estimate (SEE) ≤ 13.7% for indoor and outdoor settings. Moreover, the power estimation and VO2 agreement was reduced over both conditions (body weight, SEE = 10.3%; slope, SEE = 18.5%). Regarding data collection, it is worth highlighting that the authors did not specify the placement of the Runscribe wearable sensors affecting, as previously discussed, the possible interpretation of the measured outcomes. 5. Conclusions The previous works on running PW and the theoretical approaches provided for its estimation are, from a practical standpoint, hard to include in the everyday routine of an athlete. This study provides a critical evaluation of available scientific information regarding PW quantification in endurance running as well as the different accessible devices for its estimation. The inexistence of studies attempting to evaluate concurrent validity of PW estimation measured by wearable sensors when running (apart from non-peer-reviewed manufacturer’s white papers), the limited available information about the dynamic of PW during running and its short-term response to acute influencing factors (e.g., velocity, slope, fatigue) and long-term training adaptations (i.e., PW as a tool for monitoring training adaptations) made the analysis reported here especially difficult. However, it is arguable that the outcomes stated here are tremendously useful as PW stabilises earlier than other variables commonly used (i.e., heart rate or VO2). Furthermore, running power increases alongside velocity, resembling their linear relationship at different submaximal speeds. Additionally, the reliability of commercially available wearables has been assessed, finding Stryd to be the most reliable and accurate wearable device for running PW estimation. Ultimately, given their novelty and potential application, the analysis of PW while running and its estimation by wearable devices needs more attention from a research perspective in order to provide practitioners a reliable, valid, and friendly tool to improve both training and performance quality in running. Author Contributions: Conceptualization, D.J.-C., A.C.-L. and F.G.-P.; Methodology, D.J.-C., L.E.R.-S. and F.G.-P.; Software, A.C.-L. and L.E.R.-S.; Validation, R.R.-C., A.C.-L; Formal Analysis, D.J.-C., L.ER.-S. and F.G.-P; Investigation, A.C.-L., R.R.-C.; Resources, D.J.-C., L.E.R.-S and A.C.-L.; Data Curation, F.G.-P., D.J.-C., L.E.R.-S. and R.R.-C.; Writing-Original Draft Preparation, D.J.-C., A.C.-L. and F.G.-P.; Writing-Review & Editing, D.J.-C. and F.G.-P.; Visualization, A.C.-L.; Supervision, L.E.R.-S., F.G.-P. and R.R.-C.; Project Administration, L.E.R.-S. and A.C.-L.; Funding Acquisition, D.J.-C. and A.C.-L. All authors have read and agreed to the published version of the manuscript. Funding: The authors declare no funding has been received for this research. Conflicts of Interest: The authors declare no conflict of interest. References 1. Passfield, L.; Hopker, J.G.; Jobson, S.; Friel, D.; Zabala, M. Knowledge is power: Issues of measuring training and performance in cycling. J. Sports Sci. 2017, 35, 1426–1434. 2. Halliday, D.R.; Resnick, R. Fundamentals of Physics; Wiley: New York, NY, USA, 2007. 3. Vance, J. Run with Power: The Complete Guide to Power Meters for Running; VeloPress: Boulder, CO, USA, 2016. Sensors 2020, 20, 6482 18 of 20 4. Paquette, M.R.; Napier, C.; Willy, R.W.; Stellingwerff, T. Moving Beyond Weekly ‘Distance’: Optimizing Quantification of Training Load in Runners. J. Orthop. Sports Phys. Ther. 2020, 1–20. [CrossRef] 5. Kram, R.; Griffin, T.M.; Donelan, J.M.; Chang, Y.H. Force treadmill for measuring vertical and horizontal ground reaction forces. J. Appl. Physiol. 1998, 85, 764–769. 6. Cavanagh, P.R.; Kram, R.J.M.S.S.E. Stride length in distance running: Velocity, body dimensions, and added mass effects. Med. Sci. Sports Exerc. 1989, 21, 467–479. 7. Cavagna, G.; Mantovani, M.; Willems, P.; Musch, G.J.P.A. The resonant step frequency in human running. Pflügers Arch. 1997, 434, 678–684. 8. Clark, K.P.; Ryan, L.J.; Weyand, P.G.J.J. A general relationship links gait mechanics and running ground reaction forces. J. Exp. Biol. 2017, 220, 247–258. 9. Samozino, P.; Rabita, G.; Dorel, S.; Slawinski, J.; Peyrot, N.; Saez de Villarreal, E.; Morin, J.B. A simple method for measuring power, force, velocity properties, and mechanical effectiveness in sprint running. Scand. J. Med. Sci. Sports 2016, 26, 648–658. 10. Van Dijk, H.; Van Megen, R. The Secret of Running: Maximum Performance Gains through Effective Power Metering and Training Analysis; Meyer & Meyer Sport: Aachen, Germany, 2017. 11. Garcia-Pinillos, F.; Roche-Seruendo, L.E.; Marcen-Cinca, N.; Marco-Contreras, L.A.; Latorre-Roman, P.A. Absolute Reliability and Concurrent Validity of the Stryd System for the Assessment of Running Stride Kinematics at Different Velocities. J. Strength Cond. Res. 2018. [CrossRef] 12. Navalta, J.W.; Montes, J.; Bodell, N.G.; Aguilar, C.D.; Radzak, K.; Manning, J.W.; DeBeliso, M. Reliability of Trail Walking and Running Tasks Using the Stryd Power Meter. Int. J. Sports Med. 2019, 40, 498–502. 13. Hollis, C.R.; Koldenhoven, R.M.; Resch, J.E.; Hertel, J. Running biomechanics as measured by wearable sensors: Effects of speed and surface. Sports Biomech. 2019, 1–11. [CrossRef] 14. García-Pinillos, F.; Chicano-Gutiérrez, J.M.; Ruiz-Malagón, E.J.; Roche-Seruendo, L.E. Technology. Influence of RunScribe™ placement on the accuracy of spatiotemporal gait characteristics during running. J. Sports Eng. Technol. 2019, 17543371–19876513. [CrossRef] 15. Cerezuela-Espejo, V.; Hernández-Belmonte, A.; Courel-Ibáñez, J.; Conesa-Ros, E.; Mora-Rodríguez, R.; Pallarés, J.G. Are we ready to measure running power? Repeatability and concurrent validity of five commercial technologies. Eur. J. Sport Sci. 2020, 1–22. [CrossRef] 16. Jones, A.M.; Vanhatalo, A. The ‘critical power’concept: Applications to sports performance with a focus on intermittent high-intensity exercise. Sports Med. 2017, 47, 65–78. 17. Burnley, M.; Jones, A.M. Power–duration relationship: Physiology, fatigue, and the limits of human performance. Eur. J. Sport Sci. 2018, 18, 1–12. 18. Peters, M.D.J.; Godfrey, C.M.; Khalil, H.; McInerney, P.; Parker, D.; Soares, C.B. Guidance for conducting systematic scoping reviews. Int. J. Evid.-Based Healthc. 2015, 13, 141–146. [CrossRef] 19. Munn, Z.; Peters, M.D.J.; Stern, C.; Tufanaru, C.; McArthur, A.; Aromataris, E. Systematic review or scoping review? Guidance for authors when choosing between a systematic or scoping review approach. BMC Med. Res. Methodol. 2018, 18, 143. [CrossRef] 20. Liberati, A.; Altman, D.G.; Tetzlaff, J.; Mulrow, C.; Gøtzsche, P.C.; Ioannidis, J.P.; Clarke, M.; Devereaux, P.J.; Kleijnen, J.; Moher, D. The PRISMA statement for reporting systematic reviews and meta-analyses of studies that evaluate health care interventions: Explanation and elaboration. Ann. Intern. Med. 2009, 151, W-65–W-94. 21. Aromataris, E.; Riitano, D. Constructing a search strategy and searching for evidence. A guide to the literature search for a systematic review. Am. J. Nurs. 2014, 114, 49–56. [CrossRef] 22. Lundh, A.; Gøtzsche, P.C. Recommendations by Cochrane Review Groups for assessment of the risk of bias in studies. Bmc Med Res. Methodol. 2008, 8, 22. 23. Downs, S.H.; Black, N. The feasibility of creating a checklist for the assessment of the methodological quality both of randomised and non-randomised studies of health care interventions. J. Epidemiol. Community Health 1998, 52, 377–384. 24. Landis, J.R.; Koch, G.G. The measurement of observer agreement for categorical data. Biometrics 1977, 33, 159–174. 25. Snyder, K.L.; Mohrman, W.P.; Williamson, J.A.; Li, K. Methodological Flaws in Aubry, RL, Power, GA, and Burr, JF. An Assessment of Running Power as a Training Metric for Elite and Recreational Runners. J. Strength Cond. Res. 2018, 32, e61. Sensors 2020, 20, 6482 19 of 20 26. Aubry, R.L.; Power, G.A.; Burr, J.F. An assessment of running power as a training metric for elite and recreational runners. J. Strength Cond. Res. 2018, 32, 2258–2264. 27. Dobrijevic, S.; Ilic, V.; Djuric, S.; Jaric, S. Force-velocity relationship of leg muscles assessed with motorized treadmill tests: Two-velocity method. Gait Posture 2017, 56, 60–64. 28. García-Pinillos, F.; Soto-Hermoso, V.M.; Latorre-Román, P.Á.; Párraga-Montilla, J.A.; Roche-Seruendo, L.E. How Does Power During Running Change when Measured at Different Time Intervals? Int. J. Sports Med. 2019, 40, 609–613. 29. Austin, C.L.; Hokanson, J.F.; McGinnis, P.M.; Patrick, S. The relationship between running power and running economy in well-trained distance runners. Sports 2018, 6, 142. 30. Mulligan, M.; Adam, G.; Emig, T. A minimal power model for human running performance. PLoS ONE 2018, 13, e0206645. 31. Vandewalle, H. Modelling of Running Performances: Comparisons of Power-Law, Hyperbolic, Logarithmic, and Exponential Models in Elite Endurance Runners. Biomed Res. Int. 2018, 2018, 1–23. [CrossRef] 32. Cerezuela-Espejo, V.; Hernández-Belmonte, A.; Courel-Ibáñez, J.; Conesa-Ros, E.; Martínez-Cava, A.; Pallarés, J.G. Running power meters and theoretical models based on laws of physics: Effects of environments and running conditions. Physiol. Behav. 2020, 112972. [CrossRef] 33. Leuchanka, A.; Switaj, Z.; Clark, T. Exploring kinematic asymmetry by means of wearable sensors during marathon race. Footwear Sci. 2019, 11, S193–S194. 34. Leuchanka, A.; Switaj, Z.; Clark, T. Use of wearable sensors for measurement of spatiotemporal variables during marathon race. Footwear Sci. 2019, 11, S191–S192. 35. Gregory, C.; Koldenhoven, R.M.; Higgins, M.; Hertel, J. External ankle supports alter running biomechanics: A field-based study using wearable sensors. Physiol. Meas. 2019, 40, 044003. 36. García-Pinillos, F.; Latorre-Roman, P.A.; Roche-Seruendo, L.E.; García-Ramos, A. Prediction of power output at different running velocities through the two-point method with the Stryd™ power meter. Gait Posture 2019, 68, 238–243. 37. Brayne, L.; Barnes, A.; Heller, B.; Wheat, J. Using a wireless consumer accelerometer to measure tibial acceleration during running: Agreement with a skin-mounted sensor. Sports Eng. 2018, 21, 487–491. 38. Koldenhoven, R.M.; Hertel, J. Validation of a wearable sensor for measuring running biomechanics. Digit. Biomark. 2018, 2, 74–78. 39. Cavagna, G.A.; Saibene, F.P.; Margaria, R. Mechanical work in running. J. Appl. Physiol. 1964, 19, 249–256. 40. Driss, T.; Vandewalle, H. The measurement of maximal (anaerobic) power output on a cycle ergometer: A critical review. Biomed. Res. Int. 2013, 2013, 589361. [CrossRef] 41. Jenny, D.F.; Jenny, P. On the mechanical power output required for human running—Insight from an analytical model. J. Biomech. 2020, 110, 109948. [CrossRef] 42. Vandewalle, H.; Vautier, J.F.; Kachouri, M.; Lechevalier, J.M.; Monod, H. Work-exhaustion time relationships and the critical power concept. A critical review. J. Sports Med Phys. Fit. 1997, 37, 89–102. 43. Scherrer, J.; Samson, M.; Paleologue, A. Etude du travail musculaire et de la fatigue. 1. Donnes ergometriques obtenues chez lhomme. J. De Physiol. 1954, 46, 887–916. 44. Kennelly, A.E. An approximate law of fatigue in the speeds of racing animals. JSTOR 1906, 42, 275–331. 45. García-Manso, J.; Martín-González, J.; Vaamonde, D.; Da Silva-Grigoletto, M. The limitations of scaling laws in the prediction of performance in endurance events. J. Theor. Biol. 2012, 300, 324–329. 46. Monod, H.; Scherrer, J. The work capacity of a synergic muscular group. Ergonomics 1965, 8, 329–338. 47. Moritani, T.; Nagata, A.; Devries, H.A.; Muro, M. Critical power as a measure of physical work capacity and anaerobic threshold. Ergonomics 1981, 24, 339–350. 48. Morton, R.H. The critical power and related whole-body bioenergetic models. Eur. J. Appl. Physiol. 2006, 96, 339–354. 49. Weyand, P.G.; Lin, J.E.; Bundle, M.W. Sprint performance-duration relationships are set by the fractional duration of external force application. Am. J. Physiol. Regul. Integr. Comp. Physiol. 2006, 290, R758–R765. 50. Wilkie, D. Man as a source of mechanical power. J. Ergon. 1960, 3, 1–8. 51. Willy, R.W. Innovations and pitfalls in the use of wearable devices in the prevention and rehabilitation of running related injuries. J. Phys. Ther. Sport 2018, 29, 26–33. 52. Skiba, P.F. calculation of power output and quantification of training stress in distance runners: The development of the gOVSS algorithm. J Revis. Sep 2006, 16, 9. Sensors 2020, 20, 6482 20 of 20 53. Batliner, M.E.; Kipp, S.; Grabowski, A.M.; Kram, R.; Byrnes, W.C.J.S.m.i.o. Does metabolic rate increase linearly with running speed in all distance runners? Sports Med. Int. Open 2018, 2, E1–E8. 54. Bramble, D.M.; Lieberman, D.E.J.N. Endurance running and the evolution of Homo. Nature 2004, 432, 345–352. 55. Zuccarelli, L.; Porcelli, S.; Rasica, L.; Marzorati, M.; Grassi, B. Comparison between slow components of HR and V· O2 kinetics: Functional significance. Med. Sci. Sports Exerc. 2018, 50, 1649–1657. 56. Jaén-Carrillo, D.; García-Pinillos, F.; Cartón-Llorente, A.; Almenar-Arasanz, A.J.; Bustillo-Pelayo, J.A.; Roche-Seruendo, L.E. Test–retest reliability of the OptoGait system for the analysis of spatiotemporal running gait parameters and lower body stiffness in healthy adults. Proc. Inst. Mech. Eng. Part P: J. Sports Eng. Technol. 2020, 234, 17543371–19898353. Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Mechanical Power in Endurance Running: A Scoping Review on Sensors for Power Output Estimation during Running.

11-13-2020

Jaén-Carrillo, Diego,Roche-Seruendo, Luis E,Cartón-Llorente, Antonio,Ramírez-Campillo, Rodrigo,García-Pinillos, Felipe

eng

PMC8871887




 Citation: Fernández-Galván, L.M.; Prieto-González, P.; Sánchez-Infante, J.; Jiménez-Reyes, P.; Casado, A. The Post-Activation Potentiation Effects on Sprinting Abilities in Junior Tennis Players. Int. J. Environ. Res. Public Health 2022, 19, 2080. https:// doi.org/10.3390/ijerph19042080 Academic Editor: Paul B. Tchounwou Received: 28 December 2021 Accepted: 8 February 2022 Published: 13 February 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). International Journal of Environmental Research and Public Health Article The Post-Activation Potentiation Effects on Sprinting Abilities in Junior Tennis Players Luis Miguel Fernández-Galván 1 , Pablo Prieto-González 2,* , Jorge Sánchez-Infante 3 , Pedro Jiménez-Reyes 4 and Arturo Casado 4 1 Department of Physical Education, Sport, and Human Movement, Autonomous University of Madrid, 28049 Madrid, Spain; luisdepucela@gmail.com 2 Health and Physical Education Department, Prince Sultan University, Riyadh 11586, Saudi Arabia 3 Performance and Sport Rehabilitation Laboratory, Faculty of Sport Sciences, University of Castilla-La Mancha, 45071 Toledo, Spain; jorge.fisio.uclm@gmail.com 4 Center for Sport Studies, Rey Juan Carlos University, 28032 Madrid, Spain; peterjr49@hotmail.com (P.J.-R.); arturo.casado@urjc.es (A.C.) * Correspondence: pprieto@psu.edu.sa; Tel.: +966-114-948-661; Fax: +966-114-548-317 Abstract: Objective: This study aimed to compare the acute effects of a full squat (SQ) or hip thrust (HT) with two different loading intensities (60% and 85% 1 RM) on sprint ability in junior male tennis players. Methods: Nineteen tennis players were included in this research. They underwent four different experimental conditions: HT at 60% 1 RM, HT at 85% 1 RM, SQ at 60% 1 RM, or SQ at 85%. The force–velocity (F–V) profile was used to assess tennis players’ sprint acceleration ability before and after applying the conditioning stimulus. The variables registered were as follows: 5 m test (5 m), 10 m test (10 m), maximum theoretical force (F0), maximum power (Pmax), and the maximal ratio of horizontal-to-resultant force (RFpeak). Results: Significant improvements in 5 m, Pmax, and RFpeak were observed when the conditioning stimulus was performing one set of seven reps of HT at 60% 1 RM. When the activation protocol was one set of seven reps of SQ at 60% 1 RM, significant improvements in 5 m, 10 m, F0, Pmax (N), and RFpeak were detected. Additionally, performing one set of three reps of SQ at 85% 1 RM as an activation protocol provided significant improvements in F0. Conclusion: The use of HT and SQ with a load of 60% 1 RM improved the sprint F–V profile components related to the acceleration phase of the sprint in junior tennis players. Using intensity loads of 85% 1 RM is not adequate to increase acute sprint performance in this population. HT presents a higher transferability to sprinting in the first 5 m of sprinting, whereas SQ provides acute improvements in different sprinting phases. Keywords: post-activation potentiation; tennis; sprinting; acute performance; force-velocity profile 1. Introduction Tennis is a complex racquet sport played by two opposing players or pairs that perform intermittent efforts. In tennis, players compete against one opponent in singles or two opponents in doubles who condition the motor actions of each player [1]. Tennis has significantly evolved during the last decades [2,3]. In addition to the well-known technical and tactical requirements, physical fitness is now also a relevant performance factor [3]. During the effective playing time, among all the technical skills and movements performed by the tennis players, serving, accelerations, and changes of directions are the key performance actions. Therefore, performance in this sport is largely conditioned by power, agility, and speed abilities [2–5]. In youth tennis, the internal and external loads slightly differ with respect to elite tennis [6]. Even so, it has been determined that explosive and ballistic actions are also key performance aspects in youth tennis [7]. Thus, junior tennis players, coaches, and physical trainers should aim to improve strength, power, and sprinting abilities. Additionally, as Int. J. Environ. Res. Public Health 2022, 19, 2080. https://doi.org/10.3390/ijerph19042080 https://www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2022, 19, 2080 2 of 11 young athletes acquire a higher level of sports maturity, physical fitness training becomes increasingly relevant [8]. In this regard, one of the strategies currently implemented to attain acute increases in athletes’ functional performance in explosive and ballistic exercise is post-activation poten- tiation (PAP). PAP has been defined by Tillin and Bishop as the situation in which muscle performance is acutely improved as a result of a prior voluntary contraction [9]. Similarly, Goła´s et al. define PAP as an acute enhancement of performance or an enhancement of factors determining an explosive sports activity following a preload stimulus [10]. PAP enhances sports performance through the following mechanisms: (a) increased number of active motor units (motor unit recruitment), enhanced motor unit synchronization, de- creased presynaptic inhibition, and increased number of nerve impulses transmitted [9,11]; (b) modifications in pennation angle; and (c) phosphorylation of myosin regulatory light chains [9,11]. After receiving the appropriate conditioning stimulus, performance opti- mization depends on the balance between fatigue and potentiation [12]. PAP depends on athletes’ training level, muscle fiber type, muscle contraction type and duration, and stimulus volume and intensity [13]. Thus, it has been verified that well-trained athletes respond better to PAP than recreational athletes. Subjects with a higher percentage of type II muscle fibers also obtain better PAP responses. Likewise, voluntary contractions provide better responses than electrostimulation. Additionally, it must be taken into account that there is a consistent and significant inter-individual variability in the responses obtained by the athletes after the implementation of PAP strategies. For this reason, it is not possible to establish the optimal PAP conditioning protocol for each group of athletes [10,14]. Moreover, substantial differences have been observed in the PAP interventions imple- mented in different studies. This great variability could explain the discrepancies observed in the degree of improvement attained by the athletes. In this respect, conflicting results have been found, including improvements in some studies and no effects or decreases in sports performance in other research [13]. Consequently, further research is required to determine which is the optimal PAP protocol for different groups of athletes and different sports disciplines because each type of exercise induces different effects, and the stimuli applied to young athletes differ from those of professional athletes [13,14]. As a result, it is necessary to clarify the effect of certain aspects that still remain unclear, such as the appropriate conditioning activity intensity, the number of sets to be performed, type of exercise used, and rest period between the conditioning stimulus and the activity [15,16]. The adequate dosage of these parameters has not been standardized in previous research [9]. In this regard, Seitz and Haff indicate that stronger athletes show greater PAP responses with shorter rest periods between the PAP stimulus and the activity, whereas weaker athletes would need a longer rest. They also stated that maximum loads induce better PAP responses in stronger athletes and submaximal loads in weaker athletes [13]. Weaker individuals obtain better PAP effects with multiple sets [13,17]. However, this could increase fatigue [10]. As for the type of exercise included in the conditioning activities, the squat (SQ) is commonly used to improve jumping ability since athletes have to apply strength in the vertical vector [13,17]. The SQ also has a greater range of motion than the hip thrust (HT), which can result in greater muscle tension [18]. However, in the case of running, the hip thrust (HT) exercise (which involves greater activation of the hip extensor muscles) has higher specificity and transferability to activities that require applying strength on a horizontal vector, such as sprinting [19–21]. In tennis, to our knowledge, only one recent study conducted by Terraza-Rebollo and Baiget has analyzed the effects of PAP [22]. This study verified that a PAP intervention did not affect serve velocity and accuracy in young competition tennis players. Likewise, no studies analyzing the influence of PAP on sprint performance in tennis players have been conducted. This ability, as previously mentioned, is a key performance factor in tennis. Additionally, since stronger athletes obtained greater PAP effects than weaker subjects in previous studies, it is necessary to find suitable protocols for the weaker athletes to attain Int. J. Environ. Res. Public Health 2022, 19, 2080 3 of 11 acute performance improvements. Therefore, further research is warranted to verify the effects of PAP on key performance skills and abilities in tennis. Moreover, it is essential to use quality assessment instruments to assess the improve- ments obtained in sprinting. For this purpose, the force–velocity (F–V) profile provides valuable information about the relationship between the force applied by one athlete and the speed at which his or her neuromuscular system generates it in ballistic or explosive movements (i.e., sprinting and running) performed with his or her lower limbs [23,24]. The F–V is calculated by a linear regression over a distance of 30 m [25], and it has proven to be reliable in youth athletes [26]. The profile is composed of different variables, some of them related to the sprint acceleration ability, such as maximum theoretical force (F0), maximum power (Pmax), and the maximal ratio of horizontal-to-resultant force (RFpeak) [23,27]. In this context, the objective of the present study was to compare the acute effects of performing a full SQ and HT using two different loads (60% and 85% 1 RM) on sprinting ability in 19 junior male tennis players. The F–V was used to estimate the potential improvements obtained through the PAP. We hypothesized that both SQ and HT would effectively improve the sprint F–V profile and that more significant improvements would be obtained with loads of 60% 1 RM rather than 85% 1 RM. 2. Materials and Methods 2.1. Participants Nineteen male tennis players of Benicarló Tennis Club (Castellón, Spain), with a minimum of three years of tennis training experience (4.47 ± 1.54) but without resistance training experience participated in the present study. Subjects´ characteristics are shown in Table 1. All study participants underwent an annual medical examination in the health services of the Valencian Tennis Federation, and none of them presented any injury or health condition that could prevent them from participating in this research. Study par- ticipants and their parents or guardians received detailed verbal and written information about the experimental protocol and the potential risks and benefits of participating in it. They were also allowed to withdraw from the study at any stage without penalty. All participants´ parents or guardians gave their written informed consent to be included in this research. The present study was conducted in accordance with the Declaration of Helsinki Ethical Principles. It was also approved by the Institutional Review Board of the Bio-ethics Committee at Prince Sultan University (Riyadh, Saudi Arabia) (ethical clearance number: PSU IRB-2021-02-0070). Table 1. Participants descriptive information. Experimental Group (n = 19) Age (years) 15.61 (1.35) Height (cm) 173.89 (8.24) Weight (kg) 68.31 (13.34) Tennis training experience (years) 4.47 (1.54) Note: Data is presented as mean (standard deviation). 2.2. Procedures Before proceeding with data collection, anthropometric variables were recorded in the laboratory. Height and body mass were measured to the nearest 0.1 cm and 0.1 kg, respec- tively. Both height and body mass were measured with a digital measuring station—Seca 284, Hamburg (Germany). The experimental sessions were carried out always at the same time of the day (between 4.00 and 6.00 p.m.) to avoid the possible effect of circadian rhythm, and also because the participants´ training sessions were usually conducted at that time. The sessions were separated by a minimum of 72 h of rest time to avoid the impact of fatigue on speed test results. The tests were preceded by a 20 min warm-up consisting of seven minutes of jogging at a self-selected pace, eight minutes of dynamic stretching, and five minutes of progressive sprint bouts, with and without change of direction (60%, 70%, Int. J. Environ. Res. Public Health 2022, 19, 2080 4 of 11 and 85% of perceived maximum). The participants performed maximal-effort 30 m linear sprints on a synthetic outdoor track. A smartphone application (My Sprint, Apple Inc., Cupertino, CA, USA) was used to record and analyze the trials’ split times (5, 10, 15, 20, 25, and 30 m). The recording was conducted with an iPhone 7 (iOS 10.0.2), mounted on a tripod, and located 10 m perpendicular to the sprint direction, just in front of the 15 m marker. The system is based on high-speed video analysis (240 frames per second), and it has proven to be valid and reliable to assess linear sprint performance in relation to two different reference systems: timing photocells and radar gun [25]. Subjects started from a crouching position (staggered-stance) with their right hand on the track. The beginning of the sprint was set when the right thumb of the athlete left the ground (this was detected by visual inspection with MySprint). Two independent observers were asked to select the first frame in which participants’ right thumb left the ground (i.e., the start of the sprint) and, subsequently, the frame in which the pelvis was aligned with each of the three different markers for each of the recorded sprints [23]. Split times, participants’ body mass, and height were used by the MySprint app to calculate F0, Pmax, and RFpeak following previously validated formulas [23–25]. 2.3. Familiarization and Maximal Dynamic Strength Test During the four weeks prior to the study commencement, 12 familiarization sessions were carried out to ensure the proper technique in the full SQ and HT exercises. These familiarization sessions consisted of four sets of seven, five, and three repetitions with loads of 60%, 70%, and 85% 1 RM, respectively, for the full SQ and HT exercises. The encoder Speed4Lifts (v2.0., Speed4Lifts, Madrid, Spain) was used to calculate the peak dynamic force, which uses the load–velocity relationship evaluation method, which involves measur- ing concentric velocity with two different weight loads, and then, through linear regression equations, it predicts the load (1 RM percentage) from velocity data [28,29]. All reported repetition velocities in this study correspond to the mean propulsive velocity (MPV) of the concentric phase [29]. The MPV was used in the present study for the following two reasons: (i) it has been proven to have a very high intra-and inter-participant reliability, similar to mean and peak velocity [30]; (ii) regarding the mean values of the propulsive phase (i.e., MPV), when assessing the velocity with which a load is lifted in a concentric action, it avoids underestimating individuals´ neuromuscular ability, especially when lifting light and medium loads [31]. 2.4. Full Squat and Hip Thrust The full SQ was performed starting from the upright position with the knees and hips fully extended. Each participant descended in a continuous movement until his upper thighs were below the horizontal plane and then immediately ascended back to the upright position. Participants were always required to execute the concentric phase of full SQ at maximal velocity. An SQ rack Fitness Line (Collado-Villalba, Spain) and a standard Olympic bar and weight plates (Eleiko, Halmstad, Sweden) were used for all sessions. To perform the HT, subjects were instructed to start by sitting on the ground with their legs flat on the floor, feet shoulder-width apart, and their upper back against a padded exercise bench. Using the same Olympic bar and disks utilized in the previous exercise, the bar was covered with a pad for comfort, and it was placed above the participants´ lower legs, slightly below their knees [32]. Once the subjects positioned the barbell above their pelvis, they assumed the starting position of the exercise by bringing their heels toward the bench and bending their knees. Then, subjects lifted their hips until their knee joints formed a 90◦ angle with their tibias. 2.5. Methodology The activities carried out during the intervention process are shown in Figure 1. In session 1, the 1 RM full SQ test was performed, and one week later, in session 2, the 1 RM HT test was conducted. A random selection of the participants was used. Thus, Int. J. Environ. Res. Public Health 2022, 19, 2080 5 of 11 subjects were assigned to four groups, so that on each day, one group performed one type of exercise with a different load to avoid the learning effect, which could represent a threat to internal validity. All tests were performed at an outdoor facility maintained at standard environmental conditions. To simulate an “active” athletic setting, instead of seating during the rest period, the tennis players were instructed to perform an active recovery with short displacements in different directions at low intensity. Thirty seconds before testing, athletes were notified to be prepared. Session 3 began with the warm-up explained in Section 2.2. Subsequently, the 30 m sprint test was performed. Then, study participants rested for four minutes and performed three repetitions at 85% 1 RM of a full SQ, and after resting for four minutes, they performed the 30 m sprint test again. The session structure used in session 3 was applied in sessions 4, 5, and 6, but using a different conditioning stimulus. Thus, session 4 began 72 h later than session 3, and the activation protocol consisted of performing one set of seven repetitions of a full SQ at 60% of 1 RM. Session 5 started 96 h later, and the conditioning stimulus was one set of three repetitions of an HT at 85% of 1 RM. Finally, session 6 was performed 72 h later, and the activation protocol was one set of seven repetitions of an HT at 60% 1 RM. Moreover, the MySprint application was used to measure the F–V profile when the 30 m tests were performed. The activities carried out during the intervention process are shown in Figure 1. In session 1, the 1 RM full SQ test was performed, and one week later, in session 2, the 1 RM HT test was conducted. A random selection of the participants was used. Thus, subjects were assigned to four groups, so that on each day, one group performed one type of ex- ercise with a different load to avoid the learning effect, which could represent a threat to internal validity. All tests were performed at an outdoor facility maintained at standard environmental conditions. To simulate an “active” athletic setting, instead of seating dur- ing the rest period, the tennis players were instructed to perform an active recovery with short displacements in different directions at low intensity. Thirty seconds before testing, athletes were notified to be prepared. Session 3 began with the warm-up explained in Section 2.2. Subsequently, the 30 m sprint test was performed. Then, study participants rested for four minutes and performed three repetitions at 85% 1 RM of a full SQ, and after resting for four minutes, they performed the 30 m sprint test again. The session structure used in session 3 was applied in sessions 4, 5, and 6, but using a different conditioning stimulus. Thus, session 4 began 72 h later than session 3, and the activation protocol con- sisted of performing one set of seven repetitions of a full SQ at 60% of 1 RM. Session 5 started 96 h later, and the conditioning stimulus was one set of three repetitions of an HT at 85% of 1 RM. Finally, session 6 was performed 72 h later, and the activation protocol was one set of seven repetitions of an HT at 60% 1 RM. Moreover, the MySprint applica- tion was used to measure the F–V profile when the 30 m tests were performed. Figure 1. Description of the intervention process and the activities performed in each session. 2.6. Statistical Analysis Data are presented using the format of the mean SD (standard deviation). The Shapiro–Wilk test was used to contrast the normality of the variables. To determine the consistency between the measurements made in the pre- and post-test, the interclass cor- relation coefficient (ICC) was calculated for all the assessed parameters. ICC values were interpreted as follows: ICC ≤ 0.49, poor; ≥ 0.50 ICC < 0.75, moderate; ≥0.75 ICC < 0.9, good; ICC ≥ 0.9, excellent (Koo and Li, 2016). To verify whether there were differences between groups in the baseline, a one-way ANOVA test was conducted. Furthermore, to assess the effects of PAP on the three different conditions (time: pre- vs. post-test; load: 85% vs. 60%; and exercise: HT vs. SQ), a factorial repeated measures ANOVA (2 × 2 × 2) was conducted. When statistically significant p values were found (interaction effects or significant main effects), a post hoc pairwise comparison was conducted with Sidak correction to identify those differences. The effect size was calculated using the partial eta squared (η2p). Values of η2p = 0.01, η2p = 0.06, and η2p = 0.14 were considered as small, medium, and large effect sizes, respectively [33]. The level of significance established was p = 0.05. The statistical Figure 1. Description of the intervention process and the activities performed in each session. 2.6. Statistical Analysis Data are presented using the format of the mean SD (standard deviation). The Shapiro– Wilk test was used to contrast the normality of the variables. To determine the consistency between the measurements made in the pre- and post-test, the interclass correlation coeffi- cient (ICC) was calculated for all the assessed parameters. ICC values were interpreted as follows: ICC ≤ 0.49, poor; ≥ 0.50 ICC < 0.75, moderate; ≥0.75 ICC < 0.9, good; ICC ≥ 0.9, excellent (Koo and Li, 2016). To verify whether there were differences between groups in the baseline, a one-way ANOVA test was conducted. Furthermore, to assess the effects of PAP on the three different conditions (time: pre- vs. post-test; load: 85% vs. 60%; and exercise: HT vs. SQ), a factorial repeated measures ANOVA (2 × 2 × 2) was conducted. When statistically significant p values were found (interaction effects or significant main effects), a post hoc pairwise comparison was conducted with Sidak correction to identify those differences. The effect size was calculated using the partial eta squared (η2p). Values of η2p = 0.01, η2p = 0.06, and η2p = 0.14 were considered as small, medium, and large effect sizes, respectively [33]. The level of significance established was p = 0.05. The statistical analysis of the data was performed using the program IBM SPSS V.26® computing (IBM Corp., Armonk, NY, USA). Int. J. Environ. Res. Public Health 2022, 19, 2080 6 of 11 3. Results The results obtained by the subjects in the 1 RM tests are shown in Table 2, and the assessed F–V variables are presented in Table 3. The ICC values between test and retest were higher in all cases than 0.9, which reflects excellent reliability. In addition, the one- way ANOVA confirmed the absence of significant differences between the four different measurements conducted at the baseline. Table 2. Results obtained by the study participants in the 1 RM tests. Exercise 1 RM Test Result (kg) Hip thrust 53.11 (15.71) Squat 66.89 (17.16) Table 3. Results obtained by the study participants in the force–velocity variables assessed before (pre-) and after (post-) applying the post-activation potentiation. Moment Results Obtained before Applying the Post-Activation Potentiation 5 m 10 m 30 m F0 (N) Pmax (N) RFpeak Pre-85%1 RM-HIP THRUST 1.62 (0.11) 2.4 3(0.15) 5.45 (0.39) 429.90 (128.82) 791.20 (274.11) 44.50 (4.13) Pre-PAP-60%1 RM-HIP THRUST 1.60 (0.12) 2.41 (0.14) 5.40 (0.37) 441.90 (114.83) 818.20 (245.02) 45.23 (3.90) Pre-PAP-85%1 RM-SQUAT 1.55 (0.12) 2.33 (0.16) 5.23 (0.41) 467.66 (120.11) 895.69 (268.72) 47.16 (4.33) Pre-PAP-60%1 RM-SQUAT 1.60 (0.17) 2.42 (0.20) 5.42 (0.44) 430.47 (132.40) 797.86 (288.61 45.14 (5.65) Results obtained after applying the post-activation potentiation 5 m 10 m 30 m F 0(N) Pmax (N) RFpeak Post-PAP-85%1 RM-HIP THRUST 1.59 (0.11) 2.40 (0.16) 5.42 (0.39) 448.90 (128.36) 824.79 (276.91) 45.44 (4.08) Post-PAP-60%1 RM-HIP THRUST 1.55 (0.11) * 2.36 (0.17) 5.38 (0.45) 475.95 (124.44) 873.73 (285.38) * 46.97 (4.28) * Post-PAP-85%1 RM-SQUAT 1.55 (0.12) 2.32 (0.15) 5.26 (0.39) 479.18 (127.05) * 904.35 (274.73) 47.53 (4.38) Post-PAP-60%1 RM-SQUAT 1.55 (0.15) * 2.35 (0.20) * 5.32 (0.45) * 473.17 (147.74) 885.14 (329.37) * 46.94 (5.37) * 5 m: 5 m split sprint time; 10 m: 10 m split sprint time; 30 m: 30 m split sprint time; F0(N): maximal theoretical velocity; Pmax (N): maximal power; RFPeak (%): maximal ratio of horizontal-to-resultant force; PAP: post-activation potentiation; 1 RM: 1 maximum repetition; *: significant effect found (p < 0.05). The 2 × 2 × 2 ANOVA confirmed the absence of interaction effects for all the assessed variables. However, a main effect of time was found for the following parameters: 5 m (F1–18 = 7.35; p = 0.014; 268 η2p = 0.290), 10 m (F1–18 = 8.62; p = 0.009; η2p = 0.324), Pmax (F1–18 = 7.22; p = 0.015; η2p = 0.286), and RFpeak (F1–18 = 8.36; p = 0.010; η2p = 0.317). A main effect of exercise was also found for F0 (N) (F1–18 = 7.71; p = 0.035; η2p = 0.223). The results of the pairwise comparisons are shown below: 3.1. 5 m Significant differences were observed after applying the PAP load of seven reps at 60% 1 RM in HP (p = 0.018; 95%CI = 0.010 to 0.090), and in SQ (p = 0.006; 95%CI = 0.015 to 0.78). 3.2. 10 m A significant difference was found in the SQ (p = 0.003; 95%CI = 0.023 to 0.101) after applying the PAP load of seven reps at 60% 1 RM. Int. J. Environ. Res. Public Health 2022, 19, 2080 7 of 11 3.3. F0 (N) A significant difference was observed between the HT and SQ in favor of the latter condition (p = 0.024; 95%CI = −0.415 to −0.033) after applying the PAP load of three repetitions at 85% 1 RM. 3.4. Pmax (Wkg-1) Significant differences in the HT (p = 0.013; 95%CI = −98.05 to −13.01) and in SQ (p = 0.037; 95%CI = −205.92 to −7.06) were found after applying the PAP load of seven reps at 60% 1 RM. 3.5. RFpeak Significant differences were observed in HP (p = 0.016; 95%CI = −3.10 to −0.36), and in SQ (p = 0.007; 95%CI = −3.04 to −0.54) after applying the PAP load of seven reps at 60% 1 RM. 4. Discussion The main finding of the present study was that the PAP protocol applied was effective in improving the 5 m time with 60% 1 RM using both an HT and SQ with an ES of 0.43 and 0.31, respectively. The PAP was also effective in improving 10 m times when the SQ was applied with a load of 60% 1 RM (ES = 0.34), confirming our hypothesis that moderate loads are sufficient to produce performance improvements in young athletes without previous strength training experience. PAP has proven to be an effective warm-up to enhance maximal strength and speed of strength development [9,34]. In the present study, significant differences with a small effect size were found when the study participants performed the PAP with loads of 60% 1 RM using both the HT and SQ. These findings indicate that moderate loads may suffice in this population to enhance motor unit recruitment and synchronization, which are the main mechanisms associated with improved performance when applying PAP protocols [9]. The mentioned results were somewhat expected. Indeed, in previous studies, light stimuli ap- plied to young athletes with little or no strength training experience significantly enhanced sprint performance by eliciting lower fatigue [13]. In this regard, a recent meta-analysis that included 32 primary studies showed greater effects (ES = 1.06) with moderate loads (60 to 84% 1 RM) than with high loads (ES = 0.31) (>85% 1 RM) [35] in 141 subjects aged 20 ± 5 years [20]. Likewise, it was observed that an HT is effective in 5 m, but its effect declines in 10 m. We consider that the HT is more effective in improving sprinting ability over shorter distances due to the nature of this exercise because, because unlike SQ, it mainly activates the hip extensor muscles [20]. Some authors also attributed the effective- ness of an HT in the first sprinting meters to the application of strength in the horizontal vector [20,21]. However, Fitzpatrick et al. state that this theory is flawed [36]. They argue that, according to the principle of dynamic correspondence, the forces applied by an athlete must be considered in relation to the coordinate system set by the athlete. While it is true that during accelerations, athletes apply force basically in the horizontal plane, that is because their bodies lean forward, meaning that the direction of the force applied both in accelerations and high-speed running is basically the same [36]. By contrast, SQ was effective both in 5 m, 10 m, and 30 m. In this case, the improvement could be attributed to the enhanced intramuscular coordination by increasing eccentric strength in extensor muscles, which results in a decreased ground contact time and consequently improves their stride frequency (i.e., V0) [19]. Based on the results, we interpret that SQ presents great levels of transferability to all sprinting phases. The study results revealed the absence of significant differences in sprinting perfor- mance after applying the PAP using both the HT and SQ with a load of 85% 1 RM. One possible explanation is that the accumulated fatigue could have overridden the PAP ef- fect [9]. It is plausible that applying heavy loads to subjects without previous strength training experience produces a degree of fatigue that prevents them from optimizing sprint Int. J. Environ. Res. Public Health 2022, 19, 2080 8 of 11 performance. However, it must be taken into account that fatigue can result from using high-intensity conditioning stimuli but could also be due to factors such as excessive vol- ume, short recovery, strenuous warm-up, or personal characteristics of each subject. In this regard, it should be noted that greater benefits have been observed after applying PAP protocols in trained subjects [13,35,37]. Another possible explanation of these results could be that the age of the subjects (15.61 ± 1.35) coincides with the mid-PHV (peak height velocity) [37]. At this stage, the natural growth of the adolescents can lead to poorer training results due to the temporary disruption in basic motor skills caused by the accelerated growth of long bones [38]. This stage is known as adolescent awkwardness [39] and impacts neuromuscular function and physical performance [38]. In the present study, the deep SQ was performed as it generates greater motor unit activation and synchronization [13]. In this regard, the study results also show that the subjects improved the F0 and Pmax variables, which coincides with other studies where a deep SQ was also performed [14,17]. However, Seitz et.al. [13] obtained greater performance improvements using a half SQ (ES = 0.58) than full SQ (ES = 0.25). This discrepancy suggests that a deep SQ produces higher levels of acute fatigue [19] and therefore PAP become less effective. Even so, certain discrepancies have been found in this regard in the scientific literature since some authors did not observe significant differences in sprint performance after using either a half SQ or full SQ [20–22]. Therefore, it is not possible to determine the effect of PAP by performing a half SQ instead of full SQ. In this sense, it should be noted that the acute performance improvements that can be obtained depending on SQ depth vary according to the athlete’s level. Stronger subjects perform better with shallow SQs, whereas subjects with lower strength levels obtain greater improvements with deep SQs [3]. For this reason, the full SQ exercise was selected in the training protocol of the current research. As for the F–V profile components, a significant increase in F0 after performing a SQ with a load of 85% was observed. Additionally, the Pmax and RFpeak were also significantly increased after performing a SQ and HT with a load of 60% 1 RM. Improving the last two variables is crucial, since acceleration in short distances is a key performance factor in tennis. These results again suggest that both HT and SQ are effective in improving various F–V profile components, and the most appropriate intensity for both exercises is 60%, rather than 85%. Therefore, this reinforces the idea that in youth athletes without previous strength training experience, moderate loads should be used in the PAP protocols. Importantly, Seitz et al. [13] verified that resting time duration used after the condition- ing activity should be set depending on the subject’s maximal strength. Thus, the PAP effect is greater in stronger subjects when short recovery times are used (i.e., 2–3 min), whereas longer rest intervals produce greater improvements in individuals with lower strength levels. This is because of the ratio of type II muscle fibers [40,41], which in turn is associated with a higher myosin light chain phosphorylation and represents the peripheral factor on which the PAP effects are based [42]. In our study, the resting time was four minutes. This selection is justified based on the results of previous studies [43–45]. However, it cannot be ruled out that a more extended recovery period could have produced lower fatigue to study participants, and consequently, they could have obtained better results. In this sense, Wilson et al. [35], after conducting a meta-analysis, verified that long rest periods (i.e., 7 to 10 m) could be more effective than short periods (3 to 7 min) (ES = 0.54 vs. 0.14). Finally, only one set of HTs or SQs was undertaken in the conditioning protocol. However, some studies found greater effects of PAP when more than one set was performed (ES = 0.69 vs. ES = 0.24) [13,35,37], particularly in stronger athletes. Therefore, it is also possible that further PAP enhancements could have been attained if more than one set had been performed. However, it should also be considered that the study participants had no previous strength training experience. That means that performing more sets could have caused them great fatigue, as Wilson et al. state [35]. This study has some limitations that must be mentioned. The muscle activity was not measured. Thus, the results and conclusions are exclusively based on the PAP effects. Int. J. Environ. Res. Public Health 2022, 19, 2080 9 of 11 Moreover, we must consider that, since the study participants knew the study’s objective, the occurrence of a placebo effect cannot be excluded. Future research should include control groups and verify the effect of performing full vs. half SQs and the results of using a different number of sets and different rest periods between the conditioning activity and the task. 5. Conclusions The PAP protocols applied to junior tennis players without previous strength training experience effectively improved the F–V profile when the loading intensity used in the conditioning activity was 60% 1 RM and the exercises performed were either an HT or SQ. However, 85% 1 RM loads were not adequate to increase acute sprinting performance in this population group. In addition, HT presented a higher level of transferability in the first 5 m of sprinting, whereas SQ provided acute improvements in different sprinting phases. Practical Applications Both the HT and the SQ are used in sports training to improve sprinting speed or other sport skills and fitness components. The present study verified that both exercises could be useful in PAP protocols aiming to enhance sprinting ability. Practitioners and trainers can use them as a suitable PAP stimulus to induce acute effects on subsequent ballistic or explosive activities. However, the greatest effect occurs when moderate loads are applied in youth tennis players without previous strength training experience (60% 1 RM). Therefore, heavy loads may reduce their adaptative reserve prematurely and limit future performance improvements. Author Contributions: Conceptualization, A.C., P.J.-R. and L.M.F.-G.; Methodology, A.C. and L.M.F.-G.; Soft-ware, L.M.F.-G.; Validation, L.M.F.-G. and A.C.; Investigation, L.M.F.-G. and J.S.-I.; Resources, L.M.F.-G.; Data Curation, L.M.F.-G. and P.P.-G.; Writing—Original Draft Preparation, L.M.F.-G. and P.P.-G.; Writing—Review and Editing, J.S.-I.; Supervision, J.S.-I., P.J.-R. and A.C.; Fund- ing Acquisition, P.P.-G. All authors have read and agreed to the published version of the manuscript. Funding: The authors would like to recognize the efforts made by Prince Sultan University (Riyadh, Saudi Arabia), for its support in funding the research either with fees, incentives, or seed grants. Institutional Review Board Statement: The present study was conducted according to the principles set out in the Helsinki Declaration, and it was also approved by the Institutional Review Board of the Bioethics Committee at Prince Sultan University (Riyadh, Saudi Arabia) (approval no. PSU IRB-2021-02-0070). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: Not applicable. Conflicts of Interest: The authors declare no conflict of interest. References 1. Crespo, M. Estructura Funcional del Tenis; Conferencia presentada al Congreso de la Asociación de Profesores de Tenis: Guatemala, Guatemala, 2009. 2. Fernández-Fernández, J.; Méndez-Villanueva, A.; Pluim, B.M.; Fernández-García, B.; Terrados, N. Physical and physiological aspects of tennis competition (I). Arch. Med. Deporte 2006, 23, 451–454. 3. Fernandez-Fernandez, J.; Sanz-Rivas, D.; Mendez-Villanueva, A. A Review of the Activity Profile and Physiological Demands of Tennis Match Play. Strength Cond. J. 2009, 31, 15–26. [CrossRef] 4. Kovacs, M.S. Tennis physiology: Training the competitive athlete. Sports Med. 2007, 37, 189–198. [CrossRef] 5. Williams, J.A. Effect of specific strength and power training on serving velocity in tennis players. JASC 2020, 28, 80–92. 6. Torres-Luque, G.; Sanchez-Pay, A.; Moya, M. Competitive analysis of requirement of young tennis players. J. Sport Health Res. 2011, 3, 71–78. 7. Ulbricht, A.; Fernandez-Fernandez, J.; Mendez-Villanueva, A.; Ferrauti, A. Impact of Fitness Characteristics on Tennis Performance in Elite Junior Tennis Players. J. Strength Cond Res. 2016, 30, 989–998. [CrossRef] Int. J. Environ. Res. Public Health 2022, 19, 2080 10 of 11 8. Granacher, U.; Lesinski, M.; Büsch, D.; Muehlbauer, T.; Prieske, O.; Puta, C.; Gollhofer, A.; Behm, D.G. Effects of Resistance Training in Youth Athletes on Muscular Fitness and Athletic Performance: A Conceptual Model for Long-Term Athlete Development. Front. Physiol. 2016, 7, 164. [CrossRef] 9. Tillin, N.A.; Bishop, D. Factors modulating post-activation potentiation and its effect on performance of subsequent explosive activities. Sports Med. 2009, 39, 147–166. [CrossRef] 10. Goła´s, A.; Maszczyk, A.; Zajac, A.; Mikołajec, K.; Stastny, P. Optimizing post activation potentiation for explosive activities in competitive sports. J. Hum. Kinet. 2016, 52, 95–106. [CrossRef] 11. Aagaard, P. Training-induced changes in neural function. Exerc. Sport Sci. Rev. 2003, 31, 61–67. [CrossRef] 12. Boullosa, D.; Del Rosso, S.; Behm, D.G.; Foster, C. Post-activation potentiation (PAP) in endurance sports: A review. Eur. J. Sport Sci. 2018, 18, 595–610. [CrossRef] [PubMed] 13. Seitz, L.B.; Haff, G.G. Factors Modulating Post-Activation Potentiation of Jump, Sprint, Throw, and Upper-Body Ballistic Performances: A Systematic Review with Meta-Analysis. Sports Med. 2016, 46, 231–240. [CrossRef] [PubMed] 14. Kobal, R.; Pereira, L.A.; Kitamura, K.; Paulo, A.C.; Ramos, H.A.; Carmo, E.C.; Roschel, H.; Tricoli, V.; Bishop, C.; Loturco, I. Post-Activation Potentiation: Is there an Optimal Training Volume and Intensity to Induce Improvements in Vertical Jump Ability in Highly-Trained Subjects? J. Hum. Kinet. 2019, 66, 195–203. [CrossRef] [PubMed] 15. Hughes, D.C.; Ellefsen, S.; Baar, K. Adaptations to Endurance and Strength Training. Cold Spring Harb. Perspect Med. 2018, 8, a029769. [CrossRef] [PubMed] 16. Zauner, C.W.; Maksud, M.G.; Melichna, J. Physiological Considerations in Training Young Athletes. Sports Med. 1989, 8, 15. [CrossRef] [PubMed] 17. Weber, K.R.; Brown, L.E.; Coburn, J.W.; Zinder, S.M. Acute effects of heavy-load squats on consecutive squat jump performance. J. Strength Cond. Res. 2008, 22, 726–730. [CrossRef] [PubMed] 18. Barbalho, M.; Coswig, V.; Souza, D.; Serrão, J.C.; Hebling-Campos, M.; Gentil, P. Back Squat vs. Hip Thrust Resistance-training Programs in Well-trained Women. Int. J. Sports Med. 2020, 41, 306–310. [CrossRef] 19. Cronin, J.; Hansen, K.; Kawamori, N.; McNair, P. Effects of weighted vests and sled towing on sprint kinematics. Sports Biomech. 2008, 7, 160–172. [CrossRef] [PubMed] 20. Neto, W.K.; Vieira, T.L.; Gama, E.F. Barbell hip thrust, muscular activation and performance: A systematic review. J. Sport. Sci. Med. 2019, 18, 198–206. 21. Carbone, L.; Garzón, M.; Chulvi-Medrano, I.; Bonilla, D.A.; Alonso, D.A.; Benítez-Porres, J.; Petro, J.L.; Vargas-Molina, S. Effects of heavy barbell hip thrust vs back squat on subsequent sprint performance in rugby players. Biol Sport. 2020, 37, 325–331. [CrossRef] 22. Terraza-Rebollo, M.; Baiget, E. Effects of Postactivation Potentiation on Tennis Serve Velocity and Accuracy. Int. J. Sports Physiol. Perform. 2019, 1–6. [CrossRef] [PubMed] 23. Samozino, P.; Edouard, P.; Sangnier, S.; Brughelli, M.; Gimenez, P.; Morin, J.B. Force-velocity profile: Imbalance determination and effect on lower limb ballistic performance. Int. J. Sports Med. 2014, 35, 505–510. [CrossRef] [PubMed] 24. Jiménez-Reyes, P.; Samozino, P.; García-Ramos, A.; Cuadrado-Peñafiel, V.; Brughelli, M.; Morin, J.B. Relationship between vertical and horizontal force-velocity-power profiles in various sports and levels of practice. PeerJ 2018, 6e, 5937. [CrossRef] [PubMed] 25. Romero-Franco, N.; Jiménez-Reyes, P.; Castaño-Zambudio, A.; Capelo-Ramírez, F.; Rodríguez-Juan, J.J.; González-Hernández, J.; Toscano-Bendala, F.J.; Cuadrado-Peñafiel, V.; Balsalobre-Fernández, C. Sprint performance and mechanical outputs computed with an iPhone app: Comparison with existing reference methods. Eur. J. Sport Sci. 2017, 17, 386–392. [CrossRef] [PubMed] 26. Runacres, A.; Mackintosh, K.A.; McNarry, M.A. The effect of constant-intensity endurance training and high-intensity interval training on aerobic and anaerobic parameters in youth. J. Sports Sci. 2019, 37, 2492–2498. [CrossRef] [PubMed] 27. Morin, J.B.; Edouard, P.; Samozino, P. Technical ability of force application as a determinant factor of sprint performance. Med. Sci. Sports Exerc. 2011, 43, 1680–1688. [CrossRef] 28. Bosquet, L.; Porta-Benache, J.; Blais, J. Validity of a Commercial Linear Encoder to Estimate Bench Press 1 RM from the Force-Velocity Relationship. J. Sports Sci. Med. 2010, 9, 459–463. 29. González-Badillo, J.J.; Marques, M.C.; Sánchez-Medina, L. The importance of movement velocity as a measure to control resistance training intensity. J. Hum. Kinet. 2011, 29A, 15–19. [CrossRef] 30. Banyard, H.G.; Nosaka, K.; Vernon, A.D.; Haff, G.G. The Reliability of Individualized Load-Velocity Profiles. Int. J. Sports Physiol. Perform. 2018, 13, 763–769. [CrossRef] 31. Sanchez-Medina, L.; Perez, C.E.; Gonzalez-Badillo, J.J. Importance of the propulsive phase in strength assessment. Int. J. Sports Med. 2010, 31, 7. [CrossRef] 32. Orjalo, A.J.; Callaghan, S.J.; Lockie, R.G. The Effects of the Barbell Hip Thrust on Post-Activation Performance Enhancement of Change of Direction Speed in College-Aged Men and Women. Sports 2020, 12, 151. [CrossRef] [PubMed] 33. Cohen, J. The statistical power of abnormal-social psychological research: A review. J. Abnorm. Psychol. 1988, 65, 9. [CrossRef] [PubMed] 34. Hernández-Davó, J.L.; Loturco, I.; Pereira, L.A.; Cesari, R.; Pratdesaba, J.; Madruga-Parera, M.; Sanz-Rivas, D.; Fernández- Fernández, J. Relationship between Sprint, Change of Direction, Jump, and Hexagon Test Performance in Young Tennis Players. J. Sports Sci. Med. 2021, 20, 197–203. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2022, 19, 2080 11 of 11 35. Wilson, J.M.; Duncan, N.M.; Marin, P.J.; Brown, L.E.; Loenneke, J.P.; Wilson, S.M.C.; Jo, E.; Lowery, R.P.; Ugrinowitsch, C. Meta-Analysis of Postactivation Potentiation and Power. J. Strength Cond. Res. 2013, 27, 854–859. [CrossRef] [PubMed] 36. Fitzpatrick, D.A.; Cimadoro, G.; Cleather, D.J. The Magical Horizontal Force Muscle? A Preliminary Study Examining the "Force-Vector" Theory. Sports 2019, 7, 30. [CrossRef] 37. Lesinski, M.; Muehlbauer, T.; Büsch, D.; Granacher, U. Acute effects of postactivation potentiation on strength and speed performance in athletes. Sportverletz. Sportschaden Organ Der Ges. Fur Orthop.-Traumatol. Sportmed. 2013, 27, 147–155. 38. Moran, J.; Parry, D.A.; Lewis, I.; Collison, J.; Rumpf, M.C.; Sandercock, G.R.H. Maturation-related adaptations in running speed in response to sprint training in youth soccer players. J. Sci. Med. Sport 2018, 21, 538–542. [CrossRef] 39. Lloyd, R.S.; Oliver, J.L.; Faigenbaum, A.D.; Myer, G.D.; De Ste Croix, M.B. Chronological age vs. biological maturation: Implications for exercise programming in youth. J. Strength Cond. Res. 2014, 28, 1454–1464. [CrossRef] 40. Maughan, R.J.; Watson, J.S.; Weir, J. Relationships between muscle strength and muscle cross-sectional area in male sprinters and endurance runners. Eur. J. Appl. Physiol. Occup. Physiol. 1983, 50, 309–318. [CrossRef] 41. Aagaard, P.; Andersen, J.L. Correlation between contractile strength and myosin heavy chain isoform composition in human skeletal muscle. Med. Sci. Sports Exerc. 1998, 30, 1217–1222. [CrossRef] 42. Grange, R.W.; Vandenboom, R.; Houston, M.E. Physiological significance of myosin phosphorylation in skeletal muscle. Can. J. Appl. Physiol. 1993, 18, 229–242. [CrossRef] [PubMed] 43. Xenofondos, A.; Laparidis, K.; Kyranoudis, A.; Galazoulas, C.; Bassa, E.; Kotzamanidis, C. Post-activation potentiation: Factors affecting it and the effect on performance. J. Phys. Educ. Sport. 2010, 28, 32–38. 44. Nibali, M.L.; Chapman, D.W.; Robergs, R.A.; Drinkwater, E.J. Considerations for determining the time course of post-activation potentiation. Appl. Physiol. Nutr. Metab. 2015, 40, 1163–1170. [CrossRef] [PubMed] 45. Linder, E.E.; Prins, J.H.; Murata, N.M.; Derenne, C.; Morgan, C.F.; Solomon, J.R. Effects of preload 4 repetition maximum on 100-m sprint times in collegiate women. J. Strength Cond. Res. 2010, 24, 1184–1190. [CrossRef]

The Post-Activation Potentiation Effects on Sprinting Abilities in Junior Tennis Players.

02-13-2022

Fernández-Galván, Luis Miguel,Prieto-González, Pablo,Sánchez-Infante, Jorge,Jiménez-Reyes, Pedro,Casado, Arturo

eng

PMC9377456

TYPE Original Research PUBLISHED 01 August 2022 DOI 10.3389/fpubh.2022.966578 OPEN ACCESS EDITED BY Juel Jarani, Sports University of Tirana, Albania REVIEWED BY Mücahit Fi¸sne, Sivas Cumhuriyet University, Turkey Ugur Ödek, Nev¸sehir Haci Bekta¸s Veli University, Turkey *CORRESPONDENCE Marko Joksimovi´c nicifor007@outlook.com SPECIALTY SECTION This article was submitted to Aging and Public Health, a section of the journal Frontiers in Public Health RECEIVED 11 June 2022 ACCEPTED 11 July 2022 PUBLISHED 01 August 2022 CITATION Goranovi´c K, Hadži´c R, Petkovi´c J and Joksimovi´c M (2022) Exploring trends of running performance during matches of professional soccer players in Montenegro: A longitudinal study. Front. Public Health 10:966578. doi: 10.3389/fpubh.2022.966578 COPYRIGHT © 2022 Goranovi´c, Hadži´c, Petkovi´c and Joksimovi´c. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Exploring trends of running performance during matches of professional soccer players in Montenegro: A longitudinal study Kosta Goranovi´c1, Rašid Hadži´c1, Jovica Petkovi´c1 and Marko Joksimovi´c1,2* 1Department of Physical Education, Faculty of Sports and Physical Education, University of Montenegro, Podgorica, Montenegro, 2Institute of Sports and Sports Medicine, Podgorica, Montenegro The practical value of monitoring is that well-chosen performance indicators can help coaches identify the good and bad performance of individuals or teams. External monitoring of matches is useful in establishing the physiological requirements of the sport and assessing how a player compares to the requirements of the event in this regard. This study aimed to analyze the trend component of running performance during a game of professional soccer in Montenegro. The research included a sample of 82 professional soccer players. The first subsample included 44 professional soccer players of the club Budu´cnost from Podgorica, height 185.89 ± 6.29 cm, mass 81.06 ± 5.47 kg, BMI 23.47 ± 0.96 kg/m², age 28.86 ± 3.85 yrs. The second subsample included 38 professional soccer players from the Sutjeska club from Nikši´c, height 181.88 ± 6.35 cm, mass 77.28 ± 6.78 kg, BMI 23.32 ± 1.08 kg/m², age 29.43 ± 5.68 yrs. The InStat kinematic system captured the outfield players by using six cameras placed around the perimeter of the field at the minimal height of 12 m. The frame frequency was 25 frames per second; data were centralized for further analysis. Statistically significant diferences were noted only in the variable sprint distance in the 2017 season. The results of the current research indicate that the soccer players who compete in Montenegro are below the values achieved by those who compete in Europe. KEYWORDS performance analysis, external monitoring, time-motion analysis, high intensity running, soccer Introduction Soccer is one of the most complex sports in the world; players need technical, tactical, and physical skills to achieve successful performance and eventually win a game. The cooperative relationships between players who play different positions are critical to a team’s success. For instance, the main role of midfielders is to organize the offense with proper ball control and passes, while the main duties of defenders are to win aerial duels Frontiers in Public Health 01 frontiersin.org Goranovi´c et al. 10.3389/fpubh.2022.966578 and tackles or to perform interceptions of the balls passed to attackers. Understanding these position-specific demands is crucial in the evaluation of players’ achievements (1). Modern soccer requires a high level of endurance, speed, strength, and coordination (2). Therefore, players must have well-developed physical fitness. Given that the energy used by soccer players is mainly produced by aerobic metabolism (3, 4), it is essential that players have well-developed aerobic fitness. Running in-game performance is a set of variables used in soccer performance analysis and is defined “as the choice and combination of variables that define an aspect of performance and help achieve sporting success” (5), in which the player’s duties are passing, shooting, throwing the ball, dribbling, etc. Currently, several video-based platforms are available to track player performance indicators; some of the most commonly used platforms are InStat, Optasport, and Wyscout. Such platforms quickly and accurately provide a wide range of data on game performance indicators, enabling simultaneous analysis of physical effort, movement patterns, and technical actions of players, with and without the ball (6). Various studies have examined these characteristics and requirements within a soccer team (7). Yi et al. (8) explored the technical requirements of different playing positions for play in the UEFA Champions League. In contrast, Modri´c et al. (6) identified running performance specific to each playing position in professional soccer players. Dellal et al. (9) identified positional requirements from technical and physical aspects in the French premier league. All studies indicated high applicability of running performance in evaluating team-specific achievements and team position. It is known that running performance during the game is an essential determinant of success in professional soccer, which has been studied repeatedly, although some studies have been done with different aims (10–12). However, due to its importance, more research is required in different countries according to different levels of players and leagues. This is the first study to monitor the performance of running during the game in the first Telekom Montenegrin league. In this study, we hypothesized that examining the differences in the variables mentioned in different matches could provide a useful, practical report to coaches and trainers in Montenegro. Therefore, this study aimed to explore trends of running performance during the match in professional soccer players in Montenegro in three competitive matches of different seasons. Materials and methods Participants The research included a sample of 82 professional soccer players. The first subsample included 44 professional soccer players of the soccer club Budu´cnost from Podgorica, height 185.89 ± 6.29 cm, mass 81.06 ± 5.47 kg, BMI 23.47 ± 0.96 kg/m², age 28.86 ± 3.85 yrs. The second subsample included 38 professional soccer players from the Sutjeska soccer club from Nikši´c, height 181.88 ± 6.35 cm, mass 77.28 ± 6.78 kg, BMI 23.32 ± 1.08 kg/m², age 29.43 ± 5.68 yrs. All soccer players compete in the first Telekom Montenegrin league, the highest competitive rank in Montenegro. The study is longitudinal in nature, and testing was done in three seasons: 2014/2015, 2016/2017, and 2019/2020, where derby matches between Budu´cnost and Sutjeska were observed each season. The criteria for inclusion were that the first team’s players had been team members for at least 6 months, that all the players went through the preparation period with the team, were without injuries in the previous 6 months, and that they played one half-season before testing. Exclusion criteria were athletes in the recovery phase from some form of acute or chronic injury and athletes who did not complete the entire preparation period. All respondents were first informed about the study and the purpose and goal of the research; the possible consequences were explained to them. Also, the procedure and the course of the testing itself were explained to the respondents. Prior to the survey, each respondent signed a consent form to participate. For this research, the consent and approval of the head coach and the club president were obtained, and testing was started. The research was in accordance with the Declaration of Helsinki (13). Study design InStat Kinematic System—“Currently, various video-based systems track performance indicators of soccer players (InStat, Optasport, Wyscout). Such platforms quickly and accurately provide a large range of match-related performance measures, allowing the simultaneous analysis of the physical efforts, movement patterns, and technical actions of players, both with and without the ball” (6). “The match performance indicators for each player were determined by the position-specific InStat system. The InStat tracking system was previously employed to analyze the association between running performance and game performance indicators in professional soccer players” (6). “The InStat kinematic system captured the outfield players using six cameras placed around the perimeter of the field at the minimal height of 12 m. The frame frequency was 25 frames per second; data were centralized for further analysis. InStat Autocrop allows filming matches without a cameraman. The footage covered every player on the field. There is minimum human involvement in the process; a person is only needed to set up a panoramic camera at the required height, connect it to a computer, and check the Internet connection before the start of the match. An Autocrop camera is set at a height of 8–10 meters and 23–24 meters away from the sideline. A special algorithm allows the camera to cover the entire field. The Frontiers in Public Health 02 frontiersin.org Goranovi´c et al. 10.3389/fpubh.2022.966578 program analyzes every frame and centers the image depending on the players’ positions, without any sudden zooming. The following parameters of running performance were selected to estimate the match performance of players: total distance covered per match and during each half (m), the average speed per match and during each half (km/h), maximal speed (km/h); the total distance covered at high-intensity (m) (speed range 19.8–25.2 km/h) per match and for each half, the total distance covered sprinting (m) (speed above 25.2 km/h) per match and for each half, and the number of sprints. The speed thresholds for each category are similar to those reported previously” (6) and have been universally accepted. Statistical analysis All data collected by the survey were processed using descriptive and comparative statistics. Regarding descriptive statistics, mean and standard deviation were measured for each variable. Regarding comparative statistics, a discriminant parametric procedure was used: analysis of variance with one-factor Anova and Post Hoc, which determined the differences in running performance every year separately. The statistical program for personal computers SPSS for Windows version 20.0 was used for data processing. TABLE 1 Descriptive data of performance running. Variables Team 2015 2017 2020 F Sig. Mean ± SD Mean ± SD Mean ± SD TD (m) Budu´cnost 8.274 ± 3.87 8.041 ± 3.40 9.129 ± 2.46 0.760 0.541 Sutjeska 9.441 ± 3.11 7.081 ± 3.12 7.019 ± 3.45 WD (m) Budu´cnost 2.776 ± 1.23 2.899 ± 1.23 3.431 ± 0.85 0.004 0.996 Sutjeska 3.194 ± 0.99 2.995 ± 1.04 2.498 ± 1.28 JD (m) Budu´cnost 3.436 ± 1.67 3.175 ± 1.39 3.589 ± 1.15 1.168 0.422 Sutjeska 3.841 ± 1.46 3.070 ± 1.37 2.795 ± 1.50 RD (m) Budu´cnost 1.378 ± 0.71 1.281 ± 0.72 1.395 ± 0.53 1.585 0.339 Sutjeska 1.552 ± 0.72 1.283 ± 0.63 1.155 ± 0.64 HSRD(m) Budu´cnost 719 ± 0.44 583 ± 0.30 617 ± 0.28 5.389 0.102 Sutjeska 794 ± 0.30 538 ± 0.23 461 ± 0.26 SD (m) Budu´cnost 92.75 ± 93.2 437 ± 0.32† 119 ± 0.09 0.401 0.010 Sutjeska 105 ± 72.1 347 ± 0.23‡ 66 ± 0.05 †2017 vs. 2015, 2020; ‡2017 vs. 2015, 2020; TD, total distance; WD, walk distance; JD, jog distance; RD, run distance; HSRD, high speed runs distance; SD, sprint distance. FIGURE 1 Trend in mean total distance by years. Frontiers in Public Health 03 frontiersin.org Goranovi´c et al. 10.3389/fpubh.2022.966578 FIGURE 2 Trend in mean walk distance by years. Results Table 1 shows the basic central and dispersion data on running performance during in-game soccer players. Analyzing the results in Table 1, it is evident that the players of both clubs achieved identical results in running performance during the game. Analyzing the derby match from 2020, it is evident that the soccer players of Buducnost ran more (9,129 m) in relation to the players of Sutjeska (7,019 m). Comparing the derbies from 2015 and 2017, it is clear that in the previous two derbies, the players from Sutjeska ran a greater distance compared to the derby from 2020, while the players from Buducnost ran the most in the derby in 2020. Also, in the derby in 2017, the players of both teams achieved a higher number of sprints compared to the derbies in 2015 and 2020. Applying appropriate statistical procedures, it was found that there are no statistically significant differences in running performance. Trends in running performance during the game by year are shown in the figures (Figures 1–6). Figure 1 shows the trend of the total length of running during the match. Unlike the soccer players of Buducnost, the soccer players of Sutjeska have a sharp drop in the total length of running in 2017. Figure 2 shows the trend of walking in the game. The analysis of the graph shows that the number of meters spent walking during the game varies from year to year. The soccer players of Sutjeska reduced the trend of walking, while the soccer players of Buducnost increased the trend of walking during the game. Unlike Figure 2, which shows a walk during the game, Figure 3 shows the total jog distance of the course during the game. Inspecting Figure 3 shows that the soccer players of Buducnost have a continuous trend of jogging, while the soccer players of Sutjeska have a trend of declining jogging in all 3 years. Figure 4 shows the downward trend in the running among Buducnost soccer players in all 3 years. The Sutjeska soccer players have seen a downward trend in all 3 years. Figure 5 shows the high-speed running distance for the soccer players of Buducnost and Sutjeska. Looking at Figure 5, it is evident that the players of both clubs have a downward trend in the most important zone for success in top soccer with one characteristic that the players of Buducnost have a minimal increase in 2020 compared to 2017, while the players of Sutjeska have a declining trend throughout the analyzed period. In contrast, Figure 6, which provides an insight into sprint distance, shows an increase in the number of sprints at both clubs in 2017, where the players of the Buducnost made a larger number of sprints, while in 2020 there is a decline and return to identical values as in 2015. Discussion The practical value of such analyses is that well-chosen performance indicators can help coaches identify the good and bad performance of individuals or teams. In this regard, match analyses help identify the physiological requirements of the sport and in examining how a particular player compares to the requirements of their event. Understanding the physiological load imposed on top players in accordance with their positional role during competitive matches (activity profile, distance traveled, intensity, energy systems, and muscles involved) is necessary when developing a sport-specific training protocol. Especially with elite athletes, the most important form of training is the one that corresponds to the use of energy and biomechanics of the planned competitive effect. Therefore, Frontiers in Public Health 04 frontiersin.org Goranovi´c et al. 10.3389/fpubh.2022.966578 FIGURE 3 Trend in mean jog distance by years. FIGURE 4 Trend in mean run distance by years. match analyses are helpful for the development of a specific training program that mimics the physiological conditions imposed by the game. Elite sports performances in soccer are a composite of the elite characteristics of physical performance, which in turn depend on several physiological characteristics, as well as on the training and health status of the individual athlete (14). The current study aimed to analyze trends of running performance in professional soccer players in Montenegro in three competitive matches of different seasons. During the game, soccer players perform different types of movement, ranging from resting to running at maximum speed, the intensity of which can change at any time. The distance covered during the match with elite soccer players is in the range of 10,000–12,000 m (15). The results of this study indicate that the trend component for the variable total distance is on an upward trajectory for Buducnost soccer players, ranging from 8,274 m in 2015 to 9,129 m in 2020, while for Sutjeska soccer players, there is a declining trend component of 9,441 m in 2015 to 7,019 m in 2020. Di Salvo et al. (14) recorded an average distance of 11,393 m for players competing in the Spanish Premier League in the 2003/2004 season. Osgnach et al. (16) recorded an average distance of 10,950 m for soccer players competing in the Italian Serie A in the 2007/2008 season. Comparing the stated results with the current research, it is evident that the soccer players who compete Frontiers in Public Health 05 frontiersin.org Goranovi´c et al. 10.3389/fpubh.2022.966578 FIGURE 5 Trend in mean high speed runs distance by years. FIGURE 6 Trend in mean sprint distance by years. in Montenegro are below the values achieved by those who compete in Europe. In the current study, the distances covered were categorized into five levels of intensity. The trend component in the walking distance variable for Buducnost players ranges from 2,776 m in 2015 to 3,431 m in 2020, while for Sutjeska players, there is a trend component of declining walking during the game from 3,194 m in 2015 to 2,498 m in 2020. In the variables jog distance and run distance, there is a continuous trend component without large oscillations in the players of both clubs. Withers et al. (17) state that 26.3% of the total game time falls on the intensity up to 14 km/h, 64.6% on the running intensity of 14.1–19 km/h, and 18.9% on the intensity of 19.1–23> km/h. Mayhew and Wenger (18) established that a soccer player walks 46.6%, runs slowly 38%, runs quickly or sprints 11.3%, and stands without moving 2.3% of the total playing time of a game. During a match, soccer players perform different types of behavior, ranging from standing still to maximum speed runs, the intensity of which may change at any given time. However, intensity parameters are not precisely defined in these papers. “Soccer is a non-cyclical and intermittent sport in which short-duration maximum-intensity activities, for example, sprint runs over a distance of 10–20 m, and high-intensity actions, such as counterattacks, are intertwined with activities Frontiers in Public Health 06 frontiersin.org Goranovi´c et al. 10.3389/fpubh.2022.966578 of low and moderate intensity (marching and jogging) and with pauses, for example, standing. Sprinting is one of the most important activities in soccer, although it merely constitutes between 1 and 12% of the mean total distance covered by a player during a match, that is, from only 0.5–3% of playing time. During a competitive game, players perform 2- to 4-s long-sprint runs every 90–180 s on average. It is assumed that players of higher ability cover longer sprinting distances with higher intensity” (19). The results of our study indicate that there is a downward trend in the most important zone for success in top soccer (high-speed runs distance), with one characteristic that the players of Buducnost have a minimal increase in 2020 compared to 2017, while Sutjeska players have a noticeable declining trend throughout the analyzed period. In contrast, an increase in the number of sprints at both clubs was recorded in 2017, while in 2020, there is a decline and return to identical values as in 2015. “The amount of high-speed running is what distinguishes top-class players from those at a lower level. Computerized time-motion analysis has demonstrated that international top-class players perform 28% more high- intensity running (2.43 vs. 1.90 km) and 58% more sprinting (650 vs. 410 m) than professional players at a lower level” (20). Furthermore, Ingebrigtsen et al. (21) “found that top teams in the Danish League covered 30–40% more high-speed running distance compared to the middle and bottom teams.” In contrast, Di Salvo et al. (22) “observed that Championship players did more high-speed running and sprinting than players in the Premier League, even though the differences were small. Along the same lines, a study comparing the match performance of players in the top three competitive standards of English soccer found that players in the second (Championship) and third (League 1) categories performed more high-speed running (>19 km/h) than those in the Premier League (803, 881, and 681 m, respectively), which was also the case for sprinting (308, 360, and 248 m, respectively)” (23). From the physiological aspect, the results of our study can be explained by the following fact: “During repetitive speed exercises, the contribution of phosphocreatine hydrolysis to the meeting of energy the demand of working muscles increases after each loading. The cool-down phase duration depends not only on the stimulation of the central nervous system but also on the rate of recovery of the autonomic nervous system functions related to the payoff of oxygen debt run up during physical exercise and on the rate of phosphocreatine resynthesis” (19). In contrast, soccer players perform significantly less high-intensity activities when they win than when they lose or when the result is a draw. Also, if the players score a goal in the early phase of the match, they do not use the maximum of their capacities during the match. Since winning is a pleasant situation for the team, it is possible that the players have set a strategy of keeping the ball, which results in fewer sprints (24). The limitations of this study are that only two soccer clubs from the first Telekom Montenegrin league were analyzed. Nevertheless, these two clubs are the most trophy-winning in the Montenegrin league, so they are included in the analysis. Future studies are recommended to enlarge the database. Such studies might be more suitable for detecting evolutionary trends in match-related variables. Conclusions The conclusion of this study provided information on performance in Montenegrin soccer, which could consequently improve the applicability of running performance in training and competitions. Based on the obtained results, the coaches will be advised in which direction the training process should go in order to increase the performance of Montenegrin elite soccer players. Data availability statement The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author. Ethics statement Ethical review and approval was not required for the study of human participants in accordance with the local legislation and institutional requirements. Written informed consent was obtained from the participants. Author contributions MJ formulated the research goals and aims, developed and designed the methodology, prepared the published work, and specifically wrote the initial draft. KG, JP, RH, and MJ prepared the published work, specifically with critical reviews, editing, and revisions. All authors commented on the draft and contributed to the final version, approved the publication of the manuscript, and agreed to be accountable for all aspects of the work. Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Publisher’s note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated Frontiers in Public Health 07 frontiersin.org Goranovi´c et al. 10.3389/fpubh.2022.966578 organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. References 1. Modri´c T, Versi´c S, Sekuli´c D. Aerobic fitness and game performance indicators in professional football players; playing position specifics and associations. Heliyon. (2020) 6:e05427. doi: 10.1016/j.heliyon.2020.e05427 2. Chmura P, Konefal M, Kowalczuk E, Andrzejewski M, Rokita A, Chmura J. Distances covered above and below the anaerobic threshold by professional football players in different competitive conditions. Central Eur J Sport Sci Med. (2015) 10:25–31. doi: 10.18276/cej 3. Silva JFD, Dittrich N, Guglielmo LGA. Aerobic evaluation in soccer. Rev Brasil Cineantr Desempenho Humano. (2011) 13:384–91. doi: 10.5007/1980-0037.2011v13n5p384 4. Garcia-Tabar I, Rampinini E, Gorostiaga EM. Lactate equivalent for maximal lactate steady state determination in soccer. Res Q Exerc Sport. (2019) 90:678–89. doi: 10.1080/02701367.2019.1643446 5. Lago-Peñas C, Lago-Ballesteros J. Game location and team quality effects on performance profiles in professional soccer. J Sports Sci Med. (2011) 10:465–71. 6. Modric T, Versic S, Sekulic D, Liposek S. Analysis of the association between running performance and game performance indicators in professional soccer players. Int J Environ Res Public Health. (2019) 16:4032. doi: 10.3390/ijerph16204032 7. Lago-Peñas C, Lago-Ballesteros J, Rey E. Differences in performance indicators between winning and losing teams in the UEFA champions league. J Hum Kinet. (2011) 27:135–46. doi: 10.2478/v10078-011-0011-3 8. Yi Q, Jia H, Liu H, Angel-Gomez N. Technical demands of different playing positions in the UEFA champions league. Int J Perform Anal Sport. (2018) 18:926– 93. doi: 10.1080/24748668.2018.1528524 9. Dellal A, Wong DP, Moalla W, Chamari K. Physical and technical activity of soccer players in the French first league-with special reference to their playing position. Int Sport Med J. (2010) 11:278–90. Available online at: https://hdl.handle. net/10520/EJC48393 10. Nobari H, Banoocy NK, Oliveira R, Pérez-Gómez J. Win, draw, or lose? Global positioning system-based variables’ effect on the match outcome: a full- season study on an iranian professional soccer team. Sensors. (2021) 21:5695. doi: 10.3390/s21175695 11. Akyildiz Z, Nobari H, González-Fernández FT, Moreira Praça G, Sarmento H, Hikmet Guler A, et al. Variations in the physical demands and technical performance of professional soccer teams over three consecutive seasons. Sci Rep. (2022) 12:1–24. doi: 10.1038/s41598-022-06365-7 12. Rodrigues Garcia G, Guilherme LCG, Clemente MF, Nakamura FY, Nobari H, Luiz Souza Bedo B, et al. Effects of congested fixture and matches’ participation on internal and external workload indices in professional soccer players. Sci Rep. (2022) 12:1864. doi: 10.1038/s41598-022-05792-w 13. World Medical Associations. Declaration of Helsinki ethical principles for medical research involving human subjects. JAMA. (2013) 310:2191–4. doi: 10.1001/jama.2013.281053 14. Di Salvo V, Baron R, Tschan H, Calderon-Montero FJ, Bachl N, Pigozzi F. Performance characteristics according to pla ying position in elite soccer. Int J Sports Med. (2007) 28:222–7. doi: 10.1055/s-2006-924294 15. Stolen T, Chamari K, Castanga C, Wisloff U. Physiology of soccer. Sports Med. (2005) 35:501–36. doi: 10.2165/00007256-200535060-00004 16. Osgnach C, Poser S, Bernardini R, Rinaldo R, Di Prampero PE. Energy cost and metabolic power in elite soccer: a new match analysis approach. Med Sci Sports Exerc. (2010) 42:170–8. doi: 10.1249/MSS.0b013e3181ae5cfd 17. Withers RT, Maricie Z, Wasilewski S, Kelly L. Match analyses of Australian professional soccer players. J Hum Move Stud. (1982) 8:159–76. doi: 10.1589/rika.23.407 18. Mayhew SR, Wenger HA. Time-motion analysis of professional soccer. J Hum Move Stud. (1985) 11:49–52. 19. Andrzejewski M, Chmura J, Pluta B, Strzelczyk R, Kasprzak A. Analysis of sprinting activities of professional soccer players. J Strength Cond Res. (2013) 27:2134–40. doi: 10.1519/JSC.0b013e318279423e 20. Mohr M, Krustrup P, Bangsbo J. Match performance of high-standard soccer players with special reference to development of fatigue. J Sport Sci. (2003) 21:439–49. doi: 10.1080/0264041031000071182 21. Ingebrigtsen J, Bendiksen M, Randers MB, Castagna C, Krustrup P, Holtermann A. Yo-Yo IR2 testing of elite and sub-elite soccer players: performance, heart rate response and correlations to other interval tests. J Sports Sci. (2012) 30:1337–45. doi: 10.1080/02640414.2012.711484 22. Di Salvo V, Pigozzi F, González-Haro C, Laughlin MS, De Witt JK. Match performance comparison in top english soccer leagues. Int J Sports Med. (2013) 34:526–32. doi: 10.1055/s-0032-1327660 23. Bradley PS, Carling A, Gomez Diaz P, Hood C, Barnes J, Ade M, et al. Match performance and physical capacity of players in the top three competitive standards of English professional soccer. Hum Mov Sci. (2013) 32:808–21. doi: 10.1016/j.humov.2013.06.002 24. Minano-Espin J, Casais L, Lago-Penas C, Gomez-Ruano MA. High speed running and sprinting profiles of elite soccer players. J Human Kinet. (2017) 58:169–76. doi: 10.1515/hukin-2017-0086 Frontiers in Public Health 08 frontiersin.org

Exploring trends of running performance during matches of professional soccer players in Montenegro: A longitudinal study.

08-01-2022

Goranović, Kosta,Hadžić, Rašid,Petković, Jovica,Joksimović, Marko

eng

PMC10593817

1 Vol.:(0123456789) Scientific Reports | (2023) 13:18083 | https://doi.org/10.1038/s41598-023-45055-w www.nature.com/scientificreports Differences in race history by distance of recreational endurance runners from The NURMI Study (Step 2) Beat Knechtle 1,2*, Derrick Tanous 3,4, Mabliny Thuany 5, Mohamad Motevalli 3,4, Gerold Wirnitzer 6, Claus Leitzmann 7, Katja Weiss 1, Thomas Rosemann 1 & Katharina Wirnitzer 3,4,8 Few studies were developed to understand the relationship between running characteristics and motivation. The purpose of this study was to assess the relationship between running event history, running experience, and best race performances in recreational distance runners. We used a web survey to obtain information regarding running experience, racing history, and periodization training routines/exercise habits, including weekly volumes and daily mileage and duration across periods and conditions. Associations between variables were conducted with the Chi-square test (χ2; nominal scale) and Wilcoxon test. Multiple linear regression analysis and multivariate linear regression were performed. Concerning the participants’ motive for exercising, a significant difference was identified between the race distance subgroups (p < 0.001), where 58% of M/UM runners exercised for performance (n = 38) and 64% of HM runners (n = 57) and 57% of 10 km runners (n = 52) exercised for recreational purposes. A significant difference was found in the number of years of running completed without taking a break (p = 0.004), with marathoners/ultramarathoners reporting the most years. Runners competing in different race distances such as 10 km, half-marathon, marathon, and ultra-marathon presented differences in training background and habits according to the distance of preference. Running is a global market, with an increase in the participation of athletes in running events and the number of events over the last year worldwide1–3. In the European context, a range of 5% to 31% rate of participation was shown between different countries4. In a scientific context, this growth was associated with a higher interest for understanding runners’ profiles, behaviors, and training habits5,6. The runner’s profile was previously studied in different contexts, including differences in economic level7,8, the profiles consumption and use of sports watches9, training characteristics5,10, nutritional behaviors11–13, and health outcomes14. As a social phenomenon, and with the potential to improve general physical (i.e., lower risks of all-cause and cardiovascular mortality)15 and mental health (i.e., well-being, self-confidence), running is also related to social cohesion16 and used as a potential strategy to improve physical activity levels in an epidemiological context17. In this way, the reasons to start running and to be engaged in running training were also investigated previously18. For non-professional runners, motivational differences were shown in athletes competing in differ- ent race distances19–22. For runners in 5 km, fun and health were the most important factors for training23, while ultra-marathoners had higher scores in affiliation, life meaning, and lower body weight concerns24. Based on previous studies, a body of evidence is available regarding motivational characteristics and run- ners’ profiles19,25,26. However, few studies were developed to understand the relationship between running back- ground and motivation27. Understanding the motives and habits considering training and competing that enable OPEN 1Institute of Primary Care, University of Zurich, 8000 Zurich, Switzerland. 2Medbase St. Gallen Am Vadianplatz, Vadianstrasse 26, 9000 St. Gallen, Switzerland. 3Department of Sport Science, University of Innsbruck, 6020 Innsbruck, Austria. 4Department of Research and Development in Teacher Education, University College of Teacher Education, Tyrol, 6020 Innsbruck, Austria. 5Faculty of Sports, University of Porto, Porto, Portugal. 6AdventureV & change2V, 6135 Stans, Austria. 7Institute of Nutrition, University of Gießen, 35390 Gießen, Germany. 8Research Center Medical Humanities, Leopold-Franzens University of Innsbruck, 6020 Innsbruck, Austria. *email: beat.knechtle@hispeed.ch 2 Vol:.(1234567890) Scientific Reports | (2023) 13:18083 | https://doi.org/10.1038/s41598-023-45055-w www.nature.com/scientificreports/ non-professional runners to be engaged in physical exercise is an important feature to provide support and to understand why people are or not engaged in running, as well as to develop strategies to maintain the training commitment. Therefore, this is the first exploratory investigation to assess the aspects of motivation, education, training, previous experience, and performance in different running groups such as 10 km, half-marathon (HM), and marathon (M)/ultramarathon (UM) recreational distance runners. Based on previous studies28–30, it is assumed that there would be differences in these aspects in recreational endurance runners of different distances (10 km, HM, M/UM)10. Materials and methods Please see the subsequent description of the methodology for the complete profile for this investigation (Part A of the arrangement)31, as well as previous publications10,32–34. Following a protocol35, the Nutrition and Running High Mileage (NURMI) Study has been approved by the ethics board of St. Gallen, Switzerland on the 6th of May in 2015 (EKSG 14/145) with a retrospective trial registration (number: ISRCTN73074080). It was required that the participants provided informed consent before taking part in the NURMI Study. For the participants’ recruit- ment and study procedures, the responsive reader is kindly referred to Part A of the arrangement publication31. Figure 1 shows the enrollment and categorization of participants, and their characteristics are shown in Table 1. Measures Race performances, training routines, and exercise habits of active distance runners were expressed using the following parameters: running experience (total number of years of running fully completed without taking a break); racing history (overall number of completed races, ratio of HM/M events to total races, age at time of the first running event, the first race distance completed: 10 km, HM, M, best HM/M times, the number of planned races completed in the previous two years: HM/M/UM); periodization training routines/exercise habits, includ- ing weekly volumes (number of running sessions, and breadth of training in km and hours) and daily mileage and duration across periods and conditions. Running performance was related to best finishing HM and M time based on a normalized aggregate mean transformed to an index (ranging 0–100). The latent variable of run- ning history was derived by both factors: (1) “running-experience” (by pooled items: “age.first.running event”, “age.run”, “age.first.half-marathon”, “age.first.marathon”) and (2) “racing-experience” (by pooled items: “years. running”, “completed.half-marathon.number”, “completed.marathon.number”), which were defined by specific items that were based on manifest variables. As running experience (e.g., years of running fully completed, age at the first race event, total number of races completed) is dependent upon age, the respective items were operationalized with age (e.g., age-related years of running, age-related number of completed races over half-marathon distance). Based on this, the respective items (e.g., age-related beginning of running, first marathon race completed) were centered by median values, and were z-transformed creating a new scale through summarizing the respective items (e.g., years of running fully completed, completed races over specified distances). From this the values were categorized with the latent factors “running-experience” and “racing-experience” into low (values below − 1), medium (values ranging Figure 1. Enrollment and Categorization of Participants by Race Distance. 3 Vol.:(0123456789) Scientific Reports | (2023) 13:18083 | https://doi.org/10.1038/s41598-023-45055-w www.nature.com/scientificreports/ from − 1 to + 1), and high (values higher + 1). A principal component analysis (PCA as heuristic approach) was performed to identify the respective factors. The PCA was justified by sufficient high correlations (0.79 by the Kaiser–Meyer–Olkin-Kriterium, and p < 0.001 by the Bartlett-Test as highly significant) to derive the extraction of two factors. The “Eigen”-Wert > 1 (declaration of 73.4% of total variance of both the latent factors) was defined to justify to model two latent factors: “running-experience” (from items: “age.first.running event”, “age.run”, “age. first.half-marathon”, “age.first.marathon”) and “racing-experience” (from items: “years.running”, “completed.half- marathon.number”, “completed.marathon.number”). Statistical analysis The statistical analyses were all performed with R software (version 3.6.2 Core Team 2019; R Foundation for Statistical Computing; Vienna, Austria). The exploratory analysis was performed with descriptive statistics, including median with interquartile range (IQR) and mean with standard deviation (SD). PCA was used for identifying the latent factors.Significant differences in running and racing activity (experience, training, rac- ing, etc.) between race distance subgroups were calculated with a non-parametric test. Associations between variables were conducted with Chi-square test (χ2; nominal scale) and Wilcoxon test (ordinal and metric scale) have been approximated by using F distributions and ordinary least squares. Multiple linear regression analysis and multivariate linear regression were performed to test the differences in performance, health, and leisure Table 1. Runner characteristics, including motive to race, experience, and history displayed race distance. Note Results are presented as percentage (%), total numbers, and median (IQR). χ2 statistic calculated by Pearson’s Chi-squared test and F statistic calculated by Kruskal–Wallis test. 10 km 10 km. HM half-marathon. M/UM marathon/ultra-marathon. Total 100% (245) 10 km37% (91) HM 36% (89) M/UM 27% (65) Statistics Age (Years) 39 (IQR 17) 37 (IQR 18) 37 (IQR 18) 44 (IQR 17) F(2,242) = 4.87 p = 0.008 BMI (kg/m2) 21.7 (IQR 3.5) 21.3 (IQR 3.94) 22 (IQR 3.28) 22.2 (IQR 3.25) F(2,242) = 1.22 p = 0.296 Civil status  Single 27% (66) 26% (24) 31% (28) 22% (14) χ2 (4) = 1.95 p = 0.744  With spouse/married 67% (164) 67% (61) 63% (56) 72% (47)  Separated/divorce 6% (15) 7% (6) 6% (5) 6% (4) Motive to race  Leisure 46% (106) 41% (36) 47% (41) 51% (29) χ2 (2) = 1.34 p = 0.512  Performance 54% (125) 59% (51) 53% (46) 49% (28) Favorite season of racing  Winter < 1% (2) 1% (1) 1% (1) / χ2 (6) = 9.04 p = 0.171  Spring 46% (106) 36% (31) 55% (48) 47% (27)  Summer 23% (52) 28% (24) 15% (13) 26% (15)  Autumn 31% (71) 36% (31) 29% (25) 26% (15) Running experience (years) 7 (IQR 7) 5 (IQR 8) 7 (IQR 6) 8 (IQR 9) F(2, 241) = 5.77 p = 0.004 First event age (years)  10 km 30 (IQR 16) 30 (IQR 17) 28 (IQR 15) 33 (IQR 17) F(2, 151) = 0.69 p = 0.502 F(2, 216) = 1.17 p = 0.313 F(2, 135) = 0.18 p = 0.836 F(2, 239) = 1.77 p = 0.172  HM 32 (IQR 16) 33 (IQR 15) 30 (IQR 18) 35 (IQR 13)  M 35 (IQR 13) 33 (IQR 15) 34 (IQR 17) 35 (IQR 12)  Total 30 (IQR 16) 30 (IQR 17) 28 (IQR 18) 34 (IQR 13) First event  10 km 65% (157) 81% (74) 59% (52) 48% (31) χ2 (4) = 46.24 p < 0.001  HM 27% (65) 18% (16) 38% (33) 25% (16)  M 9% (21) 1% (1) 3% (3) 27% (17) Total races completed 8 (IQR 11) 7 (IQR 11) 6 (IQR 11) 10 (IQR 11) F(2, 242) = 2.90 p = 0.057 Ratio of HM/M to total races 40 (IQR 50) 20 (IQR 35) 48 (IQR 43) 53 (IQR 49) F(2, 242) = 18.44 p < 0.001 Completion of planned events (previous 2 years)  HM 2 (IQR 3) 1 (IQR 2) 3 (IQR 4) 2 (IQR 3) F(2, 242) = 7.04 p = 0.001 F(2, 242) = 75.19 p < 0.001 F(2, 242) = 28.84 p < 0.001  M 1 (IQR 2) 0 (IQR 1) 0 (IQR 1) 2 (IQR 2)  UM 0 (IQR 0) 0 (IQR 0) 0 (IQR 0) 0 (IQR 1) 4 Vol:.(1234567890) Scientific Reports | (2023) 13:18083 | https://doi.org/10.1038/s41598-023-45055-w www.nature.com/scientificreports/ motivations based on race distance subgroups. The regression results are displayed as effect plots with a 95% confidence interval (95%-CI). The level of statistical significance was set at p ≤ 0.05. Institutional review board The study protocol is available online via https:// sprin gerpl us. sprin gerop en. com/ artic les/ 10. 1186/ s40064- 016- 2126-4 and was approved by the ethics board of St. Gallen, Switzerland on May 6, 2015 (EKSG 14/145). The study was conduct-ed in accordance with the ethical standards of the institutional review board, medical professional codex, and with the 1964 Helsinki declaration and its later amendments as of 1996, the Data Security Laws, and good clinical practice guidelines. Study participation was voluntary and could be canceled at any time without the provision of reasons or negative consequences. In-formed consent was obtained from all individual participants included in the study considering the data collected, used, and analyzed exclusively and only in the context of the NURMI Study for scientific publication. Results The total sample included 317 runners of various long distances who finished and submitted the questionnaire. A sum of 72 participants were excluded due to failing to meet the inclusion criteria following data clearance. The final sample was comprised of 245 runners (10 km: n = 91; NURMI runners: HM: n = 89; M/UM: n = 65), including 104 males and 141 females. Together the participants had a BMI of 21.7 kg/m2 (body weight of 65 kg, height of 1.7 m) and were aged 39 years. Regarding the participants’ nationalities, 72% came from Germany (n = 177), 18% were from Austria (n = 44), and 9% were from Switzerland (n = 13) or another country (n = 11). Significant differences were observed across the race distance subgroups for height (p = 0.007), body weight (p = 0.007), and age (p = 0.008) with the M/UM participants being taller (1.8 m, IQR 0.1), heavier (67.5 kg, IQR 17.5), and older (44 years, IQR 17). No significant difference was observed across the race distance subgroups for BMI (p = 0.296) or for civil status (p = 0.744), most participants were married or living with their spouse (67%; n = 144) or single (27%; n = 66). No significant differences were found for race distance subgroups regarding the participants’ educational background (p = 0.177): 1 (< 1%) had no qualification, 53 (22%) held an A-Levels (or similar degree), 83 (34%) held an upper secondary school/technical education degree, 83 (34%) held a university degree (or possibly higher), and 25 (10%) did not answer. Concerning the participants’ motive for exercising, a significant difference was identified between the race distance subgroups (p < 0.001), where 58% of M/UM runners exercised for performance (n = 38) and 64% of HM runners (n = 57) and 57% of 10 km runners (n = 52) exercised for recreational purposes. The participants’ characteristics, including their motive to race and running experiences are shown in Table 1 based on their self-reported race distances. In Part A, additional details on the total sample’s profile and the race distance-specific subgroups are provided31. No significant differences were found across the race distance subgroups for the motive to race (p = 0.512) or the current motive to run (p = 0.583); performance was the most frequently reported racing motive (54%; n = 125) among the whole sample. No significant difference was observed for the favorite race season (p = 0.171); spring- time was the most favored season for racing for all participants (46%; n = 106), while winter was the least favored (< 1%; n = 2). A significant difference was found in the number of years of running completed (consecutively or inconsecutively) without taking a break (p = 0.004), with M/UM runners reporting the most years (8; IQR 9) and 10 km runners reporting the least (7 IQR 11). Regarding racing history, significant differences between race distance subgroups were found in (i) the ratio of completed HM/M events to the total races, where M/UM run- ners had the highest reports (53; IQR 49; p < 0.001); (ii) the first race distance, where most 10 km (81%; n = 74) and HM (59%; n = 52) runners first completed a 10 km race (p < 0.001); (iii) the best time for a HM race, where M/UM runners were the fastest on average (99 min ± 13; p < 0.001); (iv) the best time for a M race, where M/UM runners were the fastest on average (218 min ± 34; p = 0.029); (v) the completion of HM (p = 0.001), M (p < 0.001), and UM (p < 0.001) races in the previous two years, where HM runners completed the most HM races (3; IQR 4) and M/UM runners completed the most M (2; IQR 2) and UM (0; IQR 1) races. No significant differences in racing history between race distance subgroups were identified in overall completed races (p = 0.057), first event age in total (p = 0.172), or regardless of 10 km (p = 0.502), HM (p = 0.313), or M distance (p = 0.836). Non-significant relationships were identified in multivariate linear regression, as seen in Fig. 2, between (i) the motives of performance, the 10 km subgroup, and the HM subgroup (b = − 4.21; 95% CI [− 15.2 to 6.81]; p > 0.05) or the M/UM subgroup (b = 2.5; 95% CI [− 9.89 to 14.9]; p > 0.05); (ii) the motives of health, the 10 km subgroup, and the HM subgroup (b = − 3.07; 95% CI [− 11.5 to 5.39]; p > 0.05) or the M/UM subgroup (b = − 7.9; 95% CI [− 17.4 to 1.6]; p > 0.05); (iii) the motives of leisure, the 10 km subgroup, and the HM subgroup (b = 4.98; 95% CI [− 3.42 to 13.4]; p > 0.05) or the M/UM subgroup (b = 4.86; 95% CI [− 4.59 to 14.3]; p > 0.05). Multivariate linear regression was performed and the following confounders were included within different models to predict the best HM and M race time between 10 km and HM or M/UM race distance subgroups: (a) years of running history and age at the first running event, which determined 21% of variance (adjusted R2 = 0.21) and a significant difference was identified for M/UM runners (b = 10.9; 95% CI [1.74–20]; p < 0.05) but not for HM runners (b = − 5.72; 95% CI [− 14.1 to 2.65]; p > 0.05); (b) training routines and exercise habits (including preparation condition 3, preparation condition 4, weekly kilometers of preparation condition 1, professional support, and the training extent for main race in months), which determined 22% of variance (adjusted R2 = 0.22) and no significant difference for HM (b = − 6; 95% CI [− 14.6 to 2.65]; p > 0.05) or M/UM (b = 0.679; 95% CI [− 9.09 to 10.5]; p > 0.05) race distance groups; (c) racing history (total races completed, the ratio of HM/M events to total events, HM races completed, and M races completed), which determined 16% variance (adjusted R2 = 0.16) and no significant difference for HM (b = − 6.26; 95% CI [− 15 to 2.45]; p > 0.05) or M/UM (b = 7.6; 95% CI [− 3.47 to 18.7]; p > 0.05) race distance subgroups. In Table 2, multiple linear regression analyses are provided. 5 Vol.:(0123456789) Scientific Reports | (2023) 13:18083 | https://doi.org/10.1038/s41598-023-45055-w www.nature.com/scientificreports/ Discussion This study was the first exploratory investigation aiming to analyze running event history, running experience, and best race performance between 10 km, HM, and M/UM recreational runners. The most important findings were (i) the runners had a similar BMI regardless of race distance subgroup even though M/UM runners were the tallest participants and weighed the most; (ii) no difference was found across race distance subgroups in the motive to race or for the linked motives (i.e. exercise motive, original motive to run, present motive to run, motive to race); (iii) M/UM runners tallied significantly more years of fully active running experience and completed significantly more of their planned marathon and ultra-marathon races in the previous two years; (iv) significant differences between the race distance subgroups in best time performances, where M/UM runners were fastest on average to complete HM and M events, however, when analyzing best time performances with an index and applying confounders (training routines and exercise habits; racing history) in multivariate linear regression analyses, no significant differences in performance were found between subgroups; (v) M/UM runners remained fastest on average to complete HM and M events when considering the confounders of running experience Figure 2. Effect plots displaying 95%-CI average between 10 km, HM, and M/UM subgroups in exercise/ running/racing motives (n = 231). Note 95%-CIs were computed using the multivariate regression analyses (Wald approximation). Table 2. Multiple linear regression analyses on running experience, training routines and exercise habits, and racing history. Note b = estimate (marginal effects), CI confidence interval, HM half-marathon, M marathon, UM ultra-marathon. Adjusted r2 Model 1 0.21 Model 2 0.22 Model 3 0.16 b 95%-CI p b 95%-CI P b 95%-CI p Intercept 68.8 56.2–81.5 < 0.001  Years of running experi- ence 0.765 − 0.28–1.25 < 0.01  First event age − 0.97 − 1.3 to − 0.64 < 0.001  HM Subgroup − 5.72 − 14.1 to 2.65 > 0.05  M/UM Subgroup 10.9 1.74–20 < 0.05 Intercept 32 19.4–44.7 < 0.001  Preparation condition 3 5.92 1.85–9.98 < 0.01  Preparation condition 4 − 2.14 − 5.98 to 1.7 > 0.05 Prep Condition 1: Weekly km 0.274 0.1–0.45 < 0.01  Professional support 10.6 0.17–21.1 < 0.05  Training extent for main race − 1.2 − 3.17 to 0.77 > 0.05  HM subgroup − 6 − 14.6 to 2.65 > 0.05  M/UM subgroup 0.679 − 9.09 to 10.5 > 0.05 Intercept 39.4 28.7–50.1 < 0.001  Races completed in total 0.558 0.05–1.07 < 0.05  Ratio of HM/M to total races − 0.0746 − 0.22 to 0.08 > 0.05  HM races completed 0.962 − 0.94 to 2.86 < 0.05  M races completed 0.634 − 1.7 to 2.97 > 0.05  HM subgroup −6 .26 − 15 to 2.45 > 0.05  M/UM subgroup 7.6 − 3.47 to 18.7 > 0.05 6 Vol:.(1234567890) Scientific Reports | (2023) 13:18083 | https://doi.org/10.1038/s41598-023-45055-w www.nature.com/scientificreports/ in years fully active in running without break and the participants’ age at their first running event. Thus, this exploratory investigation upholds the assumption that there is a difference in the best race performances con- sidering time to finish between recreational endurance runners of different distances (10 km, HM, M/UM). Differences in anthropometry and age across groups We found that the runners had a similar BMI regardless of race distance subgroup, even though M/UM run- ners had the highest body height and the heaviest body mass. Furthermore, M/UM runners were older. A study investigating master half-marathoners, master marathoners, and master ultra-marathoners found, however, no differences regarding their age, body mass, body height, and body mass index36. A study comparing recreational marathoners and recreational ultra-marathoners found differences in anthropometry where marathoners had a lower calf circumference but thicker skinfold thicknesses at pectoral, axilla, and suprailiacal sites compared to the ultra-marathoners37. Also, a study comparing recreational half-marathoners and marathoners reported that half-marathoners had a higher body mass, longer legs, a larger circumference of the upper arm, thicker thigh skinfolds, a higher sum of skinfold thicknesses, a higher body fat percentage, and a higher skeletal muscle mass than marathoners29. These disparate findings might be due to different sample sizes and performance levels of the subjects. Differences in motivation across groups We found no difference across race distance subgroups regarding the motivation to compete or the associated motives (i.e., exercise motive, original motive to run, present motive to run, motive to race). Interestingly, this finding disagrees with previous findings19,21,26, and different aspects might explain the discrepancy. Methodo- logical differences, including analysis stratification by sex and age groups22,38, training habits39,40, and country of residence41, can be related to the differences in the findings. Differences between the sexes were shown for mara- thoners, where women were more motivated about their weight, affiliation, psychological coping, life meaning, and self-esteem but were less driven by competition38. Ultra-marathoners presented higher scores on affiliation and life meaning and lower values for body weight concerns, personal goal achievement, and self-esteem38,42. The second running boom (1990s) increased the number of runners that are not aiming to become professional athletes but their engagement in competitions as a leisure/social activity16, which people used as a strategy to be involved in social groups as well as to know different places around the world43. Furthermore, no significant difference was observed for the favorite race season. In elite marathoners, however, the seasonal distribution for marathon running has two peaks, spring (weeks 14 to 17) and autumn (weeks 41 to 44). During these two periods, the expected temperature is close to the optimal value for marathon running44. It is well-described that interrelationships between marathon results and weather factors such as air temperature, wet bulb temperature, and human biometeorological indices exist45. Most probably, recreational runners do not focus on environmental conditions but rather on a specific event they want to compete in. Differences in running experience across groups We found a significant difference in the number of years of running completed (consecutively or inconsecutively) without taking a break, with M/UM runners reporting the highest number of years and 10 km runners reporting the lowest number. M/UM runners reported more years of fully active running experience and completed more of their planned marathon and ultra-marathon races in the previous two years compared to the 10 km runners. The higher time of experience for M/UM runners and more completed marathon and ultra-marathon races in the previous two years highlight the profile of this subgroup. Similar findings showed that long-distance runners were older than short-distance runners (i.e., 5 km, 10 km)46,47. These characteristics are also related to the age of peak performance since a positive relationship has been reported between the age of peak performance and the length of the race distance48–50. In this way, differences between the race distance subgroups regarding the best time performances can also be related to training background and running experience. Besides the genetic component51, the main physiological parameters associated with long-distance performance (i.e., maximal oxy- gen consumption (VO2max), running economy, lactate threshold, and velocity associated with VO2max) are developed during training through the increases in the mitochondrial content and skeletal muscle capillary density32,52. Besides that, marathon and ultra-marathon performance are strongly related to sex, morphological, and psychological variables53,54, which can act as confounders in the present study. Differences in previous performance across groups We found significant differences between the race distance subgroups regarding the best time performances. On average, M/UM runners were faster to complete HM and M events. This finding is not in line with previ- ous findings. Data covering 107.9 million race results, including 70,000 events held from 1986 to 2018, showed that non-professional marathoners were 18% and 17% slower compared to female and male half-marathoners, respectively55. In addition, the best performances can be related to the sex distribution among the subgroups since men are overrepresented in M/UM (62%). A body of evidence is available regarding running performance differences between sex56,57, where men tended to perform 10% better compared to women56. Data from previ- ous research from the NURMI study confirms sex differences for years of active running, the number of races completed, and best time performance, with men being faster on average at HM and M distances compared to women33. However, these differences tended to be null when training routines, exercise habits, and racing history was considered confounders. These results indicate that regardless of the subgroup distance, training background is important for the best finish time, as shown previously28. In addition, when considering the confounders of running experience in years fully active in running without a break and the participants’ age at their first running 7 Vol.:(0123456789) Scientific Reports | (2023) 13:18083 | https://doi.org/10.1038/s41598-023-45055-w www.nature.com/scientificreports/ event, M/UM runners remained the fastest on average to complete HM and M. These results highlight the mul- tifactorial and complex nature of the cause of achieved results or successes in sports disciplines58. Differences in race performance across groups We found that M/UM runners remained the fastest on average to complete HM and M events when considering the confounders of running experience in years fully active in running without a break and the participants’ age at their first running event. Thus, this exploratory investigation upholds the assumption that there is a difference in the best race performances considering time to finish between recreational endurance runners of different distances (10 km, HM, M/UM). A study comparing 10 km, half-marathon, and marathon showed differences regarding age and running speed between the groups59. Limitations Considering the limitation of the cross-sectional design, this study’s findings have some limitations that should be addressed, including that no underlying causation can be acquired from the present results. The primary limi- tation for vital consideration is the self-report feature of the survey methodological approach, which is known to result in misrepresented answers due to social expectations60. In addition, study participation was voluntary, which may have led to a non-randomized study population, although the participants were highly motivated. For the present study, the distance groups were not stratified by sex, which limits the comparisons, and suggest that different sub-groups need to be studied among runners to better understand motives, routines, and physi- cal exercise engagement. To limit the misreporting effect, the survey included control questions throughout the different parts. Additionally, highly motivated distance runners made up the study sample, which likely added to the reliability of their responses and enhanced the dataset. Moreover, the sample included 245 endurance run- ners, which was relatively small considering the commonality of running as a sport. Moreover, other individual (nutritional status or the nutritional type maintained by the participants) and environmental characteristics (the racing environment, and specific weather conditions) that affect training commitment and performance was not considered in the present study (but of the NURMI Study Step 3, not published so far). Despite this limitation, the present study presents some advances for the events organizations, coaches, and sports scientists to better understand amateur runners of different characteristics. In addition, the race distance subgroups were unequally distributed per se, considering that 37% of the total sample were 10 km runners, 36% were HM runners, and 27% were M/UM runners. Another limitation is that multiple aspects of running competitions were not con- trolled for, essentially the racing environment itself and the specific weather conditions (poor or good running weather, temperature, and humidity), the time of the event, the season, and competition region. Regardless, the best time performances were retrospectively verified under random selection. Lastly, the current investigation did not include nutritional status or the nutritional type maintained by the participants, as personal nutrition is well-known to affect performance. Even though this investigation did not include nutritional results, the NURMI study has obtained the runners nutritional evidence that was or will be published in other articles due to scientific journal publication demands. Conclusions Runners competing in different race distances such as 10 km, half-marathon, marathon, and ultra-marathon presented differences in training background and habits according to the distance of preference. Marathoners and ultra-marathoners were older, taller, and heavier, were running for more years, and had faster personal best times than 10 km runners. Further studies need to consider the second level of information, considering the role of competition in runners’ training commitment as well as environmental features related to training commitment. Data availability The data sets generated during and/or analyzed during the current study and presented in this article are not publicly available. Requests to access the datasets should be directed to info@nurmi-study.com. Subjects will receive a brief summary of the results of the NURMI Study if desired. Received: 27 December 2022; Accepted: 15 October 2023 References 1. RunRepeat. The State of Ultra Running 2020. https:// runre peat. com/ state- of- ultra- runni ng. (2021). 2. RunRepeat. 133 Stats on 5K Running Races in the US. https:// runre peat. com/ the- us- 5k- stats- page (2021). 3. RunRepeat. Marathon Statistics 2019 Worldwide. https:// runre peat. com/ resea rch- marat hon- perfo rmance- across- natio ns (2020). 4. Scheerder, J., Breedveld, K. & Borgers, J. Running across Europe: The Rise and Size of One of the Largest Sport Markets (Palgrave Macmillan, 2015). 5. Kozlovskaia, M. et al. A profile of health, lifestyle and training habits of 4720 Australian recreational runners: The case for promot- ing running for health benefits. Health Promot J. Austr. 30, 172–179. https:// doi. org/ 10. 1002/ hpja. 30 (2019). 6. Parra-Camacho, D., Alonso Dos Santos, M. & González-Serrano, M. Amateur Runners’ Commitment: An analysis of sociodemo- graphic and sports habit profiles. Int. J. Environ. Res. Public Health 17, 925. https:// doi. org/ 10. 3390/ ijerp h1703 0925 (2020). 7. Thuany, M., Malchrowicz-Mośko, E., Waśkiewicz, Z. & Gomes, T. Individual and economic characteristics as determinants of Brazilian runners’ motivation. Sustainability 13, 10178. https:// doi. org/ 10. 3390/ su131 810178 (2021). 8. Breuer, C., Hallmann, K. & Wicker, P. Determinants of sport participation in different sports. Manag 16, 269–286. https:// doi. org/ 10. 1080/ 13606 719. 2011. 613625 (2013). 9. Janssen, M., Scheerder, J., Thibaut, E., Brombacher, A. & Vos, S. Who uses running apps and sports watches? Determinants and consumer profiles of event runners’ usage of running-related smartphone applications and sports watches. PLoS ONE 12, e0181167. https:// doi. org/ 10. 1371/ journ al. pone. 01811 67 (2017). 8 Vol:.(1234567890) Scientific Reports | (2023) 13:18083 | https://doi.org/10.1038/s41598-023-45055-w www.nature.com/scientificreports/ 10. Knechtle, B. et al. Training and racing behavior of recreational runners by race distance-results from the NURMI Study (Step 1). Front. Physiol. 12, 620404. https:// doi. org/ 10. 3389/ fphys. 2021. 620404 (2021). 11. Boldt, P. et al. Quality of life of female and male vegetarian and vegan endurance runners compared to omnivores: Results from the NURMI study (step 2). J. Int. Soc. Sports Nutr. 15, 33. https:// doi. org/ 10. 1186/ s12970- 018- 0237-8 (2018). 12. Wirnitzer, K. et al. Sex differences in supplement intake in recreational endurance runners—results from the NURMI Study (Step 2). Nutrients 13, 2776. https:// doi. org/ 10. 3390/ nu130 82776 (2021). 13. Wirnitzer, K. et al. Who is running in the D-A-CH countries? An epidemiological approach of 2455 omnivorous, vegetarian, and vegan recreational runners-results from the NURMI Study (Step 1). Nutrients 14, 677. https:// doi. org/ 10. 3390/ nu140 30677 (2022). 14. Boldt, P. et al. Sex Differences in the health status of endurance runners: Results from the NURMI Study (Step 2). J. Strength Cond. Res. 33, 1929–1940. https:// doi. org/ 10. 1519/ jsc. 00000 00000 003010 (2019). 15. Lee, D.-C. et al. Leisure-time running reduces all-cause and cardiovascular mortality risk. J. Am. Coll. Cardiol. 64, 472–481. https:// doi. org/ 10. 1016/j. jacc. 2014. 04. 058 (2014). 16. Hautbois, C., Djaballah, M. & Desbordes, M. The social impact of participative sporting events: A cluster analysis of marathon participants based on perceived benefits. Sport Soc. 23, 335–353. https:// doi. org/ 10. 1080/ 17430 437. 2019. 16733 71 (2019). 17. WHO. Guidelines on Physical Activity and Sedentary Behaviour (World Health Organization, 2020) https:// www. who. int/ publi catio ns/i/ item/ 97892 40015 128. 18. Malchrowicz-Mosko, E., León-Guereño, P., Tapia-Serrano, M., Sánchez-Miguel, P. & Waśkiewicz, Z. What encourages physically inactive people to start running? An analysis of motivations to participate in parkrun and city trail in poland. Public Health Front. 8, 581017. https:// doi. org/ 10. 3389/ fpubh. 2020. 581017 (2020). 19. Whitehead, A. et al. Motivational differences between 5K, half marathon and full marathon participants in the UK and India. Manag Sport Leisure 27, 337–350. https:// doi. org/ 10. 1080/ 23750 472. 2020. 17912 36 (2022). 20. León-Guereño, P., Galindo-Domínguez, H., Balerdi-Eizmendi, E., Rozmiarek, M. & Malchrowicz-Mośko, E. Motivation behind running among older adult runners. BMC Sports Sci. Med. Rehabil. 13, 138. https:// doi. org/ 10. 1186/ s13102- 021- 00366-1 (2021). 21. Gerasimuk, D. et al. Age-related differences in motivation of recreational runners, marathoners, and ultra-marathoners. Front. Psychol. 12, 738807–738807. https:// doi. org/ 10. 3389/ fpsyg. 2021. 738807 (2021). 22. Leon-Guereno, P., Tapia-Serrano, M., Castaneda-Babarro, A. & Malchrowicz-Mosko, E. Do sex, age, and marital status influence the motivations of amateur marathon runners? The Poznan Marathon Case Study. Front. Psychol. 11, 2151. https:// doi. org/ 10. 3389/ fpsyg. 2020. 02151 (2020). 23. Manzano-Sánchez, D., Postigo-Pérez, L., Gómez-López, M. & Valero-Valenzuela, A. Study of the motivation of Spanish amateur runners based on training patterns and gender. Int. J. Environ. Res. Public Health 17, 8185. https:// doi. org/ 10. 3390/ ijerp h1721 8185 (2020). 24. Waśkiewicz, Z., Nikolaidis, P., Chalabaev, A., Rosemann, T. & Knechtle, B. Motivation in ultra-marathon runners. Psychol. Res. Behav. Manag. 12, 31–37. https:// doi. org/ 10. 2147/ prbm. s1890 61 (2019). 25. Stevinson, C. et al. Adherence and health-related outcomes of beginner running programs: A 10-week observational study. Res. Q. Exerc. Sport 93, 87–95. https:// doi. org/ 10. 1080/ 02701 367. 2020. 17999 16 (2022). 26. Rozmiarek, M. et al. Motivation and eco-attitudes among night runners during the COVID-19 pandemic. Sustainability 14, 1512 (2022). 27. León-Guereño, P., Tapia-Serrano, M. & Sánchez-Miguel, P. The relationship of recreational runners’ motivation and resilience levels to the incidence of injury: A mediation model. PLoS ONE 15, e0231628. https:// doi. org/ 10. 1371/ journ al. pone. 02316 28 (2020). 28. Fokkema, T. et al. Training for a (half-)marathon: Training volume and longest endurance run related to performance and running injuries. Scand. J. Med. Sci. Sports 30, 1692–1704. https:// doi. org/ 10. 1111/ sms. 13725 (2020). 29. Friedrich, M. et al. A comparison of anthropometric and training characteristics between female and male half-marathoners and the relationship to race time. Asian J. Sports Med. 5, 10–20. https:// doi. org/ 10. 5812/ asjsm. 34175 (2014). 30. Knechtle, B., Knechtle, P., Roseman, T. & Senn, O. Sex differences in association of race performance, skin-fold thicknesses, and training variables for recreational half-marathon runners. Percept. Motor Skills 111, 653–668. https:// doi. org/ 10. 2466/ 05. 25. PMS. 111.6. 653- 668 (2010). 31. Knechtle, B. et al. Training habits in recreational half-marathon, marathon/ultra-marathon and 10-KM distance runners (Part A)-Results from the NURMI Study (Step 2). Sci. Rep. 12, 10295 (2022). 32. Tanous, D. et al. Sex differences in training behaviors of 10 km to ultra-endurance runners (Part A)-results from the NURMI Study (Step 2). Int. J. Environ. Res. Public Health 19, 13238. https:// doi. org/ 10. 3390/ ijerp h1920 13238 (2022). 33. Motevalli, M. et al. Sex differences in racing history of recreational 10 km to ultra runners (Part B)-results from the NURMI Study (Step 2). Int. J. Environ. Res. Public Health 19, 13291. https:// doi. org/ 10. 3390/ ijerp h1920 13291 (2022). 34. Wirnitzer, K. et al. Health status of recreational runners over 10-km up to ultra-marathon distance based on data of the NURMI Study Step 2. Sci. Rep. 12, 10295. https:// doi. org/ 10. 1038/ s41598- 022- 13844-4 (2022). 35. Wirnitzer, K. et al. Prevalence in running events and running performance of endurance runners following a vegetarian or vegan diet compared to non-vegetarian endurance runners: The NURMI Study. SpringerPlus 5, 458. https:// doi. org/ 10. 1186/ s40064- 016- 2126-4 (2016). 36. Knechtle, B., Rüst, C. A., Knechtle, P. & Rosemann, T. Does muscle mass affect running times in male long-distance master run- ners?. Asian J. Sports Med. 3, 247–256. https:// doi. org/ 10. 5812/ asjsm. 34547 (2012). 37. Rüst, C., Knechtle, B., Knechtle, P. & Rosemann, T. Similarities and differences in anthropometry and training between recreational male 100-km ultra-marathoners and marathoners. J. Sports Sci. 30, 1249–1257. https:// doi. org/ 10. 1080/ 02640 414. 2012. 697182 (2012). 38. Waskiewicz, Z. et al. What motivates successful marathon runners? The role of sex, age, education, and training experience in polish runners. Front. Psychol. 10, 1671. https:// doi. org/ 10. 3389/ fpsyg. 2019. 01671 (2019). 39. Krouse, R., Ransdell, L., Lucas, S. & Pritchard, M. Motivation, goal orientation, coaching, and training habits of women ultrarun- ners. J. Strength Cond. Res. 1, 2835–2842. https:// doi. org/ 10. 1519/ JSC. 0b013 e3182 07e964 (2011). 40. Ogles, B. & Masters, K. Older vs. younger adult male marathon runners: participative motives and training habits. J. Sport Behav. 23, 2 (2000). 41. Elbe, A., Madsen, C. & Midtgaard, J. A cross-cultural comparison of motivational factors in Kenyan and Danish middle and long distance elite runners. J. Psychol. Afr. 20, 421–427. https:// doi. org/ 10. 1080/ 14330 237. 2010. 10820 394 (2010). 42. Doppelmayr, M. & Molkenthin, A. Motivation of participants in adventure ultramarathons compared to other foot races. Biol. Sport 21, 319–323 (2004). 43. Tian, H., Qiu, Y., Lin, Y., Zhou, W. & Fan, C. The role of leisure satisfaction in serious leisure and subjective well-being: Evidence from Chinese Marathon Runners. Front. Psychol. 11, 581908. https:// doi. org/ 10. 3389/ fpsyg. 2020. 581908 (2020). 44. Marca, A. et al. Marathon progress: Demography, morphology and environment. J. Sports Sci. 32, 524–532. https:// doi. org/ 10. 1080/ 02640 414. 2013. 835436 (2014). 45. Zhang, S., Meng, G., Wang, Y. & Li, J. Study of the relationships between weather conditions and the marathon race, and of mete- orotropic effects on distance runners. Int. J. Biometeorol. 36, 63–68. https:// doi. org/ 10. 1007/ bf012 08915 (1992). 46. Thuany, M., Gomes, T., Rosemann, T., Knechtle, B. & de Souza, R. No trends in the age of peak performance among the best half- marathoners and marathoners in the world between 1997–2020. Medicina 57, 409. https:// doi. org/ 10. 3390/ medic ina57 050409 (2021). 9 Vol.:(0123456789) Scientific Reports | (2023) 13:18083 | https://doi.org/10.1038/s41598-023-45055-w www.nature.com/scientificreports/ 47. Nikolaidis, P., Alvero-Cruz, J., Villiger, E., Rosemann, T. & Knechtle, B. The age-related performance decline in marathon running: The paradigm of the Berlin marathon. Int. J. Environ. Res. Public Health 16, 2022. https:// doi. org/ 10. 3390/ ijerp h1611 2022 (2019). 48. Hoffman, M. Performance trends in 161-km ultramarathons. Int. J. Sports Med. 31, 31–37 (2010). 49. Knechtle, B. & Nikolaidis, P. Physiology and Pathophysiology in Ultra-Marathon Running. Front. Physiol. 9, 634. https:// doi. org/ 10. 3389/ fphys. 2018. 00634 (2018). 50. de Souza, R. et al. Ultramarathon evaluation above 180 km in relation to peak age and performance. BioMed Res. Int. 2022, 1036775. https:// doi. org/ 10. 1155/ 2022/ 10367 75 (2022). 51. Appel, M., Zentgraf, K., Krüger, K. & Alack, K. Effects of genetic variation on endurance performance, muscle strength, and injury susceptibility in sports: A systematic review. Front. Physiol. 12, 694411. https:// doi. org/ 10. 3389/ fphys. 2021. 694411 (2021). 52. Holloszy, J. O. Biochemical adaptations in muscle. Effects of exercise on mitochondrial oxygen uptake and respiratory enzyme activity in skeletal muscle. J. Biol. Chem. 242, 2278–2282 (1967). 53. Buck, K., Spittler, J., Reed, A. & Khodaee, M. Psychological attributes of ultramarathoners. Wilderness Environ. Med. 29, 66–71. https:// doi. org/ 10. 1016/j. wem. 2017. 09. 003 (2018). 54. Knechtle, B. Ultramarathon runners: Nature or nurture?. Int. J. Sports Physiol. Perform. 7, 310–312. https:// doi. org/ 10. 1123/ ijspp.7. 4. 310 (2012). 55. Andersen, J. J. The State of Running 2019. https:// runre peat. com/ state- of- runni ng (2019). 56. Hallam, L. & Amorim, F. Expanding the gap: An updated look into sex differences in running performance. Front. Physiol. 12, 804149–804149. https:// doi. org/ 10. 3389/ fphys. 2021. 804149 (2022). 57. Besson, T. et al. Sex differences in endurance running. Sports Med. https:// doi. org/ 10. 1007/ s40279- 022- 01651-w (2022). 58. Gajda, R. et al. To be a champion of the 24-h ultramarathon race If not the heart … mosaic theory?. Int. J. Environ. Res. Public Health 18, 1–25. https:// doi. org/ 10. 3390/ ijerp h1805 2371 (2021). 59. Nikolaidis, P., Cuk, I., Clemente-Suárez, V., Villiger, E. & Knechtle, B. Number of finishers and performance of age group women and men in long-distance running: Comparison among 10km, half-marathon and marathon races in Oslo. Res. Sports Med. 29, 56–66. https:// doi. org/ 10. 1080/ 15438 627. 2020. 17267 45 (2021). 60. Althubaiti, A. Information bias in health research: Definition, pitfalls, and adjustment methods. J. Multidiscip. Healthc. 9, 211–217. https:// doi. org/ 10. 2147/ JMDH. S1048 07 (2016). Acknowledgements There are no professional relationships with companies or manufacturers who will benefit from the results of the present study. Moreover, this research did not receive any specific grant or funding from funding agencies in the public, commercial, or non-profit sectors. Author contributions K.W. conceptualized and designed the study together with B.K. and C.L. K.W. conducted data analysis and M.M. and D.T. provided statistical expertise. M.M., T.R., K.W., and D.T. drafted the manuscript. T.R., C.L., B.K., M.T., and K.W. critically reviewed it. G.W. provided technical support through data acquisition and data management. All authors have read and agreed to the published version of the manuscript. Competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to B.K. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2023

Differences in race history by distance of recreational endurance runners from The NURMI Study (Step 2).

10-23-2023

Knechtle, Beat,Tanous, Derrick,Thuany, Mabliny,Motevalli, Mohamad,Wirnitzer, Gerold,Leitzmann, Claus,Weiss, Katja,Rosemann, Thomas,Wirnitzer, Katharina

eng

PMC9368712

Citation: Nicolas, M.; Gaudino, M.; Bagneux, V.; Millet, G.; Laborde, S.; Martinent, G. Emotional Intelligence in Ultra-Marathon Runners: Implications for Recovery Strategy and Stress Responses during an Ultra-Endurance Race. Int. J. Environ. Res. Public Health 2022, 19, 9290. https://doi.org/10.3390/ ijerph19159290 Academic Editors: Javier Abián-Vicén and Britton W. Brewer Received: 24 May 2022 Accepted: 26 July 2022 Published: 29 July 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). International Journal of Environmental Research and Public Health Article Emotional Intelligence in Ultra-Marathon Runners: Implications for Recovery Strategy and Stress Responses during an Ultra-Endurance Race Michel Nicolas 1,* , Marvin Gaudino 1 , Virginie Bagneux 2, Gregoire Millet 3 , Sylvain Laborde 4 and Guillaume Martinent 5 1 Laboratory Psy-DREPI (EA 7458), University of Bourgogne Franche-Comté, 21000 Dijon, France; marvin.gaudino@u-bourgogne.fr 2 LPCN, Université de Caen Normandie, 14032 Caen, France; virginie.bagneux@unicaen.fr 3 SSUL, Institute of Sport Sciences, Faculty of Biology and Medicine, University of Lausanne, CH-1011 Lausanne, Switzerland; gregoire.millet@unil.ch 4 Department of Performance Psychology, Institute of Psychology, German Sport University Cologne, 50923 Cologne, Germany; s.laborde@dshs-koeln.de 5 Laboratory on Vulnerabilities and Innovation in Sport, University of Lyon 1, 69367 Lyon, France; guillaume.martinent@univ-lyon1.fr * Correspondence: michel.nicolas@u-bourgogne.fr Abstract: The aim of this research was to investigate the role of trait emotional intelligence (EI) in recovery stress states in a mountain ultra-marathon (MUM) race. Recovery stress states of 13 finishers were assessed before, during, and immediately after the end of an extreme MUM, whereas emotional intelligence was assessed 2 days before the MUM race. Temporal evolutions of recovery stress states were examined. Stress states increased after the race whereas recovery states decreased in all participants. In addition, recovery states were influenced by the trait EI level assessed before the competition. Results supported the hypothesis that trait EI tends to have a positive effect by boosting recovery strategies. In this perspective, trait EI could have a protective role against stress and improve pre-competition mental preparation. High scores of trait EI (in comparison to low scores of trait EI) could have helped athletes to increase recovery states in order to improve their psychological adaptation to one of the most difficult races in the world. Keywords: emotional intelligence; recovery stress states; mountain ultra-marathon 1. Introduction Extreme sports situations demand multidimensional psychological adaptive responses which could depend on recovery stress states [1], as well as individual factors such as emotional intelligence [2]. Biopsychological perspective of recovery and stress [3], em- braces physical and biopsychosocial dimensions of both stress and recovery to indicate the extent to which someone is physically and/or mentally stressed, as well as whether that person is capable of using individual strategies for recovery and which strategies are used. EI refers to a form of intelligence which aims to capture individual differences in interpersonal and intrapersonal emotional functioning [4–6]. Its potential contribution in sporting competitions has been demonstrated and is considered to be a key factor in improving individual adaptation, notably with regard to the stress process [2]. During the last few decades, the recovery process has been associated with stress states to explain how athletes may be better able to tolerate and buffer stress from training and competition [7]. Whereas the relationship between stress and EI has been largely documented, no study has investigated the relationship between EI and recovery. The aim of this paper is to evaluate the involvement of EI in recovery stress states. Int. J. Environ. Res. Public Health 2022, 19, 9290. https://doi.org/10.3390/ijerph19159290 https://www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2022, 19, 9290 2 of 13 Extreme situations are demanding and challenging. They impose on an individual the need to cope with exceptional physical or psychosocial circumstances that require adaptive responses that engage personal resources which could be overwhelmed [8]. A mountain ultra-marathon (MUM) race could be considered one of the most extreme sport- ing situations after polar expeditions [9] because it implies a complex and multidimensional adaptation defined by the dynamic impact between environmental and personal constraints and resources (i.e., physical, psychological, and social) on adjustment [8,10]. During a MUM, athletes are exposed to a variety of stressors and have to run for extended periods over long distances and dangerous terrain with changing altitudes in an uncertain and risky environment [11]. Exposure to these stressful environmental and climatic conditions tends to push participants to draw on their own resources in order to perform beyond their ordinary limits [12]. MUM is by definition a playing field for in situ ecological research investigating psychological impairments that are mirrored in multidimensional psycho- logical processes, such as emotional disturbances [13] and increases/decreases in recovery stress states [14]. These impairments were also observed during the month following the competition, highlighting that ultra-endurance sports are challenging situations with long-term repercussions [14,15]. According to Lazarus and Folkman, the seminal model of psychological stress (1984), extreme situations can exacerbate stress states [8]. Beyond a certain point, any effort to manage an excessive stress state could engage personal resources and in turn cause their potential consumption if the recovery process is not implemented [16]. However, a certain level of perceived stress is an integral part of psychological adaptation [17]. The objective is no longer to annihilate stress but to attempt to reach a balance between stress state and personal resources. The recovery process actually represents a core concept in investigating how to deal with and buffer the stress state because it helps to protect, build, refill or restore personal resources [7,18]. Recovery is defined as a multilevel process used to tolerate stress and to re-establish performance abilities and psychological and physical strength in order to optimize situational conditions [7]. Thus, recovery is based on proactive and self-initiated activities [18–20]. In the last decade, there has been an increased interest in the investigation of the interrelated dynamics of recovery stress states in order to better understand the psycho- logical adaptations in extreme situations. The theoretical model of the recovery stress process [3,19] leads to a joint measurement of the extent to which an individual is fre- quently and multidimensionally stressed (social, emotional, physical, and behavioral) and its recovery-associated activities/states. The objective is to reach an individual biopsychoso- cial balance in order to counterbalance the negative effects of stress, help to adjust to the situation, and to achieve a continuous high-level performance [3]. Results from individuals’ exposure to spatial simulations [21], polar stations, i.e., wintering in Antarctica [22], and extreme sports [14] have provided strong evidence of the importance of considering the recovery stress process. Unbalanced recovery stress states (i.e., increased stress states and decreased recovery states) can lead to dysfunctional outcomes such as chronic fatigue and concomitant overtraining, and psychological exhaustion [1,20]. Consequently, the participant’s adaptation to sports training and competition is compromised [1,19]. Results of previous studies on ultra-endurance races showed that participants have simultaneously reported an increase in stress states and a decrease in recovery states mirrored, notably, in the emotional exhaustion dimension [14,15]. The repercussions could be observed up to four weeks after the race, highlighting the long-term impact of a stressful event on the recovery stress states [15]. In particular, evolutions in the recovery stress states experienced by MUM runners in the month following a demanding MUM race have been characterized by a significant linear increase in recovery and a linear decrease in stress states [15]. Results show that the harder the situation is, the longer the need to evaluate and manage recovery stress states. However, even in extreme environments, recovery stress states are not always unbal- anced [22,23], suggesting that personal resources could be sustained. Results from previous research showed that recovery stress states could be modulated according to individual Int. J. Environ. Res. Public Health 2022, 19, 9290 3 of 13 difference factors such as perceived stress and perceived control [19,22,24]. Specifically, these studies have shown that perceived stress was positively linked to psychological, physiological, and social stress responses whereas perceived control was positively linked to recovery strategies [19,22,24]. These results provided supporting evidence that indi- vidual cognitive resources were involved in managing the stress process [10,17]. These results corroborate the cognitive–motivational–relational theory [10] and emphasize that the interaction between the person and the environment is mediated by the degree to which a situation is appraised as stressful and controllable. These findings suggest that individual differences could help explain the differences in psychological responses within a challenging situation [25]. Among individual variables identified as factors influencing stress management, EI could determine an athlete’s ability to handle psychological stress and also facilitate physical and psychological recovery [26]. The theoretical nature of EI-related constructs remains assigned to a wide array of concepts and models [27]. EI provides an interesting framework for assessing individual differences with regard to how individuals identify, express, understand, regulate and use their own and others’ emotions to ultimately guide their thinking and actions [4,5]. Among the several theoretical frameworks focusing on EI [5,28–30], the present study was grounded within the trait model of EI [29,30] based on the rationale that EI was conceptualized in the present study as an individual difference variable. The trait model [29] defines EI as a lower-order personality trait that is mainly evaluated using a self-report measure [30]. A systematic review [2] concluded that EI had a protective role with regard to the stress process in athletes. For example, EI was found to be associated with the use of more efficient coping strategies (i.e., task-oriented coping) to manage stress [31]. Furthermore, Laborde, Dosseville, Guillén, and Chavez [32] indicated that EI positively predicted per- ceived control, coping (e.g., task-oriented coping strategies, coping effectiveness), and performance satisfaction. In addition, numerous studies support the idea that EI is a key factor in improving individual adaptation [33,34]. Individuals with high EI would be more competent in coping with challenges and would perceive less stress [35] and more well-being [36]. EI may help to explain how stress is physiologically better tolerated and buffered by certain individuals [37]. Previous research within the context of MUM race also supported the notion that EI is positively associated with pleasant emotional states [38]. The connection between EI and pleasant emotions could be crucial to depicting the relationship between EI and the recovery process. In her broaden-and-build theory, Fredrickson [39] posited that experiences of pleasant emotions broaden people’s momentary thought–action repertoires in a way that serves to build their enduring personal resources and subsequent emotional well-being. Several empirical studies provide evidence supporting this theoreti- cal approach [40] including studies in sports settings e.g., [41]. Consequently, EI could be expected to play a major role in boosting recovery processes and helping to protect, build, refill or restore personal resources when individuals have faced stressful situations. Based on previous studies in extreme situations, the interplay between recovery and stress states has been shown to play a major role in the psychological adaptation processes. However, some gaps remain in the research. Specifically, the role of trait EI in the recovery process has not yet been investigated whereas promising theoretical support exists for the link between EI and recovery [39]. Consequently, the present study aims to provide experimental evidence regarding the psychological adaptation in MUM runners by investigating the recovery stress process and the relationship between recovery stress states and trait EI, especially before, during, and after one of the most extreme MUM races. This study could provide insights on how stress states could be tolerated and/or buffered in MUM in regard to recovery strategies [7]. Considering previous results on recovery stress states in ultra-marathons [14,15], it was hypothesized that (1) stress states would increase during and after the race compared to pre-race, whereas (2) recovery would decrease in the same time evolution. Furthermore, based on previous research on EI [2], we hypothesized Int. J. Environ. Res. Public Health 2022, 19, 9290 4 of 13 that: (3) athletes with low trait EI would report lesser recovery and higher stress states during the MUM race than athletes with a high trait EI. 2. Methods 2.1. Participants Thirteen athletes running an extreme MUM event (2 women and 11 men), aged from 29 to 52 years old (Mage = 40.08 yrs, SDage = 6.76), voluntarily participated in this study. Initially, 17 participants were recruited through an announcement for this study, which was part of a larger research project that focused on the physiological consequences of this race, which is considered to be the most challenging mountain ultra-marathon in the world. Of the 17 participants initially enrolled in the study, 13 completed the race (inclusion criterion) and thus constituted our sample for this study. The MUM was the Tor des Géants® (TdG) and consisted of a semi-self-sufficiency race where the runners covered a total of 338 km with a cumulated altitude of 30,959 m of positive elevation under changing climate conditions (temperatures between −9 and 15 ◦C). For the rest, rescue, and refreshment points, runners were able to rely on the seven base camps spaced approximately 50 km apart. On average, participants accomplished the race in 132.67 h (SD = 13.16). All participants signed a consent form stipulating their right to withdraw from the experiment at any time without prejudice. This study was approved by the local ethics committee in accordance with the Declaration of Helsinki (amended 2013). 2.2. Measures 2.2.1. Brief Emotional Intelligence Scale (BEIS-10) The BEIS-10 [42] is based on both the EI model of Salovey and Mayer [6] and the work of Lane et al. [43]. The BEIS-10 was administered to athletes to measure their trait EI using a 6-point Likert scale (1 = never to 6 = always). The 10-item version is a short and efficient measure to quickly assess an individual’s perception of the extent to which they appraise, regulate, and use emotions. For this study, the internal consistency of the BEIS-10 was 0.84. 2.2.2. Recovery Stress States (RestQ-36-R-Sport) Based on the original Rest-Q for athletes [7], the French version of the RestQ-36-R- Sport questionnaire [18] was used to quickly assess the multidimensional nature (physical, emotional, behavioral, and social) of recovery and stress states. This questionnaire was developed to quickly measure the frequency of current stress along with the frequency of recovery using a 6-point Likert scale (1 = never to 6 = always). Higher scores in the stress responses reflected an intense and elevated perceived stress state. Higher scores in the recovery strategies reflected a high frequency of using numerous recovery strategies. Pre- TdG instructions given to participants for the completion of the RestQ-36-R-sport referred to «the 3 last days» whereas Per- and Post-TdG instructions referred to «the 3 last hours» in order to evaluate the psychological states during the MUM. The internal consistency for total stress and recovery scores across the several measurement times ranged from 0.53 to 0.98. Cronbach alpha tends to increase with an increase in the number of participants [44], leading researchers to suggest a cut-off value of 0.60 for a low sample size [45]. Other researchers prefer the use of the raw average inter-item correlation (AIIC) as a statistical marker of internal consistency. For this, a rule of thumb is offered by Clark and Watson [44] who recommend AIIC scores higher than 0.15. In the present study, all the AIIC scores were higher than 0.15. 2.3. Procedure The thirteen participants rated their recovery stress states and trait EI on self-report questionnaires in the 3 h before the race (Pre-MUM). Secondly, while the majority of research on emotion, EI, or recovery stress states conducted in ultra-endurance sports has focused on the differences pre- and post-race e.g., [13–15], participants in the present investigation also rated their recovery stress states during the race. This measure was completed at the Donnas camp (located at the mid-race point: Per-MUM) after athletes had run 155 km with an Int. J. Environ. Res. Public Health 2022, 19, 9290 5 of 13 average running time of 51 h (SD = 3.41). Thirdly, the athletes completed the RestQ-36-R-sport questionnaire for the last time during the three-hour period after finishing the race. 2.4. Statistical Analysis Shapiro–Wilk and Levene’s tests were used to verify data normality and the homo- geneity of variances at each time point. Trait EI data were analyzed: (a) using correlational analysis with recovery stress states scores on the total sample; and (b) by dividing its scores into either a high or low group based on the median value, a common practice within the literature [46–48]. While dichotomization is sometimes criticized in the literature [49], recent re-evaluations have shown that this practice is a robust, reliable, and appropriate statistical analysis when independent variables are uncorrelated [46–48]. A median split was therefore used to dichotomize participants based on their scores of trait EI. The re- sults from our sample showed that the low trait EI scores and high trait EI scores were uncorrelated (p = 0.07) and provided evidence supporting the use of a median split in the present study. Literature has also shown that conducting a median split does not increase the likelihood of a Type I error [47]. Additionally, given that the scores for all factors at the different time measures were normally and homogeneously distributed, we conducted a set of multivariate analyses of variance (MANOVA) with repeated measures to test: (1) The effect of time on recovery stress states; (2) the effect of trait EI groups (high EI vs. low EI); and (3) the effect of the interaction of trait EI-groups * time. Follow-up univariate one-way ANOVAs were conducted in order to target significant differences detected using MANOVA. Pairwise comparisons (post hoc) were conducted using Tukey’s HSD. 3. Results 3.1. Descriptive Analyses Descriptive statistics for recovery stress states and trait EI are shown in Table 1. Results of correlational analysis for the total sample showed that recovery was negatively correlated with stress state (r = −0.59, p < 0.05) whereas trait EI was not significantly correlated with stress state and recovery state. Table 1. Descriptive statistics and inter-correlations for recovery stress states and EI scores in high trait EI (n = 6) and low trait EI (n = 7). Recovery Stress Emotional Intelligence Total sample Recovery - Stress −0.59 * - Emotional Intelligence 0.41 −0.25 - M 3.57 2.70 44.31 SD 0.47 0.30 6.79 High trait EI group Recovery - Stress −0.91 * - Emotional Intelligence 0.79 −0.90 * - M 4.26 2.61 49.96 SD 0.17 0.14 3.03 Low trait EI group Recovery - Stress −0.56 - Emotional Intelligence 0.18 0.14 - M 3.37 2.85 39.57 SD 0.16 0.13 5.86 Note. * p < 0.05. Based on the median value (Me = 45), a significant difference between the high trait EI and low trait EI groups was observed in this study, t(11) = 4.53, p = 0.008, d = 2.23. The Int. J. Environ. Res. Public Health 2022, 19, 9290 6 of 13 high EI-level group contained 6 athletes (Mage = 50.12; SD = 2.71) while the low EI group contained 7 athletes (Mage = 39.19; SD = 5.35). High trait EI and low trait EI mean scores were not significantly correlated (p > 0.05), encouraging the use of a median split. For the high trait EI group, correlations revealed that recovery was negatively correlated with stress state (r = −0.91, p < 0.05) and stress state was negatively correlated with EI (r = −0.90, p < 0.05). For the low trait EI group, no significant correlation was found. 3.2. Stress State Figure 1 presents changes in stress scores during the TdG®. Firstly, the effect of the EI group on stress state was not significant, F(1, 11) = 0.74, p = 0.408 (M low EI = 2.77, SD = 0.13; M high EI = 2.62, SD = 0.14). Secondly, stress state scores changed over time, F(2, 22) = 5.19, p = 0.014, ηp2 = 0.28. Tukey post hoc tests revealed significant increases, specifically between pre-MUM (M = 2.57, SD = 0.11) and post-MUM (M = 2.88, SD = 0.12, p = 0.017, d = 3.04) and between per-MUM (M = 2.62, SD = 0.11) and post-MUM (p = 0.043, d = 2.35). Thirdly, no significant interaction was observed, showing that stress scores were not influenced by the athletes’ EI levels throughout the race, Wilk’s λ = 0.97, F(2, 22) = 0.269, p = 0.767. Figure 1. Total stress scores during MUM in high and low EI groups. 3.3. Recovery State Figure 2 presents the changes in recovery states during the TdG®. Firstly, results showed no significant effect of the trait EI group on recovery scores, F(1, 11) = 3.77, p = 0.078 (M low EI = 3.31, SD = 0.16; M high EI = 3.78, SD = 0.17). Secondly, the effect of time on recovery scores was significant, F(2, 22) = 7.50, p = 0.003, ηp2 = 0.45. Tukey HSD post hoc tests showed that the score for recovery decreased between pre-MUM (M = 3.81, SD = 0.12) and per-MUM (M = 3.46, SD = 0.09, p = 0.02, d = 3.11) and between pre-MUM and post-MUM (M = 3.43, SD = 0.09, p = 0.003, d = 3.58). Thirdly, the interaction effect of trait EI group X time on recovery was significant, F(2, 22) = 12.21, p = 0.0003, ηp2 = 0.53. The low trait EI group reported a lower score for recovery (M = 3.26, SD = 0.60) compared to the high trait EI group (M = 4.27, SD = 0.39, p = 0.004, d = 4.35) at Pre-MUM and this effect was non-significant at per-MUM and Post- MUM. In addition, only recovery scores in the high trait EI group decreased over time. Specifically, recovery scores decreased between pre-MUM (M = 4.27, SD = 0.19) and both per-MUM (M = 3.61, SD = 0.16, p = 0.0009, d = 3.11) and between pre-MUM and post- MUM (M = 3.46, SD = 0.19, p = 0.0002, d = 3.11) among the high trait EI group whereas no significant difference was observed among the low trait EI group. Finally, all participants during the race reported high recovery levels compared to stress states, F(2, 48) = 7.74, p = 0.001, ηp2 = 0.24 (Table 2). A Tukey’s HSD post hoc test confirmed that all recovery scores were higher than the stress scores, either before, during, or after the race (Table 1). Int. J. Environ. Res. Public Health 2022, 19, 9290 7 of 13 Table 2. Results of the MANOVA analysis for recovery stress states in the low EI and high EI groups. Low Emotional Intelligence (n = 7) High Emotional Intelligence (n = 6) Tukey’s HSD Interpretation Pre-MUM (1) Per-MUM (2) Post-MUM (3) Pre-MUM (4) Per-MUM (5) Post-MUM (6) M(SD) M(SD) M(SD) M(SD) M(SD) M(SD) Recovery (R) 3.30 (0.23) * 3.32 (0.11) * 3.48 (0.14) * 4.02 (0.23) *µ 3.57 (0.11) * 3.40 (0.14) * EI-level effect F(1, 11) = 5.23, p = 0.04, ηp2 = 0.322 R in high EI > R in low EI Time effect F(2, 22) = 9.10, p = 0.001, ηp2 = 0.452 R at Pre-MUM > R at Per- and Post-MUM EI level * Time F(2, 22) = 12.53, p = 0.0002, ηp2 = 0.532 1 < 4; 4 > 5–6 Stress (S) 2.70 (0.15) 2.84 (0.18) 3.01 (0.16) 2.45 (0.16) 2.56 (0.19) 2.82 (0.17) EI-level effect F(1, 11) = 1.53, p = 0.24, ηp2 = 0.122 NS Time effect F(2, 22) = 4.36, p = 0.03, ηp2 = 0.283 S at Pre-MUM < S at Post-MUM EI level * Time F(2, 22) = 0.07, p = 0.93 NS Note. * Mean of recovery significantly higher than mean of stress. NS = non-significant; µ Mean of Pre-MUM recovery in high EI group significantly higher than Mean of pre-MUM recovery in low EI group. Int. J. Environ. Res. Public Health 2022, 19, 9290 8 of 13 Figure 2. Total recovery score during MUM in high and low trait EI groups. Notes. * Mean of recovery for the high trait EI group significantly higher than the mean of recovery for the low trait EI group. ↕ Mean of recovery for the high trait EI group significantly higher at pre-MUM than per- and post-MUM and then the mean of recovery for the low trait EI group. 4. Discussion The purpose of this study was to examine the time courses of recovery stress states before, during, and after one of the most challenging MUM races. This study also con- tributes to identifying how individuals’ high versus low trait EI affects their recovery stress states. Results showed an imbalance between recovery stress states, highlighting an increase in the stress states and a decrease in the recovery states. This confirms the first two hypotheses and reaffirms that running a MUM race is a psychologically demanding situation. However, a particularly interesting finding concerns the differences in recovery states based on trait EI scores. As expected in regard to the third hypothesis, athletes with higher trait EI scores reported higher recovery states compared to athletes with lower trait EI scores. Our findings support the positive role of trait EI on an individual’s ability to cope with challenging situations [38]. Consistent with previous research on extreme situations [14,21], the results of the present study tend to reaffirm that the stress state is increased over time regarding an ultra-endurance race. Specifically, stress states significantly increased immediately after the race compared to the start of the race, while no significant variation of stress states was observed between pre- and per-MUM. Even if athletes tended to experience a constant stress state during the first part of the race, prolonged and repeated exposure to stressful environmental conditions increased the stress state after the race. It is well established that runners completing a MUM have to push their resources beyond ordinary limits [12] to cope with the severe demands placed upon them, such as physical repercussions (e.g., fatigue, sleep deprivation) [50], emotional disturbances [51], and social stress [14]. Athletes reported higher scores of recovery than stress at every time point. These results suggest that the recovery strategies were frequently used to buffer stress states. It seems that athletes who finished the race tended to efficiently manage their resources throughout the race. Based on their higher scores of recovery compared to stress, they had to prioritize recovery to ensure their performance, health, and well-being [52] an Int. J. Environ. Res. Public Health 2022, 19, 9290 9 of 13 effective biopsychosocial adjustment during the race. However, it is also noteworthy that recovery decreased over the duration of the race, reflecting the difficulty of the runners in maintaining and using strategies to preserve physical and psychological resources throughout the race. The impairment of this balance may be explained by the fact that individuals have to draw on their own resources to achieve their goals over a long time [23]. In line with the findings for prolonged exposure to stressors, where recovery decreases and stress increases simultaneously [21], continuous effort in demanding situations could lead to exhaustion of psychological resources and in turn could prevent the use of recovery strategies [14]. Based on the median split approach, two groups were distinguished with significantly different trait EI scores. Although a significant negative correlation was observed between recovery and stress scores among the total sample, the correlation between recovery and stress scores only remained significant among participants belonging to the high trait EI group. As expected, recovery and stress states were significantly related in individuals reporting greater levels of trait EI. As suggested by Jeffrey [26], recovery stress states could be more balanced in an athlete with a high trait EI in order to find an optimal recovery within any challenge. Our results agree with this statement: Athletes who reported a high trait EI reported more recovery strategies (i.e., active, passive, and proactive), which could provide them with better control of their stress states before the race. Surprisingly, scores for stress states were not statistically different between the high and low trait EI groups. Literature suggested that EI was associated with significantly lower stress scores in stressful situations (i.e., competition), highlighting the protective role of trait EI within stressful events [2]. However, our results do not confirm this literature. This could be explained by the potentially positive impact of the stress states, which may lead to psychological adaptation and coping within stressful environmental conditions. Stressful conditions actually lead to an increase in the stress responses in ultra-endurance athletes [13–15]. However, a certain level of stress state may be necessary for a successful psychological adaptation, as long as the recovery is sufficient to help mobilize personal resources [7,17]. Stress is therefore no longer considered to be a negative consequence because it supports adjustment. Thus, the goal is not to eliminate the stress state per se but rather to use it, while maintaining high scores of recovery, to buffer, manage, and regulate stress. In this way, trait EI could play a protector role in stress through cognitive appraisals in helping individuals evaluate situations as being challenging [53]. Reaching a balance between stress state and recovery state would be a particularly relevant strategy to promote adjustments in a MUM situation. In addition, the stress state experienced by athletes could be considered as eustress to help further increase and mobilize their personal resources in a constant adjustment to the extreme situation [23]. As a reminder, all participants in this study were part of the 55% of finishers, suggesting that an optimal recovery stress state was observed. As expected, athletes who reported a high trait EI showed higher scores of recovery before the race. In other words, these runners tended to be more able to protect, build, refill or restore their personal resources compared to the low trait EI runners. This finding highlights the positive role of trait EI on the passive, active and proactive approaches to recovery, in addition to its positive influence on the use of several psychological skills, such as self-talk, imagery, or activation [43]. Our findings at the outset of the competi- tion highlight that trait EI would help to optimize psychological processes by buffering stressor effects [28] and boosting personal resources (physical, emotional, behavioral, and social) [18]. However, an alternative explanation could be provided for the fact that only recovery scores in the high trait EI group decreased over time. High trait EI participants could have a better introspective sense of their internal state, whereas low trait EI partic- ipants may not have as fine-tuned a sense of their internal states and therefore did not report changes in their recovery states over time. Int. J. Environ. Res. Public Health 2022, 19, 9290 10 of 13 5. Limits Due to limited access to the elite athlete population and finishers in these ecologically extreme situations, the small sample size of the present study represents a limitation for generalization and further analyses such as regression models. In this line, it would have been interesting to have more information on the characteristics of the participants to better understand our results by considering other potential biological, psychological, and sociological factors. For example, age, gender, and experience, but also training periodization, fitness, nutrition, and type of recovery practices may be involved in the development of recovery and stress states. Further research should consequently endeavor to recruit larger sample sizes and, more specifically, to go beyond the global EI score used in this study. This score was calculated from 5 distinct sub-dimensions: Appraisal of one’s own emotions, appraisal of others’ emotions, regulation of one’s own emotions, regulation of others’ emotions, and use of emotions. Previous research has revealed the relevance of investigating these sub-dimensions independently given that they could be differently associated with psychological responses, such as emotions [38,43]. Therefore, future research with a larger sample and different EI or emotional regulation questionnaires, e.g., CERQ [54]; PEC, [55]; TEIQue, [56] could lead to a more specific understanding of the respective influence of each dimension of trait EI on recovery strategies in stressful situations. 6. Practical Applications This study gives insight into the role of trait EI in the recovery stress states during a MUM race. Runners should be aware that ultra-endurance races lead to substantial changes in recovery stress states and that trait EI could help them to improve their preparation for a MUM race. The ability to balance recovery stress states is essential in preventing pathogenic psychological outcomes but also for the development and maintenance of skilled performance, health, and well-being [1,19]. The positive association between trait EI and the recovery process could also help to improve pre-competitive resources and mental preparation. Coaches, athletes, and psychological counselors are concerned by this result because they could conduct specific interventions in order to improve the trait EI in athletes and in turn the balance between recovery strategies and stress states. As shown in previous studies, it is possible to improve trait EI [5,57]. EI interventions [58] should first focus on the understanding of the emotional information in order to lead individuals to be aware and accumulate sufficient knowledge to transform this into practice (i.e., recovery strategies) to increase trait EI. 7. Conclusions Despite the limitations of this study, investigating the role of trait EI in MUM athletes should provide a better understanding of the balance of recovery and stress states. An added value of this study was to indicate that high trait EI was linked to higher scores of recovery before the race, suggesting that such athletes tend to be better prepared to cope with MUM. Athletes, coaches, and practitioners in sports psychology could develop trait EI in order to facilitate the use of recovery strategies and optimize personal resources in competition. Author Contributions: Conceptualization, M.N. and G.M. (Gregoire Millet); methodology, M.N. and M.G.; formal analysis, M.G. and G.M. (Guillaume Martinent); investigation, M.G.; writing— original draft preparation, M.N. and M.G. writing—review and editing, M.N., M.G., V.B., G.M. (Gregoire Millet), S.L. and G.M. (Guillaume Martinent). All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: The study was conducted in accordance with the Declaration of Helsinki, and approved by the Institutional Review Board of the University of Burgundy. Int. J. Environ. Res. Public Health 2022, 19, 9290 11 of 13 Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: The data are available and can be sent by the corresponding author. Acknowledgments: We would like to thank the participants of the study who were fully committed during their preparation time prior to the MUM. Conflicts of Interest: The authors declare no conflict of interest. References 1. Nicolas, M.; Gaudino, M.; Vacher, P. Stress and recovery in extreme situations. In Sport, Recovery, and Performance: Interdisciplinary Insights; Kellmann, M., Beckmann, J., Eds.; Routledge: Abingdon, UK, 2018. 2. Laborde, S.; Dosseville, F.; Allen, M.S. Emotional intelligence in sport and exercise: A systematic review. Scand. J. Med. Sci. Sports 2015, 26, 862–874. [CrossRef] [PubMed] 3. Kellmann, M.; Bertollo, M.; Bosquet, L.; Brink, M.; Coutts, A.J.; Duffield, R.; Erlacher, D.; Halson, S.L.; Hecksteden, A.; Heidari, J.; et al. Recovery and Performance in Sport: Consensus Statement. Int. J. Sports Physiol. Perform. 2018, 13, 240–245. [CrossRef] [PubMed] 4. Mayer, J.D.; Salovey, P. What is emotional intelligence. In Emotional Development, Emotional Literacy, and Emotional Intelligence; Salovey, P., Sluyter, D., Eds.; Basic Books: New York, NY, USA, 1997; pp. 3–31. 5. Mikolajczak, M. Going Beyond The Ability-Trait Debate: The Three-Level Model of Emotional Intelligence. E-J. Appl. Psychol. 2010, 5, 25–31. [CrossRef] 6. Salovey, P.; Mayer, J.D. Emotional intelligence. Imagin. Cogn. Personal. 1990, 9, 185–211. [CrossRef] 7. Kellmann, M.; Kallus, K.W. Recovery-Stress Questionnaire for Athletes: User Manual; Human Kinetics: Champaign, IL, USA, 2001; Volume 1. 8. Nicolas, M.; Martinent, G.; Suedfeld, P.; Gaudino, M. Assessing psychological adaptation during polar winter-overs: The isolated and confined environments questionnaire (ICE-Q). J. Environ. Psychol. 2019, 65, 101317. [CrossRef] 9. Millet, G.P.; Millet, G.Y. Ultramarathon is an outstanding model for the study of adaptive responses to extreme load and stress. BMC Med. 2012, 10, 77. [CrossRef] 10. Lazarus, R.S. Emotion and Adaptation; Oxford University Press: Oxford, UK, 1991. 11. Lièvre, P.; Gautier, A. Les registres de la logistique des situations extrêmes: Des expéditions polaires aux services d’incendies et secours. Manag. Avenir 2009, 24, 196–216. [CrossRef] 12. Pearson, H. Freaks of nature? Nature 2006, 444, 1000–1001. [CrossRef] [PubMed] 13. Lane, A.M.; Wilson, M. Emotions and trait emotional intelligence among ultra-endurance runners. J. Sci. Med. Sport 2011, 14, 358–362. [CrossRef] [PubMed] 14. Nicolas, M.; Banizette, M.; Millet, G. Stress and recovery states after a 24 h ultra-marathon race: A one-month follow-up study. Psychol. Sport Exerc. 2011, 12, 368–374. [CrossRef] 15. Gaudino, M.; Martinent, G.; Millet, G.Y.; Nicolas, M. The time courses of runners’ recovery-stress responses after a mountain ultra-marathon: Do appraisals matter? Eur. J. Sport Sci. 2019, 19, 876–884. [CrossRef] [PubMed] 16. Geuna, S.; Brunelli, F.; Perino, M.A. Stressors, stress and stress consequences during long-duration manned space missions: A descriptive model. Acta Astronaut. 1995, 36, 347–356. [CrossRef] 17. Lazarus, R.S.; Folkman, S. Coping and adaptation. In The Handbook of Behavioral Medicine; Wiley: Hoboken, NJ, USA, 1984; pp. 282–325. 18. Nicolas, M.; Vacher, P.; Martinent, G.; Mourot, L. Monitoring stress and recovery states: Structural and external stages of the short version of the RESTQ sport in elite swimmers before championships. J. Sport Health Sci. 2016, 8, 77–88. [CrossRef] 19. Kellmann, M. Preventing overtraining in athletes in high-intensity sports and stress/recovery monitoring. Scand. J. Med. Sci. Sports 2010, 20, 95–102. [CrossRef] [PubMed] 20. Martinent, G.; Decret, J.-C.; Filaire, E.; Isoard-Gautheur, S.; Ferrand, C. Evaluations of the Psychometric Properties of the Recovery-Stress Questionnaire for Athletes among a Sample of Young French Table Tennis Players. Psychol. Rep. 2014, 114, 326–340. [CrossRef] 21. Nicolas, M.; Weiss, K. Stress and recovery assessment during simulated microgravity: Effects of exercise during a long-term head-down tilt bed rest in women. J. Environ. Psychol. 2009, 29, 522–528. [CrossRef] 22. Nicolas, M.; Suedfeld, P.; Weiss, K.; Gaudino, M. Affective, Social, and Cognitive Outcomes During a 1-Year Wintering in Concordia. Environ. Behav. 2015, 48, 1073–1091. [CrossRef] 23. Nicolas, M.; Gushin, V. Stress and Recovery Responses during a 105-day Ground-based Space Simulation: Stress and recovery during mars 105. Stress Health 2014, 31, 403–410. [CrossRef] 24. Vacher, P.; Nicolas, M.; Mourot, L. Recovery and stress states: Did perceived control and goal attainment matters during tapering period? Int. J. Sport Psy. 2019, 50, 469–484. 25. Bartone, P.T.; Krueger, G.P.; Roland, R.R.; Sciarretta, A.A.; Bartone, J.V.; Johnsen, B.H. Individual Differences in Adaptability for Long Duration Space Exploration Missions; NASA Johnson Space Center: Houston, TX, USA, 2017. 26. Jeffreys, I. A multidimensional approach to enhancing recovery. Strength Cond. J. 2005, 27, 78–85. [CrossRef] Int. J. Environ. Res. Public Health 2022, 19, 9290 12 of 13 27. Hughes, D.; Evans, T.R. Putting ‘Emotional Intelligences’ in Their Place: Introducing the Integrated Model of Affect-Related Individual Differences. Front. Psychol. 2018, 9, 2155. [CrossRef] [PubMed] 28. Mayer, J.; Salovey, P.; Caruso, D.R. Emotional Intelligence: Theory, Findings, and Implications. Psychol. Inq. 2004, 15, 197–215. [CrossRef] 29. Petrides, K.V.; Furnham, A. Trait emotional intelligence: Behavioural validation in two studies of emotion recognition and reactivity to mood induction. Eur. J. Personal. 2003, 17, 39–57. [CrossRef] 30. Petrides, K.V.; Mikolajczak, M.; Mavroveli, S.; Sanchez-Ruiz, M.-J.; Furnham, A.; Pérez-González, J.-C. Developments in Trait Emotional Intelligence Research. Emot. Rev. 2016, 8, 335–341. [CrossRef] 31. Laborde, S.; You, M.; Dosseville, F.; Salinas, A. Culture, individual differences, and situation: Influence on coping in French and Chinese table tennis players. Eur. J. Sport Sci. 2012, 12, 255–261. [CrossRef] 32. Laborde, S.; Dosseville, F.; Guillen, F.; Chávez, E. Validity of the trait emotional intelligence questionnaire in sports and its links with performance satisfaction. Psychol. Sport Exerc. 2014, 15, 481–490. [CrossRef] 33. Laborde, S.; Brüll, A.; Weber, J.; Anders, L.S. Trait emotional intelligence in sports: A protective role against stress through heart rate variability? Pers. Individ. Differ. 2011, 51, 23–27. [CrossRef] 34. Matthews, G.; Zeidner, M.; Roberts, R.D. Emotional Intelligence: Science and Myth; MIT Press: Cambridge, MA, USA, 2004. 35. Mikolajczak, M.; Roy, E.; Luminet, O.; Fillée, C.; de Timary, P. The moderating impact of emotional intelligence on free cortisol responses to stress. Psychoneuroendocrinology 2007, 32, 1000–1012. [CrossRef] [PubMed] 36. Slaski, M.; Cartwright, S. Emotional intelligence training and its implications for stress, health and performance. Stress Health 2003, 19, 233–239. [CrossRef] 37. Laborde, S.; Lautenbach, F.; Allen, M.S.; Herbert, C.; Achtzehn, S. The role of trait emotional intelligence in emotion regulation and performance under pressure. Pers. Individ. Differ. 2013, 57, 43–47. [CrossRef] 38. Nicolas, M.; Martinent, G.; Millet, G.; Bagneux, V.; Gaudino, M. Time courses of emotions experienced after a mountain ultra-marathon: Does emotional intelligence matter? J. Sports Sci. 2019, 37, 1831–1839. [CrossRef] [PubMed] 39. Fredrickson, B.L. The role of positive emotions in positive psychology: The broaden-and-build theory of positive emotions. Am. Psychol. 2001, 56, 218–226. [CrossRef] [PubMed] 40. Fredrickson, B.L.; Branigan, C. Positive emotions broaden the scope of attention and thought-action repertoires. Cogn. Emot. 2005, 19, 313–332. [CrossRef] 41. Martinent, G.; Nicolas, M. Temporal ordering of affective states and coping within a naturalistic achievement-related demanding situation. Int. J. Stress Manag. 2017, 24, 29–51. [CrossRef] 42. Davies, K.A.; Lane, A.M.; Devonport, T.; Scott, J.A. Validity and Reliability of a Brief Emotional Intelligence Scale (BEIS-10). J. Individ. Differ. 2010, 31, 198–208. [CrossRef] 43. Lane, A.M.; Thelwell, R.C.; Lowther, J.; Devonport, T. Emotional intelligence and psychological skills use among athletes. Soc. Behav. Pers. Int. J. 2009, 37, 195–201. [CrossRef] 44. Clark, L.A.; Watson, D. Constructing validity: Basic issues in objective scale development. In Methodological Issues and Strategies in Clinical Research; Kazdin, A.E., Ed.; American Psychological Association: Washington, DC, USA, 2016; pp. 187–203. 45. Hair, J.F.; Black, W.C.; Babin, B.J.; Anderson, R.E. Multivariate Data Analysis, 7th ed.; Prentice-Hall: Englewood Cliffs, NJ, USA, 2010. 46. Ellis, R.; Saringer, C.; Davis, A.; Biber, D.; Ferrer, D.A. Examining the Impact of Wellness Champions on the Effectiveness of a Workplace Health and Well-Being Program. Am. J. Health Promot. 2020, 35, 121–126. [CrossRef] [PubMed] 47. Iacobucci, D.; Posavac, S.S.; Kardes, F.R.; Schneider, M.J.; Popovich, D. The median split: Robust, refined, and revived. J. Consum. Psychol. 2015, 25, 690–704. [CrossRef] 48. Kovacsik, R.; Griffiths, M.D.; Pontes, H.M.; Soós, I.; de la Vega, R.; Ruíz-Barquín, R.; Demetrovics, Z.; Szabo, A. The Role of Passion in Exercise Addiction, Exercise Volume, and Exercise Intensity in Long-term Exercisers. Int. J. Ment. Health Addict. 2019, 17, 1389–1400. [CrossRef] 49. Irwin, J.R.; McClelland, G.H. Negative Consequences of Dichotomizing Continuous Predictor Variables. J. Mark. Res. 2003, 40, 366–371. [CrossRef] 50. Vernillo, G.; Savoldelli, A.; Zignoli, A.; Trabucchi, P.; Pellegrini, B.; Millet, G.P.; Schena, F. Influence of the world’s most challenging mountain ultra-marathon on energy cost and running mechanics. Eur. J. Appl. Physiol. 2014, 114, 929–939. [CrossRef] [PubMed] 51. Parry, D.; Chinnasamy, C.; Papadopoulou, E.; Noakes, T.; Micklewright, D. Cognition and performance: Anxiety, mood and perceived exertion among Ironman triathletes. Br. J. Sports Med. 2010, 45, 1088–1094. [CrossRef] [PubMed] 52. Brink, M.S.; Nederhof, E.; Visscher, C.; Schmikli, S.L.; Lemmink, K.A. Monitoring load, recovery, and performance in young elite soccer players. J. Strength Cond. Res. 2010, 24, 597–603. [CrossRef] [PubMed] 53. Mikolajczak, M.; Luminet, O. Trait emotional intelligence and the cognitive appraisal of stressful events: An exploratory study. Pers. Individ. Differ. 2008, 44, 1445–1453. [CrossRef] 54. Garnefski, N.; Kraaij, V.; Spinhoven, P. Negative life events, cognitive emotion regulation and emotional problems. Personal. Individ. Differ. 2001, 30, 1311–1327. [CrossRef] 55. Brasseur, S.; Grégoire, J.; Bourdu, R.; Mikolajczak, M. The Profile of Emotional Competence (PEC): Development and Validation of a Self-Reported Measure that Fits Dimensions of Emotional Competence Theory. PLoS ONE 2013, 8, e62635. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2022, 19, 9290 13 of 13 56. Petrides, K.V.; Furnham, A. Trait emotional intelligence: Psychometric investigation with reference to established trait taxonomies. Eur. J. Personal. 2001, 15, 425–448. [CrossRef] 57. Nelis, D.; Quoidbach, J.; Mikolajczak, M.; Hansenne, M. Increasing emotional intelligence: (How) is it possible? Pers. Individ. Differ. 2009, 47, 36–41. [CrossRef] 58. Kotsou, I.; Mikolajczak, M.; Heeren, A.; Grégoire, J.; Leys, C. Improving Emotional Intelligence: A Systematic Review of Existing Work and Future Challenges. Emot. Rev. 2018, 11, 151–165. [CrossRef]

Emotional Intelligence in Ultra-Marathon Runners: Implications for Recovery Strategy and Stress Responses during an Ultra-Endurance Race.

07-29-2022

Nicolas, Michel,Gaudino, Marvin,Bagneux, Virginie,Millet, Gregoire,Laborde, Sylvain,Martinent, Guillaume

eng

PMC8691618

RESEARCH ARTICLE An open-source, lockable mouse wheel for the accessible implementation of time- and distance-limited elective exercise Joseph J. Bivona, IIIID1,2*, Matthew E. Poynter1* 1 Department of Medicine and Vermont Lung Center, University of Vermont Larner College of Medicine, Burlington, Vermont, United States of America, 2 Cellular, Molecular, and Biomedical Sciences Doctoral Program, University of Vermont, Burlington, Vermont, United States of America * JJ.Bivona@uvm.edu (JJB); matthew.poynter@med.uvm.edu (MEP) Abstract Current methods of small animal exercise involve either voluntary (wheel running) or forced (treadmill running) protocols. Although commonly used, each have several drawbacks which cause hesitancy to adopt these methods. While mice will instinctively run on a wheel, the distance and time spent running can vary widely. Forced exercise, while controllable, puts animals in stressful environments in which they are confined and often shocked for “encouragement.” Additionally, both methods require expensive equipment and software, which limit these experiments to well-funded laboratories. To counter these issues, we developed a non-invasive mouse running device aimed to reduce handler-induced stress, provide time- and distance-based stopping conditions, and enable investigators with limited resources to easily produce and use the device. The Lockable Open-Source Training- Wheel (LOST-Wheel) was designed to be 3D printed on any standard entry-level printer and assembled using a few common tools for around 20 USD. It features an on-board screen and is capable of tracking distances, running time, and velocities of mice. The LOST-Wheel overcomes the largest drawback to voluntary exercise, which is the inability to control when and how long mice run, using a servo driven mechanism that locks and unlocks the running surface according to the protocol of the investigator. While the LOST-Wheel can be used without a computer connection, we designed an accompanying application to provide scien- tists with additional analyses. The LOST-Wheel Logger, an R-based application, displays milestones and plots on a user-friendly dashboard. Using the LOST-Wheel, we imple- mented a timed running experiment that showed distance-dependent decreases in serum myostatin as well as IL-6 gene upregulation in muscle. To make this device accessible, we are releasing the designs, application, and manual in an open-source format. The imple- mentation of the LOST-Wheel and future iterations will improve upon existing murine exer- cise equipment and research. PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 1 / 11 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Bivona JJ, III, Poynter ME (2021) An open-source, lockable mouse wheel for the accessible implementation of time- and distance- limited elective exercise. PLoS ONE 16(12): e0261618. https://doi.org/10.1371/journal. pone.0261618 Editor: Richard Jay Smeyne, Thomas Jefferson University, UNITED STATES Received: October 23, 2021 Accepted: December 6, 2021 Published: December 21, 2021 Peer Review History: PLOS recognizes the benefits of transparency in the peer review process; therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. The editorial history of this article is available here: https://doi.org/10.1371/journal.pone.0261618 Copyright: © 2021 Bivona, Poynter. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All model, code, and manual files are available for download at: https:// github.com/jjbivona/LOSTwheel. Introduction While exercise is a safe and effective health strategy [1], the ability for laboratories to test hypotheses concerning the physiological adaptations brought about by movement in mouse models is hindered by the expense of commercial products [2]. To understand the systemic effects of untrained exercise and lower the barrier of entry for murine-exercise research, we have developed an open-source mouse running wheel that accurately tracks and displays the distance traveled and restricts wheel running at specified distances or times. The Lockable Open-Source Training Wheel (LOST-Wheel) has been designed for both standalone and com- puter connected scenarios. The only requirement for standalone mode is a USB power source. In this mode, the LOST-Wheel can log cumulative distance, which is visualized through the onboard screen. When a computer is connected to the wheel through the LOST-Wheel Logger application, more detailed data such as speed and time running can be collected, graphed, and exported. The LOST-Wheel was tested in both overnight (acute) and week-long (chronic) experiments. Samples from the acute, untrained, bout of exercise were subjected to analysis with real-time quantitative polymerase chain reaction and protein quantification through Luminex assays. This design of the LOST-Wheel was inspired as an attempt to create an inexpensive, freely accessible, and human-relevant model of exercise. One drawback of voluntary exercise is the inability to limit running distances [3]. For studies that interrogate the dose dependent effect of exercise or to restrict running to certain times and distances, we implemented a microcon- troller regulated locking mechanism to prevent wheel movement at the will of the investigator. Previous work using commercially available products has shown that wheel running produces dose-dependent effects on neuron proliferation and dendritogenesis [4]. Additionally, that the presence of an immobile wheel in a cage also elicits neurological effects in the absence of its use demands that such a device-exposure group should be included as a proper control in wheel-running experiments [4, 5]. Despite the average gait of a mouse being 5–6 cm [6], mice voluntarily run upwards of 7 hours and 20 km/night [3] when provided with a standard wheel [7, 8]. The use of a wheel for long periods cannot be explained by a single theory [9], and phys- iological effects differ substantially between strains of mice [10]. By limiting running distances, the locking capacity of the LOST-Wheel enables researchers to normalize voluntary exercise across animals. While forced exercise (treadmill running) is an alternative strategy that ensures a consistent distance across animals, the handling [11], confinement [12], and electrical shock [13] required for its implementation can induce stress and alter the biological responses being studied [14]. To verify the effectiveness of the LOST-Wheel and to evaluate the consequences of a single, untrained bout of exercise, we allowed a cohort of mice to perform voluntary exercise for a sin- gle night (their waking time) then performed RT-qPCR on gastrocnemius muscle and multi- plex analysis of several myokines in serum. We confirmed exercise-induced physiological changes, including increases in gastrocnemius Il6 gene expression and a significant, negative, relationship between serum myostatin and the distance traveled by mice. Methods LOST-Wheel design and code The LOST-Wheel was designed using Fusion360 (Autodesk, San Rafael, CA) and is composed of four pieces: main body, top face, servo pin, and wheel. Table 1 contains a component list for the electronics, bearings, axle, and hardware. Slicing the models for 3D printing was conducted using Cura (Ultimaker, Utrecht, Netherlands). All pieces were printed using fusion deposition PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 2 / 11 Funding: This study was made possible by support from the Larner College of Medicine and the Vermont Lung Center T32HL076122. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. modeling with polylactic acid (PLA) on an Ender3 V2 3D printer (Creality 3D, Shenzhen, China). The sketches used to program the LOST-Wheel were created in Arduino IDE (Ardu- ino, New York City, NY) in Arduino/C++ language with the additional libraries, U8g2 and U8x8. A computer rendering, representative image, and wiring diagram for the LOST-Wheel are shown in Fig 1A–1C, respectively. In acute exercise experiments, the Timer Mode protocol was uploaded and set to begin when the wheel was powered on. For chronic exercise, the Dis- tance Mode protocol was uploaded, and the threshold set to 106 m for unlimited running. The LOST-Wheel can be powered indefinitely; however, in the experiments of this manuscript, data was collected daily, at which time wheels were reset. All files required to build and program the LOST-Wheel are available at https://github. com/jjbivona/LOSTwheel. The LOST-Wheel design files, manual, and LOST-Wheel Logger software are licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. To access a build video and setup tutorial, visit: https://www.youtube. com/channel/UCUp9zD0H99VcX0XXl2qGUmg. LOST-Wheel logger While the LOST-Wheel can accurately display distance using its onboard screen, more detailed information can be obtained using the LOST-Wheel Logger Application. The application was created in R and RStudio (version 4.1.1 and 1.4 respectively) using Shiny [15], ggplot2 [16], and Serial [17] packages. Users are instructed to enter an ID for the wheel, the associated com- munication port (COM port), and the duration of data collection. After information is entered and the start button is pressed, the logger restarts the wheel and collects data in one second intervals until the duration is met. The program then calculates the individual slopes between each data point to determine the maximum speed in meters/second. Using this information, the Logger can determine the amount of time the mouse has run. Previous literature indicates that untrained mice have an average speed of 1–2 km/h (0.28–0.56 m/s) [18]; therefore, a threshold of 0.2 m/s is applied to exclude non-running events. Finally, the LOST-Wheel Log- ger creates distance/time and velocity/time graphs and presents all information for the user. The number of wheels that can be simultaneously connected is limited by the number of Table 1. Component list for the LOST-Wheel. Component Quantity 6 mm ID, 10 mm OD, x 3 mm bearing 3 6 mm axle cut to 65 mm 1 M3x5 self-tapping screw 1 M2x6 self-tapping screw 1 M2.3x8 self-tapping screw 8 M1.7x6 self-tapping screw 8 10 mm x 5 mm x 3 mm neodymium magnet 2 22-gauge, 2.54 mm breadboard jumper wires, 3 male, 7 female 10 Arduino Nano (or similar) 1 9g micro servo 1 KY-003 hall effect sensor 1 0.96 inch 128x64 OLED Screen I2C connection SSD1306 Driver 1 This list serves as a template for the electronics and hardware required for the device. Generic Arduino clones can be substituted as microcontrollers since they are often a fraction of the price. Magnet size and quantity can also be changed depending on availability and accuracy required. https://doi.org/10.1371/journal.pone.0261618.t001 PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 3 / 11 universal serial bus (USB) ports available on the computer. In our laboratory, we use an inex- pensive USB hub to expand the number of wheels connected. The current version of the LOST-Wheel Logger accommodates a single wheel. Therefore, to collect data from multiple wheels, the user must open a separate instance of RStudio for each wheel. Mice 12–20 week old male C57BL/6J mice purchased from The Jackson Laboratory (Bar Harbor, ME) were housed in AAALAC-accredited animal facilities at the University of Vermont, and all experimental animal procedures were approved by the University of Vermont Institutional Animal Care and Use Committee, protocol #202100027. Mice were maintained on a 12 hour- light/dark cycle, beginning at 07:00 and 19:00, respectively, and provided chow and water ad libitum. To examine the effect of a single, untrained bout of exercise, mice were brought from the vivarium and caged individually. A single LOST-Wheel, or an immobile “dummy” wheel, was introduced into each cage at 08:00. Wheels in the running group remained unlocked to accli- mate the mice until 12:00, at which point the wheels locked. At 19:00 the wheels unlocked, and mice were allowed to run voluntarily for 12 hours, at which point they were euthanized by an intraperitoneal injection of pentobarbital (Euthasol, Midwest Veterinary Supply, Lakeville, MN), followed by exsanguination. Serum and gastrocnemius muscle were collected and snap frozen in liquid nitrogen. Fig 1. Lost-Wheel assembly and testing. A) Computer rendered design of the assembled LOST-Wheel. B) Completed assembly of the LOST-Wheel. C) Simplified wiring diagram of components. 22-gauge, 2.54 mm breadboard jumper wires are soldered to the microcontroller and connected to components using the attached plugs. Should a single component fail, this allows for easy replacement without resoldering. D) The LOST-Wheel Logger application can be used in conjunction with the LOST-Wheel to collect and plot additional data. E) Mice were allowed to run unrestricted for 7 days on the LOST-Wheel. Each morning, the distance was recorded, and wheels were reset (n = 5). https://doi.org/10.1371/journal.pone.0261618.g001 PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 4 / 11 For one week running experiments, wheels remained unlocked throughout the duration of the test. Distances were recorded at 09:00 each morning and the wheels were reset. RNA isolation, quantitative real-time polymerase chain reaction (qRT-PCR), and protein quantitation Total RNA was isolated from liquid nitrogen pulverized whole muscle using TRIzol reagent followed by a chloroform-isopropanol extraction (Thermo Fisher Scientific, Waltham, MA, USA). RNA concentration and purity were measured using a NanoDrop 2000 spectrophotom- eter. (Thermo Fisher Scientific). cDNA was synthesized from 100 ng of RNA using the qScript Supermix reagent kit per manufacturer’s instructions (Quantabio, Beverly, MA, USA). Quanti- tative real-time PCR was performed using iTaq Universal SYBR Green Supermix on a CFX96 Touch (Bio-Rad, Hercules, CA, USA), with the relative mRNA expression calculated using the threshold cycle (Ct; 2−ΔΔCt) method normalized to Gapdh expression. The following primer sequences were used (Integrated DNA Technologies, Coralville, IA, USA): Il6 forward 5’-CCCGGAGAGGAGACTTCACAG-3’, reverse 5’-GAGCATTGGAAATTGGGGTA-3’; Gapdh forward 5’-ACGACCCCTTCATTGACCTC-3’, reverse 5’-TTCACACCCATCA CAAACAT-3’. Serum samples were analyzed using a Milliplex Mouse Myokine Magnetic Bead Panel (Millipore Sigma, St. Louis, MO, USA) on a Luminex 100 xMAP Instrument (Bio-Rad) according to kit instructions. Statistical analysis and figures RT-qPCR and Milliplex data were analyzed and visualized using GraphPad Prism version 9.2.0 for Windows (GraphPad Software, San Diego, CA, USA) with unpaired t-tests and a lin- ear regression, respectively. Significance is designated by p-values < 0.05. Fig 2A was created with BioRender.com. Results Building and testing the LOST-Wheel All components in Table 1 are readily available and the 3D printed models can be created using any entry-level 3D printer capable of printing in polylactic acid (PLA) filament. The wheel can be assembled using only a #1 Phillips head screwdriver, a soldering iron, and a wire cutter/stripper. Excluding a 3D printer, the entire apparatus can be created for less than 15 USD. The models can easily be modified to account for larger diameter bearings and drive shafts, additional wheel magnets, or changes in component mounting holes. A representative rendering and completed wheel are shown in Fig 1A and 1B, respectively. A simplified compo- nent wiring diagram is shown in Fig 1C. We have also created a series of videos that outline the assembly, programming, and cage setup, which can be accessed at https://www.youtube. com/channel/UCUp9zD0H99VcX0XXl2qGUmg. The LOST-Wheel was tested for 7 days to evaluate durability and animal safety, during which mice steadily increased distance traveled per day (Fig 1D), averaging 2046.38 ± 2647.11 m on the first day and 8972.71 ± 2888.14 m at the end of the trial. One mouse did not use the wheel until the second day. Locking criteria To limit and control mouse exercise, we developed three separate modes for the LOST-Wheel. These are first edited by the user based on their experimental requirements and uploaded to PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 5 / 11 the wheel through the Arduino IDE program. In Timer Mode, the user inputs the amount of time (in hours) for which they want the wheel to be sequentially locked and unlocked. This program was implemented for our acute voluntary exercise experiments. The second and third modes limit exercise by distance and time spent running, respectively, and allow for the normalization of exercise across animals within experimental groups. Acute voluntary exercise by untrained mice To examine the effects of a single bout of voluntary exercise on untrained mice, we subjected mice to a one-day acclimation and exercise protocol (Fig 2A). Mice were singly placed in cages containing either a LOST-Wheel or an immobile “dummy” wheel at 08:00. The wheel remained unlocked for four hours for acclimation, at which point the servo inserted a pin into the wheel to lock the device (12:00). At 19:00, the pin was withdrawn, and mice were allowed to voluntarily use the wheel for 12 hours. At the end of exercise, we collected serum and leg muscles, then measured expression and production of interleukin-6 (IL-6), a muscle-produced cytokine (myokine) reported to be induced in exercised human subjects [19] as well as in Fig 2. Acute exercise of untrained mice. A) Experimental setup. Mice were acclimated to the LOST-Wheel for four hours, at which point the wheel locked until the evening. Mice were allowed to run unrestricted for 12 hours, at which point the wheels relocked and mice were euthanized. B) RT-qPCR analysis of gastrocnemius muscle Il6 expression (n = 6-7/group, unpaired t test). C) Serum myostatin concentrations relative to the distance traveled of mice undergoing untrained acute exercise (n = 6, simple linear regression). https://doi.org/10.1371/journal.pone.0261618.g002 PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 6 / 11 mouse models of acute exercise [20] and myostatin. We observed a significant increase in the expression of Il6 in the gastrocnemius muscle following acute exercise (Fig 2B). Interestingly, there was a significant correlation between distance traveled on the LOST-Wheel and serum myostatin measured by Luminex (Fig 2C). One mouse did not log any distance on the LOST-- Wheel. This mouse was excluded from RT-qPCR measurements but included in measure- ments of serum myostatin as a sedentary control. Discussion This manuscript presents an open-source mouse wheel that can be affordably and efficiently assembled by investigators. The LOST-Wheel can be used in standalone mode to collect data on overall distances traveled. When used in combination with a computer, the wheel interfaces with the LOST-Wheel Logger application to calculate the amount of time spent running and top speed achieved. Finally, we verified the integrity of the device and observed physiological changes in serum and muscle gene expression from a single untrained bout of exercise. Physiological alterations after running While the primary objective was to assess the efficacy of the LOST-Wheel, there was merit in observing the effects of acute exercise. Muscle derived signaling molecules, termed myokines, are released during muscle contraction and indicate physiological adaptations following exer- cise. Most notably, IL-6 is highly upregulated and is believed to function differently from clas- sic inflammatory signaling by instead increasing glucose sensitivity and uptake [21]. We observed increases in Il6 expression in the gastrocnemius muscle (Fig 2B); however, increased IL-6 protein concentrations were not detected in serum using a myokine multiplex panel, implicating its local effect in the muscle. Additionally, we observed a significant, negative, cor- relation between distance traveled and serum myostatin concentrations (Fig 2C). As a regula- tor of muscle growth and differentiation [22], this correlation implies that myostatin is dose dependently regulated by running distance. These results align with previous studies in humans and rats, in which myostatin was transiently decreased after bouts of acute exercise [23, 24]. Voluntary exercise versus forced exercise While treadmill based forced exercise allows for controlled speed, duration, and incline of training, it increases corticosterone and norepinephrine levels, indicating a strong stress response that is not elicited by during voluntary exercise [25–28]. The shock, confinement, or handling of the mice can all contribute to the increased stress reported. Additionally, the pres- ence of an immobile wheel can have the added benefit of environmental enrichment [29, 30]. Due to this effect, we advise using a locked LOST-Wheel or creating a dummy wheel (fully assembled without electronic components) for control groups [4, 5]. Durability of the LOST-Wheel Several iterations of the LOST-Wheel were prototyped before using the Hall effect sensor and magnet combination. Originally, the wheel rotated on a rotary encoder, an electro-mechanical part that has a finite number of rotations (30,000–100,000) before wearing out. We also attempted using an infrared sensor, but cage bedding would often block the beam, rendering it useless. The magnet and Hall effect sensor bypass these problems and should remain opera- tional indefinitely. The running surface and shaft can easily be removed and sprayed with etha- nol to disinfect between uses. While mice have occasionally chewed the running surface, this PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 7 / 11 has not hindered its balance or performance. We have also designed and provided a guard for the power cord that protects it from destructive animals. Comparison to existing wheels To verify data collection on the LOST-Wheel, we have compared our 7-day protocol to the well detailed and extensive work of de Bono and colleagues [18], who developed a method to track distances run by mice but that lacks the ability to limit exercise, requires an expensive data acquisition unit with software (Spike 2, https://ced.co.uk/prices/1401options, https://ced. co.uk/prices/softwareprices), and has limited customizability and programmability. We have developed the LOST-Wheel Logger App, which allows for detailed collection, analysis, and export of data similar to the Spike 2 program. Reassuringly, our device recapitulates the dis- tances reported in the aforementioned work. Recently, an open-source wheel was introduced that tracks running distances using a similar Hall Effect sensor as the LOST-Wheel. However, this device does not allow for controlled exercise or additional data collection as it only collects cumulative distance measurements [31]. Forced exercise designs exist that can be used to exer- cise mice, but such devices require extensive machining and calibration [32] or the repurpos- ing of an existing human-treadmill [33]. Cost The LOST-Wheel provides an inexpensive alternative to commercial murine exercise devices. Similar distance tracking devices, without locking abilities, cost 300–400 USD per wheel, require an additional data acquisition unit (400–900 USD), and necessitate accompanying software (800–2500 USD) (price quotes are from correspondence with commercial retailers). The availability and shallow learning curve of Arduino-based microcontrollers allows for labo- ratories to create their own devices at a fraction of the price [34]. The 3D printed parts can be outsourced to university fabrication labs, commercial 3D print operations, or fabricated in- house as entry level fused deposition modeling printers have substantially decreased in price over the last decade, with entry level printers ranging from 200–300 USD. Future designs The LOST-Wheel was created out of necessity and to improve current research-based exercise protocols in experimental animals. We have shown that this inexpensive, open-source wheel can provide investigator-controllable exercise to small rodent research without modification of the cage. The value of an open-source project is that it allows researchers to easily imple- ment changes that fit their research goals and budgets. Future iterations of the LOST-Wheel and Logger App can include wireless transmission of data to a smartphone through commer- cially available Bluetooth-Arduino adaptors. Other changes may include on-board data stor- age, operant conditioning modifications, or using the locking pin to provide resistance against the wheel to model weighted wheel hypertrophy-inducing exercise [35]. Supporting information S1 Data. The code, manual, and 3D files in both .STL and .F3D format, can be found at: www.github.com/jjbivona/lostwheel. (TXT) S1 Video. Videos for building and setting up the LOST-Wheel can be found at: https:// www.youtube.com/channel/UCUp9zD0H99VcX0XXl2qGUmg. (TXT) PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 8 / 11 Acknowledgments The authors thank Jeffery L. Brabec and Colleen E. Yancey for their help debugging code. Author Contributions Conceptualization: Joseph J. Bivona, III. Data curation: Joseph J. Bivona, III. Formal analysis: Joseph J. Bivona, III. Funding acquisition: Joseph J. Bivona, III. Investigation: Joseph J. Bivona, III. Methodology: Joseph J. Bivona, III. Project administration: Joseph J. Bivona, III, Matthew E. Poynter. Software: Joseph J. Bivona, III. Supervision: Matthew E. Poynter. Visualization: Joseph J. Bivona, III. Writing – original draft: Joseph J. Bivona, III. Writing – review & editing: Joseph J. Bivona, III, Matthew E. Poynter. References 1. Pedersen BK, Saltin B. Exercise as medicine—evidence for prescribing exercise as therapy in 26 differ- ent chronic diseases. Scandinavian Journal of Medicine & Science in Sports. 2015; 25(S3):1–72. https://doi.org/10.1111/sms.12581 PMID: 26606383 2. Pearce JM. Chapter 2—The Benefits of Sharing—Nice Guys and Girls do Finish First. In: Pearce JM, editor. Open-Source Lab. Boston: Elsevier; 2014. p. 13–35. 3. Manzanares G, Brito-da-Silva G, Gandra PG. Voluntary wheel running: patterns and physiological effects in mice. Brazilian journal of medical and biological research = Revista brasileira de pesquisas medicas e biologicas. 2018; 52(1):e7830–e. https://doi.org/10.1590/1414-431X20187830 PMID: 30539969. 4. Dostes S, Dubreucq S, Ladevèze E, Marsicano G, Abrous DN, Chaouloff F, et al. Running per se stimu- lates the dendritic arbor of newborn dentate granule cells in mouse hippocampus in a duration-depen- dent manner. Hippocampus. 2016; 26(3):282–8. https://doi.org/10.1002/hipo.22551 PMID: 26606164 5. Dubreucq S, Marsicano G, Chaouloff F. Emotional consequences of wheel running in mice: Which is the appropriate control? Hippocampus. 2011; 21(3):239–42. https://doi.org/10.1002/hipo.20778 PMID: 20232385 6. Wertman V, Gromova A, La Spada AR, Cortes CJ. Low-Cost Gait Analysis for Behavioral Phenotyping of Mouse Models of Neuromuscular Disease. Journal of visualized experiments: JoVE. 2019;(149). Epub 2019/08/06. https://doi.org/10.3791/59878 PMID: 31380846. 7. Stranahan AM, Lee K, Martin B, Maudsley S, Golden E, Cutler RG, et al. Voluntary exercise and caloric restriction enhance hippocampal dendritic spine density and BDNF levels in diabetic mice. Hippocam- pus. 2009; 19(10):951–61. https://doi.org/10.1002/hipo.20577 PMID: 19280661. 8. Place N, Ivarsson N, Venckunas T, Neyroud D, Brazaitis M, Cheng AJ, et al. Ryanodine receptor frag- mentation and sarcoplasmic reticulum Ca2+ leak after one session of high-intensity interval exercise. Proceedings of the National Academy of Sciences of the United States of America. 2015; 112 (50):15492–7. Epub 2015/11/02. https://doi.org/10.1073/pnas.1507176112 PMID: 26575622. 9. Sherwin CM. Voluntary wheel running: A review and novel interpretation. Animal Behaviour. 1998; 56 (1):11–27. https://doi.org/10.1006/anbe.1998.0836 PMID: 9710457 10. Merritt JR, Rhodes JS. Mouse genetic differences in voluntary wheel running, adult hippocampal neuro- genesis and learning on the multi-strain-adapted plus water maze. Behav Brain Res. 2015; 280:62–71. Epub 2014/11/27. https://doi.org/10.1016/j.bbr.2014.11.030 PMID: 25435316. PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 9 / 11 11. Hurst JL, West RS. Taming anxiety in laboratory mice. Nature Methods. 2010; 7(10):825–6. https://doi. org/10.1038/nmeth.1500 PMID: 20835246 12. Morgan KN, Tromborg CT. Sources of stress in captivity. Applied Animal Behaviour Science. 2007; 102 (3):262–302. https://doi.org/10.1016/j.applanim.2006.05.032 13. Bali A, Jaggi AS. Electric foot shock stress: a useful tool in neuropsychiatric studies. Reviews in the Neurosciences. 2015; 26(6):655–77. https://doi.org/10.1515/revneuro-2015-0015 PMID: 26167978 14. Balcombe JP, Barnard ND, Sandusky C. Laboratory Routines Cause Animal Stress. Journal of the American Association for Laboratory Animal Science. 2004; 43(6):42–51. PMID: 15669134 15. Winston Chang JC, JJ Allaire, Carson Sievert, Barret Schloerke, Yihui Xie, Jeff Allen, et al. shiny: Web Application Framework for R. R package version 1.6.0. ed2021. 16. Wickham H. ggplot2: Elegant Graphics for Data Analysis. New York: Springer-Verlag; 2016. 17. Seilmayer M. serial: The Serial Interface Package. 2020. 18. DeBono JP, Adlam D, Paterson DJ, Channon KM. Novel quantitative phenotypes of exercise training in mouse models. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology. 2006; 290(4):R926–R34. https://doi.org/10.1152/ajpregu.00694.2005 PMID: 16339385. 19. Gusba JE, Wilson RJ, Robinson DL, Graham TE. Interleukin-6 and its mRNA responses in exercise and recovery: relationship to muscle glycogen. Scandinavian Journal of Medicine & Science in Sports. 2008; 18(1):77–85. https://doi.org/10.1111/j.1600-0838.2006.00635.x PMID: 17346285 20. Tominaga T, Ma S, Saitou K, Suzuki K. Glucose Ingestion Inhibits Endurance Exercise-Induced IL-6 Producing Macrophage Infiltration in Mice Muscle. Nutrients. 2019; 11(7):1496. https://doi.org/10.3390/ nu11071496 PMID: 31262006 21. Pal M, Febbraio MA, Whitham M. From cytokine to myokine: the emerging role of interleukin-6 in meta- bolic regulation. Immunology & Cell Biology. 2014; 92(4):331–9. https://doi.org/10.1038/icb.2014.16 PMID: 24751614 22. Allen DL, Hittel DS, McPherron AC. Expression and function of myostatin in obesity, diabetes, and exer- cise adaptation. Medicine and science in sports and exercise. 2011; 43(10):1828–35. https://doi.org/10. 1249/MSS.0b013e3182178bb4 PMID: 21364474. 23. Harber MP, Crane JD, Dickinson JM, Jemiolo B, Raue U, Trappe TA, et al. Protein synthesis and the expression of growth-related genes are altered by running in human vastus lateralis and soleus mus- cles. American Journal of Physiology-Regulatory, Integrative and Comparative Physiology. 2009; 296 (3):R708–R14. https://doi.org/10.1152/ajpregu.90906.2008 PMID: 19118097. 24. Matsakas A, Friedel A, Hertrampf T, Diel P. Short-term endurance training results in a muscle-specific decrease of myostatin mRNA content in the rat. Acta Physiol Scand. 2005; 183(3):299–307. Epub 2005/03/04. https://doi.org/10.1111/j.1365-201X.2005.01406.x PMID: 15743390. 25. Sasaki H, Hattori Y, Ikeda Y, Kamagata M, Iwami S, Yasuda S, et al. Forced rather than voluntary exer- cise entrains peripheral clocks via a corticosterone/noradrenaline increase in PER2::LUC mice. Scien- tific Reports. 2016; 6(1):27607. https://doi.org/10.1038/srep27607 PMID: 27271267 26. Hayes K, Sprague S, Guo M, Davis W, Friedman A, Kumar A, et al. Forced, not voluntary, exercise effectively induces neuroprotection in stroke. Acta Neuropathologica. 2008; 115(3):289–96. https://doi. org/10.1007/s00401-008-0340-z PMID: 18210137 27. Svensson M, Rosvall P, Boza-Serrano A, Andersson E, Lexell J, Deierborg T. Forced treadmill exercise can induce stress and increase neuronal damage in a mouse model of global cerebral ischemia. Neuro- biology of Stress. 2016; 5:8–18. https://doi.org/10.1016/j.ynstr.2016.09.002 PMID: 27981192 28. Allen JM, Miller MEB, Pence BD, Whitlock K, Nehra V, Gaskins HR, et al. Voluntary and forced exercise differentially alters the gut microbiome in C57BL/6J mice. Journal of Applied Physiology. 2015; 118 (8):1059–66. https://doi.org/10.1152/japplphysiol.01077.2014 PMID: 25678701. 29. Pham TM, Brene´ S, Baumans V. Behavioral assessment of intermittent wheel running and individual housing in mice in the laboratory. J Appl Anim Welf Sci. 2005; 8(3):157–73. Epub 2006/02/14. https:// doi.org/10.1207/s15327604jaws0803_1 PMID: 16468945. 30. Fox C, Merali Z, Harrison C. Therapeutic and protective effect of environmental enrichment against psy- chogenic and neurogenic stress. Behav Brain Res. 2006; 175(1):1–8. https://doi.org/10.1016/j.bbr. 2006.08.016 PMID: 16970997 31. Edwards J, Olson B, Marks DL. Constructing and programming a cost-effective murine running wheel with digital revolution counter. Lab Animal. 2021; 50(8):202–4. https://doi.org/10.1038/s41684-021- 00795-y PMID: 34145438 32. Williams M, Sater S, Burkhalter C, Schoonen S, Miller J, Shrestha D, et al. Low-cost, open-source, vari- able speed and incline treadmill for studying impacts of neonatal locomotion. HardwareX. 2020; 7: e00097. https://doi.org/10.1016/j.ohx.2020.e00097 PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 10 / 11 33. Bouganim S, Bergdahl A. Constructing an inexpensive and versatile homemade rodent treadmill. Lab Anim (NY). 2017; 46(3):67–9. Epub 2017/02/18. https://doi.org/10.1038/laban.1196 PMID: 28211864. 34. D’Ausilio A. Arduino: a low-cost multipurpose lab equipment. Behav Res Methods. 2012; 44(2):305–13. Epub 2011/11/01. https://doi.org/10.3758/s13428-011-0163-z PMID: 22037977. 35. Call JA, McKeehen JN, Novotny SA, Lowe DA. Progressive resistance voluntary wheel running in the mdx mouse. Muscle & nerve. 2010; 42(6):871–80. https://doi.org/10.1002/mus.21764 PMID: 21104862. PLOS ONE A locking mouse wheel for investigator-limited voluntary exercise PLOS ONE | https://doi.org/10.1371/journal.pone.0261618 December 21, 2021 11 / 11

An open-source, lockable mouse wheel for the accessible implementation of time- and distance-limited elective exercise.

12-21-2021

Bivona, Joseph J,Poynter, Matthew E

eng

PMC10675285

Citation: Del Arco, A.; Martinez Aguirre-Betolaza, A.; Malchrowicz- Mo´sko, E.; Gogojewicz, A.; Castañeda-Babarro, A. Are Supplements Consumed by Middle-Distance Runners Evidence-Based? A Comparative Study between Level of Competition and Sex. Nutrients 2023, 15, 4839. https://doi.org/10.3390/nu15224839 Academic Editors: Valentín E. Fernández-Elías and Olga López Torres Received: 16 October 2023 Revised: 16 November 2023 Accepted: 17 November 2023 Published: 20 November 2023 Copyright: © 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). nutrients Article Are Supplements Consumed by Middle-Distance Runners Evidence-Based? A Comparative Study between Level of Competition and Sex Asier Del Arco 1, Aitor Martinez Aguirre-Betolaza 1 , Ewa Malchrowicz-Mo´sko 2 , Anna Gogojewicz 3 and Arkaitz Castañeda-Babarro 1,* 1 Health, Physical Activity and Sports Science Laboratory, Department of Physical Activity and Sports, Faculty of Education and Sport, University of Deusto, 48007 Bilbao, Spain; asier.delarco@opendeusto.es (A.D.A.); a.martinezdeaguirre@deusto.es (A.M.A.-B.) 2 Institute of Sport Sciences, Poznan University of Physical Education, 61-871 Pozna´n, Poland; malchrowicz@awf.poznan.pl 3 Institute of Health Sciences, Poznan University of Physical Education, 61-871 Pozna´n, Poland; gogojewicz@awf.poznan.pl * Correspondence: arkaitz.castaneda@deusto.es Abstract: Background: Middle-distance running events have special physiological requirements from a training and competition point of view. Therefore, many athletes choose to take sport supplements (SS) for different reasons. To date, few studies have been carried out that review supplementation patterns in middle-distance running. The aim of the present study is to analyze the consumption of SS in these runners with respect to their level of competition, sex and level of scientific evidence. Methods: In this descriptive cross-sectional study, data was collected from 106 middle-distance runners using a validated questionnaire. Results: Of the total sample, 85.85% responded that they consumed SS; no statistical difference was found regarding the level of competition or sex of the athletes. With respect to the level of competition, differences were observed in the total consumption of SS (p = 0.012), as well as in that of medical supplements (p = 0.005). Differences were observed between sexes in the consumption of medical supplements (p = 0.002) and group C supplements (p = 0.029). Conclusions: Higher-level athletes consume SS that have greater scientific evidence. On the other hand, although the most commonly consumed SS have evidence for the performance or health of middle-distance runners, runners should improve both their sources of information and their places of purchase. Keywords: middle-distance; supplementation; nutrition; performance; health 1. Introduction Middle-distance running events are highly complex from a bioenergetic, training and tactical point of view [1]. The level of energy intensity is in a middle ground between aerobic and anaerobic metabolism [2], with the aerobic contribution in the 800 m being between 60 and 75% and slightly higher (77–85%) in the 1500 m [3]. In addition, due to the type of muscle fibers these athletes have (Mainly IIX and IIA [4]), most middle- distance runners can reach lactate peaks of >20 mmol/L, leading to muscle pH levels as low as 6.6 [5]. However, the high speed requirements make both aerobic and anaerobic metabolism contribute significantly during these events [6]. This can be reflected in the distribution of training intensities throughout the season. Middle-distance runners work a very wide spectrum of training zones, ranging from low-intensity running sessions to very-high-intensity glycolytic workouts [7]. In this way, elite middle-distance runners develop aerobic capacities similar to those of long-distance runners, mechanical skills close to those of sprinters, as well as a highly enhanced anaerobic capacity [1]. Some of these Nutrients 2023, 15, 4839. https://doi.org/10.3390/nu15224839 https://www.mdpi.com/journal/nutrients Nutrients 2023, 15, 4839 2 of 11 characteristics make them adopt different race strategies [8,9]. However, sometimes the difference between being a medalist or not is minimal [10], and the improvements seen with some SS are very worthwhile in terms of performance [11]. Supplements are defined as “A food, food component, nutrient, or nonfood compound that is purposefully ingested in addition to the habitually-consumed diet with the aim of achieving a specific health and/or performance benefit” [11]. Although many athletes use SS to improve their performance, there are other underlying reasons for their use [12]. According to the Australian Institute of Sport (AIS), supplements are classified into four groups using the “ABCD” system [13]. This is based on the latest scientific evidence for determining whether a product is safe, permitted and effective in improving performance or health: (A) supplements with solid scientific evidence in specific situations under estab- lished protocols; (B) components with emerging evidence that should be used in research or clinical settings; (C) supplements with limited evidence and effects on performance; (D) prohibited products or those with a high risk of contamination by doping substances. Regarding middle-distance races, some of the supplements that have shown the most evidence in improving performance are caffeine [14,15], β-Alanine [16–18] and sodium bicarbonate [19–21]. However, these SS are not among the most consumed by middle- distance runners, with the consumption of vitamins, minerals and amino acids being higher than the previously mentioned ones [22]. Although SS can provide both health and performance benefits, athletes’ knowledge of them is sometimes limited [23,24]. In the same way, it has been shown that the use of some SS with less scientific evidence is greater than those with higher levels of supporting research [25]. Finally, some of the main motivators for their consumption are unqualified individuals, such as friends, teammates or the runners themselves [26–29]. To our knowledge, few studies have been conducted to analyze supplementation patterns in athletes, and no one exclusively in middle-distance runners. Thus, the objective of this research is to know the supplementation trends in those athletes with respect to their level and gender. On the other hand, it aims to assess whether the SS taken by middle- distance runners are those with the most scientific evidence, thus reducing the existing gap in the literature [30]. 2. Materials and Methods 2.1. Type of Study The research was a descriptive and cross-sectional study. The sample was selected using non-probabilistic, non-injurious and convenience sampling among training groups and individual middle-distance athletes at the national level. 2.2. Participants and Study Sample A total of 106 middle-distance runners (800–1500 m) participated, of which 74 were men and 32 were women (gender assigned at birth). Only two requirements were estab- lished to participate in the study, which were as follows: (1) be over 18 years of age (legal age in Spain); (2) be currently performing middle-distance disciplines. The level of the ath- letes was differentiated by their area of competition, which could be regional (competing at regional or provincial level), national (competitions in Spain) or international (competitions at European and World level). Table 1 describes the age, basic anthropometric data and best performances in middle-distance events of the participants involved in the research. 2.3. Instruments The questionnaire chosen for this research has been previously used in studies with the same objectives carried out in other sports [26,31,32]. This one was chosen for two main reasons; on the one hand, for its contents, structure, applicability and ease of completion for the athletes. The second reason was the quality of the questionnaire, which was created by 25 experts from different areas and achieved a 54% methodological validity, being one of the 57 questionnaires (out of 167) validated to obtain accurate data on supplement Nutrients 2023, 15, 4839 3 of 11 consumption [33]. The questionnaire has 4 main parts and a total of 32 questions. The first one asks for personal (e.g., sex), anthropometric (e.g., height, weight) and sociodemographic (e.g., region of residence) data, with a total of 8 questions. The second, with a total of 5 questions, covers topics about the sport practice (e.g., years of practice, level of competition). The third part, with a similar objective, collects information about your best times in the different middle-distance disciplines or about your training days and number of competitions and has a total of 8 questions. Finally, the fourth part (11 questions) covers the area of supplementation, with questions such as: what supplements do they consume, reason for consumption, and place of purchase. This questionnaire collects data about all types of supplements, among which we find sports foods (e.g., energy bars, sports gels), medical supplements (e.g., iron, vitamin D, multivitamins) or performance supplements (e.g., caffeine, creatine, ß-Alanine). These different types of supplements are defined as sport supplements in the current study. From this last section, different questions related to diet were eliminated from the original questionnaire because they did not contribute to the objective of the study and in order to limit the response time. This questionnaire can be obtained in: Suplementación nutricional en la actividad físico-deportiva: análisis de la calidad del suplemento proteico consumido [34]. Table 1. Characteristics and personal times of the different subjects. Sex (n) Category (n) Age Height * Weight * BMI * PB 800 m PB 1500 m Male (74) Regional (29) 22.9 ± 5.7 176.3 ± 8.2 64.8 ± 9.1 18.3 ± 2.0 2:01.74 ± 6.80 4:17.22 ± 16.31 National (43) 24.5 ± 7.6 177.6 ± 6.4 65.1 ± 6.1 18.3 ± 1.4 1:56.42 ± 5.25 4:02.92 ± 16.50 International (2) 20.0 ± 2.8 189.0 ± 5.7 68.5 ± 4.9 18.1 ± 0.8 1:48.38 ± 1.15 3:47.50 Female (32) Regional (11) 23.6 ± 8.2 165.0 ± 4.8 52.4 ± 7.0 15.8 ± 1.7 2:26.04 ± 7.04 5:13.66 ± 21.95 National (17) 21.8 ± 3.1 164.8 ± 4.3 52.5 ± 4.1 15.9 ± 1.1 2:15.56 ± 6.55 4:59.61 ± 43.06 International (4) 21.0 ± 2.9 167.5 ± 4.8 55.0 ± 3.2 16.4 ± 0.8 2:05.27 ± 3.84 4:15.23 ± 9.97 Results are expressed as mean ± SD. BMI: body mass index. * Self-reported height and weight. BMI calculated from self-reported height and weight. PB: personal best. Gender assigned at birth. 2.4. Procedures For the data collection, the questionnaire was distributed via training groups, known athletes and social networks. The questionnaire was distributed online so that runners could complete it remotely, voluntarily and anonymously. The protocol complied with the provisions of the Declaration of Helsinki for human research and was approved by the ethical committee of the University of Deusto (ETK-14/23-24) dated 26 October 2023. 2.5. Statistical Analysis To verify whether the variables had a normal distribution, a Kolmogorov–Smirnov test was applied, and Levene’s test was used to verify homoscedasticity. The quantitative data were presented as mean + SD, while the qualitative variables were expressed as percentages and frequencies. A two-way ANOVA was performed for the sex factor (male–female) and level of competition (regional, national and international) to analyze the differences in the total consumption of SS, as well as the SS consumed from the different categories. To assess sex differences, a t-test for independent variables was performed, while to assess differences among competition levels, a one-way ANOVA was performed. For those variables in which significant differences were found, the Bonferroni post hoc analysis was used. Regarding the analysis of the athletes who consumed SS, the reason for consumption, the place where they obtained them and who advised them to consume them, a chi-square (χ2) test was used to verify the existence or not of differences between athletes of different sex and level of competition. As for the SS that were consumed by at least 10% of the sample, a χ2 test was performed to verify possible differences according to sex or level of competition. The level of statistical significance was established as p < 0.05. The statistical analysis was carried out using the Statistical Package for Social Sciences (SPSS) software v.28.0.0 (IBM, Armonk, NY, USA) for Windows. Nutrients 2023, 15, 4839 4 of 11 3. Results 3.1. General Consumption of Sport Supplements Of the total sample, 85.85% reported consuming supplements, while 15 of the 106 sub- jects responded that they did not consume any type of sport supplement. Regarding sex, supplement consumption was higher in men (89.2%) than in women (78.1%), with no statistical differences between them (p = 0.143). In the analysis of the results by level of competition, the percentage of autonomous athletes who consumed supplements was 77.5%, in athletes at the national level it was 90.0%, while in athletes who competed at the international level the consumption was 100%, with no differences between levels (p = 0.126). Table 2 shows the supplements consumed according to the different categories es- tablished by the AIS. With respect to total supplement consumption, differences were observed at the competitive level between international and regional athletes (p = 0.011). However, no differences were appreciated based on sex (F = 2.248; p = 0.466), with a total consumption of 4.8 ± 3.7 and 5.4 ± 5.7 for men and women, respectively. No interactions were observed between level and sex (F = 0.306; p = 0.737). Table 2. Descriptive data of the SS consumed according to the different categories defined by the AIS as a function of sex and level of competition. Variable Sex Level of Competition M F R N I Total Mean ± SD Mean ± SD Mean ± SD Mean ± SD Mean ± SD Med IQ Mean ± SD Med IQ Total SS 4.8 ± 3.7 5.4 ± 5.7 4.0 ± 3.8 5.1 ± 3.7 9.5 ± 9.6 6.0 26 5.0 ± 4.4 5.0 27 Group A Sports food 1.1 ± 1.0 1.0 ± 1.2 1.0 ± 1.1 1.1 ± 1.0 1.2 ± 1.5 0.5 3 1.1 ± 1.1 1.0 4 Medical supplement 0.4 ± 0.6 0.8 ± 0.9 0.3 ± 0.5 0.6 ± 0.8 1.3 ± 1.0 1.0 3 0.52 ± 0.7 0.0 3 Performance supplement 1.0 ± 1.1 0.9 ± 1.1 0.8 ± 1.1 1.0 ± 1.9 1.7 ± 1.6 1.5 4 1.0 ± 1.1 1.0 4 Total Group A 2.5 ± 1.9 2.7 ± 2.4 2.1 ± 2.0 2.7 ± 1.9 4.2 ± 3.7 3.0 10 2.5 ± 2.1 2.0 10 Group B 0.5 ± 0.7 0.5 ± 0.7 0.5 ± 0.6 0.5 ± 0.7 1.2 ± 1.2 1.0 3 0.5 ± 0.7 0.0 3 Group C 0.5 ± 0.6 0.3 ± 0.4 0.3 ± 0.6 0.5 ± 0.6 0.4 ± 0.6 0.5 1 0.4 ± 0.6 0.0 2 AIS: Australian Institute of Sport; SS: sport supplements; SD: standard deviation; M: male; F: female; R: regional; N: national; I: international; Group A: supplements with solid scientific evidence in specific situations under established protocols; Group B: components with emerging evidence that should be used in research or clinical settings; Group C: supplements with limited evidence and effects on performance; gender assigned at birth. For Group A, no differences were observed between sexes or levels or for the sex– level interaction for the sports food, performance supplement or total intake. However, differences were observed for the group of medical supplements between competition levels (international athletes, p = 0.004 vs. regional and p = 0.037 vs. national athletes), with consumption being higher as the level of the athletes increased. Likewise, differences between sexes were noted in this group (F = 3.797; p = 0.002), with higher consumption in women than in men (0.4 ± 0.6 vs. 0.8 ± 0.9). Table 3 describes the differences between supplement consumption according to level of competition, sex and the interaction between both. Regarding Group B, no differences were observed between sexes (F = 1.591; F = 0.860), levels of competition (F = 2.656; p = 0.075) or the interaction between sex and level of competition (F = 0.279; p = 0.860). Finally, for group C supplement consumption, differences were observed with respect to sex (F = 13.297; p = 0.029), with 0.5 ± 0.6 vs. 0.3 ± 0.4 for males and females, respectively. However, no differences were seen between levels or for the sex–level-of-competition interaction. 3.2. Most-Consumed Supplements by Competitive Level and Sex Table 4 shows those supplements that were consumed by more than 10% of the sample. The most-consumed supplements were caffeine (37%), followed by energy bars and sport drinks (34% for both) and creatine (31.1%). With respect to sex, differences were only observed for iron consumption (p < 0.001), with higher consumption in women than in men (17.6% vs. 56.3%). Differences between levels were observed for recovery shakes Nutrients 2023, 15, 4839 5 of 11 (83.3% vs. 20% vs. 7.5%, p < 0.001; for international, national and regional athletes) and vitamin D (50.0% vs. 18.3% vs. 10.0%, p = 0.047; for international, national and regional athletes). The most-consumed supplements in the sport food subgroup for women were sport drinks (34%), contrary to men where the use of sports bars was a little bit higher (36.5%). Regarding medical supplements, iron was the main supplement for both sexes (17.6 vs. 56.3 for male and female). Table 3. ANOVA of the SS consumed according to the different categories defined by the AIS as a function of sex, level of competition and their interaction. Variable Sex Level of Competition Sex–Level-of-Competition (Mean ± SD) F p F p R N I F p M F M F M F Total SS 2.248 0.466 4.582 0.012 # 4.0 ± 3.6 3.9 ± 4.5 5.1 ± 3.4 5.2 ± 4.5 7.5 ± 9.1 10.5 ± 11.1 0.306 0.737 Group A Sports food 1.330 0.726 0.102 0.903 1.1 ± 1.1 0.8 ± 1.1 1.1 ± 0.9 1.1 ± 1.4 1.5 ± 2.1 1.0 ± 1.4 0.270 0.764 Medical supplement 3.797 0.002 * 5.693 0.005 #$ 0.24 ± 0.4 0.6 ± 0.7 0.5 ± 0.7 0.8 ± 0.9 0.5 ± 0.7 1.8 ± 1.0 1.138 0.325 Performance supplement 0.014 0.592 2.167 0.120 0.8 ± 1.1 0.6 ± 1.3 1.1 ± 1.1 0.8 ± 0.7 1.5 ± 2.1 1.8 ± 1.7 0.162 0.850 Total group A 0.671 0.560 3.066 0.051 2.1 ± 2.0 2.0 ± 2.3 2.7 ± 1.8 2.8 ± 2.1 3.5 ± 4.9 4.5 ± 3.7 0.163 0.849 Group B 1.591 0.860 2.656 0.075 0.5 ± 0.7 0.5 ± 0.5 0.5 ± 0.8 0.4 ± 0.5 1.0 ± 1.4 1.3 ± 1.3 0.279 0.757 Group C 13.297 0.029 * 1.884 0.157 0.3 ± 0.7 0.1 ± 0.3 0.6 ± 0.6 0.3 ± 0.5 0.5 ± 0.7 0.5 ± 0.6 0.151 0.860 AIS: Australian Institute of Sport; SS: sport supplements; SD: standard deviation; M: male; F: female; R: regional; N: national; I: international; Group A: supplements with solid scientific evidence in specific situations under established protocols; Group B: components with emerging evidence that should be used in research or clinical settings; Group C: supplements with limited evidence and effects on performance; gender assigned at birth. * Statistical difference at p < 0.05 between male and female. # Statistical difference at p < 0.05 between regional and international athletes. $ Statistical difference at p < 0.05 between national and international athletes. Table 4. Distribution (%) of the most-consumed supplements (>10%) as a function of sex and level of competition according to the categories defined by the AIS. Category Supplement Name Total (%) Sex (%) Level of Competition (%) M F p R N I p Group A Sports foods Sport bars 34.0 36.5 28.1 0.273 30.0 38.3 16.7 0.451 Sport drinks 34.0 33.8 34.4 0.561 27.5 38.3 33.3 0.533 Sports gel 21.7 21.6 21.9 0.582 22.5 21.7 16.7 0.949 Whey protein 30.2 29.7 31.3 0.525 25.0 31.7 50.0 0.429 Recovery shakes 18.9 20.3 15.6 0.394 7.5 20.0 83.3 <0.001 * Medical supplements Iron 29.2 17.6 56.3 <0.001 * 22.5 30.0 66.7 0.084 Vitamin D 17.0 14.9 21.9 0.269 10.0 18.3 50.0 0.047 * Performance supplements β-Alanine 20.8 20.3 21.9 0.521 12.5 23.3 50.0 0.081 Caffeine 37.7 36.5 40.6 0.424 35.0 36.7 66.7 0.318 Creatine 31.1 36.5 18.8 0.054 20.0 38.3 33.3 0.151 Group B Vit C 19.8 20.3 18.8 0.542 17.5 20.0 33.3 0.662 Group C BCAA 10.4 12.2 6.3 0.295 10.0 10.0 16.7 0.873 Glutamine 11.3 12.2 9.4 0.482 5.0 15.0 16.7 0.276 AIS: Australian Institute of Sport; M: male; F: female; R: regional; N: national; I: international; Group A: sup- plements with solid scientific evidence in specific situations under established protocols; Group B: components with emerging evidence that should be used in research or clinical settings; Group C: supplements with limited evidence and effects on performance; gender assigned at birth. * Statistical difference at p < 0.05. For performance supplements, differences were observed with caffeine and creatine being the most consumed for men (36.5%) and only caffeine for women (40.6%). Finally, for group C, both BCAA and glutamine were the most-consumed ones for males (12.2%), but not for females (glutamine = 9.4%). As for the level of the athlete, the most-consumed supplements for international athletes were recovery shakes (83.3%), followed by iron and caffeine (66.7%). The national-level athletes’ most-consumed supplements were creatine, sports bars and sport drinks (38.3%), while caffeine was the most-consumed one by regional athletes (35%). Nutrients 2023, 15, 4839 6 of 11 3.3. Information about the Place of Purchase, Recommendations and Consumption Patterns Most athletes took supplements on training and competition days (39.62%). The daily consumption of supplements was 26.42%, followed by training (14.15%) and competition (11.32%). No differences were observed between genders (p = 0.106) as opposed to between categories for daily consumption (p = <0.00). Thus, 33.3% of the international athletes consumed it daily, while only 18.3% or 7.5% did so in the case of national and regional ones. In analyzing the moment of consumption, most of the sample used them after (56.60%) or before (50.94%) practicing exercise, followed by during training (20.75%). Only a small percentage responded that it was taken during the holiday period (1.89%) or indifferently (7.55%). No differences were observed for levels but between levels for pre- and post- training consumption (p = 0.007), which varied according to the level of competition (33% vs. 20% vs. 25% for international, national and regional athletes, respectively). The principal objective of consumption was to improve performance (70.75%), fol- lowed by taking care of their health (35.85%) and palliating dietary deficits (16.98%). Finally, of the 106 middle-distance runners, only 6.60% consumed them for health problems or necessity (3.77%). In this area, no differences were observed between sexes (p = 0.564) or levels of competition (p = 0.086). The primary place of purchase was the internet (51.89%), followed by specialized stores (26.42%) or a pharmacy (24.54%). Other minority sources of purchase were herbalists (12.26%), sports monitors (3.77%), friends (1.89%) or parapharma- cies (0.94%), with no statistically significant differences (p = 0.082 and p = 0.545 for gender and level). Finally, those who encouraged the use of SS were mainly coaches (37.74%), followed by dieticians–nutritionists (26.42%), teammates (21.70%) or physicians (16.04%). There were other people and sources that recommended its use such as friends and the internet (8.49%) or social network profiles (4.72%). Likewise, there were no differences between levels (p = 0.919) or genders (p = 0.410). 4. Discussion The main objective of this study was to analyze the supplementation patterns in middle-distance runners, as well as the differences between genders and level of compe- tition. The results indicate that the main differences between levels are observed both in total consumption and in the intake of medical supplements, with these being greater as the level of the athlete increases. Similarly, differences between levels were also observed in the consumption of medical supplements, as well as in pre- and post-training intake. This indicates that, although most athletes place emphasis on performance enhancement via supplementation, higher-level athletes also use these aids to maintain a better state of health and recover between sessions. Of the total sample, 85.85% responded that they consumed SS, which was higher than the consumption in other disciplines such as fencing or sailing [26,35], but lower than in sports such as rowing, trail running or tennis (100%, 93.8% and 88.6%, respec- tively) [25,29,31]. No differences were noted for sexes or competition levels, in line with recent research [25,35]. Comparing the data obtained with a sample of athletes from dif- ferent disciplines, the consumption of SS in middle-distance runners is higher (85% vs. 77%) [36]. Although there have been previous attempts to investigate supplementation patterns in athletes [22], one contained a limited sample while the other had only a few supplements [33,37] and no one has conducted it exclusively in middle-distance athletes. Therefore, this is the first to do so using a representative sample of middle-distance event participants and a broad list of SS. With respect to total SS consumption, there were differences between international and regional athletes, which had been previously noted in all types of sportsmen and women [33]. With respect to the different groups established by the AIS according to the level of evidence [13], in group A, no differences were observed between levels and genders, contrary to other recent studies [32,35,38]. However, differences between levels were close to being statistically significant (p = 0.051). Within the subgroups that exist in group A, only differences in medical supplements are observed for gender, which may be Nutrients 2023, 15, 4839 7 of 11 primarily due to the higher consumption of iron among women compared to men (56.3% vs. 17.6%). For the level of competition, differences were also observed in this subgroup, being higher as the level increased. These two findings are contrary to the results from other sports, where no differences have been observed for this subgroup between athlete levels or genders [26,29,35,39]. Regarding group B, which includes SS with emerging evidence but in need of future research, no differences are observed for level and gender, in line with the results in other sports [25,26,29,35,39]. Finally, in group C (supplements with insufficient scientific evidence to support its use), differences were noted between sexes, in line with some [25], but not all, recent evidence [26,35,39]. This could be due to the athlete’s knowledge, which is worse as the level of competition decreases [40]. Taking into account the days of sport practice when they usually take the SS, 39.62% responded that they take them during training and competition, followed by daily consump- tion and solely on training days, at 26.42% and 14.15%, respectively. Although the main sporting day is similar to that of other sports such as mountain running or rowing [25,29,39], the second and third causes differ between sports. This could be due to differences in the physiological demands of each event, as well as the average duration and energetic require- ments of training sessions. Differences were noted in daily consumption between levels of competition, indicating that the main difference between higher- and lower-level athletes was the use of medical supplements on a daily basis. On the other hand, the majority of middle-distance athletes take SS after (56.60%) or before sports practice (50.94%), while a lower percentage take them during sports practice. The duration of middle-distance sessions rarely surpasses 90–120 min [7], while other sports training sessions usually exceed this time, in which they will need to provide higher nutrition and hydration [25]. Here too, differences between levels are observed for pre- and post-consumption, demonstrating how top-level runners place greater importance on preparing for training or recovering for an upcoming workout. In analyzing the reasons for its consumption, the main one is to improve their performance (70.75%), followed by health care (35.85%) and to palliate a dietary deficit (16.98%), similar to other sport disciplines [26,29,31,32,35,36]. Concerning the person who motivated the consumption of SS, the main motivator was the coach (37.74%), followed by dietitians–nutritionists (26.42%), which showed a worse advisor in the case of middle-distance runners compared to other sports [25,39]. The next advisors were teammates, followed by physicians, indicating the existence of other sports modalities with a worse source of support [26,35]. In this sense, dietitians– nutritionists are the most appropriate when choosing one supplement or another regardless of the level of scientific evidence [26,41,42]. Finally, most athletes purchased SS on the internet (51.89%), followed by specialized stores (26.42%) and pharmacies (24.53%). In this sense, both pharmacy and internet products may contain quantities different from those advertised or contaminated substances, which may also put the athletes at risk of unintended doping [43], so athletes tend to go to specialized stores in order to avoid these problems [12,42]. Finally, with regard to the most-consumed SS, we can appreciate caffeine in the first place. Caffeine is a natural stimulant for the central nervous system, possesses various suggested benefits for enhancing performance and is one of the supplements with the highest scientific evidence supporting its use [15]. These advantages encompass enhanced neuromuscular functionality and a decrease in fatigue and perceived effort levels during physical exertion, among others [44]. The following most-consumed SS were sport drinks (formulated to provide a balanced combination of carbohydrates and liquids, facilitating athletes in rehydrating and replenishing energy simultaneously during and after their workout) and sport bars (created as a portable source of carbohydrates, helping meeting carbohydrate intake goals) [13], which also belong to Group A, such as caffeine. These two supplements help mainly in carbohydrate replenishment post or during training or to reach the recommended daily intake of carbohydrates, which can be up to 70% of the total diet or around 6–12 CHO · kg−1 · BW · day−1. In this sense, carbohydrate intake both Nutrients 2023, 15, 4839 8 of 11 during [45] and immediately after [46] exercise limits fatigue and improves performance in the following training sessions. The next most-consumed supplement was isolated protein, with considerable scientific evidence supporting its use [13], which appears necessary both for the recovery and repair of damaged myofibrillar proteins and to optimize mitochondrial and possibly sarcoplasmic protein synthesis [47]. However, this seems unnecessary in most cases, since athletes tend to consume more protein than any high recommendation [47]. Continuing with the SS that can provide more benefits among those consumed by more than 10% of the sample, we find iron or β-Alanine. Iron plays a fundamental role in the transport of oxygen and a high prevalence of anemia has been observed among middle-distance runners [22]. A small decrease in hemoglobin content (subclinical anemia) leads to a significant decrease in oxygen transport capacity and, therefore, a decrease in performance [48]. Thus, it is important to monitor these variables on a recurring basis in order to supplement if necessary. On the other hand, β-Alanine acts as an intracellular buffer by increasing the concentration of muscular carnosine [49]. Since high-intensity exercise (usually performed by middle- distance runners both in training and competition [7]) increases the amount of hydrogen ions and lowers the intracellular pH from 7.0 to 6.6, supplementation with β-Alanine may improve the ability to withstand this drop, limiting muscular fatigue. However, the determination of whether supplementation enhances performance in elite middle-distance athletes is challenging due to insufficient data and non-performance-related tests [47]. Despite this, considering the absence of side effects and potential performance benefits, individual athletes and their support teams may want to try β-Alanine supplementation to assess its effectiveness for them [1]. Finally, it is important to note the very low percentage of athletes using inorganic nitrates or beetroot juice as SS (6.60%). This supplementation seems to improve performance via the bioavailability of nitric oxide, improving exercise efficiency (decreased O2 cost at the same absolute workload) [50]. However, this low use may be due to variability in the response to its supplementation [1] or decreased effects as the physiological capabilities of the athletes increase [50]. It is important to mention that, although a large part of the SS consumed by middle- distance runners in this study belong to group A, it is also observed that there is still a fairly large consumption of supplements with little or no scientific evidence (groups B and C). This has also been observed in other sports, so it is important that athletes use reliable sources of information when deciding which supplements to consume [25,51]. In addition, the present research has several limitations. First of all, the sample is larger than that of other studies with the same population, but a greater participation of international athletes is necessary. In addition, it was the athletes themselves who responded retrospectively to the consumption of SS, which could lead to errors in the number or type of supplements. Therefore, it is necessary to compare and have the support of different federations or institutions worldwide to check if the consumption is similar depending on the competitive level or gender. 5. Conclusions Supplement consumption in middle-distance running is similar to that in other sports. The main differences between levels are seen in the total supplement consumption and in the consumption of medical supplements, as well as in daily or pre- and post-exercise con- sumption, with these being higher as the level of competition increases. On the other hand, the differences between sexes are found in the consumption of both medical supplements and supplements with limited evidence. Middle-distance runners should improve both their sources of information and places of purchase in order to avoid supplements with low scientific evidence or contaminated/fraudulent products. Author Contributions: Conceptualization, A.D.A. and A.C.-B.; Methodology, A.D.A., A.M.A.-B. and E.M.-M.; Software, A.M.A.-B.; Formal Analysis, A.D.A., A.M.A.-B. and A.G.; Investigation, A.D.A. and A.M.A.-B., Data Curation, A.D.A. Writing—Original Draft Preparation, A.D.A.; Writing—Review Nutrients 2023, 15, 4839 9 of 11 & Editing, A.M.A.-B., E.M.-M., A.G. and A.C.-B.; Supervision; A.C.-B. Project Administration; A.C.-B. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. The protocol complied with the provisions of the Declaration of Helsinki for human research and was approved by the ethical committee of the University of Deusto (ETK-14/23-24) dated 26 October 2023. Data Availability Statement: Data of the article are available in the tables of this paper or on request from the corresponding author. Acknowledgments: Asier del Arco gives thanks for the distribution of the questionnaire and ded- icates the article to the memory of Carlos and Angel Basas, for their contributions to the world of athletics at the national and world levels. Conflicts of Interest: The authors of the present article declare no conflict of interest. References 1. Stellingwerff, T.; Bovim, I.M.; Whitfield, J. Contemporary nutrition interventions to optimize performance in middle-distance runners. Int. J. Sport Nutr. Exerc. Metab. 2019, 29, 106–116. [PubMed] 2. Billat, L.V. Interval training for performance: A scientific and empirical practice. Special recommendations for middle- and long-distance running. Part I: Aerobic interval training. Sports Med. 2001, 31, 13–31. [CrossRef] [PubMed] 3. Sandford, G.N.; Stellingwerff, T. “Question Your Categories”: The Misunderstood Complexity of Middle-Distance Running Profiles with Implications for Research Methods and Application. Front. Sports Act. Living 2019, 1, 28. 4. Costill, D.L.; Daniels, J.; Evans, W.; Fink, W.; Krahenbuhl, G.; Saltin, B. Skeletal muscle enzymes and fiber composition in male and female track athletes. J. Appl. Physiol. 1976, 40, 149–154. [CrossRef] [PubMed] 5. Hermansen, L.; Osnes, J.B. Blood and muscle pH after maximal exercise in man. J. Appl. Physiol. 1972, 32, 304–308. [CrossRef] 6. Brandon, L.J. Physiological factors associated with middle distance running performance. Sports Med. 1995, 19, 268–277. [CrossRef] 7. Haugen, T.; Sandbakk, Ø.; Enoksen, E.; Seiler, S.; Tønnessen, E. Crossing the Golden Training Divide: The Science and Practice of Training World-Class 800- and 1500-m Runners. Sports Med. 2021, 51, 1835–1854. 8. Bellinger, P.; Derave, W.; Lievens, E.; Kennedy, B.; Arnold, B.; Rice, H.; Minahan, C. Determinants of Performance in Paced and Maximal 800-m Running Time Trials. Med. Sci. Sports Exerc. 2021, 53, 2635–2644. 9. Bellinger, P.; Derave, W.; Lievens, E.; Kennedy, B.; Arnold, B.; Rice, H.; Minahan, C. Determinants of last lap speed in paced and maximal 1500-m time trials. Eur. J. Appl. Physiol. 2021, 121, 525–537. 10. Mujika, I.; Padilla, S. Scientific bases for precompetition tapering strategies. Med. Sci. Sports Exerc. 2003, 35, 1182–1187. [CrossRef] 11. Maughan, R.J.; Burke, L.M.; Dvorak, J.; Larson-Meyer, D.E.; Peeling, P.; Phillips, S.M.; Rarson, E.S.; Walsh, N.P.; Garthe, I.; Geyer, H.; et al. IOC consensus statement: Dietary supplements and the high-performance athlete. Br. J. Sports Med. 2018, 52, 439–455. [PubMed] 12. Garthe, I.; Maughan, R.J. Athletes and supplements: Prevalence and perspectives. Int. J. Sport. Nutr. Exerc. Metab. 2018, 28, 126–138. 13. Australian Institute of Sport Position Statement. Supplements and Sports Foods in High Performance Sports; The Australian Institute of Sport: Bruce, Australia, 2021; 9p. 14. Ganio, M.; Klau, J.; Casa, D.; Armstrong, L.E.; Maresh, C. Effect of Caffeine on Sport Endurance: A Systematic Review. J. Strength Cond. Res. 2009, 23, 315–324. [CrossRef] [PubMed] 15. Christensen, P.M.; Shirai, Y.; Ritz, C.; Nordsborg, N.B. Caffeine and bicarbonate for speed. A meta-analysis of legal supplements potential for improving intense endurance exercise performance. Front. Physiol. 2017, 8, 1–16. 16. Saunders, B.; Elliott-Sale, K.; Artioli, G.G.; Swinton, P.A.; Dolan, E.; Roschel, H.; Sale, C.; Gualano, B. β-Alanine supplementation to improve exercise capacity and performance: A systematic review and meta-Analysis. Br. J. Sports Med. 2017, 51, 658–669. [PubMed] 17. Hobson, R.M.; Saunders, B.; Ball, G.; Harris, R.C.; Sale, C. Effects of β-alanine supplementation on exercise performance: A meta-analysis. Amino Acids 2012, 43, 25–37. 18. Ducker, K.J.; Dawson, B.; Wallman, K.E. Effect of beta-alanine supplementation on 800-m running performance. Int. J. Sport Nutr. Exerc. Metab. 2013, 23, 554–561. [CrossRef] 19. Schauf, M.; Ball, T.E. Effects of sodium bicarbonate ingestion on anaerobic performance of women: Dosage effect. J. Hum. Mov. Stud. 1996, 31, 179–187. 20. Bird, S.R.; Wiles, J.; Robbins, J. The effect of sodium bicarbonate ingestion on 1500-m racing time. J Sports Sci. 1995, 13, 399–403. [CrossRef] Nutrients 2023, 15, 4839 10 of 11 21. Wilkes, D.; Gledhill, N.; Smyth, R. Effect of acute induced metabolic alkalosis on 800-m racing time. Med. Sci. Sports Exerc. 1983, 15, 277–280. [CrossRef] 22. Tabata, S.; Yamasawa, F.; Torii, S.; Manabe, T.; Kamada, H.; Namba, A.; Kato, J.; Kaneko, H.; Tahara, K.; Tsukahara, Y. Use of nutritional supplements by elite Japanese track and field athletes. J. Int. Soc. Sports Nutr. 2020, 17, 38. 23. Escribano-Ott, I.; Mielgo-Ayuso, J.; Calleja-González, J. A glimpse of the sports nutrition awareness in Spanish basketball players. Nutrients 2022, 14, 27. [CrossRef] 24. Sekulic, D.; Tahiraj, E.; Maric, D.; Olujic, D.; Bianco, A.; Zaletel, P. What drives athletes toward dietary supplement use: Objective knowledge or self-perceived competence? Cross-sectional analysis of professional team-sport players from Southeastern Europe during the competitive season. J. Int. Soc. Sports Nutr. 2019, 16, 1–9. [CrossRef] 25. Jiménez-Alfageme, R.; Rubio-Quintanilla, N.; Romero-García, D.; Sanchez-Oliver, A.J.; Sospedra, I.; Martínez-Sanz, J.M. Are the Consumption Patterns of Sports Supplements Similar among Spanish Mountain Runners? Nutrients 2023, 15, 262. [CrossRef] 26. Mata, F.; Domínguez, R.; López-Samanes, Á.; Sánchez-Gómez, Á.; Jodra, P.; Sánchez-Oliver, A.J. Analysis of the consumption of sports supplements in elite fencers according to sex and competitive level. BMC Sports Sci. Med. Rehabil. 2021, 13, 50. 27. Wiens, K.; Erdman, K.A.; Stadnyk, M.; Parnell, J.A. Dietary supplement usage, motivation, and education in young Canadian athletes. Int. J. Sport Nutr. Exerc. Metab. 2014, 24, 613–622. [CrossRef] [PubMed] 28. Aguilar-Navarro, M.; Baltazar-Martins, G.; Brito de Souza, D.; Muñoz-Guerra, J.; del Mar Plata, M.; Del Coso, J. Gender Differences in Prevalence and Patterns of Dietary Supplement Use in Elite Athletes. Res. Q. Exerc. Sport 2021, 92, 659–668. [PubMed] 29. Domínguez, R.; López-Domínguez, R.; López-Samanes, Á.; Gené, P.; González-Jurado, J.A.; Sánchez-Oliver, A.J. Analysis of Sport Supplement Consumption and Body Composition in Spanish Elite Rowers. Nutrients 2020, 12, 3871. [PubMed] 30. Fleming, J.A.; Naughton, R.J.; Harper, L.D. Investigating the Nutritional and Recovery Habits of Tennis Players. Nutrients 2018, 10, 443. [CrossRef] 31. Sánchez-Oliver, A.J.; Mata-Ordoñez, F.; Domínguez, R.; López-Samanes, A. Use of nutritional supplements in amateur tennis players. J. Phys. Educ. Sport 2018, 18, 775–780. 32. Comes, A.V.; Sánchez-Oliver, A.J.; Martínez-Sanz, J.M.; Domínguez, R. Analysis of Nutritional Supplements Consumption by Squash Players. Nutrients 2018, 10, 1341. [CrossRef] [PubMed] 33. Knapik, J.J.; Steelman, R.A.; Hoedebecke, S.S.; Austin, K.G.; Farina, E.K.; Lieberman, H.R. Prevalence of Dietary Supplement Use by Athletes: Systematic Review and Meta-Analysis. Sports Med. 2016, 46, 103–123. [PubMed] 34. Sánchez-Oliver, A.J. Suplementación Nutricional en la Actividad Físico-Deportiva: Análisis de la Calidad del Suplemento Proteico Consumido; Universidad de Granada: Granada, Spain, 2012; p. 179. 35. Caraballo, I.; Domínguez, R.; Guerra-Hernandez, E.J.; Sánchez-Oliver, A.J. Analysis of Sports Supplements Consumption in Young Spanish Elite Dinghy Sailors. Nutrients 2020, 12, 993. [CrossRef] [PubMed] 36. Baltazar-Martins, G.; Brito de Souza, D.; Aguilar-Navarro, M.; Muñoz-Guerra, J.; Del Mar Plata, M.; Del Coso, J. Prevalence and patterns of dietary supplement use in elite Spanish athletes. J. Int. Soc. Sports Nutr. 2019, 16, 30. [CrossRef] 37. Tscholl, P.; Alonso, J.M.; Dollé, G.; Junge, A.; Dvorak, J. The use of drugs and nutritional supplements in top-level track and field athletes. Am. J. Sports Med. 2010, 38, 133–140. [CrossRef] 38. Sánchez-Oliver, A.J.; Domínguez, R.; López-Tapia, P.; Tobal, F.M.; Jodra, P.; Montoya, J.J.; Guerra-Hernández, E.J.; Ramon-Álvarez, J.J. A Survey on Dietary Supplement Consumption in Amateur and Professional Rugby Players. Foods 2021, 10, 7. [CrossRef] 39. Jiménez-Alfageme, R.; Domínguez, R.; Sánchez Oliver, A.J.; Sospedra, I.; Gil Izquierdo, Á.; Martínez-Sanz, J.M. Sports supplements use in mountain runners: Differences by sex and competitive level. Nutr. Hosp. 2022, 39, 1341–1348. 40. Trakman, G.L.; Forsyth, A.; Devlin, B.L.; Belski, R. A Systematic Review of Athletes’ and Coaches’ Nutrition Knowledge and Reflections on the Quality of Current Nutrition Knowledge Measures. Nutrients 2016, 8, 570. [CrossRef] 41. Wardenaar, F.C.; Ceelen, I.J.M.; Van Dijk, J.W.; Hangelbroek, R.W.J.; Van Roy, L.; Van Der Pouw, B.; De Vries, J.H.M.; Mensink, M.; Witkamp, R.F. Nutritional Supplement Use by Dutch Elite and Sub-Elite Athletes: Does Receiving Dietary Counseling Make a Difference? Int. J. Sport Nutr. Exerc. Metab. 2017, 27, 32–42. [CrossRef] 42. Martínez-Sanz, J.M.; Mata, F.; Ripoll, M.S.; Braza, J.M.P.; Martinez Segura, A.; Sánchez-Oiver, A.J.; Cortell Tormo, J.M. Fraud in nutritional supplements for athletes: A narrative review. Nutr. Hosp. 2021, 38, 839–847. 43. Martínez-Sanz, J.M.; Sospedra, I.; Ortiz, C.M.; Baladía, E.; Gil-Izquierdo, A.; Ortiz-Moncada, R. Intended or Unintended Doping? A Review of the Presence of Doping Substances in Dietary Supplements Used in Sports. Nutrients 2017, 9, 1093. [CrossRef] [PubMed] 44. Burke, L.M. Caffeine and sports performance. Appl. Physiol. Nutr. Metab. 2008, 33, 1319–1334. [CrossRef] [PubMed] 45. Clauss, M.; Skattebo, Ø.; Dæhli, M.R.; Valsdottir, D.; Bastani, N.E.; Johansen, E.I.; Kolnes, K.J.; Skålhegg, B.S.; Jensen, J. Carbohy- drate Ingestion during Prolonged Cycling Improves Next-Day Time Trial Performance and Alters Amino Acid Concentrations. Med. Sci. Sports Exerc. 2023. online ahead of print. [CrossRef] 46. Blom, P.C.S.; Høstmark, A.T.; Vaage, O.; Kardel, K.R.; Mæhlum, S. Effect of different post-exercise sugar diets on the rate of muscle glycogen synthesis. Med. Sci. Sports Exerc. 1987, 19, 491–496. [CrossRef] 47. Stellingwerff, T.; Boit, M.; Res, P. Nutritional strategies to optimize training and racing in middle-distance athletes. J. Sports Sci. 2007, 25, 17–28. [CrossRef] [PubMed] 48. San-Millán, I. Blood Biomarkers in Sports Medicine and Performance and the Future of Metabolomics. Methods Mol. Biol. 2019, 1978, 431–446. Nutrients 2023, 15, 4839 11 of 11 49. Harris, R.C.; Stellingwerff, T. Effect of β-Alanine Supplementation on High-Intensity Exercise Performance. Nestle Nutr. Inst. Workshop 2013, 76, 61–71. 50. Jones, A.M. Influence of dietary nitrate on the physiological determinants of exercise performance: A critical review. Appl. Physiol. Nutr. Metab. 2014, 39, 1019–1028. [CrossRef] 51. Jiménez-Alfageme, R.; Domínguez, R.; Sanchez-Oliver, A.J.; Tapia-Castillo, P.; Martínez-Sanz, J.M.; Sospedra, I. Analysis of the Consumption of Sports Supplements in Open Water Swimmers according to the Competitive Level. Nutrients 2022, 14, 5211. [CrossRef] Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Are Supplements Consumed by Middle-Distance Runners Evidence-Based? A Comparative Study between Level of Competition and Sex.

11-20-2023

Del Arco, Asier,Martinez Aguirre-Betolaza, Aitor,Malchrowicz-Mośko, Ewa,Gogojewicz, Anna,Castañeda-Babarro, Arkaitz

eng

PMC7510966

RESEARCH ARTICLE Area per player in small-sided games to replicate the external load and estimated physiological match demands in elite soccer players Andrea RiboliID1,2*, Giuseppe Coratella2, Susanna Rampichini2, Emiliano Ce´ ID2, Fabio EspositoID2 1 Performance Department, Atalanta B.C., Bergamo, Italy, 2 Department of Biomedical Sciences for Health, Università degli Studi di Milano, Milano, Italy * riboliandrea@outlook.com Abstract The current study determined the area-per-player during small- or large-sided games with or without goalkeeper that replicates the relative (m
min-1) total distance, high-inten- sity running distance, sprint distance and metabolic power covered during official matches. Time-motion analysis was performed on twenty-five elite soccer-players during 26 home-matches. A total of 2565 individual samples for SSGs using different pitch sizes and different number of players were collected and classified as SSGs with (SSG-G) or without goalkeeper (SSG-P). A between-position comparison was also performed. The area-per-player needed to replicate the official match demands was largely higher in SSG-G vs SSG-P for total distance [187±53 vs 115±35 m2, effect size (ES): 1.60 95%CI 0.94/2.21], high-intensity running distance [262±72 vs 166±39 m2, ES: 1.66(0.99/2.27)] and metabolic power [177±42 vs 94±40, ES: 1.99(1.31/2.67)], but similar for sprint dis- tance [(316±75 vs 295±99 m2, ES: 0.24(-0.32/0.79)] with direction of larger area-per- player for sprint distance > high-intensity running > total distance metabolic power for both SSG-G and SSG-P. In SSG-G, forwards required higher area-per-player than cen- tral-defenders [ES: 2.96(1.07/4.35)], wide-midfielders [ES: 2.45(0.64/3.78)] and wide- defenders [ES: 3.45(1.13/4.99)]. Central-midfielders required higher area-per-player than central-defenders [ES: 1.69(0.20/2.90)] and wide-midfielders [ES: 1.35(-0.13/ 2.57)]. In SSG-P, central defenders need lower area-per-player (ES: -6.01/-0.92) to overall replicate the match demands compared to all other positions. The current results may be used to gain knowledge of the SSGs relative to the match demands. This imply manipulating SSGs using higher or lower ApP, the presence of the goalkeeper or design specific rules to increase or decrease the position-specific demands with respect to the desired external load outcomes. PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 1 / 15 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Riboli A, Coratella G, Rampichini S, Ce´ E, Esposito F (2020) Area per player in small-sided games to replicate the external load and estimated physiological match demands in elite soccer players. PLoS ONE 15(9): e0229194. https://doi. org/10.1371/journal.pone.0229194 Editor: Luca Paolo Ardigò, Universita degli Studi di Verona, ITALY Received: January 31, 2020 Accepted: September 4, 2020 Published: September 23, 2020 Peer Review History: PLOS recognizes the benefits of transparency in the peer review process; therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. The editorial history of this article is available here: https://doi.org/10.1371/journal.pone.0229194 Copyright: © 2020 Riboli et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Introduction Small- or large-sided games are frequently used to replicate the soccer-specific match demands in terms of technical proficiency, tactical awareness, speed, acceleration/deceleration, and endurance performance [1]. To assess these demands, contemporary player-tracking technolo- gies such as global positioning system (GPS) or semi-automatic video-based multi-camera image system (MCIS), are typically used [2]. In small- or large-sided games (SSGs), the manip- ulation of pitch size, number of players per team, goalkeeper presence and technical rules modulate the soccer-specific demands depending on the aims of each practice session [1, 3]. Increments in pitch size or reduction in the number of players increases total distance (TD) covered, total high-intensity running distance (HIRD) and total sprint distance (TSD) [4, 5]. Conversely, when pitch size is reduced or the number of players is increased, players get more ball touches but they have not the space to reach the high-speed running, and the total distance covered is rather characterized by acceleration and deceleration (Acc/Dec) [5, 6]. To possibly combine the pitch size and number of players, the area per player (ApP, expressed as m2
player-1) has been introduced [1]. Lastly, SSGs can be performed with (SSG-G) or without goalkeepers (SSG-P), when the aim is to out-score the opponent team or to maintain ball pos- session as long as possible, respectively [1].Some authors reported higher TD and distances covered at different speed-thresholds during 2-, 3- and 4-a-side SSG-P than SSG-G [7]. Simi- larly, higher HIRD was found comparing 3-a-side [8], 5- and 7-a-side SSG-P than SSG-G [9]. Although TD, HIRD and TSD were found to be higher in SSG-P than SSG-G using the same pitch size [1, 10], other studies found lower HIRD in 3-a-side SSG-P than SSG-G [9], no differ- ences in TSD in 3-a-side vs 5-a-side SSG-P than SSG-G [9] or higher TSD and lower Acc/Dec in SSG-G compared to SSG-P [6]. These conflicting findings suggest that further investigation is needed [11]. The metabolic power (Pmet) approach has been recently proposed as a tool to estimate the energetic demands of variable-speed and accelerated/decelerated locomotion activities typi- cally seen in team sports [12, 13]. While it is difficult to measure directly the exact energy cost of changing speed, a metabolic power calculation based on a theoretical model has been used to estimate the energy cost of locomotion in team sports [12, 14]. However, this model was questioned since it may underestimate the actual net energy demand of soccer-specific exer- cises [15–17]. Additionally, the traditional speed-threshold approach was shown to provide similar external load compared to Pmet [18, 19]. Nevertheless, the metabolic power approach could capture the high-demanding locomotor activities independently of the actual speed reg- istered by GPS [16, 20], and it was shown to be a useful tool for the classification of the loco- motion intensity in team sports [21]. Previous studies have provided evidence for concurrent ecological validity to this approach, reporting correlations between Pmet and aerobic fitness variables during professional soccer matches [22] and with time above 85% of the maximal heart-rate in elite hockey matches [21]. Moreover, Pmet can be sensitive to decrements in run- ning performance during competition [23–25] and it could be used to account for positional differences [23, 25]. Therefore, the combination of the Pmet approach and the traditional speed-threshold metrics should be used to provides a more comprehensive assessment of the intermittent running demands typically occurring in team sports [15–17, 21, 24, 26–28]. An accurate comparison of the match vs training loads may help to plan the training ses- sions to condition the locomotor activities typically required during the official match and to optimize performance goals [5, 29]. Quantifying TD, HIRD, TSD, Acc/Dec and Pmet training loads relative to the game demands was suggested to be an important strategy when attempting to optimize position-specific loads in elite soccer practice [29]. Particularly, the locomotor activities during different SSGs compared to official matches are still under investigation. PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 2 / 15 Funding: The author(s) received no specific funding for this work Competing interests: The authors have declared that no competing interests exist Additionally, discriminating such locomotor activities by position could help to tailor the training session. Therefore, the present study aimed to: i) determine the ApP that could be used to replicate the official matches TD, HIRD, TSD, Acc/Dec (normalized as meters covered in one minute) and Pmet (normalized as W
kg-1) during both SSG-P and SSG-G; and ii) differ- entiate the ApP according to playing position. To increase the ecological validity, this was assessed in elite Serie A soccer players. Materials and methods Participants Twenty-five elite soccer players competing in Italian Serie A were involved in the present study (age: 27 ± 5 yrs; body mass: 79 ± 7 kg; body height: 1.84 ± 0.06 m). All participants were classified according to their position: central-defenders (n = 6), wide-defenders (n = 4), cen- tral-midfielders (n = 5), wide-midfielders (n = 5) and forwards (n = 5). The goalkeepers were excluded from data collection. The club’s medical staff certified the health status of each player. An injured player was excluded from data collection for at least one month after their return to full training. All participants gave their written consent after a full explanation of the pur- pose of the study and the experimental design. The Ethics Committee of the Università degli Studi di Milano approved the study and was performed in accordance with the principles of the Declaration of Helsinki (1975). Design The present investigation was carried out during the competition period across two seasons (August 2014 –May 2016). The participants undertook their traditional weekly training rou- tine. All sessions were performed on two grass pitches preserved by qualified operators and were conducted at the same time of day to limit the effects of circadian variation. A specialized and high-qualified physician staff recommended and monitored the diet regime of each player before and after every training session. Two different formats of SSGs were analyzed: SSG-G and SSG-P. A total of 2565 (1033 and 1532, respectively) individual GPS samples with a median of 37 (range = 12 to 62) and 56 (range = 25 to 86) in SSG-G and SSG-P respectively were undertaken for each player. The number of players ranged from 5vs5 to 10vs10, with a pitch area ranging from 800 m2 to 6825 m2 for SSG-G and 3v3 to 10vs10 with a pitch area from 400 m2 to 4550 m2 for SSG-P. Hence, ApP ranged from 67 m2 to 341 m2 for SSG-G and from 43 m2 to 341 m2 for SSG-P (for a detailed description of these parameters, see S1 and S2 Tables). ApP was calculated excluding the goalkeepers in SSG-G. Both small- or large-sided games were abbreviated as SSGs and specified by ApP. The SSGs were performed under the supervision and motivation of several coaches to keep up a high work-rate [3]. For the same reason, a ball was always available by prompt replacement when it went out-of-play [1]. In SSG-G, the corners were replaced by a prompt ball-in-game from the goalkeeper [9]. The SSGs were completed after a standardized 20-min warm-up under the guidance of club staff. Only official home matches (N = 26; indi- vidual samples = 228; individual sample range = 6 to 24) were assessed to ensure data consis- tency [11]. The home-match pitch size was 105 x 66 m, with a grass surface. To determine the ApP in both SSG-G and SSG-P that replicates the normalized TD, HIRD, TSD, Acc/Dec (m
min-1) and Pmet (W
kg-1) recorded during the official matches, we first recorded these variables during the official matches. Thereafter, we separately plotted each relationship between ApP and the normalized TD, HIRD, TSD, Acc/Dec and Pmet during SSG-G or SSG-P. Then, the mean values recorded during the official matches were used to PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 3 / 15 intersect each ApP/ TD, HIRD, TSD, Acc/Dec or Pmet relationship recorded in SSG-G or SSG-P to calculate the ApP that corresponded to the official match demands (Fig 1). Procedures For the aims of this study, the interchangeability of GPS and MCIS for TD, HIRD, TSD, Acc/ Dec and Pmet needed to be calculated as first step. A 10Hz GPS (K-Sport, Montelabbate, Italy) unit was used to collect data during the training sessions [30]. The GPS unit was placed within a dedicated pouch between the player’s shoulder blades (upper thoracic-spine) in a sports vest and worn under the playing jersey. Each device was turned on at least 15-min before each ses- sion to allow for acquisition of the satellite signal [6]. To reduce the inter-unit differences, each player wore the same unit for every training session over the whole investigation [31]. The locomotor activities during the official matches were collected using a computerized semi- automated MCIS (STATS LLC, Chicago, Illinois, USA) and processed by a dedicated software (K-SportOnline, K-Sport, Montelabbate, Italy). The system has previously been shown to pro- vide valid and reliable measurements of the match activity in soccer [32, 33]. During both training sessions and home-matches, total distance, total high-intensity run- ning distance (>15 km
h-1), total sprint distance (>24 km
h-1) [3, 11, 33] were measured. Fig 1. Graphical representation of the procedures used to determine the area per player in SSG-G or SSG-P that matches the official match demands. X-axis: the area per player in SSG-G or SSG-P; Y-axis: the SSG-G or SSG-P demands. The regression line shows how the area per player influences the SSGs demands. The horizontal dashed line represents the official match demands. From the intersection point of the regression line with the horizontal line (i.e. when the SSGs demands equate the official match demands), a vertical dotted line is drawn to the X-axis. The intersection point between the X-axis and the vertical dotted line is the calculated area per player in SSGs necessary to replicate the official match demands. https://doi.org/10.1371/journal.pone.0229194.g001 PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 4 / 15 Additionally, the total distance of velocity changes calculated using >2 m
s-2 accelerations and decelerations (Acc/Dec) were measured [4, 5]. The average metabolic power (Pmet) was calcu- lated following previous procedures [13, 27]. TD, HIRD, TSD and Acc/Dec were normalized as relative distance covered in one minute (m
min-1), while Pmet were normalized as watt per kilogram (W
kg-1); then all parameters were inserted into the data analysis. TD, HIRD, TSD and Acc/Dec were measured using either GPS or the MCIS. Therefore, to check the interchangeability of these two tracking technologies, a 10-min simulated match was monitored using both GPS and MCIS simultaneously [2, 34, 35]. All data were collected in the stadium where the official matches were played. For each dependent locomotor activity, a cali- bration equation was calculated to compare GPS and MCIS, as previously proposed [2, 34]. Statistical analysis Statistical analysis was performed using a statistical software package (SigmaPlot v-12.5, Systat Software Inc., San Jose, CA, USA). To check the normal distribution of the sampling, a Sha- piro-Wilk test was used. A Bland-Altman analysis was used to display the degrees of bias and the limits of agreement between the GPS and the MCIS. A linear regression analysis was used to calculate the correlation between GPS and MCIS. The Pearson’s product moment and the typical error of the estimate (TEE) were calculated to determine the relationship between the GPS and the MCIS. The correlation coefficient was interpreted as follows: r = 0.00–0.09 trivial, 0.10–0.29 small, 0.30–0.49 moderate, 0.50–0.69 large, 0.70–0.89 very large, 0.90–0.99 nearly per- fect; the threshold values for the TEE were interpreted as follows: >0.2 small, >0.6 moderate, >1.2 large and >2 very large [36]. A linear regression analysis was used to calculate the correla- tion between TD, HIRD, TSD, Acc/Dec, Pmet and the ApP during both SSG-G and SSG-P. Thereafter, a two-way ANOVA was used to calculate the difference in the optimal ApP in TD, HIRD, TSD, Acc/Dec, Pmet calculated for SSG (SSG-G vs SSG-P) and position (central-defend- ers, wide-defenders, central-midfielders, wide-midfielders and forwards). A post-hoc analysis (Holm-Sidak correction) was used to calculate the differences in the independent factors. The effect size with 95% confidence intervals (CI) was calculated and interpreted as follows: <0.20: trivial; 0.20–0.59: small; 0.60–1.19: moderate; 1.20–1.99: large; 2.00: very large [36]. Statistical significance was set at α < 0.05. Unless otherwise stated, all values are presented as mean ± standard deviation (SD). Results The magnitude of the GPS vs MCIS bias is shown in Fig 2. For each dependent parameter, Bland-Altman analysis and correlation graph with the respective calibration equation are shown. The bias between GPS vs MCIS were trivial for TD (-3.0 ± 1.3%, ES = -0.18, CI: -0.80/ 0.44), HIRD (-3.3 ± 1.6%, ES = -0.12, CI: -0.74/0.51), TSD (-3.9 ± 10.9%, ES = -0.11, CI: -0.44/ 0.22) and Acc/Dec (-4.1 ± 6.3%, ES = -0.19, CI: -0.80/0.44) and small for Pmet (-4.0 ± 0.6%, ES = -0.38, CI: -1.27/0.49). A small TEE was found between the MCIS and GPS for TD (TEE: 0.09, CI: 0.07/0.14), HIRD (TEE: 0.04, CI: 0.03/0.06), TSD (TEE: 0.08, CI: 0.07/0.10) and Pmet (TEE: 0.07, CI: 0.04/0.13), while a moderate TEE was found for Acc/Dec (TEE: 0.75, CI: 0.56/ 1.10). In addition, a nearly perfect correlation was observed for TD, HIRD, TSD and Pmet and a moderate correlation for Acc/Dec measured using GPS and MCIS (Fig 2). As shown in Fig 3, in SSG-P a very large correlation between the relative distance and the ApP was found for TD, HIRD, TSD and Pmet while a moderate correlation was found for Acc/ Dec. In SSG-G, a very large correlation between the relative distance and the ApP for TD, HIRD and TSD, a large correlation for Pmet and a moderate negative correlation for Acc/Dec PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 5 / 15 PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 6 / 15 were found. Because of the moderate correlations observed for Acc/Dec in both SSG-G and SSG-P, we did not perform the calculation or the ApP for Acc/Dec, given the high risk of bias. For both SSG-P and SSG-G, the ApP necessary to replicate the relative distance recorded during the matches for TD, HIRD, TSD and Pmet is shown in Table 1. No SSG × position inter- action was found (p = 0.674) for ApP for TD. A main effect for SSG (p < 0.001) and position (p = 0.024) was detected. The between-SSG post-hoc analysis is reported in Table 1. In SSG-P, a larger ApP is required for forwards vs central-defenders (p = 0.023; ES = 4.35, CI: 1.93/6.01), with no other between-position differences. In SSG-G, no between-position difference occurred. No SSG × position interaction was found (p = 0.065) for ApP for HIRD. A main effect for SSG (p < 0.001) and position (p < 0.001) was detected. The between-exercise post-hoc analy- sis is reported in Table 1. In SSG-P, a higher ApP is required for forwards vs central-defenders (p = 0.024; ES = 2.92, CI: 1.04/4.29), with no other between-position differences. In SSG-G, forwards required higher ApP than central-defenders (p < 0.001; ES = 2.96, CI: 1.07/4.35), wide-midfielders (p = 0.002; ES = 2.45, CI: 0.64/3.78) and wide-defenders (p = 0.029, ES = 3.45, CI: 1.13/4.99). Central-midfielders required a higher ApP than central-defenders (p = 0.002; ES = 1.69, CI: 0.20/2.90) and wide-midfielders (p = 0.019, ES = 1.35, CI: 0.13/2.57). No SSG × position interaction was found (p = 0.803) for ApP for TSD, not even a main effect for exercise (p = 0.415). A main effect for position (p = 0.049) was detected. The between-exercise post-hoc analysis is reported in Table 1. In both SSG-P and SSG-G, no between-position difference occurred. No SSG × position interaction was found (p = 0.167) for ApP for Pmet. A main effect for SSG (p < 0.001) and position (p = 0.002) was detected. The between-SSG post-hoc analysis is reported in Table 1. In SSG-P, a lower ApP is required for central-defenders vs wide-defenders (p = 0.031; ES = -2.69, CI: -4.32/-1.05), wide-midfielders (p < 0.001; ES = -2.64, CI: -4.35/- 0.92), central-midfielders (p = 0.028; ES = -5.10, CI: -7.53/-2.66), forwards (p = 0.024; ES = -1.89, CI: -3.32/0.47). In SSG-G, no between-position difference occurred. Discussion The first novel finding observed in the present study was a detailed calculation of the ApP in SSG-P or SSG-G necessary to replicate the TD, HIRD, TSD or Pmet recorded during the official matches. It is shown here that, irrespective of the SSG type, the higher the speed threshold, the larger the ApP required (i.e., TSD > HIRD > TD  Pmet). Secondly, the inclusion of the goal- keeper increases the ApP for TD, HIRD and Pmet, while no difference was observed in SSG-P vs SSG-G for TSD. Additionally, central defenders required the lowest ApP compared to all other positions, both in SSG-P and SSG-G. Lastly, both central-midfielders and forwards need the highest ApP compared to all other positions, both in SSG-G and SSG-P, to replicate the match demands. During official matches, total high-intensity running distance covered [37], technical skills to maintain greater ball possession [38], the total distance covered with ball possession [39] and tactical behaviours [40] are key factors for success in soccer performance. Within weekly training routines, SSGs are largely used to elicit high-intensity running [1], a high number of technical drills with the ball possession [41] and to improve tactical behaviours [40]. Fig 2. Bland-Altman analysis and linear regression analysis with calibration equation for the GPS vs MCIS bias for each locomotor activity. The linear regression analysis is shown with 95% confidence bands. Panels A-B: total distance; C-D: high-intensity running distance; E-F: total sprint distance; G-H: acceleration/deceleration; I-L: metabolic power. https://doi.org/10.1371/journal.pone.0229194.g002 PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 7 / 15 PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 8 / 15 Interestingly, SSGs were shown to lead to similar enhancement in aerobic fitness than high- intensity running interval training [42]. In SSGs, manipulating the number of players, the pitch size and the goalkeeper presence results in different physiological, technical and tactical outcomes [1, 40]. For example, while increments in ApP was shown to increases TD, HIRD and TSD [4, 5], decreasing ApP leads to more ball touches and Acc/Dec [5, 6]. Determining the ApP that replicates the match external-load demands may help sport physiologists and practitioners to properly plan SSGs for specific performance objectives [5]. Therefore, the cur- rent results may be used to gain knowledge of the SSGs relative to the match demands. Unsur- prisingly, both in SSG-P and in SSG-G, higher ApP leads to greater distance covered whatever the speed threshold [5]. Accordingly, the present findings highlight that the ApP in SSGs to replicate the TSD match demands is very close to the official match ApP ( 340 m2). In line with the present outcomes, it was shown that the larger the pitch size, the greater the distance covered at speed >18 km
h-1 [43]. Other authors found that TD and the distance covered at 19.8–25.2 km
h-1 and >25.2 km
h-1 increased proportionally with the pitch size [44]. A recent study reported that ApP  311 m2 was able to replicate the high-speed match demands during SSG-G [5]. The exposure to high-demanding activities was shown to improve the players’ Fig 3. The relationship between area per player (m2
player) and relative speed distance (m
min-1) or estimated metabolic power (W
kg-1) for each locomotor activity. The linear regression analysis with 95% confidence bands and the correlation between the area per player and the relative distance or metabolic power are also reported. SSG-P, closed circles: small-sided games possession-play without goalkeepers; SSG-G, open circles: small-sided games with goalkeepers. Panel A: total distance; B: high-intensity running distance; C: total sprint distance, D: acceleration/deceleration; E: metabolic power. https://doi.org/10.1371/journal.pone.0229194.g003 Table 1. Area per player (m2
player) to replicate official-match load using SSGs for relative speed distances or estimated metabolic power. Data are presented as mean(SD), effect size (95% confidence intervals). TD HIRD TSD Pmet Position SSG-P SSG-G p ES (CI) SSG-P SSG-G p ES (CI) SSG-P SSG-G p ES (CI) SSG-P SSG-G p ES (CI) Total 115 (35) 187 (53)a <0.001 -1.60 (-2.21/- 0.94) 166 (39) 262 (72)a <0.001 -1.66 (-2.27/- 0.99) 295 (99) 316 (75) 0.415 -0.24 (-0.79/ 0.32) 94(40) 177 (42)a <0.001 -1.99 (-2.67/- 1.31) CD 65 (24)b 165 (26)a <0.001 -4.00 (-5.55/- 1.83) 122 (30)b 205 (57)abc <0.001 -1.82 (-3.00/- 0.37) 257 (76) 278 (51) 0.672 -0.32 (-1.44/ 0.84) 31(11) 151 (23)a <0.001 -6.14 (-8.85/- 3.44) WD 121 (21) 193 (71)a 0.023 -1.38 (-2.70/ 0.31) 163 (30) 246 (36)ab 0.003 -2.50 (-3.92/- 0.43) 297 (26) 274 (67) 0.696 0.45 (-1.01/ 1.79) 106 (31)d 183 (27)a 0.003 -2.39 (-4.01/- 0.77) CM 119(9) 184 (41)a 0.021 -2.12 (-3.41/- 0.42) 174 (28) 311 (69)a <0.001 -2.60 (-3.96/- 0.74) 329 (66) 340 (33) 0.834 -0.21 (-1.43/ 1.05) 107 (13)d 191 (25)a <0.001 -3.81 (-5.88/- 1.73) WM 135 (20) 183 (81)a 0.079 -0.81 (-2.02/ 0.55) 172 (19) 222 (72)abc 0.031 -0.95 (-2.15/ 0.44) 264 (52) 281 (62) 0.758 -0.30 (-1.51/ 0.98) 132 (12)d 180 (61)a 0.047 -0.95 (-2.71/ 0.51) FW 147(9) 214 (51)a 0.018 -1.83 (-3.09/- 0.22) 207 (28) 333 (12)a <0.001 -5.85 (-7.91/- 2.66) 334 (92) 407 (68) 0.176 -0.90 (-2.10/ 0.48) 99(2)d 201 (66)a <0.001 -1.97 (-3.48/- 0.46) TD, total distance; HIRD, high intensity running distance; TSD, sprint distance; Pmet, average metabolic power; SSG-P, small-sided games without goalkeepers; SSG-G, small-side games with goalkeepers; Total, team average; CD, central defenders; WD, wide defenders; CM, central midfielders; WM, wide midfielders; FW, forwards; ES, effect size; CI, confidence interval. a Significantly different (p < 0.05) from SSG-P b Significantly different (p < 0.05) from forwards c Significantly different (p < 0.05) from central midfielders d Significantly different (p < 0.05) from central defenders https://doi.org/10.1371/journal.pone.0229194.t001 PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 9 / 15 fitness level, to prepare the players to the match workload and to result in greater protection against non-contact injuries [45–47]. Therefore, manipulating ApP allows for training loads in SSGs to be managed with respect to the desired external load outcomes, both for performance and prevention purposes. The current findings also highlight that training using SSGs with or without goalkeeper affects the ApP necessary to replicate the match demands. Particularly, with the exception of TSD, the goalkeeper presence increases the ApP for TD, HIRD and Pmet, i.e. SSG-G > SSG-P. Partially in contrast with the present outcomes, it was reported that SSG-G resulted in higher TSD than found in SSG-P [6]. However, the authors investigated a maximum ApP of 135 m2, hence, this does not allow an appropriate comparison. Other researchers reported that TD and the time spent in high-intensity running (>17 km
h-1) was higher with goalkeepers [48]. Although the authors argued that the goalkeeper presence might have motivated the players, several authors found higher high-intensity running without goalkeepers in different 3-, 4-, 5-, and 7-a-side SSGs [7–9]. Moreover, two subsequent reviews [1, 10] consistently remarked that the goalkeeper presence could improve the players’ organization, thus decreasing the SSGs demands. Indeed, during SSG-G, the two teams’ aim is to outscore the opponent team, while maintaining a match-like tactical organization. In contrast, since during SSG-P the aim is to maintain the ball possession as long as possible, the players are free to move across the selected pitch size. This rule-difference seems to account for the largest ApP in SSG-G necessary to rep- licate the TD, HIRD or Pmet recorded during the official matches. Interestingly, the current results come with moderate correlation between Acc/Dec and ApP in SSG-P, while no correla- tion was observed between Acc/Dec and ApP in SSG-G. Previous results suggested that lower pitch size induced increments in Acc/Dec [5, 49]. In line with the present outcomes, other authors retrieved no differences for high-demand (>2 m
s2) Acc/Dec with the increment in pitch size during 3-, 5- and 7-a-side [9] or 3-, 5-, and 10-a-side SSG-G or SSG-P [6]. Compar- ing SSG-P vs SSG-G, higher Acc/Dec were reported during SSG-G than SSG-P using an ApP of ~210 m2 [9], while no difference in Acc/Dec between SSG-G vs SSG-P were found using an ApP from 73-to-135 m2 [6]. Despite the greater stimulation of accelerations in SSG-G vs SSG-P might be accounted for the players’ need to overpass the opponent or defensive lines in order to achieve the rival goal in larger ApP, a controversy still exists. To our knowledge, the calculation of the ApP across positions was used here for the first time. No between-position difference in ApP was found for TSD, neither in SSG-P nor SSG-G. In SSG-P, it was observed that central defenders need lower ApP than forwards for TD, and HIRD, while lower ApP than all other position for Pmet. In SSG-G, no between-position differ- ence in ApP was observed for TD, TSD and Pmet, while forwards and central midfielders need larger ApP than central defenders and wide midfielders for HIRD suggesting that these posi- tions might undergo different stimuli during similar SSGs. Defenders tend to move within a “defined” space over the official match, while central-midfielders and forwards tend to cover a greater area of the pitch in order to gain possession of the ball, marking the opponent or creat- ing space to score [33]. This might be considered for the lower ApP needed to accumulate the match demands in central defenders than forwards/central-midfielders. However, the sprint- ing activities are not influenced by position, since these appear to need large pitch areas avail- able anyhow. The different ApP recorded across position offers the possibility to tailor the training load to enhance the performance adaptations. It was previously suggested that similar high-intensity training load could lead to overload or underload different positions, so affect- ing the competition performance or possibly increasing the risk of injury [29]. Interestingly, high-intensity activities were shown to be underloaded during the training routines compared to the official matches, with a high variability across positions [29]. The present results suggest that some positions need higher or lower ApP to replicate the HIRD or TSD accumulated over PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 10 / 15 the matches. Furthermore, position-specific rule modifications within SSGs or additional exer- cises could be integrated to technical/tactical exercises to individualize high-intensity training activities. Some limitations accompany the present investigation. For replication purposes, the inter- changeability between the GPS and MCIS needs to be carefully checked, especially when recording high-speed or non-linear movements [2]. The present results are based on the trivial differences in the metrics recorded using either the GPS or MCIS and a calibration equation was provided to partially account for these differences. Secondly, due to technological limita- tion during the official matches, no internal load parameter (e.g. heart rate) was assessed. However, it was reported that Pmet maintains a strong and consistent relationship with the measures of internal load during low-to-high intensity locomotor activities [21]. Therefore, Pmet could be a satisfactory way to estimate with accuracy the training and match demands [12, 22] and to classify the locomotion intensity in team sports [21, 28]. Conclusions The current results suggest that soccer players need a specific ApP during SSGs with or with- out goalkeeper to replicate the match demands, especially to perform each locomotor activity (i.e., TSD > HIRD > TD  Pmet). Moreover, SSG-G need higher ApP than SSG-P to replicate the match demands. Lastly, position-difference in ApP were found, so that central defenders need lower and forwards and central midfielders higher ApP. These results allow managing the training loads towards the desired players’ fitness compo- nent to maximize transfer to the game-like and performance goal using SSGs. Indeed, soccer training methodology are evolving to an alternation of the training objectives with the aim to overload the desired fitness component relative to the match demands [5, 29]. When aware of the training/matches differences in locomotor activities, coaches could design SSGs with the intent to replicate, underload or overload the match demands. This imply manipulating SSGs using higher or lower ApP, the presence of the goalkeeper or design specific rules to increase or decrease the position-specific demands. To synthetize, the present outcomes could be used in practice to: i) calculate an ApP that replicate an estimated match demand using Pmet for both SSG-P and SSG-G; ii) replicate the official relative match demands using the specific min- imal ApP to HIRD or TSD be accumulated during the SSG-G/P performed in the training ses- sions; iii) differentiate the ApP when SSG-P or SSG-G are performed according to the aim of the training session (e.g. replicate, overload or underload specific training objectives); iv) add SSGs with position-specific ApP to the training routines when needed or propose specific additional exercises or rules to overload or underload each player. Supporting information S1 Table. Small-sided games with goalkeepers. The small-sided games with goalkeepers are split for the number of players and pitch size (width x length). The total pitch area and area per player have been calculated. The average number of observations per player for each condition are also reported as mean (max-min). (DOCX) S2 Table. Small-sided games without goalkeepers. The small-sided games without goalkeep- ers are split for the number of players and pitch size (width x length). The total pitch area and area per player have been calculated. The average number of observations per player for each condition are also reported as mean (max-min). (DOCX) PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 11 / 15 S1 Data. (XLSX) Author Contributions Conceptualization: Andrea Riboli. Data curation: Andrea Riboli. Formal analysis: Emiliano Ce´. Investigation: Andrea Riboli. Methodology: Andrea Riboli, Giuseppe Coratella. Software: Susanna Rampichini. Supervision: Fabio Esposito. Visualization: Giuseppe Coratella. Writing – original draft: Andrea Riboli, Giuseppe Coratella. Writing – review & editing: Andrea Riboli, Giuseppe Coratella, Fabio Esposito. References 1. Hill-Haas SV, Dawson B, Impellizzeri FM, Coutts AJ. Physiology of small-sided games training in foot- ball: a systematic review. Sports Med. 2011; 41(3):199–220. Epub 2011/03/15. https://doi.org/10.2165/ 11539740-000000000-00000 PMID: 21395363. 2. Buchheit M, Allen A, Poon TK, Modonutti M, Gregson W, Di Salvo V. Integrating different tracking sys- tems in football: multiple camera semi-automatic system, local position measurement and GPS technol- ogies. J Sports Sci. 2014; 32(20):1844–57. Epub 2014/08/06. https://doi.org/10.1080/02640414.2014. 942687 PMID: 25093242. 3. Rampinini E, Coutts AJ, Castagna C, Sassi R, Impellizzeri FM. Variation in top level soccer match per- formance. Int J Sports Med. 2007; 28(12):1018–24. Epub 2007/05/15. https://doi.org/10.1055/s-2007- 965158 PMID: 17497575. 4. Lacome M, Simpson BM, Cholley Y, Buchheit M. Locomotor and Heart Rate Responses of Floaters During Small-Sided Games in Elite Soccer Players: Effect of Pitch Size and Inclusion of Goalkeepers. Int J Sports Physiol Perform. 2018; 13(5):668–71. Epub 2017/09/28. https://doi.org/10.1123/ijspp.2017- 0340 PMID: 28952828. 5. Lacome M, Simpson BM, Cholley Y, Lambert P, Buchheit M. Small-Sided Games in Elite Soccer: Does One Size Fit All? Int J Sports Physiol Perform. 2018; 13(5):568–76. Epub 2017/07/18. https://doi.org/ 10.1123/ijspp.2017-0214 PMID: 28714774. 6. Gaudino P, Alberti G, Iaia FM. Estimated metabolic and mechanical demands during different small- sided games in elite soccer players. Human movement science. 2014; 36:123–33. Epub 2014/06/27. https://doi.org/10.1016/j.humov.2014.05.006 PMID: 24968370. 7. Koklu Y, Sert O, Alemdaroglu U, Arslan Y. Comparison of the physiological responses and time-motion characteristics of young soccer players in small-sided games: the effect of goalkeeper. J Strength Cond Res. 2015; 29(4):964–71. Epub 2013/08/15. https://doi.org/10.1519/JSC.0b013e3182a744a1 PMID: 23942169. 8. Mallo J, Navarro E. Physical load imposed on soccer players during small-sided training games. J Sports Med Phys Fitness. 2008; 48(2):166–71. Epub 2008/04/23. PMID: 18427410. 9. Castellano J, Casamichana D, Dellal A. Influence of game format and number of players on heart rate responses and physical demands in small-sided soccer games. J Strength Cond Res. 2013; 27 (5):1295–303. Epub 2012/07/28. https://doi.org/10.1519/JSC.0b013e318267a5d1 PMID: 22836601. 10. Aguiar M, Botelho G, Lago C, Macas V, Sampaio J. A review on the effects of soccer small-sided games. J Hum Kinet. 2012; 33:103–13. Epub 2013/03/15. https://doi.org/10.2478/v10078-012-0049-x PMID: 23486554; PubMed Central PMCID: PMC3588672. PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 12 / 15 11. Gregson W, Drust B, Atkinson G, Salvo VD. Match-to-match variability of high-speed activities in pre- mier league soccer. Int J Sports Med. 2010; 31(4):237–42. Epub 2010/02/17. https://doi.org/10.1055/s- 0030-1247546 PMID: 20157871. 12. Polglaze T, Hoppe MW. Metabolic Power: A Step in the Right Direction for Team Sports. Int J Sports Physiol Perform. 2019:1–5. Epub 2019/02/09. https://doi.org/10.1123/ijspp.2018-0661 PMID: 30732493. 13. di Prampero PE, Botter A, Osgnach C. The energy cost of sprint running and the role of metabolic power in setting top performances. Eur J Appl Physiol. 2015; 115(3):451–69. Epub 2015/01/01. https:// doi.org/10.1007/s00421-014-3086-4 PMID: 25549786. 14. Minetti AE, Pavei G. Update and extension of the ’equivalent slope’ of speed-changing level locomotion in humans: a computational model for shuttle running. J Exp Biol. 2018; 221(Pt 15). Epub 2018/06/14. https://doi.org/10.1242/jeb.182303 PMID: 29895678. 15. Brown DM, Dwyer DB, Robertson SJ, Gastin PB. Metabolic Power Method: Underestimation of Energy Expenditure in Field-Sport Movements Using a Global Positioning System Tracking System. Int J Sports Physiol Perform. 2016; 11(8):1067–73. Epub 2016/03/22. https://doi.org/10.1123/ijspp.2016- 0021 PMID: 26999381. 16. Buchheit M, Manouvrier C, Cassirame J, Morin JB. Monitoring Locomotor Load in Soccer: Is Metabolic Power, Powerful? Int J Sports Med. 2015; 36(14):1149–55. Epub 2015/09/24. https://doi.org/10.1055/s- 0035-1555927 PMID: 26393813. 17. Stevens TG, De Ruiter CJ, Van Maurik D, Van Lierop CJ, Savelsbergh GJ, Beek PJ. Measured and estimated energy cost of constant and shuttle running in soccer players. Med Sci Sports Exerc. 2015; 47(6):1219–24. Epub 2014/09/12. https://doi.org/10.1249/MSS.0000000000000515 PMID: 25211365. 18. Castagna C, Varley M, Povoas SCA, D’Ottavio S. Evaluation of the Match External Load in Soccer: Methods Comparison. Int J Sports Physiol Perform. 2017; 12(4):490–5. Epub 2016/09/13. https://doi. org/10.1123/ijspp.2016-0160 PMID: 27618733. 19. Dubois R, Paillard T, Lyons M, McGrath D, Maurelli O, Prioux J. Running and Metabolic Demands of Elite Rugby Union Assessed Using Traditional, Metabolic Power, and Heart Rate Monitoring Methods. J Sports Sci Med. 2017; 16(1):84–92. Epub 2017/03/28. PMID: 28344455; PubMed Central PMCID: PMC5358036. 20. Coutts AJ, Kempton T, Sullivan C, Bilsborough J, Cordy J, Rampinini E. Metabolic power and energetic costs of professional Australian Football match-play. J Sci Med Sport. 2015; 18(2):219–24. Epub 2014/ 03/05. https://doi.org/10.1016/j.jsams.2014.02.003 PMID: 24589369. 21. Polglaze T, Hogan C, Dawson B, Buttfield A, Osgnach C, Lester L, et al. Classification of Intensity in Team Sport Activity. Med Sci Sports Exerc. 2018; 50(7):1487–94. Epub 2018/02/13. https://doi.org/10. 1249/MSS.0000000000001575 PMID: 29432324. 22. Manzi V, Impellizzeri F, Castagna C. Aerobic fitness ecological validity in elite soccer players: a meta- bolic power approach. J Strength Cond Res. 2014; 28(4):914–9. Epub 2013/12/19. https://doi.org/10. 1519/JSC.0000000000000239 PMID: 24345968. 23. Malone S, Solan B, Collins K, Doran D. The metabolic power and energetic demands of elite Gaelic football match play. J Sports Med Phys Fitness. 2017; 57(5):543–9. Epub 2016/04/01. https://doi.org/ 10.23736/S0022-4707.16.06233-2 PMID: 27029959. 24. Polglaze T, Dawson B, Buttfield A, Peeling P. Metabolic power and energy expenditure in an interna- tional men’s hockey tournament. J Sports Sci. 2018; 36(2):140–8. Epub 2017/03/12. https://doi.org/10. 1080/02640414.2017.1287933 PMID: 28282747. 25. Kempton T, Sirotic AC, Coutts AJ. An integrated analysis of match-related fatigue in professional rugby league. J Sports Sci. 2015; 33(1):39–47. Epub 2014/05/27. https://doi.org/10.1080/02640414.2014. 921832 PMID: 24857235. 26. di Prampero PE, Osgnach C. Metabolic Power in Team Sports—Part 1: An Update. Int J Sports Med. 2018; 39(8):581–7. Epub 2018/06/15. https://doi.org/10.1055/a-0592-7660 PMID: 29902808. 27. Osgnach C, Poser S, Bernardini R, Rinaldo R, di Prampero PE. Energy cost and metabolic power in elite soccer: a new match analysis approach. Med Sci Sports Exerc. 2010; 42(1):170–8. Epub 2009/12/ 17. https://doi.org/10.1249/MSS.0b013e3181ae5cfd PMID: 20010116. 28. Young D, Malone S, Collins K, Mourot L, Beato M, Coratella G. Metabolic power in hurling with respect to position and halves of match-play. PLoS One. 2019; 14(12):e0225947. Epub 2020/01/01. https://doi. org/10.1371/journal.pone.0225947 PMID: 31891945; PubMed Central PMCID: PMC6938404. 29. Martin-Garcia A, Gomez Diaz A, Bradley PS, Morera F, Casamichana D. Quantification of a Profes- sional Football Team’s External Load Using a Microcycle Structure. J Strength Cond Res. 2018. Epub 2018/09/11. https://doi.org/10.1519/JSC.0000000000002816 PMID: 30199452. PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 13 / 15 30. Castagna C, D’Ottavio S, Cappelli S, Araujo Povoas SC. The Effects of Long Sprint Ability-Oriented Small-Sided Games Using Different Ratios of Players to Pitch Area on Internal and External Load in Soccer Players. Int J Sports Physiol Perform. 2019:1265–72. Epub 2019/03/13. https://doi.org/10.1123/ ijspp.2018-0645 PMID: 30860405. 31. Buchheit M, Al Haddad H, Simpson BM, Palazzi D, Bourdon PC, Di Salvo V, et al. Monitoring accelera- tions with GPS in football: time to slow down? Int J Sports Physiol Perform. 2014; 9(3):442–5. https:// doi.org/10.1123/ijspp.2013-0187 PMID: 23916989. 32. Rampinini E, Alberti G, Fiorenza M, Riggio M, Sassi R, Borges TO, et al. Accuracy of GPS devices for measuring high-intensity running in field-based team sports. Int J Sports Med. 2015; 36(1):49–53. https://doi.org/10.1055/s-0034-1385866 PMID: 25254901. 33. Di Salvo V, Baron R, Tschan H, Calderon Montero FJ, Bachl N, Pigozzi F. Performance characteristics according to playing position in elite soccer. Int J Sports Med. 2007; 28(3):222–7. Epub 2006/10/07. https://doi.org/10.1055/s-2006-924294 PMID: 17024626. 34. Harley JA, Lovell RJ, Barnes CA, Portas MD, Weston M. The interchangeability of global positioning system and semiautomated video-based performance data during elite soccer match play. J Strength Cond Res. 2011; 25(8):2334–6. Epub 2011/07/13. https://doi.org/10.1519/JSC.0b013e3181f0a88f PMID: 21747299. 35. Randers MB, Mujika I, Hewitt A, Santisteban J, Bischoff R, Solano R, et al. Application of four different football match analysis systems: a comparative study. J Sports Sci. 2010; 28(2):171–82. Epub 2010/ 04/15. https://doi.org/10.1080/02640410903428525 PMID: 20391091. 36. Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc. 2009; 41(1):3–13. Epub 2008/12/19. https://doi.org/10. 1249/MSS.0b013e31818cb278 PMID: 19092709. 37. Krustrup P, Mohr M, Ellingsgaard H, Bangsbo J. Physical demands during an elite female soccer game: importance of training status. Med Sci Sports Exerc. 2005; 37(7):1242–8. Epub 2005/07/15. https://doi. org/10.1249/01.mss.0000170062.73981.94 PMID: 16015145. 38. Castellano J, Casamichana D, Lago C. The Use of Match Statistics that Discriminate Between Suc- cessful and Unsuccessful Soccer Teams. J Hum Kinet. 2012; 31:139–47. Epub 2013/03/15. https://doi. org/10.2478/v10078-012-0015-7 PMID: 23487020; PubMed Central PMCID: PMC3588662. 39. Hoppe MW, Slomka M, Baumgart C, Weber H, Freiwald J. Match Running Performance and Success Across a Season in German Bundesliga Soccer Teams. Int J Sports Med. 2015; 36(7):563–6. Epub 2015/03/12. https://doi.org/10.1055/s-0034-1398578 PMID: 25760152. 40. Moniz F, Scaglia A, Sarmento H, Garcia-Calvo T, Teoldo I. Effect of an Inside Floater on Soccer Players Tactical Behaviour in Small Sided and Conditioned Games. J Hum Kinet. 2020; 71:167–77. Epub 2020/ 03/10. https://doi.org/10.2478/hukin-2019-0080 PMID: 32148581; PubMed Central PMCID: PMC7052718. 41. da Silva CD, Impellizzeri FM, Natali AJ, de Lima JR, Bara-Filho MG, Silami-Garcia E, et al. Exercise intensity and technical demands of small-sided games in young Brazilian soccer players: effect of num- ber of players, maturation, and reliability. J Strength Cond Res. 2011; 25(10):2746–51. Epub 2011/09/ 14. https://doi.org/10.1519/JSC.0b013e31820da061 PMID: 21912285. 42. Kunz P, Engel FA, Holmberg HC, Sperlich B. A Meta-Comparison of the Effects of High-Intensity Inter- val Training to Those of Small-Sided Games and Other Training Protocols on Parameters Related to the Physiology and Performance of Youth Soccer Players. Sports Med Open. 2019; 5(1):7. Epub 2019/ 02/23. https://doi.org/10.1186/s40798-019-0180-5 PMID: 30790134; PubMed Central PMCID: PMC6384288. 43. Hill-Haas SV, Dawson BT, Coutts AJ, Rowsell GJ. Physiological responses and time-motion character- istics of various small-sided soccer games in youth players. J Sports Sci. 2009; 27(1):1–8. Epub 2008/ 11/08. https://doi.org/10.1080/02640410902761199 PMID: 18989820. 44. Gaudino P, Iaia FM, Alberti G, Hawkins RD, Strudwick AJ, Gregson W. Systematic bias between run- ning speed and metabolic power data in elite soccer players: influence of drill type. Int J Sports Med. 2014; 35(6):489–93. Epub 2013/10/30. https://doi.org/10.1055/s-0033-1355418 PMID: 24165959. 45. Bowen L, Gross AS, Gimpel M, Li FX. Accumulated workloads and the acute:chronic workload ratio relate to injury risk in elite youth football players. Br J Sports Med. 2017; 51(5):452–9. Epub 2016/07/28. https://doi.org/10.1136/bjsports-2015-095820 PMID: 27450360; PubMed Central PMCID: PMC5460663. 46. Malone S, Roe M, Doran DA, Gabbett TJ, Collins K. High chronic training loads and exposure to bouts of maximal velocity running reduce injury risk in elite Gaelic football. J Sci Med Sport. 2017; 20(3):250– 4. Epub 2016/08/25. https://doi.org/10.1016/j.jsams.2016.08.005 PMID: 27554923. 47. Bowen L, Gross AS, Gimpel M, Bruce-Low S, Li FX. Spikes in acute:chronic workload ratio (ACWR) associated with a 5–7 times greater injury rate in English Premier League football players: a PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 14 / 15 comprehensive 3-year study. Br J Sports Med. 2019. Epub 2019/02/23. https://doi.org/10.1136/ bjsports-2018-099422 PMID: 30792258. 48. Dellal A, Chamari K, Pintus A, Girard O, Cotte T, Keller D. Heart rate responses during small-sided games and short intermittent running training in elite soccer players: a comparative study. J Strength Cond Res. 2008; 22(5):1449–57. Epub 2008/08/21. https://doi.org/10.1519/JSC.0b013e31817398c6 PMID: 18714244. 49. Castellano J, Casamichana D. Differences in the number of accelerations between small-sided games and friendly matches in soccer. J Sports Sci Med. 2013; 12(1):209–10. Epub 2013/10/19. PMID: 24137079; PubMed Central PMCID: PMC3761762. PLOS ONE External load using small-sided games in elite soccer PLOS ONE | https://doi.org/10.1371/journal.pone.0229194 September 23, 2020 15 / 15

Area per player in small-sided games to replicate the external load and estimated physiological match demands in elite soccer players.

09-23-2020

Riboli, Andrea,Coratella, Giuseppe,Rampichini, Susanna,Cé, Emiliano,Esposito, Fabio

eng

PMC1277951

VOLUME 1: NO. 4 OCTOBER 2004 Use of a Community Trail Among New and Habitual Exercisers: A Preliminary Assessment ORIGINAL RESEARCH Suggested citation for this article: Gordon PM, Zizzi SJ, Pauline J. Use of a community trail among new and habit- ual exercisers: a preliminary assessment. Prev Chronic Dis [serial online] 2004 Oct [date cited]. Available from: URL: http://www.cdc.gov/pcd/issues/2004/oct/04_0058.htm. PEER REVIEWED Abstract Introduction We evaluated physical activity patterns and trail use among new and habitually active exercisers using onsite trail interviews. Methods Using a cross-sectional study design, 414 adults who accessed two new trails that bisect a rural community of 26,809 residents were interviewed during the first summer of the trails’ official operation (2001). The trails comprise 12 miles of level and paved surface and run parallel to adja- cent water sheds, businesses, and neighborhoods. Recent trail activity patterns were obtained, including the follow- ing: frequency of use, mode of activity, duration, distance traveled on trail, access points, time of day used, use of exercise companions, and distance traveled to get to trail. Perceived enablers and barriers related to trail use were also obtained. Data were compared between newly adopted exercisers (new exercisers) and individuals active prior to development of the trails (habitually active exercisers). Results Twenty-three percent of the trail users were new exercis- ers. New exercisers were more dependent on the trails as a primary outlet for physical activity than were habitually active exercisers (P < .001). New exercisers traveled short- er distances to access the trails and rated convenience as a primary reason for using them. Both safety and terrain issues emerged as enablers for trail use, and unsafe condi- tions emerged as a concern among new exercisers. Conclusion A community trail may be an important vehicle for pro- moting physically active lifestyles. However, new exercis- ers must overcome issues of proximal and safe access from residential areas in addition to other safety concerns to achieve regular physical activity. Introduction Although the health benefits of physical activity are now well established (1), 55% of Americans do not meet the minimal physical activity recommendations for health (2). Environmental and policy approaches to promoting physi- cal activity have been recommended to change the physi- cal and social environments that individuals inhabit. Public health officials theorize that when suitable facilities are available to community residents, physical activity lev- els increase (3,4). Healthy People 2010 objectives recom- mend creating and enhancing access to places and facili- ties where people can be physically active (5). Furthermore, the Task Force on Community Preventive Services has recently issued a strong recommendation for policy and environmental approaches that create or enhance access to places for physical activity, along with information outreach activities, as an intervention to increase community physical activity levels (6). The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. www.cdc.gov/pcd/issues/2004/oct/04_0058.htm • Centers for Disease Control and Prevention 1 Paul M. Gordon, PhD, MPH, Samuel J. Zizzi, EdD, Jeff Pauline, EdD VOLUME 1: NO. 4 OCTOBER 2004 One example of an environmental and policy approach to increase physical activity in the community is the devel- opment of a walking/bicycling trail. A community walk- ing/bicycling trail can be a relatively low-cost intervention that may facilitate physical activity by reducing barriers related to cost, convenience, and accessibility (7,8). Moreover, because the trail is a permanent fixture within the community, it may facilitate the maintenance of a physically active lifestyle. Brownson et al examined the characteristics and possible impact of walking-trail devel- opment and suggested that walking trails may be particu- larly effective at reaching populations at high risk for inac- tive behaviors (9). Although recent studies have included trails as examples of physical environmental attributes of an active community (10), community walking/biking trails in particular have not been well studied. One recent investigation in Australia found that a newly constructed rail trail accompanied by a local promotional campaign increased cycling (11). More studies are needed to assess the importance of a community walking/biking trail on influencing physical activity levels. It is not known how important a trail is among individ- uals who have newly adopted exercise habits. Nor is it known if the types of physical activity and patterns of trail use differ between new exercisers and habitually active exercisers. Although health officials have theorized that community recreation trails can provide convenient and accessible opportunities for engaging in regular physical activity, little data are available to describe the trails’ importance, particularly among those who are transition- ing toward an active lifestyle. In addition, the barriers and enablers to trail use, which may differ between new and habitually active exercisers, are important to understand- ing how to facilitate this transition. This information will provide health officials with insights that may be useful for promoting trail use and active lifestyles among resi- dents within their communities. Methods Design A cross-sectional study used data from an onsite inter- view survey of physical activity patterns, barriers, and enablers to trail use among adults using two new rail trails within the city of Morgantown, WVa. The Caperton and Decker’s Creek trails comprise 12 miles of paved trails that bisect the town and run adjacent to the Monongahela River and Decker’s Creek, respective- ly. These trails also extend outside the city limits with an additional 14 miles of unpaved trails. Construction on these trails was completed in spring 2001. Rail trails are multiuse pathways constructed on abandoned railway beds and can be used for both recreational and trans- portation-related physical activity (12). In addition to stretching along waterways, these level trails intersect neighborhoods and business establishments within city limits. Sample An interceptor-based survey approach was used instead of a population-based survey approach because of its better ability to identify and probe for trail users’ perceptions and attitudes. Trained interviewers admin- istered the Recreation Trail Evaluation Survey (RTES) to a sample of 414 adult trail users who lived in Monongalia County, West Virginia. Graduate students were trained to interview participants using skills train- ing developed from other physical-activity interview- driven questionnaires (13). During training, interview- ers reviewed and discussed the RTES questionnaire, rehearsed several practice interviews, and received grades on proficiency. Important features of the training sessions included clear explanations of the frame of ref- erence for each question, how to control the pace and structure of the interview, and how and when to use prompts and other questions. To assure consistency, the same interviewers participated in the RTES pilot study prior to the study’s initiation. Interviews were conducted two times per day using a randomized schedule that included predetermined blocks of time (7-10:00 AM, 11- 2:00 PM, 3-6:00 PM, and 6-9:00 PM) and five different trail access points to ensure that samples fairly represented time of day, location on trail, and time of week (i.e., week- end vs weekday). The influence of weather was recognized as a possible limitation to data collection, but poor weath- er rarely occurred during data collection. The trail inter- view took approximately five to 10 minutes per participant to complete. Trail interviews took place for four weeks from June–July 2001. A true survey response rate (num- ber of participants divided by total number of individuals who used the trails during the interview sessions) was not attained because of the way data were collected: some indi- viduals may have passed by while interviews were being 2 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2004/oct/04_0058.htm The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. conducted. Nevertheless, 98% of all individuals who were approached were willing to participate. Willingness to par- ticipate in the study was high perhaps due to the novelty of newly developed trails in a smaller community and because the investigation took place shortly after their opening. Moreover, we did not infringe upon the partici- pants’ right to exercise. Rather, participants were inter- viewed as they entered or exited the various trailheads, and interviewers sometimes walked along with the exer- cisers during interviews. To prevent duplication, partici- pants were asked at the start of the survey if they had pre- viously been interviewed. Measures The RTES measured recent trail physical activity pat- terns and included information on up to two types of phys- ical activity performed on the trails (Appendix). The sur- vey’s exercise components queried participants on fre- quency and duration of activity, distance traveled on trail, and points of access for each type of activity. The question format used for the exercise components was similar to the format used by the Behavioral Risk Factor Surveillance System (BRFSS) (14). In addition, information was obtained on time of day, exercise companions, and method and distance traveled to get to the trail. Additionally, all respondents were asked if they participated in any of 10 non-trail physical activities in the previous month. These activities included walking, aerobic dance, bicycling, golf- ing, strength training, gardening, jogging/running, swim- ming/water exercises, organized team sports, and house- work. Non-trail recreational patterns were assessed based on each activity’s type, duration, and frequency. In addition to self-reported distance traveled on the trail, actual distances were also calculated by premeasuring dis- tances between access points and landmarks using an odometer wheel. Subjects were asked to identify points traveled on the trail (entry, turnaround, and exit) and the actual distance was calculated. Because there were no sig- nificant differences between self-reported and actual dis- tance traveled on the trail, actual distance traveled is reported in the present study. Each perceived enabler and barrier to trail use was measured using a five-point Likert scale ranging from 1 = not at all important to 5 = most important. Enablers were defined as reasons for using the trail and included safety, scenery/environment, terrain (e.g., flat, paved), conven- ience, and atmosphere. Barriers were defined as items that may prevent participants from using the trail more and included safety issues, parking, accessibility, facilities, maintenance, and congestion. Using an open-ended question format, interviewers also asked participants to identify their primary enabler or barrier. Social and demographic information was collected on age, sex, marital status, race, employment status, educational attainment, and individual income level. An initial pilot survey was developed from several existing documents that were obtained from similar stud- ies (9,15) and tested over a three-week period. A sample was obtained at five key access points along the trail within the city limits to yield 161 users that included 90 female and 71 male adult respondents ranging in age from 18 to 82. Three expert reviewers analyzed results from the pilot survey to identify possible issues of clarity. Minor revisions to the trail user survey were made to address problems. While reliability measures are known for questions obtained from the BRFSS, no specific psy- chometric measures were obtained for the completed RTES survey. The finalized survey consisted of 33 closed and open-ended items. Of primary interest to this investigation was to deter- mine if the addition of the trail into the community caused any trail users to adopt new physical activity pro- grams. Consequently, participants were asked, “Did you exercise regularly [more than three times per week for 20 minutes] before using this trail?” Three times per week was used as the frequency threshold for regular exercisers because of the associated health benefits that may exist among vigorous exercisers (1). This construct was designed to determine whether participants were cur- rently engaged in a pattern of regular physical activity rather than to identify the prevalence of individuals meet- ing physical activity recommendations for health. Ninety- three (22.5%) trail users responded “no” to this question and were classified as new exercisers. The remaining 321 (77.5%) participants who answered “yes” were classified as habitually active exercisers. To determine differences that might exist between new exercisers and habitually active exercisers, comparisons of physical activity pat- terns and preferences for trail use were analyzed. Among all survey respondents, 94% were attaining 150 minutes of leisure-time physical activity per week, the amount recommended by the surgeon general (1). VOLUME 1: NO. 4 OCTOBER 2004 www.cdc.gov/pcd/issues/2004/oct/04_0058.htm • Centers for Disease Control and Prevention 3 The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. VOLUME 1: NO. 4 OCTOBER 2004 This investigation was approved by the Human Subject’s Institutional Review Board at West Virginia University. Analysis Survey data were analyzed to determine the uses and usefulness of newly developed trails for physical activity within a community. The primary research question relat- ed to how many of the trail users in the sample were new exercisers and how many were habitually active prior to trail completion. After grouping participants, a series of analyses were conducted to explore potential demograph- ic, behavioral, and motivational differences related to trail use between groups. All data were coded and entered into an SPSS 10.0 (SPSS Inc, Chicago, Ill) statistical software database for analyses. Chi-square analyses were conduct- ed to determine differences in proportions. In addition, an independent t-test was used to test for differences in phys- ical activity variables (e.g., frequency, duration, distance) between groups. Results The sample (n = 414) was 94.4% white (n = 391), 44.9% male (n = 186), and 55.1% female (n = 228). Table 1 sum- marizes the primary demographic characteristics of the community trail users in this survey. These characteristics are representative of the community population. According to the 2000 U.S. Census, Monongalia County, West Virginia, is 93% white and 50% female (16). The age distribution for the county is as follows: 18–25 years = 22.0%; 26–35 years = 19.2%; 36–45 years = 17.8%; 46–64 years = 26.8%; and older than 65 years = 14.2%. The age distribution of the survey sample is comparable to the cen- sus distribution, except the sample had fewer respondents older than 65 (6.5%). Impact of trail on physical activity rates Ninety-three (22.5%) trail users were classified as new exercisers, and 321 (77.5%) participants were classified as habitually active exercisers. A two-way chi-square analy- sis was performed to determine differences between groups across sex, age, and employment status. These analyses revealed no significant differences, suggesting that new exercisers and habitually active exercisers share similar demographic profiles. Analyses were also used to compare the frequency of additional physical activity reported between new and previously active exercisers. All respondents were asked if they had participated in any of 10 various physical activities (e.g., aerobic dance, swim- ming, team sports, housework, gardening) in the previous month. The total number of activities for each participant was computed, and an independent t-test was conducted to test the hypothesis. Habitually active exercisers reported significantly more frequency of additional physical activi- ty (mean = 1.83 occurences; SD = 1.2) than new exercisers (mean = 1.2 occurences; SD = 1.1), t (412) = 4.51, P < .001. Additionally, more than twice as many new exercisers (31%) than habitually active exercisers (15%) reported that the trail was their only form of physical activity. Nearly all (98%) of the new exercisers reported that their exercise amounts had increased when asked, “Since using the trail, has the amount of exercise that you do increased, decreased, or stayed the same?” Only 52% of the habitually active exercisers reported an increase. Conversely, 48% of habitually active exercisers and only 2% of new exercisers reported that their exercise amounts stayed the same. These data suggest that the physical activity patterns of nearly one half of habitually active exercisers were not impacted by the addition of the trail. Moreover, the perceived improvement in physical activity levels between new and habitually active exercisers was significantly different (X2[4] = 120.54, P < .001), with new exercisers reporting much greater increases in physical activity than habitually active exercisers with the addi- tion of the trail (Figure). Types and patterns of physical activity on the trail New exercisers traveled shorter distances to access the trails compared with habitually active exercisers (2.9 ± 3.4 miles vs 3.9 ± 6.0 miles; P = .03). The majority of respon- dents traveled to the trails by vehicle (81%). However, new exercisers were more likely to walk (18%) to the trails than habitually active exercisers (10.1%) (P = .04). Overall, these two groups differed in their patterns of physical activity on the trails. New exercisers were also more like- ly to walk (58% to 42%), less likely to run or jog (11% to 17%), and less likely to in-line skate (4% to 11%) than habitually active exercisers (X2[3] = 9.15, P = .02). Comparisons of average time and distance on the trails provide further support to the hypothesis that habitually active exercisers are engaging in different modes or high- 4 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2004/oct/04_0058.htm The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. er exercise intensities com- pared with new exercisers. Habitually active exercisers traveled greater distances (P = .03) on the trail (6.64 ± 5.7 miles) than new exercis- ers (5.41 ± 3.7 miles) but did not spend a longer amount of time exercising (57.2 ± 30.1 min) than new exercis- ers (59.6 ± 30.2 min). Additionally, the frequency of weekly trail use averaged 3.4 (± 2.1) days per week in the entire sample. No sig- nificant difference in week- ly trail use was observed between new exercisers (3.63 ± 1.5 days) and habit- ually active individuals (3.3 ± 2.3 days). Enablers and barriers to trail use Table 2 presents the mean Likert-scale ratings of per- ceived enablers and barriers to trail use among new and habitually active exercisers. Participants were asked to rank each enabler and barrier, and rankings for enablers and barriers based on their aggregate level of importance were assigned by the investigators. New exercisers ranked enablers in the following order of importance: 1) conven- ience, 2) terrain, 3) safety, 4) scenery, and 5) atmosphere. In contrast, habitual exercisers ranked enablers in this order: 1) terrain, 2) convenience, 3) scenery, 4) safety, and 5) atmosphere. Mean ratings of enablers differed between groups. New exercisers rated safety (P = .03), terrain (P = .04), and convenience (P = .001) as significantly more important than habitually active exercisers. New exercis- ers rated unsafe conditions as a significantly higher barri- er than habitually active exercisers (P = .04), although mean scores (3.1 ± 1.6) were in the middle of the five-point scale. All other perceived enablers and barriers were sim- ilar for both groups. Discussion In this preliminary investigation, improvements in physical activity behavior occurred as a result of adding a community walking/biking trail, particularly among previously inactive partici- pants. Approximately 25% of the trail users became regu- lar exercisers (three or more times a week) as a result of the development of the trail. Moreover, new exercisers were much more dependent on the trail as a principal place for engaging in physi- cal activity than those who exercised regularly prior to trail development. Thirty- one percent of new exercisers used the trail as the only venue for physical activity. This suggests that recre- ational trails may be a pow- erful vehicle for physical activity promotion, particu- larly among previously inactive individuals. Brownson et al suggested that within rural communities, sedentary individuals may be the most likely to benefit from walking trails (9). Although Morgantown, WVa, is a city of 26,809 residents, it is located in a rural region where there is lit- tle opportunity to safely engage in walking for physical activity. With narrow streets that lack traffic-calming strategies, bike lanes, and sidewalks, the community is not conducive for walking or bicycling. The introduction of a safe and convenient area to walk may be an excellent physical activity promotion tool. In a recent review of the effectiveness of interventions to increase physical activity, the Guide to Community Preventive Services proposed that creating access to places for physical activity, combined with informational outreach, is an effective means for increasing physical activity levels (6). The current investi- gation supports this recommendation. New exercisers also traveled shorter distances to access the trail, implying that residential proximity to the trail may play an important role in whether individuals will use the trail. In further support of this, new exercisers were more likely to rate convenience as a primary reason for using the trail. Residential proximity to trails and their usage has previously been documented (10,11,17). Increases in self-reported and geospatial distance were associated with a decreased likelihood of using a bikeway VOLUME 1: NO. 4 OCTOBER 2004 www.cdc.gov/pcd/issues/2004/oct/04_0058.htm • Centers for Disease Control and Prevention 5 The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. Figure. Percentage of increase in physical activity reported by new and habitually active exercisers when asked, “Since using the trail, approximately how much has your exercise level increased?” VOLUME 1: NO. 4 OCTOBER 2004 (17). Moreover, King et al found that walking levels among older women were higher among those living in areas where parks or trails existed (10). Their study, however, did not specifically measure the impact of a walking trail. Nevertheless, they concluded that the ability to engage in walking trips from home and the perception of having favorable neighborhood surroundings for walking are associated with increased physical activity levels (10). Merom et al found that trail usage was increased among cyclists, particularly among individuals in close proximity to a trail (11). In our study, data suggest that convenient, safe, and proximal community walking trails provide an incentive for community residents to engage in regular physical activity. This offers further support to the impor- tance of closely linking recreational trails with residential communities to provide safe and convenient access. The type and pattern of physical activity on the trail also differed between new exercisers and habitually active indi- viduals. It appears that newer exercisers begin with a more conservative physical activity (walking), whereas habitually active trail users more commonly select moder- ate- to high-intensity activities (e.g., running, in-line skat- ing). Choosing more conservative physical activities like walking may also be related to a concern for personal safe- ty and injury prevention. Both safety and terrain issues emerged as significant enablers for trail use among new exercisers. Consequently, new exercisers may be more con- cerned with injury prevention during physical activity and may use the trail because they feel it is safe and appropri- ate for exercise. Similarly, new exercisers were more like- ly to rate unsafe conditions as a barrier when asked, “What issues may prevent you from using the trail more frequently?” These data suggest that new exercisers are more sensitive to safety concerns than habitually active individuals. How individuals perceive their environment may be more important in persuading a physically active lifestyle (18,19). Carnegie et al (20) identified a link between per- ceptions of the environment and stage of change for phys- ical activity (21). In their study, contemplators (21) (inac- tive but intend to become more physically active) had more negative perceptions of the environment for physical activ- ity. Similarly, it is reasonable to believe that the new exer- cisers in the present study were still embracing more neg- ative perceptions of the environment than those who are habitually active. Developing strategies to address safety concerns along with other negative perceptions may be necessary if individuals are to progress to being habitual- ly active. As such, trail advocates should prioritize and address safety concerns among new exercisers to promote the appeal of a trail for the long-term pursuit of enhancing physical activity within a community. Although this preliminary investigation found that new exercisers appear to be more dependent on a recreational trail for achieving a pattern of regular physical activity compared with habitually active exercisers, this study has the following limitations: • This investigation used a cross-sectional design that pro- hibited us from obtaining a baseline assessment of phys- ical activity levels prior to the development of the trail. • We relied on trail interviews, which may be subject to a potential response bias. Although we were unable to determine a true response rate, nearly all individuals (98%) approached on the trail were willing to partici- pate. • We used self-reported physical activity data, so there is no direct evidence that trail activities reported were actually performed. Nevertheless, every effort was made to conduct the interviews in a standardized format. • The construct used to classify new vs habitual exercisers was not validated. We relied on individual recall. Consequently, it is possible that some trail users were misclassified. However, nearly all of the respondents (94%) were meeting physical activity recommendations (engaged in 150 minutes of leisure-time physical activi- ty per week). Furthermore, to prevent a response bias, we asked participants about the type, frequency, and duration of their physical activity before asking them whether they were exercising regularly (more than three times per week for 20 minutes). • Finally, we used an interceptor-based survey approach to probe respondents’ views of the trail and identify their perceptions of the environment. Thus, while the infor- mation presented helps to identify perceptions of the environment for the trail user, it does not necessarily reflect the impact of the trail on the overall community. However, community-wide phone-survey data (unpub- lished data), which were obtained during the same time, indicate that 20% of regular exercisers use the trail as their primary exercise venue and only neighborhood 6 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2004/oct/04_0058.htm The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. streets provided a more common exercise location among community residents. Perhaps a lack of connectivity to the trail prevented many community members from using the trail as a primary site for regular physical activity. Given that there are very few walkable neigh- borhoods (e.g, no sidewalks, bike lanes, traffic-calming strategies) within the community, trail use would likely further increase if pedestrian connectivity from the trail to residential areas improved. Regardless, these data provide a preliminary assess- ment of the importance of physical environmental changes, such as the development of a walking and biking trail, for promoting physically active lifestyles. Although a community trail can provide opportunities for all residents to engage in regular physical activity, both proximal and safe access from residential areas and safety on the trail may be important issues to encourage trail use among new exercisers. Acknowledgments This study was funded by the West Virginia University Prevention Research Center and the Centers for Disease Control and Prevention. The authors would like to acknowledge David Goodrich, Emily Spangler, and Amy Sindler for their contributions. Author Information Corresponding author: Paul M. Gordon, PhD, MPH, West Virginia University, School of Medicine, Department of Human Performance and Exercise Science, P.O. Box 9227, Morgantown, WV 26506. Telephone: 304-293-0442. E-mail: pgordon@hsc.wvu.edu. Author affiliations: Samuel J. Zizzi, EdD, West Virginia University, School of Physical Education, Morgantown, WVa; Jeff Pauline, EdD, Ball State University, School of Physical Education, Muncie, Ind. References 1. U.S. Department of Health and Human Services. Physical activity and health: a report of the Surgeon General executive summary. Atlanta (GA): Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, The President's Council on Physical Fitness and Sports; 1996. 2. Macera CA, Jones DA, Yore MM, Ham SA, Kohl HW, Kimsey CD Jr, et al. Prevalence of physical activity, including lifestyle activities among adults - United States 2000-2001. MMWR Morb Mortal Wkly Rep 2003;52(32):764-769. 3. Schmid T, Pratt M, Howze E. Policy as intervention: environmental and policy approaches to the preven- tion of cardiovascular disease. Am J Pub Health 1995;85(9):1207-11. 4. King AC. Community and public health approaches to the promotion of physical activity. Med Sci Sports Exerc 1994;26(11):1405-12. 5. U.S. Department of Health and Human Services. Healthy people 2010: understanding and improving health. Washington (DC): U.S. Government Printing Office; 2000. 6. Kahn EB, Ramsey LT, Brownson RC, Heath GW, Howze EH, Powell KE, et al. The effectiveness of inter- ventions to increase physical activity. A systematic review. Am J Prev Med 2002;22(4 Suppl):73-107. 7. Wang G, Macera C, Scudder-Soucie B, Schmid T, Pratt M, Buchner D. Cost effectiveness of a bicycle/pedestrian trail development in health promo- tion. Prev Med 2004;38(2):237-42. 8. Wang G, Macera CA, Scudder-Soucie B, Schmid T, Pratt M, Buchner D, et al. Cost analysis of the built environment: the case of bike and pedestrian trials in Lincoln, Neb. Am J Public Health 2004;94(4):549-53. 9. Brownson RC, Housemann RA, Brown D, Jackson- Thompson J, King A, Malone B, et al. Promoting phys- ical activity in rural communities: walking trail access, use, and effects. Am J Prev Med 2000;18(2):235-41. 10. King WC, Brach JS, Belle S, Killingsworth R, Fenton M, Kriska AM. The relationship between convenience of destinations and walking levels in older women. Am J Health Promot 2003;18(1):74-82. 11. Merom D, Bauman A, Vita P, Close G. An environ- mental intervention to promote walking and cycling - the impact of a newly constructed Rail Trail in Western Sidney. Prev Med 2003;36(2):235-42. 12. Flink CA, Olka K, Searns RM. Trails for the twenty- first century: planning, design, and management man- ual for multi-use trails. 2nd ed. Washington (DC): Island Press; 2001. VOLUME 1: NO. 4 OCTOBER 2004 www.cdc.gov/pcd/issues/2004/oct/04_0058.htm • Centers for Disease Control and Prevention 7 The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. VOLUME 1: NO. 4 OCTOBER 2004 13. Kriska AM, Casperson C. A collection of physical activ- ity questionnaires for health-related research. Med Sci Sports 1997;29(6):s94-s99. 14. Hahn RA, Heath GW, Chang MH. Cardiovascular dis- ease risk factors and preventive practices among adults - United States, 1994: a behavioral risk factor atlas. Behavioral Risk Factor Surveillance System State Coordinators. MMWR CDC Surveill Summ 1998;47(5):35-69. 15. Neff LJ, Ainsworth BE, Wheeler FC, Krumwiede SE, Trepal AJ. Assessment of Trail Use in a Community Park. Fam Community Health 2000;23(3):76-84. 16. Compliance with physical activity recommendations by walking for exercise--Michigan, 1996 and 1998. MMWR Morb Mortal Wkly Rep 2000;49(25):560-5. 17. Troped PJ, Saunders RP, Pate RR, Reininger B, Ureda JR, Thompson SJ. Associations between self- reported and objective physical environmental factors and use of a community rail-trail. Prev Med 2001;32(2):191-200. 18. Kirtland KA, Porter DE, Addy CL, Neet MJ, Williams JE, Sharpe PA, et al. Environmental measures of physical activity supports: perception versus reality. Am J Prev Med 2003;24(4):323-31. 19. Ball K, Bauman A, Leslie E, Owen N. Perceived envi- ronmental aesthetics and convenience and company are associated with walking for exercise among Australian adults. Prev Med 2001;33(5):434-40. 20. Carnegie MA, Bauman A, Marshall AL, Mohsin M, Westley-Wise V, Booth ML. Pereptions of the physical environment, stage of change for physical activity, and walking among Australian adults. Res Q Exerc Sport 2002;73(2):146-55. 21. Marcus BH, Simkin LR. The stages of exercise behav- ior. J Sports Med Phys Fitness 1993;33(1):83-8. Tables Table 1. Socio-demographic Characteristics of Trail Users (n = 414), Morgantown, WVa, 2001 8 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2004/oct/04_0058.htm The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. Characteristic No. (%) Sex Female 228 (55.1) Male 186 (44.9) Age (years) 18-25 96 (23.2) 26-35 92 (22.2) 36-45 89 (21.5) 46-65 110 (26.6) 65+ 27 (6.5) Race/ethnicity White 391 (94.4) Black 7 (1.7) Other 13 (3.1) Declined 3 (0.7) Annual household income, $ <$10,000 111 (26.8) $10,000 to 30,000 105 (25.4) $31,000 to 60,000 114 (27.5) >$60,000 54 (13.0) Declined 30 (7.2) Education High school/GED 145 (35.0) Technical school 16 (3.9) College graduate 160 (38.6) Graduate school 61 (14.7) Professional degree 30 (7.2) Declined 2 (0.5) Employment Status Homemaker 28 (6.8) Self-employed 30 (7.2) Student 100 (24.2) Employed for wages 213 (51.4) Retired 33 (8.0) Unemployed 7 (1.7) Declined/Other 3 (0.7) Table 2. Perceived Enabling Factors and Personal Barriers to Trail Use for New Exercisers and Habitually Active Exercisers (n = 414), Morgantown, WVa, 2001 VOLUME 1: NO. 4 OCTOBER 2004 www.cdc.gov/pcd/issues/2004/oct/04_0058.htm • Centers for Disease Control and Prevention 9 The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. Difference between Habitually active New exerciser new and habitually (n = 321) (n = 93) active exercisers Mean Mean Characteristic (±SD)a Rankb (±SD) Rank Pc Enablers Safety 3.9 (1.3) 4 4.2 (1.0) 3 .03 Scenery/environment 4.0 (1.0) 3 4.1 (1.0) 4 .16 Terrain (flat, paved) 4.3 (0.9) 1 4.6 (0.7) 2 .04 Convenience 4.3 (0.9) 2 4.7 (0.5) 1 <.001 Atmosphere 3.8 (1.1) 5 4.1 (1.2) 5 .19 Barriers Unsafe 2.7 (1.7) 3 3.1 (1.6) 2 .04 Parking 2.1 (1.3) 6 2.1 (1.4) 6 .78 Accessibility 2.2 (1.3) 5 2.4 (1.6) 5 .11 Facilities 3.1 (1.4) 1 3.4 (1.4) 1 .08 Maintenance 3.0 (1.5) 2 2.8 (1.6) 3 .28 Congestion 2.6 (1.4) 4 2.5 (1.4) 4 .53 aMean values represent a five-point Likert scale ranging from 1 = not important at all to 5 = most important. bRank is based on the aggregate level of importance placed on each variable. cBased on independent t-test. VOLUME 1: NO. 4 OCTOBER 2004 Appendix Recreational Trail Evaluation Survey, Morgantown, WVa, 2001 Interviewer name: Interview date: Interview time: Trailhead location: Statement: Hello, We are conducting an interview about the recre- ational trail on behalf of the Division of Exercise Physiology and the Prevention Research Center at West Virginia University. We would like to get your opinions about the usage of the trail. The interview will take approximately five minutes to complete. Your responses are confidential and no identifying information will be obtained. Participation is voluntary and you may refuse to answer any ques- tions. 1. Have you already been interviewed? Yes (Stop — not eligible) or No (Continue) 2. Would you like to participate? Yes (Continue to question 3) or no (Stop — not eligible) 3. Are you 18 or older? Yes (Continue to question 4) or no (Stop — not eligible) 4. How long have you been using the trail? (Weeks, months, or year) 5. What type of activity do you usually do on the trail? (Walk, run, bike, or inline skate) 6. How far do you usually perform [stated activity]? (Miles) 7. Where do you usually enter and exit the trail? (Caperton Trail: start, turn around, or finish) (Decker’s Creek Trail: start, turn around, or finish) 8. How many minutes does this usually take you? 9. How many times (days) per week do you use the trail for [stated activity]? 10. Is there a second activity that you do on the trail? (If no, skip to 15) 11. How far do you usually perform [stated activity]? (Miles) 12. Where do you usually enter and exit the trail? (Caperton Trail: start, turn around, or finish) (Decker’s Creek Trail: start, turn around, or finish) 13. How many minutes does this usually take you? 14. How many times (days) per week do you use the trail for [stated activity]? 15. Did you exercise regularly (three or more times per week for 20 minutes per session) before using this trail? Yes or no 16. a. Since using the trail, has the amount of exercise that you do: Increased (Skip to question 16b); decreased (why?); stayed the same; or don’t know b. Since using the trail, approximately how much has your exercise increased? (0–25%; 26–50%; 51–75%; 76–100%; over 100%) 17. On most days, where do usually come from to get to the trail? (Work, home, school, or other [identify other]) 18. On most days how do you get to the trail? (Walk, drive, bicy- cle, bus, or other [identify other]) 19. How far do you travel to use the trail? (Miles) 20. How long does it take you to get to the trail by walking? (Minutes) 21. While on the trail do you usually exercise: with others or alone? (If alone, skip to question 23) 22. Who do you usually exercise with? (Friends, other family members/relatives, spouse/partner, walk/run club, children, pets, or other [identify other]) 23. What time of the day do you usually use the trail? [Read cat- egories aloud] Early morning (5–8:00 AM), Morning (8–11:00 AM), Midday (11:00 AM–2:00 PM), Afternoon (2-6:00 PM), Evening (after 6:00 PM), Varies, Refuse to answer 24. Please rate the following reasons on why you use this trail instead of other facilities on a scale of 1 to 5: 25. What is the primary reason why you use the trail instead of other facilities? 26. Please rate the following concerns you have about the trail on a scale of 1 to 5: Least important = 1 to most important = 5 Safety (free from personal injury) 1 2 3 4 5 Scenery (beauty of environment) 1 2 3 4 5 Access (no cost associated with use) 1 2 3 4 5 Terrain (paved, flat) 1 2 3 4 5 Convenience (location) 1 2 3 4 5 Friendly atmosphere (social environment) 1 2 3 4 5 Other (please identify) 1 2 3 4 5 Least important = 1 to most important = 5 Unsafe 1 2 3 4 5 Parking (cost, lack of) 1 2 3 4 5 Accessibility of the trail 1 2 3 4 5 Facilities (restrooms, water fountains) 1 2 3 4 5 Maintenance 1 2 3 4 5 Space/congestion on the trail 1 2 3 4 5 Fear of injury 1 2 3 4 5 Lack of police patrol 1 2 3 4 5 Visibility of distance/mile markers 1 2 3 4 5 Other (please identify) 1 2 3 4 5 10 Centers for Disease Control and Prevention • www.cdc.gov/pcd/issues/2004/oct/04_0058.htm The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. 27. What concerns you the most about the trail? 28. Apart from your trail activities, in the past month, have you participated in any of the following? 29. Age: 18–25, 26–35, 36–45, 46–65, 65 and above, declined to answer 30. Sex: Male, female, declined to answer 31. Race/ethnic origin: White, African American, Asian American, Hispanic, other (identify other), declined to answer 32. Employment status: Homemaker, self-employed, student, employed for wages, retired, unemployed, other (identify other), declined to answer 33. Educational attainment: Eighth grade or less, high school or GED, technical school, college graduate, graduate school, professional degree, declined to answer 34. Income level: Under $10,000; $10-30,000; $31-60,000; more than $60,000; declined to answer VOLUME 1: NO. 4 OCTOBER 2004 www.cdc.gov/pcd/issues/2004/oct/04_0058.htm • Centers for Disease Control and Prevention 11 The opinions expressed by authors contributing to this journal do not necessarily reflect the opinions of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors’ affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above. Number of Minutes per Yes No days per week session Aerobic dance Bicycling Strength training Golf Jogging/running Walking Gardening Swimming/water exercises Organized team sports Housework Other

Use of a community trail among new and habitual exercisers: a preliminary assessment.

09-15-2004

Gordon, Paul M,Zizzi, Samuel J,Pauline, Jeff

eng

PMC8834746




 Citation: Muniz-Pardos, B.; Zelenkova, I.; Gonzalez-Aguero, A.; Knopp, M.; Boitz, T.; Graham, M.; Ruiz, D.; Casajus, J.A.; Pitsiladis, Y.P. The Impact of Grounding in Running Shoes on Indices of Performance in Elite Competitive Athletes. Int. J. Environ. Res. Public Health 2022, 19, 1317. https://doi.org/10.3390/ ijerph19031317 Academic Editors: Roberto Alonso González Lezcano, Francesco Nocera and Rosa Giuseppina Caponetto Received: 24 November 2021 Accepted: 23 January 2022 Published: 25 January 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). International Journal of Environmental Research and Public Health Article The Impact of Grounding in Running Shoes on Indices of Performance in Elite Competitive Athletes Borja Muniz-Pardos 1,2,3 , Irina Zelenkova 2,3 , Alex Gonzalez-Aguero 1,2,3 , Melanie Knopp 4 , Toni Boitz 4, Martin Graham 4, Daniel Ruiz 4 , Jose A. Casajus 2,3,5 and Yannis P. Pitsiladis 3,6,7,8,* 1 Faculty of Health and Sports Science (FCSD), Department of Physiatry and Nursing, University of Zaragoza, 50009 Zaragoza, Spain; bmuniz@unizar.es (B.M.-P.); alexgonz@unizar.es (A.G.-A.) 2 GENUD (Growth, Exercise, Nutrition and Development) Research Group, Department of Physiatry and Nursing, University of Zaragoza, 50009 Zaragoza, Spain; iz@i1.ru (I.Z.); joseant@unizar.es (J.A.C.) 3 International Federation of Sports Medicine (FIMS), 1007 Lausanne, Switzerland 4 adidas Innovation, adidas AG, 91074 Herzogenaurach, Germany; Melanie.Knopp@adidas.com (M.K.); Toni.Boitz@adidas.com (T.B.); martin.william.graham@adidas.com (M.G.); daniel.ruiz@adidas.com (D.R.) 5 Faculty of Medicine, Department of Physiatry and Nursing, University of Zaragoza, 50009 Zaragoza, Spain 6 School of Sport and Health Sciences, University of Brighton, Eastbourne BN20 7SN, UK 7 Centre for Exercise Sciences and Sports Medicine, FIMS Collaborating Centre of Sports Medicine, University of Rome “Foro Italico”, 00135 Rome, Italy 8 European Federation of Sports Medicine Associations (EFSMA), 1007 Lausanne, Switzerland * Correspondence: y.pitsiladis@brighton.ac.uk Abstract: The introduction of carbon fiber plate shoes has triggered a plethora of world records in running, which has encouraged shoe industries to produce novel shoe designs to enhance running performance, including shoes containing conductor elements or “grounding shoes” (GS), which could potentially reduce the energy cost of running. The aim of this study was to examine the physiological and perceptual responses of athletes subjected to grounding shoes during running. Ten elite runners were recruited. Firstly, the athletes performed an incremental running test for VO2max and anaerobic threshold (AT) determination, and were familiarized with the two shoe conditions (traditional training shoe (TTS) and GS, the latter containing a conductor element under the insole). One week apart, athletes performed running economy tests (20 min run at 80% of the AT) on a 400 m dirt track, with shoe conditions randomized. VO2, heart rate, lactate, and perceived fatigue were registered throughout the experiment. No differences in any of the physiological or perceptual variables were identified between shoe conditions, with an equal running economy in both TTS and GS (51.1 ± 4.2 vs. 50.9 ± 5.1 mL kg−1 min−1, respectively). Our results suggest that a grounding stimulus does not improve the energy cost of running, or the physiological/perceptual responses of elite athletes. Keywords: earthing; environmental physiology; running performance; running economy; shoe technology; grounding 1. Introduction During the past five years, shoe designs have experienced a great technological revolution, which has been accompanied by a plethora of world records in all long- distance running events (i.e., from 5000 m to marathons, in both male and female athletes). Joyner et al., recently suggested that the factors potentially explaining the recent records in long-distance running are the physiological and training factors, in addition to shoe technology and drafting [1]. However, the abrupt drop in world records across all distances since 2017 suggests that shoe technology has a major contribution when compared to the other factors (i.e., training methods, the physiology of athletes, and drafting are factors that have not substantially changed in the last 5 years) [2]. Int. J. Environ. Res. Public Health 2022, 19, 1317. https://doi.org/10.3390/ijerph19031317 https://www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2022, 19, 1317 2 of 10 The most popular shoe technology for road running includes a carbon fiber plate (CFP) within the sole, a light and highly reactive foam, and a stack up to 40 mm in thickness. This technology has been shown to reduce the energy cost of running during a fixed exercise intensity (traditionally between 14 and 18 km h−1) by approximately 4%, when compared to non-CFP shoes [3–5]. This improved running economy (RE) seems to be elicited by an increase in energy return caused by the action of passive elastic recoil, which in turn increases stride length and contact times, reduces step frequencies, and slightly increases the peak forces upon ground contact, when compared to non-CFP shoes [3,6,7]. The great popularity and effectiveness of CFP shoes has encouraged the shoe industry to explore new forms of shoe designs to optimize both health and performance during running. The implementation of “grounding” in humans purports to take advantage of the prolonged contact between an individual and the ground, and the potential transmission of energy between the two. Previous research states that the “direct contact of humans with the earth or using a metal conductor changes the electric potential on the surface of the body, as well as within the entire human organism” [8]. While the etiology of this potential effect is difficult to explain from a biophysiological perspective, previous findings have shown that the direct contact of an individual with the ground may reduce inflammatory processes, mood, pain, and stress at rest [9–11] and during exercise [8,9], with some studies suggesting that grounding technology may have a medical application. For example, previous research has suggested that the implementation of grounding is beneficial for mood, and may be especially beneficial in cases of depression, anxiety, stress, and trauma [11,12]. In relation to the existing research on grounding and exercise, an informative pilot study examined the effects of grounding on muscle physiology in response to exercise- induced muscle damage, and observed faster muscle recovery times under the grounding condition compared to the placebo [13]. The same group performed a more comprehensive follow-up study [14], observing that grounding significantly reduced creatine kinase (CK) levels 24 h post-exercise when compared to the placebo, suggesting that grounding may reduce acute muscular damage post-exercise. Following these early studies on grounding and muscle damage, a further study focused on the impact that this technology may have during aerobic exercise [8]. Sokal et al. claimed that the indirect contact of cyclists with the ground (through a metal conductor) while exercising elicited an increase in the electrical potential of the body when compared to those in the control group (not grounded). This study further reported that the observed increase in electrical potential with was accompanied by a greater decrease in blood urea concentrations during and after a 30 min cycling test at 50% of VO2max, indicating, according to the authors, a decreased physiological stress [8]. While these previous studies showed a benefit of grounding on the muscle recovery and physiological stress of healthy subjects in response to different modes of exercise (i.e., resistance training and cycling), the impact of this technology while running is unknown. Given the imminent introduction of grounding technology in running shoes, and the absence of rigorous scientific evidence of its effects, adding conductor elements within the shoe and employing a well-controlled experimental design, would allow for the assessment of any putative effects of this technology (i.e., grounding technology in running shoes) during running. This is especially important given the recent controversy that novel shoe technologies are negatively impacting the integrity and fairness within sport [2,15]. A recent critical review [2] highlighted how novel shoe designs are revolutionizing the world of sport, as numerous National, European, World, and Olympic records have been broken over an extraordinarily short time period (i.e., since the introduction of CFP shoes). In addition to this controversy, there is a lack of well-controlled and rigorous studies in the field that focus on the impact of shoe designs on running performance [2], which makes the true performance benefits of certain shoe technologies difficult to determine. Considering the reduced physiological stress and muscle damage witnessed in subjects while performing other physical activities (i.e., strength exercises and cycling), it is impor- tant to examine the impact of grounding on the physiological and perceptual responses to Int. J. Environ. Res. Public Health 2022, 19, 1317 3 of 10 running, especially considering the interest of shoe companies in incorporating grounding technology into running shoes, and the potential fairness/integrity issues that may result if a performance benefit is demonstrated. Therefore, the main aim of the present study was to compare the RE and physiological stress of well-trained runners while running in either grounding shoes (GS) or traditional training shoes (TTS). 2. Materials and Methods 2.1. Participants Ten highly-trained male runners (age = 27 ± 7 years; weight = 64.6 ± 6 kg; height = 176.3 ± 5.4 cm) were recruited for the present study. Upon recruitment, all subjects received and signed an informed consent form in order to participate in the study. Subjects were required to meet the following inclusion criteria: (1) to train a minimum of 50 km week−1, (2) to have a personal best under 35:00 min:s in 10 km or 17:30 min:s in 5 km, (3) to be healthy and without any musculoskeletal injury. 2.2. Procedures The present study design required runners to visit either the laboratory or the track on two occasions, both separated by a period of 7 days to avoid any residual fatigue. Visit 1 included a VO2max test, ventilatory threshold determination, and shoe familiarization in the laboratory; Visit 2 included 20 min RE tests at 80% of the anaerobic threshold, on a 400 m dirt track, with the order of the two shoe conditions randomized (Figure 1). A dirt track was selected over a traditional synthetic PU rubber track to avoid any material interference between the ground and the athlete. The present study was approved by the Ethics Committee of Aragon (CEICA, num. 17/2021). 2.3. Shoe Conditions Two shoe conditions were tested: the traditional training shoe (TTS) and the grounding shoe (GS), with these being visually identical as shown in Figure 1. Shoes with grounding potential contained a conductor element around the insole, and aimed to diminish the physiological stress experienced by the athlete during running as they run in closer contact with the ground. The insulation and thermal permeability of the shoes were considered similar, given that the same material was used for both experimental and non-experimental shoes, with the exception of the conductor element. Both uppers consisted of the same knitted textile, produced and supplied at the same time for both types of shoe (Figure 1). The GS upper included a textile webbing containing yarn that encouraged electrical charge to flow through the material. The material was stitched into the collar area, and ran through the midsole to connect with the rubber on the outsole that contacts the ground. The TTS outsole included conventional rubber, while the GS outsole included rubber that encouraged the flow of electrical charge. The manufacturers labelled the shoes with a number in red or blue according to the two shoe conditions, and this setting was used by the research team to keep the study design double-blinded (See Figure 1). Additionally, as each athlete may have become subjectively biased during the familiarization trial, all blue/red labels were obscured with tape in Visit 2. All athletes had their own pair of shoes for each shoe condition. 2.4. Visit 1. Maximal Oxygen Uptake and Ventilatory Threshold Determination On the first day, athletes were subjected to a skin temperature test and a SARS-CoV-2 antigen test, in order to participate in this study. Upon testing negative, informed consent was signed by all participants, and medical history and pre-participation screening was also completed. The laboratory assessments performed during the first day included: Anthropometric and body composition assessments. The parameters measured were as follows: weight, height, height from sitting position, foot length, calf circumference and fold, and thigh circumference and fold. Percent body fat, muscle mass, and bone mass were assessed with a DXA scan (Hologic Corp., Bedford, MA, USA). Body fat, body water, and Int. J. Environ. Res. Public Health 2022, 19, 1317 4 of 10 muscle mass were also assessed via bioimpedance (TANITA BC 780-S MA, Tanita Corp., Tokyo, Japan). Public Health 2022, 19, x FOR PEER REVIEW 4 of 11 Figure 1. Image of the right grounding shoe (A) and traditional training shoe (B) for one of the elite athletes. 2.3. Shoe Conditions Two shoe conditions were tested: the traditional training shoe (TTS) and the grounding shoe (GS), with these being visually identical as shown in Figure 1. Shoes with grounding potential contained a conductor element around the insole, and aimed to diminish the physiological stress experienced by the athlete during running as they run in closer contact with the ground. The insulation and thermal permeability of the shoes were considered similar, given that the same material was used for both experimental and non-experimental shoes, with the exception of the conductor element. Both uppers consisted of the same knitted textile, produced and supplied at the same time for both types of shoe (Figure 1). The GS upper included a textile webbing containing yarn that encouraged electrical charge to flow through the material. The material was stitched into the collar area, and ran through the midsole to connect with the rubber on the outsole that contacts the ground. The TTS outsole included conventional rubber, while the GS outsole included rubber that encouraged the flow of electrical charge. The manufacturers labelled the shoes with a number in red or blue according to the two shoe conditions, and this setting was used by the research team to keep the study design double-blinded (See Figure 1). Additionally, as each athlete may have become subjectively biased during the familiarization trial, all blue/red labels were obscured with tape in Visit 2. All athletes had their own pair of shoes for each shoe condition. Figure 1. Image of the right grounding shoe (A) and traditional training shoe (B) for one of the elite athletes. Maximal aerobic capacity test. All subjects were previously familiarized with VO2max testing. Prior to the VO2max test, subjects laid down for 5 min, and resting electrocar- diograms and blood pressure tests were performed and assessed by experienced medical doctors to ensure athletes did not have any cardiological issues. Participants breathed through a low dead space mask, with air sampled at 60 mL min−1. Before each test, two-point calibrations of the gas sensors were completed, using a known gas mixture (16% O2 and 5% CO2) and ambient air. Ventilatory volume was calibrated using a 3 L (±0.4%) syringe. Firstly, subjects performed a self-paced warm-up, and prior to the com- mencement of the test, subjects were instrumented with a portable metabolic analyzer (Cosmed K5, Cosmed Srl, Rome, Italy) and a heart rate device (Polar H10, Polar Electro, Kempele, Finland). A short-ramp incremental protocol was used (i.e., 13–16 min) as this has been shown to be the most appropriate assessment for identifying individual physiological events in well-trained runners [16–18]. The protocol consisted of a 3 min run at 10 km h−1 and a 1% gradient on a treadmill (h/p/cosmos, Nussdorf—Traunstein, Germany), followed by increases of 1 km h−1 min−1 until volitional exhaustion. Heart rate was monitored throughout the test, and overall perception of effort (RPE) and specific RPE for the legs were registered immediately after the test. This test enabled the determination of VO2max (defined as the highest 30 s mean values obtained during the test) and individual anaerobic threshold (IAT), determined through visual assessment conducted by two experienced exercise physiologists. Each individual speed for subsequent shoe trials were determined Int. J. Environ. Res. Public Health 2022, 19, 1317 5 of 10 at the 80% of the IAT velocity. This VO2max test involved the subjects’ preferred shoe, and served to objectively quantify individual running speed for subsequent RE trials (avoiding the impact of the slow component of oxygen uptake given the repeated square-wave design of the RE tests on the second visit). Visit 1 also involved the familiarization of the different running shoes during a light, 5 min run with each pair of shoes, in preparation for Visit 2. 2.5. Visit 2. Running Economy Tests During the second visit, indices of performance, with particular focus on RE, were assessed for each shoe condition, determined on a 400 m dirt track. Air temperature and humidity were recorded at the beginning and end of the experimental sessions using a portable meteorological station, and all trials were performed either in the early morning or late evening to avoid extreme environmental conditions. Participants breathed through a low dead space mask, with air sampled at 60 mL min−1. Before each subject’s first trial, the portable metabolic analyzer was calibrated following the calibration procedures aforementioned. The shoe conditions were randomly assigned, and both runners and assessors were blinded to the shoe condition. Brand new socks were used for each RE trial to avoid excessive humidity within the shoe, as this could impact grounding effect. Body mass was measured before and after each test. Each runner warmed up for 15 min with their preferred training shoes prior to being equipped with the portable metabolic analyzer. Pre-trial blood lactate was measured from a single drop of whole blood from the fingertip using a lactate meter (Lactate Pro 2, Arkray Europe, B.V., Amstelveen, the Netherlands), and pre-trial heart rate and RPE were also collected. Athletes performed two 20 min exercise bouts at 80% of their IAT velocity for each shoe condition, with a 20 min rest in between (Figure 2). The duration of this RE protocol was longer than traditional RE tests (4–6 min) used in previous studies examining shoe designs [3–5]. The reason for this was to allow for a longer contact time between the athlete and the earth, which is crucial for obtaining a dose–response relationship. Lactate, whole-body RPE, and legs-only RPE (1–10 scale) were recorded at min 1, 3, and 15 of recovery following both trials, and heart rate and ventilatory parameters were monitored throughout the test. A researcher (and experienced cyclist) paced all runners at their individual speed using a bicycle. The RE elicited by each shoe condition was determined as the mean VO2 between min 10 to min 15, as steady state was ensured during this period. To reduce the noise in the ventilatory measurements, a 7-breath averaging method was performed. Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW 6 of 11 fingertip using a lactate meter (Lactate Pro 2, Arkray Europe, B.V., Amstelveen, the Netherlands), and pre-trial heart rate and RPE were also collected. Athletes performed two 20 min exercise bouts at 80% of their IAT velocity for each shoe condition, with a 20 min rest in between (Figure 2). The duration of this RE protocol was longer than traditional RE tests (4–6 min) used in previous studies examining shoe designs [3–5]. The reason for this was to allow for a longer contact time between the athlete and the earth, which is crucial for obtaining a dose–response relationship. Lactate, whole-body RPE, and legs-only RPE (1–10 scale) were recorded at min 1, 3, and 15 of recovery following both trials, and heart rate and ventilatory parameters were monitored throughout the test. A researcher (and experienced cyclist) paced all runners at their individual speed using a bicycle. The RE elicited by each shoe condition was determined as the mean V̇O2 between min 10 to min 15, as steady state was ensured during this period. To reduce the noise in the ventilatory measurements, a 7-breath averaging method was performed. Figure 2. Protocol for the running economy trials at 80% of the anaerobic threshold (AT). 2.6. Statistical Analysis Means and standard deviations (mean ± SD) were calculated for all variables. An a priori sample size calculation (G*Power software, version 3.1.9.3, Heinrich-Heine- Universität Düsseldorf, Düsseldorf, Germany) was performed using the running economy data reported in a previous study testing different shoe designs in well-trained athletes (Barnes et al., 2018). The V̇O2 data for both the control and grounded shoe (53.61 ± 2.20 vs. 51.26 ± 2.23 mL kg−1 min−1, respectively) were used to generate a correlation 20 min 20 min 20 min 15 min warm-up 20 min Shoe 1 Shoe 2 80% AT 80% AT 15’ 15’ 3’ 3’ 1’ 1’ rest rest Rating of perceived exertion Lactate sample Heart rate recording Gas analysis Shoe condition Body mass Figure 2. Protocol for the running economy trials at 80% of the anaerobic threshold (AT). Int. J. Environ. Res. Public Health 2022, 19, 1317 6 of 10 2.6. Statistical Analysis Means and standard deviations (mean ± SD) were calculated for all variables. An a pri- ori sample size calculation (G*Power software, version 3.1.9.3, Heinrich-Heine-Universität Düsseldorf, Düsseldorf, Germany) was performed using the running economy data re- ported in a previous study testing different shoe designs in well-trained athletes (Barnes et al., 2018). The VO2 data for both the control and grounded shoe (53.61 ± 2.20 vs. 51.26 ± 2.23 mL kg−1 min−1, respectively) were used to generate a correlation coefficient of 0.45 and a Cohen’s d of 1.01. A two-tailed t-test revealed that a total sample size of 10 subjects was required to obtain statistical power of 0.80 and an alpha of 0.05. A Shapiro– Wilk test revealed normal data distributions across all studied variables. Student’s t-tests for paired samples were applied between TTS and GS shoe conditions in order to examine the differences between metabolic and RE data (HR, VO2, RER). Significant values were set at p ≤ 0.05 and effect sizes (Cohen’s d) were also calculated. The Statistical Package for the Social Sciences (SPSS) version 23.0 (SPSS Inc., Chicago, IL, USA) was used to perform the statistical analyses. 3. Results A final sample of 10 athletes completed the present study, with no drop-outs. These athletes were national to international level runners/triathletes, with two of them having participated in major sporting events (Olympic Games and World Championships). Table 1 presents the mean and individual descriptive characteristics of the sample, showing a fairly homogeneous fitness level across all runners (i.e., mean VO2max of 78.4 ± 3.8 mL kg−1 min−1). Table 1. Descriptive characteristics of the participants. ID Age (years) Weight (kg) Height (cm) BMI (kg m−2) Bioimpedance (Fat %) VO2max (mL kg−1 min−1) Athlete 1 31.0 78.5 180.3 24.1 12.7 76.0 Athlete 2 25.7 65.7 177.8 20.8 5.5 82.3 Athlete 3 35.0 64 174.3 21.1 10.4 80.3 Athlete 4 20.8 68.9 186.3 19.9 11.8 83.6 Athlete 5 31.1 57.0 171.0 19.5 3.0 78.0 Athlete 6 26.2 59.3 170.2 20.5 11.2 77.8 Athlete 7 38.2 66.0 176.5 21.2 3.8 78.5 Athlete 8 25.0 72.5 177.7 23.0 7.0 77.3 Athlete 9 20.6 64.9 171.2 22.1 8.9 80.5 Athlete 10 18.1 64.0 183.0 19.1 8.5 69.9 Mean ± SD 27.2 ± 6.6 66.1 ± 6.2 176.8 ± 5.4 21.1 ± 1.6 8.3 ± 3.4 78.4 ± 3.8 A Student’s t-test for paired samples revealed no significant difference in RE values between TTS and GS conditions (51.1 ± 4.2 vs. 50.9 ± 5.1 mL kg−1 min−1, respectively, p = 0.779, Cohen’s d = 0.092). Figure 3 shows both mean and individual values for VO2. Additionally, blood lactate was not different between shoe conditions at min 1 (p = 0.793), min 3 (p = 0.250), and min 15 (p = 0.641) post-exercise (Figure 4). Both whole-body and legs-only RPE values were also not significantly different between TTS and GS at min 1 (p = 1.0 and p = 0.273, respectively), min 3 (p = 0.443 and p = 0.591, respectively), and min 15 (p = 0.168 and p = 0.591, respectively) post-exercise (Figure 4). Finally, HR val- ues were not significantly different between TTS and GS during exercise (150.1 ± 15 vs. 151.0 ± 16 bpm, respectively, p = 0.461, Cohen’s d = 0.244; Figure 4). Int. J. Environ. Res. Public Health 2022, 19, 1317 7 of 10 Additionally, blood lactate was not different between shoe conditions at min 1 (p 0.793), min 3 (p = 0.250), and min 15 (p = 0.641) post-exercise (Figure 4). Both whole-body and legs-only RPE values were also not significantly different between TTS and GS at min 1 (p = 1.0 and p = 0.273, respectively), min 3 (p = 0.443 and p = 0.591, respectively), and min 15 (p = 0.168 and p = 0.591, respectively) post-exercise (Figure 4). Finally, HR values were not significantly different between TTS and GS during exercise (150.1 ± 15 vs. 151.0 ± 16 bpm, respectively, p = 0.461, Cohen’s d = 0.244; Figure 4). Figure 3. Mean and individual running economy values (mL kg−1 min−1) of the 10 athletes running in traditional training shoes (grey column) and in grounding shoes (black column). Figure 3. Mean and individual running economy values (mL kg−1 min−1) of the 10 athletes running in traditional training shoes (grey column) and in grounding shoes (black column). Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW 8 of 11 Figure 4. Blood lactate (A), whole-body rate of perceived exertion (RPE; B), and legs-only RPE (C) during the recovery period after running in the traditional training shoe (TTS, gray solid line) or grounding shoe (GS, black solid line). Heart rate during the running economy trial in both TTS and GS trials (D). Dashed lines represent overlapping mean values between shoes. 4. Discussion The main findings of the present study show that grounding technology applied to shoe designs does not provide a physiological/perceptual response over traditional training shoes in well-trained athletes. The RE, blood lactate, heart rate, and perceptual response of these athletes, exercising at 80% of their IAT during 20 min on a 400 m dirt track, were not different between shoes conditions. De ite e iou o i i fi di u e ti that ou di te h olo y ha 0 1 2 3 4 5 6 min 1 min 3 min 15 RPE value (1–10) Recovery time 135 140 145 150 155 160 165 170 Heart rate (bpm) 0 0.5 1 1.5 2 2.5 min 1 min 3 min 15 Blood lactate concentration (mmol·L–1) Recovery time A C TTS GS B 0 1 2 3 4 5 6 7 min 1 min 3 min 15 RPE legs value (1–10) Recovery time D Figure 4. Blood lactate (A), whole-body rate of perceived exertion (RPE; B), and legs-only RPE (C) during the recovery period after running in the traditional training shoe (TTS, gray solid line) or grounding shoe (GS, black solid line). Heart rate during the running economy trial in both TTS and GS trials (D). Dashed lines represent overlapping mean values between shoes. Int. J. Environ. Res. Public Health 2022, 19, 1317 8 of 10 4. Discussion The main findings of the present study show that grounding technology applied to shoe designs does not provide a physiological/perceptual response over traditional training shoes in well-trained athletes. The RE, blood lactate, heart rate, and perceptual response of these athletes, exercising at 80% of their IAT during 20 min on a 400 m dirt track, were not different between shoes conditions. Despite previous promising findings suggesting that grounding technology has posi- tive effects on the physiological responses (i.e., reduced acute inflammatory processes) of humans at rest [7,8], very limited research has focused on the implementation of ground- ing during exercise, with only two studies focusing on the effectiveness of grounding in reducing muscular damage after exercise-induced DOMS. This is the first study to examine the impact of grounding in shoes during running, which makes the comparison with previous studies challenging due to the unique nature of running for the implementation of this technology (i.e., intermittent contact time with the ground). Our findings, however, differ from those of Sokal et al. [8], who claimed that all recreational cyclists within their study experienced physiological attenuation at rest, during a 30 min exercise at 50% of their VO2max, and during recovery, indicated by decreases in blood urea; however, these authors failed to include any individual data. It is also worth noting that these biochemical parameters were not measured immediately prior to grounding/placebo conditions, and therefore group-by-time interactions could not be determined, which limits the interpre- tation of these results. Additionally, one would expect both blood urea and creatinine concentrations to remain unchanged following the exercise protocol used by these authors (a single bout of light exercise for 30 min). Blood urea and creatinine levels have been shown to increase after prolonged, strenuous exercise as a result of increased protein catabolism and/or impaired renal function [19], which is unlikely to have occurred during the ex- ercise protocol proposed by Sokal et al. The difference between the groups observed by Sokal et al., interpreted in the context of our present findings, are more likely due to day- to-day inter-individual variability in blood urea, or some potential methodological issues during data collection, rather than due to physiological stress attenuation during exercise. In a subsequent study, Sokal et al. presented additional data from the same aforementioned experiment [20], focusing on the effects of grounding on VO2 uptake, blood glucose, lactate, and bilirubin concentrations. The 42 subjects included in this study were divided into two subgroups (n = 21) according to their VO2max, therefore, both groups had a comparable cardiorespiratory fitness (Group A = 50.8 vs. Group B = 50.7 mL kg−1 min−1). The study design followed a double-blind, crossover protocol between Groups A and B. During the first testing day, Group A was under the placebo condition and Group B was under the grounding stimulus, with these conditions interchanged during the second day of testing. These authors reported a significantly reduced VO2 uptake (numeric data not shown by the authors) at the end of the exercise with the grounding stimulus only in Group B, when com- pared to the placebo. The study design employed by Sokal et al. [8,20] has limited reliability, given that their experimental tests were performed on different days, which may have biased the results. Day-to-day variability and the lack of a familiarization trial may have potentiated the learning effects only for Group B (i.e., the group with the grounding stimulus during the second day). These results should, therefore, be interpreted with caution. To our knowledge, the two aforementioned studies are the only two experiments focusing on the effects of grounding on the biophysiological responses of humans during submaximal exercise. However, the important methodological issues described above, and the use of cycling being the only mode of exercise, limits the interpretation of the current literature and its comparison with the present study. In our experiment, we used a double- blind, randomized, crossover design, with tests for all experimental conditions performed on the same day. We are aware that the conductor element within the shoe was not in permanent contact with the ground (i.e., intermittent contact time during running), and we did not measure muscle activity, nor foot/stride mechanics, during running, which may have provided more information and potentially revealed an effect. However, to ensure a sufficient Int. J. Environ. Res. Public Health 2022, 19, 1317 9 of 10 contact time, we designed a longer than usual RE protocol (i.e., 20 min bouts; Figure 2), so that we could identify a potential dose–response relationship over time. Despite these rigorous experimental procedures, our results show that grounding technology did not have any impact on the measured responses during running when compared to traditional training shoes. Previous research showed a decrease in muscle damage in response to high-intensity strength exercises in subjects under grounding conditions [13,14] when compared to a placebo. These findings would suggest that grounding technology may have a role to play as a muscle recovery method, which in turn could translate into a benefit for runners when performing higher intensity exercise (i.e., above the anaerobic threshold) in which muscle fatigue and acidosis occur to a greater extent. Nonetheless, future research using larger sample sizes and examining foot mechanics (especially contact times) would be required to confirm our findings. Other shoe designs currently available on the market that include a CFP and a high midsole stack height made of compliant, resilient, and lightweight foam, seem the most effective shoe modality to date. This technology has shown to improve RE by increasing the midsole longitudinal bending stiffness, favoring a decrease in the range of motion of the metatarsophalangeal joint [3,21,22]. 5. Conclusions In conclusion, our results suggest that grounding in shoe designs is not an effective alternative for well-trained athletes to improve their running efficiencies, and/or their physiological/perceptual responses during submaximal exercise. However, there are intrinsic limitations that should be considered. Potential grounding effects could have been missed during our study as running does not allow constant contact between the athlete and the ground, which could have potentially biased the results. In relation to this, lower caliber athletes may have benefited from this technology given their ground contact times are greater than faster, elite athletes; an issue that could not be addressed in the current study. Future research may therefore consider additional sports in which athletes remain in constant contact with the ground (e.g., race-walking, cross-country skiing, powerlifting). Despite these limitations, our study followed a high-quality methodological protocol (double-blind, randomized, crossover design) using a homogeneous sample of highly trained athletes (as represented in Table 1), which suggests that our conclusions are reliable for this specific population. Author Contributions: Conceptualization and methodology: B.M.-P., I.Z., M.K., D.R., J.A.C. and Y.P.P.; formal analysis: A.G.-A., J.A.C., I.Z., B.M.-P.; writing—original draft preparation: B.M.-P., I.Z., A.G.-A.; review and editing: B.M.-P., I.Z., A.G.-A., M.K., T.B., M.G., D.R., J.A.C. and Y.P.P.; supervision: Y.P.P. and J.A.C. All authors have read and agreed to the published version of the manuscript. Funding: This study was supported by a contract from adidas AG with the University of Zaragoza, Spain (Project: “Testing support for innovation project”; number 2021/0348). Institutional Review Board Statement: The present study was approved by the Ethics Committee of Aragon, Spain (CEICA, num. 17/2021). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: The datasets used and analyzed within the present manuscript will be available from the corresponding author/first author upon request. Acknowledgments: We wish to thank the athletes involved in this study for participating. Conflicts of Interest: M.K., T.B., M.G., D.R. are employees of adidas AG. B.M.P., I.Z., A.G.A., J.A.C., Y.P.P. have no conflicts of interest relevant to the content of this article. References 1. Joyner, M.J.; Hunter, S.K.; Lucia, A.; Jones, A.M. Physiology and fast marathons. J. Appl. Physiol. 2020, 128, 1065–1068. [CrossRef] [PubMed] 2. Muniz-Pardos, B.; Sutehall, S.; Angeloudis, K. Recent Improvements in Marathon Run Times Are Likely Technological, Not Physiological. Sports Med. 2021, 51, 371–378. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2022, 19, 1317 10 of 10 3. Hoogkamer, W.; Kipp, S.; Frank, J.H.; Farina, E.M.; Luo, G.; Kram, R. A comparison of the energetic cost of running in marathon racing shoes. Sports Med. 2018, 48, 1009–1019. [CrossRef] 4. Barnes, K.R.; Kilding, A.E. A Randomized Crossover Study Investigating the Running Economy of Highly-Trained Male and Female Distance Runners in Marathon Racing Shoes versus Track Spikes. Sports Med. 2019, 49, 331–342. [CrossRef] [PubMed] 5. Hunter, I.; McLeod, A.; Valentine, D.; Low, T.; Ward, J.; Hager, R. Running economy, mechanics, and marathon racing shoes. J. Sport Sci. 2019, 37, 2367–2373. [CrossRef] [PubMed] 6. Hoogkamer, W.; Kram, R.; Arellano, C.J. How biomechanical improvements in running economy could break the 2-hour marathon barrier. Sports Med. 2017, 47, 1739–1750. [CrossRef] 7. Rodrigo-Carranza, V.; González-Mohíno, F.; Santos-Concejero, J.; González-Ravé, J.M. The Effects of Footwear Midsole Longitudi- nal Bending Stiffness on Running Economy and Ground Contact Biomechanics: A Systematic Review and Meta-analysis. Eur. J. Sport Sci. 2021, 1, 26. [CrossRef] 8. Sokal, P.; Jastrz˛ebski, Z.; Jaskulska, E.; Sokal, K.; Jastrz˛ebska, M.; Radzimi´nski, Ł. Differences in blood urea and creatinine concentrations in earthed and unearthed subjects during cycling exercise and recovery. Evid.-Based Complement Altern Med. Hindawi 2013, 2013, 445. [CrossRef] 9. Menigoz, W.; Latz, T.T.; Ely, R.A.; Kamei, C.; Melvin, G.; Sinatra, D. Integrative and lifestyle medicine strategies should include Earthing (grounding): Review of research evidence and clinical observations. Explore 2020, 11, 152–160. [CrossRef] 10. Oschman, J.L.; Chevalier, G.; Brown, R. The effects of grounding (earthing) on inflammation, the immune response, wound healing, and prevention and treatment of chronic inflammatory and autoimmune diseases. J. Inflamm. Res. 2015, 8, 83. [CrossRef] 11. Chevalier, G. The effect of grounding the human body on mood. Psychol. Reports 2015, 116, 534–542. [CrossRef] [PubMed] 12. de Tord, P.; Bräuninger, I. Grounding: Theoretical application and practice in dance movement therapy. Arts Psychother. 2015, 43, 16–22. [CrossRef] 13. Brown, D.; Chevalier, G.; Hill, M. Pilot study on the effect of grounding on delayed-onset muscle soreness. J. Altern. Complementary Med. 2010, 16, 265–273. [CrossRef] 14. Brown, R.; Chevalier, G.; Hill, M. Grounding after moderate eccentric contractions reduces muscle damage. Open Access J. Sports Med. 2015, 6, 305. [PubMed] 15. Pitsiladis, Y.; Muniz-Pardos, B.; Miller, M.; Verroken, M. Sport integrity opportunities in the time of Coronavirus. Sports Med. 2020, 50, 1701–1702. [CrossRef] 16. Myers, J.; Bellin, D. Ramp exercise protocols for clinical and cardiopulmonary exercise testing. Sport Med. 2000, 30, 23–29. [CrossRef] 17. Myers, J.; Buchanan, N.; Walsh, D.; Kraemer, M.; McAuley, P.; Hamilton-Wessler, M. Comparison of the ramp versus standard exercise protocols. J. Am. Coll. Cardiol. 1991, 17, 1334–1342. [CrossRef] 18. Cerezuela-Espejo, V.; Courel-Ibañez, J.; Moran-Navarro, R.; Martınez-Cava, A.; Pallares, J.G. The relationship between lactate and ventilatory thresholds in runners: Validity and reliability of exercise test performance parameters. Front. Physiol. 2018, 9, 1320. [CrossRef] 19. Warburton, D.; Welsh, R.; Haykowsky, M.; Taylor, D.; Humen, D. Biochemical changes as a result of prolonged strenuous exercise. Br. J. Sports Med. 2002, 36, 301. [CrossRef] 20. Sokal, P.; Jastrz˛ebski, Z.; Sokal, K.; Dargiewicz, R.; Jastrz˛ebska, M.; Radzimi´nski, Ł. Earthing modulates glucose and erythrocytes metabolism in exercise. Age 2016, 21, 21. 21. Weyand, P.G. Now Afoot: Engineered Running Economy. J. Appl. Physiol. 2020, 128, 1083. 22. Stefanyshyn, D.J.; Nigg, B.M. Mechanical energy contribution of the metatarsophalangeal joint to running and sprinting. J. Biomech. 1997, 30, 1081–1085. [CrossRef]

The Impact of Grounding in Running Shoes on Indices of Performance in Elite Competitive Athletes.

01-25-2022

Muniz-Pardos, Borja,Zelenkova, Irina,Gonzalez-Aguero, Alex,Knopp, Melanie,Boitz, Toni,Graham, Martin,Ruiz, Daniel,Casajus, Jose A,Pitsiladis, Yannis P

eng

PMC9447911

RESEARCH ARTICLE Performance in youth track and field is associated with birth quartile. A register- based study among athletes in Norway from 10 years to senior level Hilde GundersenID1*, Anette Harris2, Halvard GrendstadID3, Morten Kristoffersen1, Atle Guttormsen4, Terje Dalen5, Cecilie Brekke Rygh6 1 Department of Sport, Food and Natural Sciences, Western Norway University of Applied Sciences, Bergen, Norway, 2 Department of Psychosocial Science, Faculty of Psychology, University of Bergen, Bergen, Norway, 3 Department of Physical Performance, Norwegian School of Sport Sciences, Oslo, Norway, 4 NMBU School of Economics and Business, Norwegian University of Life Sciences (NMBU), Ås, Norway, 5 Department of Physical Education and Sport Science, Faculty of Teacher Education and Arts, Nord University, Levanger, Norway, 6 Department of Health and Functioning, Western Norway University of Applied Sciences, Bergen, Norway * hsg@hvl.no Abstract Introduction Earlier studies have demonstrated that the oldest in a competition class are more likely to succeed than the youngest, a phenomenon called relative age effect (RAE). Track and field give us an opportunity to investigate the advantage of being born early in the year based upon actual performance, since objective criteria are the performance indicators. Hence, the aim of the present study was to investigate the occurrence of RAE in Norwegian track and field athletes in events where physical capacity is important for success. Methods All individual season best results from the register of The Norwegian Athletics Federation (n = 28 999) obtained in all competition classes from the age of 10 years to senior in both sexes on 60m and 600m from 2011 to 2020 were downloaded. One-way ANOVA and LSD post hoc analyses were used to analyze performance differences according to birth quartiles between athletes. Further, odds ratios (OR) were used to calculate the odds of being among the top-100 for athletes for those born in the first quartile of the year compared to the last. Results The RAE was present in several of the competition classes in sprint compared to middle-dis- tance running, and in more male than female competition classes. Overall, the OR of being among the top-100 in one of the competition classes on 60m sprint when born in first quartile compared to last quartile was 2.88 [2.30–3.62] for males and 1.54 [1.26–1.89] for females. PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 1 / 12 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Gundersen H, Harris A, Grendstad H, Kristoffersen M, Guttormsen A, Dalen T, et al. (2022) Performance in youth track and field is associated with birth quartile. A register-based study among athletes in Norway from 10 years to senior level. PLoS ONE 17(9): e0273472. https:// doi.org/10.1371/journal.pone.0273472 Editor: Caroline Sunderland, Nottingham Trent University, UNITED KINGDOM Received: February 3, 2022 Accepted: August 9, 2022 Published: September 6, 2022 Copyright: © 2022 Gundersen et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: Data from athletes from 13 to 20 years can be retrieved from: https:// www.minfriidrettsstatistikk.info/php/ LandsStatistikk.php?showclass=4&showevent= 0&outdoor=Y&showseason=0&showclub=0 Data from 10-12 years require permission from The Norwegian Athletics Federation. If interested, you may apply to friidrett@friidrett.no. Funding: The author(s) received no specific funding for this work. Conclusion Being born early in the year in events with high demand for specific physical capacities is an advantage in both sexes in most of the youngest competition classes. In males, the advan- tage of being born early in the year lasted longer in sprint than in middle-distance running, indicating that puberty affects performance in sprint and middle-distance running differently. Introduction Organized sport is one of the most popular forms of leisure-time activities worldwide, and at least one-third of children and adolescents are participating in one or more sports in most countries [1]. To provide equal opportunities and competition, children and youth usually compete in age classes based on their chronological age. However, previous studies have shown that children born early in their birth year are likely to perform better than children born later in the same age cohort, a phenomenon called relative age effect (RAE) [2]. The mat- urational hypothesis is perhaps the most common explanation for the RAE, i.e. that chrono- logically older children have a higher chance of being more physically mature, with subsequent anthropometric and physiological characteristics that aid performance [3–9]. It is reasonable to expect that the RAE would be more prominent for younger athletes because of the age differences being relatively larger. During puberty, the RAE is often found to be strongest in this period related to the large variation in physique and anthropometry due to age [10]. After puberty, RAE are often found to subside [11], however, secondary effects may maintain the persistence during senior carrier [12]. Early maturation and success may increase training motivation and thus the sport-specific skills and experience [11, 13]. Further- more, those who perform at highest level in youth sports may be favoured by more competent coaches, more systematic training and better training facilities [14, 15]. Consequences of the RAE may include favouring the physically mature at the expense of those less mature, and in worst case athletes dropout from the sports because of late maturation [11, 16]. RAE is well documented in team sports [2, 17], but there is still less evidence regarding RAE in individual sports [17]. There are, however, some important differences between team sports and individual sports that may influence the occurrence of RAE. The selection that occurs in team sports are not seen to the same extent in individual sports. In individual sports, success is also related to individual performance and not confounded by inter-individual fac- tors like team formations, tactics and positional roles [18]. Performances are further judged on objective data in most individual sports, rather than of a subjective evaluation of an individu- al’s contribution to a team performance. RAE-studies in team sports are most often investigat- ing the advantage/disadvantage for the selected/non-selected players, and this selection are based upon subjective criteria. Track and field give us an opportunity to investigate the advan- tage of being born early in the year based upon actual performance, since objective criteria are the performance indicators. There are few previous studies investigating RAEs in track and field, and as far as we know, only one study from the Scandinavian countries [19]. In addition, the majority of previous studies in track and field have focused on athletes who are 14 years (U15) or older who com- pete at international level [20–26]. The occurrence of RAE has further been investigated more in males than females [19, 20, 23, 27, 28]. Since the Norwegian Athletics Federation has regis- tered seasonal best for each athlete in all events obtained in official competitions from the age of 10 years to senior the last decades, the register will provide as an unique opportunity to PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 2 / 12 Competing interests: The authors have declared that no competing interests exist. investigate RAE in track and filed from child to adult in both sexes. Since biological matura- tion is suggested as an explanation for the RAE, the aim of the present study was to investigate the occurrence of RAE in events where physical capacity is important for the result in all com- petition classes, in both sexes. We aimed to use two different approaches 1) to investigate per- formance differences between athletes born in different quartiles of the year in all competition classes in sprint (60m) and middle-distance running (600m) in both sexes, and 2) to calculate the OR of being among the top-100 athletes in each of the competition class for those born in first vs. last quartile of the year in all competition classes for both sexes. Methods The Norwegian Athletics Federation has the last decades registered all results obtained in offi- cial competitions by Norwegian athletes. The register includes an overview of the best result of each athlete in each event in all competition classes, from 10 years to senior. For each result, competition class and competition date are registered. Birth year is registered for all athletes, and birth date for most of the athletes. The present study was conducted in accordance with the declaration of Helsinki and approved by the Norwegian centre for research data (NSD) (324455) and the Norwegian Athletics Association. Since the data are based on publicly avail- able resources, no informed consent was obtained. Procedures Results available on 60m and 600m in the database from 2011 to 2020 were downloaded on the 1st of January 2021. Only indoor results were included to avoid differences in wind and temperature conditions, and since 60m is the main sprint distance indoor in all competition classes. 60m is an international distance indoors. In Norway, youth athletes are organized within one-year age bands from the age of 10 until the age of 17 years, and in two-years age band for 18 and 19 years old (U20). Senior included results obtained from the age between 20 and 34 years, although athletes can compete in the senior class when 15 years old in Norway. Only the best result for each athlete in each competition classes were included in the analyses. Results obtained without electronic timing and athletes without registered birth date were excluded. Birth months was categorised in birth quartiles; Q1: January-March, Q2: April-June, Q3: July-September, Q4: October-December. Data analyses Data are presented as mean with standard deviation (SD) or as frequencies. Visual inspection confirmed that all data were normally distributed. One-way ANOVA analyses were performed to analyse whether there were differences between results obtained from athletes born in dif- ferent quartiles (Q1, Q2, Q3 and Q4) for each competition class and event, separated by sex. Post hoc analyses were performed with the least significant difference (LSD) test. Frequencies of the top 100 athletes born in each competition class and event for females and males were calculated. Crosstab analyses were performed to assess odds ratio (OR) and 95% confidence interval [CI] of being among the top 100 athletes (yes/no) for those born in first vs. fourth quartile of the year. IBM SPSS Statistics (version 27) was used for all statistical analyses and sta- tistical significance was accepted at p < 0.05. Results Totally, 28 999 results were registered along with birth date for 60m sprint and 600m middle distance running, 15 244 and 13 755 results for females and males respectively. For 60m sprint, PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 3 / 12 a total of 21 419 results were registered (11 512 for females and 9907 for males), and for 600m middle distance running, a total of 7580 results were registered (3732 for females and 3848 for males). Result difference between athletes born in different quartiles of the year There were significant differences in results obtained by athletes born in different quartiles, especially among the youngest athletes. Being born in the first quartile was an advantage regarding performance compared to being born in the last quartile for both events in all com- petition classes from 11 to 13 years (See Fig 1 and Tables 1 and 2 for more specific informa- tion). In sprint, there were better results in first quartile compared to the last quartile in all completion classes until the age of 16 years in females (Table 1 and Fig 1), and in most of the competition classes until U20 in males (Table 2 and Fig 1). In middle-distance running (600m), there were better results in first quartile compared to the last quartile in all competi- tion classes until the competition class 13 year in females (Table 1 and Fig 1), and until 14 years in males (Table 2 and Fig 1). In females, the largest main effect was seen in the competi- tion class 13 years on 60m sprint performance and in the competition class 12 years on 600m middle distance running. There was no significant result difference between quartiles from the age of 17 years in any of the two events. In males, the largest main effect was seen in the com- petition class 13 years for both events. No significant result difference was found at senior level in sprint performance and from the age of 15 years on middle-distance running performance. For an overview of number of athletes born in each quartile see S1 and S2 Tables. OR of being among the top-100 athletes There were higher ORs for being among the top-100 athletes when born in first quartile of the year compared to the last quartile. Overall, the OR of being among the top-100 in one of the competition classes on 60m sprint when born in first quartile compared to last quartile was 2.88 [2.30–3.62] for males and 1.54 [1.26–1.89] for females. Similar, overall OR of being among top-100 athletes on 600m middle-distance running was 2.05 [1.64–2.54] and 1.88 Fig 1. Mean of best indoor running times on 60-meter sprint (A) and 600-meter middle-distance running (B) for different age groups in male and female track and field athletes from 2011–2020. Q1: born in January-March, Q2: born April-June, Q3: born in July-September, Q4: born in October-December. https://doi.org/10.1371/journal.pone.0273472.g001 PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 4 / 12 [1.48–2.38] for males and females, respectively. In females, highest ORs were seen in the com- petition class 10 years in both events, although significantly higher OR was seen until competi- tion class 14 years on 60m and until 13 years on 600m. In males, highest ORs were seen in competition class 14 years on sprint, and in competition class 13 years on 600m. Significant higher OR was seen in most competition classes until senior level on 60m and until the compe- tition class 14 years on 600m (see Table 3 and Fig 2 for more details). Discussion In the present study, we aimed to investigate performance differences between athletes born in various quartiles of the year in each competition class in sprint (60 meter) and middle-distance running (600 meter), for both female and male track and field athletes. Moreover, we wanted to explore the OR of being among the top-100 athletes in each of the competition class for those born in first vs. last quartile of the year. Our main findings were that being born in the first quartile was an advantage for both sprint and middle-distance running success, especially in the youngest age classes. The RAE was present in more of the competition classes in sprint versus middle-distance running, and in more competition classes in males than in females. This is to the best of our knowledge, the first study investigating result performance differ- ences among athletes born in different quartiles of the year in all competition classes from 10 years to senior for both sexes. Our results indicate that chronologically older children and Table 1. Differences in performance on 60m sprint and 600m middle-distance running in females born in different birth quartiles (Q) for each competition class. Data are presented as mean ± standard deviation. Competition classes (years) Q1 Q2 Q3 Q4 Main effect Post hoc analyses LSD January-March April-June July-August Sept.-October 60m sprint (sec) 10 (n = 842) 10.52 ±0.67 10.56 ±0.67 10.80 ±0.78 10.90 ±0.73 F(3,838) = 13.45, p < .001 Q1>Q3, Q4, Q2>Q3, Q4 11 (n = 1367) 10.09 ±0.68 10.14 ±0.63 10.31 ±0.70 10.39 ±0.68 F(3,1363 = 14.64, p < .001 Q1>Q3, Q4, Q2>Q3, Q4 12 (n = 1795) 9.69 ±0.63 9.71 ±0.60 9.93 ±0.70 9.99 ±0.71 F(3,1791) = 22.38, p < .001 Q1>Q3, Q4, Q2>Q3, Q4 13 (n = 2184) 9.39 ±0.60 9.41 ±0.59 9.58 ±0.67 9.67 ±0.69 F(3,2180) = 23.68, p < .001 Q1>Q3, Q4, Q2>Q3, Q4 14 (n = 1828) 9.13 ±0.63 9.11 ±0.55 9.19 ±0.63 9.26 ±0.61 F(3,1824) = 5.19, p < .001 Q1>Q4, Q2>Q4 15 (n = 1300) 8.89 ±0.60 8.88 ±0.51 8.91 ±0.57 9.03 ±0.71 F(3,1296) = 4.02, p = .007 Q1>Q4, Q2>Q4 16 (n = 920) 8.69 ±.050 8.71 ±0.49 8.72 ±0.54 8.82 ±0.52 F(3,916) = 2.90, p = .034 Q1>Q4 17 (n = 604) 8.51 ±0.44 8.56 ± 0.45 8.63 ±0.57 8.56 ±0.44 F(3,600) = 1.55, p = .201 - U20 (n = 508) 8.42 ±0.54 8.42 ±0.43 8.43 ±0.46 8.48 ±0.52 F(3,504) = 0.40, p = .752 - Senior (n = 162) 8.09 ±0.47 8.18 ±0.45 8.23 ±0.61 8.22 ±0.54 F(3,158) = 0.64, p = .592 - 600m middle distance running (min) 10 (n = 347) 2.19.87 ±11.15 2.22.17 ±11.40 2.22.31 ±9.99 2.26.24 ±11.58 F(3,343) = 3.95, p < .009 Q1>Q4, Q2>Q4, Q3>Q4 11 (n = 631) 2.13.96 ±11.60 2.15.64 ±11.39 2.17.37±12.56 2.19.68 ±11.36 F(3,627) = 6.17, p < .001 Q1>Q3, Q4, Q2>Q4 12 (n = 758) 2.07.23 ±11.79 2.08.19 ±10.32 2.11.49±11.55 2.11.87 ±12.58 F(3,754) = 7.66, p < .001 Q1>Q3, Q4, Q2>Q3, Q4 13 (n = 762) 2.01.19 ±10.01 2.01.16 ±10.31 2.03.69±11.64 2.04.71 ±10.44 F(3,758) = 5.12, p < .002 Q1>Q3, Q4, Q2>Q3, Q4 14 (n = 587) 1.57.02 ±11.23 1.56.11 ±8.73 1.56.22 ±8.73 1.58.12 ±9.65 F(3,583) = 1.07, p = .360 - 15 (n = 256) 1.53.27 ±8.97 1.52.31 ±8.59 1.52.71 ±8.54 1.55.56 ±10.42 F(3,252) = 1.28, p = .282 - 16 (n = 155) 1.48.27 ±6.99 1.50.61 ±8.19 1.48.25 ±6.47 1.53.48 ± 8.71 F(3,151) = 3.53, p = .016 Q1>Q4, Q3>Q4 17 (n = 120) 1.50.04 ±10.09 1.46.67 ±8.18 1.46.94 ±8.50 1.51.10 ±7.08 F(3,116) = 1.77, p = .158 - U20 (n = 77) 1.43.71 ±6.86 1.43.81 ±6.04 1.42.75 ±7.17 1.44.03 ±6.67 F(3,73) = 0.13, p = .940 - Senior (n = 39) 1.42.42 ±11.17 1.38.49 ±2.65 1.42.47 ±10.24 1.41.88 ±7.74 F(3,35) = 0.47, p = .706 - U20: 18–19 years, Senior: 20–34 years. > indicate significant better performance for athletes born in quartiles at the left side of the sign compared to those born in quartiles at the right side. https://doi.org/10.1371/journal.pone.0273472.t001 PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 5 / 12 adolescents of both sexes have an advantage in youth track and field both in sprint and mid- dle-distance running. Better performance in both sprint and middle-distance running among athletes born early in the year compared to those born late can probably be explained by the Table 2. Differences in performance on 60m sprint and 600m middle-distance running for males born in different birth quartiles (Q) for each competition class. Data are presented as mean ± standard deviation (SD). Competition classes (years) Q1 Q2 Q3 Q4 Main effect Post hoc analyses LSD January-March April-June July-August Sept.-October 60m (sec) 10 (n = 710) 10.39 ±0.76 10.40 ±0.72 10.51 ±0.84 10.60 ±0.70 F(3,706) = 2.59, p = .052 - 11 (n = 1118) 9.96 ±0.76 10.05 ±0.70 10.15 ±0.79 10.23 ±0.73 F(3,1114) = 6.19, p < .001 Q1>Q3, Q4, Q2>Q4 12 (n = 1406) 9.56 ±0.75 9.71 ±0.75 9.80 ±0.76 9.97 ±0.83 F(3,1402 =) 16.31, p < .001 Q1>2Q, 3Q, 4Q, 2Q>4Q 3Q>4Q 13 (n = 1735) 9.18 ±0.75 9.28 ±0.73 9.44 ±0.74 9.66 ±0.73 F(3,1731) = 31.21, p < .001 Q1>Q2, Q3, Q4, Q2>Q3, Q4, Q3>Q4 14 (n = 1335) 8.57 ±0.66 8.69 ±0.62 8.83 ±0.71 8.98 ±0.67 F(3,1331) = 22.81, p < .001 Q1>Q2, Q3, Q4, Q2>Q3, Q4, Q3>Q4 15 (n = 1081) 8.12 ±0.46 8.24 ±0.55 8.28 ±0.56 8.41 ±0.60 F(3,1077) = 11.49, p < .001 Q1>Q2, Q3, Q4, Q2>Q4, Q3>Q4 16 (n = 880) 7.93 ±0.44 7.94 ±0.43 7.99 ±0.48 8.03 ±0.47 F(3,876) = 2.15, p = .092 - 17 (n = 645) 7.68 ±0.34 7.71 ±0.33 7.72 ±0.32 7.81 ±0.36 F(3,641) = 4.29, p = .005 Q1>Q4, Q2>Q4, Q3>Q4 U20 (n = 634) 7.55 ±0.35 7.58 ±0.38 7.64 ±0.44 7.67 ±0.37 F(3,630) = 3.25, p = .022 Q1>Q3, Q4, Q2>Q4 Senior (n = 363) 7.45 ±0.48 7.49 ±0.47 7.51 ±0.45 7.52 ±0.47 F(3,359) = 0.45, p = .718 - 600m middle distance running (min) 10 (n = 363) 2.14.06 ±11.41 2.13.03 ±10.61 2.17.61 ±13.79 2.21.28 ±13.47 F(3,359) = 7.67, p < .001 Q1>Q3, Q4, Q2>Q3, Q4 11 (n = 628) 2.07.94 ±12.04 2.09.20 ±13.03 2.12.11 ±12.46 2.13.09 ±12.62 F(3,624) = 5.41, p < .001 Q1>Q3, Q4, Q2>Q3, Q4 12 (n = 710) 2.00.92 ±11.50 2.02.26 ±11.74 2.04.17 ±11.39 2.04.59 ±12.29 F(3,706) = 3.72, p < .011 Q1>Q3, Q4 13 (n = 808) 1.53.56 ±10.14 1.55.55 ±10.61 1.57.38 ±11.09 2.00.45 ±11.11 F(3,804) = 14.85, p < .001 Q1>Q3, Q4, Q2>Q4, Q3>Q4 14 (n = 559) 1.45.66 ±8.43 1.47.86 ±9.24 1.50.21 ±10.98 1.51.04 ±9.48 F(3,555) = 9.12, p < .001 Q1>Q2, Q3, Q4 Q2>Q3, Q4 15 (n = 250) 1.37.57 ±8.04 1.38.30 ±6.69 1.39.00 ±6.58 1.41.20 ±9.35 F(3,246) = 2.34, p = .074 - 16 (n = 193) 1.35.02 ± 8.71 1.36.22 ± 7.43 1.34.45 ±6.06 1.35.45 ±6.82 F(3,189) = 0.51, p = .675 - 17 (n = 129) 1.30.58 ±4.16 1.31.63 ±4.45 1.31.12 ±5.44 1.33.56 ±5.26 F(3,125) = 2.06, p = .110 - U20 (n = 128) 1.27.15 ±4.15 1.28.84 ±5.96 1.28.05 ±4.59 1.28.98 ±4.05 F(3,124) = 1.04, p = .379 - Senior (n = 80) 1.27.42 ±9.16 1.31.54 ±12.02 1.28.74 ±9.12 1.26.28 ±4.56 F(3,76) = 1.00, p = .398 - U20: 18–19 years, Senior: 20–34 years> indicate significant better performance for athletes born in quartiles at the left side of the sign compared to those born in quartiles at the right side. https://doi.org/10.1371/journal.pone.0273472.t002 Table 3. On overview over OR 95% [CI] of being among the top-100 athletes and born in Q1 compared to Q4. Females Males Competition class (years) 60m sprint 600m 60m sprint 600m 10 2.89 [1.36–6.13] 5.15 [2.03–13.06] 2.54 [1.18–5.49] 4.26 [1.77–10.29] 11 2.49 [1.29–4.81] 3.50 [1.64–7.497] 2.26 [1.10–4.66] 3.17 [1.42–7.09] 12 1.60 [.85–3.01] 2.16 [1.11–4.179] 7.82 [3.08–19.85] 1.64 [.88–3.06] 13 1.76 [.87–3.59] 2.06 [1.03–4.149] 5.81 [2.62–12.87] 4.97 [2.37–10.43] 14 2.07[1.06–4.05] 1.69 [.83–3.432] 11.27 [3.49–36.45] 3.21 [1.52–6.75] 15 .96 [.53–1.74] 1.78 [.81–3.931] 2.59 [1.23–5.45] 1.50 [.72–3.14] 16 1.19 [.64–2.22] 2.04 [.77–5.393] 1.88 [.93–3.80] 1.68 [.74–3.82] 17 1.01 [.52–1.94] .69 [.19–2.520] 2.28 [1.14–4.57] 3.18 [.92–11.03] U20 (18–19) 1.34 [.70–2.59]  1.62 [.86–3.05] 1.87 [.55–6.32] Senior (20–34) 1.19 [.49–3.03]  1.98 [1.00–3.91]   less than 100 results. https://doi.org/10.1371/journal.pone.0273472.t003 PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 6 / 12 fact that the chronologically oldest athletes in each competition class, on average, are more bio- logically mature and therefore have more developed anthropometric and physical attributes that aid performance [5–9, 29–32]. Indeed, another possible explanation for our findings could also be linked with possible skewed birth date distribution in the greater Norwegian population. Thus, we checked the Medical Birth Registry of Norway (MBRN) (http:// statistikkbank.fhi.no/mfr/) for quarterly distribution of births in Norway between 1996 and 2010, and found that 24.8%, 26.0%, 26.1% and 23.0% were born in Q1-Q4 respectively. Even though the number of births was not totally equal between quartiles, it is highly unlikely that the RAE in the present study can be explained by the distribution of births. Our findings are thus probably linked to the maturational hypothesis. In sprint, the advantage of being born early in the year lasted longer in males than in females. Better results for those born early in the year compared to those born late was seen until competition class 16 years in females, and until competition class U20 in males. In boys, higher OR of being among the top-100 sprinters when born in the first quartile compared to the last quartile was seen in most competition classes and even at senior level. Highest OR among males was found in the competition class 14 years, where it was 11 times more likely to be among the top-100 athletes for those born in the first quartile of the year compared to those born in the last. In girls, higher OR was only seen in the competition classes 10, 11 and 14 years, with the highest OR in the youngest competition class. The above-mentioned sex differ- ences may be explained by the timing of puberty. The timing of puberty could be one explana- tion for the high OR for boys in the age of 12 to 14 years in explosive events like 60m sprint, and also an explanation for high OR in earlier ages (10 and 11 years) in girls. Therefore, we speculate that the competition classes with high OR are partly due to the fact that these age groups have larger differences in physical capacity between athletes that are born early and late in the year. Indeed, girls enter puberty at a younger age compared to boys [33], and the longer occurrence and larger RAE in males may be explained by the later onset of puberty and the more pronounced increase in muscle mass which is an advantage in explosive events [34, 35]. Further, increased motivation, more systematic training and better training facilities as a con- sequence of success in adolescence may explain the persistence of RAE in the present study after puberty among male senior sprinters [11, 13–15]. Fig 2. An overview of odds ratio (OR) of being among the top 100 athletes and born in Q1 compared to Q4 for girls (A) and boys (B). https://doi.org/10.1371/journal.pone.0273472.g002 PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 7 / 12 Kearney and colleagues (2018) found a similar trend among males at highest performance level in sprint in the UK, showing higher OR of being born in the first quartile of the year com- pared to the last in all competition classes (U13, U15, U17, U20 and senior), with the highest OR in U15 (13 and 14 years old). Also, Romann and colleagues (2015) found higher OR among Swiss boys between 8 and 15 years at highest performance level in sprint. Among girls, Kearney and colleagues (2018) showed higher OR in the competition classes U13, U15 and U17, with the highest OR among the youngest (11 and 12 years old). Unlike the present study, they found RAE in girls over 14 years of age. The longer persisting RAE among girls in the study by Kearney and colleagues (2018) may be due to the different organization of competi- tion classes in the UK with two year-bands instead of annual competition classes as in Norway. In middle-distance running, significant better result among athletes born in the first quar- tile of the year compared to those born in last quartile was seen until competition class 14 years in males. In middle-distance running, significant better result among athletes born in the first quartile of the year compared to those born in last quartile was seen until competition class 14 years in males. In girls RAE was present up to the competition class 16 years, with exception of in competition the classes 14 and 15 years. The OR of being among the top-100 middle distance runners when born in first quartile compared to last quartile was in the pres- ent study higher in most competition classes until the age of 14 years in boys. Highest OR was seen in the competition class 13 years, where it was 5 times more likely to be among the top- 100 athletes for those born in the first quartile of the year compared to those born in the last. Kearney and colleagues (2018) found higher OR in male middle distance running in the com- petition classes U13, U15 and U17 in the UK. Highest OR was found in the competition class U15 (13 and 14 years old), which is in line with our findings. However, again we cannot exclude that the different organization of competition classes between Norway and the UK can explain the longer existence of RAE in the UK. In girls, OR progressively decreased from the competition class from 10 to 13 years. Similar, Kearny and colleagues (2018) found higher OR in girls U13 (11 and 12 years old) and U15 (13 and 14 years old), with the highest OR among the youngest. As in sprint highest OR was seen at earlier age in girls than in boys, and may be due to earlier puberty onset in girls as mentioned above [33]. Unlike in male sprint, RAE did not last into adulthood in middle-distance running. The advantage of being born early in the year lasted longer in sprint than in middle-dis- tance running in males, indicating that puberty affects performance in sprint and middle-dis- tance running differently in males. This is most likely due to the different physical demands in the two events. In sprint, force and power are important factors for performance [29, 32], whereas maximal oxygen uptake is important in middle-distance running [36]. The increase in body mass/muscle mass during puberty improves force and power relevant for sprint events. Although the increase in muscle mass associated with growth and maturation facilitates the use of oxygen and thus improves the absolute VO2max (L
min-1), the concomitant increase in body mass results in an almost unchanged VO2max with age during puberty when expressed in relation to body mass (ml
kg-1
min-1) [37, 38]. For girls, increased body mass during puberty is not necessarily an advantage in explosive events or in aerobic events as more body fat mass is accumulated compared to boys [38]. Increased body mass may affect the occurrence of RAE, as those born early in the year, on average, might gain weight before those born later in the year if entering puberty earlier. One causal explanation for the decreasing OR at older age in this study may be because the relative age difference in each age group are lower at higher age. i.e. the relative age difference within an age group are higher for the 10-year-old athletes, than for athletes that are twenty years old. The stronger RAE found, especially in males, in explosive events compared to in endurance events, have previously been shown by others [23, 39]. Stronger RAEs among males at higher PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 8 / 12 level athletes than in females as in the present study are also in accordance with previous stud- ies [18, 20–23, 39]. However, some researchers have proposed strategies to limit the RAE in track and field. For instance, Romann and colleagues (2015) showed a very small or no RAE in sprint among 8 to 15 years old boys when correcting for relative age within each chronological age group, so that children were divided into age groups based on rotating cut-off dates. More recently, Brustio and Boccia [24] investigated an approach where they applied a corrective adjustment procedure in 16- and 17-year-old male and female top-level sprinters. From their retrospective analysis based on longitudinal data, the estimated expected performance changes for each annual group minimized or removed the RAE. Specifically, the results suggested an annual percentage performance difference between males of 2.02% - 2.23% and for female of 1.65% - 1.78%, and that when race results were adjusted based on these findings a more equal birth date distribution were evident compared to uncorrected performance times. Thus, to avoid that child born late in the year always are the youngest in their competition class with lesser chance to succeed, it could be possible to either adjust their performance times based on previous longitudinal data [24] or implement an alternative competition structure where com- petition is based on rolling cut-off birth dates so that children alternate in being the oldest and youngest in their age group [18]. These strategies may reduce drop-out in athletes born late in the year. In conclusion, being born early in the year in events with high demand for specific physical capacities is an advantage in both sexes in most of the youngest competition classes. The age with the highest odds of being among the top-100 athletes corresponds with the age where growth and maturation naturally affect the ability to perform physically in both sexes. Although RAE and maturation are two different constructs, our findings propose that the older individuals in each competition class might have benefitted from natural improvements in performance due to puberty. It is important that both athletes, parents and coaches are aware of the RAE, and focus on mastery and progress in training and competitions regardless of performance level. Identifying talents at an early age is a difficult task thus the main goal for practitioners in track and field events dependent on high physical capacities should be to ensure that all athletes are given a chance to reach their potential, regardless of chronological age or maturational status. This could keep children and adolescence within the sport for lon- ger, which also has implications for lifelong physical activity. It will be of interest in a further study to investigate the occurrence of RAEs in events as hurdle, high jump and shot-put where technique may be of greater importance than physical capacities. Supporting information S1 Table. Number of athletes born in each quartile who ran 60m. (TIF) S2 Table. Number of athletes born in each quartile who ran 600m. (TIF) Acknowledgments Thanks to Trond Engevik for organizing the database and to the The Norwegian Athletics Fed- eration for given the permission to use the data. Author Contributions Conceptualization: Hilde Gundersen, Terje Dalen. PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 9 / 12 Data curation: Hilde Gundersen. Formal analysis: Hilde Gundersen, Anette Harris, Terje Dalen. Methodology: Terje Dalen. Project administration: Hilde Gundersen. Writing – original draft: Hilde Gundersen, Atle Guttormsen, Cecilie Brekke Rygh. Writing – review & editing: Anette Harris, Halvard Grendstad, Morten Kristoffersen, Atle Guttormsen, Terje Dalen, Cecilie Brekke Rygh. References 1. Aubert S, Barnes JD, Abdeta C, Abi Nader P, Adeniyi AF, Aguilar-Farias N, et al. Global Matrix 3.0 Physical Activity Report Card Grades for Children and Youth: Results and Analysis From 49 Countries. J Phys Act Health. 2018; 15(S2):S251–S73. https://doi.org/10.1123/jpah.2018-0472 PMID: 30475137 2. Cobley S, Baker J, Wattie N, McKenna J. Annual age-grouping and athlete development: a meta-analyt- ical review of relative age effects in sport. Sports Med. 2009; 39(3):235–56. https://doi.org/10.2165/ 00007256-200939030-00005 PMID: 19290678 3. Rogol AD. Androgens and puberty. Mol Cell Endocrinol. 2002; 198(1–2):25–9. https://doi.org/10.1016/ s0303-7207(02)00365-9 PMID: 12573811 4. Rogol AD, Roemmich JN, Clark PA. Growth at puberty. J Adolesc Health. 2002; 31(6 Suppl):192–200. 5. Malina RM, Eisenmann JC, Cumming SP, Ribeiro B, Aroso J. Maturity-associated variation in the growth and functional capacities of youth football (soccer) players 13–15 years. Eur J Appl Physiol. 2004; 91(5–6):555–62. https://doi.org/10.1007/s00421-003-0995-z PMID: 14648128 6. Malina RM, Martinho DV, Valente-Dos-Santos J, Coelho ESMJ, Koziel SM. Growth and Maturity Status of Female Soccer Players: A Narrative Review. Int J Environ Res Public Health. 2021; 18(4). https://doi. org/10.3390/ijerph18041448 PMID: 33557121 7. Armstrong N, Welsman JR. Cardiovascular responses to submaximal treadmill running in 11 to 13 year olds. Acta Paediatr. 2002; 91(2):125–31. https://doi.org/10.1080/080352502317285081 PMID: 11951996 8. Geithner CA, Thomis MA, Vanden Eynde B, Maes HH, Loos RJ, Peeters M, et al. Growth in peak aero- bic power during adolescence. Med Sci Sports Exerc. 2004; 36(9):1616–24. https://doi.org/10.1249/01. mss.0000139807.72229.41 PMID: 15354046 9. Gil SM, Badiola A, Bidaurrazaga-Letona I, Zabala-Lili J, Gravina L, Santos-Concejero J, et al. Relation- ship between the relative age effect and anthropometry, maturity and performance in young soccer players. J Sports Sci. 2014; 32(5):479–86. https://doi.org/10.1080/02640414.2013.832355 PMID: 24050650 10. Rubia A, Bjorndal CT, Sanchez-Molina J, Yague JM, Calvo JL, Maroto-Izquierdo S. The relationship between the relative age effect and performance among athletes in World Handball Championships. PLoS One. 2020; 15(3):e0230133. https://doi.org/10.1371/journal.pone.0230133 PMID: 32214322 11. Musch J, Grondin S. Unequal competition as an impediment to personal development: a review of the relative age effect in sport. Developmental Review. 2001; 21(2):147–67. 12. Brustio PR, Lupo C, Ungureanu AN, Frati R, Rainoldi A, Boccia G. The relative age effect is larger in Ital- ian soccer top-level youth categories and smaller in Serie A. PLoS One. 2018; 13(4):e0196253. https:// doi.org/10.1371/journal.pone.0196253 PMID: 29672644 13. Harter S. Effectance motivation reconsidered: Toward a developmental model. Human Development. 1978; 1:34–64. 14. Helsen WF, Starkes JL, Van Winckel J. The influence of relative age on success and dropout in male soccer players. American Journal of Human Biology. 1998; 10(6):791–8. https://doi.org/10.1002/(SICI) 1520-6300(1998)10:6<791::AID-AJHB10>3.0.CO;2-1 PMID: 28561412 15. Pena-Gonzalez I, Fernandez-Fernandez J, Moya-Ramon M, Cervello E. Relative Age Effect, Biological Maturation, and Coaches’ Efficacy Expectations in Young Male Soccer Players. Res Q Exerc Sport. 2018; 89(3):373–9. https://doi.org/10.1080/02701367.2018.1486003 PMID: 30015598 16. Muller L, Gonaus C, Perner C, Muller E, Raschner C. Maturity status influences the relative age effect in national top level youth alpine ski racing and soccer. PLoS One. 2017; 12(7):e0181810. https://doi.org/ 10.1371/journal.pone.0181810 PMID: 28759890 PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 10 / 12 17. Smith KL, Weir PL, Till K, Romann M, Cobley S. Relative Age Effects Across and Within Female Sport Contexts: A Systematic Review and Meta-Analysis. Sports Med. 2018; 48(6):1451–78. https://doi.org/ 10.1007/s40279-018-0890-8 PMID: 29536262 18. Romann M, Cobley S. Relative age effects in athletic sprinting and corrective adjustments as a solution for their removal. PLoS One. 2015; 10(4):e0122988. https://doi.org/10.1371/journal.pone.0122988 PMID: 25844642 19. Jakobsson J, Julin AL, Persson G, Malm C. Darwinian Selection Discriminates Young Athletes: the Rel- ative Age Effect in Relation to Sporting Performance. Sports Med Open. 2021; 7(1):16. https://doi.org/ 10.1186/s40798-021-00300-2 PMID: 33650038 20. Brazo-Sayavera J, Martinez-Valencia MA, Muller L, Andronikos G, Martindale RJJ. Identifying talented track and field athletes: The impact of relative age effect on selection to the Spanish National Athletics Federation training camps. J Sports Sci. 2017; 35(22):2172–8. https://doi.org/10.1080/02640414.2016. 1260151 PMID: 27879175 21. Brazo-Sayavera J, Martinez-Valencia MA, Muller L, Andronikos G, Martindale RJJ. Relative age effects in international age group championships: A study of Spanish track and field athletes. PLoS One. 2018; 13(4):e0196386. https://doi.org/10.1371/journal.pone.0196386 PMID: 29689117 22. Brustio PR, Kearney PE, Lupo C, Ungureanu AN, Mulasso A, Rainoldi A, et al. Relative Age Influences Performance of World-Class Track and Field Athletes Even in the Adulthood. Front Psychol. 2019; 10:1395. https://doi.org/10.3389/fpsyg.2019.01395 PMID: 31275208 23. Hollings SC, Hume PA, Hopkins WG. Relative-age effect on competition outcomes at the World Youth and World Junior Athletics Championships. Eur J Sport Sci. 2014; 14 Suppl 1:S456–61. https://doi.org/ 10.1080/17461391.2012.713007 PMID: 24444241 24. Brustio PR, Boccia G. Corrective procedures remove relative age effect from world-class junior sprint- ers. J Sports Sci. 2021; 39(22):2603–10. https://doi.org/10.1080/02640414.2021.1947618 PMID: 34210248 25. Boccia G, Cardinale M, Brustio PR. Performance progression of elite jumpers: Early performances do not predict later success. Scand J Med Sci Sports. 2021; 31(1):132–9. https://doi.org/10.1111/sms. 13819 PMID: 32881090 26. Boccia G, Cardinale M, Brustio PR. Elite Junior Throwers Unlikely to Remain at the Top Level in the Senior Category. Int J Sports Physiol Perform. 2021; 16(9):1281–7. https://doi.org/10.1123/ijspp.2020- 0699 PMID: 33647881 27. Nakata H, Sakamoto K. Sex differences in relative age effects among Japanese athletes. Percept Mot Skills. 2012; 115(1):179–86. https://doi.org/10.2466/10.05.17.PMS.115.4.179-186 PMID: 23033755 28. Romann M, Fuchslocher J. The need to consider relative age effects in women’s talent development process. Percept Mot Skills. 2014; 118(3):651–62. https://doi.org/10.2466/30.10.PMS.118k24w8 PMID: 25068738 29. Colyer SL, Nagahara R, Takai Y, Salo AIT. The effect of biological maturity status on ground reaction force production during sprinting. Scand J Med Sci Sports. 2020; 30(8):1387–97. https://doi.org/10. 1111/sms.13680 PMID: 32285541 30. Handelsman DJ. Sex differences in athletic performance emerge coinciding with the onset of male puberty. Clin Endocrinol (Oxf). 2017; 87(1):68–72. https://doi.org/10.1111/cen.13350 PMID: 28397355 31. Radnor JM, Oliver JL, Waugh CM, Myer GD, Moore IS, Lloyd RS. The Influence of Growth and Matura- tion on Stretch-Shortening Cycle Function in Youth. Sports Med. 2018; 48(1):57–71. https://doi.org/10. 1007/s40279-017-0785-0 PMID: 28900862 32. Rumpf MC, Cronin JB, Oliver J, Hughes M. Kinematics and Kinetics of Maximum Running Speed in Youth Across Maturity. Pediatr Exerc Sci. 2015; 27(2):277–84. https://doi.org/10.1123/pes.2014-0064 PMID: 25389204 33. Oehme N, Bruserud IS, Madsen A, Juliusson PB. Is puberty starting earlier than before? Tidsskr Nor Laegeforen. 2020; 140(12). 34. Ramos E, Frontera WR, Llopart A, Feliciano D. Muscle strength and hormonal levels in adolescents: gender related differences. Int J Sports Med. 1998; 19(8):526–31. https://doi.org/10.1055/s-2007- 971955 PMID: 9877143 35. Moran JJ, Sandercock GR, Ramirez-Campillo R, Meylan CM, Collison JA, Parry DA. Age-Related Vari- ation in Male Youth Athletes’ Countermovement Jump After Plyometric Training: A Meta-Analysis of Controlled Trials. J Strength Cond Res. 2017; 31(2):552–65. https://doi.org/10.1519/JSC. 0000000000001444 PMID: 28129282 36. Bassett DR Jr., Howley ET. Limiting factors for maximum oxygen uptake and determinants of endur- ance performance. Med Sci Sports Exerc. 2000; 32(1):70–84. https://doi.org/10.1097/00005768- 200001000-00012 PMID: 10647532 PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 11 / 12 37. Armstrong N, Tomkinson G, Ekelund U. Aerobic fitness and its relationship to sport, exercise training and habitual physical activity during youth. Br J Sports Med. 2011; 45(11):849–58. https://doi.org/10. 1136/bjsports-2011-090200 PMID: 21836169 38. Landgraff HW, Riiser A, Lihagen M, Skei M, Leirstein S, Hallen J. Longitudinal changes in maximal oxy- gen uptake in adolescent girls and boys with different training backgrounds. Scand J Med Sci Sports. 2021; 31 Suppl 1:65–72. https://doi.org/10.1111/sms.13765 PMID: 33871085 39. Kearney PE, Hayes PR, Nevill A. Faster, higher, stronger, older: Relative age effects are most influen- tial during the youngest age grade of track and field athletics in the United Kingdom. J Sports Sci. 2018; 36(20):2282–8. https://doi.org/10.1080/02640414.2018.1449093 PMID: 29513142 PLOS ONE RAE in track and field PLOS ONE | https://doi.org/10.1371/journal.pone.0273472 September 6, 2022 12 / 12

Performance in youth track and field is associated with birth quartile. A register-based study among athletes in Norway from 10 years to senior level.

09-06-2022

Gundersen, Hilde,Harris, Anette,Grendstad, Halvard,Kristoffersen, Morten,Guttormsen, Atle,Dalen, Terje,Rygh, Cecilie Brekke

eng

PMC5330462

BASAL A CARDIO A1 TD1 TAU1 A2 TD2 TAU2 MRT A BASAL A CARDIO A1 TD1 TAU1 A2 TD2 TAU2 MRT 97,5% MLSS (s) (s) (ml.kg.min-1) (s) (s) s (ml.kg.min-1) (ml.kg.min-1) (ml.kg.min-1) (s) (s) (ml.kg.min-1) (s) (s) s Swimmer 1 6,29 13,76 26,61 26,85 8,65 1,71 113,19 25,74 35,50 11,48 8,25 20,14 20,00 3,00 2,87 179,65 195,05 23,00 Swimmer 2 5,76 23,19 34,32 12,31 21,62 2,26 112,17 37,02 33,93 20,51 15,55 18,88 9,99 6,02 0,71 197,96 298,36 16,01 Swimmer 3 7,36 16,55 46,05 19,74 14,99 1,07 95,00 1,01 34,73 20,51 7,76 32,95 20,00 7,52 27,52 Swimmer 4 7,04 15,37 43,18 9,94 14,86 24,80 13,50 12,11 34,64 9,15 13,70 0,85 90,00 11,53 22,85 Swimmer 5 5,58 12,87 41,70 14,88 16,14 4,15 94,99 298,09 31,02 13,98 18,17 32,89 14,09 8,26 0,87 135,01 12,61 22,35 Swimmer 6 5,86 13,66 35,31 14,11 17,05 31,16 23,73 5,43 15,98 9,99 13,77 0,84 100,74 257,99 23,76 Swimmer 7 12,74 6,90 28,01 7,84 26,31 34,15 17,79 3,01 21,01 5,00 13,49 1,21 112,02 102,47 18,49 Swimmer 8 7,18 26,51 24,72 12,44 10,67 23,11 7,11 15,04 23,67 6,63 8,57 1,73 195,02 298,05 15,20 Swimmer 9 7,74 12,69 37,28 14,34 20,80 2,49 105,00 58,27 35,14 19,85 11,23 26,99 15,85 17,56 1,28 200,00 299,99 33,41 Swimmer 10 6,27 22,28 40,04 10,00 12,17 1,41 97,26 0,15 22,17 11,03 7,24 36,21 9,12 15,89 25,01 Média 7,2 16,4 35,7 14,2 16,3 2,2 102,9 70,0 30,6 15,9 10,4 26,3 12,0 10,8 1,3 151,3 184,5 22,8 DP 2,1 5,9 7,3 5,5 5,4 1,1 8,4 113,9 5,2 5,3 4,9 7,4 5,3 4,7 0,7 46,9 125,9 5,4 BASAL A CARDIO A1 TD1 TAU1 A2 TD2 TAU2 MRT A BASAL A CARDIO A1 TD1 TAU1 A2 TD2 TAU2 MRT 100% MLSS (s) (s) (ml.kg.min-1) (s) (s) s (ml.kg.min-1) (ml.kg.min-1) (ml.kg.min-1) (s) (s) (ml.kg.min-1) (s) (s) s Swimmer 1 5,61 21,02 36,79 9,99 9,01 2,25 85,07 16,27 19,00 16,14 18,02 26,35 5,00 11,72 3,83 82,49 96,36 16,72 Swimmer 2 5,90 7,90 35,47 20,00 9,91 4,37 93,04 35,87 29,91 20,27 14,00 25,41 6,32 10,78 17,10 Swimmer 3 8,24 13,71 43,62 15,03 14,40 2,60 155,02 25,55 29,43 25,40 5,17 26,21 9,19 11,43 20,62 Swimmer 4 5,21 20,08 48,80 28,35 7,88 36,23 19,94 8,59 31,87 9,80 9,77 19,57 Swimmer 5 5,34 7,88 45,63 15,00 14,95 2,79 80,00 200,00 29,95 11,12 6,71 39,48 9,99 8,64 0,82 165,59 199,60 18,63 Swimmer 6 6,53 19,63 35,74 5,00 10,69 15,69 16,72 15,05 24,39 14,99 3,87 0,84 193,25 299,85 18,86 Swimmer 7 6,17 8,56 36,08 15,61 18,77 34,38 13,56 18,85 27,44 20,00 7,37 0,88 199,25 34,67 27,37 Swimmer 8 5,23 11,11 32,67 4,37 16,64 2,77 145,00 47,14 21,01 7,54 16,75 29,18 16,32 5,93 1,06 200,00 300,00 22,25 Swimmer 9 6,36 22,77 40,13 8,56 13,87 2,60 133,32 28,90 22,43 24,18 20,00 20,41 17,10 7,20 1,49 200,00 300,00 24,30 Swimmer 10 4,88 27,88 54,56 1,92 22,03 23,95 21,52 18,58 31,73 10,48 20,23 1,45 80,00 0,89 30,71 Média 5,95 16,05 40,95 12,38 13,82 2,90 115,24 58,96 26,20 17,64 14,17 28,25 11,92 9,69 1,48 160,08 175,91 21,61 DP 0,97 7,13 7,02 8,07 4,53 0,75 32,98 69,87 6,78 5,72 5,43 5,22 4,92 4,47 1,07 55,20 131,42 4,59 BASAL A CARDIO A1 TD1 TAU1 A2 TD2 TAU2 MRT A A CARDIO A1 TD1 TAU1 A2 TD2 TAU2 MRT 102,5% MLSS (s) (s) (ml.kg.min-1) (s) (s) s (ml.kg.min-1) (ml.kg.min-1) (ml.kg.min-1) (s) (s) (ml.kg.min-1) (s) (s) s Swimmer 1 6,81 10,95 39,70 10,00 16,96 4,49 80,05 72,66 26,96 21,87 17,76 27,51 4,95 8,74 1,61 185,01 78,11 13,69 Swimmer 2 6,36 10,38 36,52 14,93 13,80 3,74 110,00 36,68 28,73 23,67 6,20 21,36 6,84 20,38 27,22 Swimmer 3 6,72 8,89 46,26 14,82 9,46 4,37 105,00 9,67 24,28 19,11 11,01 36,16 12,83 7,59 20,42 Swimmer 4 6,97 25,32 51,36 5,91 18,61 24,52 17,85 16,73 36,78 7,10 13,03 20,13 Swimmer 5 5,19 8,50 47,12 5,00 16,68 2,91 87,57 128,61 21,68 27,85 17,79 23,40 10,02 1,42 11,44 Swimmer 6 6,74 22,27 37,05 19,20 20,18 1,90 150,00 7,39 39,38 14,98 16,41 29,58 14,82 8,98 2,17 190,00 299,00 23,80 Swimmer 7 6,52 17,87 35,25 24,63 18,18 7,20 120,00 57,94 42,81 11,71 10,49 34,62 14,99 11,72 1,08 87,99 69,46 26,71 Swimmer 8 5,15 9,49 37,75 4,99 23,33 4,46 82,48 245,09 28,32 8,61 25,82 34,03 19,99 4,23 24,22 Swimmer 9 7,76 13,64 39,24 5,00 19,16 6,10 92,86 96,34 24,16 22,07 9,34 24,57 7,17 12,74 0,76 199,99 24,18 19,91 Swimmer 10 5,98 23,25 42,37 9,99 3,32 5,14 80,23 144,22 13,31 20,50 17,40 30,18 11,84 7,50 1,82 172,80 64,80 19,34 Média 6,42 15,06 41,26 11,45 15,97 4,48 100,91 88,73 27,42 18,82 14,90 29,82 11,06 9,63 1,49 167,16 107,11 20,69 DP 0,80 6,54 5,35 6,81 5,80 1,59 23,23 75,79 8,46 5,76 5,67 5,53 4,72 5,25 0,57 45,33 109,26 5,15 10min TESTE ÚLTIMOS 20MIN - PÓS 10min 10min

Oxygen uptake kinetics and energy system's contribution around maximal lactate steady state swimming intensity.

02-28-2017

Pelarigo, Jailton Gregório,Machado, Leandro,Fernandes, Ricardo Jorge,Greco, Camila Coelho,Vilas-Boas, João Paulo

eng

PMC6048134

1 SCiEntiFiC RepoRTS | (2018) 8:10742 | DOI:10.1038/s41598-018-29050-0 www.nature.com/scientificreports The exercise sex gap and the impact of the estrous cycle on exercise performance in mice Aderbal S. Aguiar Jr 1,2, Ana Elisa Speck1,2, Inês M. Amaral1, Paula M. Canas1 & Rodrigo A. Cunha 1,3 Exercise physiology is different in males and females. Females are poorly studied due to the complexity of the estrous cycle and this bias has created an exercise sex gap. Here, we evaluated the impact of sexual dimorphism and of the estrous cycle on muscle strength and running power of C57BL/6 mice. Like men, male mice were stronger and more powerful than females. Exercise-induced increase of O2 consumption (VO2) and CO2 production (VCO2) were equal between sexes, indicating that running economy was higher in males. Thermoregulation was also more efficient in males. In females, proestrus increased exercise VO2 and VCO2 at low running speeds (30–35% female VO2max) and estrus worsened thermoregulation. These differences translated into different absolute and relative workloads on the treadmill, even at equal submaximal VO2 and belt speeds. In summary, our results demonstrate the better muscle strength, running power and economy, and exercise-induced thermoregulation of males compared to females. Proestrus and estrus still undermined the running economy and exercise-induced thermoregulation of females, respectively. These results demonstrate an important exercise sex gap in mice. The importance of differences between sexes/genders is recognized in biology and medicine. Sex describes bio- logical differences, while gender includes social, cultural and economic aspects1. The historical gender differences in motivation/opportunity to practice physical activity (including physical exercise and training) limited the best women exercise/sport performance, a phenomenon known as exercise gender gap in humans2. For instance, women are more prone to physical inactivity3, a risk factor for many diseases4,5. The historical evolution of exer- cise gender gap in modern Olympic Games (World Record and 10 best performances) also reveals a systemati- cally lower sport performance of females compared to males; nowadays, the differences varies between 10.7% for running and 36.8% for weightlifting2. The exercise gender gap is greatest in sports that require running economy, muscle strength, and exercise power2. Running economy is the energy demand for a submaximal running speed6, higher in men7,8 but it is unknown if this sex difference is also present in laboratory animals. A review of ≈1400 manuscripts involving more than 6 million people revealed an under-representation of women in studies of exercise and sports (35–37%)9. However, sex is a major determinant of exercise performance through the impact of anthropometry (height, weight, body fat, and muscle mass), aerobic power and anaerobic threshold, besides genetic and hormonal factors2–4,10. The minor representation of females also translates into less knowledge about the biology of exercise in this sex. So far, the main features of sexual dimorphism important for exercise described in rodents are differences in skeletal muscle kinetics and fiber-type composition10 and energy metabolism11,12. In fact, the biological mechanisms underlying the benefits of exercise were investigated in numerous animal studies in a laboratory setting, with a strong tendency to only use males probably to avoid deal- ing with the possible influence of the menstrual/estrous cycle9,13. Exercise-induced thermoregulation, submaxi- mal and maximal VO2 and VCO2, and running economy are gold physiological indexes for exercise, but have never been studied in females at different phases of the estrous cycle. In humans, the exercise sex gap is greatest in sports that require running economy, strength, and power. Similarly, we investigated the role of sex and estrous cycle in maximum (and submaximal) muscle strength and running power/economy of mice. We also evaluated exercise-induced thermoregulation. This knowledge is 1Purines at CNC-Center for Neuroscience and Cell Biology, University of Coimbra, 3004-517, Coimbra, Portugal. 2Research Group on Biology of Exercise, Department of Health Sciences, UFSC-Federal University of Santa Catarina, Araranguá, SC, 88905-120, Brazil. 3FMUC – Faculty of Medicine, University of Coimbra, 3004-504, Coimbra, Portugal. Correspondence and requests for materials should be addressed to A.S.A. (email: aderbal.aguiar@ufsc.br) Received: 22 January 2018 Accepted: 21 June 2018 Published: xx xx xxxx OPEN www.nature.com/scientificreports/ 2 SCiEntiFiC RepoRTS | (2018) 8:10742 | DOI:10.1038/s41598-018-29050-0 essential to advance the knowledge of exercise physiology in female sex. We will demonstrate that the simple extrapolation of male knowledge is not correct. Results Mouse morphology and the estrous cycle. The body mass of males was 22.7 ± 1.1% higher than females (F4,51 = 12.6, P < 0.05, Fig. 1A). Rodent tails are important for regulating body temperature14–16. The length of the tail of the males (6.5 ± 0.2 cm) was shorter than females (8.1 ± 0.3 cm, t27 = 4.5, P < 0.05). These parameters were independent of the estrous cycle, which was devoid of effects on body weight (F3,43 = 0.1, P > 0.05; Fig. 1A) and tail length (F4.51 = 0.4, P > 0.05; data not shown). The prominent estrous cycle was estrus (H2 = 8.1, P < 0.05, Fig. 1B), where vaginal smears were marked by clusters of cornified squamous epithelial cells (Fig. 1E). The vag- inal smears also allowed the morphological identification of diestrus, proestrus and metestrus, as exemplified in Fig. 1C,D and F, respectively. Male are stronger and more powerful than females. Figure 2 shows the basal motor behavior and ergometric performance of male and female mice. The open field test did not reveal significant differences in locomotion (F4,53 = 0.39, P > 0.05; Fig. 2A), average (F4,53 = 0.38, P > 0.05; Fig. 2B) and maximum speed of males and females, independently of their estrous cycle (F4,53 = 0.43, P > 0.05; Fig. 2B). Absolute exercise performance of females was curtailed in relation to males, being 27.2 ± 1.1% (F4,32 = 14.2, P < 0.05; Fig. 2C) and 40.5 ± 0.9% lower (F4,32 = 9.9, P < 0.05; Fig. 2F) in the absolute grip strength and treadmill power test, respectively. Moreover, the absolute exercise performance of females was independent of the estrous cycle in the two tests (grip strength F3,27 = 0.27, P > 0.05; Fig. 2C) (treadmill power test F3,27 = 0.19, P > 0.05; Fig. 2F). Although the absolute exercise performance of males was higher, the submaximal comparisons indicated a different conclusion. The ergometric test applied progressive running speeds for males and females through serial acceleration (F21,310 = 3.2, P < 0.05; Fig. 2E). The treadmill running power in males and females was statistically similar up to 15 m/min (F21,310 = 3.2, P > 0.05; Fig. 2E), when the relative intensity was 50 ± 3.7% of the maximum power for females, and 35 ± 3.9%% for males. The lower running power of females appeared at speeds 18 → 30 m/ min (F28,252 = 18.1, P < 0.05; Fig. 2E, gray area). At 30 m/min, the maximum overload of females (100 ± 5.7%) corresponded to a relative overload of males (71 ± 2.2% of maximum). Males reached maximum overload at speeds 39 → 42 m/min (Fig. 2E). The normalization of exercise performance by body mass eliminates sexual dimorphism. We then normalized the exercise performance by the body mass. This transformation eliminated the sex differences for muscle strength (F4,32 = 0.78, P > 0.05; Fig. 2D) and running power at speeds 15 → 30 m/min (F4,32 = 0.63, P > 0.05; Fig. 2G). Figure 1. Impact of sexual dimorphism on body mass (A) and analysis of the estrous cycle based on a morphological analysis of vaginal smears (C–F) that revealed that the prominent estrous cycle was estrus (B). Values are expressed as mean ± standard error of the mean (SEM). N = 8–10 animals/group. *P < 0.05 vs. male (ANOVA, Bonferroni post hoc test). @P < 0.05 (Kruskal-Wallis test). www.nature.com/scientificreports/ 3 SCiEntiFiC RepoRTS | (2018) 8:10742 | DOI:10.1038/s41598-018-29050-0 Males show a better running economy. There were no differences in V O2 and V CO2 kinetics between sexes. The progressive running speeds of the ergospirometry increased the V O2 (F7,126 = 2.8, P < 0.05; Fig. 3A) and VCO2 (F7,126 = 2.4, P < 0.05; Fig. 3E) of males and females at all comparative intensities (9 → 30 m/min, Fig. 3A and E). Thus, the higher submaximal running power developed at speeds 18 → 30 m/min, associated to the same submaximal V O2, showed a better running economy in males compared to females at (Fig. 3A and E, gray area). Importantly, males ran up to higher speeds (33 → 42 m/min; Fig. 2E), which resulted in a higher V O2 (F4,33 = 2.7, P < 0.05; Fig. 3D) and V CO2 (F4,33 = 2.7, P < 0.05; Fig. 3G), but not V O2max in relation to females. Proestrus increases VO2 and VCO2 during submaximal exercise testing. The submaximal V O2 (F18,192 = 2.5, P < 0.05; Fig. 3B) and V CO2 (F18,192 = 2.3, P < 0.05; Fig. 3C) of females during proestrus were signif- icantly larger at lower exercise intensities (30–35% V O2max females. We also detected these differences in total V O2 (F3,35 = 3.8, P < 0.05; Fig. 3D) and V CO2 (F3,35 = 3.2, P < 0.05; Fig. 3G) for females at proestrus during these low exercise intensities (30–35% V O2max females). The higher intensities (30–100% VO2max females) presented similar kinetics for VO2 and VCO2 in the different phases of the estrous cycle. Exercise-induced thermoregulation is less effective in estrus females. Thermoregulation requires the dissipation of heat produced during exercise. Exercise increased the heat production of males and females (F7,126 = 264, P < 0.05; Fig. 4A), without influence of the estrous cycle (F7,94 = 0.32, P > 0.05; Fig. 4B). Environment temperature and humidity did not interfere in the thermography results, since they were similar before and after the exercise test session (Fig. 4C). The thermal image (Fig. 4D) shows a female at rest, with the body and tail heated after a maximum exercise test (Fig. 4E). Resting body and tail infrared temperatures did not differ between sexes or in females at different phases of the estrous cycle (body, F4,19 = 0.53, P > 0.05; Fig. 4F) (tail, F4,19 = 2.01, P > 0.05; Fig. 4G). The maximum exercise was not enough to heat the body of males and females on metestrus, diestrus and proestrus cycle (F4,43 = 3.4, P < 0.05; Fig. 4H). Moreover, all males and females (all cycles) presented significant tail warm up after maximal exercise (F4,43 = 2.8, P < 0.05). The temperature scores (Fig. 4H and I) reinforced the prominent exercise-induced hyperthermia of females at estrus. Estrus female body heating was larger than that of males and females in other cycles (F4,43 = 3.3, P < 0.05; Fig. 4H). The tail warming of estrus females was superior to males and females at metestrus after exercise (F4,43 = 2.3, P < 0.05; Fig. 4H). Figure 2. Motor and ergometric data. Sex and estrous cycle did not influence the basal locomotion (A) and speed (B) of mice. Males were stronger (C) and more powerful on the treadmill (E,F) than females, regardless of the phase of the estrous cycle. The normalization of the performance per body mass dissipated the sexual dimorphism (D and G). Values are expressed as mean ± standard error of the mean (SEM). N = 8–10 animals/group. *P < 0.05 vs. male (ANOVA, Bonferroni post hoc test). www.nature.com/scientificreports/ 4 SCiEntiFiC RepoRTS | (2018) 8:10742 | DOI:10.1038/s41598-018-29050-0 Discussion Sex matters. Sexual dimorphism and the estrous cycle influenced exercise performance and metabolism of mice implying that these factors should be considered in experimental designs and data interpretation involving exercise biology. We showed that males were stronger and more powerful than females at moderate-high intensi- ties of exercise, when evaluating strength and running. Since submaximal and maximum overloads of exercise were different for males and females, but submaximal VO2 and VCO2 were similar, this means that the running economy of females was lower than that of males. The estrous cycle did not influence muscle strength, but under- mined the running economy and exercise-induced thermoregulation. Size matters. The sex-related exercise differences disappeared after normalization of exercise performance by size (body mass). This had already been described for muscle strength17–19, but not for running power and economy. However, body size and muscle strength are well-known secondary sexual characteristics, influenced primarily by the anabolic action of the hormone testosterone, a major determinant of sexual dimorphism20. Skeletal muscle mass and strength are lower in females19,21. Likewise, normalization of exercise performance by specific muscle mass (rather than body mass) makes sexual dimorphism disappear19,21. Male mammals are larger, with larger cross-sectional muscle area8,10. Several studies also showed that muscle length (and the length of the long bones) is also higher in male mammals, important for greater tetanic strength of the anterior mas- seter muscle8. Larger levers determine higher torques and muscle strength. Sex is also important for muscle fiber-type composition, especially the myosin IIB gene (fast muscle fiber)10. Evidence shows threefold more IIB muscle fibers in the masseter of male mice8,22. In addition, testosterone signals hypertrophy in this musculature20. Conversely, estrogen decreases muscle contractile force in female mice23,24. Thus, muscle strength and running power depends on size and sex: males have large muscles and bones, responsible for great muscle strength; this difference is further amplified by the anabolic effects of testosterone, resulting in larger muscle strength, speed and power. The testosterone also seems to influence running endurance, but not the running economy. Castration of mouse testicles deplete blood testosterone and impair running wheel endurance (10–30% males with intact gonads)25, a model of submaximal physical activity. Testosterone replacement completely reversed this impair- ment25. The antiandrogen Flutamide decreases the treadmill endurance of rats, but does not change VO2max and running economy26. Here, the exercise-induced submaximal VO2 and VCO2 up to V O2max were similar between sexes, as previously described11,12,27. Only one study demonstrated increased female submaximal V O2 during treadmill test, which further reinforces the hypothesis of females’ worst running economy27. These testosterone evidences support the best physical performance (power and endurance) of running male mice, but not the best running economy. Figure 3. Respiratory gases during an incremental test. Running similarly increased general O2 consumption (V O2, A,B) and CO2 production ( VCO2, E,F) in males and females at different speeds up to 33 m/min. V O2max was similar between the sexes (C) total VO2 (D) and V CO2 (G) was only higher in males due to higher running speeds. Proestrus increased submaximal VO2 (B and D) and VCO2 (G) at lighter intensities of ergospirometry (30–35% VO2max). Values are expressed as mean ± standard error of mean (SEM). N = 8–10 animals/group. *P < 0.05 vs. male, #P < 0.05 vs. females (ANOVA, Bonferroni post hoc test). www.nature.com/scientificreports/ 5 SCiEntiFiC RepoRTS | (2018) 8:10742 | DOI:10.1038/s41598-018-29050-0 On the other hand, estrogen seems to influence VO2 and possibly the running economy of mouse. Similar to our results, submaximal VO2 was higher in female rats during the estrogen-dominant proestrus at low treadmill speeds 5–12 m/min (6° grade, without acceleration)17, which may be considered as a low intensity exercise. We also found these differences at near speeds 9–12 m/min. A possibility is the effect of estradiol in the lung gas-exchange surface area (GSA). VO2 is directly proportional to GSA28,29; which increases during proestrus with high estradiol levels28,29. Estradiol also increases lung’s GSA and V O2 in ovariectomized rats29. Our results suggest that estrogen can increase V O2 during exercise, and worsen the running economy, especially at proestrus. Exercise-induced hyperthermia is a biological response due to greater muscle activation, mitochondrial uncoupling and proton leak30. We now report that sex and the estrous cycle do not modify the calories consumed by exercise, another important variable for running economy; however, our results showed that male thermoreg- ulation was more efficient, since the infrared dissipation of males was more effective. Literature suggests two important points for mouse thermoregulation: body surface area (BSA) and tail dry heat loss. BSA is estimated by the Meeh’s formula (BSA = body weight0.667)25. The greater body mass of males assists in better heat dissipation during/after exercise. Moreover, tail size seems to be related to thermal stress14–16,31, with animals that live in warm environments having longer tails15,32. Female tails, even longer, warmed up more during exercise than that of males. The tail length of C57BL/6 female mice was similar to that described in female BALB/c mice15. Thus, a longer tail length in female mice is suggestive of a required adaptation to compensate for their lower body mass (and area). Sanchez-Alavez33 demonstrated that body warming during exercise was higher in female mice at estrous. We saw it in the tails. Progesterone promotes heat conservation and higher body temperatures at rest34,35. Bilateral ovariectomy eliminated this estrous-associated change14,33. We suggest that this may apply to body temperature of running female mice during estrus, characterized by high progesterone levels. Thus, sex seems to be a crucial factor also for the exercise-induced thermoregulation of mice. Some of our results are similar to those reported in humans, since the physical performance of women is gen- erally lower than in men, in accordance with the exercise gender gap2,36,37. The woman’s menstrual cycle is divided into three phases: follicular, ovulation and luteal. The follicular phase can be divided into initial and late, corre- sponding to metestrus and diestrus, respectively. Ovulation corresponds to proestrus, and the luteal phase to estrus. The woman’s follicular and luteal menstrual cycle does not seem to influence muscle strength, power, and V O2 1,38–40. Human studies still allow evaluating rate of perceived effort (RPE), which also does not differ in the different menstrual phases38,41. However, the differences we found are close to ovulation, virtually impossible to Figure 4. Exercise heat production and dissipation, or thermoregulation. Upon exercise, male and female mice consumed similar calories during the incremental test (A) without any evident impact of the estrous cycle. (B) Experiments were conducted in a controlled temperature and humidity environment. (C) The thermal IR image shows an evident tail heating after the maximum exercise (or recovery time, REC, E) in relation to rest. (D) The body and tail temperature was not different at rest (F and G, respectively). Exercise warmed the body of females at estrus (H) and the tails of all groups of mice (I). Again, female tail heating was larger at estrous (I). Values are expressed as mean ± standard error of the mean (SEM). N = 8–10 animals/group. *P < 0.05 (ANOVA, Bonferroni post hoc test). www.nature.com/scientificreports/ 6 SCiEntiFiC RepoRTS | (2018) 8:10742 | DOI:10.1038/s41598-018-29050-0 evaluate in women. We demonstrated that the mouse proestrus (or human ovulation) increased VO2 and heat production at light exercise. In summary, our results highlight differences in exercise performance and metabolism between male and female mice. Sex influences size, which appear to be the main factor for mice exercise sex gap. Mouse sexual dimorphism also influenced exercise workload, but not V O2 and VCO2, implying a finest running economy in males. Males also presented better thermoregulation after exercise. The estrous cycle played a subtle role in mouse physical performance: proestrus impaired running economy and estrus impaired exercise heat loss. This implies that the impact of the estrous cycle on the performance of females should not be considered a limiting factor for their use in experimental designs. In fact, size is the main factor that should be considered in the construction of experimental designs involving exercising male and female mice. For running, a light-intensity exercise seems similar between the sexes (except proestrus), but the performance of females at moderate-intensity running cor- responded to the performance of males at low-moderate intensity; the performance of females at high-intensity running corresponded to the performance of males at moderate-high for males, and male high-intensity running was supra-maximal for females. Failure to consider these differences by measuring only the running speed, as done in most studies, introduces an error to compare performance between sexes. These results are of particular interest to counteract the underrepresentation of females in exercise experimental designs. Methods Animals. Male and female C57BL/6 mice (10–12 weeks old) were obtained from Charles River (Barcelona, Spain). Mice were housed under controlled environment (12 h light-dark cycle, lights on at 7:00 AM, and room temperature of 21 ± 1 °C) with ad libitum access to food and water. Animals were housed and handled accord- ing to European Union guidelines and the study was approved by the Ethical Committee of the Center for Neuroscience and Cell Biology (University of Coimbra). The animals were accustomed to the treadmill for 3 days. The open field or grip strength test was performed on the 4th day in independent groups of animals. Ergospirometry was performed on the 5th day. All tests were carried out between 9:00 and 17:00 hours in a sound-attenuated and temperature controlled observation room under low-intensity light (≈10 lux), where mice had been habituated for at least 1 hour. The apparatuses were cleaned with 10% ethanol between animals. Within the time window of the tests, we did not record any significant impact of the time of day (morning vs. afternoon) on the treadmill vertical power, VO2max and temperature of the tail at rest in either males or females (data not showed). Vaginal cytology. We evaluated the estrous cycle immediately after the behavioral and exercise experiments, through 4–5 consecutive vaginal lavages (with 40–50 μL of distillated H2O) then mounted on gelatinized slides (76 × 26 mm). These procedures lasted no more than 3–5 minutes, and there were no major temporal delays between behavioral experiments and fluid collection for vaginal cytology. The vaginal smear were desiccated at room temperature and covered with 0.1% crystal violet for 1 min, then twice washed with 1 mL H2O and desiccated at room temperature. The slides were mounted with Eukitt medium (Sigma-Aldrich) and evaluated under an optical microscope at 1x, 5x and 20x (Zeiss Axio Imager 2). The char- acterization of the estrous cycle was performed according to literature20,42. Females were categorized for initial (metestrus) or late (diestrus) follicular phase, ovulation (proestrus), or luteal phase (estrus)20,42. Open field. The exploration of an open field (38 × 38 cm) was analyzed for 15 min using the ANY-maze™ video tracking system (Stoelting Co.)41. Grip strength. The animal was hung with its forepaws to the central position of a 300 g metal grid and the grip strength was determined as the weight pushed (in grams)41. The computed result was the average of 3 trials, expressed in kgf. Ergospirometry. Mice were accustomed with a single-lane treadmill (Panlab LE8710, Harvard apparatus) for 3 consecutive days (speed 15 cm/s, 10 min, slope 8.7%, 0.2 mA), with 24 h interval between each habituation session. The ergospirometry test was carried out on 5th day, 48 hours after the last habituation session. The incremental protocol started at 15 cm/s with an increment of 5 cm/s every 2 min, with a constant inclination of 8.7% (5° for the LE8710 model). The exercise test lasted until running exhaustion, defined by the inability of the animal to leave the electrical grid for 5 seconds43,44. We estimated the power output for treadmill running based on a standard conversion of the vertical work, body weight and running speed45,46. Power is the 1st derivative of work relative to time (run time at each stage). Oxygen uptake ( VO2) and carbon dioxide production ( VCO2) were estimated during treadmill running in a metabolic chamber47 (Gas Analyzer ML206, 23 × 5 × 5 cm, AD Instruments, Harvard) coupled to treadmill. The animals remained in the chamber for 15 min prior to exercise testing. Atmospheric air (≈21% O2, ≈0.03% CO2) was renewed at a rate 120 mL/min, using the same sampling rate for the LASER oxygen sensor (Oxigraf X2004, resolution 0.01%) and infrared carbon dioxide sensor (Servomex Model 15050, resolution 0.1%). Heat (calories) was estimated according to the equations of Lusk48. Thermal imaging. An infrared (IR) camera (FLiR C2, emissivity of 0.95, FLiR Systems) placed overtop (25 cm height) of a plastic tube (25 cm diameter) was used to acquire a static dorsal thermal image49. IR images were taken immediately before and after exercise tests, namely at rest (Fig. 4D) and recovery (REC, Fig. 4E) peri- ods, respectively. IR images were analyzed with FLiR Tools software (Flir, Boston). www.nature.com/scientificreports/ 7 SCiEntiFiC RepoRTS | (2018) 8:10742 | DOI:10.1038/s41598-018-29050-0 Tail length. The FLiR C2 camera also captures digital pictures (640 × 480 pixels) that were loaded and cali- brated (plastic tube, 25 cm diameter) in the ImageJ software (v1.51j8, NIH, USA) for tail length measurement of live animals (ImageJ software). Statistics. Data are presented as mean ± Standard Error of the Mean (SEM). A test for normality was per- formed by Kolmogorov–Smirnov test. For each test, the experimental unit was an individual animal. The fre- quency of the estrous cycle was assessed using the Kruskal-Wallis test. The role of sex and estrous cycle in the dependent variables body mass, open field, grip strength and vertical power, V O2 and VCO2, and body and tail temperature was evaluated using on-way ANOVA. The repeated measures of ANOVA were performed to evaluate the effect of different treadmill speeds, sex and estrous cycle on the vertical power, V O2 and VCO2, and heat. The Bonferroni post hoc test was applied for significant F values. The accepted level of significance was p < 0.05. Statistics were performed using Dell Statistica (data analysis software system), version 13. Data availability. The datasets generated and analyzed during the current study are available from the cor- responding author on reasonable request. References 1. Oertelt-Prigione, S., Gohlke, B. O., Dunkel, M., Preissner, R. & Regitz-Zagrosek, V. GenderMedDB: an interactive database of sex and gender-specific medical literature. Biol Sex Differ 5, 7, https://doi.org/10.1186/2042-6410-5-7 (2014). 2. Thibault, V. et al. Women and Men in Sport Performance: The Gender Gap has not Evolved since 1983. J Sports Sci Med 9, 214–223 (2010). 3. Althoff, T. et al. Large-scale physical activity data reveal worldwide activity inequality. Nature 547, 336–339, https://doi.org/10.1038/ nature23018 (2017). 4. Booth, F. W., Roberts, C. K., Thyfault, J. P., Ruegsegger, G. N. & Toedebusch, R. G. Role of Inactivity in Chronic Diseases: Evolutionary Insight and Pathophysiological Mechanisms. Physiol Rev 97, 1351–1402, https://doi.org/10.1152/physrev.00019.2016 (2017). 5. Kokkinos, P. Physical activity, health benefits, and mortality risk. ISRN Cardiol 2012, 718789, https://doi.org/10.5402/2012/718789 (2012). 6. Saunders, P. U., Pyne, D. B., Telford, R. D. & Hawley, J. A. Factors affecting running economy in trained distance runners. Sports Med 34, 465–485 (2004). 7. Helgerud, J. Maximal oxygen uptake, anaerobic threshold and running economy in women and men with similar performances level in marathons. Eur J Appl Physiol Occup Physiol 68, 155–161 (1994). 8. Daniels, J. T. A physiologist’s view of running economy. Med Sci Sports Exerc 17, 332–338 (1985). 9. Costello, J. T., Bieuzen, F. & Bleakley, C. M. Where are all the female participants in Sports and Exercise Medicine research? Eur J Sport Sci 14, 847–851, https://doi.org/10.1080/17461391.2014.911354 (2014). 10. Haizlip, K. M., Harrison, B. C. & Leinwand, L. A. Sex-based differences in skeletal muscle kinetics and fiber-type composition. Physiology (Bethesda) 30, 30–39, https://doi.org/10.1152/physiol.00024.2014 (2015). 11. Brager, A. J. et al. Homeostatic effects of exercise and sleep on metabolic processes in mice with an overexpressed skeletal muscle clock. Biochimie 132, 161–165, https://doi.org/10.1016/j.biochi.2016.11.014 (2017). 12. Rezende, E. L., Kelly, S. A., Gomes, F. R., Chappell, M. A. & Garland, T. Jr. Effects of size, sex, and voluntary running speeds on costs of locomotion in lines of laboratory mice selectively bred for high wheel-running activity. Physiol Biochem Zool 79, 83–99, https:// doi.org/10.1086/498187 (2006). 13. Bruinvels, G. et al. Sport, exercise and the menstrual cycle: where is the research? Br J Sports Med 51, 487–488, https://doi. org/10.1136/bjsports-2016-096279 (2017). 14. Gordon, C. J. Influence of heating rate on control of heat loss from the tail in mice. Am J Physiol 244, R778–784 (1983). 15. Gordon, C. J. et al. Behaviorally mediated, warm adaptation: a physiological strategy when mice behaviorally thermoregulate. J Therm Biol 44, 41–46, https://doi.org/10.1016/j.jtherbio.2014.06.006 (2014). 16. Serrat, M. A. Allen’s rule revisited: temperature influences bone elongation during a critical period of postnatal development. Anat Rec (Hoboken) 296, 1534–1545, https://doi.org/10.1002/ar.22763 (2013). 17. Daniels, D. W., Tian, Z. & Barton, E. R. Sexual dimorphism of murine masticatory muscle function. Arch Oral Biol 53, 187–192, https://doi.org/10.1016/j.archoralbio.2007.09.006 (2008). 18. McLean, A. C., Valenzuela, N., Fai, S. & Bennett, S. A. Performing vaginal lavage, crystal violet staining, and vaginal cytological evaluation for mouse estrous cycle staging identification. J Vis Exp, e4389, https://doi.org/10.3791/4389 (2012). 19. Ueberschlag-Pitiot, V. et al. Gonad-related factors promote muscle performance gain during postnatal development in male and female mice. Am J Physiol Endocrinol Metab 313, E12–E25, https://doi.org/10.1152/ajpendo.00446.2016 (2017). 20. Bardin, C. W. & Catterall, J. F. Testosterone: a major determinant of extragenital sexual dimorphism. Science 211, 1285–1294 (1981). 21. MacLean, H. E. et al. Impaired skeletal muscle development and function in male, but not female, genomic androgen receptor knockout mice. FASEB J 22, 2676–2689, https://doi.org/10.1096/fj.08-105726 (2008). 22. Deasy, B. M. et al. A role for cell sex in stem cell-mediated skeletal muscle regeneration: female cells have higher muscle regeneration efficiency. J Cell Biol 177, 73–86, https://doi.org/10.1083/jcb.200612094 (2007). 23. Moran, A. L., Warren, G. L. & Lowe, D. A. Removal of ovarian hormones from mature mice detrimentally affects muscle contractile function and myosin structural distribution. J Appl Physiol (1985) 100, 548–559, https://doi.org/10.1152/japplphysiol.01029.2005 (2006). 24. Suzuki, S. & Yamamuro, T. Long-term effects of estrogen on rat skeletal muscle. Exp Neurol 87, 291–299 (1985). 25. Bowen, R. S. et al. Effects of Supraphysiological Doses of Sex Steroids on Wheel Running Activity in Mice. J Steroids Horm Sci 3, 110, https://doi.org/10.4172/2157-7536.1000110 (2012). 26. Georgieva, K. N. et al. The effect of flutamide on the physical working capacity and activity of some of the key enzymes for the energy supply in adult rats. Asian J Androl 19, 444–448, https://doi.org/10.4103/1008-682X.177842 (2017). 27. Molinero, A. et al. Role of muscle IL-6 in gender-specific metabolism in mice. Plos One 12, e0173675, https://doi.org/10.1371/ journal.pone.0173675 (2017). 28. Massaro, G. D., Mortola, J. P. & Massaro, D. Sexual dimorphism in the architecture of the lung’s gas-exchange region. Proc Natl Acad Sci USA 92, 1105–1107 (1995). 29. Massaro, G. D., Mortola, J. P. & Massaro, D. Estrogen modulates the dimensions of the lung’s gas-exchange surface area and alveoli in female rats. Am J Physiol 270, L110–114 (1996). 30. Gaesser, G. A. & Brooks, G. A. Metabolic bases of excess post-exercise oxygen consumption: a review. Med Sci Sports Exerc 16, 29–43 (1984). 31. Conley, K. E. & Porter, W. P. Heat loss regulation: role of appendages and torso in the deer mouse and the white rabbit. J Comp Physiol B 155, 423–431 (1985). www.nature.com/scientificreports/ 8 SCiEntiFiC RepoRTS | (2018) 8:10742 | DOI:10.1038/s41598-018-29050-0 32. Harrison, G. A. The adaptability of mice to high environmental temperatures. J. Exper. Bio. 35, 10 (1958). 33. Sanchez-Alavez, M., Alboni, S. & Conti, B. Sex- and age-specific differences in core body temperature of C57Bl/6 mice. Age (Dordr) 33, 89–99, https://doi.org/10.1007/s11357-010-9164-6 (2011). 34. Charkoudian, N., Hart, E. C. J., Barnes, J. N. & Joyner, M. J. Autonomic control of body temperature and blood pressure: influences of female sex hormones. Clin Auton Res 27, 149–155, https://doi.org/10.1007/s10286-017-0420-z (2017). 35. Opas, E. E., Gentile, M. A., Kimmel, D. B., Rodan, G. A. & Schmidt, A. Estrogenic control of thermoregulation in ERalphaKO and ERbetaKO mice. Maturitas 53, 210–216, https://doi.org/10.1016/j.maturitas.2005.04.006 (2006). 36. Cureton, K. et al. Sex difference in maximal oxygen uptake. Effect of equating haemoglobin concentration. Eur J Appl Physiol Occup Physiol 54, 656–660 (1986). 37. Maldonado-Martin, S., Mujika, I. & Padilla, S. Physiological variables to use in the gender comparison in highly trained runners. J Sports Med Phys Fitness 44, 8–14 (2004). 38. De Souza, M. J., Maguire, M. S., Rubin, K. R. & Maresh, C. M. Effects of menstrual phase and amenorrhea on exercise performance in runners. Med Sci Sports Exerc 22, 575–580 (1990). 39. Grucza, R., Pekkarinen, H., Titov, E. K., Kononoff, A. & Hanninen, O. Influence of the menstrual cycle and oral contraceptives on thermoregulatory responses to exercise in young women. Eur J Appl Physiol Occup Physiol 67, 279–285 (1993). 40. Williams, T. J. & Krahenbuhl, G. S. Menstrual cycle phase and running economy. Med Sci Sports Exerc 29, 1609–1618 (1997). 41. Stephenson, L. A., Kolka, M. A. & Wilkerson, J. E. Perceived exertion and anaerobic threshold during the menstrual cycle. Med Sci Sports Exerc 14, 218–222 (1982). 42. Caligioni, C. S. Assessing reproductive status/stages in mice. Curr Protoc Neurosci Appendix 4, Appendix 4I, https://doi. org/10.1002/0471142301.nsa04is48 (2009). 43. Ayachi, M., Niel, R., Momken, I., Billat, V. L. & Mille-Hamard, L. Validation of a Ramp Running Protocol for Determination of the True VO2max in Mice. Front Physiol 7, 372, https://doi.org/10.3389/fphys.2016.00372 (2016). 44. Lee-Young, R. S. et al. Obesity impairs skeletal muscle AMPK signaling during exercise: role of AMPKa2 in the regulation of exercise capacity in vivo. Int J Obes (Lond) 35, 982–989, https://doi.org/10.1038/ijo.2010.220 (2011). 45. Barbato, J. C. et al. Spectrum of aerobic endurance running performance in eleven inbred strains of rats. J Appl Physiol 85, 530–536 (1998). 46. Workman, J. M. & Armstrong, B. W. Oxygen cost of treadmill walking. J Appl Physiol 18, 798–803(1963). 47. Kemi, O. J., Loennechen, J. P., Wisloff, U. & Ellingsen, O. Intensity-controlled treadmill running in mice: cardiac and skeletal muscle hypertrophy. J Appl Physiol 93, 1301–1309, https://doi.org/10.1152/japplphysiol.00231.2002 (2002). 48. Lusk, G. Animal csalorimetry. Twenty-fourth paper. Analysis of the oxidation of mixtures of carbohydrate and fat. A correction. J Biol Chem 59, 41–42 (1924). 49. Crane, J. D., Mottillo, E. P., Farncombe, T. H., Morrison, K. M. & Steinberg, G. R. A standardized infrared imaging technique that specifically detects UCP1-mediated thermogenesis in vivo. Mol Metab 3, 490–494, https://doi.org/10.1016/j.molmet.2014.04.007 (2014). Acknowledgements The work was supported by Prémio Maratona da Saúde, CAPES-FCT (039/2014), FCT (PTDC/NEU-NMC/4154) and ERDF through Centro 2020 (project CENTRO-01-0145-FEDER-000008:BrainHealth 2020). A.S.A.Jr is a CNPq fellow. We would like to acknowledge Flávio N.F. Reis and Frederico C. Pereira (IBILI - Institute for Biomedical Imaging and Life Sciences, University of Coimbra) for making available the treadmill and gas analyzer. Author Contributions A.S.A. Jr. designed and performed the experiments, prepared the figures, and wrote the manuscript. A.E.S. and I.A. performed the experiments. P.M.C. designed the experiments and wrote the manuscript. R.A.C. designed the experiments and wrote the manuscript. All authors revised the manuscript. Additional Information Competing Interests: The authors declare no competing interests. Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Cre- ative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not per- mitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. © The Author(s) 2018

The exercise sex gap and the impact of the estrous cycle on exercise performance in mice.

07-16-2018

Aguiar, Aderbal S,Speck, Ana Elisa,Amaral, Inês M,Canas, Paula M,Cunha, Rodrigo A

eng

PMC6239296

RESEARCH ARTICLE A minimal power model for human running performance Matthew Mulligan1, Guillaume Adam2, Thorsten EmigID2,3* 1 Claremont McKenna College, W.M. Keck Science Department, Claremont, California, United States of America, 2 Massachusetts Institute of Technology, MultiScale Materials Science for Energy and Environment, Joint MIT-CNRS Laboratory (UMI 3466), Cambridge, Massachusetts, United States of America, 3 Laboratoire de Physique The´orique et Modèles Statistiques, CNRS UMR 8626, Baˆt. 100, Universite´ Paris-Saclay, Orsay cedex, France * emig@mit.edu Abstract Models for human running performances of various complexities and underlying principles have been proposed, often combining data from world record performances and bio-ener- getic facts of human physiology. The purpose of this work is to develop a novel, minimal and universal model for human running performance that employs a relative metabolic power scale. The main component is a self-consistency relation for the time dependent maximal power output. The analytic approach presented here is the first to derive the observed loga- rithmic scaling between world (and other) record running speeds and times from basic prin- ciples of metabolic power supply. Our hypothesis is that various female and male record performances (world, national) and also personal best performances of individual runners for distances from 800m to the marathon are excellently described by this model. Indeed, we confirm this hypothesis with mean errors of (often much) less than 1%. The model defines endurance in a way that demonstrates symmetry between long and short racing events that are separated by a characteristic time scale comparable to the time over which a runner can sustain maximal oxygen uptake. As an application of our model, we derive per- sonalized characteristic race speeds for different durations and distances. Introduction Scientists have been fascinated by trying to explain running performance and to predict its limitations for more than 100 years. A purely descriptive approach was employed by Kennelly as early as 1906 for speeds in racing events of animals and humans. For men running events from 20 yards up to a few hundred miles he found a power law relation between distance d and duration T with T * d9/8 with a relative large error of up to 9% for distances from 100m to 50 miles (and larger errors for shorter and longer distances) [1]. Almost a century ago, in 1925 noted mathematician and physiologist A.V. Hill proposed a power model based on metabolic energy considerations to describe the maximal power output Pmax(T) over a given duration T by a hyperbolic function Pmax(T) = P0 + P1/T with constants P0 and P1 (known as the “running curve”) [2]. Ward-Smith introduced a model, PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 1 / 26 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Mulligan M, Adam G, Emig T (2018) A minimal power model for human running performance. PLoS ONE 13(11): e0206645. https://doi.org/10.1371/journal.pone.0206645 Editor: Barbora Piknova, National Institutes of Health, National institute of Diabetes and Digestive and Kidney Diseases, UNITED STATES Received: July 18, 2018 Accepted: October 16, 2018 Published: November 16, 2018 Copyright: © 2018 Mulligan et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: This work was supported by Centre national de la recherche scientifique, grant EMERGENCE2017 of INP, www.cnrs.fr (T.E.) and Agence nationale de la recherche, grant ANR-11- IDEX-0001-02 (T.E.). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. based on the first law of thermodynamics, to describe performances at Olympic Games from 1960 to 1976 with an average absolute error for the predicted times of 0.86% for distances from 100m to 10,000m [3]. In 1973 the mathematician Keller formulated a purely mechani- cal model that is based on the runner’s equation of motion with a damping term [4]. The propulsive force is connected to the mechanical power utilized for running which is different from the overall metabolic power requirement. In analogy to purely mechanical problems, Keller assumed that the damping is linear in velocity and that the damping coefficient is con- stant over time. The justification for these assumptions is not validated given that a compari- son of his model to world track records from 50yards to 10,000m yields a relative large errors of about 3% for distances larger than 5000m. Furthermore, both Hill’s and Keller’s models predict the existence of a maximal speed that can be sustained for an infinite dura- tion, which is not possible from a physiological point of view and incompatible with data on running records. Similarly, a threshold power has been proposed by Jones et al. in the critical power model [5]. In fact, existing models appear to be unable to explain an important observation that has been made already by Hill in the context of his above mentioned model: The average frac- tional utilization of maximal power (or the average running speed) of world record perfor- mances scales linearly with the logarithm of the duration of the performance [2]. An interesting model that interpolates between fundamental knowledge of human bioenergetics during exercise and actual world record running performance was proposed by Peronnet and Thibault [6, 7]. Their model combines characteristics of energy metabolism, based on Hill’s hyperbolic “running curve” and the dynamics of oxygen uptake. However, the frac- tional utilization of maximal power over a given duration is described in their model by a phenomenological logarithmic term that is based on observations in running records. The latter term accounts for endurance limited sustainability of maximal aerobic power. Cur- rently, this model is most effective in reproducing world record running performances. However, it uses a number of fixed parameters that are assumed to be equal for all world record performances although they have been achieved by different athletes. In fact, many parameters can be different among individuals. For example, running economy, i.e., the energy cost of running at a given velocity, shows substantial inter-individual variation [8]. These variations are observed even among well trained elite runners. Another quantity that is modeled as a constant in Peronnet’s and Thibault’s model is the duration over which max- imal aerobic power (or VO2max) can be maintained during running which they assumed to be 7 minutes. However, direct measurements of oxygen uptake have demonstrated varia- tions of the order of one to two minutes among individuals [9, 10]. From a fundamental per- spective it is desirable to derive a model from basic principles of metabolic power generation and utilization that predicts human performances without additional phenomenological input. This is the objective of the present work. For the development of our model it is instructive to review some facts and experimental observations from exercise physiology. When developing a model that can describe run- ning performances as obtained in world records up to the marathon distance one should realize at what relative intensities these races are performed. All Olympic endurance events require intensities above 85% of VO2max which corresponds to the effort reached approxi- mately in the marathon [11]. When looking at record performances, we can also assume that runner has followed an optimal carbohydrate loading strategy so that the stored amount of glycogen is permitting best possible performance. This is of importance for the half marathon and in particular the marathon distance which is raced predominantly on carbohydrate fuel with an average respiratory gas exchange ratio of close to one for faster runners [12]. A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 2 / 26 An important physiological observation is that the total energy cost of running increases lin- early with the covered distance with no or a very small dependence on the running velocity [13, 14]. Hence the power output changes linearly with speed, with the slope quantifying running economy. It is known that this running economy can vary about 30–40% among individuals [11]. An important observation that is essential for the construction of our model is that run- ning economy usually becomes worse with the duration of a running event. The magnitude of the change in economy increases with duration and intensity. The actual change is probably subject dependent and also influenced by external conditions. We shall see below that this is an important factor in determining race velocities and endurance. This drift in running economy has been quantified in treadmill studies with a change of 4.4% for 40min at 80% VO2max, a change of 6.6% for 60min at 70% VO2max, and a change of 9.5% for 60min at 80% VO2max [15]. An other study found for 60min treadmill running near 80% of VO2max a shift of about 3% in oxygen uptake [16]. Changes in running economy have been also observed during a 5km run at a constant pace eliciting about 80–85% of VO2max with an average increase in oxygen uptake of 3.3% for men and 2.0% for women [17]. The reason for the increase in oxygen uptake and reduction in running economy is unknown. A number of mechanisms have been postulated in the literature but most of them are speculative [12, 18–20], including an increase in oxygen uptake due to neuromuscular fatigue [21]. Without discussing here the various attempts that have been made for explaining this observation, we just conclude that every activated physio- logical system increases its own particular energy consumption with the duration of exercise. Methods A minimal model for running performance In view of the current status of theoretical descriptions of human running performances, it appears useful to construct a minimal and universal model for human running performance that fulfills the following two requirements: 1. Based on basic concepts and observations on metabolic power generation and utilization during running 2. Minimal number of physiological parameters that are not fixed a priori In order to eliminate irrelevant normalization parameters from the model (that would depend on the choice of units for energy, power, etc.), we express our model in terms of rela- tive quantities. We shall base the model on expedited power measured as oxygen uptake per time since this quantity can be measured directly under real conditions by mobile spirometry. This implies a slight time dependence of oxygen uptake during prolonged exercise, even when the power output is constant, due to a change of the respiratory quotient with substrate utiliza- tion [22]. Also, since body weight usually changes during prolonged exercise, we measure power or oxygen uptake always per body weight. While the basal metabolic rate Pb is close to 1.2W/kg [6], its actual value is not required in the following. In fact, in the parameterization of running economy to be employed below, we chose to associate Pb with the power that is obtained by linearly extrapolating the running economy to zero velocity. Hence we neglect the non-linear dependence of the energy cost on sub-running (walking) velocities which causes no problem since our model uses the energy cost of motion only in the linear running regime. In our model there exists a crossover power Pm that we expect to be close to the maximal aerobic power associated with maximal oxygen uptake VO2max which is typically in the range of 75 to 85ml/(kg min) for elite runners [6]. The power Pm should not be confused with the critical or the maximal power that occurs in the 3-parameter critical power model of Morton [23]. A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 3 / 26 We measure power relative to the base value Pb, in units of the aerobic power reserve Pm − Pb that is available to the runner, hence defining the relative running power (or intensity) as p ¼ P To construct our model, we start from the following self-consistency relation PmaxðTÞ þ PsupðTÞ ¼ 1 T Z T 0 PTðtÞdt ; ð4Þ which states that the sum of the nominal average power and the additional supplemental power Psup equals the time average of the instantaneously utilized power. We make the impor- tant conjecture that the instantaneous power utilized at time t equals the maximal power that can be sustained for the remaining time T − t of the event [26], i.e., PTðtÞ ¼ PmaxðT leading to TðvÞ ¼ tc exp vmbetween mean race velocity v and race distance d is not a power law as assumed in some stud- ies [28, 29]. For example, Riegel’s formula corresponds in above notation to τ(δ) = αδ − L, υ(δ) = −(α − 1)δ + L with a constant L and an exponent α close to 1.06. Our model predicts that α = 1 exactly and that the very small deviation from α = 1, observed by Riegel and others, is due to a hierarchy of logarithmic corrections, giving rise to a non-constant L. It is interesting to observe from Eq (13) that the endurance measuring parameter γl or γs is the only quantity which determines the time to distance and velocity to distance relations when time is mea- sured in units of tc and velocity in units of vm. We note that for the comparison of our model to record performances and personal best performances of individual runners, we always use the exact expressions involving the Lambert W-function. Interpretation of supplemental power Psup, and of γl, γs The supplemental power defined in Eq (3) can be expressed relative to the aerobic power reserve Pm − Pb as PsupðTÞ Pm qualitative difference in interpretation. We shall come back to these endurance measures when we discuss personalized characteristic race paces. Estimation of physiological model parameters Our model depends on the four independent parameters vm, tc, γs and γl that characterize a group of runners (for example world record holders) or individual runners. Otherwise our model is universal in the sense that it contains no additional fixed parameters or constants. The four parameters can be estimated from a given set of results (distance and time) from exercise performed at maximal intensity, i.e., races. These sets can be either records, like world records, involving a group of different runners or personal records (best performances) from individual runners. To check the accuracy of our model and to compute the model parameters, we minimize numerically the sum of the squared differences between the actual race time and the one predicted by Eq (11) for all results in a given set. This method will be used to recon- struct individual physiological profiles (running economy and endurance) from race perfor- mances in Application 1 below. Prediction of race times and characteristic paces for given times and distances Once the model parameters for a given set of performance results have been determined, the model can be applied to compute a number of interesting quantities that could guide racing and training of a runner. For example, by comparing the time difference between the actual Fig 1. Definition of endurance for long and short duration, El and Es, respectively, from the duration T(p) over which a relative power p can be sustained. Shown is a typical range of endurances for long and short duration (gray regions, with lower and upper limits for γl and γs) and an example curve that visualizes the definition of El and Es. https://doi.org/10.1371/journal.pone.0206645.g001 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 8 / 26 race time and the model’s prediction for all raced distances, preferred or optimal distances for a runner can be identified. For distances that have not been raced before, or only prior to a newly focused training program, the formula of Eq (11), or its approximative version in Eq (13), can be used to predict racing times. Another application of our model is the estimation of characteristic velocities that corre- spond to a prescribed relative power output ^p, measured in percent of aerobic power reserve that is available over a given duration. Generally, running velocities v in training units depend on the purpose of the training session and hence on duration T or distance d of the workout intervals. Suppose that a runner trains at a relative power ^p. This relative power relates the tar- get power output P(v) to the maximal power above basal power, Pmax(T) − Pb, that can be maintained for the duration T by the relation PðvÞ ¼ ^pðPmaxðTÞ this requirement, and in this section we shall validate its accuracy by comparing it to various record performances. World and other records have been analyzed before and found to follow an approximate power law. However, the exponent of this power law shows variations with gender and dis- tance which renders its universality and general applicability questionable. Also, there is no physiological foundation for a simple power law. In fact, the existence of a crossover velocity vm implies different scaling of performances below and above this velocity due to distinct phys- iological and bio-energetic processes involved. We have analyzed record performances for eight distances, from 1000m to the marathon, for world records (current as of Oct. 2018, 2000, 1990, and 1980), current European records, and current national records (USA, Germany) see Table 1 for male records, and Table 2 for female records. Following the method described in the previous section, we have estimated the parameters of our model for each group of records. The resulting parameters tc, vm, γs, and γl together with the endurances Es and El are summarized in Tables 1 and 2. The mean relative error between our model prediction and the VDOT prediction for the race times for 13 dis- tances between 1000m and the marathon are 0.15%, 0.11%, and 0.18% for VDOT = 40, 60, and 80, respectively. These small errors suggest that the race times predicted by the VDOT model are mutually consistent. This presumably reflects that the times were obtain from a mathemati- cal model that is based on physiological observations made by Daniels among well trained and elite runners. A number of interesting observations can be made from the results: There is a high level of agreement between actual and predicted times with the relative error being larger than 1% only for a single event (Half-marathon, WR 1980) for male records, and four events for female records. The mean of the absolute value of the relative error is always smaller than 1% with the exception of the female WR from 1990 where it is 1.05%. For the male WR a decrease of the absolute value of the relative error from 1980 to today can be observed, indicating an increas- ing optimization towards the maximally possible performance (within current level of technol- ogy and training methods) that is described by our model. Hence, the record times have become more consistent with our model over time which might be also due to an increasing number of attempts to achieve best possible performances. A similar observation is made for the female WR from 1990 to 2000. However, from the 2000 WR some results (Chinese run- ner’s results for 1500m, 3.000m, 5.000m, and 10.000m) have been excluded due to the use of performance-enhancing drugs [30], and the current WR for 1500m and 10.000m are also con- troversial [31]. For the latter two distances our model predicts more than 0.5% slower times than actually raced. It is interesting to observe that our predictions are very sensitive to excep- tional performances for a particular distance compared to the other distances, and hence is able to identify suspicious race results. Due to the women’s shorter history of endurance run- ning, the female world records for 1980 are less consistent than more recent records and hence have been excluded them from our analysis. It also instructive to compare the physiological model parameters obtained from the record performances. For the male records, the obtained values for tc vary between five and six min- utes, which is in very good agreement with laboratory testing [32]. However, for female rec- ords, we observe a larger variation in tc with values around 10min being not unusual. However, in cases with such long tc the crossover velocity vm is reduced proportionally. The endurance parameter El for long distances varies between 5 and 6 for male records, implying that 90% of maximal aerobic power can be maintained for a duration between approximately 25min and 36min, for the values of tc observed here. For female records, the endurance param- eter El is significantly larger with variations in an interval of approximately 6 to 8.5, implying that 90% of maximal aerobic power can be maintained for durations up to 85min. A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 10 / 26 Table 1. Race times and model parameters for various male running records, as of Oct. 2018. Record WR men WR 2000 men WR 1990 men tc[min] 6.26 5.50 5.90 vm[m/min] 411.72 417.07 405.00 100 γs 9.99 9.87 11.76 100 γl 5.36 6.19 5.93 Es 0.37 0.36 0.43 El 6.46 5.04 5.41 distance T Tmodel % T Tmodel % T Tmodel % 1000 02:11.96 02:11.94 -0.02 02:11.96 02:11.94 -0.02 02:12.80 02:12.82 +0.01 1500 03:26.00 03:26.24 +0.12 03:26.00 03:26.24 +0.12 03:29.46 03:29.26 -0.10 1609.34 03:43.13 03:42.91 -0.10 03:43.13 03:42.91 -0.10 03:46.32 03:46.50 +0.08 3000 07:20.67 07:20.99 +0.07 07:20.67 07:19.38 -0.29 07:29.45 07:30.88 +0.32 5000 12:37.35 12:37.10 -0.03 12:39.36 12:38.37 -0.13 12:58.39 12:56.88 -0.19 10000 26:17.53 26:18.84 +0.08 26:22.75 26:33.98 +0.71 27:08.23 27:08.68 +0.03 21097.5 58:23.00 58:11.94 -0.32 59:22.00 59:18.82 -0.09 1:00:46.00 1:00:25.03 -0.58 42195 2:01:39.00 2:01:52.99 +0.19 2:05:42.00 2:05:26.57 -0.20 2:06:50.00 2:07:21.71 +0.42 mean 0.12 0.21 0.22 Record WR 1980 men US men EU men tc[min] 5.26 6.08 4.97 vm[m/min] 405.27 406.06 412.81 100 γs 12.74 10.35 12.24 100 γl 6.21 5.67 5.76 Es 0.46 0.38 0.44 El 5.00 5.83 5.67 distance T Tmodel % T Tmodel % T Tmodel % 1000 02:13.40 02:13.41 +0.01 02:13.90 02:13.87 -0.02 02:12.18 02:12.17 -0.01 1500 03:31.36 03:31.26 -0.05 03:29.30 03:29.61 +0.15 03:28.81 03:28.90 +0.04 1609.34 03:48.80 03:48.89 +0.04 03:46.91 03:46.62 -0.13 03:46.32 03:46.24 -0.04 3000 07:32.10 07:34.44 +0.52 07:29.00 07:28.52 -0.11 07:26.62 07:26.39 -0.05 5000 13:08.40 13:04.65 -0.48 12:53.60 12:51.55 -0.26 12:49.71 12:48.61 -0.14 10000 27:22.47 27:30.09 +0.46 26:44.36 26:53.60 +0.58 26:46.57 26:49.72 +0.20 21097.5 1:02:16.00 1:01:26.52 -1.32 59:43.00 59:41.08 -0.05 59:32.00 59:38.51 +0.18 42195 2:09:01.00 2:10:02.03 +0.79 2:05:38.00 2:05:26.28 -0.16 2:05:48.00 2:05:34.12 -0.18 mean 0.46 0.18 0.11 Record GER men tc[min] 4.79 vm[m/min] 411.05 100 γs 11.22 100 γl 6.11 Es 0.41 El 5.14 distance T Tmodel % 1000 02:14.53 02:14.52 -0.01 1500 03:31.58 03:31.71 +0.06 1609.34 03:49.22 03:49.10 -0.05 3000 07:30.50 07:30.28 -0.05 5000 12:54.70 12:57.09 +0.31 10000 27:21.53 27:13.07 -0.52 (Continued) A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 11 / 26 The impact of endurance alone on running performances can be highlighted by measuring the mean race velocity vðdÞ in units of the crossover velocity vm and the race distance d in units of the crossover distance dc = vmtc. The resulting relation between vðdÞ=vm and d/dc is shown in Fig 2 for the current world records. Our model predicts that this relation depends only on the endurance parameters γl and γs, see Eq (12). The corresponding model curves are also plotted in Fig 2, showing good agreement with the data from world records. The better Table 1. (Continued) 21097.5 1:00:34.00 1:00:45.41 +0.31 42195 2:08:33.00 2:08:28.19 -0.06 mean 0.17 https://doi.org/10.1371/journal.pone.0206645.t001 Table 2. Race times and model parameters for various female records, as of Oct. 2018. † For the women WR of 2000 the result of Chinese runners for the distances 1500m, 3000m, 5000m and 10000m have been excluded due to use of performance-enhancing drugs [30]. Record WR women WR 2000 women† WR 1990 women tc[min] 8.30 10.01 5.50 vm[m/min] 361.37 352.14 364.74 100 γs 9.60 10.27 12.13 100 γl 4.85 5.53 5.74 Es 0.35 0.38 0.44 El 7.88 6.10 5.70 distance T Tmodel % T Tmodel % T Tmodel % 1000 02:28.98 02:28.78 -0.13 02:28.98 02:29.07 +0.06 02:30.67 02:30.17 -0.33 1500 03:50.07 03:52.05 +0.86 03:52.47 03:52.94 +0.20 03:52.47 03:57.27 +2.06 1609.34 04:12.56 04:10.70 -0.74 04:12.56 04:11.75 -0.32 04:21.68 04:16.95 -1.81 3000 08:20.68 08:18.13 -0.51 08:21.64 08:21.94 +0.06 08:22.62 08:25.93 +0.66 5000 14:11.15 14:12.40 +0.15 14:31.48 14:29.77 -0.20 14:37.33 14:31.08 -0.71 10000 29:17.45 29:29.07 +0.66 30:13.74 30:14.88 +0.06 30:13.74 30:24.18 +0.58 21097.5 1:04:51.00 1:04:50.60 -0.01 1:06:40.00 1:06:56.85 +0.42 1:08:32.00 1:07:34.87 -1.39 42195 2:15:25.00 2:15:00.95 -0.30 2:20:43.00 2:20:18.48 -0.29 2:21:06.00 2:22:16.17 +0.83 mean 0.42 0.20 1.05 Record US women EU women GER women tc[min] 10.80 10.19 5.87 vm[m/min] 347.42 351.63 356.56 100 γs 9.39 10.25 13.91 100 γl 5.17 4.63 5.01 Es 0.34 0.38 0.49 El 6.92 8.66 7.35 distance T Tmodel % T Tmodel % T Tmodel % 1000 02:31.80 02:32.01 +0.14 02:28.98 02:29.08 +0.07 02:30.67 02:30.48 -0.13 1500 03:56.29 03:56.68 +0.17 03:52.47 03:52.92 +0.20 03:57.71 03:59.58 +0.79 1609.34 04:16.71 04:15.62 -0.42 04:12.56 04:11.73 -0.33 04:21.59 04:19.83 -0.67 3000 08:25.83 08:26.40 +0.11 08:21.42 08:21.75 +0.07 08:29.89 08:34.62 +0.93 5000 14:38.92 14:37.27 -0.19 14:23.75 14:27.22 +0.40 14:42.03 14:41.99 -0.00 10000 30:13.17 30:24.74 +0.64 29:56.34 29:56.03 -0.02 30:57.00 30:34.60 -1.21 21097.5 1:07:34.00 1:07:03.57 -0.75 1:06:25.00 1:05:40.14 -1.13 1:07:58.00 1:07:25.42 -0.80 42195 2:19:36.00 2:20:00.22 +0.29 2:15:25.00 2:16:23.60 +0.72 2:19:19.00 2:20:46.06 +1.04 mean 0.34 0.37 0.70 https://doi.org/10.1371/journal.pone.0206645.t002 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 12 / 26 endurance of women for distances longer than the crossover distance dc is clearly visible. The gray cones in the figure indicate the range of endurance parameters that could potentially be realized in practice by runners from the recreational to the elite level with suitable event spe- cialization. This type of visualization of race performances allows one to evaluate runner’s endurance independently of their maximal aerobic power and running economy which are described by the parameters vm and tc. Estimate of supplemental power We have seen that supplemental power is responsible for a slow logarithmic decline of racing velocities with distance. In Fig 3 the supplemental factor of Eq (15) (square brackets in this equation) is plotted for various record performances as function of the race duration T. The variation range of the factor implies a supplemental power between  6% and 10% above the nominal power, with the European male records (EU men) being an outlier. The curves have their maximum at the crossover time T = tc. During supra-maximal exercise (for times shorter than tc), the oxygen uptake cannot stabilize and continues to increase until the end of the race [33]. Hence we observe an increasing deviation from the nominal power with increasing Fig 2. Mean race velocity vðdÞ as function of race distance. Velocity is re-scaled by vm, and distance d is re-scaled by dc = vmtc. Shown are the male and female world records (WR, dots), model prediction from Eq (12) (solid lines), and a typically expected maximal range of velocities (gray regions). Indicated are the lower and upper limits of γs and γl for these regions. Due to the re-scaling of vðdÞ and d, this graph highlights endurance for short and long duration, independently of the velocity vm at maximal aerobic power. https://doi.org/10.1371/journal.pone.0206645.g002 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 13 / 26 duration. However, at very short times below about 1 minute, oxygen uptake kinetics limit oxy- gen supply, and the energy deficit is compensated by the anaerobic system. After 30 to 60 sec- onds, the oxygen uptake can reach 90% of VO2max [33]. This short term kinetic effect is not included in our model. Above tc, i.e., for sub-maximal velocities, oxygen uptake stabilizes and the supplemental factor decreases. However, it does not decrease to one and this is likely related to the fact that the energy cost of running starts to increase above a nominal linear curve when the lactate threshold is approached [34]. For even longer race durations, we observe a slight increase in the supplemental factor that is presumably linked to the increase of the energy cost of running with increasing distance, as discussed in the Introduction. For a marathon or a 2 hour run at about 80% VO2max the supplemental power was measured to be between 5% and 7% in terms of oxygen uptake [35, 36] which is consistent with our model prediction for T * 120min. We note that for male records, the supplemental factor shows a shallow minimum around one hour. For female records this minimum is displaced to times above two hours. Application 1: Reconstruction individual physiological profiles After we have validated the accuracy of our model against record performances, we would like to find out if it can be also applied to individual runners. If that is the case then one could Fig 3. Plot of the supplemental factor of Eq (15) for as predicted by our model for male and female world records (WR), US records (US), and European records (EU). The cusp in the curves occurs at the time tc. https://doi.org/10.1371/journal.pone.0206645.g003 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 14 / 26 compute from their personal best performances their individual physiological parameters that characterize their training state and future performance potential. The assessment of the train- ing state of an individual is important not only for performance optimization but also beyond competitive athletics for the monitoring of the health status of recreational runners. There have been performance models developed for individual runners. A popular model is the so-called VDOT model by Daniels [37]. This model and other approaches employ maximal oxygen uptake as single factor determining performance [38, 39] and this parameter is then used to determine the training state and to predict running performances. A notable exception is the model Peronnet and Thibault which has been also applied to individual runners [7]. It turns out that their model yields comparable but somewhat larger errors than the present model. Partially, this might be due their model’s assumption that the energy cost of running and the crossover time tc would be identical for all runners. Other physiological factors that determine an individual’s performance include blood lactate concentration, and the anaerobic threshold. However, these parameters require laboratory measurements that are not always available, particularly on sufficiently short time intervals and for recreational athletes. With the advent of large online databases for personal best performances, it becomes possi- ble to probe the accuracy of performance models for a large set of individual athletes. Similar to our analysis of running records, our model predictions for individual runners can be validated through comparison with their personal best performances. First we reconstruct running econ- omy and endurance profiles of an individual runner from personal best performances for a few race distances and then estimate projected race times for other distances and also some charac- teristic paces. This eliminates physiological uncertainties that result from the use of universal, typical physiological parameters in previous models. In fact, the present model provides a gen- eral scheme that can be applied to any endurance runner over a range of distances and it is not based on observations made for only a small sample of trained athletes. Our approach also yields individual relative intensities, in percent of the aerobic power reserve Pm − Pb, at which a runner performs races. This is important for the relative use of fat and carbohydrate as fuels, and hence the total carbohydrate consumption for a given race distance. In the following, we apply our model to personal best performances of British runners that are available online in the database www.thepowerof10.info [40]. As a first test of our model for individual runners, we have considered the personal bests of the top nine male and female marathon runners from this database, according to the 2015 ranking. Their personal best times for seven distances from 800m to the marathon are summarized in Tables 3 and 4. With the same methodology that we used for running records above, we obtain the four model parameters for each runner that are also listed in the tables. From these parameters we com- pute the predicted race times. We find that the agreement between the predicted and actual race times are the most accurate to date, with an average mean error of less than 1% for each individual runner for all seven distances, see Tables 3 and 4. This suggests that our model can describe the running performance of individual runners with reliable accuracy. The slightly larger mean error for individuals than for groups of runners (record holders) appears natural since an individual runner can hardly reach optimized performance for all distances. When analyzing personal bests of an individual runner one should also realize that the best times on various distances have been probably obtained over a large time span of many years. Especially at the beginning of the career of a runner, when he races predominantly shorter distances, per- formance might not be optimal. Alternatively, one could consider only best performances obtained within a short time interval like a year which however limits presumably the available distances. Hence the individual variations of the parameters tc and vm can be large but they are strongly correlated. This suggests that tc gives a rather precise estimate of the time over which A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 15 / 26 Table 3. Personal best times and model parameters for individuals (Leading male marathon runners from UK, ranking 2015, http://www.thepowerof10.info/ rankings). Runner 01 02 03 tc[min] 23.84 11.28 4.57 vm[m/min] 353.28 360.35 373.49 100 γs 8.16 11.65 10.26 100 γl 4.67 5.07 4.85 Es 0.29 0.42 0.38 El 8.52 7.20 7.86 distance T Tmodel % T Tmodel % T Tmodel % 800 01:52.08 01:52.52 +0.39 01:49.98 01:49.94 -0.04 01:58.32 01:58.32 +0.00 1500 03:41.88 03:41.06 -0.37 03:40.80 03:40.95 +0.07 03:57.48 03:57.48 -0.00 3000 07:48.90 07:46.84 -0.44 08:00.48 08:00.34 -0.03 08:16.62 08:16.24 -0.08 5000 13:28.32 13:31.65 +0.41 13:57.66 14:01.83 +0.50 14:13.32 14:09.91 -0.40 10000 28:49.02 28:32.80 -0.94 29:23.04 29:09.19 -0.79 29:18.48 29:26.13 +0.43 21097.5 1:01:25.02 1:02:32.11 +1.82 1:04:07.02 1:04:12.25 +0.14 1:04:30.00 1:04:49.91 +0.51 42195 2:10:55.02 2:09:41.73 -0.93 2:13:40.98 2:13:52.45 +0.14 2:15:51.00 2:15:11.79 -0.48 mean 0.76 0.24 0.27 Runner 04 05 06 tc[min] 19.88 8.57 6.88 vm[m/min] 357.87 349.94 382.82 100 γs 8.27 4.84 6.93 100 γl 5.70 4.19 5.70 Es 0.30 0.13 0.24 El 5.78 10.86 5.79 distance T Tmodel % T Tmodel % T Tmodel % 800 01:51.78 01:52.20 +0.38 02:09.48 02:08.54 -0.73 01:55.20 01:55.20 +0.00 1500 03:41.94 03:40.71 -0.56 04:05.22 04:08.44 +1.31 03:45.66 03:45.66 -0.00 3000 07:46.74 07:46.78 +0.01 08:40.50 08:34.37 -1.18 08:00.12 07:53.93 -1.29 5000 13:31.20 13:32.52 +0.16 14:38.58 14:36.90 -0.19 13:33.00 13:35.29 +0.28 10000 28:42.18 28:31.86 -0.60 30:04.02 30:10.06 +0.33 27:57.24 28:25.13 +1.66 21097.5 1:02:22.98 1:03:06.46 +1.16 1:04:46.98 1:05:55.65 +1.77 1:03:00.00 1:03:04.39 +0.12 42195 2:12:57.00 2:12:10.52 -0.58 2:18:21.00 2:16:24.13 -1.41 2:13:40.02 2:12:33.95 -0.82 mean 0.49 0.99 0.60 Runner 07 08 09 tc[min] 8.44 8.17 5.28 vm[m/min] 355.63 367.03 347.39 100 γs 5.72 8.15 15.25 100 γl 5.62 5.81 4.82 Es 0.17 0.29 0.52 El 5.93 5.59 7.95 distance T Tmodel % T Tmodel % T Tmodel % 800 02:05.10 02:04.97 -0.11 01:57.42 01:57.12 -0.26 02:00.42 02:00.42 +0.00 1500 04:02.40 04:02.87 +0.19 03:49.98 03:51.04 +0.46 04:10.08 04:10.08 -0.00 3000 08:28.62 08:26.15 -0.49 08:14.04 08:10.42 -0.73 08:47.70 08:51.43 +0.71 5000 14:35.94 14:30.06 -0.67 14:01.02 14:04.01 +0.36 15:18.30 15:09.93 -0.91 10000 30:04.02 30:17.72 +0.76 29:32.70 29:26.30 -0.36 31:30.90 31:30.08 -0.04 21097.5 1:06:04.02 1:07:09.06 +1.64 1:04:28.02 1:05:23.05 +1.42 1:09:12.00 1:09:20.92 +0.21 42195 2:22:55.98 2:20:56.89 -1.39 2:18:49.02 2:17:31.77 -0.93 2:24:31.02 2:24:32.68 +0.02 mean 0.75 0.65 0.27 https://doi.org/10.1371/journal.pone.0206645.t003 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 16 / 26 Table 4. Personal best times and model parameters for individuals (Leading female marathon runners from UK, ranking 2015, http://www.thepowerof10.info/ rankings). Runner 01 02 03 tc[min] 13.01 9.48 3.85 vm[m/min] 319.66 316.60 368.12 100 γs 5.45 6.80 5.82 100 γl 4.70 4.28 4.41 Es 0.16 0.23 0.18 El 8.39 10.34 9.64 distance T Tmodel % T Tmodel % T Tmodel % 800 02:17.28 02:17.17 -0.08 02:18.60 02:18.31 -0.21 02:05.94 02:05.94 -0.00 1500 04:25.56 04:25.95 +0.15 04:29.58 04:30.60 +0.38 04:05.40 04:05.12 -0.11 3000 09:13.08 09:12.71 -0.07 09:32.82 09:28.54 -0.75 08:22.20 08:26.50 +0.86 5000 15:44.22 15:47.11 +0.31 16:13.02 16:09.74 -0.34 14:29.10 14:25.36 -0.43 10000 32:39.36 32:42.02 +0.14 33:01.98 33:23.16 +1.07 30:01.08 29:51.82 -0.51 21097.5 1:12:36.00 1:11:45.67 -1.16 1:12:28.02 1:13:01.34 +0.77 1:05:40.02 1:05:30.03 -0.25 42195 2:28:04.02 2:29:05.72 +0.69 2:32:40.02 2:31:12.50 -0.96 2:15:25.02 2:16:00.78 +0.44 mean 0.37 0.64 0.37 Runner 04 05 06 tc[min] 9.45 5.92 12.36 vm[m/min] 317.48 297.47 303.80 100 γs 6.93 9.70 9.00 100 γl 5.06 4.62 6.19 Es 0.24 0.36 0.33 El 7.23 8.71 5.02 distance T Tmodel % T Tmodel % T Tmodel % 800 02:18.72 02:17.68 -0.75 02:28.80 02:28.80 -0.00 02:17.40 02:17.16 -0.18 1500 04:26.04 04:29.59 +1.33 04:57.42 04:57.42 -0.00 04:30.84 04:31.69 +0.31 3000 09:36.72 09:26.96 -1.69 10:22.86 10:21.12 -0.28 09:40.44 09:39.64 -0.14 5000 16:08.10 16:11.38 +0.34 17:43.02 17:42.23 -0.07 16:47.82 16:46.53 -0.13 10000 33:24.72 33:39.59 +0.74 36:40.02 36:42.61 +0.12 35:18.00 35:11.88 -0.29 21097.5 1:13:21.00 1:14:10.86 +1.13 1:19:55.02 1:20:39.07 +0.92 1:17:43.02 1:18:25.22 +0.90 42195 2:36:39.00 2:34:47.30 -1.19 2:48:55.98 2:47:45.34 -0.70 2:46:19.02 2:45:29.11 -0.50 mean 1.03 0.30 0.35 Runner 07 08 09 tc[min] 16.50 5.40 14.64 vm[m/min] 281.73 300.16 272.55 100 γs 7.43 16.76 7.25 100 γl 4.28 5.16 4.45 Es 0.26 0.55 0.25 El 10.33 6.93 9.47 distance T Tmodel % T Tmodel % T Tmodel % 800 02:29.82 02:29.38 -0.29 02:20.22 02:20.22 -0.00 02:37.26 02:36.54 -0.46 1500 04:51.42 04:52.95 +0.52 04:55.20 04:55.20 -0.00 05:04.32 05:06.82 +0.82 3000 10:18.72 10:17.25 -0.24 10:08.70 10:20.47 +1.93 10:48.48 10:46.06 -0.37 5000 17:58.98 17:48.35 -0.98 18:13.98 17:44.85 -2.66 18:26.70 18:32.44 +0.52 10000 36:31.98 36:45.40 +0.61 37:07.98 36:59.37 -0.39 38:34.98 38:19.98 -0.65 21097.5 1:19:07.02 1:20:19.99 +1.54 1:20:39.00 1:21:45.39 +1.37 1:24:06.00 1:23:55.81 -0.20 42195 2:48:16.02 2:46:13.19 -1.22 2:51:46.02 2:51:06.21 -0.39 2:53:25.02 2:53:58.66 +0.32 mean 0.77 0.96 0.48 https://doi.org/10.1371/journal.pone.0206645.t004 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 17 / 26 a runner can sustain the velocity vm which, however, can deviate slightly from the actual veloc- ity at VO2max, depending on the available personal best performances in the vicinity of this crossover point. In order to measure individual endurances independently of aerobic capacity, we have computed and plotted the relation between the re-scaled race velocity vðdÞ=vm and distance d/dc in analogy to our analysis of running records, see Fig 4. Two important observa- tions can be made from this graph: (1) For each individual runner, there are two distinct rela- tions between velocity and distance above and below the crossover velocity vm and distance dc. (2) Even within the group of top UK marathon runners, there is a large variation in endur- ances as quantified by the different slopes of the re-scaled velocity-distance curves and the parameters γs and γl. They gray cones of expected maximal variations shown in Fig 4 are almost completely covered by the performances of the studied runners. For one of the female runners included in Table 4, runner 03 which is Paula Radcliffe, phys- iological data are available for a long time span of about 12 years [41]. While her personal rec- ords have been obtained over a similar period of time (800m in 1993 and marathon in 2003), and her physiological data have progressed during this time, in particular running economy, we can compare our model prediction for the speed vm to Radcliffe’s speed at VO2max, aver- aged over the time period from 1993 to 2003 which is about 22.5 km/h or 375.0 m/min [41]. This value compares very well with our finding of vm = 373.5 m/min, see Table 4. Our findings show that individual performances do not follow a unique power law as sug- gested, for example, by Riegel’s formula. There are more complex variations of physiological metrics among runners and those have to be taken into account for describing and predicting accurately performances and presumably optimal training. Our computational approach reveals the physiological parameters that determine individual performance and explains how they can be used in praxis to guide training and racing. Application 2: Personalized characteristic paces We expect that our four parameter model can measure an individual runner’s performance status for distances from 800m to the marathon more accurately than previous performance models that often assume for all runners the same (average) values for certain characteristics like running economy or endurance. An example for the latter type of models is the popular VDOT model of J. Daniels which assumes a fixed running economy and endurance curves for all runners [37, 42]. Although the VDOT model represents a good first approximation of char- acteristic paces based on a single race performance, the ability to monitor individual perfor- mances with more than just one parameter allows the runner to ascertain a better understanding of their training status and potential performance. It then becomes beneficial to have a model that makes use of larger available data sets. In the same way that one may better understand current fitness by examining relative oxygen consumption at different paces rather than absolute oxygen consumption, [43] developing an approach that makes use of perfor- mance over several races describes an individual runner better than a single race. Characteristic paces are often defined by the pace that a runner can race (at current training status) for a prescribed duration or distance. When the physiological model parameters of a runner are known from sufficiently many recent race performances, the running velocities for a prescribed intensity and duration, or intensity and distance can be computed from Eqs (17) and (18), respectively. In the following we consider race paces for a given duration or distance, corresponding to ^p ¼ 1 in these equations. In order to compare our model predictions to the characteristic paces of the VDOT model, we consider three hypothetical runners that are assumed to have achieved race performances as predicted by the VDOT model with model parameter values VDOT = 40, 60, and 80. (VDOT can be regarded as an effective value for A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 18 / 26 Fig 4. Same visualization of endurance as in Fig 2 but for individual male (top) and female (bottom) runners, see Tables 3 and 4. Colors label different runners. https://doi.org/10.1371/journal.pone.0206645.g004 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 19 / 26 VO2max, see [37] for details.) From these race performances we obtain the four parameters of our model. These parameters are given in the captions of Tables 5, 6 and 7. These tables pro- vide race paces (time per km) for various distances and durations specified in the first column. Some of the paces correspond the specific paces named in the VDOT model, and they are labeled correspondingly as R-, I-, T- and M-pace. The paces proposed by the VDOT model are given in the second column. The remaining columns provide the predictions of our model. The third column lists the paces as obtained from the values of the four model parameters that result from the hypothetical race performances of the runner with the given VDOT score. There is agreement within a few seconds per kilometer. It should be kept in mind that our model, unlike the VDOT model, does not implement any fixed parameters or constants a pri- ori. We observe that the fixed parameters of the VDOT model correspond to rather superior endurance with γl  0.05 for long distances and average endurance with γs  0.09 for short distances. As we have seen above, there is substantial variation in these parameters among individuals. Hence, characteristic paces should also determined individually. We have modi- fied the endurance parameters γl and γs independently within their typical minimal and maxi- mal values while keeping vm and tc unchanged. The resulting paces are shown in the last four columns of the tables. The fast paces for short distances (1mile and 5min paces) can change up to ±10sec/km compared to the original VDOT model which is substantial. For the slower paces (for time tc and longer) the variation can be even larger with a maximum change for the Table 5. Paces per km for a runner with VDOT = 40 score for different endurances. The original physiological parameters are tc = 12.35min, vm = 214.88m/min, γl = 0.051 and γs = 0.096. In last 4 columns the endurances El and Es are given only when they are different from the original values. pace at max. power for Ref. [37] original γl = 0.04 γl = 0.08 γs = 0.15 γs = 0.05 El = 7.1 El = 12.2 El = 3.5 Es = 0.35 Es = 0.51 Es = 0.14 1 mile (R-pace) 04:20 04:25.21 orig. orig. 04:16.72 04:32.06 5min — 04:16.96 orig. orig. 04:05.87 04:27.14 time tc (I-pace) 04:42 04:39.22 orig. orig. orig. orig. 5.000m 04:49 04:49.06 04:46.79 04:55.54 orig. orig. 10.000m 05:00 05:00.67 04:55.58 05:15.85 orig. orig. 60min (T-pace) 05:06 05:03.67 04:58.07 05:19.64 orig. orig. Half marathon 05:15 05:14.30 05:05.68 05:41.31 orig. orig. marathon (M-pace) 05:29 05:28.16 05:15.70 06:09.16 orig. orig. https://doi.org/10.1371/journal.pone.0206645.t005 Table 6. Paces per km for a runner with VDOT = 60 score for different endurances. The original physiological parameters are tc = 12.67min, vm = 298.51m/min, γl = 0.052 and γs = 0.092. The meaning of the columns is the same as in Table 5. pace at max. power for Ref. [37] original γl = 0.04 γl = 0.08 γs = 0.15 γs = 0.05 El = 6.8 El = 12.2 El = 3.5 Es = 0.34 Es = 0.51 Es = 0.14 1 mile (R-pace) 03:05 03:05.04 orig. orig. 02:54.93 03:12.36 5min — 03:05.15 orig. orig. 02:56.39 03:12.07 time tc (I-pace) 03:23 03:21.00 orig. orig. orig. orig. 5.000m 03:25 03:24.14 03:23.36 03:26.00 orig. orig. 10.000m 03:32 03:32.41 03:29.49 03:39.64 orig. orig. 60min (T-pace) 03:40 03:38.78 03:34.33 03:49.55 orig. orig. Half marathon 03:42 03:42.12 03:36.53 03:56.62 orig. orig. marathon (M-pace) 03:52 03:51.99 03:43.51 04:15.08 orig. orig. https://doi.org/10.1371/journal.pone.0206645.t006 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 20 / 26 marathon pace (M-pace). For a VDOT = 40 runner, the M-pace window between slowest and fastest pace is about 55sec/km, for a VDOT = 60 runner it is about 30sec/km and even for a high level runner with VDOT = 80 it is still about 20sec/km. These variations result from dif- ferent endurances, with the crossover speed vm unchanged. We have also studied the effect of a modification of the time tc from the original VDOT model value which appears rather long with 12 to 13min. The results are shown in Tables 8, 9 and 10. The first three columns have the same meaning as in the three tables before. The last four columns list the paces that correspond to a reduction or an increase of tc by 10% or 20%, respectively. Here we observe a smaller varia- tion by a few seconds around the original paces, relatively independent of the duration or dis- tance that defines the pace. This shows that racing paces are more dependent on endurance than on the time over which runners can sustain their crossover speed at VO2max. The reason for that is the exponential dependence on γs, γl of the duration T(p) over which a relative power p can be maintained, independently of tc and vm, see Fig 1. It is interesting to relate this observation to physiological parameters that can be measured in the laboratory and have been linked to endurance capacity, like blood lactate concentration. It is known that the running speed at the lactate threshold can improve independently of VO2max and so can the runner’s endurance. Often the lactate threshold pace is identified with the running velocity that a runner can race for about 60min. The corresponding paces are shown in Tables 5–10 as “T-pace”. The relative intensity or power output in percent of the aer- obic power reserve [see Eq (1)] at the lactate threshold is given by pLT = 100[1 − γl log(60/tc)]. For example, for a recreational runner (with VDOT = 40), described by the parameters of Table 7. Paces per km for a runner with VDOT = 80 score for different endurances. The original physiological parameters are tc = 12.92min, vm = 376.85m/min, γl = 0.053 and γs = 0.088. The meaning of the columns is the same as in Table 5. pace at max. power for Ref. [37] original γl = 0.04 γl = 0.08 γs = 0.15 γs = 0.05 El = 6.6 El = 12.2 El = 3.5 Es = 0.32 Es = 0.51 Es = 0.14 1 mile (R-pace) 02:25 02:23.98 orig. orig. 02:13.52 02:30.46 5min — 02:26.99 orig. orig. 02:19.37 02:32.00 time tc (I-pace) 02:41 02:39.22 orig. orig. orig. orig. 5.000m 02:40 02:39.45 02:39.39 02:39.59 orig. orig. 10.000m 02:46 02:45.91 02:44.14 02:49.88 orig. orig. 60min (T-pace) 02:54 02:53.33 02:49.64 03:01.52 orig. orig. Half marathon 02:53 02:53.50 02:49.59 03:02.65 orig. orig. marathon (M-pace) 03:01 03:01.22 02:54.99 03:16.46 orig. orig. https://doi.org/10.1371/journal.pone.0206645.t007 Table 8. Paces per km for a runner with VDOT = 40 score for different variations of the time tc. The original physiological parameters are tc = 12.35min, vm = 214.88m/min, γl = 0.051 and γs = 0.096. pace at max. power for Ref. [37] original tc = 12.35min 0.8tc 0.9tc 1.1tc 1.2tc 1 mile (R-pace) 04:20 04:25.21 04:31.27 04:28.03 04:22.70 04:20.46 5min — 04:16.96 04:22.12 04:19.37 04:14.82 04:12.90 time tc (I-pace) 04:42 04:39.22 04:42.43 04:40.73 04:36.70 04:34.43 5.000m 04:49 04:49.06 04:52.69 04:50.76 04:47.53 04:46.15 10.000m 05:00 05:00.67 05:04.62 05:02.52 04:59.02 04:57.52 60min (T-pace) 05:06 05:03.67 05:07.47 05:05.45 05:02.07 05:00.63 Half marathon 05:15 05:14.30 05:18.63 05:16.33 05:12.49 05:10.86 marathon (M-pace) 05:29 05:28.16 05:32.89 05:30.37 05:26.18 05:24.39 https://doi.org/10.1371/journal.pone.0206645.t008 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 21 / 26 Table 5, one has pLT = 91.94% for the original value γl = 0.051, while pLT = 93.68% for γl = 0.04, and pLT = 87.35% for γl = 0.08. These values appear rather large when compared to the lactate threshold estimates from current world records: pLT = 87.08% for male and pLT = 90.41% for female records. This implies again that the VDOT model assumes a rather optimized endurance. Conclusion Modern performance testing is often based on laboratory testing of athletes with the goal of identifying physiological metrics that correlate with performance and can be linked to funda- mental physiological processes. However, measuring physiological metrics requires time con- suming and expensive testing, often under rather idealized laboratory conditions. Hence, it appears to be very useful to extract information on power characteristics for individual run- ners or certain groups of runners from performance results in racing events or time trails. This is of particularly great interest for analyzing the effect of aging on human performance, consid- ering the enormous improvement of performance in older age groups. As stated already by A. V. Hill, world and other records constitute very interesting data sets since their accuracy by far exceeds that of laboratory measurements and they correspond to best human performances at a given time in history under realistic conditions. The model presented here provides a quantitative method for extracting characteristic parameters from race performances of a group of runners or of an individual runner. The key equations and computational steps of our model are as follows: Table 9. Paces per km for a runner with VDOT = 60 score for different variations of the time tc. The original physiological parameters are tc = 12.67min, vm = 298.51m/min, γl = 0.052 and γs = 0.092. pace at max. power for Ref. [37] original tc = 12.67min 0.8tc 0.9tc 1.1tc 1.2tc 1 mile (R-pace) 03:05 03:05.04 03:08.94 03:06.86 03:03.42 03:01.97 5min — 03:05.15 03:08.72 03:06.82 03:03.67 03:02.34 time tc (I-pace) 03:23 03:21.00 03:23.37 03:22.11 03:19.25 03:17.68 5.000m 03:25 03:24.14 03:26.73 03:25.36 03:23.06 03:22.08 10.000m 03:32 03:32.41 03:35.22 03:33.72 03:31.23 03:30.17 60min (T-pace) 03:40 03:38.78 03:41.60 03:40.10 03:37.60 03:36.54 Half marathon 03:42 03:42.12 03:45.20 03:43.56 03:40.83 03:39.66 marathon (M-pace) 03:52 03:51.99 03:55.36 03:53.57 03:50.58 03:49.31 https://doi.org/10.1371/journal.pone.0206645.t009 Table 10. Paces per km for a runner with VDOT = 80 score for different variations of the time tc. The original physiological parameters are tc = 12.92min, γl = 0.053 and γs0.088. pace at max. power for Ref. [37] original tc = 12.92min 0.8tc 0.9tc 1.1tc 1.2tc 1 mile (R-pace) 02:25 02:23.98 02:26.80 02:25.30 02:22.81 02:21.76 5min — 02:26.99 02:29.69 02:28.25 02:25.87 02:24.85 time tc (I-pace) 02:41 02:39.22 02:41.12 02:40.11 02:37.90 02:36.71 5.000m 02:40 02:39.45 02:41.48 02:40.40 02:38.17 02:36.87 10.000m 02:46 02:45.91 02:48.11 02:46.94 02:44.99 02:44.16 60min (T-pace) 02:54 02:53.33 02:55.60 02:54.39 02:52.38 02:51.53 Half marathon 02:53 02:53.50 02:55.91 02:54.63 02:52.49 02:51.58 marathon (M-pace) 03:01 03:01.22 03:03.85 03:02.45 03:00.11 02:59.12 https://doi.org/10.1371/journal.pone.0206645.t010 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 22 / 26 • The key equation for the comparison of our model to race results is the expression for the race time T(d) as function of the race distance d, given in Eq 11. • We minimize the sum of the squared relative deviations in percent between actual race times and the function T(d) by varying the four model parameters vm, tc, γs, and γl in T(d) for all distances raced by a runner or a group of runners (records). The final model parameters are those that result from this minimization. • From the four model parameters the values of which were obtained from race performances or by other input like physiological data, the function T(d) predicts the race times and Eq 12 the mean race velocities for arbitrary distances. The model parameters quantify the runner’s performance status and can be used to predict personalized fastest possible but realistic and safe racing paces for a wide range of race distances and durations. Our model provides an unified description of running events at sub- and supra- maximal velocities that are separated by a time scale tc whose value is in good agreement with independent measurements. On a fundamental level, for the first time our approach provides a derivation of the previously observed but unexplained linear relation between the mean velocity and the logarithm of the duration for running records. The mechanism underlying this loga- rithmic relation could be identified as the necessity of a supplemental power, beyond the nomi- nal power cost of running, for maintaining the mean velocity. Our findings are different from the previously postulated power law relation between the mean race speed v and distance d, v  dbreak 2 hours in the marathon. For example, the latest update of the world record in the marathon by Eliud Kipchoge in Berlin on September 16, 2018 which is included in our results of Table 1, has increased the endurance for long duration from El = 5.98 to El = 6.46, i.e., by 8%, while the speed that can be raced for 6min (413.5 m/min) and the short term endurance Es remained basi- cally unchanged. Our model predicts that the endurance for long duration had to be increased to El = 7.49 with all other parameters unchanged to obtain a marathon time of 1:59:56. This corre- sponds to another increase of 16% compared to the just updated value which appears unrealistic in near future. Another possibility, however, would be to assume the endurance of the current world record, El = 6.46, and an increased speed at VO2max. For example, increasing the speed that a runner can sustain for 6min by 1.3% to vm = 418.7 m/min would yield a marathon time of 1:59:58. This could be achieved by an increase in running economy by only *1% which seems feasible, at least by material improvements and/or suitable racing conditions (course, climate). Future studies based on our model could include the dependence of the performance state on distance specialization, altitude, air temperature, age, and other factors. With the availabil- ity of big data set on running performances, these studies could be performed with much bet- ter statistics than studies with much smaller groups of runners participating in laboratory and clinical studies. Our model could be applied to other endurance sports after a modification of the running specific dependence of power on velocity. Supporting information S1 Appendix. Solution of the integral equation for Pmax(T). (PDF) S2 Appendix. Comparison to oxygen uptake measurement. (PDF) Acknowledgments Valuable discussions with Veronique Billat and Francois Pe´ronnet on various physiological aspects of the model and with Jack Daniels on the methodology of the VDOT model are acknowledged. Author Contributions Conceptualization: Thorsten Emig. Data curation: Guillaume Adam. Formal analysis: Matthew Mulligan, Thorsten Emig. Funding acquisition: Thorsten Emig. Investigation: Matthew Mulligan, Guillaume Adam, Thorsten Emig. Methodology: Thorsten Emig. Software: Matthew Mulligan, Guillaume Adam. Supervision: Thorsten Emig. Validation: Guillaume Adam, Thorsten Emig. Writing – original draft: Matthew Mulligan, Thorsten Emig. Writing – review & editing: Matthew Mulligan, Guillaume Adam, Thorsten Emig. A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 24 / 26 References 1. Kennely AE. An Approximate Law of Fatigue in the Speeds of Racing Animals. Proceedings of the American Academy of Arts and Sciences. 1906; 42(15):275–331. https://doi.org/10.2307/20022230 2. Hill AV. The Physiological Basis Of Ahletic Records. The Lancet. 1925; 206(5323):481–486. https://doi. org/10.1016/S0140-6736(01)15546-7 3. Ward-Smith AJ. A Mathematical Theory of Running, based on the 1st Law of Thermodynamics, and its Applications to the Performance of World-Class Athletes. J Biomechanics. 1985; 18(5):337–349. 4. Keller JB. A theory of competitive running. Phys Today. 1973; 26(9):43–47. https://doi.org/10.1063/1. 3128231 5. Jones AM, Vanhatalo A, Burnley M, Morton RH, Poole DC. Critical power: implications for determination of V_O2max and exercise tolerance. Med Sci Sports Exerc. 2010; 42(10):1876–90. https://doi.org/10. 1249/MSS.0b013e3181d9cf7f PMID: 20195180 6. Peronnet F, Thibault G. Mathematical analysis of running performance and world running records. J Appl Physiol. 1989; 67(1):453–465. https://doi.org/10.1152/jappl.1989.67.1.453 PMID: 2759974 7. Peronnet F, Thibault G, Cousineau DL. A theoretical analysis of the effect of altitude on running perfor- mance. J Appl Physiol. 1991; 70(1):399–404. https://doi.org/10.1152/jappl.1991.70.1.399 PMID: 2010398 8. Fletcher JR, Esau SP, MacIntosh BR. Economy of running: Beyond the measurement of oxygen uptake. J Appl Physiol. 2009; 107:1918–1922. https://doi.org/10.1152/japplphysiol.00307.2009 PMID: 19833811 9. Billat V, Renoux J, Pinoteau J, Petit B, Koralsztein J. Reproducibility of running time to exhaustion at VO2max in subelite runners. Medicine & Science in Sports & Exercise. 1994; 26(2):254–257. https:// doi.org/10.1249/00005768-199402000-00018 10. Bosquet L, Leger L, Legros P. Methods to Determine Aerobic Endurance. Sports Med. 2002; 32 (11):675–700. https://doi.org/10.2165/00007256-200232110-00002 PMID: 12196030 11. Joyner MJ, Coyle EF. Endurance exercise performance: the physiology of champions. J Physiol. 2008; 586:35–44. https://doi.org/10.1113/jphysiol.2007.143834 PMID: 17901124 12. O’Brien MJ, Viguie CA, Mazzeo RS, Brooks GA. Carbohydarte dependence during marathon running. Medicine & Science in Sports & Exercise. 1993; 25(9):1009–1017. 13. Margaria R, Cerretelli P, Aghemo P, Sassi G. Enery cost of running. J Appl Physiol. 1963; 18:367–370. https://doi.org/10.1152/jappl.1963.18.2.367 PMID: 13932993 14. Leger L, Mercier D. Gross Energy Cost of Horizontal Treadmill and Track Running. Sports Med. 1984; 1:270–277. https://doi.org/10.2165/00007256-198401040-00003 PMID: 6390604 15. Sproule J. Running economy deteriorates following 60 min of exercise at 80% VO2max. Eur J Appl Phy- siol. 1998; 77:366–371. https://doi.org/10.1007/s004210050346 16. Hunter I, Smith G. Preferred and optimal stride frequency, stiffness and economy: Changes with fatigue during a 1-h high-intensity run. Eur J Appl Physiol. 2007; 100:653–661. https://doi.org/10.1007/s00421- 007-0456-1 PMID: 17602239 17. Thomas DQ, Fernhall B, Granat H. Changes in Running Economy During a 5-km Run in Trained Men and Women Runners. J Strength Cond Res. 1999; 13(2):162–167. https://doi.org/10.1519/1533-4287 (1999)013%3C0162:CIREDA%3E2.0.CO;2 18. Beis LY, Wright-Whyte M, Fudge B, Noakes T, Pitsiladis YP. Drinking Behaviors of Elite Male Runners During Marathon Competition. Clin J Sports Med. 2012; 22(3):254–261. https://doi.org/10.1097/JSM. 0b013e31824a55d7 19. Grimby G. Exercise in man during pyrogen-induced fever. Acta Physiol Scand Suppl. 1962; 67:1–114. 20. Dion T, Savoie FA, Asselin A, Gariepy C, Goulet EDB. Half-marathon running performance is not improved by a rate of fluid intake above that dictated by thirst sensation in trained distance runners. Eur J Appl Physiol. 2013; 113(12):3011–3020. https://doi.org/10.1007/s00421-013-2730-8 PMID: 24085484 21. Burnley M, Jones AM. Power-duration relationship: physiology, fatigue and the limits of human perfor- mance. European Journal of Sport Science. 2016; 18(1):1–12. https://doi.org/10.1080/17461391.2016. 1249524 PMID: 27806677 22. Bosch A, Goslin BR, Noakes TD, Dennis SC. Physiological differences between black and white run- ners during a treadmill marathon. Eur J Appl Physiol. 1990; 61:68. https://doi.org/10.1007/BF00236696 23. Morton RH. A 3-parameter critical power model. Ergonomics. 1996; 39:611–619. https://doi.org/10. 1080/00140139608964484 PMID: 8854981 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 25 / 26 24. Billat V, Morton R, Blondel N, Berthoin S, Bocquet V, Koralsztein J, et al. Oxygen kinetics and modelling of time to exhaustion whilst running at various velocities at maximal oxygen uptake. Eur J Appl Physiol. 2000; 82(3):178–187. https://doi.org/10.1007/s004210050670 PMID: 10929211 25. Morgan DW, Martin PE, Krahenbuhl GS. Factors affecting running economy. Sports Med. 1989; 7 (5):310–330. https://doi.org/10.2165/00007256-198907050-00003 PMID: 2662320 26. Billat V, Hamard L, Koralsztein JP, Morton RH. Differential modeling of anaerobic and aerobic metabo- lism in the 800-m and 1,500-m run. J Appl Physiol. 2009; 107(2):478–487. https://doi.org/10.1152/ japplphysiol.91296.2008 PMID: 19478190 27. Corless RM, Gonnet GH, Hare DEG, Jeffrey DJ, Knuth DE. On the Lambert W function. Adv Comp Math. 1996; 5(4):329–359. https://doi.org/10.1007/BF02124750 28. Riegel PS. Athletic Records and Human Endurance. American Scientist. 1981; 69:285–290. PMID: 7235349 29. Savaglio S, Carbone V. Human performance: scaling in athletic world records. Nature. 2000; 404:244. https://doi.org/10.1038/35005165 PMID: 10749198 30. The Guardian, https://www.theguardian.com/sport/2017/oct/22/china-compulsory-doping-olympic- athletes-claims-whistleblower-athletics. 31. The Guardian, https://www.theguardian.com/sport/2017/aug/04/doping-hotspot-ethiopia-drug-testing- epo. 32. Billat V, Binsse V, Petit B, Koralsztein JJ. High level runners are able to maintain a VO2 steady-state below VO2max in an all-out run over their critical velocity. Archives of physiology and biochemistry. 1998; 106(1):38–45. https://doi.org/10.1076/apab.106.1.38.4396 PMID: 9783059 33. Gastin PB. Energy Systems Interaction and Relative Contribution During Maximal Exercise. Sports Med. 2001; 31(10):725–741. https://doi.org/10.2165/00007256-200131100-00003 PMID: 11547894 34. Batliner ME, Kipp S, Grabowski AM, Kram R, Byrnes WC. Does Metabolic Rate Increase Linearly with Running Speed in all Distance Runners? Sports Medicine International Open. 2018; 2:E1–E8. https:// doi.org/10.1055/s-0043-122068 35. Brueckner JC, Atchou G, Capelli C, A D, Barrault D, Jousselin E, et al. The energy cost of running increases with the distance covered. Eur J Appl Physiol. 1991; 62:385–389. https://doi.org/10.1007/ BF00626607 36. Brisswalter J, Hausswirth C, Vercruyssen F, Collardeau M, Vallier JM, Lepers R, et al. Carbohydrate ingestion does not influence the change in energy cost during a 2-h run in well-trained triathletes. Eur J Appl Physiol. 2000; 81:108–113. https://doi.org/10.1007/PL00013781 PMID: 10552274 37. Daniels J. Daniels’ Running Formula. 3rd ed. Human Kinetics; 2013. 38. Conley DL, Krahenbuhl GS. Running economy and distance running performance of highly trained ath- letes. Med Sci Sports Exerc. 1980; 12(5):357–360. https://doi.org/10.1249/00005768-198025000- 00010 PMID: 7453514 39. Williams KR, Cavanagh PR. Relationship between distance running mechanics, running economy, and performance. Journal of Applied Physiology. 1987; 63(3):1236–1245. https://doi.org/10.1152/jappl. 1987.63.3.1236 PMID: 3654469 40. British Athletics. The Power of 10. Available from http://www.thepowerof10.info/rankings; 2018. 41. Jones AM. The Physiology of the World Record Holder for the Women’s Marathon. International Journal of Sports Science & Coaching. 2006; 1:101. https://doi.org/10.1260/174795406777641258 42. Daniels J, Gilbert J. Oxygen power: performance tables for distance runners. J. Daniels, J. Gilbert; 1979. 43. Bassett DR Jr, Howley ET. Limiting factors for maximum oxygen uptake and determinants of endurance performance. Medicine & Science in Sports & Exercise. 2000; 32(1):70. https://doi.org/10.1097/ 00005768-200001000-00012 A minimal power model for human running performance PLOS ONE | https://doi.org/10.1371/journal.pone.0206645 November 16, 2018 26 / 26

A minimal power model for human running performance.

11-16-2018

Mulligan, Matthew,Adam, Guillaume,Emig, Thorsten

eng

PMC9977817

Increased oxygen uptake in well-trained runners during uphill high intensity running intervals: A randomized crossover testing Steffen Held1,2, Ludwig Rappelt 1,3, René Giesen1, Tim Wiedenmann1, Jan-Philip Deutsch1, Pamela Wicker4* and Lars Donath 1 1Department of Intervention Research in Exercise Training, German Sport University Cologne, Cologne, Germany, 2Department of Fitness and Health, IST University of Applied Sciences, Duesseldorf, Germany, 3Department of Movement and Training Science, University of Wuppertal, Wuppertal, Germany, 4Department of Sports Science, Bielefeld University, Bielefeld, Germany The time spent above 90% of maximal oxygen uptake ( _VO2max) during high- intensity interval training (HIIT) sessions is intended to be maximized to improve _VO2max. Since uphill running serves as a promising means to increase metabolic cost, we compared even and moderately inclined running in terms of time ≥90% _VO2max and its corresponding physiological surrogates. Seventeen well-trained runners (8 females & 9 males; 25.8 ± 6.8yrs; 1.75 ± 0.08m; 63.2 ± 8.4kg; _VO2max: 63.3 ± 4.2 ml/min/kg) randomly completed both a horizontal (1% incline) and uphill (8% incline) HIIT protocol (4-times 5min, with 90s rest). Mean oxygen uptake ( _VO2mean), peak oxygen uptake ( _VO2peak), lactate, heart rate (HR), and perceived exertion (RPE) were measured. Uphill HIIT revealed higher (p ≤ 0.012; partial eta- squared (pes) ≥ 0.351) _VO2mean (uphill: 3.3 ± 0.6 vs. horizontal: 3.2 ± 0.5 L/min; standardized mean difference (SMD) = 0.15), _VO2peak (uphill: 4.0 ± 0.7 vs. horizontal: 3.8 ± 0.7 L/min; SMD = 0.19), and accumulated time ≥90% _VO2max (uphill: 9.1 ± 4.6 vs. horizontal: 6.4 ± 4.0 min; SMD = 0.62) compared to even HIIT. Lactate, HR, and RPE responses did not show mode*time rANOVA interaction effects (p ≥ 0.097; pes ≤0.14). Compared to horizontal HIIT, moderate uphill HIIT revealed higher fractions of _VO2max at comparable perceived efforts, heartrate and lactate response. Therefore, moderate uphill HiiT notably increased time spent above 90% _VO2max. KEYWORDS incline, intervals, performance, injury, running 1 Introduction High level endurance training requires large training volumes (Seiler, 2010). In elite athletes, commonly, a high proportion of this training volume is performed at low training intensities (Seiler, 2010). However, to achieve an optimal metabolic training stimulus on maximal oxygen uptake ( _VO2max), it has been recommended to perform a certain amount of high-intensity interval training (HIIT). This recommendation is especially relevant for well-trained endurance athletes (Laursen and Jenkins, 2002). Thereby, HIIT involves repeated bouts of high-intensity exercise interspersed with recovery periods (Laursen and Jenkins, 2002; Buchheit and Laursen, 2013). This training method mainly focuses OPEN ACCESS EDITED BY Andrea Nicolò, Foro Italico University of Rome, Italy REVIEWED BY Marcel Lemire, University of Upper Alsace, France Stéphane P Dufour, Université de Strasbourg, France *CORRESPONDENCE Pamela Wicker, pamela.wicker@uni-bielefeld.de SPECIALTY SECTION This article was submitted to Exercise Physiology, a section of the journal Frontiers in Physiology RECEIVED 06 December 2022 ACCEPTED 06 February 2023 PUBLISHED 16 February 2023 CITATION Held S, Rappelt L, Giesen R, Wiedenmann T, Deutsch J-P, Wicker P and Donath L (2023), Increased oxygen uptake in well-trained runners during uphill high intensity running intervals: A randomized crossover testing. Front. Physiol. 14:1117314. doi: 10.3389/fphys.2023.1117314 COPYRIGHT © 2023 Held, Rappelt, Giesen, Wiedenmann, Deutsch, Wicker and Donath. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Frontiers in Physiology frontiersin.org 01 TYPE Original Research PUBLISHED 16 February 2023 DOI 10.3389/fphys.2023.1117314 on _VO2max improvements (Midgley et al., 2006; Buchheit and Laursen, 2013), as the upper limit to the aerobic metabolism and a key determinant of endurance performance (Joyner and Coyle, 2008). In order to improve _VO2max in highly trained endurance athletes, it has been suggested that a prolonged time at intensities corresponding to a high percentage of maximal oxygen uptake is important (Wenger and Bell, 1986; Midgley et al., 2006). Therefore, the quality of a HIIT session can be defined by mean oxygen uptake ( _VO2mean) or accumulated training time ≥90% _VO2max (Midgley et al., 2006; Turnes et al., 2016). This adaptational potential has been attributed to the large metabolic stimulus for myocardial morphological adaptations that increases maximal cardiac stroke volume and also increased peripheral skeletal muscle adaptations (Midgley et al., 2006). In both prospective and cohort studies, a high weekly running volume has been associated with running-related injuries (Macera et al., 1989; Walter et al., 1989). Although the causes of running injuries are multifactorial, in this context, the runner’s interaction with the ground and the resulting reaction force has been considered to be one risk factor (Zadpoor and Nikooyan, 2011; Daoud et al., 2012). Thus, higher loading rates were associated with increased risk of sustaining an injury (Crowell and Davis, 2011; Futrell et al., 2018). More recently, however, in a prospective case control-study in recreational runners, the vertical impact peak and loading rate were not associated with a higher injury rate (Malisoux et al., 2022). Furthermore, in collegiate cross country runners, an higher occurrence rate of bone stress injuries has been linked to a higher step rate, but not higher ground reaction forces (Kliethermes et al., 2021). Nevertheless, besides adequate periodization and polarization models in endurance sports, reducing loading rates is still recommended as an effective means to reduce the risk of developing running injuries (Bowser et al., 2018). In this context, increasing the slope might lead to a significantly lower vertical loading rate during uphill running compared to flat level running (Gottschall and Kram, 2005; Lemire et al., 2022a). Also, increasing the slope from flat level running to 7% was found to reduce flight time and increase floor contact time, in turn resulting in highly significant increases in step frequency (Padulo et al., 2013). Apart from this, previous research revealed an increased energy cost via uphill running compared to horizontal running (Lemire et al., 2022b). Additionally, when running at the same velocity, uphill running is more metabolically demanding than horizontal running (Minetti et al., 2002; Vernillo et al., 2017), hence allowing a similar training stimulus at a lower running velocity. Against this background, this randomized crossover testing examined the peak _VO2, mean _VO2 and accumulated time spent ≥90% _VO2max during moderate slope uphill compared to horizontal HIIT running. We assumed similar _VO2 data and reduced running speed during uphill HIIT. The findings of the present study might be impactful for designing and integrating HIIT session within polarization models and in terms of training variations to minimize injury risks in runners with high training volumes. 2 Materials and methods 2.1 Participants G*Power (Version 3.1.9.6) was employed to perform an a priori power analysis. Based on increased metabolic costs via uphill running (Minetti et al., 1994; 2002; Vernillo et al., 2017) moderate effect sizes (standard mean differences (SMD) = 0.60) between horizontal and uphill HIIT running were assumed. A sample size of n = 13 was determined, using the following statistical indicators (α = 0.05; study power (1-β-error) = 0.95; one tail). Assuming moderate dropouts (15%–20%), n = 17 well- trained runners were enrolled in this acute randomized controlled crossover testing. These participants consisted of 8 female (age: 24.4 ± 3.7 yrs; height: 1.69 ± 0.07 m; body mass: 56.6 ± 5.8 kg; body fat: 14.6 ± 4.8%; _VO2max: 60.5 ± 2.3 ml/min/kg; running volume: 58.1 ± 18.5 km/week) and 9 male (age: 27.1 ± 8.8 yrs; height: 1.80 ± 0.07 m; body mass: 69.1 ± 5.6 kg; body fat: 9.7 ± 3.1%; _VO2max: 65.7 ± 4.1 ml/min/kg; running volume: 65.0 ± 20.3 km/week) trained runners. Inclusion criteria were running experience of at least 3 years, running volume of at least 40 km/week, and no medical condition that potentially impedes the completion of testing and training. The study was approved by the local ethical committee (153/2022), fulfilled the international ethical standards, and all participants signed an informed written consent prior to the start of the study. 2.2 Testing procedures The measurements were conducted within four lab visits over 3 weeks for each participant. Thereby, horizontal and uphill _VO2max tests (lab visit 1 & 2) as well as horizontal and uphill HIIT protocols (lab visit 3 & 4) were performed. Adapted from previous research (Rønnestad et al., 2019; 2022), the HIIT protocol consisted of four 5-min intervals with 90 s passive rest in between. During HIIT sessions, participants were instructed to run at their maximal sustainable intensity during all four interval bouts (isoeffort) (Seiler and Hetlelid, 2005). Therefore, participants could increase or decrease the velocity individually. All measurements were conducted on a motorized treadmill (PPS Med treadmill, Woodway, Waukesha, USA), with the horizontal conditions being performed at 1% incline and the uphill conditions being performed at 8% incline. To avoid sequencing effects, the first two and the last two lab visits were individually performed in a randomized order. At least 96 h rest was ensured between each lab visit. Participants were further instructed to avoid any strenuous exercise 2 days before each testing session. To control for potential circadian effects on performance, all measurements were conducted at similar day times for each participant. A standardized 15-min warm-up (easy running, including knee lift, heel lift, external rotation hip, internal rotation hip, 10 lunges alternating, 10 squats, individual dynamic stretching) was performed prior to each lab session. Spirometric data during all lab visits were collected using a breath-by-breath system (Zan 600 Oxi USB, Zan Messgeräte, Oberthulba, Germany). This spirometric system was calibrated prior to each test, following the manufacturer’s recommendations. To determine uphill and horizontal-running _VO2max, an incremental ramp testing protocol was performed at horizontal (1% incline) and uphill (8% incline) conditions (lab visit 1 & 2). Adapted from previous research with similar _VO2max values (Baumgartner et al., 2021), the initial velocity for both ramp tests was set based on prior running experience and estimated 10 km race Frontiers in Physiology frontiersin.org 02 Held et al. 10.3389/fphys.2023.1117314 time for each participant individually at 2, 2.5, or 3 m/s. The ramp protocol then consisted of 0.2% increases every 30 s until the participant reached exhaustion (Midgley et al., 2007). All participants were verbally encouraged and motivated in the same way towards the end of each test. The highest consecutive oxygen uptake values within 30 s during the final part of the ramp tests were considered as _VO2max. For both conditions, _VO2max and objective exhaustion were verified for each participant following the corresponding criteria (Midgley et al., 2007). All participants fulfilled these objective exhaustion criteria (i.e., at least 4 out of 6 criteria). Adapted from previous research, the quality of both HIIT sessions were defined by mean _VO2 and accumulated training time ≥90% _VO2max (Time90) (Midgley et al., 2006; Thevenet et al., 2007; Turnes et al., 2016). Since both HIIT sessions were time matched with the same work to rest ratio, mean _VO2 and Time90 were determined based on the entire training session (interval with pauses). Furthermore, to determine Time90, the entire training session (interval with pauses) was normalized to seconds, subsequently seconds with _VO2 value ≥ _VO2max were summed up. Thereby, the highest _VO2max value of the horizontal or incline ramp test was used as reference values. Furthermore, peak oxygen consumption (highest oxygen uptake during the intervals averaged over 30 s; _VO2peak) during both HIIT protocols was additionally considered. Apart from this, total respiration per minute (minute volume), respiratory frequency (breath frequency), and tidal volume were also used for further data analysis. In addition, capillary blood samples were taken from the earlobe of the participants for lactate analysis (EBIOplus; EKF Diagnostic Sales, Magdeburg, Germany), heart rate (HR) was measured using a heart rate strap (Polar, Kempele, Finland), and perceived exertion levels were assessed based on RPE (CR- 10 scale) (Foster et al., 2001) prior to the first interval and immediately after each running interval. 2.3 Statistics Data are presented as means ± standard deviations. Normal distribution was initially tested using Shapiro-Wilk tests (p ≥ 0.1). Variance homogeneity was visually confirmed via plotting sampled residuals vs. theoretical (ideal) residuals (Kozak and Piepho, 2018). Sphericity was verified via Mauchly´s tests. To examine mode differences (horizontal vs. uphill) for the respective outcome measures ( _VO2, _VO2peak, _VO2max, Time 90, minute volume, breath frequency, and tidal volume), numerous separate two-way (mode: horizontal vs. uphill) repeated measurement analysis of variances (rANOVA) were conducted. 2 (mode: horizontal vs. uphill) × 4 (time: pre vs. interval 1 vs. interval 2 vs. interval 3 vs. interval 4) rANOVAs were calculated for lactate, HR, and RPE, and running velocity data. rANOVA effect sizes are given as partial eta squared (pes) with ≥0.01, ≥0.06, and ≥0.14 indicating small, moderate, and large effects, respectively (Cohen, 1988). In case of significant mode × time interaction effects, Bonferroni post hoc tests were subsequently computed. For pairwise effect size comparison, standard mean differences (SMD) were additionally calculated as the differences between means divided by the pooled standard deviations (trivial: SMD <0.2; small: 0.2 ≤ SMD <0.5; moderate: 0.5 ≤ SMD <0.8; large SMD ≥0.8) (Cohen, 1988). Furthermore, the smallest worthwhile change was calculated as 30% of baseline standard deviation (Hopkins, 2004). Pearson correlation coefficients were calculated in order to define the relationships of the measured variables. A correlation coefficient of | r | ≈ 0.30 is interpreted as low/weak correlation, | r | ≈ 0.50 is interpreted as mean/moderate correlation and | r | ≈ 0.80 is interpreted as large/ strong correlation (Cohen, 1988). Statistical analyses were conducted using R (version 4.0.5) and RStudio (version 1.4.1106) software. 3 Results 3.1 Incremental ramp test No significant differences (p = 0.100; pes = 0.100; mean difference (MD) = 0.2 ± 0.5 L/min; SMD = 0.28) were found between horizontal (3.9 ± 0.7 L/min) and uphill _VO2max (4.1 ± 0.7 L/min) during the incremental ramp tests. 3.2 HIIT sessions rANOVA revealed significant effects (p ≤ 0.012; pes ≥0.351) regarding _VO2, _VO2peak, Time90, minute volume, breath frequency, and tidal volume (Figure 1). Thereby, uphill HIIT showed higher values than horizontal HIIT for _VO2mean (3.3 ± 0.6 vs. 3.2 ± 0.5 L/min; MD = 0.1 ± 0.1 L/min; SMD = 0.15), _VO2peak (4.0 ± 0.7 vs. 3.8 ± 0.7 L/min; MD = 0.1 ± 0.2 L/min; SMD = 0.19), Time90 (9.1 ± 4.6 vs. 6.4 ± 4.0 min; MD = 2.7 ± 2.7 L/ min; SMD = 0.62), and tidal volume (2144 ± 511 vs. 2061 ± 502 ml; MD = 83 ± 117 ml; SMD = 0.16). In contrast, uphill HIIT revealed lower values than horizontal HIIT for minute volume (94.3 ± 15.1 vs. 101.2 ± 17.3 L/min; MD = 6.9 ± 8.4 L/min; SMD = 0.43) and breath frequency (44.9 ± 6.0 vs. 50.5 ± 9.2 breaths/min, MD = 5.6 ± 5.9 breaths/min; SMD = 0.73). Furthermore, only for Time90, breath FIGURE 1 Mean difference (MD ± standard deviation) between horizontal and uphill high intensity training protocols for mean oxygen consumption ( _VO2), peak oxygen consumption ( _VO2peak), and accumulated time above 90% of maximal oxygen consumption (Time90). Smallest worthwhile change (SWC) boundaries are marked in grey. Significance levels (p) and pairwise effect sizes as standard mean differences (SMD) are presented. Frontiers in Physiology frontiersin.org 03 Held et al. 10.3389/fphys.2023.1117314 frequency and minute volume, the differences between conditions exceeded the smallest worthwhile change. Furthermore, Time90 revealed high (r = 0.82) and significant (p < 0.001) correlations between horizontal and uphill HIIT. No significant mode × time rANOVA interaction effects (p ≥ 0.097; pes ≤0.14) for lactate, HR, RPE and running velocity were found (Figure 2). Nevertheless, running velocity revealed significant time effects (p ≤ 0.001). Subsequently performed post hoc tests (p ≤ 0.001; SMD ≥3.53) revealed higher running velocity during horizontal HIIT (4.47 ± 0.33 to 4.51 ± 0.35 m/s) compared to uphill HIIT (3.17 ± 0.18 to 3.18 ± 0.21 m/s) during all intervals. 4 Discussion To the best of our knowledge, this is the first acute randomized controlled crossover study that examined _VO2, lactate, HR, and RPE response of time- and effort-matched horizontal vs. uphill HIIT running in well-trained runners. Our key findings were increased mean _VO2, _VO2peak, and accumulated training time ≥90% _VO2max via uphill HIIT compared to horizontal HIIT. In contrast, lactate, HR, and RPE revealed no significant differences between horizontal and uphill HIIT protocols. Furthermore, horizontal and uphill ramp tests yielded similar _VO2max values. A higher acute oxygen consumption during uphill running is commonly explained by the fact that the use of elastic energy may be compromised, so that in turn more mechanical energy (i.e., greater concentric muscle activity) needs to be generated, in order to lift the body’s center of gravity upward and subsequently overcome the slope (Snyder and Farley, 2011). Thus, in the present study, uphill running during a HIIT session notably increased the mean time ≥90% _VO2max by about 42%. Interestingly, this percentage increase is quite similar to previous cycling-related research, which used power-output variation within the work intervals (Bossi et al., 2020). In this previous study, two different interval training sessions, matched for duration and mean power output (6 × 5 min at a mean intensity of 84% of maximal aerobic power (MAP), with 2.5 min of rest between intervals), were performed. By performing several 30s bouts at 100% MAP within these intervals to increase the power- output variation within the work intervals, the mean time ≥90% _VO2max increased by about 43% (Bossi et al., 2020). It thus seems that variation of the power-output by performing short bouts of sprinting or by employing inclination might be an important factor to increase the time ≥90% _VO2max during HIIT sessions. In addition, and in line with our findings, lactate, HR, and RPE data reported by Bossi and colleagues (Bossi et al., 2020) were similar for both interval training conditions. However, both studies only focused on short-term effects. Therefore, Bossi and colleagues (Bossi et al., 2020) emphasized the need for longitudinal studies while speculating that performance adaptations will most likely be superior to constant-intensity work intervals. Based on our data, a 6- week period of uphill HIIT (2 sessions per week) would result in about half an hour more accumulated time ≥90% _VO2max compared to horizontal HIIT. This additional accumulated time ≥90% _VO2max via uphill HIIT is equivalent to 5 horizontal FIGURE 2 Lactate (A), heart rate (B), RPE (C), and running velocity (D) data (mean ± standard deviation) of horizontal (grey) and uphill (black) high intensity training protocols. Individual values are marked as points. In addition, p-values of time*mode interaction effects (p) of the repeated measurement variance analyses (rANOVA) and corresponding effect sizes as partial eta squared (pes) are given. Frontiers in Physiology frontiersin.org 04 Held et al. 10.3389/fphys.2023.1117314 HIIT sessions. Therefore, superior performance adaptations could be assumed via uphill HIIT. This assumption is supported by increased _VO2max and power output at the lactate threshold adaptations over a 4-week training period, if recreationally- trained cyclists spent about 100s more time above 90% _VO2max per training session (Turnes et al., 2016). In line with these findings, the accumulated training time ≥90% _VO2max is frequently considered a highly important marker for efficient HIIT sessions designed to increase _VO2max (Midgley et al., 2006; Thevenet et al., 2007; Turnes et al., 2016). Our findings of HIIT protocols performed at the maximal sustainable intensity during all four interval bouts (isoeffort) (Seiler and Hetlelid, 2005) revealed increased mean _VO2, _VO2peak, and accumulated time above 90% _VO2max at a decreased running velocity during the uphill HIIT condition and similar lactate, HR, and RPE values. However, as at a given speed, uphill running results in higher _VO2, lactate, HR, and RPE data compared to horizontal running (Minetti et al., 1994; 2002; Vernillo et al., 2017), it might be possible that the maximum oxygen uptake differs between running uphill compared to level running conditions. Nevertheless, we did not find significant differences in _VO2max in the initial incremental ramp tests performed at horizontal running condition and 8% slope. This is in line with results reported by Lemire and colleagues (Lemire et al., 2020) who reported similar _VO2max values in well-trained trail runners performing step tests on a treadmill in level and 15% uphill running conditions. However, a different study conducted in well-trained trail runners comparing the physiological responses to step tests with increasing gradient reported significantly higher _VO2max values at gradients of 40% compared to level running (Cassirame et al., 2022). This has also been described by Margaria and colleagues (Margaria et al., 1963): According to their work, when running on positive gradients up to 15% incline the minimum energy cost of running increases as a function of the incline. At slopes above 20%, however, the energy cost becomes equal to that of concentric muscular work (Minetti et al., 2002). It therefore seems, that at least in special populations (i.e., trail runners) and at very steep inclination (i.e., above 15%) the maximal oxygen uptake might significantly and relevantly differ from level running. Hence, this should be taken into account, when quantifying training load as a percentage value of the maximal oxygen uptake. Previous research revealed that 19%–79% of runners report musculoskeletal injuries of the lower extremities annually (van Gent et al., 2007). Thereby, loading rate and ground reaction force were repeatedly named as relevant risk factors (Crowell and Davis, 2011; Zadpoor and Nikooyan, 2011; Futrell et al., 2018). These relationships, however, were often established based on retrospective, cross-sectional data. More recently, in prospective case control-studies comprising recreational (Malisoux et al., 2022) and collegiate cross country runners (Kliethermes et al., 2021), the vertical impact peak and loading rate were not associated with a higher injury rate. Nevertheless, reducing loading rates is still recommended as an effective means to reduce the risk of developing running injuries (Bowser et al., 2018). In this context, uphill running revealed decreased ground reaction force data compared to horizontal running (Gottschall and Kram, 2005). Furthermore, we observed decreased running velocities during uphill HIIT compared to horizontal HIIT, which additionally decrease loading rate and ground reaction force (Keller et al., 1996). In detail, previous research revealed a 22%–39% ground reaction force decrease via an 6%–9% slope increase (Gottschall and Kram, 2005; Kowalski and Li, 2016). Furthermore, slower running resulted in reduced ground reaction force (Keller et al., 1996). Based on our running velocity differences between horizontal and uphill HIIT, this would result in a ground reaction force reduction of 11%. For the present study a possible reduction of loading rates remains, however, speculative, as these loading rates and ground reaction forces were not measured. Thus, more adequately powered prospective studies are necessary to investigate the association of musculoskeletal injuries of the lower extremities and loading rate as well as the potential prevention effect of uphill running. Horizontal running has been linked to the stretch-shortening cycle of the muscle-tendon unit of the lower limb (Schöffl et al., 2021), in which part of the mechanical energy of the center of mass (COM) is absorbed during the negative work phase to be restored during the next positive work phase (Nicol et al., 2006). This storage and release of kinetic and potential energy contributes to the acceleration of the body upwards during the propulsive phase and to the reduction of the energy production needed during the concentric phase (Snyder and Farley, 2011; Snyder et al., 2012). In contrast, during uphill running, the center of mass needs to be propelled vertically and does not oscillate around an equilibrium (Dewolf et al., 2016). In detail, the center of mass loses horizontal while simultaneously gaining vertical velocity during the first part of ground contact. Subsequently, during the second part of the contact, a fraction of the energy stored in the elastic elements of the muscle tendon unit is released to increase the kinetic and potential of the center of mass (Dewolf et al., 2016). Accordingly, differences in muscle activation patterns of the lower extremities have been reported between horizontal and uphill running (Yokozawa et al., 2007), with concentric muscle work being dominant during uphill running (Giandolini et al., 2016). Furthermore, to increase the running velocity in flat running conditions, athletes tend to increase their stride length and frequency almost linearly (Ito et al., 1983; Cavanagh and Kram, 1989; Brisswalter and Legros, 1995). Simultaneously, the floor contact time and flight time are reduced (Ito et al., 1983; Cavanagh and Kram, 1989; Brisswalter and Legros, 1995). Even though this pattern is also visible during uphill running compared to flat running, stride length and flight time are significantly reduced, since the foot touches the belt or ground earlier (Padulo et al., 2012; 2013). As the floor contact time does not seem to differ between flat and uphill running, this subsequently leads to a significant reduction in flight time during the uphill running condition (Padulo et al., 2012; 2013). Therefore, it seems possible, that prolonged training sessions running uphill might change the athlete’s kinematics, thus resulting in a reduction in running economy at horizontal conditions. Nevertheless, at least for constant running velocities, experienced athletes select an individual combination of stride length and frequency resulting in the least energy cost (Cavanagh and Kram, 1989; Cavagna et al., 1991), while providing the greatest mechanical efficiency (Morgan et al., 1994). Even though only a small fraction of the overall training time is spent on high-intensity running (Stöggl and Sperlich, 2015), a Frontiers in Physiology frontiersin.org 05 Held et al. 10.3389/fphys.2023.1117314 potential longitudinal effect on running economy induced by prolonged uphill running should be addressed in further research. A limitation that needs to be addressed is the lack of spatiotemporal running parameters including information on stride length and frequency. Thus, further research should try to disentangle the relationship between spatiotemporal running parameters and oxygen uptake during uphill running. In addition, the potential long-term training effects mentioned above should be examined in appropriate longitudinal intervention studies. In conclusion, this randomized crossover testing revealed increased mean _VO2, _VO2peak, and accumulated training time ≥90% _VO2max via uphill HIIT. Thus, uphill running during HIIT sessions appears to be an effective alternative to traditional horizontal HIIT sessions. Whether performance adaptations will be superior to horizontal running work intervals remains to be established by a longitudinal study, but similar lactate, HR, and RPE data suggest that it is unlikely that negative training outcomes occur. Nevertheless, future research should investigate whether training-induced adaptations can be improved via uphill HIIT. Furthermore, such further studies should also examine if different muscle activation patterns via uphill running (Giandolini et al., 2016) lead to adverse effects in terms of (horizontal) running economy. Data availability statement The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. Ethics statement The studies involving human participants were reviewed and approved by Ethical committee of the German Sport University Cologne (approval no. 153/2022). The patients/participants provided their written informed consent to participate in this study. Author contributions SH, RG, and LD contributed to the conception and design of the study. RG, TW, and JD led the intervention. LR, SH, and TW performed the statistical analysis. SH wrote the first draft of the manuscript. LR, TW, PW, and LD wrote sections of the manuscript. PW copyedited the draft for content, language, and format, and organized the submission and revision/resubmission process. All authors contributed to the article and approved the submitted version. Funding We acknowledge the financial support of the German Research Foundation (DFG) and the Open Access Publication Fund of Bielefeld University for the article processing charge. Acknowledgments We appreciatively acknowledge Jonas Hochstrate for his support during the data acquisition phase. Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Publisher’s note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. References Baumgartner, T., Held, S., Klatt, S., and Donath, L. (2021). Limitations of foot-worn sensors for assessing running power. Sensors (Basel) 21, 4952. doi:10.3390/s21154952 Bossi, A. H., Mesquida, C., Passfield, L., Rønnestad, B. R., and Hopker, J. G. (2020). Optimizing interval training through power-output variation within the work intervals. Int. J. Sports Physiol. Perform. 15, 982–989. doi:10.1123/ijspp.2019-0260 Bowser, B. J., Fellin, R., Milner, C. E., Pohl, M. B., and Davis, I. S. (2018). Reducing impact loading in runners: A one-year follow-up. Med. Sci. Sports Exerc 50, 2500–2506. doi:10.1249/MSS.0000000000001710 Brisswalter, J., and Legros, P. (1995). Use of energy cost and variability in stride length to assess an optimal running adaptation. Percept. Mot. Ski. 80, 99–104. doi:10.2466/pms. 1995.80.1.99 Buchheit, M., and Laursen, P. B. (2013). High-intensity interval training, solutions to the programming puzzle: Part I: Cardiopulmonary emphasis. Sports Med. 43, 313–338. doi:10.1007/s40279-013-0029-x Cassirame, J., Godin, A., Chamoux, M., Doucende, G., and Mourot, L. (2022). Physiological implication of slope gradient during incremental running test. Int. J. Environ. Res. Public Health 19, 12210. doi:10.3390/ijerph191912210 Cavagna, G. A., Willems, P. A., Franzetti, P., and Detrembleur, C. (1991). The two power limits conditioning step frequency in human running. J. Physiol. 437, 95–108. doi:10.1113/jphysiol.1991.sp018586 Cavanagh, P. R., and Kram, R. (1989). Stride length in distance running: Velocity, body dimensions, and added mass effects. Med. Sci. Sports Exerc 21, 467–479. doi:10. 1249/00005768-198908000-00020 Cohen, J. (1988). Statistical power analysis for the behavioral sciences. Hillside: Lawrence Earlbaum Associates. Crowell, H. P., and Davis, I. S. (2011). Gait retraining to reduce lower extremity loading in runners. Clin. Biomech. (Bristol, Avon) 26, 78–83. doi:10.1016/j.clinbiomech. 2010.09.003 Daoud, A. I., Geissler, G. J., Wang, F., Saretsky, J., Daoud, Y. A., and Lieberman, D. E. (2012). Foot strike and injury rates in endurance runners: A retrospective study. Med. Sci. Sports Exerc 44, 1325–1334. doi:10.1249/MSS.0b013e3182465115 Dewolf, A. H., Peñailillo, L. E., and Willems, P. A. (2016). The rebound of the body during uphill and downhill running at different speeds. J. Exp. Biol. 219, 2276–2288. doi:10.1242/jeb.142976 Foster, C., Florhaug, J., Franklin, J., Gottschall, L., Hrovatin, L., Parker, S., et al. (2001). A new approach to monitoring exercise training. J. Strength Cond. Res. 15, 109–115. doi:10.1519/1533-4287(2001)015<0109:anatme>2.0.co;2 Futrell, E. E., Jamison, S. T., Tenforde, A. S., and Davis, I. S. (2018). Relationships between habitual cadence, footstrike, and vertical load rates in runners. Med. Sci. Sports Exerc 50, 1837–1841. doi:10.1249/MSS.0000000000001629 Frontiers in Physiology frontiersin.org 06 Held et al. 10.3389/fphys.2023.1117314 Giandolini, M., Vernillo, G., Samozino, P., Horvais, N., Edwards, W. B., Morin, J.-B., et al. (2016). Fatigue associated with prolonged graded running. Eur. J. Appl. Physiol. 116, 1859–1873. doi:10.1007/s00421-016-3437-4 Gottschall, J. S., and Kram, R. (2005). Ground reaction forces during downhill and uphill running. J. Biomech. 38, 445–452. doi:10.1016/j.jbiomech.2004.04.023 Hopkins, W. (2004). How to interpret changes in an athletic performance test. Sportscience, 1–7. Available at: https://www.sportsci.org/jour/04/wghtests.htm. Ito, A., Komi, P. V., Sjödin, B., Bosco, C., and Karlsson, J. (1983). Mechanical efficiency of positive work in running at different speeds. Med. Sci. Sports Exerc 15, 299–308. doi:10.1249/00005768-198315040-00009 Joyner, M. J., and Coyle, E. F. (2008). Endurance exercise performance: The physiology of champions. J. Physiol. 586, 35–44. doi:10.1113/jphysiol.2007.143834 Keller, T. S., Weisberger, A. M., Ray, J. L., Hasan, S. S., Shiavi, R. G., and Spengler, D. M. (1996). Relationship between vertical ground reaction force and speed during walking, slow jogging, and running. Clin. Biomech. (Bristol, Avon) 11, 253–259. doi:10.1016/0268-0033(95)00068-2 Kliethermes, S. A., Stiffler-Joachim, M. R., Wille, C. M., Sanfilippo, J. L., Zavala, P., and Heiderscheit, B. C. (2021). Lower step rate is associated with a higher risk of bone stress injury: A prospective study of collegiate cross country runners. Br. J. Sports Med. 55, 851–856. doi:10.1136/bjsports-2020-103833 Kowalski, E., and Li, J. X. (2016). Lower limb joint angles and ground reaction forces in forefoot strike and rearfoot strike runners during overground downhill and uphill running. Sports Biomech. 15, 497–512. doi:10.1080/14763141.2016.1185458 Kozak, M., and Piepho, H.-P. (2018). What’s normal anyway? Residual plots are more telling than significance tests when checking ANOVA assumptions. J. Agron. Crop Sci. 204, 86–98. doi:10.1111/jac.12220 Laursen, P. B., and Jenkins, D. G. (2002). The scientific basis for high-intensity interval training: Optimising training programmes and maximising performance in highly trained endurance athletes. Sports Med. 32, 53–73. doi:10.2165/00007256- 200232010-00003 Lemire, M., Falbriard, M., Aminian, K., Pavlik, E., Millet, G. P., and Meyer, F. (2022a). Correspondence between values of vertical loading rate and oxygen consumption during inclined running. Sports Med. - Open 8, 114. doi:10.1186/s40798-022-00491-2 Lemire, M., Hureau, T. J., Remetter, R., Geny, B., Kouassi, B. Y. L., Lonsdorfer, E., et al. (2020). Trail runners cannot reach _VO2max during a maximal incremental downhill test. Med. Sci. Sports Exerc 52, 1135–1143. doi:10.1249/MSS.0000000000002240 Lemire, M., Remetter, R., Hureau, T. J., Geny, B., Lonsdorfer, E., Favret, F., et al. (2022b). Energy cost of running in well-trained athletes: Toward slope-dependent factors. Int. J. Sports Physiol. Perform. 17, 423–431. doi:10.1123/ijspp.2021-0047 Macera, C. A., Pate, R. R., Powell, K. E., Jackson, K. L., Kendrick, J. S., and Craven, T. E. (1989). Predicting lower-extremity injuries among habitual runners. Arch. Intern Med. 149, 2565–2568. doi:10.1001/archinte.149.11.2565 Malisoux, L., Gette, P., Delattre, N., Urhausen, A., and Theisen, D. (2022). Spatiotemporal and ground-reaction force characteristics as risk factors for running- related injury: A secondary analysis of a randomized trial including 800+ recreational runners. Am. J. Sports Med. 50, 537–544. doi:10.1177/03635465211063909 Margaria, R., Cerretelli, P., Aghemo, P., and Sassi, G. (1963). Energy cost of running. J. Appl. Physiol. 18, 367–370. doi:10.1152/jappl.1963.18.2.367 Midgley, A. W., McNaughton, L. R., Polman, R., and Marchant, D. (2007). Criteria for determination of maximal oxygen uptake: A brief critique and recommendations for future research. Sports Med. 37, 1019–1028. doi:10.2165/00007256-200737120-00002 Midgley, A. W., McNaughton, L. R., and Wilkinson, M. (2006). Is there an optimal training intensity for enhancing the maximal oxygen uptake of distance runners?: Empirical research findings, current opinions, physiological rationale and practical recommendations. Sports Med. 36, 117–132. doi:10.2165/00007256-200636020-00003 Minetti, A. E., Ardigò, L. P., and Saibene, F. (1994). Mechanical determinants of the minimum energy cost of gradient running in humans. J. Exp. Biol. 195, 211–225. doi:10. 1242/jeb.195.1.211 Minetti, A. E., Moia, C., Roi, G. S., Susta, D., and Ferretti, G. (2002). Energy cost of walking and running at extreme uphill and downhill slopes. J. Appl. Physiol. 93, 1039–1046. doi:10.1152/japplphysiol.01177.2001 Morgan, D., Martin, P., Craib, M., Caruso, C., Clifton, R., and Hopewell, R. (1994). Effect of step length optimization on the aerobic demand of running. J. Appl. Physiol. 77, 245–251. doi:10.1152/jappl.1994.77.1.245 Nicol, C., Avela, J., and Komi, P. V. (2006). The stretch-shortening cycle: A model to study naturally occurring neuromuscular fatigue. Sports Med. 36, 977–999. doi:10.2165/ 00007256-200636110-00004 Padulo, J., Annino, G., Migliaccio, G. M., Dʼottavio, S., and Tihanyi, J. (2012). Kinematics of running at different slopes and speeds. J. Strength Cond. Res. 26, 1331–1339. doi:10.1519/JSC.0b013e318231aafa Padulo, J., Powell, D., Milia, R., and Ardigò, L. P. (2013). A paradigm of uphill running. PLoS One 8, e69006. doi:10.1371/journal.pone.0069006 Rønnestad, B. R., Bakken, T. A., Thyli, V., Hansen, J., Ellefsen, S., and Hammarstrøm, D. (2022). Increasing oxygen uptake in cross-country skiers by speed variation in work intervals. Int. J. Sports Physiol. Perform. 17, 384–390. doi:10.1123/ijspp.2021-0226 Rønnestad, B. R., Rømer, T., and Hansen, J. (2019). Increasing oxygen uptake in well- trained cross-country skiers during work intervals with a fast start. Int. J. Sports Physiol. Perform. 1, 383–389. doi:10.1123/ijspp.2018-0360 Schöffl, I., Jasinski, D., Ehrlich, B., Dittrich, S., and Schöffl, V. (2021). Outdoor uphill exercise testing for trail runners, a more suitable method? J. Hum. Kinet. 79, 123–133. doi:10.2478/hukin-2021-0066 Seiler, S., and Hetlelid, K. J. (2005). The impact of rest duration on work intensity and RPE during interval training. Med. Sci. Sports Exerc 37, 1601–1607. doi:10.1249/01.mss. 0000177560.18014.d8 Seiler, S. (2010). What is best practice for training intensity and duration distribution in endurance athletes? Int. J. sports physiology Perform. 5, 276–291. doi:10.1123/ijspp.5. 3.276 Snyder, K. L., and Farley, C. T. (2011). Energetically optimal stride frequency in running: The effects of incline and decline. J. Exp. Biol. 214, 2089–2095. doi:10.1242/jeb. 053157 Snyder, K. L., Kram, R., and Gottschall, J. S. (2012). The role of elastic energy storage and recovery in downhill and uphill running. J. Exp. Biol. 215, 2283–2287. doi:10.1242/ jeb.066332 Stöggl, T. L., and Sperlich, B. (2015). The training intensity distribution among well- trained and elite endurance athletes. Front. Physiology 6. Available at: https://www. frontiersin.org/articles/10.3389/fphys.2015.00295 (Accessed September 14, 2022). Thevenet, D., Tardieu-Berger, M., Berthoin, S., and Prioux, J. (2007). Influence of recovery mode (passive vs. active) on time spent at maximal oxygen uptake during an intermittent session in young and endurance-trained athletes. Eur. J. Appl. Physiol. 99, 133–142. doi:10.1007/s00421-006-0327-1 Turnes, T., de Aguiar, R. A., Cruz, R. S. de O., and Caputo, F. (2016). Interval training in the boundaries of severe domain: Effects on aerobic parameters. Eur. J. Appl. Physiol. 116, 161–169. doi:10.1007/s00421-015-3263-0 van Gent, R. N., Siem, D., van Middelkoop, M., van Os, A. G., Bierma-Zeinstra, S. M. A., and Koes, B. W. (2007). Incidence and determinants of lower extremity running injuries in long distance runners: A systematic review. Br. J. Sports Med. 41, 469–480. discussion 480. doi:10.1136/bjsm.2006.033548 Vernillo, G., Giandolini, M., Edwards, W. B., Morin, J.-B., Samozino, P., Horvais, N., et al. (2017). Biomechanics and physiology of uphill and downhill running. Sports Med. 47, 615–629. doi:10.1007/s40279-016-0605-y Walter, S. D., Hart, L. E., McIntosh, J. M., and Sutton, J. R. (1989). The Ontario cohort study of running-related injuries. Arch. Intern Med. 149, 2561–2564. doi:10.1001/ archinte.149.11.2561 Wenger, H. A., and Bell, G. J. (1986). The interactions of intensity, frequency and duration of exercise training in altering cardiorespiratory fitness. Sports Med. 3, 346–356. doi:10.2165/00007256-198603050-00004 Yokozawa, T., Fujii, N., and Ae, M. (2007). Muscle activities of the lower limb during level and uphill running. J. Biomechanics 40, 3467–3475. doi:10.1016/j.jbiomech.2007. 05.028 Zadpoor, A. A., and Nikooyan, A. A. (2011). The relationship between lower- extremity stress fractures and the ground reaction force: A systematic review. Clin. Biomech. (Bristol, Avon) 26, 23–28. doi:10.1016/j.clinbiomech.2010.08.005 Frontiers in Physiology frontiersin.org 07 Held et al. 10.3389/fphys.2023.1117314

Increased oxygen uptake in well-trained runners during uphill high intensity running intervals: A randomized crossover testing.

02-16-2023

Held, Steffen,Rappelt, Ludwig,Giesen, René,Wiedenmann, Tim,Deutsch, Jan-Philip,Wicker, Pamela,Donath, Lars

eng

PMC4338213

COLLECTION REVIEW Injuries in Runners; A Systematic Review on Risk Factors and Sex Differences Maarten P. van der Worp1,2,3*, Dominique S. M. ten Haaf2, Robert van Cingel2,4, Anton de Wijer1,5, Maria W. G. Nijhuis-van der Sanden3,6, J. Bart Staal2,3 1 Academic Institute, University of Applied Sciences Utrecht, Department of Physical Therapy, Utrecht, the Netherlands, 2 HAN, University of Applied Sciences Nijmegen, Institute Health Studies, Nijmegen, the Netherlands, 3 Radboud University Medical Center, Radboud Institute for Health Science, Scientific Institute for Quality of Healthcare, Nijmegen, the Netherlands, 4 Sport Medical Center Papendal, Arnhem, the Netherlands, 5 Radboud University Medical Center, Radboud Institute for Health Science, Department of Oral Function & Prosthetic Dentistry, Nijmegen, the Netherlands, 6 Radboud University Medical Center, Radboud Institute for Health Science, Department of Rehabilitation, Nijmegen, the Netherlands * MvdWorp@Gmail.com Abstract Background The popularity of running continues to increase, which means that the incidence of running- related injuries will probably also continue to increase. Little is known about risk factors for running injuries and whether they are sex-specific. Objectives The aim of this study was to review information about risk factors and sex-specific differ- ences for running-induced injuries in adults. Search Strategy The databases PubMed, EMBASE, CINAHL and Psych-INFO were searched for relevant articles. Selection Criteria Longitudinal cohort studies with a minimal follow-up of 1 month that investigated the associ- ation between risk factors (personal factors, running/training factors and/or health and life- style factors) and the occurrence of lower limb injuries in runners were included. Data Collection and Analysis Two reviewers’ independently selected relevant articles from those identified by the system- atic search and assessed the risk of bias of the included studies. The strength of the evi- dence was determined using a best-evidence rating system. Sex differences in risk were determined by calculating the sex ratio for risk factors (the risk factor for women divided by the risk factor for men). PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 1 / 18 OPEN ACCESS Citation: van der Worp MP, ten Haaf DSM, van Cingel R, de Wijer A, Nijhuis-van der Sanden MWG, Staal JB (2015) Injuries in Runners; A Systematic Review on Risk Factors and Sex Differences. PLoS ONE 10(2): e0114937. doi:10.1371/journal. pone.0114937 Academic Editor: Amir A. Zadpoor, Delft University of Technology (TUDelft), NETHERLANDS Published: February 23, 2015 Copyright: © 2015 van der Worp et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This study was financially supported by the Netherlands Organization for Health Research and Development (ZonMw), grant no. 50-50310-98-156. Competing Interests: The authors have declared that no competing interests exist. Main Results Of 400 articles retrieved, 15 longitudinal studies were included, of which 11 were considered high-quality studies and 4 moderate-quality studies. Overall, women were at lower risk than men for sustaining running-related injuries. Strong and moderate evidence was found that a history of previous injury and of having used orthotics/inserts was associated with an in- creased risk of running injuries. Age, previous sports activity, running on a concrete surface, participating in a marathon, weekly running distance (30–39 miles) and wearing running shoes for 4 to 6 months were associated with a greater risk of injury in women than in men. A history of previous injuries, having a running experience of 0–2 years, restarting running, weekly running distance (20–29 miles) and having a running distance of more than 40 miles per week were associated with a greater risk of running-related injury in men than in women. Conclusions Previous injury and use of orthotic/inserts are risk factors for running injuries. There ap- peared to be differences in the risk profile of men and women, but as few studies presented results for men and women separately, the results should be interpreted with caution. Fur- ther research should attempt to minimize methodological bias by paying attention to recall bias for running injuries, follow-up time, and the participation rate of the identified target group. Introduction Although running has been popular since the 1970s [1], the number of runners and running events has increased steadily since 2000 [1,2]. This increase is largely due to girls and women who started running [2,3]. Running in the adult population is one of the most popular physical activities around the world and in the Western society many cities have their own recreational running event. Furthermore, running is one of the most efficient ways to achieve physical fit- ness, which is linked with longevity [1]. A drawback of the sport is the relatively high risk of in- jury, with an incidence varying between 19% and 79% [4]. This large variation is due to differences in the definition of injury, study populations, and follow-up periods [5]. Injuries di- minish pleasure in exercise and lead to a temporary or even permanent discontinuation of run- ning. Injuries furthermore lead to increased costs because of necessary medical treatment (e.g., the direct medical costs per injured runner at the emergency department is estimated at €1300 [6]), and/or absence from work. In conclusion, running is very popular in the adult population, however strategies are needed to prevent high incidences of running injuries in this group of runners. Acute running injuries are rare, consisting mainly of muscle injuries, sprain, or skin lesions (blisters and abrasions) [7]. Eighty percent of running disorders are overuse injuries, resulting from a mismatch between the resilience of the connective and supporting tissue and running [7]. Running is one of the most common sports that give rise to overuse injuries of lower back and the leg [8]. The predominant site of leg injuries is the knee, for which the location specific incidence ranged from 7.2% to 50.0% [4]. Running injuries of the lower leg, foot and upper leg are common, ranging from 9.0% to 32.2%, 5.7% to 39.3%, and 3.4% to 38.1%, respectively [4]. Less common sites of running are the ankle, the hip/pelvis/groin and lower back, ranging from 3.9% to 16.6%, 3.3% to 11.5% and 5.3% to 19.1 respectively [4,9–11]. Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 2 / 18 Poorly perfused tissues, such as ligaments, tendons and cartilage, are particularly at risk be- cause they adapt more slowly than muscles to increased mechanical load [7]. Hreljac [8] sug- gested that injury should be avoided not by minimizing the stress applied to a biological structure but by optimizing the amount and frequency of loading stress. Given the dynamic na- ture of the relationship between applied stress and injury, there must be an optimal level of ap- plied stress for any biological structure [8]. Furthermore, the multifactorial model of Meeuwisse et al. showed the importance of identify- ing predisposing factors that make a runner susceptible for injury [12]. Identifying such factors may contribute to the development of injury prevention strategies [13], especially when these can be influenced by adequate training or by optimizing training environment. Moreover, the exact causes of running injuries are likely to be diverse [4] and possibly interacting with each other [13]. Risk factors for running injuries can be clustered into three domains, 1) personal factors (e.g. age, sex, height, genetic imprinting), 2) running/training factors (e.g. weekly running days, distance, running shoes), and 3) health and lifestyle related factors (e.g. smoking, a history of comorbidity and previous injuries) [4]. Three narrative reviews [5,14,15], published in 1992, reported the occurrence of injuries to be based on multifactorial risk factors. In their systematic review, Van Gent et al. (2007) [4] found limited evidence that older age [16], differences in lower leg length [17], a larger left tubercle-sulcus angle [17] and greater knee varus [17], greater height (in men) [18], use of alcohol [16], and a positive medical history (e.g. taken medication, high blood pressure, asthma, and nervous or emotional problems) [16] are associated with a higher risk of injury in men and women. Strong evidence was found that previous injuries were associated with lower extremity running injuries [4], but the studies used different definitions of previous injury, in terms of its location, time of occurrence, etc. Also, the recent systematic review of Saragiotto et al. [19] confirmed that previous injuries are a risk factor for new run- ning injuries and no association between sex and running injuries was found in most of the in- cluded studies. In this systematic review [19] only prospective cohort studies were included and risk factors for general running-related injuries were determined. However, no distinction was made in the risk factors for specific running related injuries, e.g. medial tibial stress syn- drome, Iliotibial band syndrome, etc. Differences in the health status of women and men are of increasing concern to European health policymakers and are becoming a subject of growing interest to researchers [20]. The injury patterns between men and women differ and there are several reasons for the dif- ferences in injury rates, related to anatomic and physiologic differences [21]. Two recent Dutch prospective studies of novice runners [10,22] pinpointed at possible dif- ferences in injury risk profiles of men and women. In a study of runners (n = 629) who were preparing for a 6.7-km run, a younger age and lack of running experience were significant risk factors for running injuries in men, whereas lack of running experience, a higher body mass index (BMI), and earlier participation in sports without axial pressure (swimming and cycling) were risk factors for running injuries in women [10]. A subsequent study of a different cohort of novice runners (n = 532) also showed sex-specific risk factors, but the results were contradic- tory: significant risk factors for men were previous injuries in the past year, higher BMI, and earlier participation in sports without axial pressure, whereas in women a positive navicular drop test was the sole risk factor in adjusted analyses [22]. However, the statistical analysis used in these two studies, stepwise multiple regression, is questionable [23] and more research is needed to clarify the sex difference in risk profile. A previous study from Canada also reported sex differences in risk factors for running injuries. A BMI of > 26 kg/m2 was reported as protective in men, whereas age younger than 31 years was protective in women; running once a week and age older than 50 years were risk factors in women [24]. Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 3 / 18 From the above, it can be appreciated that it is difficult to draw conclusions about the risk factors for running injuries in general, for specific running injuries, and possible differences in risk profile between men and women. Earlier reviews [4,5,14,15] need to be updated to identify all possible factors that may predispose a runner for injury and enabling future researchers to develop, potentially sex-specific, interventions to prevent running-related injuries [13]. The present literature synthesis aims to review current evidence for risk factors for running- associated injuries in adults and to determine whether risk factors for such injuries differ be- tween men and women. Methods We used the MOOSE statement to report our systematic review of observational studies and the STARLITE statement to report our literature search [25,26]. Search Strategy Four bibliographical databases, namely, CINAHL (1982 to 26 December, 2012), EMBASE (1947 to 1 January, 2013), PubMed (1940 to 26 December, 2012), and Psych INFO (1806–1 January, 2013), were searched using search strings developed by the first author and the librari- an expert (AT) of the Radboud University Nijmegen Medical Center. The following search terms (Mesh, title- and/or abstract words) were used to identify the study population in combi- nation with lower extremity injuries: running, track and field, jogging and lower limb, lower ex- tremity, leg-, hip-, knee-, ankle- and foot injuries, soft tissue injuries, musculoskeletal pain, bursitis, sprains and strains, tendinopathy, tendinitis, Iliotibial band syndrome, patellofemoral pain syndrome, and plantar fasciitis. Keywords used to identify a relevant study design were cohort studies, longitudinal studies, follow-up, retrospective-, observational-, prospective stud- ies, risk factors, and etiology. For the PubMed search, see S1 Appendix. The search strings of the other databases are available upon request from the authors. Selection Criteria Two reviewers (MvdW & JS) independently selected relevant articles, based on titles and ab- stracts. Full papers were retrieved if the abstract provided insufficient information to decide whether the article should be included. The selection criteria were: 1) the design indicated a longitudinal cohort study with a minimal follow-up of 1 month; 2) the objective of the study was to investigate the association between risk factors (personal factors, running/training fac- tors and/or health and lifestyle factors) and the occurrence of lower limb injuries; 3) the study population consisted of novice runners, long-distance runners both recreational and/or com- petitive; 4) the article was published in a peer-reviewed journal in English or German. Studies concerning elite, professional or ultra-runners, patient populations, children, and/or young ad- olescents (age <18 years), or in which participants were predominantly exposed to other types of sporting activity than running (e.g. military training, triathlon, etc.) were excluded. If a study contained a mixed population of runners and patients, the results for the runners had to be presented separately in order for the study to be included. The reference lists of all identified relevant publications were checked for other relevant publications. Quality Assessment Articles that met the selection criteria were evaluated for risk of bias. A quality list of twelve items, based on assessment tools of the Cochrane Collaboration [27] and previous systematic reviews of risk factors for musculoskeletal disorders [28–30], was used. The list was based on Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 4 / 18 generally accepted principles of etiological research and was relevant for cohort studies. Some items were adapted to the topic of interest of this review by replacing risk factors with personal factors, running/training factors, and/or health & lifestyle factors (see S1 Table). Two reviewers (DtH & MvdW) independently assessed the quality of the studies. All items were scored as positive, negative, or unclear. A positive score indicated a well-described and well-performed item. A negative score indicated that the item was described but not well per- formed, and unclear meant the item was unclear because insufficient information was avail- able. For each item, the scores of the two reviewers were compared. Any difference in scoring was resolved in a consensus meeting. If consensus was not reached, a third reviewer (AW) made the final decision. A high-quality study was defined as scoring positive on > 50% of the items [28–30]. Data Extraction and Statistical Analysis The following information was extracted from the included studies: year of publication, study design with follow-up period, injury definition, population characteristics (age, sex, body mass index, or height and weight, and the proportion of subjects analyzed in the included studies; number of subjects analyzed, divided by the number of included subjects, multiplied with 100) and the incidence of (running) injuries; injury specific or overall and, if given, sex specific. Cohen’s Kappa (K) values were calculated for the interobserver agreement between the two reviewers with regard to risk of bias. A Kappa value of > 0.8 indicates high level of agreement between assessors, a value between 0.61 and 0.8 a substantial agreement, a value between 0.41 and 0.6 a moderate level of agreement, and a value of < 0.41 poor level of agreement [31]. SPSS 20.0 was used to calculate K values. The main dependent outcome variable was running-induced leg injury. Identified risk fac- tors were summarized per injury, overall and injury specific. All risk factors were grouped into three main categories: 1) personal factors, 2) running/training related factors, and 3) health & lifestyle related factors. To evaluate associations between risk factors and running injuries p-values, crude odds ra- tios (ORs), hazard ratios (HRs,) and relative risks (RRs) with 95% confidence intervals (CI) were retrieved from the included publications. Crude values were used for this evaluation to prevent biases and shortcomings of stepwise multiple regression analyses [23]. Adjusted risk estimates derived from multivariable regression analyses were only used when the independent variables of the model were pre-specified and not based on a stepwise selection algorithm or when crude associations were not available. Pooling and Best-Evidence Synthesis Separate meta-analyses with the random effects model [32] were planned to obtain the pooled OR, HR or pooled RR (with 95% CI) for running injuries. If pooling was not possible due to heterogeneity of the study populations, a best evidence synthesis was presented. For each identified risk factor, levels of evidence were established for the association be- tween this factor and the occurrence of running injuries. These levels of evidence were based on the guidelines of van Tulder et al. [33] and were divided into the following levels: strong evi- dence, defined as consistent findings (in
75% of the studies) in multiple (
2) high-quality studies; moderate evidence, defined as consistent findings (in
75% of the studies) in one high-quality study and multiple low-quality studies; limited evidence, defined as consistent findings (in
75% of the studies) in multiple low-quality studies or one high-quality study; and conflicting evidence, defined as conflicting findings reported by <75% of the studies re- porting consistent findings. Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 5 / 18 Sex Ratio In studies in which risk factors were presented separately for men and women, possible sex dif- ferences in risk were determined by dividing the risk factor for women by the risk factor for men, which produced a sex ratio. A ratio higher than 1.25 (i.e., women had a higher risk) or lower than 0.75 (i.e. women had a lower risk) was regarded as a relevant sex difference [34,35]. Results Literature Search A flow chart for article retrieval is given in S1 Fig. Of 400 articles retrieved as potentially rele- vant, 17 were considered eligible for full-text screening based on title and abstract. Of these 17 studies, 2 [36,37] seemed, after consultation with the authors, to have an abstract only, so 15 ar- ticles were included for quality assessment, data extraction, and analysis. Included Studies Of the 15 included longitudinal cohort studies, 13 were prospective [10,17,22,24,38–46] and 2 retrospective [9,47] studies. They were all published in English. The follow-up time of these studies ranged from 8 weeks to 1 year and the mean age of study participants ranged from 36 to 44 years. Thirteen studies had a mixed population, 1 study included only women [39], 1 study included only men [42], and 1 study did not report the sex of the study population [47]. BMI and height differed between the various reports. In the study of Bennett et al. [38], 13.6% of study participants had a low BMI (<18.5 kg/m2); the studies of Lun et al. [44] and McKean et al. [47] did not report BMI, weight, or height. The proportion of subject analyzed in the orig- inal studies ranged from 46% to 100%. Seven studies [10,22,24,39–41,43] included novice run- ners. All studies used different (running) definitions of injury, except for one research group who used the same definition in their two studies [9,10,17,22]. Five studies defined running-re- lated injuries as involving the lower limb [38,40,41,43,46], 4 studies included the influence of the symptoms on running [9,17,24,47], and 6 studies defined injuries in terms of the lower limb and the influence of symptoms [10,22,39,42,44,45]. Two studies included a specific time frame of running restriction caused by the running injury [10,22]. Four studies specifically looked at signs and symptoms related to Achilles tendinopathy [41,46] and patellofemoral dys- function/pain syndrome [39,43]. Only Bennett et al. [38] excluded traumatic injuries. Wen et al. [9,17] included overuse injuries in their definition of injury. Wen et al. [9,17] investigated the same experienced runners in a retrospective study [9] and in a longitudinal prospective study, published a year later [17]. In order to avoid duplication, the results of these two studies were considered as coming from one study population [48]. The incidence of the running inju- ries reported in the included studies where in the range of 20.6% to 79.3%, 25.0% to 79.5% and 19.8% to 79.1% for overall, men and women, respectively. The injury specific incidences were 7.8% and 14.3% for Achilles tendinopathy injuries [41,46] and 16.7% and 20.8% for patellofe- moral pain injuries [39,43]. S2 Table presents a summary of these studies including the popula- tion characteristics (age, sex, BMI, and the proportion of people analyzed), type of running, injury definition and (running) incidence; injury specific and/or overall and, if given, sex specific. Risk of Bias The overall agreement between the two reviewers was 77% with a moderate reliability (Kappa = 0.6). The agreement for the individual items ranged from 53% (item 12) to 100% (item 6). Most disagree- ment was seen for item 5 (“Were the data on system factors, running/training related factors, and/or Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 6 / 18 health and lifestyle factors collected using standardized methods of acceptable quality?”), item 7 (“Were the data on outcome collected using standardized methods of acceptable quality?”), and item 12 (“Positive, if the number of cases in the final multivariable was at least ten times the number of independent variables in the analysis.”), because of the different interpretation of the definitions of “standardized methods”, “acceptable quality”, and by miscalculating/interpretation of the number of cases in the final multivariable, respectively. Other disagreements were mostly due to differences in interpretation. All disagreements were resolved in a consensus meeting. Nine of the 13 prospective cohort studies [10,17,22,24,38,42,44–46] were considered to be high quality (> 6 items positive), as were the 2 retrospective cohort studies [9,47] (S3 Table). Risk Factors for Running Injuries; Overall and Injury Specific The heterogeneity in study populations, in operationalization of both outcomes and risk fac- tors, and time to follow-up prevented us from following a formal meta-analytical approach. Study populations varied from novice runners to recreational runner and competitive runners, outcomes from running-related injuries, overall injuries to lower leg overuse injuries and more localized injuries, e.g. Achilles Tendinopathy, back injuries (S4–S6 Tables). Follow-up time points varied from 8 weeks to 1 year (S2 Table). Across the studies different categories of inde- pendent variables were used with different cut-off points (S4–S6 Tables) or injured versus in- jured runners were compared using continuous values of risk factors (e.g. the mean age of injured runners was higher than the mean age of non-injured runners [46]). For these reasons we refrained from doing a meta-analysis. We therefore choose to present the results using a best evidence synthesis. Risk factors were divided into three categories: personal factors, run- ning/training related factors and health and lifestyle factors (see S4–S6 Tables). Personal Factors; S4 Table Sex. One low quality study [40] and five high quality studies [10,22,38,46,47] assessed sex as risk factor for running injuries. One high-quality studies [22] found men to have a significantly higher risk of running-related injuries than women, and particularly younger men (< 40 years) [47]. Thus there was limited evidence that men are at higher risk of running-related injuries. Age. Four low-quality studies [39–41,43] and four high-quality studies [9,17,44,46] investi- gated the relationship between age and running injuries. Only one study found age to have a significant effect on running injuries: Wen et al. [17] showed that lower age was significantly protective against overall (not specified) overuse injury. Thus there was only limited evidence that lower age affects the risk of running-related injuries. Wen et al. [9] and Hirschmüller et al. [46] found higher age to be a significant risk factor for hamstrings injuries and midportion Achilles tendinopathy, respectively. This indicates that there is limited evidence that age affects the risk of hamstrings injuries and midportion Achilles tendinopathy. BMI. Three low-quality studies [39,41,43] and three high-quality studies [9,38,46] exam- ined BMI as a risk factor for running injuries. BMI was not found to have significant effect on injury risk in runners overall, but Wen et al. [9] found a higher BMI to be a risk factor for back injuries in women and a lower BMI to be a risk factor for foot injuries in men. Thus there was limited evidence that BMI is a risk factor for back injuries in women and for foot injuries in men. Height. Four low-quality studies [39–41,43] and three high-quality studies [9,17,46] investi- gated height as a risk factor for running injuries. Wen et al. [9] found lower height in men to be a significant risk factor for foot injuries, indicating limited evidence. Weight. Three low quality studies [39–41,43] and three high-quality [9,17,46] study investi- gated weight as a risk factor for running injuries. Wen et al. [9] found higher weight in women Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 7 / 18 and lower weight men to be a risk factor for back injuries and foot injuries, respectively. In the same research group, Wen et al. [17] found higher weight to be protective against foot injuries. Thus there was limited evidence that higher weight in women and lower weight in men were risk factors for back and foot injuries, respectively. Furthermore, there was limited evidence that a heavier weight protects against foot injuries. Navicular drop. One high-quality study [38] investigated the influence of navicular drop on running injuries. Bennett et al. [38] found runners with a high navicular drop (>10 mm) in the left or right foot were at greater risk for medial exercise-related leg pain. Also, a navicular drop of more than 10 mm in only the left foot was significantly associated with a higher risk of medi- al exercise-related leg pain. Thus there was limited evidence that navicular drop (> 10 mm) is a risk factor for running injuries. Intratendinous blood flow. Only one high-quality study [46] investigated the influence of blood flow in the Achilles tendon on Achilles tendinopathy in runners. Runners with intraten- dinous microvessels (indicating primary neovascularization) were at greater risk of mid-por- tion Achilles tendinopathy. Thus there was limited evidence that impaired intratendinous blood flow is a risk factor for running injuries. Force distribution pattern. Three low-quality studies [40,41,43] investigated force distribu- tion patterns in relation to running injuries. Hesar et al. [40] found significantly less laterally directed force distribution at first metatarsal contact and forefoot flat, and significantly more medial directed force displacement in the forefoot contact phase, foot flat phase, and heel-off phase in runners without lower leg overuse injuries. These individuals also had a significantly quicker change in the center of force (COF) at forefoot flat, a lower force and loading under- neath the lateral border of the foot, and a significantly lower directed force displacement of the COF at forefoot flat than did runners with lower leg injuries. Van Ginkel et al. [41] found a sig- nificant decrease in the total posterior–anterior displacement of the COF and a laterally direct- ed force distribution underneath the forefoot at ‘forefoot flat’ as intrinsic gait-related risk factors for Achilles tendinopathy in novice runners. Thijs et al. [43] demonstrated that runners with a significantly higher vertical peak force underneath the second metatarsal and shorter time to the vertical peak force underneath the lateral heel were at higher risk for patellofemoral pain syndrome. In conclusion, there was limited evidence that a number of force distribution factors/patterns are risk factors for, or protective against, lower leg injuries, Achilles tendinopa- thy, and patellofemoral pain in runners [40,41,43]. Alignment. Three high-quality studies [9,17,44] investigated the influence of alignment on the occurrence of running injuries. In their prospective study, Wen et al. [17] found that run- ners in the group with the highest combined arch index were protective against, and runners in the group with the lowest leg difference were at higher risk for running injuries, respectively. In the retrospective study by the same research group [9], runners in the groups with the lowest left tubercle-sulcus angle and lowest combined (mean left and right) tubercle-sulcus angle were found to be at higher risk for ankle injuries. Runners in the groups with the lowest heel valgus, the highest heel valgus, and highest right arch index were found to be protective against knee injuries [17]. In this same prospective study, runners in the group with the highest left tuber- cle-sulcus angle and highest knee valgus were found to be significant at risk for shin injuries [17]. In subgroup analyses of this study, the highest heel valgus group was significant protective against foot injuries (expressed as injury incidence per 1000 miles running, or as injury inci- dence per 1000 hours running) [17]. In conclusion, there was limited evidence that a small dif- ference in leg length is a risk factor for overall running injuries. There was also limited evidence that a large left tubercle-sulcus angle and a large knee varus are risk factors for shin in- juries. Furthermore there was limited evidence that a low left tubercle-sulcus angle and Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 8 / 18 combined (average of left and right) tubercle-sulcus angle are risk factors for ankle injuries and that several alignment factors are protective against running injuries. Running & Training Related Factors for Running Injuries; S5 Table Running experience. Five high-quality studies investigated the relationship between running experience and running injuries [9,17,42,46,47]. Limited evidence was found that more run- ning experience was a risk factor for overall running injuries [17]. There was also limited evi- dence that running with less (< 1 year) experience was protective for running injuries [47]. Limited evidence was found that more running experience was a risk factor for knee [42] and foot injuries [17]. Training. Five high-quality studies investigated the relationship between training factors and running injuries [9,17,42,46,47]. The prospective study of Wen et al. [17] found increased hours of running per week to be protective against overall injuries (expressed in terms of inci- dence per mileage or hours run). There was limited evidence that age < 40 years combined with running
6 times a week was a significant risk factor for running injury [47], as there was for age
40 years combined with running
6 times a week [47]. There was also limited evidence that age  40 years combined with running 1–3 times a week and running < 10 miles per week were significant protective factors for running injury [47], and an age
40 years combined with running 1–3 times a week was protective [47]. Van Middelkoop et al. [42] found that interval training was protective against knee injury in men. In contrast, the two high quality studies by Wen et al. [9,17] found more interval training to be a risk factor for shin injuries. The evidence for interval training being a risk or protective factor was limited. There was also limited evidence that increasing hours of running per week is protective against knee and foot injuries [17] and that a slower training pace was a risk factor for heel injuries [9]. Surface. Only one high-quality study [9] investigated the relationship between surface and running injuries. There was limited evidence that running time on concrete surface is protec- tive against back and thigh injuries [9]. Distance. Four high-quality studies [9,42,44,46] analyzed running distance as independent variable for running injuries. There was limited evidence that higher weekly mileage is associat- ed with hip and hamstrings injuries [9] and that a training distance of 0–40 km a week is pro- tective against the incidence of calf injuries [42]. Race participation. One high-quality study [42] (= limited evidence) found the risk of run- ning injuries to be higher in men who had participated in more than six races in the last year. Shoe use. Two high-quality studies [9,17] analyzed the relationship between shoe use and running injuries. There was limited evidence that changing shoes more frequently was a risk factor for overall injuries [9] and limited evidence for using one pair of running shoes or alter- nating between two pairs versus alternating between more than two pairs of shoes as a risk fac- tor for knee injuries [9]. Furthermore, limited evidence was found for a higher number of shoes as a risk factor for shin injuries [17]. Health & Life-Factors Related for Running Injuries; S6 Table History of previous injury. Four high-quality studies [17,38,42,46] investigated the relation- ship between running injuries and previous injuries. Bennett et al. [38] found that runners with a history of exercise-related leg pain for a month or a year were at greater risk of a relapse of ex- ercise-related leg pain. Wen et al. [17] also found previous injuries to be a risk factor for run- ning injuries. In the high-quality study of Van Middelkoop et al. [42], lower extremity injury in Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 9 / 18 the previous 12 months was found to be a risk factor for running injury in men. In conclusion, there was strong evidence that previous injury is a risk factor for running injuries. Van Middelkoop et al. [42] found that a lower extremity injury in the previous 12 months was a risk factor for a knee injury, and that an injury at another location (hip, groin, thigh, knee, ankle, or/and foot) was a risk factor for calf injury. None of the other studies identified risk factors for knee and/or calf injury. Bennett et al. [38] found that runners with a history of medial exercise-related leg pain lasting longer than 1 month were at greater risk of medial exer- cise-related leg pain. A history of old shin injuries was found to be a risk factor for shin injuries in one high-quality study [17]. A previous disorder of the Achilles tendon was a significant risk factor for midportion Achilles tendinopathy in one high-quality study [46]. In conclusion, there was limited evidence that previous injury is a risk factor for specific running injuries, namely, medial exercise-related leg pain, midportion Achilles tendinopathy, shin injuries, knee and calf injuries. Orthotic/inserts. Two high-quality studies [9,47] investigated orthotic/inserts as a risk fac- tor for running injuries. Both found wearing orthotics or using shoe inserts to be a risk factor for running injuries (moderate evidence). Wen et al. [9] found the use of shoe insert to be a risk factor for foot injuries, indicating limited evidence for this association. Sex Ratio Five high-quality studies [10,22,24,45,47] analyzed data for men and women separately (see S7 Table). One study showed women to be at significantly lower risk of injuries overall than men [22]. Two studies showed men with a history of injury were at higher risk of running injuries than women with a similar history [22,45]. One high-quality study found the risk of injury to be higher in women than men if the women were older [10], had previously engaged in other sports activities [10], had the previous year participated in a marathon [45], had a weekly dis- tance running of 48–63.8 km for the preceding 3 months [45], ran on concrete surface [45], and had running shoes that were 4- to 6-months old [24], with sex ratios of 1.4, 1.9, 2.0, 2.2, 4.2, and 4.9, respectively. Men were, in comparison with women, at greater risk of injury if they restarted running [10], had less than 2 years’ running experience [45], had a weekly running distance of 32–47.8 km [45] or had a weekly running distance > 64 km [45], with a sex ratio of 0.7, 0.7, 0.7 and 0.4, respectively. Discussion The purpose of this study was to synthesize current evidence on determinants of running-in- duced injuries of the leg in adults and to determine sex differences in risk profile for running injuries. We found strong and moderate evidence that previous leg injury and use of orthotics/ inserts increase the risk of leg injuries, respectively. Furthermore, there was only limited (one high-quality study) or no (one/two low-quality studies) evidence for other potential risk factors for running injuries (overall and injury specific). Analysis of the sex ratios showed that women are at lower risk of running injuries than men. Factors that increased the risk of running-related injuries in women were older age, previ- ous participation in non-axial sports (e.g. cycling, swimming, etc.), participating last year in a marathon, running on concrete, a longer weekly running distance (48–63.8 km) and wearing running shoes for 4 to 6 months. Men were at greater risk of such injuries if they restarted run- ning, had a history of previous injuries, a running experience of 0–2 years, had a weekly run- ning distance between 32–47.8 km, and having a weekly running distance more than 64 km per week. Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 10 / 18 Running injuries have a multifactorial origin that can be subdivided into personal, running/ training, and health and/or lifestyle factors [5,14,15]. These factors can reinforce each other and their influence may also be mediated by cultural or societal factors [49]. The importance of each factor, and hence its contribution to the risk of symptoms and injuries, varies among individuals and running environments. Personal factors investigated in this review focused on sex, age, an- thropometric, and biomechanical factors; psychosocial factors were not investigated as risk fac- tors for running injuries. Psychosocial factors seem to have a role in musculoskeletal disorders [49–51] and thus future studies should investigate their role in running-related leg injuries. Most running injuries are due to overuse [7], but only Wen et al. [9,17] and Bennett et al. [38] included or excluded overuse/acute injuries in their definition of injury, respectively. Overuse injuries of the musculoskeletal system generally occur when a structure is repeatedly exposed to loading forces. Forces lower than the threshold associated with acute injury ulti- mately lead to fatigue of that specific structure [52,53]. There is no standard definition of over- use running injury [8,54], but it should minimally include a musculoskeletal ailment that can be attributed to running and that causes a restriction of running speed, distance, duration, or frequency for at least a week [8]. Of the articles included in our review, that of Buist et al. [10,22] used definitions “any musculoskeletal pain of the lower limb or back causing a restric- tion in running for at least one day [10] or one week [22]” that matches the most with these cri- teria. The other studies did not define the period during which injury restricted running. Future research should use the definition of running injuries used by Buist et al. [10,22] or in- clude a minimal time frame of running restriction when defining running-related injuries. To our knowledge, this is the third review that systematically examined risk factors for run- ning injuries. Five reviews of running injuries have been published in the past [4,5,14,15,19], and three of these narrative studies were published more than 20 years ago [5,14,15]. The most recent systematic reviews were published in 2007 [4] and 2014 [19]. Van Gent et al. [4] found strong evidence that a long training distance per week in men and previous injuries were risk fac- tors for injuries; however, a long training distance per week was a protective factor for knee inju- ries. Although we also found previous injury to be a risk factor for running-related injuries, the variety in the other results can be explained by differences in the studies included. Seventeen arti- cles, dating from 1982 to 2006, were included [4]: 10 studies were published after 2006 and were therefore not included in the study of Van Gent et al. [4]. As we used a minimal follow-up time of 1 month and an age of >18 years as inclusion criteria, the studies of Walter et al. [18] and Sat- terthwaite et al. [16] were not included in our review. The finding of Van Gent et al. [4] that lon- ger training distance per week is a protective against knee injuries could not be confirmed because studies providing evidence for this association were not included in our review. The recent published review by Saragiotto et al. [19] included only prospective studies which mentioned running or runners in the abstract/title. Moreover, articles that studied risk factors for specific injuries (e.g. medial tibial stress syndrome) were excluded in their systematic review. Fur- thermore, Saragiotto et al. [19] included all categories of runners, this in contrast with our study population consisting of novice runners, long-distance runners, both recreational and/or com- petitive. In their study [19] also pooling of data was not possible due to the large heterogeneity of the statistical methods used across studies. However, although they did not perform a best evi- dence synthesis and used different inclusion and exclusion criteria, the conclusion that previous injury is a risk factor for running injuries was the same as in our study. Risk Factors for Running Injury We decided to classify the different risk factors for running injuries according to the existing literature of systematic reviews (personal, running/training, health and lifestyle) [4,14,15], to Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 11 / 18 facilitate comparison between the reviews. However, applying a public health approach to sports injury prevention as described by Finch [55], conceptualizing risk factors as modifiable and nonmodifiable provides additional insight [56]. Modifiable risk factors associated with running injuries provide the base for developing running injury prevention interventions, whereas nonmodifiable risk factors are important for risk stratification and targeted prevention [56]. Nonmodifiable Risk Factors for Running Injuries History of injury. Previous injury was consistently associated with running injuries and espe- cially in men. The lack of association between previous injury and running injuries in women might be because most of the included studies investigated female novice runners with minimal running experience and few injuries in the past [10,22,24,39–41,43]. It is not clear whether a high rate of re-injury is due to incomplete healing of the original in- jury, an uncorrected biomechanical problem, or recall bias and/or the definition of the injury. Previous lower extremity injuries that have healed completely (i.e., the return of full, pre-injury joint range of motion, musculoskeletal strength, and proprioception) should not increase the risk of a subsequent lower extremity injury [57]. However, injuries that give rise to permanent structural or biomechanical malfunction and/or dysfunctional coordination increase the risk of future running injuries [58]. In our review, three high-quality studies [22,42,45] found a history of previous leg injury to be a risk factor in men. However, the definition of “previous injury” differed in the various studies, in terms of its nature (e.g. acute or gradual onset), whether it is running related or not, when it occurred and how long it lasted. It is essential to know the ex- tent and characteristics of recovery from a previous injury [57]. Lastly, in most studies partici- pants were asked about injuries in the previous year, which means that recall bias could be a problem. In conclusion, previous (running) leg injury seems an important risk factor for running in- juries. Further research should focus on a clear definition of “previous (running) injury” and should more focus on recovery processes to judge the possibility of re-injury including the time of occurrence, and on minimizing recall bias by reducing the time frame of recall. Modifiable Risk Factors for Running Injuries Training. Overuse running injuries are suggested to be the result of training errors [8] and our results confirm this. On the basis of this review, it seems that the ideal training intensity has not yet been established. Runners with a high training frequency and/or running distance ap- peared to be more susceptible to overuse injuries, especially those runners who have no run- ning experience and, seemingly contradictory, runners who are experienced and who have run, perhaps long distances, for a longer time. Van Gent et al. [4] found strong evidence that men with a higher weekly training frequency were more prone to running injuries. However, run- ning only once a week could lead to overuse injuries, especially in women [24]. This is probably because running stresses the musculoskeletal system [8], which does not have time to adapt to this type of exercise because of the low frequency of running. In conclusion, overuse running injuries should be prevented by optimizing and personaliz- ing training, bearing in mind the (limited) evidence that running/training-related factors influ- ence the risk of injury. Orthotic/insert. Foot orthoses are widely used to treat existing pathological conditions and to prevent overuse injuries [59]. They function in two ways: 1) the insert acts as a cushion that absorbs shock transmitted to the lower limb, and 2) they compensate for biomechanical defi- ciencies of the foot, such as excessive pronation and differences in leg length [60]. Most Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 12 / 18 findings of this review contradict these statements. McKean et al. [47] and Wen et al. [9] showed that runners with orthotic/inserts were at higher risk of running injuries, although it is possible that runners who are more prone to injury are given orthotic/inserts earlier. However, given the findings about the role of the navicular drop [22], alignment [9,17], and force distri- bution [40,41] in running-related injuries, it is doubtful that compensating biomechanical defi- ciencies with an orthotic/insert is effective in preventing running injuries. In conclusion, orthotics/inserts do not seem useful to compensate for biomechanical deficiencies. Sex Differences Differences between the health of men and women are a major concern to European health au- thorities [20]. Only five high-quality studies [10,22,24,45,47] investigated the effect of runner’s sex on the risk of running injuries. However, given the small number of studies that investigat- ed this, it was not possible to establish sex-specific profiles for risk factors. Two high-quality studies investigated the relation between previous injury and running in- juries and presented data for men and women separately, so that it was possible to calculate a sex ratio. When the criteria of Van Tulder [33] were used to determine the level of evidence for sex differences, two studies [22,47] provided moderate evidence that men (< 40 year) had a higher risk of running-related injuries and two studies [22,45] provided moderate evidence that men had a higher risk of running-related injuries when having a previous injury; the other studies did not provide evidence of sex-related differences in risk of running injuries. However, physical therapists, sports physicians, etc. can provide sex-specific advice for the prevention of running injuries, and trainers and coaches can tailor their training advice to individual runners. More prospective longitudinal studies are necessary and should analyze data for men and women separately, in order to obtain evidence-based, sex-specific risk profiles [20,61]. Risk of Bias & Study Limitations As risk factors were operationalized as dichotomous, ordinal, or even continuous variables, it was not possible to calculate a meaningful pooled summary of outcomes. Moreover, conclu- sions made after data pooling might have been of limited value given the heterogeneity in defi- nition of running injury in the various studies. Quality scoring systems are used in an attempt to address possible methodological short- comings that could threaten the validity of study results [30]. We created our quality scale based on the lists used by the Cochrane Collaboration to assess cohort studies [27] and on lists used in previous studies [28–30]. One of these lists [29] was quantified by West et al. [62] in a study that evaluated quality-rating systems for observational studies. The scoring list of Ariëns et al. [29] scored positive on six and partially positive on one out of nine domains for assessing study quality [62]. While the usefulness of quality control is disputed [62] as it is difficult to de- termine how to weight each item in an overall quality score, sum scores are considered helpful in a systematic review for distinguishing between studies with a low or a high risk of bias [62,63]. We evaluated the quality of the included studies in order to gain insight into the risk of bias and therefore to enable us to draw meaningful conclusions. A point of concern is that many of the included studies did not clearly describe the participation rate of the target group, which limits the generalizability of findings [64]. This study had some limitations. All included studies, prospective and retrospective, were assessed using the same quality list. Because it would be better to adjust the list for a retrospec- tive design, a second quality analysis was done for the two retrospective studies reviewed [9,47], such that item 2 (“participation rate is at least 80% from the identified target group”) and 3 (“the participation rate at main moment of follow-up is at least 80% or the nonresponse Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 13 / 18 is not selective”) were scored as “not applicable” in the scoring list. This did not influence the quality score of these articles (both remained high quality), and therefore had no influence on the results of our best evidence syntheses. By our inclusion criteria (e.g. long-distance runners recreational and/or competitive) for se- lecting the original studies, a broad spectrum in the type of runners (novice, track and field, etc.) was selected. When the inclusion criteria were more strictly defined, our results could be presented stratified for each group of runners. However, the number of studies per type of run- ners would be too small to give useful information and by choosing a broader spectrum of type of runners, our results are more generalizable to the total adult running population. Although we performed an extensive literature search, it is likely that both selection and publication bias influenced the results. Future research, in which running injury is uniformly defined, may indicate whether the factors found in our review are true risk factors. Conclusion and Implications More high-quality studies of risk factors for running injuries are needed before strong conclu- sions can be drawn about the relevance of specific risk factors. Furthermore, consensus must be achieved about the definition of running injuries, and large cohort studies are needed to in- vestigate different types (biomechanical, hormonal, psychological, etc.) of risk factors with em- phasis on potential differences between men and women. To minimize bias, future studies should pay attention to recall of previous running injuries, follow-up time, and the participation rate. This review found strong evidence that previous leg injury is a risk factor for running-relat- ed leg injuries. Some sex-specific risk factors were identified, but not enough studies investigat- ed differences between men and women to obtain more definite results. Running injuries seem to have a multifactorial origin, but on the basis of our findings, ef- forts to prevent injury should focus on runners, especially men, with a history of running inju- ries and provide customized training and/or specific exercises. The use of orthotics/inserts should be discouraged. Supporting Information S1 PRISMA Checklist (DOCX) S1 Fig. Flow Chart of the search of articles. (TIFF) S2 Fig. PRISMA 2009 flow diagram. (DOCX) S1 Table. Quality assessment check list [27–30]. (DOCX) S2 Table. Study Characteristics. (DOCX) S3 Table. Results of the risk of bias assessment. (DOCX) S4 Table. Significant personal risk- & protective factors for running injuries. (DOCX) Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 14 / 18 S5 Table. Significant running/training-related risk factors for running injuries. (DOCX) S6 Table. Significant health & lifestyle factors related for running injuries. (DOCX) S7 Table. Risk factors for running injuries with sex ratio. (DOCX) S1 Appendix. Search terms PubMed. (DOCX) Acknowledgments This study was financially supported by the Netherlands Organization for Health Research and Development (ZonMw), grant no. 50-50310-98-156. The authors would like to acknowledge the following persons who made substantial contributions: Alice Tillema, information special- ist at Radboud University Nijmegen Medical Centre, Library, the Netherlands, who assisted with our extensive literature search and Petra Habets, Amsterdam Medical Centre, University of Amsterdam, the Netherlands for giving structural commentary on early versions of the manuscript. Author Contributions Analyzed the data: MVDW DTH. Contributed reagents/materials/analysis tools: MVDW DTH RVC ADW MNVDS JS. Wrote the paper: MVDW. Substantial contributions to concep- tion and design, acquisition of data: MVDW MNVDS JS. Analysis and interpretation of data: MVDW DTH RVC ADW MNVDS JS. Drafting the article or revising it critically for important intellectual content: MVDW DTH RVC ADW MNVDS JS. Final approval of the version to be published: MVDW DTH RVC ADW MNVDS JS. References 1. Fields KB, Sykes JC, Walker KM, Jackson JC (2010) Prevention of running injuries. Curr Sports Med Rep 9: 176–182. doi: 10.1249/JSR.0b013e3181de7ec5 PMID: 20463502 2. Bottenburg van M, Kalmthout van J, Meulen van der R, Nuijten S, Rijnen B, et al. (2006) De tweede loopgolf. over groei en omvang van de loopsportmarkt en hoe de KNAU haar marktaandeel verder kan vergroten. W.J.H. Mulier Instituut, 's Hertogenbosch, the Netherlands. 3. Lynch SL, Hoch AZ (2010) The female runner: Gender specifics. Clin Sports Med 29: 477–498. doi: 10. 1016/j.csm.2010.03.003 PMID: 20610034 4. van Gent R, Siem D, van Middelkoop M, van Os A, Bierma-Zeinstra S, et al. (2007) Incidence and de- terminants of lower extremity running injuries in long distance runners: A systematic review. Br J Sports Med 41: 469–480. PMID: 17473005 5. Hoeberigs JH (1992) Factors related to the incidence of running injuries. A review. Sports Med 13: 408–422. PMID: 1615258 6. VeiligheidNL (2012) Hardloopblessures: Blessurecijfers. Avalaible: http://www.veiligheid.nl/nieuws/ meer-sportblessures-door-beginnende-hardlopers/$file/Cijfersfactsheet%20Hardlopen.pdf. Accessed on 7/30/2014. 7. Walther M, Reuter I, Leonhard T, Engelhardt M (2005) Verletzungen und überlastungsreaktionen im laufsport. Orthopäde 34: 3999. doi: 10.5435/JAAOS-D-14-00433 PMID: 25624366 8. Hreljac A (2004) Impact and overuse injuries in runners. Med Sci Sports Exerc 36: 845–849. PMID: 15126720 9. Wen DY, Puffer JC, Schmalzried TP (1997) Lower extremity alignment and risk of overuse injuries in runners. Med Sci Sports Exerc 29: 1291–1298. PMID: 9346158 Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 15 / 18 10. Buist I, Bredeweg SW, Bessem B, van MW, Lemmink KA, et al. (2010) Incidence and risk factors of run- ning-related injuries during preparation for a 4-mile recreational running event. Br J Sports Med 44: 598–604. doi: 10.1136/bjsm.2007.044677 PMID: 18487252 11. Hespanhol Junior LC, Pena Costa LO, Lopes AD (2013) Previous injuries and some training character- istics predict running-related injuries in recreational runners: A prospective cohort study. J Physiother 59: 263–269. doi: 10.1016/S1836-9553(13)70203-0 PMID: 24287220 12. Meeuwisse WH (1994) Assessing causation in sport injury: A multifactorial model. Clin J Sport Med 4: 66. 13. Meeuwisse WH, Tyreman H, Hagel B, Emery C (2007) A dynamic model of etiology in sport injury: The recursive nature of risk and causation. Clin J Sport Med 17: 215–219. PMID: 17513916 14. Macera CA (1992) Lower extremity injuries in runners. advances in prediction. Sports Med 13: 50–57. PMID: 1553455 15. van Mechelen W (1992) Running injuries. A review of the epidemiological literature. Sports Med 14: 320–335. PMID: 1439399 16. Satterthwaite P, Norton R, Larmer P, Robinson E (1999) Risk factors for injuries and other health prob- lems sustained in a marathon. Br J Sports Med 33: 22–26. PMID: 10027053 17. Wen DY, Puffer JC, Schmalzried TP (1998) Injuries in runners: A prospective study of alignment. Clin J Sport Med 8: 187–194. PMID: 9762477 18. Walter SD, Hart LE, McIntosh JM, Sutton JR (1989) The ontario cohort study of running-related injuries. Arch Intern Med 149: 2561–2564. PMID: 2818114 19. Saragiotto BT, Yamato TP, Hespanhol Junior LC, Rainbow MJ, Davis IS, et al. (2014) What are the main risk factors for running-related injuries? Sports Med. 20. Ostrowska A (2012) Health inequalities—gender perspective. Przegl Lek 69: 61–66. PMID: 22768415 21. Boles CA, Ferguson C (2010) The female athlete. Radiol Clin North Am 48: 1249–1266. doi: 10.1016/j. rcl.2010.07.015 PMID: 21094409 22. Buist I, Bredeweg SW, Lemmink KA, van MW, Diercks RL (2010) Predictors of running-related injuries in novice runners enrolled in a systematic training program: A prospective cohort study. Am J Sports Med 38: 273–280. doi: 10.1177/0363546509347985 PMID: 19966104 23. Steyerberg EW, Eijkemans MJ, Habbema JD (1999) Stepwise selection in small data sets: A simulation study of bias in logistic regression analysis. J Clin Epidemiol 52: 935–942. PMID: 10513756 24. Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, et al. (2003) A prospective study of running injuries: The vancouver sun run "in training" clinics. Br J Sports Med 37: 239–244. PMID: 12782549 25. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, et al. (2000) Meta-analysis of observational studies in epidemiology: A proposal for reporting. meta-analysis of observational studies in epidemiolo- gy (MOOSE) group. JAMA 283: 2008–2012. PMID: 10789670 26. Booth A (2006) "Brimful of STARLITE": Toward standards for reporting literature searches. J Med Libr Assoc 94: 421–9, e205. PMID: 17082834 27. Dutch cochrane centre. Evidence-based guideline development. Form III: To assess cohort studies. Available: http://Dcc.cochrane.org. Accessed on 7/14/2013. 28. IJmker S, Huysmans MA, Blatter BM, van der Beek AJ, van Mechelen W, et al. (2007) Should office workers spend fewer hours at their computer? A systematic review of the literature. Occup Environ Med 64: 211–222. PMID: 17095550 29. Ariens GA, van Mechelen W, Bongers PM, Bouter LM, van der Wal G (2000) Physical risk factors for neck pain. Scand J Work Environ Health 26: 7–19. PMID: 10744172 30. Bongers PM, Kremer AM, ter Laak J (2002) Are psychosocial factors, risk factors for symptoms and signs of the shoulder, elbow, or hand/wrist?: A review of the epidemiological literature. Am J Ind Med 41: 315–342. PMID: 12071487 31. Landis JR, Koch GG (1977) An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics 33: 363–374. PMID: 884196 32. Zeegers MP, Heisterkamp SH, Kostense PJ, van der Windt DA, Scholten RJ (2000) Practice of system- atic reviews. VII. pooling of results from observational studies. Ned Tijdschr Geneeskd 144: 1393– 1397. PMID: 10923147 33. van Tulder M, Furlan A, Bombardier C, Bouter L, Editorial Board of the Cochrane Collaboration Back Review Group (2003) Updated method guidelines for systematic reviews in the cochrane collaboration back review group. Spine (Phila Pa 1976) 28: 1290–1299. Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 16 / 18 34. Hooftman WE, van Poppel MN, van der Beek AJ, Bongers PM, van Mechelen W (2004) Gender differ- ences in the relations between work-related physical and psychosocial risk factors and musculoskeletal complaints. Scand J Work Environ Health 30: 261–278. PMID: 15458009 35. Altman DG, Bland JM (2003) Interaction revisited: The difference between two estimates. BMJ 326: 219. PMID: 12543843 36. Buist I, Bredeweg SW (2011) Higher risk of injury in overweight novice runners. Br J Sports Med 45: 338–338. 37. Nielsen R, Ramskov D, Sorensen H, Lind M, Rasmussen S, et al. (2011) Protocol for the dano-run study: A 1-year observational follow up study on running related injuries in 1000 novice runners. Br J Sports Med 45: 365–366. 38. Bennett JE, Reinking MF, Rauh MJ (2012) The relationship between isotonic plantar flexor endurance, navicular drop, and exercise-related leg pain in a cohort of collegiate cross-country runners. Int J Sports Phys Ther 7: 267–278. PMID: 22666641 39. Thijs Y, Pattyn E, Van Tiggelen D, Rombaut L, Witvrouw E (2011) Is hip muscle weakness a predispos- ing factor for patellofemoral pain in female novice runners? A prospective study. Am J Sports Med 39: 1877–1882. doi: 10.1177/0363546511407617 PMID: 21632979 40. Hesar NG, Van Ginckel A, Cools A, Peersman W, Roosen P, et al. (2009) A prospective study on gait- related intrinsic risk factors for lower leg overuse injuries. Br J Sports Med 43: 1057–1061. doi: 10. 1136/bjsm.2008.055723 PMID: 19228665 41. Van Ginckel A, Thijs Y, Hesar NG, Mahieu N, De Clercq D, et al. (2009) Intrinsic gait-related risk factors for achilles tendinopathy in novice runners: A prospective study. Gait Posture 29: 387–391. doi: 10. 1016/j.gaitpost.2008.10.058 PMID: 19042130 42. Van Middelkoop M, Kolkman J, van Ochten J, Bierma-Zeinstra S, Koes BW (2008) Risk factors for lower extremity injuries among male marathon runners. Scand J Med Sci Sports 18: 691–697. doi: 10. 1111/j.1600-0838.2007.00768.x PMID: 18266787 43. Thijs Y, De CD, Roosen P, Witvrouw E (2008) Gait-related intrinsic risk factors for patellofemoral pain in novice recreational runners. Br J Sports Med 42: 466–471. doi: 10.1136/bjsm.2008.046649 PMID: 18397970 44. Lun V, Meeuwisse WH, Stergiou P, Stefanyshyn D (2004) Relation between running injury and static lower limb alignment in recreational runners. Br J Sports Med 38: 576–580. PMID: 15388542 45. Macera CA, Pate RR, Powell KE, Jackson KL, Kendrick JS, et al. (1989) Predicting lower-extremity in- juries among habitual runners. Arch Intern Med 149: 2565–2568. PMID: 2818115 46. Hirschmuller A, Frey V, Konstantinidis L, Baur H, Dickhuth H-H., et al. (2012) Prognostic value of achil- les tendon doppler sonography in asymptomatic runners. Med Sci Sports Exerc 44: 199–205. doi: 10. 1249/MSS.0b013e31822b7318 PMID: 21720278 47. McKean KA, Manson NA, Stanish WD (2006) Musculoskeletal injury in the masters runners. Clin J Sport Med 16: 149–154. PMID: 16603885 48. Senn SJ (2009) Overstating the evidence: Double counting in meta-analysis and related problems. BMC Med Res Methodol 9: 10-2288-9-10. doi: 10.1186/1471-2288-9-87 PMID: 20042092 49. Bongers PM, Ijmker S, van den Heuvel S, Blatter BM (2006) Epidemiology of work related neck and upper limb problems: Psychosocial and personal risk factors (part I) and effective interventions from a bio behavioural perspective (part II). J Occup Rehabil 16: 279–302. PMID: 16850279 50. Broniecki M, Esterman A, May E, Grantham H (2010) Musculoskeletal disorder prevalence and risk fac- tors in ambulance officers. J Back Musculoskelet Rehabil 23: 165–174. doi: 10.3233/BMR-2010-0265 PMID: 21079295 51. da Costa BR, Vieira ER (2010) Risk factors for work-related musculoskeletal disorders: A systematic review of recent longitudinal studies. Am J Ind Med 53: 285–323. doi: 10.1002/ajim.20750 PMID: 19753591 52. Hreljac A, Marshall RN, Hume PA (2000) Evaluation of lower extremity overuse injury potential in run- ners. Med Sci Sports Exerc 32: 1635–1641. PMID: 10994917 53. Stanish WD (1984) Overuse injuries in athletes: A perspective. Med Sci Sports Exerc 16: 1–7. PMID: 6708775 54. Rolf C (1995) Overuse injuries of the lower extremity in runners. Scand J Med Sci Sports 5: 181–190. PMID: 7552763 55. Finch C (2006) A new framework for research leading to sports injury prevention. J Sci Med Sport 9: 3– 9. PMID: 16616614 56. Cameron KL (2010) Commentary: Time for a paradigm shift in conceptualizing risk factors in sports in- jury research. J Athl Train 45: 58–60. doi: 10.4085/1062-6050-45.1.58 PMID: 20064049 Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 17 / 18 57. Hootman JM, Macera CA, Ainsworth BE, Martin M, Addy CL, et al. (2002) Predictors of lower extremity injury among recreationally active adults. Clin J Sport Med 12: 99–106. PMID: 11953556 58. Van der Worp MP, van der Horst N, de Wijer A, Backx FJ, Nijhuis-van der Sanden MW (2012) Iliotibial band syndrome in runners: A systematic review. Sports Med 42: 969–992. doi: 10.2165/11635400- 000000000-00000 PMID: 22994651 59. Finestone A, Novack V, Farfel A, Berg A, Amir H, et al. (2004) A prospective study of the effect of foot orthoses composition and fabrication on comfort and the incidence of overuse injuries. Foot Ankle Int 25: 462–466. PMID: 15319103 60. Mattila VM, Sillanpaa PJ, Salo T, Laine HJ, Maenpaa H, et al. (2011) Can orthotic insoles prevent lower limb overuse injuries? A randomized-controlled trial of 228 subjects. Scand J Med Sci Sports 21: 804– 808. doi: 10.1111/j.1600-0838.2010.01116.x PMID: 20492587 61. Borchers AT, Gershwin ME (2012) Sociological differences between women and men: Implications for autoimmunity. Autoimmunity Reviews 11: A413–A421. doi: 10.1016/j.autrev.2011.11.016 PMID: 22155202 62. West S, King V, Carey T, Lohr K, McKoy N, et al. (2002) Systems to rate the strength of scientific evi- dence. Rockville (MD): Agency for Healthcare Research and Quality (US). PMID: 25057650 63. Van Tulder MW, Suttorp M, Morton S, Bouter LM, Shekelle P (2009) Empirical evidence of an associa- tion between internal validity and effect size in randomized controlled trials of low-back pain. Spine (Phila Pa 1976) 34: 1685–1692. 64. Vandenbroucke JP, von Elm E, Altman DG, Gotzsche PC, Mulrow CD, et al. (2007) Strengthening the reporting of observational studies in epidemiology (STROBE): Explanation and elaboration. Epidemiol- ogy 18: 805–835. PMID: 18049195 Risk Factors and Sex Differences in Running Injuries PLOS ONE | DOI:10.1371/journal.pone.0114937 February 23, 2015 18 / 18

Injuries in runners; a systematic review on risk factors and sex differences.

02-23-2015

van der Worp, Maarten P,ten Haaf, Dominique S M,van Cingel, Robert,de Wijer, Anton,Nijhuis-van der Sanden, Maria W G,Staal, J Bart

eng

PMC7875396

RESEARCH ARTICLE The effect of changing foot progression angle using real-time visual feedback on rearfoot eversion during running Seyed Hamed MousaviID1,2,3*, Laurens van Kouwenhove1, Reza Rajabi2, Johannes Zwerver3,4, Juha M. Hijmans1 1 Department of Rehabilitation Medicine, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands, 2 Department of Health and Sport Medicine, Faculty of Physical Education and Sport Sciences, University of Tehran, Tehran, Iran, 3 Center for Human Movement Science, University Medical Center Groningen, University of Groningen, Groningen, The Netherlands, 4 Department of Sports Medicine, Gelderse Vallei Hospital, Ede, The Netherlands * s.h.mousavi@umcg.nl, mousavihamed84@yahoo.com Abstract Atypical rearfoot in/eversion may be an important risk factor for running-related injuries. Prominent interventions for atypical rearfoot eversion include foot orthoses, footwear, and taping but a modification derived from gait retraining to correct atypical rearfoot in/eversion is lacking. We aimed to investigate changes in rearfoot in/eversion, subtalar pronation, medial longitudinal arch angle, and selected lower limb joint biomechanics while performing toe-in/toe-out running using real-time visual feedback. Fifteen female runners participated in this study. Subjects performed toe-in/toe-out running using real-time visual feedback on foot progression angle, which was set ±5˚ from habitual foot progression angle. 3D kinematics of rearfoot in/eversion, subtalar supination/pronation, medial longitudinal arch angle, foot pro- gression angle, hip flexion, ab/adduction and internal/external rotation, knee flexion, ankle dorsiflexion, and ankle power were analyzed. A repeated-measures ANOVA followed by pairwise comparisons was used to analyze changes between three conditions. Toe-in run- ning compared to normal and toe-out running reduced peak rearfoot eversion (mean differ- ence (MD) with normal = 2.1˚; p<0.001, MD with toe-out = 3.5˚; p<0.001), peak pronation (MD with normal = -2.0˚; p<0.001, MD with toe-out = -3.4; p = <0.001), and peak medial lon- gitudinal arch angle (MD with normal = -0.7˚; p = 0.022, MD with toe-out = -0.9; p = 0.005). Toe-out running significantly increased these kinematic factors compared to normal and toe-in running. Toe-in running compared to normal running increased peak hip internal rota- tion (MD = 2.3; p<0.001), and reduced peak knee flexion (MD = 1.3; p = 0.014). Toe-out run- ning compared to normal running reduced peak hip internal rotation (MD = 2.5; p<0.001), peak hip ab/adduction (MD = 2.5; p<0.001), peak knee flexion (MD = 1.5; p = 0.003), peak ankle dorsiflexion (MD = 1.6; p<0.001), and peak ankle power (MD = 1.3; p = 0.001). Run- ners were able to change their foot progression angle when receiving real-time visual feed- back for foot progression angle. Toe-in/toe-out running altered rearfoot kinematics and medial longitudinal arch angle, therefore supporting the potential value of gait retraining focused on foot progression angle using real-time visual feedback when atypical rearfoot in/ PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 1 / 17 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Mousavi SH, van Kouwenhove L, Rajabi R, Zwerver J, Hijmans JM (2021) The effect of changing foot progression angle using real-time visual feedback on rearfoot eversion during running. PLoS ONE 16(2): e0246425. https://doi. org/10.1371/journal.pone.0246425 Editor: Nizam Uddin Ahamed, University of Pittsburgh, UNITED STATES Received: March 4, 2020 Accepted: January 19, 2021 Published: February 10, 2021 Copyright: © 2021 Mousavi et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Funding: The author(s) received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. eversion needs to be modified. It should be considered that changes in foot progression angle when running is accompanied by changes in lower limb joint biomechanics. Introduction Running-related injuries (RRIs) are very common in athletes; sports clinicians are frequently consulted about these injuries [1]. Abnormal kinematics as intrinsic risk factors are considered to have an important role in the high incidence of RRIs [2]. Rearfoot eversion is among the most commonly reported kinematic factors in the studies investigating foot function and/or risk factors for lower-limb injuries [2–4]. As the subtalar coordination axis is not aligned with the foot coordination axes, and besides, no anatomical landmark exists on the talus, rearfoot eversion is predominantly measured as a surrogate to describe subtalar pronation [5, 6]. Much debate exists on whether atypical rearfoot eversion contributes to injury [4, 7]. This is mainly because most studies investigating rearfoot eversion for RRIs have either a case-control or a cross-sectional design that cannot prove causality; the results of prospective studies are mainly based on a small sample size [8]. There are several studies reporting that rearfoot eversion may be a potential risk factor for RRIs [2, 9, 10]. In a recent systematic review [2] we showed that peak rearfoot eversion may be associated with iliotibial band syndrome, patellar tendinopathy, and posterior tibial tendon dysfunction in runners. Female runners with atypical rearfoot ever- sion may be more prone to RRIs, specifically female runners with tibial stress fracture showed greater peak rearfoot eversion [11] and female runners with iliotibial band syndrome showed lower peak rearfoot eversion compared to non-injured runners [2, 12]. Atypical medial longitudinal arch angle (MLAA) is another contributing factor predisposing athletes to musculoskeletal overuse injuries [13]. MLAA collapse causes the calcaneus to evert in relation to the tibia, resulting in rearfoot eversion [14]. Normal rearfoot eversion and MLAA are essential for optimal shock absorption of the foot during the stance phase of gait. Atypical rearfoot eversion and MLAA may influence: 1. distribution of weight through the lower extremity, increas- ing force to the medial aspects of the foot and reducing shock absorption and postural balance abilities; and 2. alignment of the lower-limb kinematic chain, such as knee and hip mechanics [15]. For instance, excessive rearfoot eversion may result in excessive tibia internal rotation and hip internal rotation. This can result in faulty knee flexion/extension biomechanics which in turn produce a compensatory reaction in the tibiofemoral joint and may subsequently lead to patellofe- moral symptoms [16]. Reduced shock absorption and malalignment are considered to play an important role in the development of running-related overuse injuries of the lower extremity [17]. In recent decades there has been increasing clinical and scientific interest in modifying atypical rearfoot eversion in order to prevent or manage RRI [18, 19]. External supports such as foot orthoses, motion control shoes, and therapeutic adhesive taping are the most common interventions studied to reduce excessive rearfoot eversion [20–22]. These are widely pre- scribed to realign or correct atypical rearfoot eversion. It is reported that foot orthoses may cause dependency and long-term negative psychological effects [23]. Most importantly, their effectiveness remains controversial [18]. So far, only a few studies have investigated the effect of a training program on excessive rearfoot eversion [24–28], showing sensory-motor training to be superior to either foot orthoses [26] or taping for correcting excessive rearfoot eversion [28]. More research into sports-related functional training interventions that modify rearfoot eversion is thus warranted. Gait retraining is an increasingly used biomechanical modification intervention to practice a movement task [25, 29]. Gait retraining applying real-time feedback from an instrumented PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 2 / 17 treadmill and/or motion capture through a projection and sound system is a novel way to induce motor strategies to alter movement patterns. Data obtained by tracking the biomechan- ics of the body are used to produce real-time feedback. Likewise, feedback modalities such as auditory and/or visual cues are provided for a training task to practice it at a given point/posi- tion. As these feedback modalities are given for any steps and individuals can perceive their own performance in the real time, this may help perform the training task more effectively than verbal feedback given by a coach or clinician. Step rate, step width, step length, foot strike, and hip adduction are the most common gait parameters on which virtual reality feedback is currently given to modify other running-related biomechanical risk factors [30–33]. These studies report successful findings for modifying biomechanical risk factors using the afore- mentioned parameters. Moreover, real-time visual feedback of kinematics or kinetics was reported as the most successful strategy to reduce high-risk factors for RRIs [31]. Therefore, gait retraining may be a useful approach to alter rearfoot biomechanics during running. Previous studies show that rearfoot motion and MLAA might be influenced by changing foot progression angle (FPA) during running and walking [34–36]. In addition, FPA (foot abduction) is postulated as having a positive association with subtalar eversion because it is one of the subtalar pronation movement components [6, 37]. The assumption is that while running toe-out or toe-in, rearfoot kinematics are changed to more or less eversion, respec- tively. As inconsistent results have been reported about the effect/association of FPA on/with rearfoot eversion [34, 38–40], well-designed studies to investigate the effects of changing FPA on rearfoot kinematics are needed. Several studies have intervened FPA during walking using gait retraining to improve pain or reduce knee adduction moment, a contributing factor to knee osteoarthritis [41–43]. Nevertheless, it is unknown whether toe-in or toe-out positioning during running affects frontal plane rearfoot motion. Accordingly, it is postulated that running retraining with changing FPA may be useful to correct atypical rearfoot kinematics. The main aim of this study is to investigate changes in rearfoot in/eversion while perform- ing toe-in/toe-out running using real-time visual feedback. The secondary aim is to investigate changes in subtalar pronation/supination, MLAA, hip flexion, ab/adduction and internal/ external rotation, knee flexion, ankle dorsiflexion, and ankle power while performing toe-out/ in running. We hypothesized that toe-in running reduces peak rearfoot eversion, subtalar pro- nation, and MLAA, and toe-out running increases these factors relative to a natural FPA. From a clinical perspective, the findings might be helpful toward controlling atypical rearfoot kinematics when managing RRI. Methods Study design This is a cross-sectional pilot study conducted to determine the feasibility and effects of chang- ing FPA using real-time visual feedback on rearfoot in/eversion, subtalar supination/prona- tion, and MLAA. Setting Data were collected at the Motion Lab of the Center for Rehabilitation, University Medical Center Groningen between January 2019 and April 2019. Participants Seventeen female runners recruited by our advertisements and social media from the Univer- sity of Groningen and local running clubs volunteered to participate. Inclusion criteria were: PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 3 / 17 female, aged 18–40, minimum 1 year running experience, training distance >10 km/week, habitual rearfoot striker, free of self-reported lower-limb injuries or pain during the last 6 months, no musculoskeletal disorders and/or pain, no foot medial arch disorder determined using the navicular drop test, and no atypical static rearfoot in/eversion [44] prior to data col- lection. Two volunteers were excluded who had flat foot and/or excessive static rearfoot ever- sion. Fifteen volunteers who met the inclusion criteria participated in this study. Ethical approval was obtained through the local Medical Ethics Committee (METc 2018/086) of Uni- versity Medical Center Groningen. Subjects signed an informed consent form and completed a self-developed questionnaire for demographic information prior to data collection. Instrumentation Running assessments were performed on an instrumented split-belt treadmill with two inte- grated force plates of the Gait Real-time Analysis Interactive Lab (GRAIL) system (Motekforce Link, The Netherlands) [45]. Ground reaction force (GRF) signals were recorded at 1000Hz, combined with a 10-camera integrated motion capture system (Vicon Bonita 10; Vicon Motion Systems, Oxford, UK) and further processed to kinematic and kinetic variables in D-Flow (Version 3.28; Motekforce Link, The Netherlands) at a 100 Hz sampling frequency. Real-time filtering of the marker data was performed with a low-pass second-order zero phase Butterworth filter with a 6 Hz cut-off frequency. Marker placement Thirty-four markers were attached to the subject’s body by the same investigator (SHM). Of these, 26 markers were attached according to the human body model 2 (HBM2) (Fig 1) [29] and 8 markers were attached to both feet at the first metatarsal head, navicular bone tuberosity, medial side of calcaneus and posterior part of calcaneus for measuring rearfoot in/eversion and MLAA. Two markers attached along the vertical bisection of the heel counter and the Fig 1. Marker placement. Twenty-six markers were attached to the body according to the HBM model; 8 markers were attached to the feet, to be used for calculating rearfoot eversion and MLAA. https://doi.org/10.1371/journal.pone.0246425.g001 PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 4 / 17 marker on the medial aspect of the heel counter were considered for the rearfoot segment. Additionally, four holes were cut in the shoes: at the first metatarsal head, navicular bone tuberosity, medial side and posterior part of calcaneus (Fig 1). The holes were cut to uncover these aspects of the foot in order to attach markers directly to the skin, allowing measurement of foot movement and not shoe movement. All subjects wore the same brand of neutral shoes (Dr Comfort, refresh, USA) with the same neutral insole. Baseline measurement To set a certain running speed for giving real-time feedback and generalize it to all runners, treadmill speed was set at 8 km/h for all conditions. After a 5-minute warm-up period, a 20-second baseline dataset was collected containing at least 20 strides. Baseline FPA was calcu- lated using the average FPA in midstance (the value in 50% of stance phase) for the first 20 strides. FPA was considered the angle between the line connecting the markers on the calca- neus and second metatarsal head with the longitudinal axis of the treadmill. Feedback A custom-made application was developed on the D-flow software to produce real-time feedback for toeing-out and toeing-in FPA. A clock with a red pointer set to FPA (degrees) was designed to reflect FPA during midstance in real time (Fig 2). A 5˚ target range was shown on the clock (green, Fig 2). To perform toeing-out FPA, the target range was set at 5˚ more than the baseline FPA average with an area of ±2.5˚ deviation from this point. Like- wise, to perform toeing-in FPA, the target range was set at 5˚ less than the baseline FPA average with an area of ±2.5˚ deviation from this point. The 5˚ deviation from the baseline FPA was selected based on pilot testing. In the pilot testing, five runners were asked to run with various FPA (±5˚, ±10˚, and ±15˚) relative to their baseline FPA. The 5˚ deviation from the baseline FPA (±5˚ FPA) was the only FPA that all runners ran with no difficulty. Prior to the feedback session, subjects were asked to practice the conditions with their dom- inant foot in the standing position to become familiar with the feedback. When the red pointer was located within the given target range, the area became green (positive feed- back)–otherwise it became red (negative feedback). The red pointer was fixed on FPA in midstance and updated on each step. Before doing each task, subjects had a 2-min running with FPA feedback. An extra minute was allowed if needed. Subjects were then asked to run, and after 1-min running a 20-second dataset was collected. The order of the experi- mental tasks was randomized. Outcome measurement The rearfoot segment coordinate system was established according to International Society of Biomechanics (ISB) recommendations [46] and calculated as rotation of local calcaneus coordination system relative to the fixed laboratory coordinate system using the rotation sequence defined by ISB. Subtalar pronation/supination angle was calculated using the Isman and Inman method [5], with the detailed explanation described in the study of van den Bogert et al. [29]. Because the pronation/supination axis is not aligned with the foot coordinate axes, it is defined in the foot reference frame using the average subtalar joint. MLAA was calculated based on the angle formed between three markers: first metatarsal head, navicular bone tuberosity and medial aspect of calcaneus. Hip, knee and ankle kine- matics and ankle power are standard measures of HBM2 computed as explained by van den Bogert et al. [29]. GRF data were used to identify the stance phase, with a threshold of 10N vertical GRF for touchdown and toe-off. Kinematic and GRF data were filtered using PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 5 / 17 low-pass zero phase second-order Butterworth filters with the same 6Hz cutoff frequency. Outcomes of ten consecutive steps following the given FPA were calculated and averaged within subjects before being averaged within conditions. Kinematic data during the stance phase were time-normalized to 100% of stance phase. Peak angles were explained as the maximum angle during the stance phase. Timing of peak angles was expressed as percent- age of the stance phase. Angle excursions were expressed as range of motion from touch- down to peak angle. A custom MATLAB script (Version R2018a, Natick, MA, USA) was used to analyze data. Fig 2. Picture representing real-time visual feedback for changing FPA. The training process: Real-time visual feedback is provided to the subject via the big screen. The red pointer represents the FPA of the right foot that is fixed in the midstance (50% stance phase) and updated on each step. The target area is a wedge with a 5˚ range, with its middle point specifying the subject’s normal FPA +5˚ for toe-out and -5˚ for toe-in. The aim is to turn the target area green (positive feedback) by keeping the red pointer (FPA) inside the target area. If the red pointer leaves the target area, the target area turns red (negative feedback). https://doi.org/10.1371/journal.pone.0246425.g002 PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 6 / 17 Statistical analysis Data were analyzed using IBM SPSS version 23 (IBM Corp., Armonk, NY, USA). The normal- ity of data was checked by Shapiro-Wilk tests and QQ plots. A repeated-measures one-way ANOVA was performed for each outcome to identify statistically significant differences between conditions: baseline, toeing-out, and toeing-in trials. Significant main effects were fol- lowed up using pairwise comparisons with Bonferroni adjustment. Significance level (α) was set at 0.05. Results Table 1 shows participants’ characteristics. All assumptions for repeated-measures ANOVA were met (no significant outliers, normal distribution, and sphericity). Table 2 shows results of one-way repeated-measures ANOVA analysis for all measured variables; Fig 3 shows the ensemble average curves of measured variables. The average FPA in midstance for normal, toe-out and toe-in running were -6.2˚, -10.4˚, and -1.8˚, respectively; the average individual standard deviations were 1.1˚, 1˚ and 0.9˚, respectively. We found significant main effects of FPA conditions on peak rearfoot eversion (p<0.001), rearfoot eversion at touchdown (p = 0.001), and rearfoot eversion excursion (p<0.001). There was no significant main effect of FPA conditions on time to peak rearfoot eversion (p = 0.462). Post-hoc tests showed a significant difference in peak rearfoot eversion between normal and toe-out (mean difference (MD) = 1.4; p<0.001), between normal and toe-in (MD = -2.1; p<0.001), and between toe-out and toe-in (MD = -3.5; p<0.001). Post-hoc tests for rearfoot eversion excursion showed a significant difference between normal and toe-in (MD = -1.4; p = 0.001), and between toe-out and toe-in (MD = 1.9; p = 0.001). There was no significant dif- ference in rearfoot eversion excursion between normal and toe-out (MD = -0.5; p = 0.1). Post- hoc tests for rearfoot eversion at touchdown showed significant differences between normal and toe-out (MD = 1.1; p = 0.037), and between toe-out and toe-in (MD = -2.0; p = 0.011). We found significant main effects of FPA conditions on peak pronation (p<0.001), supina- tion/pronation at touchdown (p<0.001), time to peak pronation (p = 0.019), and pronation excursion (p = 0.042). Pairwise comparisons showed a significant difference in peak pronation between normal and toe-out (MD = -1.4; p = 0.002), between normal and toe-in (MD = 2; p<0.001), and between toe-out and toe-in (MD = 3.4; p<0.001). Post-hoc tests for pronation at touchdown showed a significant difference between normal and toe-out (MD = -1.4; p = 0.015), between normal and toe-in (MD = 1.2; p = 0.004), and between toe-out and toe-in (MD = 2.6; p<0.001). Post-hoc tests showed a significant difference in time to peak pronation between toe-out and toe-in (MD = 2.5; p = 0.035). There were no significant differences in time to peak pronation between normal and toe-in (MD = 0.5; p = 0.999) and between normal and toe-out (MD = -2.1; p = 0.219). We found significant main effects of FPA conditions on peak MLAA (p = 0.001), time to peak MLAA (p = 0.04), and MLAA excursion (p<0.001). There was no significant main effect Table 1. Participants characteristics. Variable Mean (SD) Range Age, y 27.5 (6.3) 21–40 Height, cm 170 (5) 164–182 Weight, kg 61.4 (6.1) 50–72 Running experience, y 6.3 (4.4) 2–17 Weekly distance, km 32.7 (17.4) 10–65 https://doi.org/10.1371/journal.pone.0246425.t001 PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 7 / 17 of FPA conditions on MLAA at touchdown (p = 0.816). Bonferroni post-hoc tests showed a significant difference in peak MLAA between normal and toe-in (MD = 0.7; p = 0.022), and between toe-out and toe-in (MD = 0.9; p = 0.005). There was no significant difference in peak MLAA between normal and toe-out (MD = -0.2; p = 0.876). Post-hoc tests also showed a sig- nificant difference in time to peak MLAA between normal and toe-out (MD = -2.6; p = 0.033) Table 2. Statistical results of the all variablesa for each FPA condition. Variable FPA condition One-way repeated measures analysis Normal FPA Toe-out 5˚ Toe-in 5˚ F-value P-value Eta squared Foot progression angle (°) -6.2 (3.1) -10.4 (3.2) † -1.9 (3.2) ‡ 444.66 < 0.001 0.97 Average individual SD for FPA(°) 1.1 (0.3) 1.0 (0.2) 0.9 (0.2) 1.77 0.189 0.11 Peak rearfoot eversion (˚) -8.5 (2.2) -9.9 (2.6) † -6.4 (2.2) ‡ 103.16 < 0.001 0.88 Time to peak rearfoot eversion (% stance) 46.3 (2.7) 47.3 (3.0) 47.3 (3.7) 0.76 0.478 0.05 Rearfoot eversion at touchdown (˚) 3.2 (2.1) 2.1 (2.1) † 4.2 (2.3) 9.51 0.001 0.41 Rearfoot eversion excursion (˚) -11.5 (3.3) -12.0 (3.6) † -10.1 (3.2) ‡ 20.46 < 0.001 0.59 Peak pronation (˚) 4.4 (4.5) 5.8 (4.5) † 2.4 (4.6) ‡ 66.36 < 0.001 0.83 Time to peak pronation (% stance) 70.1 (15.2) 72.1 (17.3) † 69.6 (16) 4.56 0.019 0.25 Supination/pronation at touchdown (˚) -2.3 (4.9) -1 (4.8) † -3.5 (4.9) ‡ 24.69 < 0.001 0.64 Pronation excursion (˚) 6.7 (4.2) 6.8 (3.4) 5.9 (4) ‡ 3.57 0.042 0.20 Peak MLAA (˚) 6.2 (2.2) 6.4 (2.2) † 5.5 (2.3) ‡ 9.7 0.001 0.41 Time to peak MLAA (% stance) 54 (8.1) 56.6 (7.1)  54.1 (9) 3.61 0.040 0.21 MLAA at touchdown (˚) -1.0 (2.3) -1 (2.5) -0.9 (2.7) 0.21 0.816 0.01 MLAA excursion (˚) 7.2 (1.7) 7.4 (1.7) † 6.5 (1.7) ‡ 13.02 < 0.001 0.48 Peak hip internal rotation 7.6 (4.8) 5.1 (5.2) † 9.8 (5.1) ‡ 71.28 < 0.001 0.84 Time to peak hip internal rotation (% stance) 58.3 (38.8) 59.9 (35.7) 51.9 (39.6) 1.70 0.214 0.11 Hip internal rotation at touchdown (˚) 4.5 (5.7) 1.6 (5.9) † 7.0 (5.8) ‡ 82.25 < 0.001 0.86 Hip internal rotation excursion (˚) 3.0 (2.7) 3.5 (2.5) 2.8 (3.0) 1.21 0.305 0.08 Peak hip ab/adduction 14.0 (3.6) 11.6 (3.4) † 13.0 (4.6) 12.38 < 0.001 0.47 Time to peak hip ab/adduction (% stance) 43.2 (3.6) 43.4 (4.0) 45.7 (5.5) 4.02 0.055 0.22 Hip ab/adduction at touchdown (˚) 7.1 (2.6) 6.3 (2.4) 6.0 (3.3) 3.06 0.063 0.18 Hip ab/adduction excursion (˚) 6.9 (2.5) 5.3 (2.3) † 7.0 (2.9) 28.45 < 0.001 0.67 Peak hip flexion 36.8 (5.5) 37.1 (5.6) 36.5 (5.2) 0.80 0.455 0.05 Time to peak hip flexion (% stance) 15.1 (16.0) 15.4 (16.3) 14.0 (16.9) 0.51 0.494 0.04 Hip flexion at touchdown (˚) 35.3 (4.9) 35.8 (5.1) 35.4 (4.8) 0.62 0.546 0.04 Hip flexion excursion (˚) 1.5 (2.2) 1.3 (1.9) 1.1 (1.6) 1.49 0.245 0.10 Peak knee flexion 41.5 (4.5) 40.0 (4.6)  40.2 (4.9) ‡ 8.81 0.001 0.39 Time to peak knee flexion (% stance) 48.1 (1.9) 48.1 (2.5) 48.6 (1.4) 0.83 0.445 0.06 Knee flexion at touchdown (˚) 9.3 (4.2) 9.9 (4.2) 10.4 (3.8) 1.83 0.179 0.11 Knee flexion excursion (˚) 32.1 (5.5) 30.2 (4.3)  30.0 (4.7) ‡ 11.46 < 0.001 0.45 Peak ankle dorsiflexion 24.0 (3.7) 22.4 (3.6) † 23.4 (3.7) 10.30 < 0.001 0.42 Time to peak ankle dorsiflexion (% stance) 56.9 (2.1) 56.8 (2.5) 57.1 (2.3) 0.22 0.807 0.02 Ankle dorsiflexion at touchdown (˚) 5.3 (3.9) 5.4 (2.9) 5.3 (3.7) 0.08 0.927 0.01 Ankle dorsiflexion excursion (˚) 18.7 (2.5) 17.0 (2.7)  18.0 (3.1) 5.50 0.010 0.28 Peak ankle power 13.7 (2.5) 12.3 (2.0) † 13.3 (2.5) 11.7 <0.001 0.46 a Values expressed as mean (SD), FPA foot progression angle, TD touchdown, MLAA medial longitudinal arch angle,  significant difference between normal and toe-out FPA p<0.05, † significant difference between toe-out and toe-in p<0.050, ‡ significant difference between normal and toe-in FPA p<0.05. https://doi.org/10.1371/journal.pone.0246425.t002 PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 8 / 17 and a significant difference in MLAA excursion between normal and toe-in (MD = 0.8; p = 0.005), and between toe-out and toe-in (MD = 1; p<0.001). Fig 3. Ensemble average curves of measured variables for three FPA conditions, with 0% representing heel strike and 100% toe-off. Solid curves = normal FPA, dotted curves = toe-out condition, dashed curves = toe-in condition. Shaded area represents ±1 SD of the normal FPA condition. https://doi.org/10.1371/journal.pone.0246425.g003 PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 9 / 17 We found significant main effects of FPA conditions on peak hip internal/external rota- tion (p<0.001), and hip internal/external rotation at touchdown (p<0.001). There were no significant main effects of FPA conditions on time to peak hip internal/external rotation (p = 0.214), and hip internal/external rotation excursion (p = 0.305). Post-hoc tests showed a significant difference in peak hip internal/external rotation between normal and toe-out (MD = 2.5; p<0.001), between toe-out and toe-in (MD = -4.8; p<0.001), and between nor- mal and toe-in (MD = -2.3; p<0.001). Post-hoc tests also showed a significant difference in hip internal/external rotation at touchdown between normal and toe-out (MD = 3.0; p<0.001), between toe-out and toe-in (MD = -5.4; p<0.001), and between normal and toe- in (MD = -2.5; p<0.001). We found significant main effects of FPA conditions on peak hip ab/adduction (p<0.001), and hip ab/adduction excursion (p<0.001). There were no significant main effects of FPA con- ditions on hip ab/adduction at touchdown (p = 0.063), and time to peak hip ab/adduction (p = 0.055). Post-hoc tests showed a significant difference in peak hip ab/adduction between normal and toe-out (MD = 2.5; p<0.001), and between toe-out and toe-in (MD = -1.4; p = 0.044). There was no significant difference in peak hip ab/adduction between normal and toe-in (MD = 1.0; p = 0.292). Post-hoc tests also showed a significant difference in hip ab/ adduction excursion between normal and toe-out (MD = 1.6; p<0.001), and between toe-out and toe-in (MD = -1.7; p = 0.044). There was no significant difference in hip ab/adduction excursion between normal and toe-in (MD = -0.1; p = 0.999). We found no significant main effects of FPA conditions on peak hip flexion (p = 0.455), time to peak hip flexion (p = 0.494), hip flexion at touchdown (p = 0.546), and hip flexion excursion (p = 0.245). We found significant main effects of FPA conditions on peak knee flexion (p = 0.001), and knee flexion excursion (p<0.001). There were no significant main effects of FPA conditions on time to peak knee flexion (p = 0.445), and knee flexion at touchdown (p = 0.179). Post-hoc tests showed a significant difference in peak knee flexion between normal and toe-out (MD = 1.5; p = 0.003), and between normal and toe-in (MD = 1.3; p = 0.014). There was no significant difference in peak knee flexion between toe-out and toe-in (MD = -0.2; p = 0.999). Post-hoc tests showed a significant difference in knee flexion excursion between normal and toe-out (MD = 2.0; p = 0.003), and between normal and toe-in (MD = 2.4; p = 0.005). There was no significant difference in knee flexion excursion between toe-out and toe-in (MD = 0.4; p = 0.999). We found significant main effects of FPA conditions on peak ankle dorsiflexion (p<0.001), and ankle dorsiflexion excursion (p = 0.010). There were no significant main effects of FPA conditions on time to peak ankle dorsiflexion (p = 0.807), and ankle dorsiflexion at touchdown (p = 0.927). Post-hoc tests showed a significant difference in peak ankle dorsiflexion between normal and toe-out (MD = 1.6; p<0.001), and between toe-out and toe-in (MD = -1.0; p = 0.011). There was no significant difference in peak ankle dorsiflexion between normal and toe-in (MD = 0.6; p = 0.653). Post-hoc tests also showed a significant difference in ankle dorsi- flexion excursion between normal and toe-out (MD = 1.8; p = 0.001). There was no significant difference in ankle dorsiflexion excursion between normal and toe-in (MD = 0.6; p = 0.916), and between toe-out and toe-in (MD = -1.1; p = 0.258). We found significant main effects of FPA conditions on peak ankle power (p<0.001). Post-hoc tests showed a significant difference in peak ankle power between normal and toe- out (MD = 1.3; p = 0.001), and between toe-out and toe-in (MD = -0.9; p<0.005). There was no significant difference in peak ankle power between normal and toe-in (MD = 0.4; p = 590). PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 10 / 17 Discussion The main aim of this study was to investigate the immediate effects of toe-in/toe-out running using real-time visual feedback on rearfoot in/eversion, subtalar pronation/supination, and MLAA during running. In support of our hypothesis, peak rearfoot eversion, peak subtalar pronation and peak MLAA were reduced by toe-in running and increased by toe-out running compared to normal running. Additionally, toe-in running reduced rearfoot eversion excur- sion, MLAA excursion, and subtalar pronation at touchdown compared to normal and toe- out running. Nowadays gait retraining is increasingly used in clinical practice to prevent and manage a variety of sports injuries. Since modifying atypical rearfoot in/eversion is of great interest to biomechanical research and clinical practices, our study constitutes a feasible and applicable basis for using real-time visual feedback to perform toe-in/toe-out running in order to change rearfoot in/eversion, subtalar supination/pronation, and MLAA. No study has so far investigated the kinematic effects of changing FPA using real-time visual feedback during running. We considered ±5˚ differences from the subject’s normal FPA as target points. Participants generally responded in accordance with the target points. The target area on the clock was set ±2.5˚ from the target point for both toe-in and toe-out conditions. The results of the averaged individual SDs show that participants successfully changed their FPA based on the target area (2SD = 2 for toe-out and 1.8 for toe-in). None reported any problems with changing the FPA when asked about any difficulties during performing tasks. The average change of FPA relative to normal FPA was 4.2˚ for toe-out running and 4.4˚ for toe-in running. In fact, the FPA display in the real-time feedback during familiarization helped subjects adapt to the experiment. Because the pointer was aligned with the subject’s FPA, it could be easily perceived. We hypothesized that toe-in running reduces peak rearfoot eversion and toe-out running increases peak rearfoot eversion. Excessive rearfoot eversion is a modifiable risk factor for overuse injuries in athletes [47]. We showed that moving from toe-out to toe-in resulted in a reduction in peak eversion and subtalar pronation, thus supporting the potential value of changing FPA as a method for gait retraining in order to modify atypical rearfoot in/eversion. These alterations in peak rearfoot in/eversion by toe-in/toe-out running might be partially due to lateral and medial shifting of foot pressure during foot roll-over by toe-in and toe-out, respectively [48]. Our results showed that toe-in running reduces peak rearfoot eversion by 2.1˚ –a promising result, as a recent systematic review and meta-analysis investigating rearfoot eversion in injured runners and controls showed that a 2˚ increase in peak rearfoot eversion distinguishes injured from healthy runners [2]. Hence toe-in/toe-out running with a 4–5˚ dif- ference from the preferred FPA may have clinical significance in the control of atypical rear- foot in/eversion when preventing and managing RRI. Subtalar pronation at touchdown and time to peak pronation were significantly reduced by toe-in running relative to normal running; however, rearfoot eversion at touchdown and time to peak rearfoot eversion were not significant between different FPA conditions. Rearfoot eversion excursion was also reduced during toe-in running compared to normal and toe-out running. Supination and/or inversion are directly attributed to tarsal joint locking in either the early or the late stance phase [49]. Accordingly, supination helps the foot turn to a rigid lever where needed. It is documented that greater pronation excursion leads to a delayed supination in the late stance [50]. According to our results, in toe-in running the foot is in a more supi- nated position relative to normal and toe-out running. Therefore, in individuals who have greater foot pronation, toe-running may help foot stabilization at touchdown and even in late stance phase during running. We found a smaller MLAA during toe-in running compared with normal and toe-out run- ning. The effect of changing FPA on MLAA during running is not yet well documented. Only PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 11 / 17 few studies with conflicting results investigated the association between MLAA and FPA dur- ing running or walking [36, 51, 52], examining only the correlation between MLAA and FPA but not how changeable MLAA is when changing FPA. We found a decreased MLAA of approximately 0.7˚ and 1˚ during toe-in running compared to normal and toe-out running, respectively. This amount of change might appear small compared to changes possibly required to correct an atypical MLAA. An explanation for a small change in MLAA might be that the current study was conducted for effects on healthy subjects in order to assess the potential of FPA modification on MLAA. Further research on individuals with atypical MLAA is therefore warranted to determine how effective FPA modifications are on MLAA. This is even more important because the correction of atypical MLAA has often been suggested as one of the potential corrective strategies for atypical rearfoot eversion [50, 53]. Our results showed that changes in FPA are accompanied by changes in lower limb joint biomechanics. Specifically, these changes occurred in the peak hip internal/external rotation, hip ab/adduction, knee flexion, ankle dorsiflexion, and ankle power. Compared to running with normal FPA, both toe-out and toe-in running reduced peak hip adduction. As increased peak hip adduction is associated with iliotibial band syndrome and patellofemoral pain syn- drome in runners [12, 54], FPA modification might be used as a potential gait retraining to reduce peak hip adduction. Toe-out running was accompanied by increased hip external rota- tion which is reported as a risk factor for medial tibial stress syndrome [55]. In contrast, toe-in running was accompanied by increased hip internal rotation. Toe-in running, therefore, may be used as potential gait retraining to reduce increased hip external rotation. Toe-out running compared to running with normal FPA reduced ankle power, possibly resulting in reducing the effectiveness of the ankle in providing positive push-off power. Gait retraining studies have shown that the peak knee adduction moment, a contributing factor to knee osteoarthritis, is increased/decreased with changes in FPA [41, 56]. This can subsequently load different aspects of the knee. As a result, clinician and researchers should consider changes in lower limb joint biomechanics when using FPA to modify rearfoot eversion. Rearfoot in/eversion has been commonly used in the literature as an alternative way to express subtalar supination/pronation during walking or running [4, 6]. Our results show that although peak rearfoot eversion can be a proper representative of peak subtalar pronation dur- ing running, there are considerable differences between the other variables such as angle at touchdown, time to peak and excursion. Rearfoot eversion only describes one aspect of subta- lar pronation and the other aspects of subtalar pronation may distinguish it from rearfoot ever- sion during running. Isman and Inman [5] presented a 3D kinematic approach defining true supination/pronation angle occurring in the subtalar joint as used in the current study. It is therefore suggested that future studies apply proper terminology (supination/pronation and/ or rearfoot in/eversion) based on the study objective. One major concern about rearfoot eversion is that no standardized norm or clinical defini- tion exists for classification of rearfoot eversion during running to specify the extent to which rearfoot eversion falls into typical or atypical movements [4]. In the current study, the use of the terms typical or atypical (greater or lower) rearfoot eversion during running is based on the results of studies investigating rearfoot eversion between injured (history of injury) and non-injured runners. Therefore, further studies warrant to determine to what extent rearfoot eversion can be considered as typical or atypical. The current study showed the feasibility and effects of toe-in/toe-out running using real- time visual feedback on rearfoot in/eversion, subtalar pronation/supination, and MLAA. To successfully achieve a target FPA during running an advanced system is needed, as running is a fast-cyclic motion that makes real-time feedback difficult. Current approaches for giving feedback on FPA mainly require camera-based motion capture, limiting FPA measurement PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 12 / 17 or/and training to laboratory settings (thus hindering FPA training outside the laboratory). Recent studies present valid means using insole- or shoe-embedded sensors to estimate FPA and give real-time feedback during over-ground gait [57, 58]. However, these means were vali- dated during walking only, therefore usability investigations during running are required. Also, compared to the motion capture values an absolute error of 1.7˚ should be taken into account when using these means. To see the effectiveness of gait retraining on rearfoot ever- sion over time or to be sure that change in rearfoot eversion is at the desired level, tracking rearfoot eversion might be useful. As application of 3D motion capture systems is not easy in clinical practice, 2D measurement of rearfoot eversion using smartphone application can be used as a surrogate to 3D measurement [59]. Limitations and recommendations There are a few limitations in the current study. All participants were healthy female runners so results cannot be extrapolated to male runners and/or injured runners. Further research is needed to investigate whether our results have the same effect on runners with atypical in/ eversion, supinated or pronated feet, and/or MLAA. All participants ran with rearfoot strike. Since foot strike pattern affects the ankle and foot biomechanics, our results cannot be general- ized to those with midfoot or forefoot strike. Based on our results regarding the differences between rearfoot eversion and subtalar pronation it was suggested that future research con- ducted to investigate the biomechanics of subtalar pronation in individuals should consider the difference between rearfoot in/eversion and subtalar supination/pronation, and choose one or both based on the study objective. Running speed was set at 8km/h, so it is not clear whether the same results would be found at higher or slower speeds. Changes in lower limb biomechanics such as hip ab/adduction, hip internal/external rotation, knee flexion, ankle dor- siflexion, and ankle power should be taken into account when changing FPA is used to modify rearfoot eversion. As our study aimed to investigate the acute effect of changing FPA, it is unknown whether runners would retain it in the long term. Therefore, further studies should be undertaken to investigate the viability of toe-in/toe-out running in the long term. Conclusion This study showed that female healthy runners were able to change their FPA when receiving real-time visual feedback for FPA. Toe-in running using real-time visual feedback reduced peak rearfoot eversion, peak pronation, and peak MLAA compared to normal and toe-out running. Toe-out running, instead, increased these kinematic factors compared to normal and toe-in running. Rearfoot in/eversion is not an appropriate surrogate to predict all supination/ pronation parameters. Our study provides new knowledge and lays the foundation for future research into modifying atypical rearfoot in/eversion, subtalar supination/pronation, and MLAA using gait retraining (toe-in and toe-out running) by real-time visual feedback during running. Clinicians and researchers should take it into account that changes in FPA when run- ning is accompanied by changes in lower limb joint biomechanics. Supporting information S1 Fig. One-way repeated measure ANOVA results for foot progression angle (FPA). (DOCX) S2 Fig. One-way repeated measure ANOVA results for rearfoot eversion variables. (DOCX) PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 13 / 17 S3 Fig. One-way repeated measure ANOVA results for pronation variables. (DOCX) S4 Fig. One-way repeated measure ANOVA results for medial longitudinal arch angle. (DOCX) S5 Fig. One-way repeated measure ANOVA results for hip internal rotation. (DOCX) S6 Fig. One-way repeated measure ANOVA results for hip ab/adduction. (DOCX) S7 Fig. One-way repeated measure ANOVA results for hip flexion. (DOCX) S8 Fig. One-way repeated measure ANOVA results for knee flexion. (DOCX) S9 Fig. One-way repeated measure ANOVA results for ankle dorsiflexion. (DOCX) S10 Fig. One-way repeated measure ANOVA results for peak ankle power. (DOCX) Acknowledgments We gratefully thank all participants for having volunteered to be part of this study. Author Contributions Conceptualization: Seyed Hamed Mousavi, Laurens van Kouwenhove, Reza Rajabi, Johannes Zwerver, Juha M. Hijmans. Data curation: Seyed Hamed Mousavi, Laurens van Kouwenhove, Reza Rajabi. Formal analysis: Seyed Hamed Mousavi. Funding acquisition: Seyed Hamed Mousavi, Reza Rajabi. Methodology: Laurens van Kouwenhove. Project administration: Seyed Hamed Mousavi, Johannes Zwerver, Juha M. Hijmans. Software: Laurens van Kouwenhove. Supervision: Laurens van Kouwenhove, Reza Rajabi, Johannes Zwerver, Juha M. Hijmans. Visualization: Seyed Hamed Mousavi, Johannes Zwerver, Juha M. Hijmans. Writing – original draft: Seyed Hamed Mousavi, Johannes Zwerver, Juha M. Hijmans. Writing – review & editing: Seyed Hamed Mousavi, Laurens van Kouwenhove, Reza Rajabi, Johannes Zwerver, Juha M. Hijmans. References 1. Taunton JE, Ryan MB, Clement DB, McKenzie DC, Lloyd-Smith DR, Zumbo BD. A retrospective case- control analysis of 2002 running injuries. Br J Sports Med. 2002; 36: 95–101. Available: http://www.ncbi. nlm.nih.gov/pubmed/11916889 https://doi.org/10.1136/bjsm.36.2.95 PMID: 11916889 PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 14 / 17 2. Mousavi SH, Hijmans JM, Rajabi R, Diercks R, Zwerver J, van der Worp H. Kinematic risk factors for lower limb tendinopathy in distance runners: A systematic review and meta-analysis. Gait Posture. 2019; 69: 13–24. https://doi.org/10.1016/j.gaitpost.2019.01.011 PMID: 30658311 3. Razeghi M, Batt ME. Foot type classification: a critical review of current methods. Gait Posture. 2002; 15: 282–91. Available: http://www.ncbi.nlm.nih.gov/pubmed/11983503 https://doi.org/10.1016/s0966- 6362(01)00151-5 PMID: 11983503 4. Nigg B, Behling AV, Hamill J. Foot pronation. Footwear Sci. 2019; 11: 131–134. https://doi.org/10.1080/ 19424280.2019.1673489 5. Isman RE, Inman VT. Anthropometric studies of the human foot and ankle. Foot Ankle,. 1969; 11: 97– 129. 6. Nester CJ. Rearfoot complex: a review of its interdependent components, axis orientation and func- tional model. Foot. 1997; 7: 86–96. https://doi.org/10.1016/S0958-2592(97)90054-7 7. Ferber R, Hreljac A, Kendall KD. Suspected mechanisms in the cause of overuse running injuries: a clinical review. Sports Health. 2009; 1: 242–6. https://doi.org/10.1177/1941738109334272 PMID: 23015879 8. Hein T, Janssen P, Wagner-Fritz U, Haupt G, Grau S. Prospective analysis of intrinsic and extrinsic risk factors on the development of Achilles tendon pain in runners. Scand J Med Sci Sport. 2014; 24: e201– e212. https://doi.org/10.1111/sms.12137 9. Vannatta CN, Heinert BL, Kernozek TW. Biomechanical risk factors for running-related injury differ by sample population: A systematic review and meta-analysis. Clinical Biomechanics. Elsevier Ltd; 2020. p. 104991. https://doi.org/10.1016/j.clinbiomech.2020.104991 PMID: 32203864 10. Kuhman DJ, Paquette MR, Peel SA, Melcher DA. Comparison of ankle kinematics and ground reaction forces between prospectively injured and uninjured collegiate cross country runners. Hum Mov Sci. 2016; 47: 9–15. https://doi.org/10.1016/j.humov.2016.01.013 PMID: 26827155 11. Milner CE, Hamill J, Davis IS. Distinct hip and rearfoot kinematics in female runners with a history of tib- ial stress fracture. J Orthop Sports Phys Ther. 2010; 40: 59–66. https://doi.org/10.2519/jospt.2010. 3024 PMID: 20118528 12. Aderem J, Louw QA. Biomechanical risk factors associated with iliotibial band syndrome in runners: a systematic review. BMC Musculoskelet Disord. 2015; 16: 1–16. https://doi.org/10.1186/s12891-015- 0454-0 PMID: 25637090 13. Neal BS, Griffiths IB, Dowling GJ, Murley GS, Munteanu SE, Franettovich Smith MM, et al. Foot posture as a risk factor for lower limb overuse injury: a systematic review and meta-analysis. J Foot Ankle Res. 2014; 7: 55. https://doi.org/10.1186/s13047-014-0055-4 PMID: 25558288 14. Nakamura H, Kakurai S. Relationship between the medial longitudinal arch movement and the pattern of rearfoot motion during the stance phase of walking. J Phys Ther Sci. 2003; 15: 13–18. https://doi.org/ 10.1589/jpts.15.13 15. Dixon SJ. Application of center-of-pressure data to indicate rearfoot inversion-eversion in shod running. J Am Podiatr Med Assoc. 2006; 96: 305–12. https://doi.org/10.7547/0960305 PMID: 16868323 16. Tiberio D. The effect of excessive subtalar joint pronation on patellofemoral mechanics: a theoretical model. J Orthop Sports Phys Ther. 1987; 9: 160–165. doi:1911 [pii] https://doi.org/10.2519/jospt.1987. 9.4.160 PMID: 18797010 17. Van Der Worp MP, Ten Haaf DSM, Van Cingel R, De Wijer A, Nijhuis-Van Der Sanden MWG, Bart Staal J. Injuries in runners; a systematic review on risk factors and sex differences. PLoS One. 2015; 10: 1–18. https://doi.org/10.1371/journal.pone.0114937 PMID: 25706955 18. Desmyttere G, Hajizadeh M, Bleau J, Begon M. Effect of foot orthosis design on lower limb joint kine- matics and kinetics during walking in flexible pes planovalgus: A systematic review and meta-analysis. Clin Biomech. 2018; 59: 117–129. https://doi.org/10.1016/j.clinbiomech.2018.09.018 PMID: 30227277 19. Naderi A, Degens H, Sakinepoor A. Arch-support foot-orthoses normalize dynamic in-shoe foot pres- sure distribution in medial tibial stress syndrome. Eur J Sport Sci. 2019; 19: 247–257. https://doi.org/10. 1080/17461391.2018.1503337 PMID: 30086684 20. Jafarnezhadgero A, Alavi-Mehr SM, Granacher U. Effects of anti-pronation shoes on lower limb kine- matics and kinetics in female runners with pronated feet: The role of physical fatigue. Boullosa D, editor. PLoS One. 2019; 14: e0216818. https://doi.org/10.1371/journal.pone.0216818 PMID: 31086402 21. Lilley K, Stiles V, Dixon S. The influence of motion control shoes on the running gait of mature and young females. Gait Posture. 2013; 37: 331–335. https://doi.org/10.1016/j.gaitpost.2012.07.026 PMID: 23122596 22. Cheung RTH, Chung RCK, Ng GYF. Efficacies of different external controls for excessive foot prona- tion: A meta-analysis. Br J Sports Med. 2011; 45: 743–751. https://doi.org/10.1136/bjsm.2010.079780 PMID: 21504966 PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 15 / 17 23. Banwell HA, Mackintosh S, Thewlis D. Foot orthoses for adults with flexible pes planus: a systematic review. J Foot Ankle Res. 2014; 7: 23. https://doi.org/10.1186/1757-1146-7-23 PMID: 24708560 24. Andreasen J, Mølgaard CM, Christensen M, Kaalund S, Lundbye-Christensen S, Simonsen O, et al. Exercise therapy and custom-made insoles are effective in patients with excessive pronation and chronic foot pain—A randomized controlled trial. Foot. 2013; 23: 22–28. https://doi.org/10.1016/j.foot. 2012.12.001 PMID: 23434214 25. Okamura K, Kanai S, Fukuda K, Tanaka S, Ono T, Oki S. The effect of additional activation of the plan- tar intrinsic foot muscles on foot kinematics in flat-footed subjects. Foot. 2019; 38: 19–23. https://doi. org/10.1016/j.foot.2018.11.002 PMID: 30530189 26. Kim E-K, Kim JS. The effects of short foot exercises and arch support insoles on improvement in the medial longitudinal arch and dynamic balance of flexible flatfoot patients. J Phys Ther Sci. 2016; 28: 3136–3139. https://doi.org/10.1589/jpts.28.3136 PMID: 27942135 27. Kim E, Choi H, Cha J-H, Park J-C, Kim T. Effects of neuromuscular training on the rear-foot angle kine- matics in elite women field hockey players with chronic ankle instability. J Sports Sci Med. 2017; 16: 137–146. Available: http://www.ncbi.nlm.nih.gov/pubmed/28344462 PMID: 28344462 28. Lee J, Yoon J, Cynn H. Foot exercise and taping in patients with patellofemoral pain and pronated foot. J Bodyw Mov Ther. 2017; 21: 216–222. https://doi.org/10.1016/j.jbmt.2016.07.010 PMID: 28167183 29. Van Den Bogert AJ, Geijtenbeek T, Even-Zohar O, Steenbrink F, Hardin EC. A real-time system for bio- mechanical analysis of human movement and muscle function. Med Biol Eng Comput. 2013; 51: 1069– 1077. https://doi.org/10.1007/s11517-013-1076-z PMID: 23884905 30. Agresta C, Brown A. Gait retraining for injured and healthy runners using augmented feedback: A sys- tematic literature review. J Orthop Sport Phys Ther. 2015; 45: 576–584. https://doi.org/10.2519/jospt. 2015.5823 PMID: 26158882 31. Napier C, Cochrane CK, Taunton JE, Hunt MA. Gait modifications to change lower extremity gait biome- chanics in runners: A systematic review. British Journal of Sports Medicine. 2015. pp. 1382–1388. https://doi.org/10.1136/bjsports-2014-094393 PMID: 26105016 32. Chan ZYS, Zhang JH, Au IPH, An WW, Shum GLK, Ng GYF, et al. Gait retraining for the reduction of injury occurrence in novice distance runners: 1-year follow-up of a randomized controlled trial. Am J Sports Med. 2018; 46: 388–395. https://doi.org/10.1177/0363546517736277 PMID: 29065279 33. Mousavi SH, van Kouwenhove L, Rajabi R, Zwerver J, Hijmans JM. The effect of changing mediolateral center of pressure on rearfoot eversion during treadmill running. Gait Posture. 2021; 83: 201–209. https://doi.org/10.1016/j.gaitpost.2020.10.032 PMID: 33171373 34. Kernozek TW, Ricard MD. Foot placement angle and arch type: effect on rearfoot motion. Arch Phys Med Rehabil. 1990; 71: 988–991. PMID: 2241547 35. Britt M. Potential relationships between lower extremity structural alignment and toe angle during static and dynamic activities. 2019. pp. 1–52. 36. Hollander K, Stebbins J, Albertsen IM, Hamacher D, Babin K, Hacke C, et al. Arch index and running biomechanics in children aged 10–14 years. Gait Posture. 2018; 61: 210–214. https://doi.org/10.1016/j. gaitpost.2018.01.013 PMID: 29413786 37. Sangeorzan A, Sangeorzan B. Subtalar Joint Biomechanics: From Normal to Pathologic. Foot and Ankle Clinics. W.B. Saunders; 2018. pp. 341–352. https://doi.org/10.1016/j.fcl.2018.04.002 PMID: 30097078 38. Levinger P, Murley GS, Barton CJ, Cotchett MP, McSweeney SR, Menz HB. A comparison of foot kine- matics in people with normal- and flat-arched feet using the Oxford Foot Model. Gait Posture. 2010; 32: 519–523. https://doi.org/10.1016/j.gaitpost.2010.07.013 PMID: 20696579 39. Barnes A, Wheat J, Milner CE. Fore- and rearfoot kinematics in high- and low-arched individuals during running. Foot Ankle Int. 2011; 32: 710–6. https://doi.org/10.3113/FAI.2011.0710 PMID: 21972767 40. McClay I, Manal K. The influence of foot abduction on differences between two-dimensional and three- dimensional rearfoot motion. Foot Ankle Int. 1998; 19: 26–31. https://doi.org/10.1177/ 107110079801900105 PMID: 9462909 41. Shull PB, Silder A, Shultz R, Dragoo JL, Besier TF, Delp SL, et al. Six-week gait retraining program reduces knee adduction moment, reduces pain, and improves function for individuals with medial com- partment knee osteoarthritis. J Orthop Res. 2013; 31: 1020–25. https://doi.org/10.1002/jor.22340 PMID: 23494804 42. Uhlrich SD, Silder A, Beaupre GS, Shull PB, Delp SL. Subject-specific toe-in or toe-out gait modifica- tions reduce the larger knee adduction moment peak more than a non-personalized approach. J Bio- mech. 2018; 66: 103–110. https://doi.org/10.1016/j.jbiomech.2017.11.003 PMID: 29174534 43. Simic M, Wrigley TV, Hinman RS, Hunt MA, Bennell KL. Altering foot progression angle in people with medial knee osteoarthritis: the effects of varying toe-in and toe-out angles are mediated by pain and PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 16 / 17 malalignment. Osteoarthr Cartil. 2013; 21: 1272–1280. https://doi.org/10.1016/j.joca.2013.06.001 PMID: 23973141 44. Langley B, Cramp M, Morrison SC. Selected static foot assessments do not predict medial longitudinal arch motion during running. J Foot Ankle Res. 2015; 8: 56. https://doi.org/10.1186/s13047-015-0113-6 PMID: 26464583 45. Gagliardi C, Turconi AC, Biffi E, Maghini C, Marelli A, Cesareo A, et al. Immersive virtual reality to improve walking abilities in cerebral palsy: A pilot study. Ann Biomed Eng. 2018; 46: 1376–1384. https://doi.org/10.1007/s10439-018-2039-1 PMID: 29704186 46. Wu G, Siegler S, Allard P, Kirtley C, Leardini A, Rosenbaum D, et al. ISB recommendation on definitions of joint coordinate system of various joints for the reporting of human joint motion—part I: ankle, hip, and spine. J Biomech. 2002; 35: 543–548. https://doi.org/10.1016/s0021-9290(01)00222-6 PMID: 11934426 47. Goo Y-M, Kim D-Y, Kim T-H. The effects of hip external rotator exercises and toe-spread exercises on lower extremity muscle activities during stair-walking in subjects with pronated foot. J Phys Ther Sci. 2016; 28: 816–819. https://doi.org/10.1589/jpts.28.816 PMID: 27134364 48. Sakaguchi M, Shimizu N, Yanai T, Stefanyshyn DJ, Kawakami Y. Hip rotation angle is associated with frontal plane knee joint mechanics during running. Gait Posture. 2015; 41: 557–561. https://doi.org/10. 1016/j.gaitpost.2014.12.014 PMID: 25572723 49. Blackwood CB, Yuen TJ, Sangeorzan BJ, Ledoux WR. The midtarsal joint locking mechanism. Foot Ankle Int. 2005; 26: 1074–1080. https://doi.org/10.1177/107110070502601213 PMID: 16390642 50. Hollander K, Zech A, Rahlf AL, Orendurff MS, Stebbins J, Heidt C. The relationship between static and dynamic foot posture and running biomechanics: A systematic review and meta-analysis. Gait Posture. 2019; 72: 109–122. https://doi.org/10.1016/j.gaitpost.2019.05.031 PMID: 31195310 51. Twomey DM, McIntosh AS. The effects of low arched feet on lower limb gait kinematics in children. Foot. 2012; 22: 60–65. https://doi.org/10.1016/j.foot.2011.11.005 PMID: 22155064 52. Twomey D, McIntosh AS, Simon J, Lowe K, Wolf SI. Kinematic differences between normal and low arched feet in children using the Heidelberg foot measurement method. Gait Posture. 2010; 32: 1–5. https://doi.org/10.1016/j.gaitpost.2010.01.021 PMID: 20172730 53. Zhang X, Aeles J, Vanwanseele B. Comparison of foot muscle morphology and foot kinematics between recreational runners with normal feet and with asymptomatic over-pronated feet. Gait Posture. 2017; 54: 290–294. https://doi.org/10.1016/j.gaitpost.2017.03.030 PMID: 28390293 54. Noehren B, Hamill J, Davis I. Prospective evidence for a hip etiology in patellofemoral pain. Med Sci Sports Exerc. 2013; 45: 1120–1124. https://doi.org/10.1249/MSS.0b013e31828249d2 PMID: 23274607 55. Reinking MF, Austin TM, Richter RR, Krieger MM. Medial tibial stress syndrome in active individuals: A systematic review and meta-analysis of risk factors. Sports Health. 2017; 9: 252–261. https://doi.org/ 10.1177/1941738116673299 PMID: 27729482 56. Shull PB, Shultz R, Silder A, Dragoo JL, Besier TF, Cutkosky MR, et al. Toe-in gait reduces the first peak knee adduction moment in patients with medial compartment knee osteoarthritis. J Biomech. 2013; 46: 122–128. https://doi.org/10.1016/j.jbiomech.2012.10.019 PMID: 23146322 57. Charlton JM, Xia H, Shull PB, Hunt MA. Validity and reliability of a shoe-embedded sensor module for measuring foot progression angle during over-ground walking. J Biomech. 2019; 89: 123–127. https:// doi.org/10.1016/j.jbiomech.2019.04.012 PMID: 31047695 58. Xia H, Xu J, Wang J, Hunt MA, Shull PB. Validation of a smart shoe for estimating foot progression angle during walking gait. J Biomech. 2017; 61: 193–198. https://doi.org/10.1016/j.jbiomech.2017.07. 012 PMID: 28780187 59. Mousavi SH, Hijmans JM, Moeini F, Rajabi R, Ferber R, van der Worp H, et al. Validity and reliability of a smartphone motion analysis app for lower limb kinematics during treadmill running. Phys Ther Sport. 2020; 43: 27–35. https://doi.org/10.1016/j.ptsp.2020.02.003 PMID: 32062587 PLOS ONE The effect of changing foot progression angle on rearfoot eversion during running PLOS ONE | https://doi.org/10.1371/journal.pone.0246425 February 10, 2021 17 / 17

The effect of changing foot progression angle using real-time visual feedback on rearfoot eversion during running.

02-10-2021

Mousavi, Seyed Hamed,van Kouwenhove, Laurens,Rajabi, Reza,Zwerver, Johannes,Hijmans, Juha M

eng

PMC4636312

RESEARCH ARTICLE Effect of a Wide Stance on Block Start Performance in Sprint Running Mitsuo Otsuka*, Toshiyuki Kurihara, Tadao Isaka Faculty of Sport and Health Science, Ritsumeikan University, 1-1-1 Nojihigashi, Kusatsu, Shiga, 525–8577, Japan * otsuka-a@st.ritsumei.ac.jp Abstract This study aimed to clarify the effect of widened stance width at the set position during the block start phase in sprint running on kinematics and kinetics at the hip joint and block- induced power. Fourteen male sprinters volunteered to participate in this study. They per- formed three block-start trials with a normal stance width (25 ± 1 cm, normal condition) and a widened stance width (45 ± 2 cm, widened condition) at the set position. The block start movements were recorded at 250 Hz with high-speed cameras and the ground reaction forces at 1250 Hz with force plates. During the block phase in the widened condition, the hip abduction and external rotation angles in both legs were significantly larger and smaller, respectively, than those in the normal condition. The positive peak value of the hip power in the rear leg was significantly greater in the widened condition than that in the normal condi- tion. However, no significant difference was seen in the normalized block-induced power between the widened and normal conditions. We conclude that a widened stance width at the set position affects the hip-joint kinematics and rear hip power generation during the block start phase, but no effect on the block-induced power when considering sprinting per- formance during the whole block start phase. Introduction During the pushing phase on starting blocks, the average value of external power in an anterior direction to translate the whole-body centre of mass (COM; hereafter, block-induced power) is important for a great performance in the 100-m dash [1,2]. This is associated with the exten- sion of front and rear legs. Sprinters are permitted to adjust the anteroposterior position and inclination of both starting blocks in accordance with the regulations for athletes [3]. Several previous studies have clarified the effect of different body postures at the set position on the subsequent sprinting motion during the block start phase [4–8]. For instance, relative to the elongated start, bunched and medium starts shorten the pushing duration during the pushing phase on the starting block, thereby shortening the subsequent sprinting time at 5 m and 10 m [7,8]. This may be due to the optimal position of body segments, which enhances the power generation of the lower limbs and block-induced power during the block start phase [4,5]. PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 1 / 13 OPEN ACCESS Citation: Otsuka M, Kurihara T, Isaka T (2015) Effect of a Wide Stance on Block Start Performance in Sprint Running. PLoS ONE 10(11): e0142230. doi:10.1371/journal.pone.0142230 Editor: Miklos S. Kellermayer, Semmelweis University, HUNGARY Received: July 6, 2015 Accepted: October 18, 2015 Published: November 6, 2015 Copyright: © 2015 Otsuka et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: This work was supported by the Kozuki Foundation 2011-9-1 to MO (http://www.kozuki.or.jp/ jigyou/spresearch/list2011_09_spres.html). The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. Sprinters extend both legs during the block start phase from a bending position [2,9]. Peak extension angular velocity of the front and rear hips, contributes to the block-induced power during the block start phase, unlike that of the knee and ankle [2]. This rapid hip extension might be associated with greater power generation at the hip joint, thereby providing greater block-induced power during the block start phase. The hip power generation is larger than those at the knee and ankle of the front and rear legs during the block start phase [5]. These findings suggest that the hip extensor kinetics is a key contributor to high block-induced power. This leg extension motion during the block start phase is similar to the double-legged squat motion [10]. Biomechanical analysis of the squatting motion has revealed that lengthen- ing the mediolateral distance between feet (140% of shoulder width) contributes to stronger isometric contractions in lower limb muscles [11]. This widened stance width enhances the mean electromyography value of the gluteus maximus during a squat [12–14]. The stance width at the set position of the block start has been reported to be 23 ± 1 cm [15]; this is shorter than the stance width in previous studies on widened stance width in squatting [11–14]. These studies may indicate that a widened stance width during the block start phase would enhance block-induced power during the block start phase attained by a greater hip joint power. Never- theless, sprinters may not be able to select enough stance width at the set position for greater block-induced power using the current competition blocks [3]. For instance, the mediolateral width of starting blocks in overall dimension is 30 cm (RM-150, Seiko, Tokyo, Japan). This width is narrower than the stance width in squats used to enhance muscle strength [11]. Thus, further investigation of squatting to block start is required to elucidate the effect of a wide stance width on block-induced power. This information would help in reconsidering block start rules [3], developing new starting block designs, and aiding sprinters in the appro- priate placement of starting blocks in the future. This study aimed to clarify the effect of wid- ened stance width at the set position during the block start phase on hip kinematics, hip kinetics and block-induced power. Our hypotheses were as follows: 1) widened step width dur- ing the block start phase would lead to sprinters changing their hip position and enhance block-induced power attained by the high hip power generation, and 2) widened step width during the block start phase would affect the relationship between changes in block-induced power and those in the hip power generation. Materials and Methods Participants Fourteen male sprinters (mean ± standard deviation [SD]; age: 21.1 ± 1.2 years, body mass: 64.5 ± 3.9 kg, height: 1.76 ± 0.04 m) volunteered to participate in this study. Three participants were international-level sprinters, who were finalists in the National Championship. All partic- ipants were sprint specialists with training experience of
6 years (8.4 ± 2.4 years), and the average personal best 100-m time was 10.99 ± 0.40 s (range: 10.21–11.65 s). The experimental protocol was approved by the Research Ethics Committee Involving Liv- ing Human Subjects at Ritsumeikan University (BKC-human-2011-011). Each participant pro- vided written informed consent before study participation. The individual in this manuscript has given written informed consent (as outlined in PLOS consent form) to publish these case details. Experimental procedure The participants were asked to perform a 10-m sprint on an indoor track, exerting maximum effort from a crouching position, with a widened (45 ± 2 cm, widened condition, Fig 1A, S1 Video) and normal (25 ± 1 cm, normal condition, Fig 1B, S2 Video) stance width (mediolateral A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 2 / 13 distance between midpoints of first and fifth metatarsals of feet). The participants were instructed to start after a gun signal by a starter [3]. The stance width in the widened condition corresponded to 140% of shoulder width during a squat at widened stance width [11,13]. Each participant performed three trials each with normal and widened stance width using their own spike shoes, and the order of trials in both conditions was randomized. To reduce the effect of body size and crouching position type on block performance, the anteroposterior distance of the starting blocks was adjusted to 12% of each participant’s height (21 ± 1 cm) so as to be cor- responded with the bunched start [7,8,16]. The anteroposterior distance between the feet in a bunched start is closer to that of a squat motion relative to those of medium and elongated starts. Throughout the experiment, the block angles of front and rear legs were set at 40° and 42°, respectively [4]. The starting blocks were securely anchored to the synthetic track surface on two separate force plates (0.40 m × 0.60 m; TF-4060-B; Tech-Gihan, Inc., Kyoto, Japan). Before the experimental trials, an appropriate 15-min warm-up including jogging and stretch- ing and at least three sprints in each stance width condition were performed. Data collection Data on the participants’ sprinting movement and ground reaction force (GRF) during the block start phase were captured simultaneously. A total of 36 retro-reflective markers sized 12 mm were attached on the pelvis and lower limbs, and the three-dimensional locations of the markers were recorded using a 16-camera motion capture system (Raptor-E digital; Motion Analysis Corporation, Santa Rosa, CA, USA) sampling at 250 Hz. The GRF data from legs were separately measured by positioning a total of 2 force plates (TF-4060-B; Tech-Gihan, Inc., Kyoto, Japan), arranged in two rows of two each, sampling at 1250 Hz. The sprinting time was measured based on signals from the gun (EP; Molten Inc, Hiroshima, Japan) and photocell Fig 1. A postero-superior view of the set position during the block start phase. (A) Set position in the widened condition. (B) Set position in normal condition. doi:10.1371/journal.pone.0142230.g001 A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 3 / 13 (E3G-R13; Omuron Inc., Kyoto, Japan) set at 2.0-m mark, and were synchronized with the GRF data. Data processing GRF data were not filtered [17]. We used a vertical GRF threshold of 10 N to determine the instant of take-off from the starting blocks. We then used these readings to divide the block start phase into the double-stance and single-stance phases (Fig 2). The marker trajectory data were filtered using a fourth-order, zero-lag, low-pass Butter- worth filter, and the cut-off frequency was set at 10 Hz [15]. A 7-segment rigid body model including the pelvis, thighs, shanks, and feet was created. The segmental data were calculated using mass properties based on cadavers [18]. The locations of the center of mass and inertial properties were obtained using a mathematical model [19]. For comparison between different sprinters, the dimensionless [20] normalized block- induced power was considered as the best indicator of sprinting performance during the block start phase (hereafter, normalized block-induced power) [1,2]. So as to clarify the relationship between changes in block-induced power and those in the hip power generation, the dimen- sionless normalized block-induced power (PN) was calculated as follows: PN ¼ P=ðm g3=2l1=2Þ ð1Þ where m is the mass of the sprinter, g is the acceleration due to gravity, and l is the leg length of the sprinter. Block-induced power (P) was calculated as follows [1,2]: P ¼ mðv2 f  v2 i Þ=ð2 DtÞ ð2Þ Fig 2. Definition of the double-leg and single-leg stance phases during the block start phase. The block start phase was divided into double-leg and single-leg stance phases based on the instant of rear-leg take-off. The black solid line represents the pelvis and front leg, the black dashed line represents the rear leg, and the grey line represents the other segments that were not analyzed in this study. doi:10.1371/journal.pone.0142230.g002 A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 4 / 13 where vi and vf are the anteroposterior velocities of COM at the start (here the vi = 0 m/s) and end of the block start phase, respectively, and Δt is the pushing duration during the block start phase. Here, the vf is equal to the GRF impulse of the anteroposterior component normalized by body mass during the block start (hereafter, normalized anteroposterior impulse, IN). The IN was calculated as follows [21]: IN ¼ Z t2 t1 apGRFSum dt=m ð3Þ where t1 and t2 are the times at which force application begins and ends, respectively, and apGRFSum is the anteroposterior component of the sum of GRFs from the legs during the block start phase. The normalized front-block-induced power, normalized rear-block-induced power, and normalized anteroposterior impulses in front and rear legs were calculated, after adjusting for Eqs (1), (2) and (3) based on the number of force plates. Mean value of anteropos- terior and mediolateral accelerations of COM during the block start phase (mean anteroposter- ior acceleration (apCOMacc) and mean mediolateral acceleration (mlCOMacc), respectively) was calculated as follows: apCOMacc ¼ apGRFSum=m ð4Þ mlCOMacc ¼ mlGRFSum=m ð5Þ where apGRFSum and mlGRFSum are anteroposterior and mediolateral components of sum of GRFs vector from the legs, respectively. Reaction time, pushing durations in front and rear legs during the block start phase, sprint times up to 2.0 m with the reaction time were calculated. Hip angle, moment and power data were calculated using an algorithm in Visual 3D (v4.86.0; C-motion, Inc., Germantown, MD, USA). The locations of the center of rotation of the hip [22], knee [23], and ankle [24] were estimated from anatomical landmarks using a pre- dictive approach. In each pelvic [24], thigh [24], shank [23], and foot [24] anatomical coordi- nate system, the x-axis represented the extension–flexion axis of segment rotation, the z-axis of the distal frame represented the external–internal rotation axis, and the axis orthogonal to the previous two at any given instant in time (y-axis) represented the abduction–adduction axis. Hip extension, abduction, and external rotation angles of front and rear legs were calculated using a hip joint coordinate system based on the x-y-z rotation sequence [24]. Knee and ankle joint coordinate systems were created using methods of Grood and Suntay [23] and Wu et al. [24], respectively. The hip extension moments of the front and rear/swing legs were calculated using a standard inverse dynamics approach [25], and were normalized by body mass (Mhip): positive value indicates extension moment whereas negative value indicates flexion moment. The hip extension angular velocity (ωhip) was calculated by Winter’s method: positive value indicates extension angular velocity whereas negative value indicates flexion angular velocity [25]. The hip joint power (Phip) for all sprinters was calculated as follows: Phip ¼ Mhipohip ð6Þ Positive power results when the hip joint moment acts in the same direction as the angular velocity of the hip joint (concentric action). Negative power results when the hip joint moment acts in the opposite direction as the angular velocity of the hip joint (eccentric action). We used cubic spline interpolation to normalize these values with respect to time, with 100% represent- ing the time of the block start phase. A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 5 / 13 Changes in normalized block-induced power were calculated (ΔNormalized block-induced power), and changes in the normalized block-induced power by legs as well as changes in the mean positive value of hip joint power in the widened condition relative to normal condition were calculated at each double- and single-stance phases (ΔNormalized block-induced power, ΔNormalized front-block-induced power, ΔNormalized rear-block-induced power, ΔFront-hip power generation, and ΔRear-hip power generation, respectively). These changes of variables in the widened condition relative to those normal condition (Δvar) were calculated as follows: Dvar ½% ¼ ðvarW=varN  1Þ  100 ð7Þ where varW and varN are mean values of time-series variables during each phase in the widened and normal conditions, respectively. Normalized block-induced powers in the two conditions to calculate ΔNormalized block-induced, ΔNormalized front-block-induced and ΔNormalized rear-block-induced powers were calculated adjusting Eqs (1), (2) and (3) based on the phase and number of force plates. For instance, when calculating the block-induced power for ΔNor- malized front-block-induced power during the double-stance phase in the widened condition, vi is 0 m/s, vf is the anteroposterior impulse on a force plate exerted by the front foot divided by body mass during the double-stance phase, and Δt is the duration during the double-stance phase in the widened condition. Statistical analysis For all parameters, the mean value of all three trials in each condition was used for further anal- ysis [2]. All parameters are shown as mean ± SD. We calculated the intraclass correlation coef- ficient (ICC(1,3)) for all parameters among three trials. The Lilliefors test was used to assess normality of variables. In the case of normally distributed samples, paired t-tests were used to assess the differences in the variables, and in the remaining, the Wilcoxon test was used for paired samples. Pearson’s correlation coefficient (r) was used to assess the relationships of changes in variables. Two-way repeated-measure analysis of variance and the post-hoc tests were used to assess the different effects of two conditions and legs on the variables (condition [widened and normal] x leg [front and rear]). The level of significance was set at P < 0.05. Results Table 1 shows the results of sprinting performance. Of the study participants, no significant difference in normalized block-induced power was seen between the widened and normal con- ditions (Fig 3). No significant difference was seen in the anteroposterior impulse, mean antero- posterior acceleration, reaction time, and duration of the block start phase between the two conditions. No significant difference was seen in the sprint time up to 2.0 m between the two conditions. All ICC(1,3) in sprinting performance exceeded 0.700 expect for reaction time (widened condition: ICC(1,3) = 0.626; normal condition (ICC(1,3) = 0.355). The hip extension angle in the front leg was not significantly different between widened and normal conditions during the double- and single-stance phases (Fig 4A). In contrast, the hip extension angle of the rear leg in the widened condition was significantly larger than that in the normal condition at the end of double-stance phase and the subsequent maximum extension angle (Fig 4B). The hip abduction angle in both legs in the widened condition was significantly larger than that in the normal condition during the block phase (Fig 4C and 4D). The hip external rotation angle of both legs in the widened condition was significantly lesser than that in the normal condition during the double-stance phase (Fig 4E and 4F). No significant differences were seen in the hip extension moments in both legs during the double-stance phase (Fig 5A and 5B), while the hip extension moment in the widened A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 6 / 13 condition was less than that in normal condition at the end of the single-stance phase. While no significant differences were seen in the hip extension angular velocity in the front leg during the block start phase (Fig 5C), the peak values of the hip extension and flexion angular velocity was significantly greater in the widened condition than that in the normal condition (Fig 5D). While no significant differences were seen in the hip power of the front leg between two condi- tions (Fig 5E), the positive peak value of the hip power of the rear leg in widened condition was significantly larger than that in the normal condition during double–stance phase (Fig 5F). ΔNormalized block-induced power during the block start phase significantly related to that during the double-stance phase (r = 0.668; P < 0.01), but did not significantly relate to that during the single-stance phase (r = −0.021; n.s.). During the double-stance phase, this ΔNor- malized block-induced power significantly related to the ΔNormalized rear-block-induced power (r = 0.793; P < 0.01), but did not significantly relate to the ΔNormalized front-block- induced power (r = 0.390; n.s.). During the double-stance phase, ΔNormalized rear-block- induced power significantly related to the ΔRear-hip power generation (r = 0.947; P < 0.01). Discussion This study aimed to clarify the effect of widened stance width at the set position during the block start phase in sprint running on kinematics and kinetics at the hip joint, and block- induced power. During the block phase, the hip abduction and external rotation angles in both legs and peak hip power generation in rear leg in the widened condition were significantly changed compared to those in the normal condition. However, in the widened condition, no significant changes in normalized block-induced power were seen compared to that in the Table 1. Mean ± SD, range and reliability of sprinting performance during the block start phase and the subsequent sprinting time in widened and normal conditions. Parameter Generator Widened condition Normal condition Δ% Mean ± SD Range ICC Mean ± SD Range ICC Mean ± SD Range Normalized block-induced power Both legs 0.543 ± 0.051 0.461–0.663 0.886 0.539 ± 0.053 0.461–0.668 0.891 0.9 ± 5.3 −8.5–10.8 Front leg 0.219 ± 0.023† 0.187–0.259 0.826 0.241 ± 0.042† 0.183–0.344 0.864 −7.5 ± 13.7† −24.8–27.9 Rear leg 0.130 ± 0.030* 0.091–0.191 0.954 0.113 ± 0.034 0.042–0.161 0.875 22.6 ± 37.0 −9.7–126.7 Anteroposterior impulse (Ns/kg) Both legs 3.20 ± 0.20 2.94−3.57 0.961 3.20 ± 0.18 2.98–3.53 0.938 −0.1 ± 2.3 −5.8–3.4 Front leg 2.03 ± 0.11*† 1.89–2.25 0.924 2.14 ± 0.14† 1.95–2.41 0.929 −4.6 ± 5.9† −14.1–5.5 Rear leg 1.15 ± 0.16* 0.89–1.40 0.910 1.05 ± 0.22 0.55–1.30 0.705 12.1 ± 18.7 −13.1–60.2 Pushing duration during the block start phase (s) Both legs 0.330 ± 0.025 0.292–0.368 0.952 0.334 ± 0.031 0.298–0.420 0.946 −0.9 ± 4.5 −12.4–4.1 Front leg 0.330 ± 0.025† 0.292–0.368 0.952 0.334 ± 0.031† 0.298–0.420 0.946 −0.9 ± 4.5 −12.4–4.1 Rear leg 0.180 ± 0.023 0.144–0.224 0.896 0.175 ± 0.034 0.132–0.276 0.932 3.7 ± 8.7 −18.8–13.6 Mean anteroposterior acceleration (m/s2) ― 9.73 ± 0.59 8.73–10.82 0.843 9.65 ± 0.72 8.10–11.03 0.923 1.0 ± 4.6 −6.2–12.8 Mean mediolateral acceleration (m/s2) ― −0.70 ± 0.47* −1.56–−0.14 0.927 −0.57 ± 0.41 −1.52–−0.05 0.924 50.0 ± 65.1 −43.5–202.0 Reaction time (s) ― 0.180 ± 0.023 0.144–0.224 0.626 0.179 ± 0.016 0.154–0.206 0.355 0.6 ± 9.0 −16.7–18.6 Sprint time up to 2 m (s) ― 0.810 ± 0.043 0.738–0.906 0.836 0.797 ± 0.046 0.734–0.900 0.883 1.7 ± 5.1 −4.8–15.7 *Significant difference from normal condition (P < 0.05). †,§ Significant difference from rear leg (P < 0.05). doi:10.1371/journal.pone.0142230.t001 A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 7 / 13 normal condition. This suggests that the response to widened stance width on the normalized block-induced power was not remarkable. Previous studies have reported a significant effect of body posture in the sagittal plane on block performance [5,7,8]. A lower block angle (40°) leads sprinters to a 3.6% higher block- induced impulse than a higher block angle (65°) [5]. Other previous study has reported that the duration during the block start phase of the elongated start was 9.8% and 11.6% longer than that of medium and bunched starts, respectively [7]. In this study, the hip abduction and internal rotation angles of the front and rear legs during the double-stance phase were larger with a widened stance width than with a normal stance width; therefore, there was a change in the body position. However, no significant difference was seen in the sprinting performance, including the normalized block-induced power during the block start (−0.9 ± 5.2%), between the widened and normal conditions. These suggest that the changes in the set position that are related to the stance width have a lesser effect on the block performance relative to those related to the anteroposterior position and block angles. We focused specifically on the hip power generation, which is considered as the key power for enhancing normalized block-induced power in lower limbs, rather than the power at the knee and ankle [2,5]. During the double-stance phase, the hip joint in the front and rear legs were generated power to induce hip extension, and during the single-stance phase, the hip in the rear leg was generated power to induce hip flexion. This was corresponding with the find- ings of the previous studies [5,9]. When the stance width is increased from narrow to wide dur- ing the squatting motion, the hip extensor muscle activity [12–14,26] and the maximum strength in lower limbs [11] increase. Demura et al. [11] compared the leg muscle strength Fig 3. Normalized block-induced power in widened and normal conditions. The bold solid line indicates the mean value of normalized block-induced power in all participants, and the thin dashed lines indicate the normalized block-induced power in each participant (n = 14). doi:10.1371/journal.pone.0142230.g003 A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 8 / 13 Fig 4. Changes in the hip angles during the block start phase normalized with respect to time. (A) Hip extension angles in front leg. (B) Hip extension angle in rear leg. (C) Hip abduction angle in front leg. (D) Hip abduction angle in rear leg. (E) Hip external rotation angle in front leg. (F) Hip external rotation angle in rear leg. The black and grey lines indicate the mean (bold) ± SD (thin) of the time-series data of the front and rear/swing legs in widened and normal A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 9 / 13 between narrow (5 cm) and wide (140% width of the shoulders) stance widths in a squat. They found that the exerted maximum lower limb force in wide stance width was greater than that in narrow stance width. Similarly, in this study, the peak power generation at the hip in the rear leg was greater during the double-stance phase in the widened condition than that in the normal condition. This was due to enhancing the hip extension angular velocity in the rear leg. During the double-stance phase in the widened condition, hip in the rear leg was abducted and internally rotated, which was not the case in the normal condition. Perhaps this changed a property of the muscle contraction in hip extensor muscles during the double-stance phase and enhanced the muscle’s contraction velocity, which associates with the hip extension angu- lar velocity, in the widened condition. In addition, the ΔRear-hip power generation was signifi- cantly associated with the ΔRear-block-induced normalized block-induced power, indicating that our second hypothesis was accepted in the rear leg. This suggests that those who preferred a widened stance width at the set position could increase the hip power generation in the rear leg by the optimal hip position and could enhance the normalized block-induced power during the double-stance phase. These findings supported the previous study demonstrating the importance of the rear leg [2]. However, when considering sprinting performance during the whole block start phase, no sig- nificant difference was seen in all sprinting performances between the two conditions. The block start phase involves the single-stance phase in which sprinters have to generate power with sin- gle-leg stance, in contrast to the squat motion. Therefore, it can be considered that during the sin- gle-stance phase in the widened condition, sprinters must push the block to a different direction relative to the normal condition. Indeed, mean mediolateral acceleration was less in the widened condition than in the normal condition, suggesting that the sprinter’s COM leaned toward the first step (rear leg) side during the single-stance phase. Thus, our first hypothesis in this study was rejected and did not correspond with the findings for the squatting position [11]. There are two limitations in this study. First, the number of combinations among stance width, block angles, and anteroposterior distance were limited for the starting blocks. The stan- dardized set up was prepared with block angles and anteroposterior distance between the blocks. Therefore, each sprinter probably was not allowed to create their individual optimal set up for the starting blocks and probably could not perform their best block start in the normal condition. Moreover, the stance width in block start was not normalized by each participant’s body characteristics in the normal or widened conditions. We did not perform normalizations by body characteristics because the same stance width is conventionally used for all sprinters in competitive races, which is corresponded with that in normal condition. These might affect sprinting performance during the block start phase, changes in variables in the widened condi- tion relative to the normal condition, and reliability of the results. However, the number of exper- imental trials should be small because the participant cannot repeatedly perform the sprint starts with maximal effort (no more than six trials) [27]. We could prepared only two conditions at the set position during the block start phase, and participant performed a total of six block-start tri- als. This was because of the unavoidable number of conditions as the block start experiment. Sec- ond, this study was conducted without the familiarization with the widened stance width in the block start. All participants were sprint specialists with enough training experience, suggesting that they were familiar with the set position in the normal condition relative to the widened conditions. These angles at the initial and end instants of double- and single-leg phases and at the instant when the peak value occurs were compared between the two conditions. Significant differences between the two conditions are shown as * (P < 0.05). Significant difference between the front and rear legs in widened condition are shown as † (P < 0.05) and that in normal condition are shown as § (P < 0.05). The vertical dashed line indicates the instant of rear leg take-off during the block start phase. doi:10.1371/journal.pone.0142230.g004 A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 10 / 13 Fig 5. Changes in the hip moment, angular velocity, and joint power normalized with respect to time. (A) Hip extension moment in front leg. (B) Hip extension moment in rear leg. (C) Hip extension angular velocity in front leg. (D) Hip extension angular velocity in rear leg. (E) Hip joint power in front leg. (F) Hip joint power in rear leg. The black and grey lines indicate the mean (bold) ± SD (thin) of the time-series data of the front and rear/swing legs in widened and A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 11 / 13 condition. Even with extensive training in the widened condition before the experimental trial, the normal condition would be optimal to enhance the block-induced power relative to the wid- ened condition. Despite these limitations, we have reported some new information that can serve as baseline data for future studies on developing the newly designed starting block. In conclusion, a widened stance width at set position which we prepared in this study affected the hip-joint kinematics in both legs and hip power generation in the rear leg during the block start phase. However, when considering sprinting performance during the whole block start phase, there were no significant effect of the widened stance width on block-induced power and the subsequent sprint time. Supporting Information S1 Video. Trial in the widened condition. (WMV) S2 Video. Trial in the normal condition. (WMV) Author Contributions Conceived and designed the experiments: MO. Performed the experiments: MO. Analyzed the data: MO. Contributed reagents/materials/analysis tools: MO TI. Wrote the paper: MO TK TI. References 1. Bezodis NE, Salo AI, Trewartha G. Choice of sprint start performance measure affects the perfor- mance-based ranking within a group of sprinters: which is the most appropriate measure? Sports Bio- mech. 2010; 9: 258–269. doi: 10.1080/14763141.2010.538713 PMID: 21309300 2. Bezodis NE, Salo AI, Trewartha G. Relationships between lower-limb kinematics and block phase per- formance in a cross section of sprinters. Eur J Sport Sci. 2015; 15: 118–124. doi: 10.1080/17461391. 2014.928915 PMID: 24963548 3. International Association of Athletic Federations. A. Competition Rules 2014–2015. 2013;154–158. 4. Guissard N, Duchateau J, Hainaut K. EMG and mechanical changes during sprint starts at different front block obliquities. Med Sci Sports Exerc. 1992; 24: 1257–1263. doi: 10.1249/00005768- 199211000-00010 PMID: 1435177 5. Mero A, Kuitunen S, Harland M, Kyrolainen H, Komi PV. Effects of muscle-tendon length on joint moment and power during sprint starts. J Sports Sci. 2006; 24: 165–173. doi: 10.1080/ 02640410500131753 PMID: 16368626 6. Milanese C, Bertucco M, Zancanaro C. The effects of three different rear knee angles on kinematics in the sprint start. Biol Sport. 2014; 31: 209–215. doi: 10.5604/20831862.1111848 PMID: 25177099 7. Slawinski J, Dumas R, Cheze L, Ontanon G, Miller C, Mazure-Bonnefoy A. 3D kinematic of bunched, medium and elongated sprint start. Int J Sports Med. 2012; 33: 555–560. doi: 10.1519/JSC. 0b013e3181ad3448 PMID: 22499565 8. Slawinski J, Dumas R, Cheze L, Ontanon G, Miller C, Mazure-Bonnefoy A. Effect of postural changes on 3D joint angular velocity during starting block phase. J Sports Sci. 2013; 31: 256–263. doi: 10.1080/ 02640414.2012.729076 PMID: 23062070 9. Debaere S, Delecluse C, Aerenhouts D, Hagman F, Jonkers I. From block clearance to sprint running: Characteristics underlying an effective transition. J Sport Sci. 2013; 31: 137–149. doi: 10.1080/ 02640414.2012.722225 normal conditions. These angles at the initial and end instants of double- and single-leg phases and at the instant when the peak value occurs were compared between the two conditions. Significant differences between the two conditions are shown as * (P < 0.05). Significant difference between the front and rear legs in widened condition are shown as † (P < 0.05) and that in normal condition are shown as § (P < 0.05). The vertical dashed line indicates the instant of rear leg take-off during the block start phase. doi:10.1371/journal.pone.0142230.g005 A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 12 / 13 10. Okkonen O, Hakkinen K. Biomechanical comparison between sprint start, sled pulling, and selected squat-type exercises. J Strength Cond Res. 2013; 27: 2662–2673. doi: 10.1519/JSC. 0b013e31829992b0 PMID: 23760361 11. Demura S, Miyaguchi K, Shin S, Uchida Y. Effectiveness of the 1RM estimation method based on iso- metric squat using a back-dynamometer. J Strength Cond Res. 2010; 24: 2742–2748. doi: 10.1519/ JSC.0b013e3181e27386 PMID: 20885196 12. Clark DR, Lambert MI, Hunter AM. Muscle activation in the loaded free barbell squat: a brief review. J Strength Cond Res. 2012; 26: 1169–1178. doi: 10.1519/JSC.0b013e31822d533d PMID: 22373894 13. McCaw ST, Melrose DR. Stance width and bar load effects on leg muscle activity during the parallel squat. Med Sci Sports Exerc. 1999; 31: 428–436. doi: 10.1097/00005768-199903000-00012 PMID: 10188748 14. Paoli A, Marcolin G, Petrone N. The effect of stance width on the electromyographical activity of eight superficial thigh muscles during back squat with different bar loads. J Strength Cond Res. 2009; 23: 246–250. doi: 10.1519/JSC.0b013e3181876811 PMID: 19130646 15. Otsuka M, Shim JK, Kurihara T, Yoshioka S, Nokata M, Isaka T. Effect of expertise on 3D force applica- tion during starting block phase and subsequent steps in sprint running. J Appl Biomech. 2014; 30: 390–400. doi: 10.1123/jab.2013-0017 PMID: 24615252 16. Harland MJ, Steele JR. Biomechanics of the sprint start. Sports Med. 1997; 23: 11–20. doi: 10.2165/ 00007256-199723010-00002 PMID: 9017856 17. Lieberman DE, Venkadesan M, Werbel WA, Daoud AI, D’Andrea S, Davis IS, et al. Foot strike patterns and collision forces in habitually barefoot versus shod runners. Nature. 2010; 463: 531–536. doi: 10. 1038/nature08723 PMID: 20111000 18. Dempster WT, Gaughran G. Properties of body segments based on size and weight. Am J Anat. 1967; 120: 33–54. 19. Hanavan EP, Jr. A mathematical model of the human body. Amrl-Tr-64-102. AMRL TR. 1964;1–149. 20. Hof AL. Scaling gait data to body size. Gait Posture. 1996; 4: 222–223. doi: 10.1016/0966-6362(95) 01057-2 21. Mendoza L, Schöllhorn W. Training of the sprint start technique with biomechanical feedback. J Sports Sci. 1993; 11: 25–29. doi: 10.1080/02640419308729959 PMID: 8450581 22. Bell AL, Pedersen DR, Brand RA. A comparison of the accuracy of several hip center location predic- tion methods. J Biomech. 1990; 23: 617–621. doi: 10.1016/0021-9290(90)90054-7 PMID: 2341423 23. Grood ES, Suntay WJ. A joint coordinate system for the clinical description of three-dimensional motions: application to the knee. J Biomech Eng. 1983; 105: 136–144. PMID: 6865355 24. Wu G, Siegler S, Allard P, Kirtley C, Leardini A, Rosenbaum D, et al. ISB recommendation on defini- tions of joint coordinate system of various joints for the reporting of human joint motion—part I: ankle, hip, and spine. J Biomech. 2002; 35: 543–548. doi: 10.1016/S0021-9290(01)00222-6 PMID: 11934426 25. Winter DA. Biomechanics and Motor Control of Human Movement. New York: Wiley and Sons Press; 2009. 26. Mirakhorlo M, Azghani MR, Kahrizi S. Validation of a musculoskeletal model of lifting and its application for biomechanical evaluation of lifting techniques. J Res Health Sci. 2014; 14: 23–28. PMID: 24402846 27. Bradshaw EJ, Maulder PS, Keogh JWL. Biological movement variability during the sprint start: Perfor- mance enhancement or hindrance? Sports Biomech. 2007; 6: 246–260. doi: 10.1080/ 14763140701489660 PMID: 17933190 A Wide Stance on Block Start Performance PLOS ONE | DOI:10.1371/journal.pone.0142230 November 6, 2015 13 / 13

Effect of a Wide Stance on Block Start Performance in Sprint Running.

11-06-2015

Otsuka, Mitsuo,Kurihara, Toshiyuki,Isaka, Tadao

eng

PMC5350167

ORIGINAL RESEARCH Heavy strength training improves running and cycling performance following prolonged submaximal work in well-trained female athletes Olav Vikmoen1, Bent R. Rønnestad1, Stian Ellefsen1 & Truls Raastad2 1 Section for Sport Sciences, Lillehammer University College, Lillehammer, Norway 2 Deparment of Physical Performance, Norwegian School of Sport Sciences, Oslo, Norway Keywords Concurrent training, cycling economy, prolonged cycling, prolonged running, running economy. Correspondence Olav Vikmoen; Norwegian Defence Research Establishment (FFI), PO Box 26 N-2027 Kjeller, Norway. Tel: +47 63807825 Fax: +47 63807892 Email: olav.vikmoen@ffi.no Funding information This work was supported by grant 203961 from the Regional Science Fund - Innlandet of Norway. Received: 10 January 2017; Accepted: 11 January 2017 doi: 10.14814/phy2.13149 Physiol Rep, 5 (5), 2017, e13149, doi: 10.14814/phy2.13149 Abstract The purpose of this study was to investigate the effects of adding heavy strength training to female duathletes’ normal endurance training on both cycling and running performance. Nineteen well-trained female duathletes (VO2max cycling: 54
3 ml∙kg1∙min1, VO2max running: 53
3 ml∙kg1∙min1) were randomly assigned to either normal endurance training (E, n = 8) or normal endurance training combined with strength training (E+S, n = 11). The strength training consisted of four lower body exercises [3 9 4-10 repetition maximum (RM)] twice a week for 11 weeks. Running and cycling performance were assessed using 5-min all-out tests, per- formed immediately after prolonged periods of submaximal work (3 h cycling or 1.5 h running). E+S increased 1RM in half squat (45
22%) and lean mass in the legs (3.1
4.0%) more than E. Performance during the 5-min all-out test increased in both cycling (7.0
4.5%) and running (4.7
6.0%) in E+S, whereas no changes occurred in E. The changes in running perfor- mance were different between groups. E+S reduced oxygen consumption and heart rate during the final 2 h of prolonged cycling, whereas no changes occurred in E. No changes occurred during the prolonged running in any group. Adding strength training to normal endurance training in well-trained female duathletes improved both running and cycling performance when tested immediately after prolonged submaximal work. Introduction During the last decade, increased attention has been given to the effects of adding strength training to endurance athletes’ normal training on running and cycling perfor- mance (e.g., Paavolainen et al. 1999; Aagaard et al. 2011; Ronnestad et al. 2011; Sedano et al. 2013). Improvements in performance have been reported in both running (Paa- volainen et al. 1999; Storen et al. 2008; Sedano et al. 2013; Damasceno et al. 2015) and cycling (Koninckx et al. 2010; Ronnestad et al. 2010a; Aagaard et al. 2011; Ronnestad et al. 2015). However, the literature is far from conclusive, and numerous studies do not report such improvements in neither running (Ferrauti et al. 2010; Roschel et al. 2015) nor cycling (Bishop et al. 1999; Basti- aans et al. 2001; Levin et al. 2009). Some methodological differences may explain these equivocal findings. To posi- tively affect cycling performance, it seems that the strength training regime needs to involve heavy loads, typically between 10 and 4 repetition maximum (RM) (Koninckx et al. 2010; Ronnestad et al. 2010a; Aagaard et al. 2011; Ronnestad et al. 2015). To improve running performance on the other hand, both explosive, plyomet- ric and heavy strength training seems effective (Paavolai- nen et al. 1999; Sedano et al. 2013; Damasceno et al. 2015). To the best of our knowledge, only one study has ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. 2017 | Vol. 5 | Iss. 5 | e13149 Page 1 Physiological Reports ISSN 2051-817X investigated the effect of strength training on performance in both cycling and running in the same athletes. This study reported increased time to exhaustion at VO2max in both cycling and running (Hickson et al. 1988). However, the study did not include an endurance training only group, and therefore the results should be interpreted with caution. The observation that somewhat different strength train- ing regimes affect performance in cycling and running indicates that some of the performance-enhancing mecha- nisms may differ between these sports. Suggested mecha- nisms by which strength training can improve cycling and running performance include changes in rate of force development, changes in tendon stiffness, changes in movement mechanics, and changes in muscular character- istics such as increased muscle strength, muscle mass, and improved anaerobic capacity (Saunders et al. 2006; Ronnestad and Mujika 2014). Some of these factors may be important for performance in both running and cycling, whereas other mechanisms may affect perfor- mance differently in these sports. For example, in run- ning, the stretch-shortening cycle in each stride enables the possibility to store and recoil elastic energy, whereas in cycling, the possibilities to take advantage of stored elastic energy is negligible. Consequently, a factor such as muscle-tendon stiffness may play a role for running per- formance, but likely not for cycling performance. On the other hand, a factor like improved anaerobic capacity should affect performance to the same degree in both running and cycling. Road races in cycling often consist of a long initial period of cycling at a moderate intensity, followed by an all-out performance at the end. Even though running competitions are ran at a more even pace, they are also often decided with an all-out effort in the end. During such efforts, a quite large proportion of the energy demand will come from anaerobic sources (Gastin 2001). Therefore, performance during a relatively short test will in addition to VO2max and other aerobic parameters also be largely influenced by anaerobic capacity. Muscle mass is an important determinant of anaerobic capacity (Bangsbo et al. 1993). We have previously reported increased CSA of m. quadriceps femoris after 11 weeks of heavy strength training in female endurance athletes together with increased mean and peak power during the Wingate test (Vikmoen et al. 2016a). This indicates improved anaerobic capacity in the same athletes included in this study. There- fore, performance in a quite short performance test should be positively affected by this strength training regime. In addition to increased muscle CSA, changes in protein levels and expression of genes coding for proteins that are involved in the anaerobic metabolism might contribute to increased anaerobic performance. Performance in an all-out effort at the end of long competitions should also be affected by the fatigue devel- oped during the competition. In Ronnestad et al. (2011), such performance was simulated by 3 h of submaximal cycling followed by a 5-min all-out test. Power output during the 5-min all-out test was improved following 12 weeks of heavy strength training in well-trained male cyclists. This was related to improved cycling economy and reduced physiological strain during the final hour of the submaximal trial, leaving the strength-trained athletes less fatigued before the 5-min all-out test (Ronnestad et al. 2011). However, no previous study has assessed effects of heavy strength training on all-out performance following a prolonged submaximal work or physiological responses during prolonged submaximal running. The primary purpose of this study was to investigate the effects of 11 weeks of heavy strength training on 5-min all-out performance after separate trials of pro- longed submaximal work in both running and cycling and on physiological responses during the prolonged work. We especially wanted to identify performance- enhancing mechanisms after strength training which acts similarly and differently on cycling and running perfor- mance. We hypothesized that the addition of heavy strength training would result in improved 5-min all-out perfor- mance in both cycling and running. Furthermore, we hypothesized that changes in 5-min all-out performance would be related to improved work economy during the prolonged trials and to changes related to anaerobic capacity such as increased muscle mass and changes in expression of genes that are involved in anaerobic pro- cesses. We also anticipated that some of the underlying mechanisms for improved work economy would differ between running and cycling. Methods Ethical approval The study was approved by the Local Ethics Committee at Lillehammer University College. Written informed con- sent was obtained from all athletes prior to inclusion, and the study was carried out in accordance with the Declara- tion of Helsinki. Participants Twenty-eight female duathletes who fulfilled at least two of Jeukendrup et al. (2000) training and race status descriptions of a well-trained athlete were recruited to this study. None of the athletes had performed systematic strength training for the last 12 months leading up to the 2017 | Vol. 5 | Iss. 5 | e13149 Page 2 ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. Strength Training and Endurance Performance O. Vikmoen et al. study. The athletes were matched on VO2max and ran- domly assigned to either adding heavy strength training to the ongoing endurance training (E + S, n = 14) or endurance training only (E, n = 14). During the study, three athletes in E + S left the project for reasons unre- lated to the project protocol: one because of an injury, one because of a prolonged period of illness during the last part of the intervention and one because of other medical reasons. In E, six athletes left the study for rea- sons unrelated to the project protocol (injuries from bicy- cle crash, pregnancy, and lack of time). Therefore, the final numbers of athletes in E + S and E were 11 and 8, respectively. Experimental overview This study is part of a larger study investigating the effects of heavy strength training on various aspects of cycling and running performance. The effect on time-trial performance and traditional performance determinants in cycling and running has been previously reported (Vikmoen et al. 2016a,b). Whenever data from these studies are utilized for correlation purposes or otherwise, it will be clearly speci- fied. The strength training program for the E+S group con- sisted of two strength training sessions per week and lasted for 11 weeks (during the competition period from April to July). The testing before and after the intervention period was organized in five test days. During pretests, test day 1 consisted of biopsy sampling from m. vastus lateralis for determination of muscle fiber type composition and mRNA expression of genes related to fat and anaerobic metabolism. Test day 2 consisted of a VO2max test in cycling followed by 1RM test in half squat. Test day 3 con- sisted of a VO2max test in running. Test day 4 consisted of a prolonged submaximal running trial followed by a 5-min all-out test. Test day 5 consisted of a prolonged submaxi- mal cycle trial followed by a 5-min all-out test. There were at least 7 days between day 1 and 2 and 3–7 days between the remaining test days. After the intervention period, the only difference in test order was that muscle biopsies were taken on the last test day. Training Endurance training duration and intensity were calculated based on heart rate (HR) recordings. Endurance training was divided into three HR zones: (1) 60%–82%, (2) 83%–87%, and (3) 88%–100% of maximal HR. For detailed information on endurance training characteris- tics, see Vikmoen et al. (2016a). Briefly, there were no significant differences between groups in their average weekly endurance training duration or distribution between intensity zones. The heavy strength training for the E + S groups tar- geted leg muscles and were performed twice per week during the 11-week intervention period. Adherence to the strength training was high, with E + S athletes completing 21.4
1.0 (range 19–22) of the planned 22 strength training sessions. The strength training program was per- formed as reported in Vikmoen et al. (2016a). Briefly, each strength training session consisted of four leg exer- cises: half squat in a smith machine, leg press with one leg at a time, standing one-legged hip flexion, and ankle plantar flexion. Three sets were performed per exercise. An investigator supervised the athletes at all workouts during the first 2 weeks and at least one workout per week thereafter. During weeks 1–3, athletes trained with 10RM sets at the first session and 6RM sets at the second session. These alternating loads were adjusted to 8RM and 5RM during weeks 4–6, and was further adjusted to 6RM and 4RM during weeks 7–11. The athletes were encouraged to increase their RM loads continually throughout the intervention period and they were allowed assistance on the last repetition. Physical performance tests The athletes were instructed to refrain from intense exer- cise the day preceding testing and to prepare for the tests as they would have done for a competition. This included consuming the same type of meal at the same time as they would do if the test was a regular competition. Fur- thermore, the participants were instructed to replicate the preparation before every test. All cycling tests were per- formed on a electromagnetically braked cycle ergometer (Lode Excalibur Sport, Lode B. V., Groningen, The Netherlands), which was adjusted according to each ath- lete preference for seat height, horizontal distance between tip of seat and bottom bracket, and handlebar position. During all cycling tests the ergometer was in a cadence-independent mode (constant watt-production); so, the power output was not affected by the cyclists‘ cho- sen cadence. The running tests were performed on a motor-driven treadmill (Woodway Desmo Evo, Wauke- sha, WI). The inclination of the treadmill was set to 5.3% at all tests. All testing were performed under similar envi- ronmental conditions (18–20°C). VO2max in cycling The cycling VO2max test protocol utilized in this study and its results has been described elsewhere (Vikmoen et al. 2016a). Briefly, the test was initiated with 1-min cycling at a power output of 100 W that was subsequently increased by 25 W every minute until exhaustion. VO2 was measured (30-sec sampling time) using a ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. 2017 | Vol. 5 | Iss. 5 | e13149 Page 3 O. Vikmoen et al. Strength Training and Endurance Performance computerized metabolic system with mixing chamber (Oxycon Pro, Erich Jaeger, Hoechberg, Germany). The gas analyzers were calibrated with certified calibration gases of known concentrations before every test. The flow turbine (Triple V, Erich Jaeger, Hoechberg, Germany) was calibrated before every test with a 3 l, 5530 series, calibration syringe (Hans Rudolph, Kansas City, USA). VO2max was calculated as the average of the two highest 30 sec VO2 measurements. Peak cycling performance dur- ing the test (Wmax) was calculated as the mean power output during the last 2 min of the incremental test. After the test, blood [la] and HRpeak was noted. [La] were analyzed in whole blood with a Lactate Pro LT-1710 analyzer (Arcray Inc., Kyoto, Japan). RPE was recorded using the Borg scale (Borg, 1982). HR was measured using a Polar S610i heart rate monitor (Polar, Kempele, Finland). Prolonged submaximal cycling followed by a 5-min all-out cycling test The prolonged cycling lasted for 180 min on a power output corresponding to 44% of Wmax (111
9 W and 116
8 W in E + S and E, respectively). The same abso- lute power output was utilized post intervention. VO2 and HR were determined during 3-min periods every 30th min throughout the prolonged cycling and RPE and [la] were measured every 30th min. Average values for each hour were calculated and used for statistical analyses. Athletes were allowed to occasionally stand in the pedals during the prolonged cycling, but not during the 3-min periods of measurements and not during the final 5-min all-out test. Athletes were allowed to consume water and a sport drink containing 60 g/L carbohydrates, ad libitum, in order to maintain fluid balance and mimic race condi- tions. The amount of sport drink consumed were similar between groups and from pre to post (across groups, val- ues were 1.24
0.57 L and 1.26
0.59 L, respectively). After conclusion of the prolonged cycling, athletes were allowed a 3-min rest before a 5-min all-out test for deter- mination of cycling performance. During the first minute of the test, the power output was set by the investigators. This individual selected power output was based on pilot work and corresponded to 85% of Wmax. Thereafter, the control unit for the power output was put next to the ergometer and the athletes were allowed to adjust the power output themselves with the instruction to cycle at the highest average power output as possible. The partici- pant received feedback regarding power output and elapsed time, but not HR or cadence. Performance was measured as the mean power output during the 5-min all-out test. At the posttest, one athlete in E + S had to withdraw during the prolonged test due to pain in the hip. Therefore, the final numbers included in the statisti- cal analysis of these tests are 10 in E + S and 8 in E. VO2max in running The VO2max test protocol utilized in this study and its results have been described elsewhere (Vikmoen et al. 2016b). Briefly, the test was initiated with 1-min running at 8 kmh1 that was subsequently increased by 1 kmh1 every minute until exhaustion. VO2max was calculated as the average of the two highest 30 sec VO2 measurements. Peak running performance during the test (Vmax) was cal- culated as the mean running velocity during the last 2 min of the incremental test. Prolonged submaximal running followed by a 5-min all-out running test The prolonged running lasted for 90 min at a speed cor- responding to 60% of Vmax (7.7
0.4 kmh1 and 7.9
0.3 kmh1 in E + S and E, respectively). Each par- ticipant ran at the same absolute speed at both pretrial and posttrial. VO2 and HR were measured during 3-min periods every 15th min throughout the prolonged run- ning and RPE and [la] were measured every 15th min. Average values for each 30-min period were calculated and used for statistical analyses. The athletes were allowed to consume water and a sport drink containing 60 gL1 carbohydrates, ad libitum, in order to maintain fluid bal- ance. The amount of sport drink consumed was similar between groups and from pre to post (across groups val- ues were 0.76
0.27 L and 0.72
0.24 L, respectively). After conclusion of the prolonged running, the athletes were allowed a 3-min rest before a 5-min all-out test was performed for determination of running performance. During the first minute of the test, the speed was set by the investigators. This individual selected speed was based on pilot work and corresponded to 85% of Vmax. There- after, the athletes were allowed to adjust the speed them- selves with the instruction to run as fast as possible. The athletes received feedback on speed and elapsed time, but not HR or distance. Performance was measured as the distance covered during the 5-min all-out test. 1RM tests Approximately 20 min after termination of the cycling VO2max test, maximal strength in the legs was tested as 1RM in half squat. The 1RM protocol used has been described elsewhere (Vikmoen et al. 2016a). Briefly, the 1RM test started with a specific warm-up, consisting of three sets with gradually increasing load (40, 75, and 85% of expected 1RM) and decreasing number of repetitions 2017 | Vol. 5 | Iss. 5 | e13149 Page 4 ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. Strength Training and Endurance Performance O. Vikmoen et al. (10?6?3). The first attempt was performed with a load approximately 5% below the expected 1RM. If a lift was successful, the load was increased by approximately 5%. The test was terminated when the athletes failed to lift the load in 2–3 attempts and the highest successful load lifted was noted as 1RM. Athletes were given a 3-min rest between lifts. Lean mass in the legs Lean mass in the legs (LegLM) was determined by dual-energy X-ray absorptiometry using a Lunar Prodigy densiometer (Prodigy Advance PA+302047, Lunar, San Francisco, CA, USA). The athletes were instructed to refrain from training for the 24 h leading up to the mea- surement. They were also instructed to not ingest any food or liquid for the 3 h preceding the measurement. The same trained technician performed all DXA scans on each participant. Care was taken to position the body at the same location at each measurement. Muscle biopsy sampling Muscle biopsies were sampled from m. vastus lateralis using the Bergstr€om procedure and treated as previously described (Vikmoen et al. 2016a). An appropriately sized muscle sample was excised and selected for quantitative real-time PCR (qRT-PCR) analyses (average wet weight
SD: 38
7 mg), and a similarly sized sample was selected for immunohistochemical analyses (average wet weight
SD: 34
13 mg). Pre- and post-biopsies were sampled at the same time of day for each particular athlete. Athletes were instructed to refrain from physical activity during the last 24 h before biopsy sampling and not to ingest any food the 3 h preceding the biopsy. Biop- sies for qRT-PCR analyses were immersed immediately in RNAlater and treated according to manufacturers’ proto- col before storage at 80°C (Ambion, Foster City, CA). Biopsies for immunohistochemical analyses were formaldehyde fixated (Chemi-teknik AS, Oslo, Norway). Muscle biopsy analyses Immunohistochemistry Protocols for immunohistochemical analyses of muscle fiber type composition and the results have been presented else- where (Vikmoen et al. 2016a). Briefly, formalin-fixed muscle biopsies were paraffin-embedded and sectioned, whereupon transverse, serial sections were labeled for MyHCI (A4.840, H. Blau, Stanford, USA; Developmental Studies Hybridoma Bank), MyHCIIA (EPR5280, Nordic Biosite), and MyHCIIX (6H1, C Lucas, Sydney, Australia; Developmental Studies Hybridoma Bank). Determination of muscle fiber composi- tion was performed using Photoshop CS6 Extended (Adobe, San Jose, CA). The investigator performing the image analy- ses were blinded as to which group the athlete belonged. Muscle fibers that were positive for both MyHCIIA and MyHCIIX are referred to as muscle fiber type IIAX-IIX (Vik- moen et al. 2016a). Because of technical problems with some analyses, the number of individuals in the immunohisto- chemistry data is eight in E + S and eight in E. Gene expression Gene expression was assessed for genes involved in fatty acid oxidation and anaerobic energy metabolism. Primer design, RNA extraction, quantitative PCR (qPCR), and evaluation of the stability of reference genes was per- formed as previously described (Ellefsen et al. 2014). b2- microglobulin and ribosomal protein L32 were found to be the two most stable references genes and were utilized for calculation of normalization factors using GeNorm, which were in turn utilized for calculation of target gene expression. All genes with associated primers are presented in Table 1. Statistics All data in the text, figures, and tables are presented as mean
standard deviation, unless otherwise stated. Prior to statistical testing, gene expression were log2-trans- formed to maximize the likelihood of normal distribution. Unpaired students t-tests were used to test for differ- ences between groups at pre and post, and differences in changes from pre to post, except for evaluating responses during the prolonged trials. Within-group analyses were performed using paired t-tests except for evaluating responses during the prolonged trials. To evaluate changes in responses during the prolonged trials within groups (pre to post) a two-way repeated measures analysis of variance (ANOVA) (time of inter- vention period and time during the prolonged trials as factors) with Sidek-Holm post hoc test was performed. To evaluate differences in changes in the responses during the prolonged trials between the groups, a two-way repeated measures ANOVA (changes from pre to post in each group and time point during the prolonged trial as factors) with Sidek-Holm post hoc test were performed. In addition, effect sizes for the key performance and physiological adaptations were calculated to compare the practical significance between the two groups. Effects size were calculated as Cohen’s d and the criteria to inter- pret the magnitude were the following: 0–0.2 = trivial, 0.2–0.6 = small, 0.6–1.2 = moderate, 1.2–2.0 = large, and ˃2 = very large (Hopkins et al. 2009). ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. 2017 | Vol. 5 | Iss. 5 | e13149 Page 5 O. Vikmoen et al. Strength Training and Endurance Performance Correlations analyses were done using the Pearson pro- duct-moment method and correlations coefficients were interpreted according to Hopkins et al. (2009); r ˂ 0.1 trivial, 0.1–0.3 = small, 0.3–0.5 = moderate, 0.5–0.7 = large, 0.7–0.9 = very large, 0.9 = nearly perfect, and 1.0 = perfect. Analyses were performed in GraphPad Prism 6 (Graph- Pad Software Inc., CA) and Excel 2013 (Microsoft Corpo- ration, Redmon, WA). All analyses resulting in P ≤ 0.05 were considered statistically significant. Results There were no significant differences between E + S and E at preintervention in any of the measured variables. Body mass, maximal strength, and legLM Body mass remained unchanged in E+S (pre: 62.4
5.2 kg, post: 63.1
5.6 kg), but was slightly reduced in E (pre: 65.6
8.4 kg, post: 64.8
8.0 kg P < 0.05). The change in body mass was different between groups (P < 0.05). E + S increased 1RM in half squat with 45
22% (P < 0.01), while no change occurred in E (3
10%, P = 0.52, Fig. 1). The change in 1RM was larger in E + S than in E (P < 0.01) and the ES analysis revealed a very large practical effect of E + S compared to E (ES = 2.4). LegLM increased in E + S with 3.1
4.0% (P < 0.05), while it decreased in E with 2.2
2.1% (P < 0.05, Fig. 1). The change in legLM was larger in E + S than in E (P < 0.01) with a large practical effect of E + S com- pared to E (ES = 1.69). Because of the reduced body mass in E, all VO2 mea- surements are presented as body mass adjusted values. Since power output measured using cycling ergometers does not correctly reflect the influence of body mass on outdoor cycling performance, especially during uphill cycling (Anton et al. 2007), power outputs measurements are reported as body mass adjusted values (Wkg1). However, running at a treadmill is influenced by body Table 1. Details of primers used for RT-qPCR. Gene Forward primer Reverse Primer LDHA1 ATTCAGCCCGATTCCGTTAC TTCCACTCCATACAGGCACAC LDHB1 CATGGATGGATTTTGGGGGAAC AACACCTGCCACATTCACAC MCT11 TTGGAGTCATTGGAGGTCTTGG CCAATGGTCGCCTCTTGTAG MCT41 AGGCAAACTCCTGGATGCG AAAATCAGGGAGGAGGTGAGC PFKM1 TGACCTCCAGAAAGCAGGTAAG AACCAGGCCCACAATGTTC GAPDH1 AAGGCTGGGGCTCATTTG ACGAACATGGGGGCATC CPT22 AGCAGATGATGGTTGAGTGC TCAAAGCCCTGGCCCATTG SLC252 GCATTGCAGGGATCTTCAACTG ATATTTCCCAGGAGGTGCAGTC LDHA, lactate dehydrogenase A; LDHB, lactate dehydrogenase B; MCT1, monocarboxylate transporter 1; MCT4, monocarboxylate transporter 4; PFKM, phosphofructokinase; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; CPT2, carnitine palmitoyltransferase 2; SLC 25, carnitine/ acylcarnitine translocase, member 20. 1Genes involved in anaerobic energy metabolism. 2Genes involved in fatty acid oxidation. Figure 1. Individual values (dotted lines) and mean values (solid lines) before (Pre) and after (Post) the intervention period for athletes adding strength training to their normal endurance training (E+S, n = 11) and athletes performing normal endurance training only (E, n = 8). A: Lean mass in the legs. B: one repetition maximum (RM) in squat. * Different than pre (P ˂ 0.05), # the percent change from pre is different in E + S than in E (P ˂ 0.05). 2017 | Vol. 5 | Iss. 5 | e13149 Page 6 ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. Strength Training and Endurance Performance O. Vikmoen et al. mass to the same degree as outdoor running (McMiken and Daniels 1976); so, no body mass adjustments are done on the reported running distances. Muscle fiber type composition The effect of the present intervention on fiber type com- position has been previously reported (Vikmoen et al. 2016a). In brief, there was a reduction in the proportions of fibers positive for both IIA and IIX MyHC from 9
7% to 0% in E+S (P < 0.01) with a concomitant increase in type IIA fibers proportions from 39
13% to 51
10% (P < 0.01). Gene expression Of the nine genes investigated, only mRNA levels for CPT2 and LDHB increased 1.8
0.5 -fold and 1.2
0.3–fold, respectively, in E + S (P < 0.05). The remainder of the genes did not change expression in response to the intervention (Fig. 2). VO2max and Wmax/Vmax The effect of the intervention used in this study on VO2max and Wmax/Vmax has been previously described (Vikmoen et al. 2016a,b). In brief, VO2max in both cycling and running and Wmax/Vmax were unchanged in both groups during the intervention period. Responses during the prolonged trials The physiological responses during the prolonged trials are displayed in Table 2 and their percent changes are dis- played in Figure 3. After the intervention, E + S reduced VO2 during the last two hours of the prolonged cycling trial (P ˂ 0.05) with no changes in E. The changes during the last two hours were different between the groups (P ˂ 0.05). In addition, the effect size analysis revealed a large practical effect of E + S compared to E during the last hour of the trial (ES = 1.2). There were no changes in VO2 for neither E + S nor E during the prolonged running. E + S had a reduced HR throughout the prolonged cycling after the intervention (P ˂ 0.05), while E had a reduced HR during the first hour only (P ˂ 0.05). There was a moderate practical effect of E + S compared to E during the last hour of the trial (ES = 1.12). The correla- tion between changes in VO2 and HR during the last hour of the prolonged cycling was large (r = 0.59). Both E+S and E had a reduced HR during the entire prolonged running trial after the intervention period (P ˂ 0.05). There was no difference in changes between the groups. Compared to the pretrial, RPE was lower during the last hour of prolonged cycling for E + S and lower during the last two hours for E (P ˂ 0.05). However, there were no differences in changes between the groups. RPE did not change during the prolonged running. There were no changes in RER in neither of the groups during the pro- longed trial in both cycling and running. In cycling, cadence did not change in either group during the intervention. 5-min all-out tests After the intervention, the mean power output during the 5-min all-out cycling test increased by 7.0
4.5% (P < 0.05) in E+S with no change in E (3.3
7.1%, P = 0.27 Fig. 4). The difference between the groups was not statistically significant, but the practical effect of E + S compared to E was moderate (ES = 0.62). E + S Figure 2. Log2-fold change in mRNA expression for genes involved in fat transport and anaerobic metabolism during the intervention period for athletes adding strength training to their normal endurance training (E + S, n = 11) and athletes performing normal endurance training only (E, n = 8). * Different than pre (P ˂ 0.05). Values are mean
95% CI. ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. 2017 | Vol. 5 | Iss. 5 | e13149 Page 7 O. Vikmoen et al. Strength Training and Endurance Performance increased running distance in the 5-min all-out running test by 4.7
6.0% (P < 0.05) with no change in E (0.6
5.0%, Fig. 4). The increase in running distance was larger in E + S than in E (P = 0.05), and the practi- cal effect of E + S compared to E was moderate (ES = 0.95). Correlation analyses revealed a large correla- tion between change in all-out cycling performance and Wmax (r = 0.54, P ˂ 0.05) and between all-out running performance and Vmax (r = 0.53, P ˂ 0.05). There was a large correlation between change in all-out performance and the training induced change in IIAX-IIX fibers in cycling (r = 0.54, P ˂ 0.05, Fig. 5) and in running (r = 0.50, P ˂ 0.05, Fig. 5) when data from both groups were included. When only E + S athletes were included, the correlation got very large in cycling (r = 0.73, P = 0.065, Fig. 5) but disappeared in running (r = 0. 28, P = 0.547, Fig. 5). The correlation between the percent change in running distance and mean power output in cycling was moderate but not statistically significant (r = 0.40, P = 0.10). Discussion The main finding of this study is that addition of heavy strength training to the regular endurance training of female duathletes improved both running and cycling performance measured as 5-min all-out performance tested immediately after prolonged submaximal work. In addition, VO2 and HR were reduced during the last two hours of a 3-h prolonged cycling trial after the addition of heavy strength training, whereas no effects of added strength training were observed on physiological responses during prolonged submaximal running. Strength, legLM, and muscle fiber type composition The observed increase in 1RM in half squat and legLM is in accordance to previously observed improvements in endurance athletes adding 8–12 weeks of heavy strength training (e.g., Bishop et al. 1999; Storen et al. 2008; Ronnestad et al. 2010a; Aagaard et al. 2011; Ronnestad et al. 2015). The results lend further support to the notion that a substantial increase in strength can be achieved with little or no change in body mass (Storen et al. 2008; Ronnestad et al. 2010a; Sunde et al. 2010; Ronnestad et al. 2015). Increased body mass is usually undesirable for performance in cycling and running and therefore a concern among endurance athletes considering adding strength training. The increase in legLM reported Table 2. Responses during the prolonged trials in cycling and running for athletes adding strength training to their normal endurance training (E + S, n = 10) and athletes performing normal endurance training only (E, n = 8). E+S E Test section First section Middle section Last section First section Middle section Last section VO2 (ml∙kg1∙min1) Cycling Pre 30.5
2.9 31.3
3.0 31.9
2.9 30.1
3.2 30.5
3.4 31.0
3.1 Post 30.0
2.5 30.2
2.9*,# 30.9
3.2*,# 29.9
2.4 30.8
2.9 31.5
3.0 Running Pre 37.3
1.8 37.7
1.8 37.7
1.8 37.0
2.1 37.3
2.0 37.3
1.8 Post 37.0
2.2 37.5
2.0 37.6
1.9 37.4
2.0 37.4
1.5 37.4
1.4 HR (beats∙min1) Cycling Pre 134
12 138
14 143
14 129
11 130
9 135
7 Post 131
12* 131
14* 137
13* 125
9* 128
10 135
9 Running Pre 158
12 163
13 165
13 152
11 157
11 158
11 Post 154
11* 158
10* 159
11* 148
13* 151
11* 153
11* RER Cycling Pre 0.85
0.03 0.84
0.03 0.82
0.03 0.87
0.03 0.84
0.03 0.81
0.04 Post 0.87
0.04 0.85
0.03 0.82
0.03 0.88
0.03 0.85
0.03 0.82
0.03 Running Pre 0.90
0.02 0.89
0.02 0.88
0.02 0.90
0.02 0.87
0.03 0.86
0.03 Post 0.91
0.03 0.88
0.03 0.86
0.03 0.90
0.02 0.88
0.03 0.86
0.03 RPE (Borg scale) Cycling Pre 11
1 12
1 13
1 11
2 12
2 13
2 Post 11
1 12
1 12
1* 10
2 11
1* 12
1* Running Pre 12
1 13
1 13
1 11
2 12
1 13
1 Post 11
1 12
1 13
1 11
1 12
1 13
1 Cadence (rev∙min1) Cycling Pre 84
8 83
10 83
10 83
10 81
12 80
13 Post 85
9 83
8 83
9 81
11 81
12 80
14 Running Pre – – – – – – Post – – – – – – Values are mean
SD. *Different than pre (P ˂ 0.05) #The change from pre to post is different in E+S than in E (P ˂ 0.05). 2017 | Vol. 5 | Iss. 5 | e13149 Page 8 ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. Strength Training and Endurance Performance O. Vikmoen et al. in this study indicates that at least some of the improved strength was due to muscle hypertrophy. In addition, we observed a fiber type shift from type IIAX-IIX toward type IIA fibers (Vikmoen et al. 2016a), a common adap- tation to strength training among both untrained and endurance trained individuals (Staron et al. 1994; Aagaard et al. 2011). The increased legLM and fiber type shift shows that the strength training program was effective in inducing adaptations at the muscular level. Physiological responses during the prolonged trials As previously observed in well-trained male cyclists (Ronnestad et al. 2011), E+S reduced VO2 during the last two hours of the prolonged cycling after the strength training intervention. Therefore, although no change in cycling economy was observed during the first hour, cycling economy was clearly improved when the athletes started to get fatigued. This is highly important in cycling where many races include prolonged submaximal intensi- ties for several hours. Improved cycling economy have also been reported in untrained individuals (Loveless et al. 2005) and trained male cyclists (Sunde et al. 2010) after strength training interventions when measured in a nonfatigued state. However, this seems not to be the case in highly trained to elite cyclists (Ronnestad et al. 2010a; Aagaard et al. 2011; Ronnestad et al. 2015). The results from this study and the study by Ronnestad et al. (2011) indicate that after a strength training intervention, cycling Figure 3. Percent change in responses during the prolonged trials in cycling (left panels) and running (right panels) for athletes adding strength training to their normal endurance training (E + S, n = 10) and athletes performing normal endurance training only (E, n = 8). Values are mean
SD. * Different than pre (P ˂ 0.05), # the percent change from pre is different in E + S than in E (P ˂ 0.05). ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. 2017 | Vol. 5 | Iss. 5 | e13149 Page 9 O. Vikmoen et al. Strength Training and Endurance Performance economy should also be tested when the athletes are somewhat fatigued. HR was reduced throughout the prolonged cycling trial after the intervention period in E + S and as for VO2 the effect was more pronounced during the last 2 h. Conse- quently, the reduced HR was probably because of the reduced VO2 and hence reduced energy cost. In fact, the reduced HR mirrored the changes in VO2 and a large correlation between change in VO2 and change in HR during the last hour was observed (r = 0.59). The mechanisms behind improved cycling economy during the last 2 h of the trial are somewhat unclear. One explanation might be delayed recruitment of type II mus- cle fibers brought on by increased muscle strength and muscle mass (Ronnestad et al. 2011). When the maximal muscle strength increases and the absolute power output and cadence remains the same, the level of force devel- oped in each pedal thrust is reduced relatively to the maximal force. Given the size principle of motor unit recruitment, this implies that the more economical type I muscle fibers can account for a larger proportion of the same absolute power output (Hickson et al. 1988; Ronnestad and Mujika 2014). This may also explain the lack of changes in cycling economy during the first hour where the relative low power output should mainly recruit type I muscle fibers, thereby leaving little potential for improvements. In fact, it has previously been reported that after exercise for 60 min at an intensity requiring 43% of VO2max, glycogen breakdown mainly occurred in the type I muscle fibers (Vollestad and Blom 1985), indi- cating limited recruitment of type II fibers. However, as the duration of the work increases and muscle fibers starts to get fatigued, additional motor units needs to be recruited to sustain the power output (Gollnick and Arm- strong 1973; Vollestad and Blom 1985). The suggested mechanisms is therefore that the strength training allowed E + S to use the more economical type I muscle fibers for a longer duration of the trial after the intervention, leading to improved cycling economy during the last part. Supporting this, 5 weeks of strength training has been shown to reduce EMG activity in m. vastus lateralis dur- ing the last hour of a 2-hour prolonged cycling trial in well-trained triathletes (Hausswirth et al. 2010). The fiber type transition from type IIAX-IIX to type IIA in E+S might also contribute to the improved cycling economy since it has been suggested that type IIA fibers are more economical than the type IIX fibers (Westerblad et al. 2010). However, there was no correlation between the changes in the proportions of type IIAX-IIX and changes in economy during the last hour of the pro- longed cycling. This may be because the relatively low power output did not recruit any type IIX fibers during the trial even before the intervention. Other possible explanations for improved cycling econ- omy during the last 2 hours of the prolonged cycling trial could have been changes in substrate utilization toward larger carbohydrate utilization (Mogensen et al. 2006) or reduction in cadence (Foss and Hallen 2004). However, there were no changes in RER or cadence during the pro- longed cycling, making these explanations unlikely. In fact, based on the increased mRNA levels of CPT2, a pro- tein involved in fatty acid oxidation in the mitochondria, an increased utilization of fat as an energy substrate might have been expected. However, in Vikmoen et al. (2016a), we did not find changes in the content of the beta-oxidation enzyme hydroxyacyl-CoA dehydrogenase (HADH) in the very same biopsy material, supporting the notion that rates of fatty acid oxidation did not change. In contrast to cycling, no changes occurred in VO2 during the prolonged running. This is surprising since the proposed mechanisms for the reduced VO2 during cycling in theory also could reduce VO2 during the prolonged running. However, some methodological differences Figure 4. Individual values (dotted lines) and mean values (solid lines) before (Pre) and after (Post) the intervention period for athletes adding strength training to their normal endurance training (E + S, n = 10) and athletes performing normal endurance training only (E, n = 8). A: Running distance during the 5-min all-out running test. B: Mean power output during the 5-min all-out cycling test. * Different than pre (P ˂ 0.05), # the percent change from pre is different in E + S than in E (P = 0.05). 2017 | Vol. 5 | Iss. 5 | e13149 Page 10 ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. Strength Training and Endurance Performance O. Vikmoen et al. might explain the different finding between cycling and running. The prolonged running was only half as long as the prolonged cycling and was performed at a higher rela- tive workload (60% vs. 44% of Vmax and Wmax, respec- tively). Because the reduced VO2 during the cycling trial was seen during the last 2 h, it may be speculated that the prolonged running were too short. However, running races do seldom last as long as cycling races, and the shorter duration was therefore chosen for the prolonged running. To compensate for the shorter duration, the prolonged running was performed at a higher relative intensity than the prolonged cycling. This may have led to a quite high recruitment of type II motor units from the start, and the potential for reduced VO2 during the last part of the trial may therefore have been limited. In fact, in a glycogen breakdown study, it was estimated that a large proportion of type IIA fibers were recruited already from the start at a power output corresponding to 61% of VO2max (Vollestad and Blom 1985). No changes in running economy after addition of strength training is in conflict with results from previous studies where improved running economy ranging from 3 to 8% have been reported (e.g., Paavolainen et al. 1999; Storen et al. 2008; Sedano et al. 2013). Some method- ological differences might explain this discrepancy. Run- ning economy was tested with an inclination of 5.3% in our study, and in combination with the relative low workload, the velocity during the prolonged running was quite low compared to previous studies. In fact, the improvements in running economy after strength training have been reported to be dependent on running velocity (Saunders et al. 2006). The lack of effect on running economy may also be because the strength training pro- gram used did not induce any changes in patellar tendon stiffness (Vikmoen et al. 2016b). Changes in muscle-ten- don stiffness is a frequently proposed mechanism behind improved running economy after strength training (Saun- ders et al. 2006; Storen et al. 2008). Performance during the 5-min all-out tests The improved cycling performance observed in the 5-min all-out test is in accordance with a similar study in male cyclists, who found increased 5-min all-out performance following prolonged cycling after adding strength training to their normal endurance training (Ronnestad et al. 2011). A novel finding in this study is that 5-min all-out running performance after a prolonged submaximal trial also seems to be affected to the same degree as in cycling. Improved running and cycling performance after strength training is in accordance with previous studies in cycling (Koninckx et al. 2010; Ronnestad et al. 2010a,b; Sunde et al. 2010, Aagaard et al. 2011; Ronnestad et al. 2015; Vikmoen et al. 2016a) and running (Paavolainen et al. 1999; Storen et al. 2008; Sedano et al. 2013; Damasceno et al. 2015) when performance is measured in a more tra- ditional way. However, other studies contradict these findings both in cycling (Bishop et al. 1999; Bastiaans Figure 5. A: Correlation between changes in type IIAX-IIX proportions and changes in mean power output during the 5-min all-out cycling test. The inserted panel shows the correlation when only the athletes adding strength training to their normal endurance training are included. B: Correlation between changes in type IIAX-IIX proportions and changes in running distance during the 5-min all-out running test. The inserted panel shows the correlation when only the athletes adding strength training to their normal endurance training are included. ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. 2017 | Vol. 5 | Iss. 5 | e13149 Page 11 O. Vikmoen et al. Strength Training and Endurance Performance et al. 2001; Levin et al. 2009) and running (Ferrauti et al. 2010; Roschel et al. 2015). Some methodological differ- ences may explain these equivocal findings. To positively affect cycling performance, it seems that the strength training regime needs to involve heavy training load (4-10RM), rather large volumes of training and last for 8 weeks or longer (Koninckx et al. 2010; Ronnestad et al. 2010a, Aagaard et al. 2011; Ronnestad et al. 2015). On the other hand, both explosive, plyometric and heavy strength training seems effective in improving running performance (Paavolainen et al. 1999; Storen et al. 2008; Sedano et al. 2013; Damasceno et al. 2015). Together, these observations indicate that the mecha- nisms behind changes in running and cycling performance after strength training may be somewhat different. How- ever, improvements in both cycling and running perfor- mance may be related to typical adaptations to prolonged periods of heavy strength training such as increased mus- cle mass and fiber type transitions from type IIX to type IIA; improvements in running performance may also rely on adaptations such as changes in leg stiffness, rate of force development, and other neuromuscular characteris- tics. Therefore, mechanisms behind the improved perfor- mance in cycling and running in this study might be different. This is supported by the fact that the correlation between changes in running and cycling performance (r = 0.40) were not statistically significant. Since the performance tests were performed right after the prolonged trials, changes in the physiological responses to the submaximal exercise was expected to affect perfor- mance. We suggest that the reduced VO2 and HR observed during the last 2 hours of the cycling trial, indicating reduced physiological strain and less fatigue, made the ath- letes in E + S capable of producing higher mean power output during the final 5-min all-out test. Furthermore, reduced VO2 in E + S means that the total energy con- sumption during the prolonged cycling trial was lower after the intervention and with no change in substrate utilization the total carbohydrate utilization was reduced. Therefore, some of the improved cycling performance in E + S may be due to a better conservation of glycogen stores during the prolonged trial. The importance of less physiological strain during the submaximal exercise is indirectly sup- ported by the fact that 5-min all-out performance, tested in the rested state, was unchanged after 16 weeks of strength training in elite cyclists (Aagaard et al. 2011). Based on the present data, the positive effect of strength training on performance in the 5-min all-out running test cannot be explained by changes in physiolog- ical responses during the submaximal running. Therefore, the improved running performance after strength training has to be through other mechanisms. During a 5-min all-out test a substantial part of the energy is derived from anaerobic metabolism (Gastin 2001). Therefore, increased anaerobic capacity might be a mechanism behind the improved performance in both cycling and running. In fact, endurance performance has been reported to correlate well with measurements of anaerobic performance (e.g., Bulbulian et al. 1986; Houmard et al. 1991). Increased anaerobic capacity can be achieved through increases in muscle mass (Bangsbo et al. 1993) and/or through increasing amount of anaerobic enzymes. Even though small to none changes were found in mRNA expression of genes coding for important proteins in anaerobic metabolic pathways in E + S, the increased muscle mass should mean that anaerobic capacity was improved. Anaerobic capacity should also affect perfor- mance in Vmax/Wmax. Even though there were no signifi- cant changes in these variables, the correlation between changes in Vmax and running performance and Wmax and cycling performance further support that improved anaer- obic capacity might play a role for the improved perfor- mance in E + S. In addition, there was a very large correlation between legLM and absolute average power output during the 5-min all-out test before the interven- tion (r = 0.71, data not shown) indicating that muscle mass is important in these kinds of tests. There were large correlations between the reduction in muscle fiber type IIAX-IIX proportions and changes in 5-min all-out performance in both cycling and running. The type IIA fibers is less fatigable than the type IIX fibers (Westerblad et al. 2010), and a fiber type transition could therefore improve performance. However, a correlation between two variables does not necessarily mean a cause and effect relationship (Greenfield et al. 1998). Perhaps, the athletes with a large reduction in fiber type IIAX-IIX proportions had a large response to the strength training and that other adaptations to the strength training actu- ally were responsible for the improved performance. Indeed, there was a large negative correlation (r = 0.65, data not shown) between the change in legLM and change in the proportion of type IIAX-IIX fibers. Notably, when only E+S was included, the correlation between 5-min all-out performance and IIAX-IIX fiber transitions got very large in cycling and disappeared in running. This indicates that the possible performance-enhancing effects from fiber type shift from type IIAX-IIX toward type IIA was more important in cycling than in running. The improved performance cannot be explained by changes in VO2max since VO2max did not change in neither cycling nor running. The lack of effect of strength training on VO2max is not surprising and is in accordance with the current literature (e.g., Storen et al. 2008; Aagaard et al. 2011; Ronnestad et al. 2015). Importantly, we expected no change in VO2max in the study, as athletes were instructed to continue their normal endurance training, having a 2017 | Vol. 5 | Iss. 5 | e13149 Page 12 ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. Strength Training and Endurance Performance O. Vikmoen et al. good base of training from their winter training consisting of running, cycling, and cross-country skiing. This is the first controlled study to demonstrate that adding heavy strength training to endurance training leads to improvements in both cycling and running per- formance in the same athletes. Performance was tested as 5-min all-out performance, measured immediately after prolonged periods of submaximal work. The improved cycling performance was probably related to reduced physiological strain during the submaximal trial. This is also the first study reporting improved running perfor- mance following a prolonged submaximal effort. How- ever, there were no changes in the physiological responses to prolonged running. Therefore, improved running per- formance was more likely related to other mechanisms like changes in anaerobic capacity and neuromuscular changes. Changes in anaerobic capacity probably also contributed to improved cycling performance. A fiber type shift from type IIAX-IIX toward type IIA in the main propulsive muscles also seemed to contribute to the improved performance, especially in cycling. Based on the results of this study, both runners and cyclists should include heavy strength training in their training programs for maximal gains in performance. This seems to be par- ticularly important for performance during late phases of long-lasting competitions. Acknowledgments The authors thank the participants for their time and effort; students Kristoffer Bergstrøm, Øyvind Trøen, Roger Kristoffersen, Allan Sørgaard Nielsen, and Sondre Prestkvern for assistance during the intervention follow- up and data sampling. A special thanks to the Hospital for Rheumatic Diseases at Lillehammer for performing the DXA scans. Olav Vikmoen also thanks his current employer, the Norwegian Defence Research Establishment (FFI) for support during the writing process. Conflict of Interests None declared. References Aagaard, P., J. L. Andersen, M. Bennekou, B. Larsson, J. L. Olesen, R. Crameri, et al. 2011. Effects of resistance training on endurance capacity and muscle fiber composition in young top-level cyclists. Scand. J. Med. Sci. Sports 21:e298–e307. Anton, M. M., M. Izquierdo, J. Ibanez, X. Asiain, J. Mendiguchia, and E. M. Gorostiaga. 2007. Flat and uphill climb time trial performance prediction in elite amateur cyclists. Int. J. Sports Med. 28:306–313. Bangsbo, J., L. Michalsik, and A. Petersen. 1993. Accumulated O2 deficit during intense exercise and muscle characteristics of elite athletes. Int. J. Sports Med. 14:207–213. Bastiaans, J. J., A. B. van Diemen, T. Veneberg, and A. E. Jeukendrup. 2001. The effects of replacing a portion of endurance training by explosive strength training on performance in trained cyclists. Eur. J. Appl. Physiol. 86:79–84. Bishop, D., D. G. Jenkins, L. T. Mackinnon, M. McEniery, and M. F. Carey. 1999. The effects of strength training on endurance performance and muscle characteristics. Med. Sci. Sports Exerc. 31:886–891. Bulbulian, R., A. R. Wilcox, and B. L. Darabos. 1986. Anaerobic contribution to distance running performance of trained cross-country athletes. Med. Sci. Sports Exerc. 18:107–113. Damasceno, M. V., A. E. Lima-Silva, L. A. Pasqua, V. Tricoli, M. Duarte, D. J. Bishop, et al. 2015. Effects of resistance training on neuromuscular characteristics and pacing during 10-km running time trial. Eur. J. Appl. Physiol. 115:1513– 1522. Ellefsen, S., O. Vikmoen, G. Slettalokken, J. E. Whist, H. Nygaard, I. Hollan, et al. 2014. Irisin and FNDC5: effects of 12-week strength training, and relations to muscle phenotype and body mass composition in untrained women. Eur. J. Appl. Physiol. 114:1875–1888. Ferrauti, A., M. Bergermann, and J. Fernandez-Fernandez. 2010. Effects of a concurrent strength and endurance training on running performance and running economy in recreational marathon runners. J. Strength Cond. Res. 24:2770–2778. Foss, O., and J. Hallen. 2004. The most economical cadence increases with increasing workload. Eur. J. Appl. Physiol. 92:443–451. Gastin, P. B. 2001. Energy system interaction and relative contribution during maximal exercise. Sports Med. 31:725– 741. Gollnick, P. D., R. B. Armstrong, C. W. T. Saubert, W. L. Sembrowich, R. E. Shepherd, and B Saltin. 1973. Glycogen depletion patterns in human skeletal muscle fibers during prolonged work. Pflugers Arch. 344:1–12. Greenfield, M. L., J. E. Kuhn, and E. M. Wojtys. 1998. A statistics primer. Correlation and regression analysis. Am. J. Sports Med. 26:338–343. Hausswirth, C., S. Argentin, F. Bieuzen, Y. Le Meur, A. Couturier, and J. Brisswalter. 2010. Endurance and strength training effects on physiological and muscular parameters during prolonged cycling. J. Electromyogr. Kinesiol. 20:330– 339. Hickson, R. C., B. A. Dvorak, E. M. Gorostiaga, T. T. Kurowski, and C. Foster. 1988. Potential for strength and endurance training to amplify endurance performance. J. Appl. Physiol. 65:2285–2290. Hopkins, W. G., S. W. Marshall, A. M. Batterham, and J. Hanin. 2009. Progressive statistics for studies in sports ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. 2017 | Vol. 5 | Iss. 5 | e13149 Page 13 O. Vikmoen et al. Strength Training and Endurance Performance medicine and exercise science. Med. Sci. Sports Exerc. 41:3–13. Houmard, J. A., D. L. Costill, J. B. Mitchell, S. H. Park, and T. C. Chenier. 1991. The role of anaerobic ability in middle distance running performance. Eur. J. Appl. Physiol. Occup. Physiol. 62:40–43. Jeukendrup, A. E., N. P. Craig, and J. A. Hawley. 2000. The bioenergetics of World Class Cycling. J Sci Med Sport 3:414–433. Koninckx, E., M. Van Leemputte, and P. Hespel. 2010. Effect of isokinetic cycling versus weight training on maximal power output and endurance performance in cycling. Eur. J. Appl. Physiol. 109:699–708. Levin, G. T., M. R. McGuigan, and P. B. Laursen. 2009. Effect of concurrent resistance and endurance training on physiologic and performance parameters of well-trained endurance cyclists. J. Strength Cond. Res. 23:2280–2286. Loveless, D. J., C. L. Weber, L. J. Haseler, and D. A. Schneider. 2005. Maximal leg-strength training improves cycling economy in previously untrained men. Med. Sci. Sports Exerc. 37:1231–1236. McMiken, D. F., and J. T. Daniels. 1976. Aerobic requirements and maximum aerobic power in treadmill and track running. Med Sci Sports 8:14–17. Mogensen, M., M. Bagger, P. K. Pedersen, M. Fernstrom, and K. Sahlin. 2006. Cycling efficiency in humans is related to low UCP3 content and to type I fibres but not to mitochondrial efficiency. J. Physiol. 571:669–681. Paavolainen, L., K. Hakkinen, I. Hamalainen, A. Nummela, and H. Rusko. 1999. Explosive-strength training improves 5-km running time by improving running economy and muscle power. J. Appl. Physiol. 86:1527–1533. Ronnestad, B. R., and I. Mujika. 2014. Optimizing strength training for running and cycling endurance performance: a review. Scand. J. Med. Sci. Sports 24:603–612. Ronnestad, B. R., E. A. Hansen, and T. Raastad. 2010a. Effect of heavy strength training on thigh muscle cross-sectional area, performance determinants, and performance in well- trained cyclists. Eur. J. Appl. Physiol. 108:965–975. Ronnestad, B. R., E. A. Hansen, and T. Raastad. 2010b. In-season strength maintenance training increases well- trained cyclists’ performance. Eur. J. Appl. Physiol. 110:1269–1282. Ronnestad, B. R., E. A. Hansen, and T. Raastad. 2011. Strength training improves 5-min all-out performance following 185 min of cycling. Scand. J. Med. Sci. Sports 21:250–259. Ronnestad, B. R., J. Hansen, I. Hollan, and S. Ellefsen. 2015. Strength training improves performance and pedaling characteristics in elite cyclists. Scand. J. Med. Sci. Sports 25: e89–e98. Roschel, H., R. Barroso, V. Tricoli, M. A. Batista, F. M. Acquesta, J. C. Serrao, et al. 2015. Effects of strength training associated with whole-body vibration training on running economy and vertical stiffness. J. Strength Cond. Res. 29:2215–2220. Saunders, P. U., R. D. Telford, D. B. Pyne, E. M. Peltola, R. B. Cunningham, C. J. Gore, et al. 2006. Short-term plyometric training improves running economy in highly trained middle and long distance runners. J. Strength Cond. Res. 20:947–954. Sedano, S., P. J. Marin, G. Cuadrado, and J. C. Redondo. 2013. Concurrent training in elite male runners: the influence of strength versus muscular endurance training on performance outcomes. J. Strength Cond. Res. 27:2433– 2443. Staron, R. S., D. L. Karapondo, W. J. Kraemer, A. C. Fry, S. E. Gordon, J. E. Falkel, et al. 1994. Skeletal muscle adaptations during early phase of heavy-resistance training in men and women. J. Appl. Physiol. 76:1247–1255. Storen, O., J. Helgerud, E. M. Stoa, and J. Hoff. 2008. Maximal strength training improves running economy in distance runners. Med. Sci. Sports Exerc. 40:1087–1092. Sunde, A., O. Storen, M. Bjerkaas, M. H. Larsen, J. Hoff, and J. Helgerud. 2010. Maximal strength training improves cycling economy in competitive cyclists. J. Strength Cond. Res. 24:2157–2165. Vikmoen, O., S. Ellefsen, O. Troen, I. Hollan, M. Hanestadhaugen, T. Raastad, et al. 2016a. Strength training improves cycling performance, fractional utilization of VO2max and cycling economy in female cyclists. Scand. J. Med. Sci. Sports 26:384–396. Vikmoen, O., T. Raastad, O. Seynnes, K. Bergstrom, S. Ellefsen, and B. R. Ronnestad. 2016b. Effects of heavy strength training on running performance and determinants of running performance in female endurance athletes. PLoS ONE 11:e0150799. Vollestad, N. K., and P. C. Blom. 1985. Effect of varying exercise intensity on glycogen depletion in human muscle fibres. Acta Physiol. Scand. 125:395–405. Westerblad, H., J. D. Bruton, and A. Katz. 2010. Skeletal muscle: energy metabolism, fiber types, fatigue and adaptability. Exp. Cell Res. 316:3093–3099. 2017 | Vol. 5 | Iss. 5 | e13149 Page 14 ª 2017 The Authors. Physiological Reports published by Wiley Periodicals, Inc. on behalf of The Physiological Society and the American Physiological Society. Strength Training and Endurance Performance O. Vikmoen et al.

Heavy strength training improves running and cycling performance following prolonged submaximal work in well-trained female athletes.

[]

Vikmoen, Olav,Rønnestad, Bent R,Ellefsen, Stian,Raastad, Truls

eng

PMC9484631

RESEARCH ARTICLE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners Jason BonacciID1*, Wayne Spratford2,3,4, Claire Kenneally-Dabrowski1, Danielle TrowellID1, Adrian Lai5 1 Centre for Sports Research, School of Exercise and Nutrition Sciences, Deakin University, Geelong, Australia, 2 Movement Science, Australian Institute of Sport, Canberra, Australia, 3 Discipline of Sport and Exercise Science, Faculty of Health, University of Canberra, Canberra, Australia, 4 University of Canberra Research Institute for Sport and Exercise (UCRISE), University of Canberra, Australia, 5 Lululemon Athletica, Vancouver, Canada * jason.bonacci@deakin.edu.au Abstract Purpose To compare the ankle plantar-flexor muscle-tendon mechanical behaviour during barefoot and shod forefoot running. Methods Thirteen highly trained forefoot runners performed five overground steady-state running tri- als (4.5 ± 0.5 m.s-1) while barefoot and shod. Three-dimensional kinematic and ground reac- tion force data were collected and used as inputs for musculoskeletal modelling. Muscle- tendon behaviour of the ankle plantar-flexors (soleus; medial gastrocnemius; and lateral gastrocnemius) were estimated across the stance phase and compared between barefoot and shod running using a two-way multivariate analysis of variance. Results During barefoot running peak muscle-tendon unit (MTU) power generation was 16.5% (p = 0.01) higher compared to shod running. Total positive MTU work was 18.5% (p = 0.002) higher during barefoot running compared to shod running. The total sum of tendon elastic strain energy was 8% (p = 0.036) greater during barefoot compared to shod running, how- ever the relative contribution of tendon and muscle fibres to muscle-tendon unit positive work was not different between conditions. Conclusion Barefoot forefoot running demands greater muscle and tendon work than shod forefoot run- ning, but the relative contribution of tendon strain energy to overall muscle-tendon unit work was not greater. PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 1 / 13 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Bonacci J, Spratford W, Kenneally- Dabrowski C, Trowell D, Lai A (2022) The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners. PLoS ONE 17(9): e0274806. https://doi.org/10.1371/journal. pone.0274806 Editor: Nili Steinberg, The Wingate College of Physical Education and Sports Sciences at the Wingate Institute, IL, ISRAEL Received: November 30, 2021 Accepted: September 3, 2022 Published: September 19, 2022 Copyright: © 2022 Bonacci et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: The author(s) received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. Introduction The human ankle plantar-flexors, the soleus (SOL), medial gastrocnemius (MG) and lateral gastrocnemius (LG), perform an important biomechanical function during running. Experi- mental and musculoskeletal modelling studies have demonstrated that the ankle plantar-flex- ors generate force up to 12 times body weight (BW) during running [1]; the greatest force of all the lower-limb muscle groups. They are also the dominant contributors to support and hor- izontal propulsion of the body’s centre of mass during running [2, 3]. The ankle plantar-flexors have relatively short muscle fibres that insert onto the calcaneus via a compliant Achilles ten- don. This configuration favours the capacity to store elastic strain energy, which contributes a considerable amount of the mechanical work performed by the musculotendon units [4, 5]. Previous modelling studies have demonstrated that elastic strain energy stored in the Achilles tendon during steady-state running provides a greater contribution to muscle-tendon unit (MTU) propulsive work compared to the muscle fibres [6–8]. The contribution of tendon strain energy to overall MTU propulsive work increases as running speed advances from slow running towards maximum sprinting [8]. However, the effects of footwear and foot-strike type on the energetics of the muscle fibres and tendon in the ankle plantar-flexors during running remain largely unknown. Sinclair et al. [9] utilised musculoskeletal modelling to examine the effect of footwear on ankle plantar-flexor muscle forces during running. They reported a 32% increase in MG mus- cle force during barefoot running compared to shod running, but no difference in LG or SOL muscle forces between conditions. Foot-strike pattern was not controlled in the study. For example, the ankle was more plantarflexed at contact in the barefoot running condition sug- gesting that participants switched from a rearfoot strike during the shod condition to a mid/ forefoot strike during the barefoot condition. It is not possible to discern if the increase in force developed by the MG was due to differences in footwear or foot-strike type. Running barefoot and in minimalist shoes has been associated with greater peak internal ankle plantar- flexion moments and plantar-flexor impulse during stance [10, 11]. These greater demands on the ankle plantar-flexors are due to a lack of shoe cushioning, which increases the muscular effort required to attenuate ground impact forces while potentially increasing the metabolic cost of running [12, 13]. However, running barefoot and in a minimalist shoe are more eco- nomical than running in a cushioned shoe, even when shoe mass is accounted for [11, 14]. If barefoot running is more economical but has greater ankle plantar-flexor demands, this dis- crepancy may suggest that the ankle plantar-flexors utilise greater Achilles tendon elastic strain energy during barefoot compared to shod running. Perl et al. [11] postulated that the elevated heel during shod running and greater lower extremity elastic energy storage may explain the moderate metabolic benefits of barefoot or minimally shod running. The authors indirectly measured Achilles tendon strain using the overall length change of the entire triceps surae MTU complex. This method does not distin- guish between the length changes of the muscle fibre and tendon components, which previous in-vivo studies of the ankle plantar-flexors during running have shown are decoupled from that of the MTU [15, 16]. As a result, it is not possible to differentiate the work done by the MTU in the ankle plantar-flexors into the elastic strain energy stored in the tendon and the work done by the muscle fibres. Therefore, the aim of the study was to investigate the effect of footwear on the mechanical behaviour of the ankle plantar-flexor muscle fibre and tendon components during running. Specifically, we used experimental kinematic and kinetic data in conjunction with musculoskeletal modelling to compute the mechanical power and work per- formed by the MTU, muscle fibres and tendon in the SOL, MG and LG during barefoot and shod running. As the purpose of this study was to examine the effect of footwear rather than PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 2 / 13 foot strike on plantar-flexor MTU behaviour, only habitual forefoot runners were recruited. This is because rearfoot strikers often switch to a forefoot strike when running barefoot [17], which could confound the results. We hypothesised that in comparison to shod running, bare- foot running would result in: (i) a greater amount of tendon elastic strain energy in the ankle plantar-flexors and; (ii) a greater relative contribution of tendon elastic strain energy contribu- tion to MTU positive work compared to muscle fibre work. Methods Participants Thirteen highly trained distance runners (8 males and 5 females; mean ± SD; age, 29.9 ± 5.9 years; height, 176.7 ± 7.5 cm; body mass, 64.9 ± 8.8 kg) were recruited for the study. Based on a priori sample size calculation, 13 participants would be sufficient to generate an effect size of 0.93 at a power of 80% and α of < 0.05 [9]. All participants were training for competition (training history: 14.3 ± 1.9 km per session, 7.4 ± 1.9 sessions per week, 109.6 ± 27.9 km per week) with the average personal best 10 km time in the previous year of 33.7 ± 3.7 minutes. All participants self-reported as a forefoot striker and this was confirmed during data collection. No participants were suffering from any pre-existing musculoskeletal injury that might affect their ability to participant in the study. Written informed consent was obtained from all partic- ipants and ethical approval was attained from Deakin University and Australian Institute of Sport human research ethics committees. Experimental data collection Running trials were conducted on a 110 m indoor synthetic track in the Biomechanics Labora- tory at the Australian Institute of Sport, Canberra. Three-dimensional kinematic data were col- lected using a 22-camera motion analysis system (VICON, Oxford Metrics Ltd, Oxford, UK) sampling at 250 Hz. The calibrated capture volume was approximately 20 m in length and situ- ated ~60 m along the track, which allowed sufficient distance for participants to accelerate, hold a steady-state speed through the capture volume and then safely decelerate to rest. Retro- reflective markers (14 mm diameter) were placed at predefined locations on the pelvis and lower limbs [10]. Individual markers were placed on the left and right anterior superior iliac spines and posterior superior iliac spines. The thigh segment was defined by a three-marker cluster affixed laterally and aligned with the head of the femur and lateral femoral condyle. The lower leg was defined by a three-marker cluster aligned with the lateral femoral condyle and lateral malleolus. Markers placed on the superoposterior aspect of the calcaneus, and first and fifth metatarsals defined the foot. In the shod condition, these markers were placed on the shoe. Individual retroflective markers were also placed on the medial and lateral femoral con- dyles and medial and lateral malleoli to define the knee and ankle joint centres, respectively. Ground reaction force (GRF) data were collected using eight in-ground force plates (Kistler Instrument Corp., Dimensions: 900 x 600 mm, Amherst, New York, USA) sampling at 1500 Hz. The force plates were embedded into the synthetic running track directly adjacent to each other spanning a total length of 7.2 m. Marker trajectories and GRF data were filtered using fourth-order, low-pass Butterworth filters with the same cut-off frequency of 20 Hz [18]. The data collection protocol involved two experimental conditions: shod (lightweight rac- ing flat, NIKE LunaRacer 2) and barefoot. The racing flat had a low heel-forefoot offset (6 mm) and mean mass of 184.2 ± 19.4 g. All participants were required to complete a 10-day familiarisation period prior to testing to get accustomed to the barefoot and shod conditions. The average distances completed by all participants during the familiarisation period in the barefoot and shod conditions were 4.3 ± 3.2 km and 20.7 ± 11.5 km in 2.6 ± 0.6 and 3 ± 0.6 PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 3 / 13 sessions, respectively. The volume of barefoot familiarisation was lower than shod running to minimise likelihood of acute overload during this unfamiliar condition [19, 20]. All partici- pants habitually wore a standard cushioned running shoe for most of their training volume. Participants performed a standardised warm-up prior to data collection that involved five overground running trials within the capture volume. Participants then performed a static cali- bration trial and five overground steady-state running trials in each of the two randomly ordered conditions. The desired steady-state running speed for each participant was set at 90% of participant’s best 10 km time in the previous year. The mean desired steady-state running speed for all participants was 4.5 ± 0.5 m.s-1. Average steady-state speed for each trial were obtained using timing gates (Speedlight Telemetry Timing, Swift Performance Equipment, Walcol, QLD, Australia) placed at the start and end of the calibrated capture volume. Trials were accepted if the average speed was within ±5% of the desired speed. Steady state running was confirmed post testing via examination of the net horizontal force impulse (S1 Dataset). Forefoot strike was defined as a foot strike in which the point of first contact of the foot with the ground was the forefoot or the front half of the shoe sole. Two classification techniques were used to confirm that all participants had forefoot strike patterns. The first classification technique used the presence of an initial ankle plantar-flexion angle in kinematic data and absence of impact peak in the vertical GRF profile [17] while the second classification tech- nique used the markers placed on the heel and the toe to define foot strike patterns [21]. The difference in vertical position of the heel and first metatarsal markers was calculated during both the static trial and at initial contact during running. The vertical difference between markers during the static trial was then subtracted from the difference at initial contact. Partic- ipants were classified as forefoot strikers if the final value was 40 mm or less [21]. Both classifi- cation techniques verified that all participants had forefoot strike patterns in both the barefoot and shod conditions. All participants completed both experimental conditions. Musculoskeletal model The skeletal system was modelled as a 12 segment, 31 degree of freedom (DOF) mechanical linkage system, similar to that described by Hamner at al. [22]. The lower limb joints were modelled as follows: the pelvis was free to translate and rotate in space (6 DOF), the hip was a ball-and-socket joint (3 DOF), the knee was a hinge joint (1 DOF), and the ankle-subtalar complex was a universal joint (2 DOF) comprised of two non-intersection hinge joints. While the model contained the metatarsophalangeal joint, this was locked for all simulations and the mid and fore- foot acted as a rigid segment. The model was actuated by 96 MTU actuators. Each MTU was modelled as a Hill-type muscle consisting of in-parallel active and passive mus- cle fibre elements attached in-series with a series elastic element. Hereafter, the series elastic element will be termed tendon as a result of the significant influence of the free tendon on series elasticity compared with other elastic connective tissue (e.g. aponeurosis). Maximum shortening velocity was set to 15 optimal fibre lengths per second to be consistent with previ- ous modelling studies investigating running [23, 24]. The maximum isometric force of all mus- cles was increased three-fold, as required for successful simulations of highly dynamic movement such as gait [25, 26]. The SOL, MG and LG were assumed to be three separate MTUs with three independent tendon elements representing the Achilles tendon. The tendon strains at maximum isometric force generation were informed by previously reported data. Experimental data from three rearfoot runners [16] were used to evaluate the effect of modifying plantar-flexor tendon strains on model-based estimates of SOL and MG muscle fibre lengths. Simulated fibre lengths were compared to that measured in-vivo using ultrasound during rearfoot running [6, 16, 27] PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 4 / 13 across a range of tendon strains from 3.3%-10%. A tendon strain of 5% for the SOL, MG and LG gave the most comparable model-based muscle fibre length changes in the SOL and MG during stance phase. Tendon strains at maximum isometric contraction for the ankle plantar- flexors were consistent with reported in-vivo measurements of the Achilles tendon of 4.9 ± 1% [28] and 5.1 ± 1.1% [29]. Furthermore, simulated tendon strains during running were within the range of tendon strains reported using dynamic ultrasound measurements of the SOL and MG at equivalent running speeds [15, 16]. These two consistencies support our decision to increase the tendon strain for the ankle plantar-flexors at maximum isometric force genera- tion. Tendon compliance in other MTUs remained unchanged at 3.3% of maximum isometric force generation [30]. Computational simulations Muscle modelling simulations were performed using OpenSimTM [31]. Subject-specific mus- culoskeletal models were attained by scaling a generic model to the participant’s height and body mass. Individual musculoskeletal models were created for each participant and experi- mental condition (i.e. barefoot and shod). The same generic model (with identical marker placement) was used during scaling for each condition. A set of joint angles for each time instant was calculated using an inverse kinematic analysis where the sum of the squares of the differences between experimental markers trajectories and virtual markers in the model was minimised [32]. An inverse dynamics analysis in conjunction with a computed muscle control algorithm (CMC) were used to predict muscle forces and activations [33, 34]. A standard inverse dynamic approach was used to compute net joint torques generated about the torso, hip, knee and ankle joints. Residual reduction analysis (RRA) was used to reduce dynamic incon- sistencies between joint kinematics and the measured GRF. The errors in the residual forces and dynamically-consistent kinematics were within the recommended bounds of the analysis [35]. Muscle forces and activations were computed using CMC in accordance with the physi- ological force-length and force-velocity properties of muscle fibre and tendon, as well as the geometric and dynamic constraints of the system. CMC solved the muscle redundancy prob- lem by predicting a set of muscle excitations that drove a model forward in time (simulation time window of 0.01 s) such that the sum of squared muscle activations was minimised and the kinematics of the model tracked the dynamically consistent joint kinematics obtained from RRA. Muscle excitations were bounded between 0 (no muscle activation) and 1 (full muscle activation) with activation and deactivation time constants of 10 and 30 ms, respec- tively [30]. The mechanical power developed by the MTU, muscle fibre and tendon elements were cal- culated by multiplying MTU, muscle fibre and tendon force by their corresponding contrac- tion velocity at each time instant. Negative and positive power represented power absorption and generation, respectively. The positive work done by the MTU, muscle fibre and tendon was found by integrating the MTU, muscle fibre and tendon power curves over the duration of the stance phase where power was generated. All participants were forefoot strikers, thus the tendon and MTU lengthened and performed negative work during early stance followed by a period of positive work during mid- to late stance. The recovery of tendon elastic strain energy was represented by the positive work done by the tendon after the tendon performed negative work and after the MTU started generating positive power. In this study, we were specifically interested in the percentage contributions of positive muscle fibre work and tendon elastic strain energy to the positive work done by the MTU (i.e., propulsion energy). The calculation of these contributions are detailed in a previous paper [8]. PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 5 / 13 Briefly, the contributions were calculated using the following equation, %MTU contribution ¼ W WMTU  100 where WMTU is the positive work done by the MTU and W is the area under the WMTU curve attributable to positive muscle fibre work or tendon elastic strain energy. Data analysis Data for each participant were averaged over five stance phases for each experimental condi- tion; time normalised to 0–100% of the stride cycle and used to calculate group mean ± SD val- ues. Muscle force was normalised by the participant’s body weight while mechanical power and work were normalised by body mass (kg). Stride parameters of stance duration and stride length were calculated, along with the relative heel-toe marker position to classify footstrike position. The outcome variables included SOL, MG and LG peak muscle force (BW), peak MTU positive power (W.kg-1), total MTU positive work (J.kg-1) and the percentage contribu- tions of positive muscle fibre work and tendon elastic strain energy to the positive work done by the MTU. The total sum of tendon elastic strain energy stored in all three ankle plantar-flex- ors was also calculated. Data are presented as mean ± SD. The Shapiro-Wilk test was con- ducted to determine if any data violated the assumption of normality. Stance duration, stride length, footstrike position and total sum of tendon elastic strain energy were compared between shod and barefoot conditions using a two-tailed paired sample t-test with α set at < 0.05. A two-way multivariate analysis of variance (MANOVA) was used to identify the effect of footwear and muscle on ankle plantar-flexor peak muscle forces, peak MTU power generation, total positive MTU work, and contribution of tendon and muscle fibre to positive MTU work. Where a main effect was found, post-hoc tests with Bonferroni correction were used to test for differences between means. Standardised mean differences (SMD) were calcu- lated to express the magnitude of difference between conditions and interpreted according to the following criteria: calculated SMD of 0.2–0.49, small change; SMD of 0.5–0.79, moderate change; and SMD  0.8, large change [36]. All statistical analysis was conducted using the Sta- tistical Package for the Social Sciences v27 (IBM Statistics, Chicago, USA). Results Stance duration and footstrike position were not different between barefoot and shod running, though there was a small decrease in stride length during barefoot compared to shod running (Table 1). Peak muscle force, peak MTU power generation, total positive MTU work and rela- tive contribution of tendon and muscle fibre to MTU positive work are plotted in Fig 1. Plan- tar-flexor muscle fibre, tendon and MTU length curves across the stance phase are presented in S1 Appendix. Table 1. Group mean ± SD values and the difference between footwear conditions for stride parameters and footstrike position. Barefoot Shod Mean difference [95% CI] p-value SMD Stance duration (s) 0.2 ± 0.01 0.2 ± 0.01 0.00 [-0.00, 0.00] 0.391 0.098 Stride length (m) 3.0 ± 0.3 3.1 ± 0.3 -0.1 [-0.11, -0.07] 0.001 0.29† Footstrike position (mm)# 34.4 ± 3.6 33.2 ± 3.6 1.2 [-0.3, 2.8] 0.099 0.34 #A lower relative value indicates more ankle plantarflexion at initial contact (forefoot strike < 40 mm). Significant difference between barefoot and shod conditions. † Small change https://doi.org/10.1371/journal.pone.0274806.t001 PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 6 / 13 The total sum of tendon elastic strain energy was 8.3% (1.2 to 15.5%, SMD = 1.17, p = 0.036) higher during barefoot (0.9 [0.1] J.kg-1) compared to shod (0.8 [0.1] J.kg-1) running. The two-way MANOVA revealed a main effect for footwear (p = 0.032) but no footwear by muscle interaction effects (p = 0.669). Univariate analysis demonstrated a significant differ- ence in peak MTU power generation (p = 0.01) and total positive MTU work (p = 0.002) for footwear conditions. Compared to shod running, peak MTU power generation was 16.5% (4 to 28.9%, SMD = 0.7) higher when running barefoot. Total positive MTU work was 18.5% (7 to 30%, SMD = 0.92) higher during barefoot compared to shod running. There was no effect of footwear on peak muscle force (p = 0.21) or tendon and muscle fibre contribution to posi- tive MTU work (p = 0.46 & 0.37, respectively). Fig 1. (A) Peak MTU muscle force; (B) peak MTU power generation; (C) total positive MTU work; (D) relative tendon and muscle fibre contributions to positive work (%) during the stance phase of barefoot and shod running. https://doi.org/10.1371/journal.pone.0274806.g001 PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 7 / 13 Discussion The aim of this study was to investigate the mechanical behaviour of the plantar-flexor muscle fibre and tendon components during barefoot and shod forefoot running. Tendon elastic strain energy was significantly greater (8.3%, SMD = 1.2) during barefoot running compared to shod running, confirming our first hypothesis. Achilles tendon elastic strain energy contrib- uted the majority of MTU positive work during both shod and barefoot running. This is con- sistent with previous studies of shod running [6–8], although estimates of SOL relative tendon contribution were approximately 10% greater than previously reported at comparative speeds [7, 8]. This discrepancy may be due to methodological differences, as previous studies did not control footstrike pattern and modelled the LG and MG as a single MTU [7, 8]. The relative contribution of the tendon to plantar-flexor MTU positive work was not different between barefoot and shod conditions. This indicates that both tendon and muscle fibre work are increased when running barefoot. This finding refutes our second hypothesis. The sum of muscle fibre and tendon contributions to MTU positive work exceeded 100% for all plantar-flexors during both barefoot and shod running. However, this is expected and can be explained by the transfer of energy from the muscle fibre to the tendon. During early stance, muscle fibres of the plantar flexors shorten and do positive work, while the tendon and MTU lengthen and do negative work [16]. This behaviour is common in MTUs where the ten- don is compliant [30], such as the plantar-flexors. Positive work done by the muscle fibres dur- ing early stance is transmitted to the tendon as it stretches. This results in energy stored in the tendon, which is later returned as the tendon shortens during propulsion, in late stance. Because of this additional energy returned via the tendon, the sum of tendon and muscle fibre positive work exceeds 100% of MTU positive work. Peak muscle force was not significantly greater during barefoot compared to shod forefoot running. In contrast, Sinclair et al. [9] found large increases in MG (32%) peak forces during barefoot compared to shod running. This disparity may be due to differences in footstrike pat- terns between studies. In the current study, all participants used a forefoot strike during shod and barefoot running. In comparison, a switch from rear to mid/forefoot was noted between shod and barefoot running in the previous study. The observed change in footstrike pattern has previously been reported in habitual rearfoot shod runners when transitioning to barefoot running [17]. The large increases observed by Sinclair et al. [9] are likely due to a lack of cush- ioning combined with a change in footstrike pattern during barefoot running. An absence of shoe cushioning during barefoot running increases the ankle plantarflexion moment during stance due to an increased muscular effort to absorb impact forces [10, 11]. The increase in MG muscle forces and Achilles tendon strain energy during barefoot forefoot running must be carefully considered as a rapid transition out of footwear may overload this complex. Peak MTU power generation and MTU total positive work increased when running bare- foot compared to shod running. Previous studies have reported that greater positive work is required at the ankle joint as shoe cushioning decreases or is removed when running barefoot [10, 37]. This is likely caused by greater propulsive forces when running barefoot [38]. As the plantar-flexors are the primary contributors to ankle joint torque and positive work generation during propulsion, it is expected that greater work is required of the plantar-flexor MTUs when running barefoot. As previously noted, this increased MTU positive work was a result of increases in both tendon elastic strain energy and muscle fibre work. There is some evidence that barefoot running has economical benefits [11, 14]. However, the mechanism behind this is unclear. Perl et al. [11] postulated that greater elastic energy storage and release could explain greater economy when running barefoot. While the current study showed an increase in tendon elastic strain energy during barefoot running, the relative contribution of the tendon PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 8 / 13 to MTU positive work was not greater. Thus, barefoot running was not associated with greater tendon contribution to overall MTU positive work. The muscle fibre contributions to overall MTU positive work were similar in both conditions. The overall increase in MTU work and similar tendon and muscle fibre contributions to MTU positive work are unlikely to contribute to an economical advantage during barefoot running. We found increased Achilles tendon elastic strain energy while running barefoot compared to shod. While increased tendon elastic strain energy may be beneficial, we must also consider how tendon compliance affects the operating range of muscle fibres on the force-length curve. Increased tendon elastic-energy storage and recovery may result in muscle fibres operating almost isometrically, which is most beneficial if near the optimal operating length on the force-length curve [39]. However, a trade-off may occur if tendon compliance results in mus- cle fibres operating at lengths which are further from optimal. This was observed by Uchida et al. [40], whereby increased tendon compliance of the MG resulted in muscle fibres remain- ing shorter during running, and operating far from their optimal lengths. As a result, greater activation was required to produce the necessary plantarflexion moment and the metabolic power requirement of running was increased. A similar trend has been observed when exam- ining plantar-flexor muscle-tendon behaviour at varying running speeds [8]. As speed increases, tendon contribution to MTU positive work increases; however, muscle fibres oper- ate at progressively unfavourable regions of the force-length curve. In the current study, mus- cle fibre lengths and shortening were similar between barefoot and shod conditions (S1 Appendix). Therefore, in both conditions the plantar-flexor muscle fibres were at similar oper- ating lengths. Both tendon contributions to work, and the resulting effects on muscle fibre length operating range must be considered when understanding the effects of fibre and tendon behaviour on running economy. Previous studies show that increased ankle work when running barefoot or in minimalist shoes is associated with a decrease in mechanical work at the knee [10, 37]. Fuller et al. [37] suggested the shift in mechanical work towards the ankle may allow greater elastic-energy stor- age and recovery in the Achilles tendon, and therefore improve mechanical efficiency. The plantar-flexors have short fibres and a long, compliant tendon while the MTUs supporting the knee are generally larger with relatively shorter tendons [41]. Therefore, it could be more eco- nomical to rely on the plantar-flexors to produce power during stance [37]. A more holistic view of lower-limb positive work production may suggest that increased reliance on the ankle will result in increased elastic-energy storage and recovery, as this is better facilitated by the plantar-flexors. Further studies which examine the muscle and tendon mechanical behaviour of the MTUs supporting both the ankle and knee during forefoot shod and barefoot running are required to further explore this theory. We observed an increase in total Achilles tendon elastic strain energy when running bare- foot. Footwear also limits the compression and recoil of the elastic elements supporting the longitudinal arch of the foot [42]. The current study used a rigid mid and fore- foot model that does not consider the compression and recoil of the foot longitudinal arch. This is a limitation of the current study, as a simplified foot model may result in overestimation of predicted ankle joint power and therefore plantar-flexor MTUs power [43]. It is possible that this could influ- ence differences in MTU power of the plantar-flexors between barefoot and shod running. Further, the small decrease (0.1 m) in stride length during barefoot running may explain the changes in plantar-flexor MTU behaviour between barefoot and shod running. Predictions of plantar-flexor energetics are sensitive to the musculoskeletal model input parameters. The model and its underlying parameters were scaled to each subject’s height and mass; however subject-specific bone geometries and musculotendon measures were not uti- lised. In particular, Achilles tendon compliance can vary greatly in humans [44] and influences PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 9 / 13 the magnitude of muscle fibre work and tendon elastic strain energy [39]. This is important to acknowledge within the context of this study. However, a robust method was used to deter- mine the most appropriate tendon compliance, and this resulted in fibre [6, 16, 27] and tendon [15, 16] strains that were consistent with ultrasound measurements during steady-state run- ning. The Achilles tendon was also represented within the model as three separate tendons, rather than one common tendon. While the exact effects of this design on predictions of plan- tar-flexor energetics is unclear, non-uniform deformation of tendinous regions arising from each plantar-flexor have been reported [45] and may support this modelling approach. We only examined the acute effects of barefoot forefoot running on plantar-flexor energetics. Long-term use of minimalist shoes and running barefoot can result in adaptations to the mechanical properties of the tendon. Runners with four years of running in minimalist shoes displayed greater Achilles tendon stiffness compared to traditionally shod runners [46] and the Achilles tendon adapts to increased loading when exposed to minimalist shoe running by increasing stiffness [47]. This suggests that greater Achilles tendon stiffness might also be seen in habitual barefoot runners. The mechanical properties of the tendon were the same during the barefoot and shod simulations. This study does not account for adaptations to the mechan- ical properties of the tendon that may occur with prolonged barefoot running, which may influence predicted plantar-flexor energetics. Finally, we used the same performance criterion to estimate muscle-tendon parameters during barefoot and shod running. It is possible that participants utilised a different cost function to minimise muscle excitations than the one used in this study; however, there is currently no evidence to support this. In conclusion, running barefoot with a forefoot strike increased plantar-flexor positive work and power generation and Achilles tendon elastic strain energy when compared to shod forefoot running. The relative contribution of tendon to plantar-flexor MTU positive work remained similar between shod and barefoot forefoot running. These results indicate that barefoot forefoot running does not preferentially favour elastic energy over muscle fibre work more than shod running. Those who adopt barefoot forefoot running should be aware of the greater demand on the plantar-flexors and Achilles tendon compared to shod forefoot run- ning. Future studies should consider how mechanical adaptations of the Achilles tendon in habitual barefoot runners may influence plantar-flexor energetics. Supporting information S1 Appendix. Plantar-flexor muscle fibre, tendon and musculotendon normalised lengths. (TIF) S1 Dataset. Individual data for the plantar-flexor muscle, tendon and musculotendon units during barefoot and shod forefoot running. (XLSX) Acknowledgments The authors would like to Dr Amy Hicks for her assistance with data collection and analysis and Cody Lindsay for his assistance in reviewing the manuscript. Author Contributions Conceptualization: Jason Bonacci. Formal analysis: Jason Bonacci, Adrian Lai. Investigation: Jason Bonacci, Wayne Spratford. PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 10 / 13 Methodology: Jason Bonacci, Wayne Spratford, Adrian Lai. Project administration: Jason Bonacci. Resources: Wayne Spratford, Adrian Lai. Visualization: Jason Bonacci, Claire Kenneally-Dabrowski, Danielle Trowell. Writing – original draft: Jason Bonacci, Claire Kenneally-Dabrowski, Danielle Trowell, Adrian Lai. Writing – review & editing: Jason Bonacci, Wayne Spratford, Claire Kenneally-Dabrowski, Danielle Trowell, Adrian Lai. References 1. Komi P. Relevance of in vivo force measurements to human biomechanics. J Biomech. 1990; 23:23– 34. https://doi.org/10.1016/0021-9290(90)90038-5 PMID: 2081741 2. Dorn T, Schache A, Pandy M. Muscular strategy shift in human running: dependence of running speed on hip and ankle muscle performance. J Exp Biol. 2012; 215(11):1944–56. 3. Hamner S, Delp S. Muscle contributions to fore-aft and vertical body mass center accelerations over a range of running speeds. J Biomech. 2013; 46(4):780–7. https://doi.org/10.1016/j.jbiomech.2012.11. 024 PMID: 23246045 4. Biewener A. Muscle function in vivo: a comparison of muscles used for elastic energy savings versus muscles used to generate mechanical power. Am Zool. 1998; 38(4):703–17. 5. Roberts TJ. The integrated function of muscles and tendons during locomotion. Comp Biochem Physiol A Mol Integr Physiol. 2002; 133(4):1087–99. https://doi.org/10.1016/s1095-6433(02)00244-1 PMID: 12485693 6. Farris D, Sawicki G. Human medial gastrocnemius force–velocity behavior shifts with locomotion speed and gait. Proc Natl Acad Sci U S A. 2012; 109(3):977–82. https://doi.org/10.1073/pnas.1107972109 PMID: 22219360 7. Hof A, Van Zandwijk J, Bobbert M. Mechanics of human triceps surae muscle in walking, running and jumping. Acta Physiologica Scandinavica. 2002; 174(1):17–30. https://doi.org/10.1046/j.1365-201x. 2002.00917.x PMID: 11851593 8. Lai A, Schache A, Lin Y, Pandy M. Tendon elastic strain energy in the human ankle plantar-flexors and its role with increased running speed. J Exp Biol. 2014; 217(17):3159–68. https://doi.org/10.1242/jeb. 100826 PMID: 24948642 9. Sinclair J, Atkins S, Richards J, Vincent H. Modelling of muscle force distributions during barefoot and shod running. J Hum Kinet. 2015; 47:9. https://doi.org/10.1515/hukin-2015-0057 PMID: 26557186 10. Bonacci J, Saunders P, Hicks A, Rantalainen T, Vicenzino B, Spratford W. Running in a minimalist and lightweight shoe is not the same as running barefoot: a biomechanical study. Br J Sports Med. 2013; 47 (6):387–92. https://doi.org/10.1136/bjsports-2012-091837 PMID: 23314887 11. Perl D, Daoud A, Lieberman D. Effects of footwear and strike type on running economy. Med Sci Sport Exer. 2012; 44(7):1335–43. https://doi.org/10.1249/MSS.0b013e318247989e PMID: 22217565 12. Franz J, Wierzbinski C, Kram R. Metabolic cost of running barefoot versus shod: is lighter better? Med Sci Sport Exer. 2012; 44(8):1519–25. https://doi.org/10.1249/MSS.0b013e3182514a88 PMID: 22367745 13. Tung K, Franz J, Kram R. A test of the metabolic cost of cushioning hypothesis during unshod and shod running. Med Sci Sport Exer. 2014; 46(2):324–9. https://doi.org/10.1249/MSS.0b013e3182a63b81 PMID: 24441213 14. Fuller J, Thewlis D, Tsiros M, Brown N, Buckley J. Effects of a minimalist shoe on running economy and 5-km running performance. J Sport Sci. 2016; 34(18):1740–5. https://doi.org/10.1080/02640414.2015. 1136071 PMID: 27328725 15. Lichtwark G, Bougoulias K, Wilson A. Muscle fascicle and series elastic element length changes along the length of the human gastrocnemius during walking and running. J Biomech. 2007; 40(1):157–64. https://doi.org/10.1016/j.jbiomech.2005.10.035 PMID: 16364330 16. Lai A, Lichtwark GA, Schache AG, Lin Y-C, Brown NA, Pandy MG. In vivo behavior of the human soleus muscle with increasing walking and running speeds. J Appl Physiol. 2015; 118(10):1266–75. https://doi. org/10.1152/japplphysiol.00128.2015 PMID: 25814636 PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 11 / 13 17. Lieberman D, Venkadesan M, Werbel W, Daoud A, D’andrea S, Davis I, et al. Foot strike patterns and collision forces in habitually barefoot versus shod runners. Nature. 2010; 463(7280):531–5. https://doi. org/10.1038/nature08723 PMID: 20111000 18. Kristianslund E, Krosshaug T, van den Bogert A. Effect of low pass filtering on joint moments from inverse dynamics: implications for injury prevention. J Biomech. 2012; 45(4):666–71. https://doi.org/10. 1016/j.jbiomech.2011.12.011 PMID: 22227316 19. Tam N, Tucker R, Astephen Wilson JL. Individual responses to a barefoot running program: insight into risk of injury. Am J Sport Med. 2016; 44(3):777–84. https://doi.org/10.1177/0363546515620584 PMID: 26744483 20. Tam N, Wilson JLA, Coetzee DR, van Pletsen L, Tucker R. Loading rate increases during barefoot run- ning in habitually shod runners: Individual responses to an unfamiliar condition. Gait Posture. 2016; 46:47–52. https://doi.org/10.1016/j.gaitpost.2016.02.013 PMID: 27131176 21. Yong J, Silder A, Delp S. Differences in muscle activity between natural forefoot and rearfoot strikers during running. J Biomech. 2014; 47(15):3593–7. https://doi.org/10.1016/j.jbiomech.2014.10.015 PMID: 25458201 22. Hamner S, Seth A, Delp S. Muscle contributions to propulsion and support during running. J Biomech. 2010; 43(14):2709–16. https://doi.org/10.1016/j.jbiomech.2010.06.025 PMID: 20691972 23. Arnold E, Hamner S, Seth A, Millard M, Delp S. How muscle fiber lengths and velocities affect muscle force generation as humans walk and run at different speeds. J Exp Biol. 2013; 216(11):2150–60. https://doi.org/10.1242/jeb.075697 PMID: 23470656 24. Thelen D, Chumanov E, Best T, Swanson S, Heiderscheit B. Simulation of biceps femoris musculoten- don mechanics during the swing phase of sprinting. Med Sci Sport Exer. 2005; 37(11):1931–8. https:// doi.org/10.1249/01.mss.0000176674.42929.de PMID: 16286864 25. Lai A. Muscle and tendon mechanical interactions during human locomotion [Doctoral thesis]. Mel- bourne, Australia: University of Melbourne; 2015. 26. Schache AG, Dorn TW, Blanch PD, Brown NA, Pandy MG. Mechanics of the human hamstring muscles during sprinting. Med Sci Sport Exer. 2012; 44(4):647–58. https://doi.org/10.1249/MSS. 0b013e318236a3d2 PMID: 21912301. 27. Giannakou E, Aggeloussis N, Arampatzis A. Reproducibility of gastrocnemius medialis muscle architec- ture during treadmill running. J Electromyog Kines. 2011; 21(6):1081–6. https://doi.org/10.1016/j. jelekin.2011.06.004 PMID: 21763154 28. Maganaris C, Paul J. Tensile properties of the in vivo human gastrocnemius tendon. J Biomech. 2002; 35(12):1639–46. https://doi.org/10.1016/s0021-9290(02)00240-3 PMID: 12445617 29. Muramatsu T, Muraoka T, Takeshita D, Kawakami Y, Hirano Y, Fukunaga T. Mechanical properties of tendon and aponeurosis of human gastrocnemius muscle in vivo. J Appl Physiol. 2001; 90(5):1671–8. https://doi.org/10.1152/jappl.2001.90.5.1671 PMID: 11299254 30. Zajac FE. Muscle and tendon: properties, models, scaling, and application to biomechanics and motor control. Crit Rev Biomed Eng. 1989; 17(4):359–411. PMID: 2676342. 31. Delp S, Anderson F, Arnold A, Loan P, Habib A, John C, et al. OpenSim: open-source software to create and analyze dynamic simulations of movement. IEEE Trans Biomed Eng. 2007; 54(11):1940–50. https://doi.org/10.1109/TBME.2007.901024 PMID: 18018689 32. Lu T, O’connor J. Bone position estimation from skin marker co-ordinates using global optimisation with joint constraints. J Biomech. 1999; 32(2):129–34. 33. Thelen D, Anderson F. Using computed muscle control to generate forward dynamic simulations of human walking from experimental data. J Biomech. 2006; 39(6):1107–15. https://doi.org/10.1016/j. jbiomech.2005.02.010 PMID: 16023125 34. Thelen D, Anderson F, Delp S. Generating dynamic simulations of movement using computed muscle control. J Biomech. 2003; 36(3):321–8. https://doi.org/10.1016/s0021-9290(02)00432-3 PMID: 12594980 35. Hicks J, Uchida T, Seth A, Rajagopal A, Delp S. Is my model good enough? Best practices for verifica- tion and validation of musculoskeletal models and simulations of movement. J Biomech Eng. 2015; 137 (2). 36. Cohen J. A power primer. Psychol Bull. 1992; 112(1):155. https://doi.org/10.1037//0033-2909.112.1. 155 PMID: 19565683 37. Fuller JT, Buckley JD, Tsiros MD, Brown NA, Thewlis D. Redistribution of mechanical work at the knee and ankle joints during fast running in minimalist shoes. J Athl Train. 2016; 51(10):806–12. https://doi. org/10.4085/1062-6050-51.12.05 PMID: 27834504 38. Paquette M, Zhang S, Baumgartner L. Acute effects of barefoot, minimal shoes and running shoes on lower limb mechanics in rear and forefoot strike runners. Footwear Sci. 2013; 5(1):9–18. PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 12 / 13 39. Lai A, Schache AG, Brown NA, Pandy MG. Human ankle plantar flexor muscle–tendon mechanics and energetics during maximum acceleration sprinting. J R Soc Interface. 2016; 13(121):20160391. https:// doi.org/10.1098/rsif.2016.0391 PMID: 27581481 40. Uchida TK, Hicks JL, Dembia CL, Delp SL. Stretching your energetic budget: how tendon compliance affects the metabolic cost of running. PloS One. 2016; 11(3):e0150378. https://doi.org/10.1371/journal. pone.0150378 PMID: 26930416 41. Horsman MK, Koopman HF, van der Helm FC, Prose´ LP, Veeger H. Morphological muscle and joint parameters for musculoskeletal modelling of the lower extremity. Clin Biomech. 2007; 22(2):239–47. 42. Stearne S, McDonald K, Alderson J, North I, Oxnard C, Rubenson J. The foot’s arch and the energetics of human locomotion. Sci Rep. 2016; 6(1):1–10. 43. Bruening DA, Cooney KM, Buczek FL. Analysis of a kinetic multi-segment foot model part II: kinetics and clinical implications. Gait Posture. 2012; 35(4):535–40. https://doi.org/10.1016/j.gaitpost.2011.11. 012 PMID: 22197290 44. Morrison SM, Dick TJ, Wakeling JM. Structural and mechanical properties of the human Achilles ten- don: Sex and strength effects. J Biomech. 2015; 48(12):3530–3. https://doi.org/10.1016/j.jbiomech. 2015.06.009 PMID: 26159060 45. Franz JR, Slane LC, Rasske K, Thelen DG. Non-uniform in vivo deformations of the human Achilles ten- don during walking. Gait Posture. 2015; 41(1):192–7. https://doi.org/10.1016/j.gaitpost.2014.10.001 PMID: 25457482 46. Histen K, Arntsen J, L’Hereux L, Heeren J, Wicki B, Saint S, et al. Achilles tendon properties of minimal- ist and traditionally shod runners. J Sport Rehabil. 2017; 26(2):159–64. https://doi.org/10.1123/jsr. 2016-0006 PMID: 27632859 47. Joseph MF, Histen K, Arntsen J, L’Hereux L, Defeo C, Lockwood D, et al. Achilles tendon adaptation during transition to a minimalist running style. J Sport Rehabil. 2017; 26(2):165–70. https://doi.org/10. 1123/jsr.2016-0007 PMID: 27632879 PLOS ONE The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners PLOS ONE | https://doi.org/10.1371/journal.pone.0274806 September 19, 2022 13 / 13

The effect of footwear on mechanical behaviour of the human ankle plantar-flexors in forefoot runners.

09-19-2022

Bonacci, Jason,Spratford, Wayne,Kenneally-Dabrowski, Claire,Trowell, Danielle,Lai, Adrian

eng

PMC9565015

Citation: Tomovic, M.; Toliopoulos, A.; Koutlianos, N.; Dalkiranis, A.; Bubanj, S.; Deligiannis, A.; Kouidi, E. Correlation between Cardiopulmonary Indices and Running Performance in a 14.5 km Endurance Running Event. Int. J. Environ. Res. Public Health 2022, 19, 12289. https://doi.org/10.3390/ ijerph191912289 Academic Editors: Pantelis T. Nikolaidis and Masatoshi Nakamura Received: 3 August 2022 Accepted: 23 September 2022 Published: 27 September 2022 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2022 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). International Journal of Environmental Research and Public Health Article Correlation between Cardiopulmonary Indices and Running Performance in a 14.5 km Endurance Running Event Milena Tomovic 1 , Alexandros Toliopoulos 1 , Nikolaos Koutlianos 1, Anastasios Dalkiranis 1, Sasa Bubanj 2 , Asterios Deligiannis 1 and Evangelia Kouidi 1,* 1 Sports Medicine Laboratory, School of Physical Education and Sports Science, Aristotle University, Thermi PC, 57001 Thessaloniki, Greece 2 Faculty of Sport and Physical Education, University of Nis, 18000 Nis, Serbia * Correspondence: kouidi@phed.auth.gr Abstract: Background: Running is a common recreational activity, and the number of long-distance- race participants is continuously growing. It is well-established that regular physical activity can prevent and manage non-communicable diseases and benefit public health. Training for a long- distance race requires development of specific aerobic abilities and should generate the desired race performance. The purpose of this study was to support the training design and motivation of recreational endurance runners, by investigating whether a 14.5 km race performance of long-distance runners correlates with their cardiopulmonary indices measured in the laboratory. Methods: To examine the relationships of a 14.5 km running performance with the cardiopulmonary parameters of amateur runners, a cross-sectional study design was applied. Fifteen (eleven men and four women) recreational long-distance runners (aged 41.3 ± 9.2 years) from Northern Greece were included in the study and were evaluated in the laboratory within one week before an endurance running race—the 14.5 km Philip Road race, in Greece. The laboratory-based examinations of the athletes consisted of a comprehensive medical pre-participation screening and maximal cardiopulmonary exercise testing. Results: The results showed that the 14.5 km race performance time (73.8 ± 9.7 min) significantly correlated with the cardiopulmonary-exercise-testing speed-related indices at specific submaximal and maximal workloads (p < 0.01, p < 0.05), while the cardiopulmonary indices of oxygen uptake did not reliably predict race running time (p > 0.05). Conclusions: There is a better correlation of the 14.5 km running performance of recreational long-distance runners with the cardiopulmonary- exercise-testing speed-related indices at specific workloads than with the indices of oxygen uptake, running economy or respiratory economy. When preparing a training strategy, amateur long-distance runners should mostly rely on specific running-speed-related laboratory data rather than on oxygen- uptake values. Keywords: running; maximal oxygen uptake; running economy; sports performance; cardiopulmonary exercise test 1. Introduction Running is a common recreational activity, and the number of long-distance-race participants is continuously growing [1–5]. In England alone, more than 3 million adults participate in recreational running each month, and the USA and Australia show similar trends [6–8]. According to the Physical Activity Council (the USA’s definitive source for sports, fitness and recreational activity participation), running is one of the top 10 recre- ational activities that inactive Americans would choose, if about to commence regular exercise [8]. Regular physical activity is one of the cornerstones of public health, as it is proven to help prevent and manage non-communicable diseases such as hypertension, obesity and several cancers. An active lifestyle respects the environment and induces behaviours that can preserve and improve environmental health [7,9]. Regular running Int. J. Environ. Res. Public Health 2022, 19, 12289. https://doi.org/10.3390/ijerph191912289 https://www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2022, 19, 12289 2 of 11 has confirmed health benefits [10–14], and a recent systematic review and meta-analysis indicates that running reduces the risk of all-cause, cardiovascular and cancer mortality by 27%, 30% and 23%, respectively [9]. The same research identifies a literature gap in studies that would include sustained running participation, reproducible assessment of running habits and accurate evaluation of running performance. Training for a long-distance race requires development of specific aerobic abilities and should generate the desired race performance. Amateur runners often use elite runners’ training methodologies, risking external overload, with consequential higher incidence of overuse injuries [15–17]. Furthermore, the studied relationship between recreational runners’ motivation and the incidence of injury emphasises the importance of adequate training methodology for injury prevention, especially among novice runners [17–19]. Thus, it is important to understand and objectively evaluate physiological and other parame- ters that, according to the existing literature, might influence or predict the long-distance running performance of amateur runners [20–22]. The available scientific evidence does not offer a reliable equation for performance prediction, and relevant studies are char- acterised with high data heterogeneity and often controversial findings. It is not clear if anthropometric [22], cardiorespiratory [21,23] or training load [21,24] indicators form valid performance-prediction models for popular long-distance amateur running events [20]. Additionally, performance prediction models do not identify the physiological parameters necessary for adequate training prescription. Data on anthropometric parameters are in- complete [22], while cardiopulmonary indices and training load indicators such as maximal oxygen uptake and kilometres of running per week, although studied repeatedly and with an immense amount of data, still do not provide the best solution for training design and follow up [22,23]. The purpose of this study was to support the training design and motivation of recreational endurance runners, by investigating whether a 14.5 km race performance of long-distance runners correlates with their cardiopulmonary indices measured in the laboratory. Hopefully, our findings will help recreational runners, not ready for or capable of a whole marathon race, reach the desired performance level. 2. Materials and Methods To examine the relationships of a 14.5 km running performance with cardiopulmonary parameters of amateur runners, a cross-sectional study design was applied. Fifteen (eleven men and four women) recreational long-distance runners from Northern Greece were included in the study and were evaluated in the laboratory within one week before an endurance running race—the 14.5 km Philip Road race, Greece. The laboratory-based examinations of the athletes consisted of a comprehensive medical pre-participation screen- ing and maximal cardiopulmonary exercise testing (CPET) performed on a treadmill. All amateur runners included were healthy and had been training regularly. They gave an informed consent and completed a questionnaire with detailed medical and training history. Participants’ training history will be presented in the results section. Anthropometric parameters were measured prior to CPET (height—SECA Leices- ter resuscitation meter; weight, body fat and muscle mass percentage—Omron Karada Scan BF511, HBF-511T-E/HBF-511B-E). Maximal treadmill (Montara Trackmaster 428, KS, USA) CPET, via a breath-by-breath gas-analyzing system (Geratherm Respiratory GmbH’s BlueCherry, Bad Kissingen, Germany), followed the clinical exam in an appropriate labora- tory environment (room temperature 20 ◦C and relative humidity between 25–50%). The gas-analyzing system was validated prior to each testing. Heart rate (HR) was monitored by Polar Receiver Pulstik system (Geratherm Respiratory GmbH, Bad Kissingen, Germany). Participants did not train nor consume any nutritional supplements or caffeinated bev- erages 24 h before testing. The maximal ramp exercise protocol [25] included: warm-up: 1 min at speed of 2 km/h, slope 0% and 3 min at speed of 5 km/h, slope 0%; test: speed 5 km/h and slope 1%. The workload was enhanced by a gradual increase in the speed, by 1.2 km/h per min, until exhaustion. Moreover, the slope increased to 2%, when participants Int. J. Environ. Res. Public Health 2022, 19, 12289 3 of 11 reached a speed of 13 km/h, and did not increase further with speed increment beyond 13 km/h. Recovery was 5 min. During the CPET, minute ventilation (VE, L/min), oxygen consumption (VO2, relative (mL/kg/min) and absolute values (mL/min)), respiratory exchange ratio (RER), VCO2 (carbon dioxide production, mL/kg/min), ventilatory equivalents for oxygen and carbon dioxide (VE/VO2, VE/VCO2), PETO2 (end tidal oxygen volume), PETCO2 (end tidal carbon dioxide volume), HR, oxygen pulse (O2 pulse, mL/beat), time to voluntary exhaustion (4 min of warm up and 5 min of recovery were excluded) and the 2nd ventilatory threshold (VT) were measured [26]. Additionally, the same parameters as well as the running speeds (“v” expressed in km/h) were noted at the VT and RER1 (RER value of 1, VO2 = VCO2) points of the CPET: VO2VT, VO2RER1, vVT, vRER1, HRVT, HRRER1 and maximal parameters: VO2max, vVO2max, vpeak, HRmax, VEmax, max O2 pulse, RERmax, tidal volume max, VO2max/VO2ref% (maximal VO2 expressed as a percentage of a VO2max value predicted according to participant’s age, sex and training habits) and VT/VO2max% (percentage of achieved maximal VO2 at VT CPET’s point). Running economy (RE)8, RE10, RE12, VO2/WR and VO2/WR (7.9–13.1 km/h) values were used as running economy indicators. The RE8, RE10 and RE12 indicators were derived from VO2 of each athlete at speeds of 8 km/h, 10 km/h and 12 km/h, respectively, and were expressed in mL/min/kg. The VO2/WR and VO2/WR (7.9–13.1) indicators were calculated directly from the ergospirometer, and they represent VO2 achieved at these speeds and converted to a work rate (WR) expressed in watts (mL/min/watt). The VO2/WR index represents the average value through the whole test load, while the VO2/WR (7.9–13.1) index represents the average oxygen uptake per work rate between speeds of 7.9 km/h and 13.1 km/h. These specific values of speeds were marked because they were reached by all participants; thus, a direct comparison between them was possible. Furthermore, the speed of 7.9 km/h is the lowest intensity that forced running over walking. Two weeks after the lab evaluation, the athletes participated in the Philip Road race— 14.5 km route from Vergina to Veria, in Greece. The competition was entirely on asphalt, and it started at 10.30 am. The athletes were offered water at 5.4 km and water and isotonic fluid at 10.4 km. The weather on the race day was clear and with optimal conditions. The race performance times of the runners were collected from the official results, and these time records were net, i.e., the time from the moment the athletes crossed the starting line to the time they arrived at the finish line. Descriptive statistics were used to describe categorical variables. Continuous variables were expressed as mean ± SD and Shapiro–Wilk test was used for testing the normality of all data. The differences between values were evaluated using the paired sample t-test. Relationships between categorical variables were tested using the chi-squared statistic. The Pearson linear correlation coefficient was used for quantitative values, and, for the non- linear data, we used Spearman non-parametric correlation. Statistical analysis was carried out with the IBM SPSS statistical program (Social Package for Social Sciences, Chicago, IL, USA, version 25.0). A two-tailed p < 0.05 was accepted as statistically significant. 3. Results All fifteen recruited recreational runners (aged 41.3 ± 9.2 years) completed the 14.5 km race and had no injuries or any health disorders during the race. They had been practicing running for the past 5.6 (±5.6) years, with a frequency of 5 (±1.4) days, 7.2 (±3.1) hours and 52.7 (±19.5) km per week. Their demographic, anthropometric and race performance data, as shown in Tables 1 and 2, contain the overview of the participants’ CPET results. Int. J. Environ. Res. Public Health 2022, 19, 12289 4 of 11 Table 1. Demographic, anthropometric and race performance characteristics of participants. Participants Race Performance (min) Average Race Speed (km/h) Age (Years) Height (cm) Weight (kg) BMI b (kg/m2) Body Fat (%) Muscle Mass (%) 1 57.93 15.02 37 189 83 23.24 19 38.3 2 59.38 14.65 37 180.5 76 23.33 21.6 37 3 64.18 13.56 48 181 77 23.50 18.7 37.7 4 66.7 13.04 51 169 68 23.81 18.1 38.4 5 67.83 12.83 52 178 74 23.36 18.6 37.4 6 71.23 12.21 38 171 73 24.96 23 37.1 7 71.42 12.18 34 176 72 23.24 17.2 40.7 8 71.85 12.11 39 184 92 27.17 28.1 33.4 9 74.38 11.70 32 156.5 57 23.27 32.9 28.4 10 78.13 11.14 39 173.5 78 25.91 26 35 11 79.83 10.90 39 169.5 78 27.15 25.7 35.8 12 81.23 10.71 66 173 79 26.40 20.7 35.4 13 86.63 10.04 33 177 66 21.70 20.3 35.5 14 86.65 10.04 40 168.5 68 23.95 30.8 30.5 15 89.03 9.77 34 171 60 20.52 22.6 33.6 Mean Value 73.8 12.0 41.3 174.5 73.4 24.1 22.9 35.6 SD a 9.7 1.6 9.2 7.8 8.8 2.0 4.8 3.2 Median 71.85 12.11 39 173.5 74.0 23.50 21.6 35.8 a standard deviation; b body mass index. Table 2. Cardiopulmonary exercise testing results of participants. Participants Test Time (min) VE b (L/min) Running Speed at VT c Point (km/h) Maximal Running Speed VO2max d (L) VO2max d (mL/min/kg) Oxygen Pulse (mL/Beat) Maximal Heart Rate (Beat/min) 1 11.17 123 15.4 16.1 4.24 51.1 24 170 2 12.28 144 17.1 17.4 3.96 52.1 22 177 3 10.08 103 11.6 15.1 3.21 41.7 22 160 4 11.75 132 16.6 16.6 3.8 55.9 21 181 5 9.97 146 13.6 14.9 3.3 44.6 23 141 6 10.15 118 15.1 15.1 3.21 44.0 25 169 7 11.17 123 16.1 16.1 3.76 52.2 22 175 8 11.32 135 15.6 16.5 4.08 44.3 29 163 9 10.10 95 14.4 15.1 2.66 46.7 14 185 10 11.25 143 15.1 16.1 3.45 44.2 24 171 11 9.63 131 14.6 14.6 3.48 44.6 20 176 12 8.97 116 13.4 13.9 3.46 43.8 21 171 13 8.82 108 13.1 13.9 3.14 47.6 17 182 14 9.10 113 13.6 14.1 3.01 44.3 17 178 15 8.03 91 12.6 13.1 2.18 36.3 12 179 Mean Value 10.3 121.4 14.5 15.2 3.4 46.2 20.9 171.9 SD a 1.2 17.4 1.5 1.2 0.5 4.9 4.4 11.0 Median 10.10 123.00 14.60 15.10 3.45 44.59 22.00 175.00 a standard deviation; b minute ventilation; c ventilatory threshold; d maximal oxygen consumption. Int. J. Environ. Res. Public Health 2022, 19, 12289 5 of 11 Correlation analysis of the anthropometric parameters showed, for the race perfor- mance times, significant coefficients only for the height (174.5 ± 7.8 cm, r = −0.534, p < 0.01) and the muscle mass percentage (35.6 ± 3.2%, r = −0.696, p < 0.01) of the participants. From the participants’ training history data, only the number of kms that participants achieved during their weekly training showed a statistically significant negative correlation (52.7 ± 19.5 km, r = −0.640, p < 0.05) with the race performance time. Data obtained during the participants’ CPET, which had a statistically significant negative correlation with the athletes’ race performance, measured as race end time, were: fatigue time (10.3 ± 1.2 min, r = −0.718, p < 0.01), vVO2max (14.5 ± 1.5 km/h, r = −0.531, p < 0.05, Figure 1), vpeak (15.2 ± 1.2 km/h, r = −0.754, p < 0.01, Figure 2), absolute VO2max (3.4 ± 0.5 L/min, r = −0.617, p < 0.05), max O2 pulse (20.9 ± 4.4 mL/beat, r = −0.607, p < 0.05, Figure 3) and tidal volume max (2.4 ± 0.4, r = −0.550, p < 0.05). The speed values (vVT, vRER) achieved by runners at the VT and RER1 points of their CPET had a signifi- cant (p < 0.01) negative correlation (r = −0.733, r = −0.671, respectively, Figures 4 and 5) with the runners’ race performance times. Relative VO2max (46.2 ± 4.9 mL/min/kg, r = −0.422, p > 0.05), VO2VT (42.5 ± 4.9 mL/min/kg, r = −0.390, p > 0.05), VT/VO2max% (92.1 ± 7.1%, r = −0.468, p > 0.05), VE (121.4 ± 17.4, r = −0.468, p > 0.05), VO2max/VO2ref% (130.2 ± 17.9%, r = 0.252, p > 0.0.5) and HRVT (162.3 ± 14.9/min, r = 0.354, p > 0.05) did not show any statistically significant positive or negative correlation. Running economy indicators were not significantly correlated with the race perfor- mance times of the studied runners: RE8 (27.2 ± 2.1 mL/min/kg, r = −0.036, p > 0.05), RE10 (33.2 ± 2.3 mL//min/kg), r = 0.232, p > 0.05), RE12 (38.5 ± 3.3 mL/min/kg), r = 0.079, p > 0.05), VO2/WR (7.1 ± 1.6 mL/min/watt, r = 0.164, p > 0.05) and VO2/WR (7.9–13.1) (5.2 ± 6.7 mL/min/watt, r = −0.181, p > 0.05). Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW  5  of  11    Data obtained during the participants’ CPET, which had a statistically significant  negative correlation with the athletes’ race performance, measured as race end time, were:  fatigue time (10.3 ± 1.2 min, r = −0.718, p < 0.01), vVO2max (14.5 ± 1.5 km/h, r = −0.531, p < 0.05,  Figure 1), vpeak (15.2 ± 1.2 km/h, r = −0.754, p < 0.01, Figure 2), absolute VO2max (3.4 ± 0.5  L/min, r = −0.617, p < 0.05), max O2 pulse (20.9 ± 4.4 mL/beat, r = −0.607, p < 0.05, Figure 3)  and  tidal  volume  max  (2.4  ±  0.4,  r  =  −0.550,  p  <  0.05).  The  speed  values  (vVT,  vRER)  achieved by runners at the VT and RER1 points of their CPET had a significant (p < 0.01)  negative correlation (r = −0.733, r = −0.671, respectively, Figures 4 and 5) with the runners’  race performance times. Relative VO2max (46.2 ± 4.9 mL/min/kg, r = −0.422, p > 0.05), VO2VT  (42.5 ± 4.9 mL/min/kg, r = −0.390, p > 0.05), VT/VO2max% (92.1 ± 7.1%, r = −0.468, p > 0.05),  VE (121.4 ± 17.4, r = −0.468, p > 0.05), VO2max/VO2ref% (130.2 ± 17.9%, r = 0.252, p > 0.0.5)  and HRVT (162.3 ± 14.9/min, r = 0.354, p > 0.05) did not show any statistically significant  positive or negative correlation.    Figure 1. Correlation between participants’ speed at maximal oxygen consumption point during  cardiopulmonary exercise testing and race performance time (r = −0.531, p < 0.05). VO2max: maximal  oxygen consumption; CPET: cardiopulmonary exercise testing.  Figure 1. Correlation between participants’ speed at maximal oxygen consumption point during cardiopulmonary exercise testing and race performance time (r = −0.531, p < 0.05). VO2max: maximal oxygen consumption; CPET: cardiopulmonary exercise testing. Int. J. Environ. Res. Public Health 2022, 19, 12289 6 of 11 Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW  6  of  11        Figure 2. Correlation between participants’ maximal speed during cardiopulmonary exercise test‐ ing and race performance time (r = −0.754, p < 0.01). CPET: cardiopulmonary exercise testing.    Figure 3. Correlation between participants’ maximal oxygen pulse during cardiopulmonary exer‐ cise testing and race performance time (r = −0.607, p < 0.05). CPET: cardiopulmonary exercise testing;  max O2 pulse: maximal oxygen pulse.  Figure 2. Correlation between participants’ maximal speed during cardiopulmonary exercise testing and race performance time (r = −0.754, p < 0.01). CPET: cardiopulmonary exercise testing. Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW  6  of  11        Figure 2. Correlation between participants’ maximal speed during cardiopulmonary exercise test‐ ing and race performance time (r = −0.754, p < 0.01). CPET: cardiopulmonary exercise testing.    Figure 3. Correlation between participants’ maximal oxygen pulse during cardiopulmonary exer‐ cise testing and race performance time (r = −0.607, p < 0.05). CPET: cardiopulmonary exercise testing;  max O2 pulse: maximal oxygen pulse.  Figure 3. Correlation between participants’ maximal oxygen pulse during cardiopulmonary exercise testing and race performance time (r = −0.607, p < 0.05). CPET: cardiopulmonary exercise testing; max O2 pulse: maximal oxygen pulse. Int. J. Environ. Res. Public Health 2022, 19, 12289 7 of 11 Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW  7  of  11        Figure 4. Correlation between participants’ speed at ventilatory threshold during cardiopulmonary  exercise testing and race performance time (r = −0.733, p < 0.01). VT: ventilatory threshold; CPET:  cardiopulmonary exercise testing.    Figure 5. Correlation between participants’ speed at respiratory exchange ratio value (RER = 1) dur‐ ing cardiopulmonary exercise testing and race performance time (r = −0.671, p < 0.01). CPET: cardi‐ opulmonary exercise testing; RER1: respiratory exchange ratio equal to one (RER = 1).  Figure 4. Correlation between participants’ speed at ventilatory threshold during cardiopulmonary exercise testing and race performance time (r = −0.733, p < 0.01). VT: ventilatory threshold; CPET: cardiopulmonary exercise testing. Int. J. Environ. Res. Public Health 2022, 19, x FOR PEER REVIEW  7  of  11        Figure 4. Correlation between participants’ speed at ventilatory threshold during cardiopulmonary  exercise testing and race performance time (r = −0.733, p < 0.01). VT: ventilatory threshold; CPET:  cardiopulmonary exercise testing.    Figure 5. Correlation between participants’ speed at respiratory exchange ratio value (RER = 1) dur‐ ing cardiopulmonary exercise testing and race performance time (r = −0.671, p < 0.01). CPET: cardi‐ opulmonary exercise testing; RER1: respiratory exchange ratio equal to one (RER = 1).  Figure 5. Correlation between participants’ speed at respiratory exchange ratio value (RER = 1) during cardiopulmonary exercise testing and race performance time (r = −0.671, p < 0.01). CPET: cardiopulmonary exercise testing; RER1: respiratory exchange ratio equal to one (RER = 1). Int. J. Environ. Res. Public Health 2022, 19, 12289 8 of 11 4. Discussion The results showed that the 14.5 km race performance time of recreational runners was significantly correlated with the running speeds achieved during CPET at VT, RER1, VO2max and the test peak point. The good predictive potential of specific running speeds (peak velocity, vVO2max, vVT) achieved during the exercise testing of recreational en- durance runners was previously documented by a recent review [22] and other stud- ies [21,27–29]. Running speeds associated with anaerobic threshold, maximal oxygen uptake and CPET peak values include, as determinants, both the aerobic and anaerobic attributes of the selected runners [22,27,30]. This can explain the great predictive poten- tial of these parameters for a long-distance running event, which, in fact, requires both well-developed aerobic and anaerobic power [30]. Additionally, as already confirmed in the literature, the speed at the anaerobic threshold depends on both the maximal oxygen uptake and running economy [29–31]. Thus, the significant correlation of this specific speed with running performance, as noted in our study, is fully and favourably justified for separating the evaluation of the maximal oxygen uptake and running economy [29–31]. We can reasonably assume that the cumulative effect of the previous parameters shows a better predictive potential and responds to the complexity of long-distance-running’s physiological requirements. The CPET indices of oxygen uptake (relative VO2max, VO2VT, VT/VO2max%, VO2max/VO2ref%) were not valid for predicting the performance times in our amateur long-distance runners. Only the absolute value of VO2max showed a significant correlation with race time. This finding should be carefully interpreted, since the VO2, in L/min, as a measure of aerobic capacity, highly depends on individual anthropometric characteris- tics [26]. Thus, the use of this value should not be considered as a primary index to predict the performance times in amateur long-distance runners. The scientific literature has acknowledged the importance of oxygen uptake and aero- bic power for endurance running activities, but amateur running performance prediction models remain dependent on running-speed indices and, according to some studies, anaer- obic threshold and RE values [31–34]. Our study did not prove the utility of the calculated RE indicators, and studies with similar results concluded that RE is not useful in defining mid-distance running performance [28,35]. Furthermore, a recent systematic review and meta-analysis found that the existing evidence is not clear on the relationship of RE and mid- and long-distance running performance, as studies on this relationship have a high level of endogenous selection bias and an absence of allocation methodologies. Additional research is needed to evaluate the relationship between RE and running performance [36–38]. The participants’ CPET data for max O2 pulse demonstrated an interesting relationship with the race performance time. This parameter is traditionally used as a predictor of good cardiovascular health and for the evaluation and prognosis of ischemic cardiomyopathies. However, the existing scientific literature did not use max O2 pulse for performance prediction until now. O2 pulse depends on HR and oxygen uptake, and its maximal value is mainly determined by aerobic capacity—HRmax is age-related and does not depend on other individual physiological and pathological characteristics [26]. Future evaluation of max O2 pulse might be interesting for performance and health-related prognosis, especially in recreational athletic and patient populations. When the cardiovascular and respiratory fitness of physically active subjects is evaluated, it can be helpful from motivational and financial perspectives to produce, besides health indices, data that can predict or proclaim individual performance determinants [39,40]. Furthermore, physical fitness as a health determinant can help in defining tools for public health improvement, as it has been proven that regular physical activity can prevent and manage non-communicable diseases [7,11]. The available scientific evidence is heterogeneous regarding the utility of the an- thropometric parameters of amateur long-distance runners for performance-prediction models [21,22]. Our study results on runners’ anthropometry should be evaluated in regard to that heterogeneity. Recreational runners differ in body composition, and their training and race participation does not require professional runners’ discipline and consequential Int. J. Environ. Res. Public Health 2022, 19, 12289 9 of 11 body image. Thus, the study of anthropometric indices in amateur runners represents a great challenge for future research, since its utility may not be limited to only performance- prediction models. Anthropometric features of amateur mid- and long-distance runners can help define healthy and injury-free running guidelines. The outcomes of this study add valuable insight on amateur long-distance running performance and answer the literature’s need for a better understanding of the relationship between recreational runners’ cardiopulmonary indices and their training design. Our research highlights the cumulative effect of maximal oxygen uptake and running economy, expressed through running speeds, as better tools of training design for long-distance amateur running events. We define the cardiopulmonary indices suitable for the laboratory control of the training adaptations that are important for safe, motivated and injury-free recreational long-distance running. When designing the training strategy of amateur long- distance runners, one should use the running speeds achieved during CPET at VT, RER1, VO2max and the test peak point for progress control and periodization plan. The main limitations of the study were the small sample size and the absence of gender segregation. 5. Conclusions There is a better correlation of the 14.5 km running performance of recreational long- distance runners with CPET speed-related indices at specific workloads than with the indices of oxygen uptake, running economy or respiratory economy. When preparing a training strategy, amateur long-distance runners should mostly rely on specific running- speed-related laboratory data rather than on oxygen-uptake values. Author Contributions: M.T., A.T., N.K., A.D. (Anastasios Dalkiranis), S.B., A.D. (Asterios Deligiannis) and E.K. have made substantial contributions to the conception and design of the work as well as the acquisition, analysis and interpretation of data. All authors have read and agreed to the published version of the manuscript. Funding: This research received no external funding. Institutional Review Board Statement: The University Ethics Committee approved the study proto- col in accordance with the Helsinki Declaration for human research (approval number EC-16/2020, Thessaloniki, 12 May 2020, Vassilis Mougios). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Data Availability Statement: The data presented in this study are available on request from the corresponding author. The data are not publicly available due to the privacy of the included subjects. Conflicts of Interest: The authors declare no conflict of interest. References 1. Hulteen, R.M.; Smith, J.J.; Morgan, P.J.; Barnett, L.M.; Hallal, P.C.; Colyvas, K.; Lubans, D.R. Global participation in sport and leisure-time physical activities: A systematic review and meta-analysis. Prev. Med. 2017, 95, 14–25. [CrossRef] [PubMed] 2. Borgers, J.; Breedveld, K.; Tiessen-Raaphorst, A.; Thibaut, E.; Vandermeerschen, H.; Vos, S.; Scheerder, J. A study on the frequency of participation and time spent on sport in different organisational settings. Eur. Sport Manag. Q. 2016, 16, 635–654. [CrossRef] 3. Anthony, D.; Rüst, C.A.; Cribari, M.; Rosemann, T.; Lepers, R.; Knechtle, B. Differences in participation and performance trends in age group half and full marathoners. Chin. J. Physiol. 2014, 57, 209–219. [CrossRef] [PubMed] 4. Knechtle, B.; Nikolaidis, P.T.; Zingg, M.A.; Rosemann, T.; Rüst, C.A. Half-marathoners are younger and slower than marathoners. SpringerPlus 2016, 5, 76. [CrossRef] 5. Lima, M.G.; Malta, D.C.; Monteiro, C.N.; da Silva Sousa, N.F.; Stopa, S.R.; Medina, L.D.P.B.; de Azevedo Barros, M.B. Leisure-time physical activity and sports in the Brazilian population: A social disparity analysis. PLoS ONE 2019, 14, e0225940. [CrossRef] [PubMed] 6. Stamatakis, E.; Chaudhury, M. Temporal trends in adults’ sports participation patterns in England between 1997 and 2006: The health survey for England. Br. J. Sports Med. 2008, 42, 601–608. [CrossRef] 7. Eime, R.M.; Harvey, J.T.; Charity, M.J.; Casey, M.M.; Van Uffelen, J.G.Z.; Payne, W.R. The contribution of sport participation to overall health enhancing physical activity levels in Australia: A population-based study. BMC Public Health 2015, 15, 806. [CrossRef] Int. J. Environ. Res. Public Health 2022, 19, 12289 10 of 11 8. Wisconsin Office of Outdoor Recreation. 2018 Participation Report: The Physical Activity Council’s Annual Study Tracking Sports, Fitness and Recreation Participation in the US. Available online: https://outdoorrecreation.wi.gov/Documents/ Research%20Library%20Page%20files/US%20-%20Demographics%20%26%20Participation/Physical%20Activity%20Coucil% 20Participation%20Report_2018.pdf (accessed on 15 June 2022). 9. Pedisic, Z.; Shrestha, N.; Kovalchik, S.; Stamatakis, E.; Liangruenrom, N.; Grgic, J.; Titze, S.; Biddle, S.J.H.; Bauman, A.E.; Oja, P. Is running associated with a lower risk of all-cause, cardiovascular and cancer mortality, and is the more the better? A systematic review and meta-analysis. Br. J. Sports Med. 2019, 54, 898–905. [CrossRef] 10. Lavie, C.J.; Lee, D.C.; Sui, X.; Arena, R.; O’Keefe, J.H.; Church, T.S.; Milani, R.V.; Blair, S.N. Effects of running on chronic diseases and cardiovascular and all-cause mortality. Mayo Clin. Proc. 2015, 90, 1541–1552. [CrossRef] 11. Lee, D.C.; Brellenthin, A.G.; Thompson, P.D.; Sui, X.; Lee, I.M.; Lavie, C.J. Running as a key lifestyle medicine for longevity. Prog. Cardiovasc. Dis. 2017, 60, 45–55. [CrossRef] 12. Wang, Y.; Lee, D.C.; Brellenthin, A.G.; Eijsvogels, T.M.; Sui, X.; Church, T.S.; Lavie, C.J.; Blair, S.N. Leisure-time running reduces the risk of incident type 2 diabetes. Am. J. Med. 2019, 132, 1225–1232. [CrossRef] 13. Hespanhol, J.L.C.; Pillay, J.D.; van Mechelen, W.; Verhagen, E. Meta-analyses of the effects of habitual running on indices of health in physically inactive adults. Sport Med. 2015, 45, 1455–1468. [CrossRef] [PubMed] 14. Lee, D.C.; Pate, R.R.; Lavie, C.J.; Sui, X.; Church, T.S.; Blair, S.N. Leisure-time running reduces all-cause and cardiovascular mortality risk. J. Am. Coll. Cardiol. 2014, 4, 472–481. [CrossRef] 15. Videbæk, S.; Bueno, A.M.; Nielsen, R.O.; Rasmussen, S. Incidence of running-related injuries per 1000 h of running in different types of runners: A systematic review and meta-analysis. Sport Med. 2015, 45, 1017–1026. [CrossRef] [PubMed] 16. Kluitenberg, B.; van Middelkoop, M.; Diercks, R.; van der Worp, H. What are the differences in injury proportions between different populations of runners? A systematic review and meta-analysis. Sport Med. 2015, 45, 1143–1161. [CrossRef] [PubMed] 17. Gauffin, H.; Tillander, B.; Dahlström, Ö.; Lyth, J.; Raysmith, B.; Jacobsson, J.; Timpka, T. Maintaining motivation and health among recreational runners: Panel study of factors associated with self-rated performance outcomes at competitions. J. Sci. Med. Sport 2019, 22, 1319–1323. [CrossRef] 18. León-Guereño, P.; Tapia-Serrano, M.A.; Sánchez-Miguel, P.A. The relationship of recreational runners’ motivation and resilience levels to the incidence of injury: A mediation model. PLoS ONE 2020, 15, e0231628. [CrossRef] 19. Linton, L.; Valentin, S. Running with injury: A study of UK novice and recreational runners and factors associated with running related injury. J. Sci. Med. Sport 2018, 21, 1221–1225. [CrossRef] 20. Keogh, A.; Smyth, B.; Caulfield, B.; Lawlor, A.; Berndsen, J.; Doherty, C. Prediction equations for marathon performance: A systematic review. Int. J. Sports Physiol. Perform. 2019, 14, 1159–1169. [CrossRef] 21. Gómez-Molina, J.; Ogueta-Alday, A.; Camara, J.; Stickley, C.; Rodríguez-Marroyo, J.A.; García-López, J. Predictive variables of half-marathon performance for male runners. J. Sport Sci. Med. 2017, 16, 187–194. 22. Boullosa, D.; Esteve-Lanao, J.; Casado, A.; Peyré-Tartaruga, L.A.; Gomes da Rosa, R.; Del Coso, J. Factors affecting training and physical performance in recreational endurance runners. Sports 2020, 8, 35. [CrossRef] [PubMed] 23. Esteve-Lanao, J.; Del Rosso, S.; Larumbe-Zabala, E.; Cardona, C.; Alcocer-Gamboa, A.; Boullosa, D.A. Predicting recreational runners’ marathon performance time during their training preparation. J. Strength Cond. Res. 2021, 35, 3218–3224. [CrossRef] [PubMed] 24. Salinero, J.J.; Soriano, M.L.; Lara, B.; Gallo-Salazar, C.; Areces, F.; Ruiz-Vicente, D.; Abián-Vicén, J.; González-Millán, C.; Del Coso, J. Predicting race time in male amateur marathon runners. J. Sports Med. Phys. Fit. 2017, 57, 1169–1177. [CrossRef] [PubMed] 25. Winter, U.J.; Gitt, A.K.; Fritsch, J.; Berge, P.G.; Pothoff, G.; Hilger, H.H. Methodologic aspects of modern, computerized ergospirometry (CPX): Ramp program, constant workload test and CO2 rebreathing method. Z. Kardiol. 1994, 83, 13–26. [PubMed] 26. Wasserman, K.; Hansen, J.E.; Sue, D.Y.; Stringer, W.W.; Whipp, B.J. Principles of Exercise Testing and Interpretation: Including Pathophysiology and Clinical Applications, 4th ed.; Lippincott Williams & Wilkins: Philadelphia, PA, USA, 2004. 27. Scott, B.K.; Houmard, J.A. Peak running velocity is highly related to distance running performance. Int. J. Sports Med. 1994, 15, 504–507. [CrossRef] [PubMed] 28. Bragada, J.A.; Santos, P.J.; Maia, J.A.; Colaço, P.J.; Lopes, V.P.; Barbosa, T.M. Longitudinal study in 3,000 m male runners: Relationship between performance and selected physiological parameters. J. Sport Sci. Med. 2010, 9, 439–444. 29. Nicholson, R.M.; Sleivert, G.G. Indices of lactate threshold and their relationship with 10-km running velocity. Med. Sci. Sports Exerc. 2001, 33, 339–342. [CrossRef] 30. Billat, L.V.; Koralsztein, J.P. Significance of the velocity at VO2max and time to exhaustion at this velocity. Sport Med. 1996, 22, 90–108. [CrossRef] 31. Bassett, D.R.; Howley, E.T. Limiting factors for maximum oxygen uptake and determinants of endurance performance. Med. Sci. Sports Exerc. 2000, 32, 70–84. [CrossRef] 32. Conley, D.L.; Krahenbuhl, G.S. Running economy and distance running performance of highly trained athletes. Med. Sci. Sports Exerc. 1980, 12, 357–360. [CrossRef] 33. Daniels, J.; Daniels, N. Running economy of elite male and elite female runners. Med. Sci. Sports Exerc. 1992, 24, 483–489. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2022, 19, 12289 11 of 11 34. Bassett, D.R.; Howley, E.T. Maximal oxygen uptake: “Classical” versus “contemporary” viewpoints. Med. Sci. Sports Exerc. 1997, 29, 591–603. [CrossRef] [PubMed] 35. Grant, S.; Craig, I.; Wilson, J.; Aitchison, T. The relationship between 3 km running performance and selected physiological variables. J. Sports Sci. 1997, 15, 403–410. [CrossRef] [PubMed] 36. Borgen, N.T. Running performance, VO2max, and running economy: The widespread issue of endogenous selection bias. Sport Med. 2018, 48, 1049–1058. [CrossRef] [PubMed] 37. O Sullivan, I.J.; Johnson, M.I.; Hind, K.; Breen, S.; Francis, P. Are changes in running economy associated with changes in performance in runners? A systematic review and meta-analysis. J. Sports Sci. 2019, 37, 1521–1533. [CrossRef] [PubMed] 38. Alvero-Cruz, J.R.; Carnero, E.A.; García, M.A.G.; Alacid, F.; Correas-Gómez, L.; Rosemann, T.; Nikolaidis, P.T.; Knechtle, B. Predictive performance models in long-distance runners: A narrative review. Int. J. Environ. Res. Public Health 2020, 17, 8289. [CrossRef] 39. Lee, L.L.; Arthur, A.; Avis, M. Using self-efficacy theory to develop interventions that help older people overcome psychological barriers to physical activity: A discussion paper. Int. J. Nurs. Stud. 2008, 45, 1690–1699. [CrossRef] 40. André, N.; Agbangla, N.F. Are barriers the same whether I want to start or maintain exercise? A narrative review on healthy older adults. Int. J. Environ. Res. Public Health 2020, 17, 6247. [CrossRef]

Correlation between Cardiopulmonary Indices and Running Performance in a 14.5 km Endurance Running Event.

09-27-2022

Tomovic, Milena,Toliopoulos, Alexandros,Koutlianos, Nikolaos,Dalkiranis, Anastasios,Bubanj, Sasa,Deligiannis, Asterios,Kouidi, Evangelia

eng

PMC7775063

RESEARCH ARTICLE Effects of body movement on yaw motion in bipedal running lizard by dynamic simulation Jeongryul Kim1, Hongmin Kim2¤, Jaeheung Park3,4*, Hwa Soo KimID5*, TaeWon Seo6 1 Center for Healthcare Robotics, Korea Institute of Science and Technology, Seoul, South Korea, 2 School of Mechanical and Aerospace Engineering, Seoul National University, Seoul, South Korea, 3 Department of Intelligence and Information, Seoul National University, Seoul, South Korea, 4 Advanced Institutes of Convergence Technology (AICT), Suwon, Gyeonggi-do, South Korea, 5 Department of Mechanical System Engineering, Kyonggi University, Suwon-si, South Korea, 6 School of Mechanical Engineering, Hanyang University, Seoul, South Korea ¤ Current address: Institute of Advanced Machines and Design (IAMD), Seoul National University, Seoul, South Korea * park73@snu.ac.kr (JP); hskim94@kgu.ac.kr (HSK) Abstract Lizards run quickly and stably in a bipedal gait, with their bodies exhibiting a lateral S-shaped undulation. We investigate the relationship between a lizard’s bipedal running and its body movement with the help of a dynamic simulation. In this study, a dynamic theoretical model of lizard is assumed as a three-link consisting of an anterior and posterior bodies, and a tail, with morphometrics based on Callisaurus draconoides. When a lizard runs straight in a sta- ble bipedal gait, its pelvic rotation is periodically synchronized with its gait. This study shows that the S-shaped body undulation with the yaw motion is generated by minimizing the square of joint torque. Furthermore, we performed the biomechanical simulation to figure out the relationship between the lizard’s lateral body undulation and the bipedal running locomotion. In the biomechanical simulation, all joint torques significantly vary by the waist and tail’ motions at the same locomotion. Besides, when the waist and tail joint angles increase, the stride length and duration of the model also increase, and the stride frequency decreases at the same running speed. It means that the lizard’s undulatory body move- ments increase its stride and help it run faster. In this study, we found the benefits of the lizard’s undulatory body movement and figured out the relationship between the body move- ment and the locomotion by analyzing the dynamics. In the future works, we will analyze body movements under different environments with various simulators. 1. Introduction The running patterns of vertebrates vary depending on the species. The development of these running patterns is known to be an evolutionary result of adapting to the environment. Biolo- gists use several methods, such as comparative approaches, to study the principles of observed running patterns. In the field of biomechanics, the dynamics analysis technique is used to ver- ify the principles of running patterns by simulation. In this study, we explain the principle of PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 1 / 23 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Kim J, Kim H, Park J, Kim HS, Seo T (2020) Effects of body movement on yaw motion in bipedal running lizard by dynamic simulation. PLoS ONE 15(12): e0243798. https://doi.org/ 10.1371/journal.pone.0243798 Editor: Marc H.E. de Lussanet, University of Mu¨nster, GERMANY Received: June 1, 2020 Accepted: November 26, 2020 Published: December 31, 2020 Copyright: © 2020 Kim et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are available at https://github.com/imationzzz/ LizardSimulation. Funding: H. S. Kim was supported by the GRRC program of Gyeonggi province [GRRC KGU 2020- B02, Research on Innovative Intelligent Manufacturing Systems]. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. the running pattern of a fast bipedal running lizard. More specifically, we investigate the effect of lateral body undulation on bipedal running at high speed by simulating a theoretical mechanical model of a running lizard. A lizard can run in a straight line with a very stable posture even at a high speed. While run- ning, it has been observed that a lizard bends its body and tail in the horizontal plane, which is referred to as a lateral body undulation, in the same manner as snakes and fish. Because this movement is quite large and dynamic, we hypothesize that such a body movement plays an important role in the bipedal lizard’s locomotion. In previous studies, the lateral undulation of lizards has been observed. Ritter observed the unique S-shaped lateral body undulation of a lizard with its various running speeds [1]. Whether such a lateral undulation is a result of active movement or passive movement has also been investigated. Ritter analyzed the electromyography and kinematic data of the epaxial muscle of a running lizard and concluded that the lizard actively bends its waist to stabilize its trunk [2,3]. Similarly, by analyzing the electromyography of the epaxial muscle, Bennett et al. confirmed that at a low speed, the epaxial muscle resists the body’s torsional force, whereas at a high speed, the muscle is used to bend the waist [4]. These studies showed that the lateral body undulation is not a result of a passive external force, but of active internal moments. However, these studies did not explain the actual purpose of such a lateral body undulation or its effect on locomotion. The effect of a lizard’s tail was also studied. Jusufi et al. and Libby et al. studied the balanc- ing effect of the tail when a lizard free-falls or jumps [5,6]. Gillis et al. showed that after tail loss, a lizard could no longer control its body when it jumps [7]. Kim et al. improved the motion stability of a six-legged water-running robot in the yaw and pitch direction using a two degrees-of-freedom (DOF) active tail [8]. MSU Tailbot can control body angle for a safe land- ing with swing tail in the air [9]. Such studies have focused on the tail’s role in the lizards’ pitch and roll. On the other hand, TAYLRoACH can change its running direction as 90˚ turns at a constant rotational speed of 360˚/s using a 1 DOF active tail in the horizontal plane [10]. Sala- mandra Robotica controlled by the Central Pattern Generator (CPG) is similar to that of a real salamander in the horizontal plane [11]. These robots studied the yaw movement of the body, but the effect of the periodic lateral bending of the body during running was not investigated. The contributions of this paper are to 1) find the benefits of the lateral undulations of the lizard body during bipedal running, and 2) confirm the relationship between the lizard’s undu- latory body movement and the bipedal running locomotion. The target of this study, Calli- saurus draconoides, is a creature that can run very fast compared to its size, and the body of the lizard dynamically moves in the yaw direction. We analyze the effect of the lateral body undu- lation of a steady-state bipedal running lizard on its yaw motion via a dynamic theoretical sim- ulation. Besides, we simulate the effect of the undulatory body motion on bipedal running locomotion through a dynamic simulation using a lizard biomechanical model with the legs. The biomechanical model derived in this study has 33 DOFs, based on the morphometrics of Callisaurus draconoides. The extensive simulations verify that the undulatory body movements dominantly affects its bipedal running locomotion. 2. Analysis for the benefit of the lizard’s undulatory body movement When a lizard runs straightly with a bipedal gait, its posterior body swings symmetrically and periodically in the horizontal plane. Furthermore, the lateral undulations of a lizard generate a periodical S-shaped movement. When the lizard’s running is stable, the angle between the trunk and the tail is very close to zero [12]. It means that if the lizard ideally balances the body, PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 2 / 23 the undulatory body movement will occur only in the horizontal plane. Therefore, we assumed an ideal situation that the lizard was stably balanced, and modeled the lizard’s undulatory body movement only in the horizontal plane. In addition, we studied the relationship between the theoretical model’s running performance and the movement of the lizard body using the optimization method and assuming the data obtained from the previous studies. A. Three-link theoretical model of a lizard After observing the Callisaurus draconoides, a theoretical model comprising a three-link sys- tem with an anterior and posterior bodies, and a tail is devised as shown in Fig 1. The joints are located such that each link can rotate in the horizontal plane relative to its neighboring links. Animal muscles have both rigid and flexible properties so that when generating force actively, they are rigid but when absorbing forces like springs, they become passive and flexi- ble. This study modeled a lizard’s joint as the rotating joint to mimic the active muscle and extract joint torque. The forelegs are attached to the anterior body and the hind legs to the pos- terior body. The moment of inertia for the segment of the body is calculated by assuming the segment as homogenous cylinder. The morphometrics of Callisaurus draconoides obtained from the real measured data in [13]. For such lizards as Anolis sagrei and Anolis carolinensis, which are similar to the Callisaurus draconoides and run with a bipedal gait, Legreneur et al. found that the mass percentages of the forelegs relative to the total body mass were 2.05% and 1.8%, and those for the hind limbs were 6.2% and 4.25%, respectively [14]. Thus, the effect of the leg motion is assumed to be negligible, and the leg masses are included in the anterior and posterior bodies. The morphometrics of Callisaurus draconoides used for the three-link theo- retical model are summarized in Table 1, and the details of the derivation of the three-link the- oretical model are presented in Appendix A1. B. Kinematic analysis of a lizard’s bipedal running Fig 2 shows the forces and moments acting on each body segment of the lizard when it runs straightly with a bipedal gait. The roles of the ground reaction force (GRF), ground reaction moment (GRM), and lateral undulation of the waist and tail of a lizard must be uncovered for the analysis. When a lizard runs, its legs exert a force on the ground with every stride, and as a result, a fore-aft ground reaction force GRFx (fx) and lateral force GRFy (fy) are produced at its feet. Besides, GRFz (fz) occurs in the height direction of the lizard’s foot. GRFz is closely related to the roll and pitch motion of the lizard body. However, the body of the lizard is assumed to have relatively little movement in the roll and pitch directions compared to the yaw direction from the previous studies [12,15]. Therefore, the effect of GRFz on the lizard’s undulatory body movement was judged to be small, and the force analysis of the lizard was performed in Fig 1. Three-link theoretical model of a lizard, Callisaurus draconoides, in the horizontal plane. The moment of inertia for the segment of the body is assumed as a homogenous cylinder. Since the mass percentages of the forelegs and the hind limbs are low relative to the total body mass, the leg motion’s effect is assumed to be negligible. https://doi.org/10.1371/journal.pone.0243798.g001 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 3 / 23 the horizontal plane, excluding GRFz, as shown in Fig 2. If we consider that the right leg swings first while the left leg is in the stance stage, the GRF produces both a translational force and a moment Mf, which acts on the posterior body. In the case of the swinging of the left leg, while the right leg is in the stance stage, the lizard must bring its left leg into the appropriate position by moving it by an amount equal to the distance from the position of the left leg at the first swing, which is called the stride length. In order to implement the stride length for the left leg, the posterior body is required to rotate clockwise. Hence, a moment Mswing, which acts on the posterior body, is required to realize the second swing. The posterior body is unable to rotate by itself, which implies that Mswing is required to be produced relative to the posterior body. Because Mswing is not equal to Mf, the GRM (Mz) generated by the twisting of the foot relative to the ground and the body tor- que (Tbw, Tbt) generated by the joint movement of the bodies relative to the posterior body are used to generate Mswing. A lizard is able to run forward by repeating these steps. C. Dynamics analysis for the lizard’s undulatory body movement with optimization method The dynamics analysis of the lizard theoretical model is performed according to the process in Fig 3(a). We set the initial desired value of the lizard body theoretical model. In the theoretical Fig 2. Moments and forces periodically acting on the lizard in a period in the horizontal plane. When a lizard runs, its legs exert a force on the ground with every stride, a fore-aft ground reaction force GRFx (fx), lateral force GRFy (fy), and GRM are produced at its feet. Also, the body torques (Tbw, Tbt) are generated by the bodies’ joint movement. A lizard controls these forces, moments, and torques to run forward by repeating the gait. https://doi.org/10.1371/journal.pone.0243798.g002 Table 1. Morphometrics of Callisaurus draconoides used for the three-link theoretical model. Callisaurus draconoides Total length (mm) 192 Anterior body length (mm) 66 Posterior body length (mm) 42 Tail length (mm) 84 Anterior body width (mm) 20 Posterior body width (mm) 20 Tail width (mm) 6 Total mass (g) 10 Anterior body mass (g) (with fore leg added) 4 Posterior body mass (g) (with hind leg added) 4 Tail mass (g) 2 https://doi.org/10.1371/journal.pone.0243798.t001 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 4 / 23 simulation, torque is generated in a proportional-derivative (PD) controller to follow the body joint’s initial desired angle. We extracted the lizard joint torque function over one cycle in the simulation. Next, we have defined a function that integrates the square of this torque value over one period. As a result, jd_minimized of the body joint is derived when the function is mini- mized by changing the desired value of the lizard body joint through iteration. After both the desired waist and tail movements are generated, the theoretical simulation can be formulated as shown in Fig 3(b). This type of periodic angular motion is assumed to be a sine function in their study on the design of a machine with fish-like biomimetic locomotion in a liquid environment [16]. The waist and tail movements of a lizard can be realized via the use of a sine function similar to the previous research method. These assumptions are not per- fectly suited to mimic target animals but have the advantage of being able to reveal the effects and relationships of animal movements through simple calculations. The absolute angle of the posterior body is controlled by the GRM, but Tbw and Tbt generated by the waist and tail move- ments affect the magnitude of the GRM and the model’s running dynamics. GRM (Mz), Tbw, and Tbt are described in Fig 3(c). According to Kubo et al. in [17], the pelvic rotation angle varies with the species of lizard and ranges from 11.7˚ to 25.8˚. Reilly et al. observed the pelvic rotation to comprise a periodic sinusoidal movement [18]. The objective of our study is not to obtain a numerically perfect simulation of a running lizard, but to verify the effect of its lateral body undulation. Accord- ingly, the pelvic rotation is approximated as a sinusoidal movement of 20 sin(2πt/Tp) ˚ (Tp = 0.092 s). The magnitude of the GRF is mathematically derived while referring to related studies. GRFx, which acts in the forward-aft direction, was modeled as a sine function by Aerts et al. in [19]. The same mathematical derivation of GRFx is used in the simulation presented herein. In the case of GRFy, the direction of the force is always towards the midline of the body. The maximum value of GRFy is close to that of GRFx, whereas the impulse is much larger [20]. For the simulation, the direction, relative magnitude, and impact of GRFy are modeled according to the studies in [20,21], and mathematically formulated as a sine function. The GRF depends on the running speed of the lizard model, and thus, in the steady-state running, the GRF is periodically repeated in every stride. Therefore, the GRF was approximated as a sine function similar to the previous studies, as shown in Fig 4. The position of the foot relative to the posterior body for every stride is determined from the leg length and angle data as presented in [13]. The location of the right foot is initially set as 38 mm away and at −45˚ from the center of mass of the posterior body, while the left foot is 38 mm away and at +45˚. Without the loss of generality, the feet are modeled in a no-slip state Fig 3. (a) Overall architecture of the dynamics analysis for benefits of the undulatory body movement, (b) simulation for one cycle of the gait of the lizard model with angular body movement (waist, tail, and leg controller are a proportional- derivative (PD) controller) and (c) forces and moments in the lizard model. https://doi.org/10.1371/journal.pone.0243798.g003 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 5 / 23 during contact with the ground. The foot position is fixed, and the center of mass moves rela- tive to the feet according to the GRF. The movement of the waist and tail is minimized by the optimization, as shown in Fig 3(a). The desired angular movements of the waist and tail, which are assumed to be represented by sine functions, can be formulated as shown in Eq (1). Jd;i¼1;2 ¼ ai  sinð2p TP t þ biÞ ð1Þ The angular body movement is achieved by combining the amplitude (ai) and phase differ- ence (bi). As this study focuses on the effect of the body motion at high velocities, the fre- quency during running is fixed The angular movements of the waist and tail vary depending on the constants ai and bi in Eq (1). The values of the ai and bi determine the sinusoidal profile of the desired angular movement over time; then, the PD controller generates the waist and tail joints torques Tbw and Tbt to follow the desired profiles, as shown in Fig 3(b). Therefore, we determined ai and bi as the design parameters for the optimization. Also, we derived the objective function with the appropriate assumptions, as shown in Eq (2). Minimize FðTbwðtÞ; TbtðtÞÞ ¼ Ð Tp 0 ½T2 bwðtÞ þ T2 btðtފdt ð2Þ Subject to GRM  Const1 and q2 ¼ 20  sin 2p Tp t   . The objective function to be minimized is set as (Tbw 2 + Tbt 2). The term (Tbw 2 + Tbt 2) can be expanded as ((Tbw + Tbt)2 + (Tbw − Tbt)2)/2, and thus, minimizing ðT2 bw þ Tbt 2Þ implies that the magnitude (Tbw + Tbt) is minimized along with the uniform distribution of the moments to the joints via the term (Tbw − Tbt). The moment generated at either the waist or tail joint acts as a burden on the active joint because it has to generate a relatively large moment. The uni- form distribution of the moments over the joints reduces the load, and because a real lizard is presumed to act in the same manner, (Tbw 2 + Tbt 2) is selected as the objective function. GRM (Mz) is a constant that can be varied to produce different outcomes. Previously, we assume that the GRM (Mz) produced by the lizard is very small. Thus, in the optimization pro- cess, the GRM (Mz) is set to a minimum value. Here, q2 is the required absolute angle for the posterior body when the lizard is running at 4 m/s, as determined via a simple calculation Fig 4. Profiles of (a) GRFx and (b) GRFy approximated as a sine function similar to the previous studies [20,21]. https://doi.org/10.1371/journal.pone.0243798.g004 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 6 / 23 comprising the stride length and time. We performed the optimization by sequential quadratic programming (SQP) using fmincon in MATLAB1. D. Result of the dynamic analysis for a lateral undulation of a bipedal running lizard For various lizard theoretical models of different lengths and mass ratios, their lateral body undulations are generated in the simulations. We chose the size and weight of the lizard model based on the real lizard. It is worthwhile to note that the optimized motions may be affected by the size and weight of lizard model. Therefore, in this study, lizard models of different sizes and weights were used for simulations. The dynamic theoretical simulation is implemented in MATLAB1 with the numerical integration by the fourth-order Runge–Kutta method. The ini- tial velocity of the model is set as V0 = 4 m/s. The model movement resulting from the optimi- zation is presented in Fig 5. As expected, to minimize ðT2 bw þ Tbt 2Þ, the waist and tail joints are required to generate moments that oppose each other, which results in a natural S-shaped movement of the body. We have obtained the joint angles of the actual lizard’s waist and tail by capturing the images from a video file related with its running motion, as shown in Fig 6(a) from [22]. Then, we normalized them for comparison with the simulation. Since the shape of the lizard’s undu- latory body movement is determined by the phases of the waist and tail movement angles, we compared the phase difference between the actual lizard’s and the simulated waist and tail joint angles as shown in Fig 6(b). The phase difference of the waist was 0.002T (T is the period), and the phase difference of the tail was 0.067T, indicating that the dynamics theoreti- cal model and the real lizard’s waist and tail movements were very similar. In conclusion, we found that real lizards can minimize the sum of squares of torque by moving their body in an S-shape. 3. Lizard biomechanical modeling: For relationship between the undulatory body movement and the locomotion In the previous section, we observe that the body movement of the lizard’s theoretical model when the joint torques are minimized. In addition, we have studied the relationship between Fig 5. Movements of the lizard models calculated by the dynamics analysis when the joint torques are minimized. ( indicates the point at which the foot contacts the ground). (a) l1 = 66 mm, l2 = 42 mm, l3 = 84 mm, m1 = 4.0 g, m2 = 4.0 g, m3 = 2.0 g, (b) l1 = 60 mm, l2 = 40 mm, l3 = 92 mm, m1 = 4.0 g, m2 = 4.0 g, m3 = 2.0 g, (c) l1 = 72 mm, l2 = 42 mm, l3 = 78 mm, m1 = 4.0 g, m2 = 4.0 g, m3 = 2.0 g, (d) l1 = 66 mm, l2 = 42 mm, l3 = 84 mm, m1 = 4.2 g, m2 = 4.0 g, m3 = 1.8 g, (e) l1 = 66 mm, l2 = 42 mm, l3 = 84 mm, m1 = 3.8 g, m2 = 4.0 g, m3 = 2.2 g. https://doi.org/10.1371/journal.pone.0243798.g005 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 7 / 23 the body movement and the joint torques of the theoretical model. In this section, to confirm the relationship between the lizard’s undulatory body movement and bipedal running locomo- tion, we established the lizard biomechanical model including its legs. To clearly identify the effect of the body movement, we have constructed a biomechanical simulation to generate the motion of the biomechanical model without control and to extract the joint torque through inverse dynamics, to exclude the effect of control performance. A. Bipedal running lizard biomechanical model The biomechanical model of a bipedal running lizard is established by morphometrics of the Callisaurus draconoides based on [13], as in the previous section. The kinematics model is pre- sented in Fig 7. The body of the model consists of three links referred to as body1, body2, and body3. In addition, the fore and hind legs of the biomechanical model comprise three links. Fig 6. (a) Captured image of the running lizard for the joint angles of waist and tail from [22] and (b) comparison of the phase differences of the waist and tail joint angles between the real lizard and the dynamic model. https://doi.org/10.1371/journal.pone.0243798.g006 Fig 7. Kinematic biomechanical model of the lizard to confirm the relationship between the undulatory body movement and bipedal running locomotion: (a) bodies, (b) joints, and (c) lengths. The number of links used in the model is fifteen, and the number of joints is 33. https://doi.org/10.1371/journal.pone.0243798.g007 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 8 / 23 The total number of the links used in the kinematics model of the lizard is fifteen, as shown in Fig 7(a). The joints used for describing the angles of each body are set as rotational joints with one DOF. The fore and hind legs comprise shoulder, knee, and ankle joints. The shoulder and ankle joints are spherical joints with three DOFs. The knee joint is a rotational joint with one DOF. We add two linear joints as virtual joints to calculate the position of the biomechanical model in the fixed coordinate frame. The number of joints in the kinematics model is 33, as shown in Fig 7(b). The length and mass of the links of the biomechanical model, based on [13], are listed in Table 2. The center of the mass of each link is in the middle of the link. The positions of the center of the mass are explained in the Appendix A2 using the lengths and the joint angles of the liz- ard biomechanical model. B. Joint angles of the biomechanical model for the bipedal running locomotion We simulate the locomotion and joint movement of the lizard’s biomechanical model. In this simulation, the joint angles of the robot model are simplified as a sine function according to the actual lizard’s joint angle. The sine functions of the joint motion are defined as follows: qi ¼ ai  sin 2p period t þ bi   þ ci; i ¼ 3;


; 33 ð3Þ where ai is the magnitude, bi is the phase difference, and ci is the offset of the sine function. As we change these variables, the joint angle of the model changes dramatically. The movement of the biomechanical model is laterally symmetric because it is assumed that the model runs in a straight line without any deviation. To realize symmetric movement in the lateral direction, the body joint angle has two variables of magnitude ai and phase differ- ence bi without an offset ci. In addition, the left and right hind legs perform the same symmet- ric movement with a 180˚ phase difference. The joints of the forelegs are fixed at a constant angle to mimic a real lizard’s posture. We summarized the motion of each joint in Table 3. The ankle joint of the lizard model is set with the foot always facing forward. This setting is determined by observing that the foot of a real lizard faces forward when the lizard’s foot touches on the ground. Therefore, we set the ankle joint angle to x, z = 0 and fix the y-axis value in the fixed coordinate frame, such that lizard’s foot is facing forward when it touches the ground at any time. Table 2. Size and mass of a lizard biomechanical model. Lizard model links Length (mm) Width (mm) Mass (g) Body1 66 16.5 3.6 Body2 42 13 2.9 Body3 84 4.5 2.0 Body4, 7 16 4 0.11 Body5, 8 12 3 0.06 Body6, 9 15 3 0.03 Body10, 13 19 7.5 0.36 Body11, 14 21 4.5 0.18 Body12, 15 13 3.5 0.08 https://doi.org/10.1371/journal.pone.0243798.t002 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 9 / 23 In the biomechanical simulation, it is assumed that the body of the model is maintained at a constant distance from the ground. The lizard’s body is observed to have a small up and down movement [12,15,23]. Thus, in order to simplify the simulation and focus on the lateral body undulation of the biomechanical model, we assume that the body of the biomechanical model only moves in the horizontal plane. The distance between the body of the biomechanical model and the ground is determined based on the trajectory of the hind legs. Based on the body position, the z-direction position of the ground should be between the minimum of the foot end and the minimum of the hind ankle. If the ground position is lower than the minimum position of the foot end of the hind foot, the hind foot of the model cannot reach the ground. On the other hand, if the ground position is higher than the minimum position of the hind ankle, then the model’s ankle touches the ground, which is different from the actual lizard’s movement. Therefore, in this study, we set the ground position of the body from the center of the foot at the minimum foot position in the z-direction. Before the hind foot touches the ground, the posture of the foot is fixed, as described in the previous section and as shown in Fig 8(a). When the position of the foot end of the hind foot is lower than the position of the ground, the angle of the ankle is changed such that the position of the foot end is the same as the position of the ground, as shown in Fig 8(b). If the position of the foot end of the back foot is higher than the position of the ground, the hind foot returns to its original posture as shown in Fig 8(c). Therefore, when the foot touches the ground, the position of the foot is fixed, and after the foot leaves the ground, it returns to its original pos- ture. While in contact with the ground, the movement of the model is implemented based on the fixed position of the foot end of the hind leg. When the model is in the air without contacting the ground, we assume that the center of the mass of the biomechanical model moves linearly. Since the biomechanical model is not Table 3. Joint motion of the lizard biomechanical model. Joint’s name Symbol Joint motion Waist, tail joint q4, q5 qi ¼ ai  sin 2p period t þ bi   (4) Shoulder X, Z axis & Knee of Left hind leg q20, q21 q22 qi ¼ ai  sin 2p period t þ bi   þ ci (5) Shoulder X, Z axis & Knee of Right hind leg q27, q28 q29 qi ¼ aiforced in the air, the velocity of the model does not change. The velocity and the direction of the model in the air are calculated from the previous velocity and direction before falling off the ground. C. Inverse dynamics of the lizard biomechanical model The number of joints in the biomechanical model is k = 30 and with three virtual joints, the total number of joints is n = k+3 = 33. When the biomechanical model does not touch the ground as shown in Fig 9(a), the equation of motion is derived as follows. AðqÞ€q þ bðq; _qÞ þ gðqÞ ¼ G ð10Þ where q 2 R33×1 is the joint vector and Γ 2 R33×1 is the torque vector. A(q) 2 R33×33 is the mass and inertia matrix and bðq; _qÞ 2 R331 is the Coriolis and centrifugal vector.g(q) 2 R33×1 is the gravity vector. When the foot of the lizard biomechanical model contacts the ground, as shown in Fig 9(b), the reaction forces and moments are taken into consideration in the equa- tion of motion from the ground as follows. AðqÞ€q þ bðq; _qÞ þ gðqÞ þ JT c fc ¼ G ð11Þ JT c is the transpose of the Jacobian of the position and the angle at the foot on the ground. In addition, fc is the reaction force and moment vector.Jc must satisfy Eq (12). _xc ¼ Jc _q ð12Þ where xc is the position vector of the foot that contacts the ground and q is the joint vector. Then, by replacing €q with €q ¼ Jwhere, Lc ¼ ðJcAwith a 180˚ phase difference. The phase difference of the waist joint angle is different from that of the tail joint angle. In Fig 11, it appears that the waist joint moves first, followed by the tail joint. For the readers, the MATLAB code used for the simulation in Fig 11 can be accessed with the instruction file (https://github.com/imationzzz/LizardSimulation). Fig 12 shows the trajectory of the left hind tiptoe of the lizard biomechanical model running at 4 m/s in place and compares it with that of the actual lizard captured from its running loco- motion [22]. As shown in Fig 12(a) and 12(b), in the stance mode, both the lizard biomechani- cal model and the real lizard move their tiptoes back in a straight line but bring them forward in the swing mode. The trajectory length of the tiptoe of the lizard biomechanical model in the top view is 58.5 mm in the horizontal direction, and its maximum width is 25.8 mm in the ver- tical direction. On the other hand, the trajectory length of the real lizard in the top view is 199 mm in the horizontal direction, and its maximum width is 49 mm in the vertical direction. As shown in Fig 12(c) and 12(d), both the lizard biomechanical model and the real lizard raise Fig 10. Squares of the torques of all joints obtained by varying the body movement. The magnitude of the waist joint motion changes in the vertical direction, and the magnitude of the tail joint motion changes in the horizontal direction. https://doi.org/10.1371/journal.pone.0243798.g010 Table 4. Design parameters for optimization. Order Description Variable 1 Magnitude of the waist joint angle a4 2 Phase difference of the waist joint angle b4 3 Magnitude of the tail joint angle a5 4 Phase difference of the tail joint angle b5 5 Magnitude of the shoulder z joint angle a20 6 Offset of the shoulder z joint angle b20 7 Magnitude of the shoulder x joint angle a21 8 Magnitude of the shoulder y joint angle a22 9 Magnitude of the knee joint angle a23 https://doi.org/10.1371/journal.pone.0243798.t004 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 13 / 23 their tiptoes upward in swing mode. The trajectory length of the lizard biomechanical model in the side view is 58.5 mm in the horizontal direction, and its maximum height is 8.7 mm in the vertical direction. The trajectory length of the real lizard in the side view is 199 mm in the horizontal direction, and its maximum height is 56 mm in the vertical direction. Through this simulation, we found that when the lizard biomechanical model ran in the bipedal locomotion with minimizing the sum of squares of the torque, the resulting trajectory of tiptoe is similar to that of the actual lizard even though their sizes are different. We performed a simulation to confirm the bipedal running locomotion of the lizard bio- mechanical model by gradually increasing the size of the waist and tail joint angle, as shown in Fig 13(a). The body movement is a sine function, and the waist and tail joints are set to be same. Excluding the waist and tail joints, the legs’ movements were derived to minimize the square of the torque, as in the previous simulation. As a result, the characteristics of locomo- tion were obtained, as shown in Fig 13(b)–13(d). When the lizard biomechanical model runs at the same speed, Fig 13(b) shows that the stride length increases from 97.8 mm to 135.2 mm as the body movement increases from 2 degrees to 40 degrees. The stride duration also Fig 11. Snapshot of a moving video clip of a simulated lizard when the square of the joint torque is minimized. The lizard biomechanical model has bipedal running locomotion at a speed of 4 m/s. https://doi.org/10.1371/journal.pone.0243798.g011 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 14 / 23 increases from 24.3 ms to 34.0 ms (see Fig 13(c)). The stride frequency is the reciprocal num- ber of the stride duration and decreases from 41.2Hz to 29.4Hz. Therefore, as the body move- ment increases, the stride length increases, so the lizard can achieve a high running speed with fewer moves. On the other hand, Fig 13(d) shows that the duty factor of the lizard maintains its value, about 25.7%, independent of the body’s movement. When the joint angle of the lizard’s waist and the tail are 40 degrees, the characteristics of simulated locomotion are compared with those of actual lizard obtained from [13] in Table 5. The stride length of the lizard biomechanical model is smaller than that of the real lizard. It is guessed that the difference between the simulation model and the real lizard may stem from the modeling error. On the other hand, since other values are similar between the biomechani- cal model and the lizard, the modeling error may be negligible. 5. Conclusion This study shows that the steady-state bipedal straight running of a lizard is highly related to the symmetric and periodical angular movement of its posterior body. It is hypothesized that such pelvic rotation is produced owing to the GRM and internal torque. The dynamic theoreti- cal simulation is used to discover the relationship between the GRM and internal torque. In Fig 12. Comparison of the tiptoe’s trajectory between the dynamic biomechanical model and the real lizard. (a) The trajectory of the tiptoe of the lizard biomechanical model in the top view has a length of 58.5 mm in the horizontal direction, and 25.8 mm in the vertical direction. (b) The trajectory of the tiptoe of the real lizard in the top view has a length of 199 mm in the horizontal direction, and 49 mm in the vertical direction. (c) The trajectory of the tiptoe of the lizard biomechanical model in the side view has a length of 58.5 mm in the horizontal direction, and 8.7 mm in the vertical direction. (b) The trajectory of the tiptoe of the real lizard in the side view has a length of 199 mm in the horizontal direction, and 56 mm in the vertical direction. https://doi.org/10.1371/journal.pone.0243798.g012 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 15 / 23 the lizard theoretical model of bipedal and steady-state running, the GRF associated with the running speed is assumed to be identical for every stride based on previous studies. When the GRM, torque distribution, and torque magnitude are minimized, an S-shaped lateral body undulation similar to that of a real lizard is observed in the model. Furthermore, we established the lizard biomechanical model to figure out the relationship between the lizard’s undulatory body movement and the bipedal running locomotion. The biomechanical model consists of 15 bodies and 33 DOFs. When the biomechanical model per- forms running in bipedal locomotion, the square of the joint torque is changed by the waist and tail joints. It means that the lizard’s undulatory body movement significantly affects all joint torques when the lizard performs bipedal running locomotion. When the square of the joint torque was minimized, we observed that the undulatory body movement are quite similar to those of a lizard. The trajectory of the lizard’s biomechanical model tiptoe is small than a real lizard’s trajectory; however, the trajectory shape is similar to a real lizard’s tiptoe trajec- tory. We performed a simulation to confirm the bipedal running locomotion of the lizard bio- mechanical model by gradually increasing the size of the waist and tail joint angle. As a result, when the lizard biomechanical model runs at the same speed, the stride length and stride Fig 13. The locomotion characteristics of the lizard biomechanical model changed by increasing the undulatory body movement. (a) Concept of the increase of the lizard’s undulatory body movement. The waist and tail joint angles are set to the same angle. (b) The stride length and (c) duration are increased by increasing the lizard’s undulatory body movement. (c) The duty factor does not be changed by the undulatory body movement of the lizard biomechanical model. https://doi.org/10.1371/journal.pone.0243798.g013 Table 5. Comparison of locomotion specification between the biomechanical model and the real lizard. Variable Lizard model with 40˚ of body joint Callisaurus draconoides [13] Speed (m/s) 4.0 4.0±0.1 Stride length (mm) 135.2 319±11 Stride width (mm) 40.9 51±4 Stride duration (ms) 34.0 80±3 Duty factor (%) 25.7 24±1 Hip height (mm) 30.3 28±1 https://doi.org/10.1371/journal.pone.0243798.t005 PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 16 / 23 duration increase, and the stride frequency decrease. Therefore, as the body movement increases, the lizard can achieve a high running speed with fewer moves. It means that the liz- ard’s undulatory body movements increase its stride and help it run faster. As a result, we identify the benefits of the lizard-body movements that minimize the square of the joint torque. Besides, we confirmed that the undulatory body movements of lizard affect the lizard’s locomotion. To the best of our knowledge, there is no research to figure out the principle of the running lizard’s undulatory body movement and relationship with the loco- motion. In future work, we will analyze body movements in various ways through various environments and simulators. 6. Appendix: A1 As shown in Fig 14, for dynamic equation, the vectors for position, angle, mass and moment of inertia are defined in R3×1. The horizontal position vector of center of mass of the link is defined as x = [x1, x2, x3]T and the lateral position vector is defined as y = [y1, y2, y3]T. The angle vector of the link in global frame is q = [q1, q2, q3]T, the joint angle vector between each links is j = [j1, j2]T. The mass vector is defined as m = [m1, m2, m3]T. For numerical calculation, M = diag(m) is defined. The moment of the inertia vector is defined by i = [i1, i2, i3]T and in the same manner, I = diag(i). The sine and cosine vectors are defined as sinq = [sinq1, sinq2, sinq3]T, Sinq = diag(sinq), cosq = [cosq1, cosq2, cosq3]T cosq = diag(cosq). The model has a boundary as shown in Fig 14 ([24,25]) xiþ1 C and L are defined in Eqs (26) and (27). C ¼ From the assumption that the reaction force exists only in hind leg, the vectors of the reac- tion force are Fext,x = [0, fext,x, 0]T, Fext,v = [0, fext,v, 0]T. The inner force vector are defined as Fx = [fx1, fx2]T and Fv = [fv1, fv2]T. Eqs (32) and (33) are derived by differentiating Eqs (24) and (25) twice: C€x ¼ X4 ¼ q1 q2 0 2 64 3 75 þ Rot q3; z ð Þ  l2 2 þ shoulderX X10 ¼ q1 q2 0 2 6664 3 7775 þ Rot q3; z ð Þ  l1 0 0 2 6664 3 7775 þ Rot q3; z ð Þ  Rot q4; z ð Þ  l2 2 þ pelvicX X15 ¼ X14 þ Rot q3; z ð Þ  Rot q4; z ð Þ  Rot q27; z ð Þ  Rot q28; x ð Þ  Rot q29; y ð Þ  Rot q30; z ð Þ  l14 2 0 0 2 66664 3 77775 þRot q31; z ð Þ  Rot q32; x ð Þ  Rot q33; y ð Þ  l15 2 0 0 2 66664 3 77775 ð51Þ where Rot(qi, axis) means that {ith} joint rotates in the axis direction. ShoulderX and shoulderY are the distances from the center of the mass of body1. PelvicX and pelvicY are the distances in the x and y directions from the center of the mass of body2. Supporting information S1 File. (TXT) S1 Video. (WMV) Author Contributions Investigation: Jeongryul Kim, Hongmin Kim, Jaeheung Park, TaeWon Seo. Supervision: Hwa Soo Kim. Writing – original draft: Jeongryul Kim. Writing – review & editing: Jaeheung Park, Hwa Soo Kim, TaeWon Seo. References 1. RITTER D, Ritter R. Lateral bending during lizard locomotion. J Exp Biol. 1992; 173: 1–10. 2. Ritter D. Epaxial muscle function during locomotion in a lizard (Varanus salvator) and the proposal of a key innovation in the vertebrate axial musculoskeletal system. J Exp Biol. 1995; 198: 2477–2490. PMID: 9320404 3. Ritter D. Axial muscle function during lizard locomotion. J Exp Biol. 1996; 199: 2499–2510. PMID: 9320426 4. Bennett WO, Simons RS, Brainerd EL. Twisting and bending: the functional role of salamander lateral hypaxial musculature during locomotion. J Exp Biol. 2001; 204: 1979–1989. PMID: 11441039 5. Jusufi A, Goldman DI, Revzen S, Full RJ. Active tails enhance arboreal acrobatics in geckos. Proc Natl Acad Sci. 2008; 105: 4215–4219. https://doi.org/10.1073/pnas.0711944105 PMID: 18347344 6. Libby T, Moore TY, Chang-Siu E, Li D, Cohen DJ, Jusufi A, et al. Tail-assisted pitch control in lizards, robots and dinosaurs. Nature. 2012; 481: 181–184. https://doi.org/10.1038/nature10710 PMID: 22217942 7. Gillis GB, Bonvini LA, Irschick DJ. Losing stability: tail loss and jumping in the arboreal lizard Anolis car- olinensis. J Exp Biol. 2009; 212: 604–609. https://doi.org/10.1242/jeb.024349 PMID: 19218510 8. Kim H, Sitti M, Seo T. Tail-Assisted Mobility and Stability Enhancement in Yaw and Pitch Motions of a Water-Running Robot. IEEE/ASME Trans Mechatronics. 2017; 22: 1207–1217. 9. Zhao J, Zhao T, Xi N, Mutka MW, Xiao L. MSU tailbot: Controlling aerial maneuver of a miniature-tailed jumping robot. IEEE/ASME Trans Mechatronics. 2015; 20: 2903–2914. PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 22 / 23 10. Kohut NJ, Pullin AO, Haldane DW, Zarrouk D, Fearing RS. Precise dynamic turning of a 10 cm legged robot on a low friction surface using a tail. 2013 IEEE International Conference on Robotics and Auto- mation. IEEE; 2013. pp. 3299–3306. 11. Ijspeert AJ, Crespi A, Ryczko D, Cabelguen J-M. From swimming to walking with a salamander robot driven by a spinal cord model. Science (80-). 2007; 315: 1416–1420. https://doi.org/10.1126/science. 1138353 PMID: 17347441 12. Irschick DJ, Jayne BC. Effects of incline on speed, acceleration, body posture and hindlimb kinematics in two species of lizard Callisaurus draconoides and Uma scoparia. J Exp Biol. 1998; 201: 273–287. PMID: 9405318 13. Irschick DJ, Jayne BC. Comparative three-dimensional kinematics of the hindlimb for high-speed bipedal and quadrupedal locomotion of lizards. J Exp Biol. 1999; 202: 1047–1065. PMID: 10101105 14. Legreneur P, Homberger DG, Bels V. Assessment of the mass, length, center of mass, and principal moment of inertia of body segments in adult males of the brown anole (Anolis sagrei) and green, or Carolina, anole (Anolis carolinensis). J Morphol. 2012; 273: 765–775. https://doi.org/10.1002/jmor. 20022 PMID: 22461036 15. Druelle F, Goyens J, Vasilopoulou-Kampitsi M, Aerts P. Compliant legs enable lizards to maintain high running speeds on complex terrains. J Exp Biol. 2019;222. https://doi.org/10.1242/jeb.195511 PMID: 30796100 16. y Alvarado PV, Youcef-Toumi K. Performance of machines with flexible bodies designed for biomimetic locomotion in liquid environments. Proceedings of the 2005 IEEE International Conference on Robotics and Automation. IEEE; 2005. pp. 3324–3329. 17. Kubo T, Ozaki M. Does pace angulation correlate with limb posture? Palaeogeogr Palaeoclimatol Palaeoecol. 2009; 275: 54–58. https://doi.org/10.1016/j.palaeo.2009.02.001 18. Reilly S, Delancey M. Sprawling locomotion in the lizard Sceloporus clarkii: quantitative kinematics of a walking trot. J Exp Biol. 1997; 200: 753–765. PMID: 9318518 19. Aerts P, Van Damme R, D’Aouˆt K, Van Hooydonck B. Bipedalism in lizards: whole–body modelling reveals a possible spandrel. Philos Trans R Soc London Ser B Biol Sci. 2003; 358: 1525–1533. https:// doi.org/10.1098/rstb.2003.1342 PMID: 14561343 20. McElroy EJ, Wilson R, Biknevicius AR, Reilly SM. A comparative study of single-leg ground reaction forces in running lizards. J Exp Biol. 2014; 217: 735–742. https://doi.org/10.1242/jeb.095620 PMID: 24198262 21. McElroy EJ, Hickey KL, Reilly SM. The correlated evolution of biomechanics, gait and foraging mode in lizards. J Exp Biol. 2008; 211: 1029–1040. https://doi.org/10.1242/jeb.015503 PMID: 18344476 22. https://www.youtube.com/watch?v=ExyMxKDxT9M. 23. Fieler CL, Jayne BC. Effects of speed on the hindlimb kinematics of the lizard Dipsosaurus dorsalis. J Exp Biol. 1998; 201: 609–622. PMID: 9438835 24. Liljeback P, Pettersen KY, Stavdahl O. Modelling and control of obstacle-aided snake robot locomotion based on jam resolution. 2009 IEEE International Conference on Robotics and Automation. IEEE; 2009. pp. 3807–3814. 25. Liljeback P, Pettersen KY, Stavdahl Ø, Gravdahl JT. Controllability and stability analysis of planar snake robot locomotion. IEEE Trans Automat Contr. 2010; 56: 1365–1380. PLOS ONE Effects of body movement on yaw motion in bipedal running lizard PLOS ONE | https://doi.org/10.1371/journal.pone.0243798 December 31, 2020 23 / 23

Effects of body movement on yaw motion in bipedal running lizard by dynamic simulation.

12-31-2020

Kim, Jeongryul,Kim, Hongmin,Park, Jaeheung,Kim, Hwa Soo,Seo, TaeWon

eng

PMC10256252

Review began 05/05/2023 Review ended 05/06/2023 Published 05/10/2023 © Copyright 2023 Mittal et al. This is an open access article distributed under the terms of the Creative Commons Attribution License CC-BY 4.0., which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Divulging the Impetus of Yoga on Cardiorespiratory Fitness and Its Persona in Alleviating Anxiety Experienced by Youth: A Cohort Interventional Study Gaurav Mittal , Ruchi Kothari , Akshay Yadav , Pradeep Bokariya , Prashanth A 1. Physiology, Mahatma Gandhi Institute of Medical Sciences, Wardha, IND 2. Anaesthesiology, Lokmanya Tilak Municipal Medical College and General Hospital, Mumbai, IND 3. Anatomy, Mahatma Gandhi Institute of Medical Sciences, Wardha, IND Corresponding author: Ruchi Kothari, ruchi@mgims.ac.in Abstract Background: Globalization endangers youngsters worldwide with new standards and possibilities. Hereat of being exposed to greater demands and expectations, when it comes to performance review, their life may become more distressed. Yoga with revolutionary methods may assist youngsters in bettering their physical health regarding their maximal oxygen uptake, and also help manage their anxiety. This study ascertains the effect of yoga on youth's anxiety levels and cardio-respiratory fitness. Methods: It was a longitudinal interventional study recruiting 99 medical students wherein VO2 max (maximal oxygen uptake) on the treadmill/ergometer exercise and anxiety scores through Spielberger's anxiety scale was assessed at baseline and evaluated after 6 months of a regular yogic regime. The VO2 max was recorded by the metabolic module of Labchart software (Bella Vista, New South Wales, Australia). Findings: The VO2 max evaluated by incremental exercise to volitional fatigue was found to be 2.64 ± 0.49 L/min in males and 1.51 ± 0.44 L/min in females pre-yoga and 2.81 ± 0.52 L/min in males and 1.69 ± 0.47 L/min in females post yoga. The difference in the endline and baseline VO2 max values of yoga-performing males (t=6.595, p<0.001) and females (t = 2.478, p = 0.017) was found to be significantly higher than non- yoga performers. The METS value obtained in males was 11.96 and in females was 7.68 before yoga. Post- yoga values were 13.44 and 8.37, respectively. The difference in total anxiety scores post-intervention was 34.6 which was statistically significant (t= 4.959, p <0.001). Conclusion: From the viewpoint of a physiologist, higher VO2 max in young adults links to better physical fitness which is the potential outcome of regular yogic practice. As a result of regular yogic practice, initial soaring anxiety levels of subjects culminated in a drastic observable reduction in anxiety, which helped inculcate a judicious acumen in youngsters. Categories: Physical Medicine & Rehabilitation, Psychiatry, Integrative/Complementary Medicine Keywords: medical student training, psychiatry and mental health, spielberger’s scale, anxiety score, vo2 max, cardiorespiratory fitness, yoga research Introduction This article was previously presented as an oral presentation at the 2022 Asia-Singapore Conference on Sports Sciences on December 6, 2022. Youth are considered the blossoming buds of progression for any dwelling society, population, or even a nation as a whole. The fast-flourishing metamorphosing world of cut-throat competition is boggling with stress. It is being manifested as the most common problem among the modern generation. On entering into professional colleges, students find themselves in a new, challenging, and stressful environment. Particularly, it has been observed in medical students that they experience significant stress during their course [1-2]. Numerous factors that contribute to soaring levels of stress in medical students could be a highly competitive curriculum, intense academic competition, and excessive demands on coping abilities in physical, emotional, intellectual, financial, and social terms. As per literature available from the West [3-5], and from Asia [6] it has been documented that medical training is highly stressful, particularly for those who are in the cradle stages of their medical schooling. It is well known that stress modifies the autonomic nervous system's fineness, with sympathetic activity predominating in anxious temperament. Yogic praxis has benefitted in maintaining a physiological milieu pertaining to cardiovascular indices [7]. Yogic relaxation can moderate sympathetic preponderance [8-9]. By 1 1 2 3 1 Open Access Original Article DOI: 10.7759/cureus.38847 How to cite this article Mittal G, Kothari R, Yadav A, et al. (May 10, 2023) Divulging the Impetus of Yoga on Cardiorespiratory Fitness and Its Persona in Alleviating Anxiety Experienced by Youth: A Cohort Interventional Study. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 minimizing sympathetic activity, yoga asanas, and pranayama can tip the autonomic equilibrium in favor of relative parasympathetic control [9-11]. The objective manifestations of anxiety -- a racing heart, palpitations, sweating, elevated blood pressure, dry mouth, avoidance behavior, signs of restlessness, and heightened responsiveness reduce and eventually vanish. Medical school comes in like a soft breeze after the post-entrance inertia of rest, holidays, and fun in the life of medical students who are no longer just medical aspirants. In no time this ‘soft’ breeze gets transformed into a storm and forsooth wreaks havoc on their mental and physical well-being. Here the age-old praxis comes to their rescue anon, which is none other than yogic practice. Yoga in today's so-called “sophisticated cosmos” is described as an antediluvian practice, especially among the youth who are at crossroads. In the age of mobile phones, beepers, and 24 x 7 browsing, the yogic practice seems a more pertinent way out. Yoga practitioners have asserted its effect on balancing emotional, physical, and spiritual health for decades, but only recently has there been a move to substantiate these claims through research [12]. A state of mental tranquility is achieved by the practice of yoga as revealed by an increase in alpha waves of electroencephalogram after yoga [13-14]. At the physical level, consistent practice of asanas and pranayama confers a proportionate, flexible, typically relaxed body with an ability to combat stress efficiently [15]. Recent research has made some assertions that yoga has revolutionary methods for assisting youngsters in bettering their physical health as measured by their VO₂ max, which could also impact how they manage their anxiety. According to MI [16], the maximal oxygen uptake, i.e. VO₂ max is the single best measure of cardiorespiratory efficiency and the gold standard of physical fitness for any individual. Keeping in mind the above-stated facts, it was thought pertinent to probe an answer to the arising question as to whether long- term yoga practice could prove to be a boon for the young generation of today's world with intense academic aspirations. This study aimed at assessing the effect of a regular yogic regime on youth's degree of anxiety and cardio- respiratory fitness in terms of aerobic capacity as assessed in the Sports Physiology Laboratory. Materials And Methods Study design and setting It was a cohort interventional study with pre- and post-design. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines were used for reporting and preparing the manuscript. The study was carried out in the Sports Physiology Laboratory of a rural medical college in central India. It was undertaken where the parameters of cardiorespiratory fitness in subjects were evaluated first at baseline and later after completion of a yogic regime. During the first month, all the recruited subjects practiced yoga together under the leadership of the investigator and a trained yoga expert. Then they were advised an hour of daily yoga for a duration of 6 months, and the same subjects were re- evaluated. Similarly, baseline assessments of anxiety levels were made (pre-yoga). Subsequently, the anxiety scores were assessed following a month of yoga then there was a six-month (post-yoga) follow-up period during which the students engaged in independent praxis. We obtained signed written informed consent from all study participants. Prior approval from the Institutional Ethics Committee was ensured before the beginning of the study. Study population and selection criteria Students pursuing Bachelor of Medicine, Bachelor of Surgery (MBBS) through a rural medical college were recruited for the study. Initially, students were explained about the study and how it can benefit them in the longer run. The students who volunteered and were willing to participate in the study were shortlisted. The research was then undertaken accordingly. The sample size was estimated using OpenEpi 3.01 statistical software (Centers for Disease Control and Prevention - CDC, Atlanta) with the assumptions as the confidence level of 95%, alpha of 0.05, and power of study as 80%. The minimum sample size came out to be 84 according to the statistical software. During scrutinizing, 300 MBBS students were screened. As per the inclusion and the exclusion categorical imperative, only 99 students were considered for the study which was well over the calculated sample size. Considering a 10% shift of mindset of the students for not participating in the study, all 99 were recruited for the baseline assessments. The MBBS students in the age range of 17-25 years who gave written informed consent were included in the study. Subjects should not be involved in heavy physical activity or sports for at least a year. They should not have any already ongoing yogic regime which might already have an effect on the baseline values. Subjects suffering from any acute illness, recent surgery, endocrine disorders, cardiovascular disorders, COPD/asthma, chronic debilitating diseases such as cardiac arrhythmias, diabetes, persons receiving any drug that may affect the autonomic reflexes; not giving consent, and not willing to participate were excluded. 2023 Mittal et al. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 2 of 10 The flow diagram in Figure 1 shows the final recruitment of participants after fulfilling the selection procedure. FIGURE 1: Flow diagram showing recruitment of participants. 'n' denotes number of participants While skeletonizing the subjects of the study it was found that 33 had a totally sedentary lifestyle while the other 66 were not involved in any kind of heavy physical pursuits which could have caused bias or might have confounded the outcomes. According to the final outcome as described above, the initial cohort was divided into two groups namely a yoga performer group and a non-yoga performer one. The yoga performer group comprised subjects who followed a regular 6-month independent yogic regime and the non-yoga performer group involved students who irregularly practiced yoga. Regularity was kept a check upon by the investigators by asking for regular reports about the same. The ones who reported yogic practice for more than 75% of the days during the 6-month independent praxis were considered regular practitioners. Data sources and measurement of variables The following parameters were investigated pre- and post-yoga to determine the relationship between yogic 2023 Mittal et al. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 3 of 10 practice and physical and mental fitness. i) O₂ max - The level of oxygen consumption beyond which no further increase in oxygen consumption occurs with a further increase in the severity of exercise. Procedure for VO2 max After familiarization with the laboratory and procedures, the subjects performed an incremental ramp exercise test for volitional fatigue on a motorized treadmill/ergometer. Treadmill/ergometer speeds were predefined to increase incrementally from moderate to maximal effort. VO2 max was thus measured by this symptom-limited running exercise. A valid VO2 max was considered to have been attained when the following criteria were achieved: a. Plateau or 'peaking over' in oxygen uptake / O₂ consumption (VO₂). b. Achievement of maximum heart rate: 220 - Age ii) METS - One metabolic equivalent (MET) is defined as the amount of oxygen consumed while sitting at rest and is equal to 3.5 mL O₂ per kg body weight times the minutes of exercise. The MET concept represents a simple, practical, and easily understood procedure for expressing the energy cost of physical activities. Calculation of METS Energy cost/expenditure (METS) is used as an indicator that the participants are nearing exhaustion and the limits of their cardio-respiratory system. METS values were recorded as per the maximal oxygen consumption output data. The metabolic module of Labchart software was used to process the data and give the output readings. The Power lab 8/35 data acquisition system was used for the recording of VO2 max and deriving METS Values. Increasing workloads are used to reach exhaustion in the subject and determine a maximal level of oxygen consumption (VO2 max). A motorized treadmill (Aerofit AF 101, Nityasach Fitness Pvt Ltd, Mumbai, India) was used for the subjects to perform and reach maximal exercise levels. Baseline clinical parameters - resting pulse, blood pressure, and resting respiratory rate were measured. The height and weight were recorded as per standard procedures. iii) Anxiety levels were measured into three broad categories namely: a) Trait anxiety which is an enduring characteristic or pattern of behavior and refers to the more stable tendency to attend to, experience, and report negative emotions such as fears, worries, and anxiety across many situations. This is part of the personality dimension of neuroticism versus emotional stability. b) State anxiety which implies that state is a temporary way of being (i.e., thinking, feeling, behaving, and relating) c) Combined anxiety score Tool for Assessment of Anxiety Spielberger's anxiety scale [17], which is a standardized, validated, and widely used measure to determine the anxiety score of students, was employed. It includes a questionnaire called the ‘State-Trait Anxiety Inventory’ (STAI). A self-report assessment device, which includes two separate subscales containing 20 items each. It measures state and trait anxiety using a four-point Likert scale. Essential qualities evaluated were - feelings of apprehension, tension, nervousness, and worry. Anonymous feedback was also taken at the end of the intervention to understand students’ experience of yoga using a Proforma in which 13 parameters were assessed. Students were asked to tick against the column which was most appropriate with regard to their experience for each of the parameters. The number of students who have chosen a particular grade is expressed as a percentage of students. A consensus measure was calculated for the items on the Likert scale using the method of Tastle et al. [18]. Statistical data analysis Initially, Kobo Toolbox was used to collect anthropometric and historical data while screening the students. For quantitative data collection of variables like VO2 max, METS, and results of the anxiety scores according to the Likert scale, also Kobo toolbox was made use of. After that, R Software [19] (R Foundation for Statistical Computing, Vienna, Austria) was utilized for statistical analysis and preparation of the graphs. Certain scores of anxiety levels were found to deviate from normal distribution after their combined results 2023 Mittal et al. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 4 of 10 were checked for normal distribution. Subsequently, the non-parametric Wilcoxon signed-rank test was performed, which is used to compare two averages. Mean and standard deviation, the difference in means of various parameters, correlation analysis using Pearson’s coefficient which was deemed significant depending upon the outcome of the p-value. If the p-value was < 0.05, it was considered to be significant. Paired t-test was employed for analyzing pre- and post-intervention scores of variables. Results Participants' descriptive data The mean age of yoga performers was 19.43 ± 2.62 years and 19.54 ± 2.59 years for the non-yoga performers. There was no statistically significant difference found between these (p-value = 0.76). Reasonably mean height and weight of both the groups were found to be on similar lines which proved the point that demographic parameters did not confound the readings. There were 25 males and 21 females in the non- yoga performing group of participants and 33 males and 21 females in the yoga performers. The VO2 Max readings were taken for both the groups after 6-month intervention while the anxiety scores were recorded only for the students who regularly performed yoga (n=54). Main outcomes and results Tables 1-2 below give a comprehensive insight into the maximal oxygen consumption data (VO2 max) readings of the yoga performers as well as non-yoga performers respectively. Pre- and post-yoga readings categorized as per sex and in two defined units, have been depicted. The METS value was found to be higher post-intervention which is 11.52 compared to the baseline value of 10.34 and this difference was found to be statistically significant (p<0.05). Sex Pre-yoga reading Post-yoga reading VO2 max (mL/kg/min) VO2 max (L/min) VO2 max (mL/kg/min) VO2 max (L/min) Males (n=33) 41.86 ± 6.16 2.64 ± 0.49 44.52 ± 6.21 2.81 ± 0.52 Females (n=20) 26.95 ± 4.94 1.51 ± 0.44 29.99 ± 4.94 1.69 ± 0.47 TABLE 1: Maximal oxygen consumption data of yoga performers group. Sex Pre-yoga reading Post-yoga reading VO2 max (mL/kg/min) VO2 max (L/min) VO2 max (mL/kg/min) VO2 max (L/min) Males (n=25) 27.95 ± 5.24 1.99 ± 0.50 28.29 ± 5.39 1.93 ± 0.48 Females (n=21) 17.08 ± 2.40 1.14 ± 0.28 18.33 ± 8.10 1.24 ± 0.67 TABLE 2: Maximal oxygen consumption data of non-yoga performers group. Table 3 shows the difference in VO₂ max in mL/kg/min and VO₂ max in L/min and METS before and after intervention in both groups. For VO₂ max in mL/kg/min in the yoga performer group, the mean difference was found to be higher 2.8 compared to 0.15 in the non-intervention group and the difference was found to be statistically significant. Similarly, for METS and VO₂ max in L/min, before and after intervention value difference was found to be higher in the yoga performer group compared to the non-performer group, and the difference was found to be statistically significant (p-value = 0.0004). 2023 Mittal et al. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 5 of 10  Parameters Yoga performers group Non-Yoga performers group Difference in VO₂ max (mL/kg/min) 2.80 0.15 Difference in VO₂ max (L/min) 0.17 0.02 Difference in METS value 1.18 0.02 TABLE 3: Difference in difference analysis of fitness parameters. METS, one metabolic equivalent Anxiety scores also showed a commendable drop in the values after the intervention. Mean anxiety scores on day 1, day 30, and at the end of 6 months have been depicted. This difference was statistically significant (p < 0.001) as is clear from the Table 4 and its graphical representation is depicted as a box-plot graph in Figure 2. Mean anxiety score Difference in scores between day 1 & day 30 (with 95% confidence interval) p-value Mean anxiety score Difference in scores between Day 1 and at the end of 6 months (with 95% confidence interval) p-value Pre- yoga (on day 1) Post- yoga (on day 30) Pre- yoga (on day 1) Post-yoga (at the end of 6th month) State anxiety scores 46.82± 8.95 32.54± 7.70 14.3 (11.4-17.1) p<0.001 46.8 ± 9.0 29.4 + 6.6 17.4 (14.7 - 20.2) p<0.001 Trait anxiety scores 47.02± 10.70 33.06± 8.29 13.9 (11.2-16.7) p<0.001 46.96 ± 10.7 29.9 + 7.5 17.1 (14.5 - 19.8) p<0.001 Total anxiety scores 93.8 ± 17.73 65.62 ± 15.26 28.2 (23.1-33.4) p<0.001 93.78 ± 17.7 59.3 + 12.9 34.6 (29.8 - 39.3) p<0.001 TABLE 4: Comparison of mean of pre- and post-yoga anxiety scores. 2023 Mittal et al. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 6 of 10 FIGURE 2: Box-plot graph of a comparison of means of pre- and post- yoga anxiety scores. Pre score - mean anxiety score on day 1 and before the intervention of yoga. Post score - mean anxiety score at the end of 6-month yogic regime. To give the study more credibility and authenticity from the subject's viewpoint, participants were made to fill out a feedback questionnaire. The results were extrapolated in percentages. Table 5 gives a glance through the questionnaire. 2023 Mittal et al. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 7 of 10 Parameters Highly positive change Moderately positive change No change Moderate negative change Highly negative change Consensus Sense of contentment & well being 52 36 12 0 0 0.75 A feeling of calmness & relaxation 66 32 2 0 0 0.81 Level of concentration in studies 54 38 8 0 0 0.77 Hours required to rejuvenate 0 46 48 4 2 0.77 Self-confidence 34 44 22 0 0 0.75 Competence in any task 42 50 8 0 0 0.78 Irritability levels 42 42 16 0 0 0.75 Stamina 42 50 14 0 0 0.76 Lethargy 18 60 18 2 2 0.77 Appetite 6 22 70 2 0 0.80 Optimistic outlook in life 34 46 20 0 0 0.76 Headache, body ache 34 48 14 2 2 0.73 Mutual Interpersonal relationship 52 30 18 0 0 0.72 TABLE 5: Feedback score for various parameters expressed as a percentage of participants. Apart from the improvement observed in mental well-being score, the students also reported other beneficial effects of yoga in their anonymous feedback such as: 1. Better sleep 2. Better concentration in studies 3. Better control of anger and other negative symptoms 4. More relaxed and active throughout the day 5. Getting positive energy at the beginning of the day. Discussion A study incorporating the estimation of VO2 max along with psychological assessment was long due for youngsters. Once evaluated in conjunction, this could prove helpful for the pupils to get an idea of their aerobic capacity so as to modulate the intensity of different yoga practices according to their needs. This interventional study analyzed the dynamics of the cardiorespiratory responses in medical students and it was revealed that the yoga group had statistically significant higher VO₂ max and improved METS values when the fitness metrics were calculated pre- and post-yoga intervention. Cardiopulmonary fitness assessed as VO2 max, is regarded as a critical marker for youth health. Elevated VO2 max level in yoga performers is eventually linked to better physical fitness and was a potential outcome of regular yogic practice. This finding of our study is in accordance with the studies performed by Loganathan et al. [20] and Parikh et.al. [21]. Due to a consistent yoga practice, respondents' initial sky-high anxiety levels dramatically decreased, which contributed to the inculcation of a judicious insight in young adults. Despite the fact that the youngsters who were selected for the research were healthy, their reduced aerobic capacity seemed nerve-wracking to call for a solution. When introduced to them at this juncture, yoga curbed this issue through an extensive regime tailored to their need and aimed at improving their aerobic fitness and mitigating anxiety. 2023 Mittal et al. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 8 of 10 Our results are consistent with previous studies [22-25] which examined the effects of yoga on the health of medical students. The paucity of data on the impact of yoga on young medical undergraduates' functional aerobic capacity during treadmill/ergometer exercise in the literature available so it was thought pertinent to utilize such a methodology, which became the study's unique selling feature. Additionally, the yoga group reported plausible improvements in parameters like an improved sense of well- being, a feeling of relaxation, enhanced concentration, self-confidence, better efficiency, sound interpersonal relationships, augmented attentiveness, reduced irritability levels, and an upbeat outlook on life [26-27]. A study by Bansal et al. reported significant improvement in general and mental well-being following the intervention which again corroborates our findings [28]. Akin to any research, this study too has certain limitations. We have used only a single composite questionnaire-based measure of anxiety and have not studied psychological factors such as appraisal and coping mechanisms that may influence the stress response. The stress scores were obtained at one point of time while they were in medical school and hence the status of the mental health of students prior to their entry to the medical course could have influenced the levels of stress. Other sources of stress such as familial or interpersonal problems were not examined. Biochemical parameters of stress such as plasma or salivary cortisol were not measured. Nutrition was also one such factor that was not taken into sight, considering the fact that all students had the same source of food and kitchen being a part of the medical school. Hence that did not affect the readings drastically. Moreover, as a future prospect, it confers an avenue for further studies that can be undertaken in which a nutritive intervention could be used. Conclusions To meet the modern lifestyle full of challenges and tensions, it has become imperative especially for medical students to bail out of this turmoil and emerge with a whole new persona. There was a significant improvement in the VO2 max and a markedly discernible decrease in anxiety levels of yoga performers. To spell out the crux of the current research, it can be conjectured that yoga can not only expound aerobic fitness but can concurrently serve to be beneficent in achieving a tranquil state of mind during young age, yet providing the concentration and arousal essential in this demanding or stressful vivency of youth. Additional Information Disclosures Human subjects: Consent was obtained or waived by all participants in this study. Institutional Ethics Committee for Research on Human Subjects of Mahatma Gandhi Institute of Medical Sciences, Wardha issued approval MGIMS/IEC/PHY/101/2022. Consent was obtained or waived by all participants in this study. Institutional Ethics Committee for Research on Human Subjects of Mahatma Gandhi Institute of Medical Sciences, Wardha issued approval MGIMS/IEC/PHY/101/2022. . Animal subjects: All authors have confirmed that this study did not involve animal subjects or tissue. Conflicts of interest: In compliance with the ICMJE uniform disclosure form, all authors declare the following: Payment/services info: All authors have declared that no financial support was received from any organization for the submitted work. Financial relationships: All authors have declared that they have no financial relationships at present or within the previous three years with any organizations that might have an interest in the submitted work. Other relationships: All authors have declared that there are no other relationships or activities that could appear to have influenced the submitted work. Acknowledgements Ruchi Kothari and Gaurav Mittal contributed equally to the work and should be considered co-first authors. The data are stored as de-identified participant data, which are available on reasonable request to Ruchi Kothari (ruchi@mgims.ac.in). The authors would like to thank Dr. Arjun Kumar Jakasania, Department of Community Medicine for assistance in statistical analysis and outcomes. We are thankful to the yoga trainer, Arogyadham for training the students and carrying out the month-long yogic regime. The authors also thank Ms. Maitri Gopani, Ms. S Sushmitha, Mr. Manish Rathod, Mr. Shreyash Yedke, Mr. Pratyaksh Gurnani, and the whole of the Agnivesh batch for their constant support for procuring and managing the subjects. The authors acknowledge all the participants of the study. References 1. Abraham RR, Zulkifli EM, Fan ES, Xin GN, Lim JT: A report on stress among first year students in an Indian medical school. South East Asian J Med Educ. 2009, 30:78-81. 2. Woolery A, Myers H, Sternlieb B, Zeltzer L: A yoga intervention for young adults with elevated symptoms of depression. Alternat Ther Health Med. 2004, 10: 3. Miller PM: The first year at medical school: some findings and student perceptions . Med Educ. 1994, 28:5-7. 10.1111/j.1365-2923.1994.tb02678.x 4. Rosenberg PP: Students' perceptions and concerns during their first year in medical school . J Med Educ. 2023 Mittal et al. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 9 of 10 1971, 46:211-218. 10.1097/00001888-197103000-00005 5. Vitaliano PP, Russo J, Carr JE, Heerwagen JH: Medical school pressures and their relationship to anxiety . J Nervous Mental Dis. 1984, 172:730-736. 6. Stewart SM, Betson C, Marshall I, Wong CM, Lee PW, Lam TH: Stress and vulnerability in medical students. Med Educ. 1995, 29:119-127. 10.1111/j.1365-2923.1995.tb02814.x 7. Udupa KN: Stress and Its Management by Yoga, 2nd ed. . Narendra Prakash Jain, New Delhi; 1985. 8. Anantharaman V, Subrahmanyam S: Physiological benefits in hatha yoga training. Yoga Rev. 1983, 3:9-24. 9. Joseph S, Sridharan K, Patil SK, et al.: Study of some physiological and biochemical parameters in subjects undergoing yogic training. Indian J Med Res. 1981, 74:120-124. 10. Selvamurthy W, Nayar HS, Joseph NT, Joseph S: Physiological effects of yogic practices . NIMHANS J. 1983, 1:71-80. 11. Supe AN: A study of stress in medical students at Seth G.S. Medical College . J Postgrad Med. 1998, 44:1-6. 12. Field T: Yoga clinical research review . Complement Ther Clin Pract. 2011, 1:1-8. 10.1016/j.ctcp.2010.09.007 13. Udupa KN, Singh RH, Yadav RA: Certain studies on psychological and biochemical responses to the practice in Hatha Yoga in young normal volunteers. Indian J Med Res. 1973, 61:237-244. 14. Selvamurthy W: Yoga for everyone: a physiologist's view. Souvenir, 2nd Congress of Asian and Oceanian Physiological Societies. 1990, 12-15. 15. Brain and Psychophysiology of Stress. Sharma KN, Selvamurthy W, Battacharya N (ed): Indian Council of Medical Research, New Delhi; 1983. 16. MI JH, SP BJ, CH CB: The physiological meaning of the maximal oxygen intake test . J Clin Invest. 1958, 37:538-547. 10.1172/JCI103636 17. Spielberger CD, Gorsuch RL, Lushena RE: Manual for the State Trait Anxiety Inventory (Self Evaluation Questionnaire). Consulting Psychologists Press , Palo Alto, CA; 1970. 18. Tastle WJ, Wierman MJ, Dumdum UR: Ranking ordinal scales using the consensus measure . Issues in Inform Syst. 2005, 6:96-102. 10.48009/2_iis_2005_96-102 19. R Development Core Team: R: language and environment for statistical computing . R Foundation for Statistical Computing, Vienna; 2010. 20. Loganathan N, Aruchunan M, Manjunath NK: Effects of yoga for cardiovascular and respiratory functions: a pilot study. Integr Med Res. 2019, 8:180. 10.1016/j.imr.2019.05.004 21. Parikh HN, Patel HM, Pathak NR, Chandwani S: Effect of yoga practices on respiratory parameters in healthy young adults. Natl J Integr Res Med. 2014, 5:3. 22. Srinivasan K, Vaz M, Sucharita S: A study of stress and autonomic nervous function in first year undergraduate medical students. Indian J Physiol Pharmacol. 2006, 50:257-264. 23. Gupta N, Khera S, Vempati RP, Sharma R, Bijlani RL: Effect of yoga based lifestyle intervention on state and trait anxiety. Indian J Physiol Pharmacol. 2006, 50:41-47. 24. Malathi A, Damodaran A: Stress due to exams in medical students--role of yoga . Indian J Physiol Pharmacol. 1999, 43:218-224. 25. Divya TS, Vijayalakshmi MT, Mini K, Asish K, Pushpalatha M, Suresh V: Cardiopulmonary and metabolic effects of yoga in healthy volunteers. Int J Yoga. 2017, 10:115. 10.4103%2F0973-6131.186162 26. Eppley KR, Abrams Al, Shear J: Differential effects of relaxation techniques on trait anxiety: a meta analysis . J Clin Psychol. 1989, 45:957-974. 27. Brown RP, Gerbarg PL: Sudershan kriya yogic breathing in the treatment of stress, anxiety, and depression: part 1-neurophysiological model. J Alt Compl Med. 2005, 11:189-201. 10.1089/acm.2005.11.189 28. Bansal R, Gupta M, Agarwal B, Sharma S: Impact of short term yoga intervention on mental well being of medical students posted in community medicine: a pilot study. Indian J Commun Med. 2013, 38:105. 10.4103/0970-0218.112445 2023 Mittal et al. Cureus 15(5): e38847. DOI 10.7759/cureus.38847 10 of 10

Divulging the Impetus of Yoga on Cardiorespiratory Fitness and Its Persona in Alleviating Anxiety Experienced by Youth: A Cohort Interventional Study.

05-10-2023

Mittal, Gaurav,Kothari, Ruchi,Yadav, Akshay,Bokariya, Pradeep,A, Prashanth

eng

PMC8523042

RESEARCH ARTICLE Spatiotemporal inflection points in human running: Effects of training level and athletic modality Yuta Goto1, Tetsuya Ogawa2, Gaku Kakehata1, Naoya Sazuka3, Atsushi Okubo4, Yoshihiro Wakita4, Shigeo Iso5, Kazuyuki Kanosue5* 1 Graduate School of Sport Sciences, Waseda University, Saitama, Japan, 2 Department of Clothing, Faculty of Human Sciences and Design, Women’s University Tokyo, Japan, 3 Tokyo Laboratory 25, R&D Center, Sony Group Corporation, Tokyo, Japan, 4 Tokyo Laboratory 07, R&D Center, Sony Group Corporation, Tokyo, Japan, 5 Faculty of Sport Sciences, Waseda University, Saitama, Japan * kanosue@waseda.jp Abstract The effect of the different training regimes and histories on the spatiotemporal characteris- tics of human running was evaluated in four groups of subjects who had different histories of engagement in running-specific training; sprinters, distance runners, active athletes, and sedentary individuals. Subjects ran at a variety of velocities, ranging from slowest to fastest, over 30 trials in a random order. Group averages of maximal running velocities, ranked from fastest to slowest, were: sprinters, distance runners, active athletes, and sedentary individu- als. The velocity-cadence-step length (V-C-S) relationship, made by plotting step length against cadence at each velocity tested, was analyzed with the segmented regression method, utilizing two regression lines. In all subject groups, there was a critical velocity, defined as the inflection point, in the relationship. In the velocity ranges below and above the inflection point (slower and faster velocity ranges), velocity was modulated primarily by alter- ing step length and by altering cadence, respectively. This pattern was commonly observed in all four groups, not only in sprinters and distance runners, as has already been reported, but also in active athletes and sedentary individuals. This pattern may reflect an energy sav- ing strategy. When the data from all groups were combined, there were significant correla- tions between maximal running velocity and both running velocity and step length at the inflection point. In spite of the wide variety of athletic experience of the subjects, as well as their maximum running velocities, the inflection point appeared at a similar cadence (3.0 ± 0.2 steps/s) and at a similar relative velocity (65–70%Vmax). These results imply that the influence of running-specific training on the inflection point is minimal. Introduction Human running has been studied extensively from the viewpoint of how its temporal (cadence) and spatial (step length) components contribute to velocity [1–10]. Velocity equals the product of cadence and step length, and the relative contribution of each component to PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 1 / 12 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Goto Y, Ogawa T, Kakehata G, Sazuka N, Okubo A, Wakita Y, et al. (2021) Spatiotemporal inflection points in human running: Effects of training level and athletic modality. PLoS ONE 16(10): e0258709. https://doi.org/10.1371/journal. pone.0258709 Editor: Leonardo A. Peyre´-Tartaruga, Universidade Federal do Rio Grande do Sul, BRAZIL Received: February 11, 2021 Accepted: October 4, 2021 Published: October 18, 2021 Peer Review History: PLOS recognizes the benefits of transparency in the peer review process; therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. The editorial history of this article is available here: https://doi.org/10.1371/journal.pone.0258709 Copyright: © 2021 Goto et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. changing velocity differs across the velocity range. A previous study reported that, at slower velocities, speed is modulated primarily by adjusting step length, whereas, at faster velocities, speed is modulated more by changes in cadence [6]. At velocities close to maximum, step length shows only a small increase or even a decrease as running velocity approaches the maxi- mum [8]. These characteristics are considered to indicate the spontaneous recruitment of an adequate motor pattern which minimizes energy expenditure at a given running velocity [5, 11–13]. Mechanical approaches, such as Fenn’s approach, have been used as useful tools to elu- cidate these energy cost determinants with many practical applications [14]. Yanai and Hay [12], utilizing a two-dimensional simulation, evaluated the relative contribution of cadence and step length in the optimization of power production utilizing both anatomical (range of motion in the hip joint) and spatiotemporal (duration of the stance phase) determinants. Indeed, if the cadence is voluntarily modified from that occurring under the natural move- ment pattern at a given running velocity, metabolic rate is lowest when the cadence is in the range of ±10% of the preferred cadence [15–18]. In addition, in the slower velocity range, Cavagna et al. [19] reported that preferred cadences take place in the proximity of 3 Hz. However, the extent to which the above characteristics occur in different populations and in persons with different physical backgrounds remains unclear. Most of the above-mentioned studies focused on well-trained individuals, especially those trained for running [7–9, 12, 20]. Therefore, the purpose of the present study was to investigate how a change in running velocity altered the spatiotemporal adjustment between step length and cadence in subjects with different histories of engagement in running training. Namely, we studied:1. sprinters, 2. distance runners, 3. active athletes who had received no running-specific training, and 4. sed- entary, untrained subjects. The relationships among running velocity, cadence, and step length over a wide range of running velocities were compared across these subjects. Among the four groups, the distance runners would be expected to run as efficiently (either mechanically or metabolically) as possible. As noted above, in the slower velocity ranges, altering stride length is a more energy saving strategy for changing velocity than is altering cadence [12]. Therefore, we hypothesized: 1. the running step length/cadence patterns of individuals would be influ- enced by their running training experience and overall physical activity levels and 2. distance runners would exhibit the greatest tendency to change velocity by altering step length in the slower velocity range. Methods Subjects A total of eighty volunteers (69 males and 11 females) with different backgrounds, in terms of their running experience, participated in the study. They were assigned into one of four groups depending on their current/previous running training. We utilized four groups of subjects with different histories of running training. The first and second groups consisted of twenty sprinters (all men) and twenty distance runners (all men), respectively. The participants in the third group were twenty active athletes (16 males and 4 females). Although running is involved in many of the sports, all subjects informed us that they had received no special training for improving their running speed. For reference, the sports that the participants in the third group engaged in were: soccer, basketball, softball, weightlifting, boxing, lacrosse, volleyball, American football, badminton, handball, rowing, judo, and golf. They had all participated in their sport for at least 5 years. The fourth group consisted of sedentary individuals without a history of any regular participation in sports activities (13 males and 7 females). Table 1 lists the characteristics of participants in each group. All participants were informed of the pur- poses and procedures, and signed an informed consent form. This study was approved by the PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 2 / 12 Funding: This work was supported by Japan Society for the Promotion of Science (JSPS), KAKENHI Grant Number 19K22822 (K.K) and by Grant-in-Aid for JSPS Fellows Number 20J11122 (Y.G) from Ministry of Education, Culture, Sports, Science and Technology of Japan. Sony Group Corporation provided support in the form of salaries for authors [NS, AO, and YW], but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of these authors are articulated in the ‘author contributions’ section. Sony Group Corporation has a patent (US20180039751A1) on apparatuses for helping runners modify the V-C-S property. This patent does not interfere with the usage of any data or knowledge presented in the paper. Competing interests: Sony Group Corporation provided support in the form of salaries for authors [NS, AO, and YW].Sony Group Corporation does not alter the adherence to PLOS ONE policies on sharing data and materials presented in this paper. Human Research Ethics Committee in Faculty of Sport Sciences, Waseda University. The experiments were conducted in accordance with the Declaration of Helsinki. Experimental setup and tasks Experiments were conducted on a 30 m all-weather straight track (only 20 m for the sedentary group in consideration of their physical strength and lack of stamina) on which color markers were placed every 0.5 m for video analysis. A sagittal view of each participant was recorded by panning with a video camera (HDR-CX630V, SONY) placed approximately 10 m lateral to the center of the running path. An additional 10–30 meters was provided before and after the filming zone (of 30m or 20m) so that the subjects could accelerate and decelerate and thus maintain run- ning velocity as constant as possible throughout the recording area. This acceleration distance differed between trials and was selected by the subject. The video sampling frequency was 60 Hz. Participants were asked to run along the path 30 times at a variety of velocities, which varied from slow to the fastest possible. The order of running with different velocities was random- ized on a subject-by-subject basis. The subjects were directed to run at a particular percentage of their maximal effort [21]. This instruction included requesting a subjective effort from 10% to 100% of maximum, as well as “run faster or slower than the previous trial”. The actual run- ning speed did not necessarily match the exact percentage of their maximal speed. However, this method did produce the necessary array of running speeds and the subjects might run more than once at an intensity. When running at the minimum velocity, subjects followed our instruction to run as slowly as they could while still maintaining a running gait (as opposed to walking, jumping, hopping, or bounding). The interval between trials ranged from 30 seconds to 5 minutes, depending on the speed of the previous trial. A 5-minute rest was taken after 15 trials. The participants used their own running shoes. Spiked shoes were not allowed. Data analysis Offline data analysis was performed by using video administration software (PlayMemories, SONY, Japan). On the basis of the video analysis, the running velocity, cadence, and step length were calculated on a trial-by-trial basis for each subject. Mean running velocity (m/sec) was calcu- lated by dividing the length of the path (m) by the time taken (sec) to run over the path. The instant at which the subject passed the start and the end point were identified from the position of the chest relative to the color markers. Mean cadence (steps/sec) was calculated by dividing the number of steps by the time taken to cover that distance. The number of steps was counted from the first ground contact with the path to the last ground contact before passing the end point. The duration utilized was defined as the time between the instant of first foot-contact after the start position and that of the last foot-contact before the end. Mean step length (m) was calculated by dividing the mean running velocity (m/sec) by the mean cadence (steps/sec). Step length was also Table 1. Physical characteristics and sport activity history of each subject group. N age, years height, cm sports activity history, years Sprinters 20 22 ± 2 176.2 ± 6.1b, c, d 9.7 ± 3.0 Distance runners 20 20 ± 1 171.0 ± 4.5 7.4 ± 2.0 Active athletes 20 23 ± 2 170.1 ± 5.8 10.2 ± 4.4 Sedentary individuals 17 22 ± 2 166.0 ± 6.2 Values are means ± SD. N, number of subjects. b, c, d: values are significantly different from distance runners, active athletes, and sedentary individuals, respectively (p < 0.05). The sport activity history of the active athletes indicates the number of years of participation in that sport for each subject. https://doi.org/10.1371/journal.pone.0258709.t001 PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 3 / 12 expressed as the ratio of the step length (m) to the height (m) of each subject in order to examine the influence of the physical characteristics of the subjects. For the running velocity, the fastest among the 30 trials by each subject was designated as their maximal running velocity. In the present study, the principal analyses for the spatiotemporal running characteristics of each subject were performed with MATLAB version R2018a (The MathWorks, Inc., USA). For each subject, the data were plotted as shown in Fig 1 in order to examine the relationship between cadence and step length (horizontal axis: cadence, vertical axis: step length). This cor- respondence involved the Velocity (m/s, dotted line), Cadence (steps/s, horizontal), and Step length (m, vertical), and is defined as the V-C-S relationship. To quantitatively analyze the crit- ical point at which the relative contribution of spatiotemporal adjustment changed (cadence vs. step length), we utilized the segmented regression method which has previously been used to detect lactate threshold [22] and ventilation threshold [23] during aerobic exercise. This is a statistical method for determining the point at which a line suddenly changes slope at some unknown point. We used a segmented regression procedure [23, 24] in which the N data points were divided into two segments (the lower x data and the upper N-x data, x = 3, 4, . . ., or N-2). Each segment was fitted with a regression line using the Deming regression [25, 26]. This regression method was adopted to exclude the effects of measurement errors in cadence and step length. That is, one regression line was obtained with x data points from the ascend- ing order starting with the minimum velocity, and the other one with N-x data points from the descending order starting with the maximum velocity. The critical point (“inflection point”), then, was the intersection of the two regression lines with an x value that minimized orthogo- nal distance between measurement data and regression line for two data sets (segments) (Fig 1, cross; X). We assumed that the regression lines below and above the inflection point would adequately represent the spatiotemporal characteristics of running for each subject and group. Subjects with inflection points, thus obtained, that differed largely from the measured points, were excluded from the analysis (#18, #19, and #20, as seen in S4 Fig). Therefore, the final analysis involved 20 sprinters, 20 distance runners, 20 active athletes, and 17 sedentary individuals. For these subjects, running velocity, cadence, and step length at Fig 1. The relationship between cadence (steps/s, horizontal) and step length (m, vertical) relative to running velocity (pale broken line and the second vertical axis) in a single sprinter. The inflection point (cross) was computed from two regression lines from different data sets by combining the segmented regression method of Deming regression. The filled and open circle markers represent the data sets below and above the inflection point at which the relationship between cadence and step length changed abruptly. Inflection point was obtained as the intersection point of the two regression lines. https://doi.org/10.1371/journal.pone.0258709.g001 PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 4 / 12 the inflection point were calculated. Normalized values were determined for each parameter at the maximal running velocity. Statistical analysis Statistical analysis was performed using SPSS Statistics 23 software (IBM, USA). Maximal run- ning velocity, height of subjects, and all variables related to inflection point in each group were tested for a normal distribution using the Shapiro-Wilk test. Maximal running velocity, height, and normalized cadence at the inflection point were found to have non-normal distributions. Thus, group mean data for maximal running velocity, height of subjects, and all variables related to inflection point were analyzed among the four subject groups by using a non-parametric Krus- kal-Wallis test. Next, post-hoc pairwise comparisons using the Dunn-Bonferroni approach were made to identify additional differences between the groups. In order to further investigate the possible mechanisms responsible for the inflection point, correlational analyses were performed. All variables across all subjects related to the inflection point and maximal running velocity were tested for a normal distribution using the Shapiro-Wilk test. Maximum running velocity, and step length at maximal running velocity exhibited normal distributions. Likewise, running velocity (both unnormalized and normalized), step length (both unnormalized and normal- ized), and unnormalized cadence at the inflection point exhibited normal distributions. How- ever, cadence at maximal running velocity and normalized cadence at the inflection point exhibited non-normal distributions. Pearson’s and Spearman’s correlations were performed to analyze the relationship between maximal running velocity and other parameters at the inflec- tion point. Significance was set at p < 0.05. The data are presented as mean and standard devi- ation (mean ± SD). Results Fig 1 shows a typical example of the relationship between running velocity, cadence, and step length for a single sprinter. Both cadence and step length show specific changes in relation to changing running velocity. The inflection point (cadence: 2.97 steps/s, step length: 1.78 m) was computed from two regression lines. Fig 2A shows an inter-group comparison of the mean values of Vmax. A Kruskal-Wallis test revealed significant differences between the groups in terms of maximum running velocity (χ2 (3) = 52.463, p < 0.001). The post-hoc comparisons revealed that the maximal velocity of the sprinters was faster compared to all the other subject groups (distance runner: p = 0.009, active athlete: p < 0.001, sedentary: p < 0.001). The distance runner group exhibited signifi- cantly faster maximal running velocity in comparison with the sedentary individual group. Fig 2B–2D illustrates the correlation between maximal running velocity and cadence, absolute step length and step length normalized to height at the maximal running velocity. There were significant positive correlations between Vmax and cadence as well as step length both in the unnormalized and normalized forms (cadence: r = 0.514, p < 0.001; step length (unnorma- lized): r = 0.843, p < 0.001; step length (normalized): r = 0.803, p < 0.001). Fig 3A shows mean values of cadence and step length at maximal running velocity (Vmax), the inflection point, and minimal running velocity (Vmin) for each subject group. As shown in Fig 3A, maximal running velocity was different across the groups and was the fastest in the sprinters (I, around 10 m/s) and slowest in the sedentary individuals (IV, mostly less than 8 m/ s). All groups tended to increase step length predominately at the velocities between Vmin (velocity: 2.17 ± 0.45 m/s, cadence: 2.62 ± 0.14 steps/s, step length: 0.82 ± 0.17 m) and the inflection point, and then to increase cadence until they reached Vmax. Fig 3B depicts mean values of cadence and step length normalized to the values obtained under maximal running PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 5 / 12 velocity. The characteristics of the increase in velocity were similar to those from Fig 3A. Due to differences in the absolute value (Fig 3A) of maximal running velocity, the normalized cadence varied considerably across the subject groups, while variability in step length below the inflection point was less evident. Table 2 shows inter-group comparison of the mean values of all variables related to the inflection point. A Kruskal-Wallis test revealed significant difference of running velocity, step length, normalized cadence (χ2 (3) = 31.215, p < 0.001; χ2 (3) = 42.68, p < 0.001; χ2 (3) = 23.623, p < 0.001, respectively). The post-hoc comparisons revealed significant differences between the subject groups. In the group of sprinters, the running velocity was significantly faster as compared to the active athlete, and sedentary subject groups (active athlete: p < 0.01, sedentary: p < 0.001). For the same parameter, the group of distance runners showed signifi- cantly faster in comparison to the sedentary group (p < 0.01). The step length was significantly longer in the sprinter group in comparison to all the other subject groups (distance runner: p < 0.01, active athletes: p < 0.001, sedentary: p < 0.001). For the same parameter, the group of distance runners was significantly longer than the sedentary group (p < 0.05). In the group Fig 2. Inter-group comparison of mean values (diamond) of the maximum running velocity (Vmax) (A), and correlation between the maximal running velocity and the cadence (B), step length (C), and step length normalized by height (D) at maximal running velocity. In Fig 2A, open circles indicate each individual subject. Significant difference; p < 0.001, p < 0.01. In Fig 2B–2D, filled circles, open circles, filled triangles, and open triangles represent the sprinters, distance runners, active athletes, and sedentary individuals, respectively. There are significant positive correlations between Vmax and the cadence (B) and between Vmax and step length, both absolute velocity and velocity normalized to maximal running velocity (r = 0.514, p < 0.001; r = 0.843, p < 0.001; r = 0.803, p < 0.001, respectively). https://doi.org/10.1371/journal.pone.0258709.g002 PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 6 / 12 of sprinters, the normalized cadence was lower as compared to distance runner and sedentary subject groups (distance runner: p < 0.01, sedentary: p < 0.001). Fig 4A–4C depicts correlations between maximal running velocity and running velocity, cadence, and step length at the inflection point. There were significant positive correlations between Vmax and both velocity and step length at the inflection point (velocity: Fig 4A, r = 0.738, p < 0.001; step length: Fig 4C, r = 0.827, p < 0.001). Cadence at the inflection point had no correlation with Vmax, and was approximately constant at 3.0 ± 0.2 steps/s regardless of the subject group (Fig 4B). Fig 4D–4F illustrates correlation for the same parameters shown in Fig 4A–4C, but with values normalized to Vmax. Velocity and cadence show negative corre- lations (velocity: r = -0.300, p < 0.01; cadence: r = -0.621, p < 0.001), while step length has a positive correlation with Vmax (r = 0.290, p < 0.05). Discussion We investigated the relative contribution of cadence and step length changes as running veloc- ity was modulated in four groups of subjects with different histories of engagement in Fig 3. Mean values of cadence and step length at the maximal running velocity (Vmax), inflection point (IP), and minimal running velocity (Vmin) (A), and those with cadence and step length normalized to those under Vmax (B) for each subject group. The error bars depict the standard deviation. The filled circles, open circles, filled triangles, and open triangles represent sprinters, distance runners, active athletes and sedentary individuals, respectively. Pale broken lines represent running velocity (A) and running velocity normalized by maximal running velocity (B). The thick broken line in B illustrates the limiting situation, in which velocity change is only done with a step length change in the velocity range below the inflection point, and only with a cadence change above the inflection point. https://doi.org/10.1371/journal.pone.0258709.g003 Table 2. Kinematic variables at the inflection point. Sprinters (N = 20) Distance runners (N = 20) Active athletes (N = 20) Sedentary individuals (N = 17) velocity, m/s 5.86 ± 0.59c, d 5.36 ± 0.60 d 5.00 ± 0.50 4.50 ± 0.65 step length, m 2.03 ± 0.13 b, c, d 1.75 ± 0.14 d 1.69 ± 0.18 1.52 ± 0.21 cadence, steps/s 2.88 ± 0.26 3.06 ± 0.17 2.97 ± 0.15 2.96 ± 0.18 normalized velocity, % 64.7 ± 7.1 67.0 ± 7.5 66.7 ± 4.8 68.6 ± 7.4 normalized step length, % 96.5 ± 7.2 92.2 ± 8.2 94.1 ± 5.8 90.3 ± 7.7 normalized cadence, % 67.0 ± 4.7 b, d 72.6 ± 4.2 71.2 ± 6.6 76.0 ± 6.0 Values are means ± SD. N, number of subjects. b, c, d: values are significantly larger, from distance runners, active athletes, and sedentary individuals, respectively. Normalized velocity, step length, and cadence were obtained by normalizing with corresponding values at the maximal running velocity, respectively. https://doi.org/10.1371/journal.pone.0258709.t002 PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 7 / 12 running-specific training, utilizing the segmented regression method with two regression lines (Fig 1). In spite of a large variation in maximal running velocity, the general characteristics of the V-C-S relationship were similar across the subject groups (Fig 3) as well as across the data of individuals (S1–S4 Figs). Basic characteristics of the V-C-S relationship As expected, compared to the sprinters, maximal running velocities were progressively slower in the distance runners, active athletes and sedentary groups. There were significant differ- ences between the sprinters and the other three groups, as well as between the distance runners and the sedentary individuals (Fig 2A). Both cadence and step length at Vmax were well corre- lated with Vmax (Fig 2B and 2C, respectively). Among the subject groups, the sprinters were the tallest and the sedentary group was the shortest. The strong correlation of step length with Vmax was well-preserved, however, even when step length was normalized to the subjects’ heights (Fig 2D). Thus, faster maximum running velocities were generally accomplished with both a higher cadence and longer steps. The minimum running velocity was common to all subject groups at 2.17 ± 0.45 m/s with a cadence of 2.62 ± 0.14 steps/s and a step length of Fig 4. Correlation between maximal running velocity (Vmax) and: running velocity (A), cadence (B), and step length (C), as well as the same three parameters normalized to the Vmax (D–F) at the inflection point. Filled and open circles, and filled and open triangles represent the sprinters, distance runners, active athletes, and sedentary individuals, respectively. The correlations are all significant except for cadence (B). https://doi.org/10.1371/journal.pone.0258709.g004 PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 8 / 12 0.82 ± 0.17 m (Fig 3A). It appears that a slower cadence would have required “hopping” rather than running, and for shorter step lengths it became similar to “jogging in place”. In all four subject groups, an abrupt change in the V-C-S relationship took place at the inflection point (Fig 3 and Table 2). Velocity changes below the inflection point occurred mainly by modulating step length and velocity changes above the inflection point occurred mainly via cadence modulation. These characteristics were demonstrated in preceding studies conducted on sprinters and distance runners [7, 9], and are particularly prominent in sprinters. Running velocity at the inflection point has a significant positive correlation with Vmax (Fig 4A). Thus, the faster the Vmax, the faster the velocity at the inflection point. A faster velocity at the inflection point is mainly attained by longer step length (Fig 4C). However, this correlation was weak when it is normalized with the step length at the Vmax (Fig 4F). Overall, regardless of the training history, all groups had a similar relative step length quite close to the maximum step length (about 90%). Interestingly, the cadence at the inflection point has no correlation with Vmax and remained constant at about 3 steps/sec (Fig 4B). The history of the training influenced normalized cadence at the inflection point, that is, sprinters had a lower normalized cadence at the inflection point than the others, although in absolute terms cadence was the same. In the normalized plane (Fig 3B) inflection points of the different groups are lined along the isovelocity curve of 65–70%. Scatter plots of all subjects of all the groups showed only a weak correlation between the Vmax and the velocity at the inflection point normalized with Vmax (Fig 4D). In spite of the wide range of sports, and thus athletic modality of the subjects as well as their maximum running velocity, the inflection point appeared at a similar cadence (3.0 ± 0.2 steps/s) as well as at similar relative velocity (65–70% Vmax), across all groups. These results imply that the influence of running-specific training on the inflection point is minimal. Functional meaning of the V-C-S relationship Although the basic characteristics of the V-C-S relationship are common across different sub- ject groups, the quantitative difference could be related to quality/quantity difference in run- ning-specific training among groups. In the present study, four groups of subjects, sprinters, distance runners, active athletes uti- lizing varying degrees of running but no running training, and sedentary individuals, were studied. Of course, the above order would also be expected for the maximal velocity from fast- est to the slowest (Fig 2A). Sprinting and distance training involves running on a daily basis, and running (generally without specific running instruction) forms one aspect of training for many of the active athletes as well. It seems reasonable that some portion of the observed maxi- mal velocities reflect differences in training. Interestingly, step length at the inflection point also follows the same order as the maximal velocity (Figs 3A and 4C and 4F). In the velocity range below the inflection point, velocity change is mainly done with a change in step length; for energy-saving this is a more efficient strategy than is changing the cadence [12]. It would be beneficial for distance runners to run within this range as much as possible when their velocity is below the inflection point. Indeed, it was shown that at 4.4 m/s velocity, in the range below the inflection point, the stride length was associated with better running economy in distance runners [27]. Therefore, we had hypothesized that the ability to run below the inflection point would be particularly developed in distance runners. However, sprinters and not distance runners increased velocity by elon- gating both absolute step length (Fig 4C) and relative step length (Fig 4F), all the way to the upper running speed limit. Thus, our working hypothesis was rejected. Sprinters rarely train PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 9 / 12 in the velocity range below the inflection point. Obviously, maximal velocity is crucial for sprinters. A faster velocity cannot be accomplished only with power, especially at the highest levels. Sprinters need to develop both power and economy to the upper limit, and inevitably and unintentionally develop mechanically efficient movements. Future studies Why and by what means are there differences in the various parameters of the V-C-S relation- ship? In particular, the neural as well as physiomechanical mechanisms of differences in the V-C-S relationship should prove very interesting. In the future, motion analysis together with measurements of muscle activity and ground reaction forces could help to answer our overall question. Although numerical simulation of running and walking has many limitations [11, 12, 28], the differences in the V-C-S relationship could be analyzed with numerical models in terms of various energy costs. Furthermore, it is very interesting that even in the sedentary subjects, the basic pattern of V-C-S relationship, which is considered to reflect efficiency [12, 13], was seen. Is the V-C-S pattern innate or does it develop along the development? This, and also fatigue [29], aging [30, 31], and sex differences [32], if any, are topics that merit future analysis. Conclusions In the present study we analyzed the V-C-S relationship of running with the segmented regres- sion method and made a quantitative comparison of the “spatiotemporal running characteris- tics” in subjects with different histories of running-specific training. The common characteristic of the V-C-S relationship is, in the slower and faster velocity ranges, that velocity is mainly mod- ulated by altering step length and cadence, respectively. This was observed not only in the sprinters and distance runners, as shown in previous studies, but in active (general sport) ath- letes and sedentary subjects as well. In spite of the wide range of athletic modalities of the sub- jects, and their maximum running velocity, the inflection point appeared at a similar cadence (3.0 ± 0.2 steps/s) and at similar a relative velocity (65–70%Vmax), across all groups. These results imply that the influence of running-specific training on the inflection point is minimal. Supporting information S1 Fig. The relationship between cadence and step length for all the sprinters. The two dashed lines depict the regression lines computed from different data below and above the inflection point, respectively. (PDF) S2 Fig. The relationship between cadence and step length for all the distance runners. The two dashed lines show the regression lines computed from different data below and above the inflection point, respectively. (PDF) S3 Fig. The relationship between cadence and step length for the active athletes. The two dashed lines show the regression lines computed from different data below and above the inflection point, respectively. The title of each figure corresponds to each subject’s sports expe- rience. Characters in parentheses signify male or female subjects. (PDF) S4 Fig. The relationship between cadence and step length for the sedentary individuals. The two dashed lines show the regression lines computed from different data below and above the inflection point, respectively. In the sedentary group, three subjects were excluded from PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 10 / 12 data analysis: two subjects (No. 18 and No. 19) had estimated inflection point fell outside the range of the original data, and one subject (No. 20) showed two regression lines with almost the same slope giving the inflection point completely outside the range of measured data. Characters in parentheses signify male or female subjects. (PDF) Acknowledgments The authors thank Dr. Larry Crawshaw for English editing of the manuscript. Author Contributions Conceptualization: Yuta Goto, Tetsuya Ogawa, Gaku Kakehata, Kazuyuki Kanosue. Formal analysis: Yuta Goto, Naoya Sazuka, Yoshihiro Wakita. Funding acquisition: Yuta Goto. Investigation: Yuta Goto, Gaku Kakehata. Methodology: Yuta Goto, Tetsuya Ogawa, Naoya Sazuka, Yoshihiro Wakita. Project administration: Yuta Goto, Atsushi Okubo, Kazuyuki Kanosue. Software: Naoya Sazuka. Supervision: Kazuyuki Kanosue. Visualization: Yuta Goto. Writing – original draft: Yuta Goto, Naoya Sazuka, Yoshihiro Wakita. Writing – review & editing: Tetsuya Ogawa, Gaku Kakehata, Atsushi Okubo, Shigeo Iso, Kazuyuki Kanosue. References 1. Dillman CJ. Kinematic analyses of running. Exerc Sport Sci Rev. 1975; 3:193–218. PMID: 1175666 2. Luhtanen P, Komi P. Mechanical factors influencing running speed. In: Asmussen E, Jo¨rgensen K, edi- tors. Biomechanics VI-B. Baltimore: University Park Press.; 1978. p. 23–9. 3. Cavagna GA, Franzetti P, Heglund NC, Willems P. The determinants of the step frequency in running, trotting and hopping in man and other vertebrates. J Physiol. 1988; 399:81–92. https://doi.org/10.1113/ jphysiol.1988.sp017069 PMID: 3404473 4. Kaneko M. Mechanics and energetics in running with special reference to efficiency. J Biomech. 1990; 23 Suppl 1:57–63. https://doi.org/10.1016/0021-9290(90)90041-z PMID: 2081745 5. Cavagna GA, Willems PA, Franzetti P, Detrembleur C. The two power limits conditioning step fre- quency in human running. J Physiol. 1991; 437:95–108. https://doi.org/10.1113/jphysiol.1991. sp018586 PMID: 1890660 6. Hay JG. Cycle rate, length, and speed of progression in human locomotion. J Appl Biomech. 2002; 18 (3):257–70. 7. Weyand PG, Sternlight DB, Bellizzi MJ, Wright S. Faster top running speeds are achieved with greater ground forces not more rapid leg movements. J Appl Physiol (1985). 2000; 89(5):1991–9. 8. Hunter JP, Marshall RN, McNair PJ. Interaction of step length and step rate during sprint running. Med Sci Sports Exerc. 2004; 36(2):261–71. https://doi.org/10.1249/01.MSS.0000113664.15777.53 PMID: 14767249 9. Nummela A, Keranen T, Mikkelsson LO. Factors related to top running speed and economy. Int J Sports Med. 2007; 28(8):655–61. https://doi.org/10.1055/s-2007-964896 PMID: 17549657 10. Salo AI, Bezodis IN, Batterham AM, Kerwin DG. Elite sprinting: are athletes individually step-frequency or step-length reliant? Med Sci Sports Exerc. 2011; 43(6):1055–62. https://doi.org/10.1249/MSS. 0b013e318201f6f8 PMID: 20980924 PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 11 / 12 11. Minetti AE, Alexander RM. A theory of metabolic costs for bipedal gaits. J Theor Biol. 1997; 186 (4):467–76. https://doi.org/10.1006/jtbi.1997.0407 PMID: 9278722 12. Yanai T, Hay JG. Combinations of Cycle Rate and Length for Minimizing the Muscle Power Require- ment in Human Running. J Appl Biomech. 2004; 20(1):51–70. 13. Peyre-Tartaruga LA, Coertjens M. Locomotion as a Powerful Model to Study Integrative Physiology: Efficiency, Economy, and Power Relationship. Front Physiol. 2018; 9:1789. https://doi.org/10.3389/ fphys.2018.01789 PMID: 30618802 14. Peyre-Tartaruga LA, Dewolf AH, di Prampero PE, Fabrica G, Malatesta D, Minetti AE, et al. Mechanical work as a (key) determinant of energy cost in human locomotion: recent findings and future directions. Exp Physiol. 2021; 106(9):1897–908. https://doi.org/10.1113/EP089313 PMID: 34197674 15. Cavanagh PR, Williams KR. The effect of stride length variation on oxygen uptake during distance run- ning. Med Sci Sports Exerc. 1982; 14(1):30–5. https://doi.org/10.1249/00005768-198201000-00006 PMID: 7070254 16. de Ruiter CJ, Verdijk PW, Werker W, Zuidema MJ, de Haan A. Stride frequency in relation to oxygen consumption in experienced and novice runners. Eur J Sport Sci. 2014; 14(3):251–8. https://doi.org/10. 1080/17461391.2013.783627 PMID: 23581294 17. Connick MJ, Li FX. Changes in timing of muscle contractions and running economy with altered stride pattern during running. Gait Posture. 2014; 39(1):634–7. https://doi.org/10.1016/j.gaitpost.2013.07.112 PMID: 23948332 18. van Oeveren BT, de Ruiter CJ, Beek PJ, van Dieen JH. Optimal stride frequencies in running at different speeds. PLoS One. 2017; 12(10):e0184273. https://doi.org/10.1371/journal.pone.0184273 PMID: 29059198 19. Cavagna GA, Mantovani M, Willems PA, Musch G. The resonant step frequency in human running. Pflugers Arch. 1997; 434(6):678–84. https://doi.org/10.1007/s004240050451 PMID: 9305998 20. Dorn TW, Schache AG, Pandy MG. Muscular strategy shift in human running: dependence of running speed on hip and ankle muscle performance. J Exp Biol. 2012; 215(Pt 11):1944–56. https://doi.org/10. 1242/jeb.064527 PMID: 22573774 21. Kakehata G, Kobayashi K, Matsuo A, Kanosue K, Iso S. Relationship between subjective effort and kinematics/kinetics in the 50 m sprint. Journal of Human Sport and Exercise. 2020; 15(1):52–66. 22. Ivy JL, Withers RT, Van Handel PJ, Elger DH, Costill DL. Muscle respiratory capacity and fiber type as determinants of the lactate threshold. J Appl Physiol Respir Environ Exerc Physiol. 1980; 48(3):523–7. https://doi.org/10.1152/jappl.1980.48.3.523 PMID: 7372524 23. Neder JA, Stein R. A simplified strategy for the estimation of the exercise ventilatory thresholds. Med Sci Sports Exerc. 2006; 38(5):1007–13. https://doi.org/10.1249/01.mss.0000218141.90442.6c PMID: 16672856 24. Chen CWS, Chan JSK, Gerlach R, Hsieh WYL. A comparison of estimators for regression models with change points. Stat Comput. 2011; 21(3):395–414. 25. Deming W. Statistical adjustment of data. New York: Wiley (Dover Publications Edition, 1985)1943. https://doi.org/10.1126/science.97.2514.209 PMID: 17780136 26. Martin RF. General deming regression for estimating systematic bias and its confidence interval in method-comparison studies. Clin Chem. 2000; 46(1):100–4. PMID: 10620577 27. Tartaruga MP, Brisswalter J, Peyre-Tartaruga LA, Avila AO, Alberton CL, Coertjens M, et al. The rela- tionship between running economy and biomechanical variables in distance runners. Res Q Exerc Sport. 2012; 83(3):367–75. https://doi.org/10.1080/02701367.2012.10599870 PMID: 22978185 28. Minetti AE, Saibene F. Mechanical work rate minimization and freely chosen stride frequency of human walking: a mathematical model. J Exp Biol. 1992; 170:19–34. PMID: 1402610 29. Fischer G, Storniolo JLL, Peyre´-Tartaruga LA. Effects of Fatigue on Running Mechanics: Spring-Mass Behavior in Recreational Runners After 60 Seconds of Countermovement Jumps. Journal of Applied Biomechanics. 2015; 31(6):445–51. 30. Pantoja PD, Morin JB, Peyre-Tartaruga LA, Brisswalter J. Running Energy Cost and Spring-Mass Behavior in Young versus Older Trained Athletes. Med Sci Sports Exerc. 2016; 48(9):1779–86. https:// doi.org/10.1249/MSS.0000000000000959 PMID: 27116643 31. Cavagna GA, Legramandi MA, Peyre-Tartaruga LA. Old men running: mechanical work and elastic bounce. Proc Biol Sci. 2008; 275(1633):411–8. https://doi.org/10.1098/rspb.2007.1288 PMID: 18077249 32. Stiffler-Joachim MR, Wille C, Kliethermes S, Heiderscheit B. Factors Influencing Base of Gait During Running: Consideration of Sex, Speed, Kinematics, and Anthropometrics. J Athl Train. 2020; 55 (12):1300–6. https://doi.org/10.4085/1062-6050-565-19 PMID: 33064810 PLOS ONE Spatiotemporal inflection points in human running PLOS ONE | https://doi.org/10.1371/journal.pone.0258709 October 18, 2021 12 / 12

Spatiotemporal inflection points in human running: Effects of training level and athletic modality.

10-18-2021

Goto, Yuta,Ogawa, Tetsuya,Kakehata, Gaku,Sazuka, Naoya,Okubo, Atsushi,Wakita, Yoshihiro,Iso, Shigeo,Kanosue, Kazuyuki

eng

PMC3868388

Supporting Materials and Methods 1) The double Gaussian and double Lorentzian In order to estimate the two dates where the highest number of performances occurs for both thermal and cultural peak, we investigated two functions: (i) The double Gaussian function f1(x) that is the sum of two Gaussian functions: f1(x) = f1,1(x) + f1,2(x) (1) where f1(x) = a1 · exp −  x − b1 c1   2 +a2 · exp −  x − b2 c2   2 (2) (ii) The double Lorentz function f2(x) that is the sum of two Lorentzian functions: f1(x) = f2,1(x) + f2,2(x) (3) where f2(x) = c1 2 a1π 1 + x − b1 a1/2 2 + c2 2 a2π 1 + x − b2 a2/2 2 (4) The functions f1 and f2 are two-peak functions and x is the number of elite performances in a week. Resulting adjusted R2 and RMSE were gathered for both functions and the elected function for a given percent category was the function that presented the best statistics. 2) Estimates of x01, x02, p1, p2 For each elected function in each percent category, the two peaks x01, x02 were estimated, and the proportions p1, p2 were given by estimating the area under the curve in the interval [0, 52] for the functions f1,1, f1,2, f2,1, f2,2. For notation convenience, we denoted i and j as the indexes of the functions, such as when i = 1 and j = 1, fi,j referred to f1,1. Integration of the functions was given by: Z 52 0 fi,j(x) = Fi,j(52) − Fi,j(0) (5) where F1,j(x) = −√π × aj × cj × erf bj − x cj  (6) and F2,j(x) = −2 cj × tan−1 2 (bj − x) aj  π (7) where erf(x) is the error function and tan−1(x) the inverse tangent function. The proportion of performances in the two peaks was estimated by computing the area under the curve (proportion of Performances) of each elected model and for each PC: Z 52 0 f1(x) = Z 52 0 f1,1(x) + Z 52 0 f1,2(x) (8) Z 52 0 f2(x) = Z 52 0 f2,1(x) + Z 52 0 f2,2(x) (9) 1 And the proportions were calculated in percentages using: p1 = R 52 0 fi,1(x) R 52 0 fi(x) × 100 (10) p2 = R 52 0 fi,2(x) R 52 0 fi(x) × 100 (11) Where i = 1 or 2, depending on the elected function f1(x) or f2(x) at each PC. 2

Environment and scheduling effects on sprint and middle distance running performances.

11-20-2013

Haïda, Amal,Dor, Frédéric,Guillaume, Marion,Quinquis, Laurent,Marc, Andy,Marquet, Laurie-Anne,Antero-Jacquemin, Juliana,Tourny-Chollet, Claire,Desgorces, François,Berthelot, Geoffroy,Toussaint, Jean-François

eng

PMC3805569

The Influence of Sex, Stroke and Distance on the Lactate Characteristics in High Performance Swimming Benjamin Holfelder*, Niklas Brown, Dieter Bubeck Department of Sport and Exercise Science, University of Stuttgart, Stuttgart, Germany Abstract Background: In order to achieve world-class performances, regular performance diagnostics is required as an essential prerequisite for guiding high performance sport. In high performance swimming, the lactate performance diagnostic is an important instrument in testing the sport specific endurance capacity. Although the role of lactate as a signaling molecule, fuel and a gluconeogenic substrate is accepted, lactate parameters are discussed concerning stability, explanatory power and interpretability. Methods: We calculated the individual anaerobic threshold (IAT) of Bunc using the swimming-specific lactate threshold test by Pansold. Results: The cross-sectional analysis (ANOVA) of n = 398 high performance swimmers showed significant effects for sex, stroke and distance on the IAT, the percentage of personal best time on the IAT (% of PB on IAT) and maximal lactate values (max. bLA). For the freestyle events the IAT decreased, % of PB on IAT and max. bLA increased from 100 to 400 m significantly in men and women. Women showed significantly higher % of PB on IAT with descriptive lower IAT in 7 of 8 analyzed events. Men showed significantly higher max. bLA in 5 of 8 events. In the second step, the analysis of 1902 data sets of these 398 athletes with a multi-level analysis (MLA) showed also significant effects for sex, swimming distance and stroke. For initial status and development over time, the effect sizes for the variables distance and sex were medium to large, whereas for stroke there were no or small effect sizes. Discussion: These significant results suggest that lactate tests in swimming specifically have to consider the lactate affecting factors sex and distance under consideration of the time period between measurements. Anthropometrical factors and the physiology of women are possible explanations for the relative better performance for lower lactate concentrations compared to men. Citation: Holfelder B, Brown N, Bubeck D (2013) The Influence of Sex, Stroke and Distance on the Lactate Characteristics in High Performance Swimming. PLoS ONE 8(10): e77185. doi:10.1371/journal.pone.0077185 Editor: Jonatan R. Ruiz, University of Granada, Spain Received March 6, 2013; Accepted September 2, 2013; Published October 22, 2013 Copyright:  2013 Holfelder et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Funding: This work was supported by the German Research Foundation (DFG) within the funding programme Open Access Publishing. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing Interests: The authors have declared that no competing interests exist. * E-mail: benjamin.holfelder@inspo.uni-stuttgart.de Introduction In order to achieve maximum performance in important competitions, regular performance diagnostics is required as an essential prerequisite for guiding high performance sport [1]. It is used to determine the actual performance and thus enhance the planning and periodization of the training process [2], to recognize the athletes actual stress-recovery balance and integrate it into the training schedule. For analyzing the endurance capacity in swimming, measuring lactate is common, due to the difficult conditions for spirometric testing in a pool. However, lactate parameters are currently discussed concerning their stability, explanatory power, validity and interpretability, because factors like the training state, in particular overtraining [3], diet and nutritional status [4] and the types and sizes of muscle groups and fibers [5] are affecting the individual lactate kinetics. Although research on lactate is far away from complete [1,6] the role of lactate as a signaling molecule, fuel and a gluconeogenic substrate is accepted [5,7,8]. However, the determination of the individual anaerobic threshold (IAT) by means of lactate concentration is still a gold standard [9,10]. Besides, there is currently no adequate method in swimming to substitute the lactate diagnostic in the field, thus it seems important to increase the knowledge of lactate affecting factors before, during and after exercise to further optimize the interpretation [1]. Thus, this article evaluates the use of the IAT of Bunc et al. [11] on the lactate threshold test of Pansold [12], which is used for assessing the sport specific endurance capacity in the German Swimming Association (DSV), since the beginning of the 1990 s. The choice of the threshold concept of Bunc et al. [11] can be explained by the exponential function as a common functional basis with the Pansold-test and the calculation of the IAT regarding the characteristics of the whole lactate curve. The consideration of these seems important for more reliable statements in performance diagnostics, because thereby it is easier to differentiate between sprinters, endurance athletes and untrained people [4,13]. PLOS ONE | www.plosone.org 1 October 2013 | Volume 8 | Issue 10 | e77185 Influence of Sex Because of anthropometric, hormonal and genetic differences, sex is a major factor influencing best performances [14]. Specifically for swimming, several studies [15,16] reported that the technique of women is more economical than the technique of men. This could be explained by anthropometrical factors like body density, a lower hydrodynamic torque and the better ability to adapt to a horizontal body alignment [15,17]. For example, a higher body fat content, naturally observed in women [18], increases the prone gliding distance [19]. It can be assumed, that a more economical technique will cause lower lactate values of comparable load situations. However, could only identify sex- specific differences for the freestyle events, with greater post-race lactate concentrations in men. Crewther et al. [4] reported that men react to a bout of resistance training with higher lactate concentrations when compared to women. Thus, men exhibit a greater lean muscle mass and can train with heavier relative loads than women [4]. It was shown that performance differences between men and women decrease with increasing distance, also explained by physiological and morphological factors [21]. It seems that in women the aerobic metabolism and in men the anaerobic metabolism is better developed [22]. The higher content of muscle tissue and the better-developed anaerobic metabolism in men, result in higher lactate concentrations especially for 50 and 100 m events [22]. examined the muscle fiber type distribution in m. vastus lateralis of 140 healthy untrained subjects (55 women and 95 men) at the age between 19.0 and 23.9 years. No sex specific differences were found for the muscle fiber type distribution, but the area occupied by each type differs (women I.IIA.IIB; men IIA.I.IIB). In addition type IIA fibers were the largest in men, whereas type I fibers tended to be the largest in women [23]. From a physiological perspective, in an active state, glycolytic muscle fibers act as the main producers of lactate [5,7]. In contrast, oxidative fibers serve as lactate consumers [6], also enclosing the muscle fibers of the heart within the scope of the cell-to-cell lactate shuttle [6,7,24]. Influence of Stroke Most studies about lactate in swimming were conducted in freestyle; only a few studies analyzed the influence of the other strokes on lactate. Sawka et al. [25] found similar mean lactate concentrations after 200 yd races (182,88 m) for all strokes in 23 competitive athletes. Capelli et al. [26] measured the lactate concentration after maximal swim of 50 yd, 100 yd and 200 yd in 20 male college swimmers. The descriptive data, which bases on only 3 to 8 subjects per stroke, show different orders depending on the distance. Issurin et al. [27] reported the highest lactate concentrations in butterfly, followed by breaststroke, backstroke and freestyle across three different tests with 22 highly trained swimmers (14 male, 8 females). The study of Vescovi et al. [20] with 100 swimmers (50 male and 50 females) showed significantly lower post-race lactate concentrations for breaststroke compared to butterfly and backstroke in 50 and 100 m. Regarding the four different swimming strokes it seems to be clear that freestyle followed by backstroke show the most economic energy expendi- ture [15,26]. An explanation could be that freestyle and backstroke are characterized by a lower intracyclic variation of the swimming velocity compared to butterfly and breaststroke [16,26,28]. Butterfly and breaststroke are characterized by a gliding phase after the arm action, resulting in a greater relative loss of speed in every cycle but also underwater recoveries, especially in breast- stroke [29]. A classification between butterfly and breaststroke is unclear at present [15]. Though it could be supposed, that the economy of butterfly is the slightest on account of the high technical-coordinative demand. Especially the importance of the ability to coordinate arm and leg action for a rhythmical body motion and the high demand of potential energy raising the upper body out of the water seems to be key factors for an economic technique [30]. Although, at higher swimming speeds, breaststroke seems to be less economic [15]. An explanation could be, that breaststroke is the only stroke in which great body masses are moved against the swimming direction, which means that a lot of energy will be utilized to overcome the increased drag with increasing velocity [31]. Furthermore, breaststroke is character- ized by different styles of the flat and undulating technique, which influence the energy expenditure differently but also making it difficult to classify this stroke clearly [29]. Influence of Distance With higher swimming distance, the aerobic endurance capacity becomes more important [32]. Vescovi et al. [20] describe the post-race lactate concentrations of 50, 100, 200, 400, 800 and 1500 m events as an inverted U-shape pattern with similar concentrations for 100, 200 and 400 m. They also showed the highest post-race lactate concentrations after 200 m for backstroke and breaststroke. Similar results were shown in the study of Capelli [26], where the highest values were achieved in 200 yd for three of four strokes. From a physiological perspective, muscular power is highly determined by the muscle fiber type distribution [33]. The velocity and strength development of a muscle fiber is associated with the myosin heavy chain (MyHC) isoforms [34]. A higher content of type II fibers causes a bigger strength development [33,35], which is essential for sprinters. Because type IIA fibers act as main producers of lactate in an active state [5,7], an increased IAT is to be observed in sprinters with a larger amount of type II fibers. In contrast, it is supposed that longer distances require a training contribution with the trend towards achieving a maximum aerobic capacity with a greater content of type I fibers. Thus, higher training extents in low mean intensities are recommended, promoting a fiber shift towards the slower type I fibers. Type I fibers influence the lactate clearance positively [6]. This also explains, why with a higher endurance capacity the lactate curve shifts to the right [35]. Nevertheless, the right shift alone does not necessarily implicate an improvement in aerobic metabolism [1]. At a cellular level, the mitochondrial biogenesis seems to be important concerning the muscle fiber differentiation [36]. In oxidative fibers, the mitochondria can occupy 20–40% of the cell volume, whereas in glycolytic fibers only down to 1% of the cell volume is filled by mitochondria [37]. Within the scope of the intracellular lactate shuttle hypothesis, mitochondria have an important function for the lactate metabolism [7,24]. It has to be added, that the classification of the MyHC does not completely correlate with the oxidative capacity [38]. The work of supports this impression with swimming taking a special position. The overlappings at molecular and cellular level are reflected in competitions, with some athletes achieving world-class perfor- mances in several disciplines (e.g. 100–400 m swimming) with different performance profiles. To current knowledge, specific training has to be planned for each discipline, avoiding endurance and strength specific signaling pathways overlapping and thus reducing or even eliminating training effects [38,40,41]. Summa- rized, to understand the physiological adaptations as a result of specific training content seems to be very important for interpreting lactate tests [1]. Aims of the Study The first aim of this article is to improve the interpretability of lactate diagnostics in swimming by identifying lactate-affecting Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 2 October 2013 | Volume 8 | Issue 10 | e77185 Table 1. Information about the Pansold test protocol [42] (p. 169). distance number of steps number of repititions stroke recommended intensity for the first step in % of personal best time break between repititions break between steps lactate measurement men women 100 m 1 3 Bu 60–65 70–75 1 min 3 min directly 2 2 Ba 70–75 75–80 1 min 3 min directly 3 1 Br 70–75 80–85 5 min after 1 min 4 1 Fr 65–70 70–75 approx. 20 min after 1–3 min 5 1 increase of 3–4 s/per steplast step = maximum speed after 4, 7 and 10 min 200 m 1 3 Bu 70–75 75–80 1 min 3 min directly 2 2 Ba 75–80 80–85 1 min 3 min directly 3 1 Br 75–80 83–87 5 min after 1 min 4 1 Fr 75–80 80–85 approx. 20 min after 1–3. min 5 1 increase of 5–8 s/step. last step = maximum speed after 4, 7 and 10 min 400 m 1 1 Fr 80–85 85–90 3 min after 1 min 2 1 increase of 8–12 s/step. last step = maximum speed 5 min after 3 min 3 1 up to 30 min after 3 min 4 1 after 4, 7 and 10 min Abbreviations: Ba = Backstroke, Br = Breaststroke, Bu = Butterfly, Fr = Freestyle. doi:10.1371/journal.pone.0077185.t001 Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 3 October 2013 | Volume 8 | Issue 10 | e77185 variables. Hence, the influence of sex, distance and swimming stroke on the IAT, percentage of personal best time on IAT (% of PB on IAT) and maximum lactate (max. bLA) concentration is evaluated. The second aim of this study is to present the Multi- Level Analysis (MLA) as a statistical method which is able to analyze the typical data structure in high performance sports in a formally correct way. Materials and Methods Subjects This investigation is based on a retrospective analysis of lactate tests from the data pool of the Olympic Training Centre Hamburg/Schleswig Holstein (GER), Department of Training Science. Because the whole data set was collected by the Olympic Training Centre Hamburg/Schleswig Holstein, a unified data collection and assessment of qualified personnel is assumed. The 18-year data-collection period itself was not monitored, thus the exact procedure of data acquisition cannot be described here. In the analysis, 2063 data sets measured between 1992 and 2010 were examined. 1902 data sets of 398 athletes [female n = 170 (42.7%, age 16.9462.78 years, age range 13–26 years), male n = 228 (57.3%, age 19.1063.17 years, age range = 15–36 years) met the inclusion criteria. As an inclusion criterion, only data sets of athletes with personal bests around 700 points of the LEN point table (1000 points = world record for the period of validity) were included. Furthermore only data sets were examined in which the coefficients of determination were r2$0.92 (5 steps) or r2$0.95 (4 steps) [42]. The data were collected as part of the regular performance diagnostics of the swimming association. After consultation of the Ethics Committee of the University Tu¨bingen Medical School (GER), the retrospective and anonymous analysis of data of own patients, which where collected as part of diagnostics, therapy or therapy control need no guidance after the Professional Code for Physicians in Germany (115 (1)) and no informed consent of the patients. There are no concerns of the commission about collecting, processing and publishing such data. Test Protocols Lactate threshold test by pansold. The lactate test by Pansold [12,43] is a swimming-specific field test for diagnosing the endurance capacity, accounting for the different structures of swimming disciplines. This test protocol is used in the DSV since the beginning of the 1990 s. The test is carried out for 100 and 200 m disciplines in five steps, for 400 m in four steps in a 50 m pool, usually in the athletes’ main event. The load specification is determined by a percentage of the individual best, whereby the rest periods between the steps are fixed [42] (cf. table 1). The step duration is reduced from step to step, because of the constant distance and increasing swimming speed. After every step the lactate concentration of the capillary blood is measured with blood samples from the ear lobe. The lactate concentration of the last step (maximum speed) represents the highest value of the measurements after 4, 7 and 10 min. Therefore this lactate concentration represents the maximum individual lactate concen- tration of the test. The analysis of the lactate kinetic is based on the exponential function y = a*e(b*x) [y = lactate in mmol*L21; a = free coefficient; b = slope coefficient, x = speed in m/s]. For the calculation of the regression coefficients ‘‘a’’ and ‘‘b’’ a quasilinear regression analysis (method of the smallest squares) is carried out. In addition, the coefficient of determination is calculated, which provides information about the reliability of the Pansold-test. With five steps r2$0.92 (100 m and 200 m disciplines) and with four steps r2$0.95 (400 m disciplines) [42]. The parameters of the Pansold-function were calculated with MS Access 2007. The Individual Threshold Concept of Bunc et al. (1985) The IAT of Bunc et al. [11] is represented by the point in which the inclination of the lactate-load function changes the most. Faude et al. [44] report in a recent review, that there is a high correlation between the IAT of Bunc et al. and MLSS of r = 0.98– 0.99 in 16 healthy male runners and r = 0.89 in n = 22 healthy cyclists. In both cases, the running speed and the power output in watt at IAT are higher than at MLSS (+0,14–0,31 m/s/+71,5 W). For the calculation of the IAT, based on the lactate curve and the given exponential function, the following steps are recommended (cf. figure 1): 1. Tangent t1 to the point with the lowest load (y = a*e(b*x 1.step )) and tangent t2 to the point of 15 mmol/l (15 = a*e(b*x)) 2. Calculate intersection (S1) of both tangents 3. Angle bisector by the intersection of the tangents 4. Intersection between angle bisector and lactate curve (y = a*e(b*x)) represents the IAT of Bunc et al. [11]. The use of this threshold concept on the exponential function of the Pansold-test y = a*e(b*x) requires the extension of the Pansold- function with the lactate concentration of the first step (L1. step). This corresponds to the point of the lowest load in which the tangent t1 is calculated. The angle bisector is described by the graphical bisector with an axis relation between x-axis (km/h) to y- axis (lactate in mmol/l) from 1:2 [11]. The extrapolation of the IAT was carried out with MATLABH. Statistic Analysis All statistics were performed using SPSS (version 19.0 for Macintosh). For analyzing the influence of sex, swimming stroke and distance on the IAT, the percentage of the personal best time on IAT (% of PB on IAT) and on maximum lactate value (max. bLA), a three-factor analysis of variance (ANOVA) was conducted for each variable for 398 subjects. Because the correlations between the three variables were only weak, three ANOVAs were calculated instead of a MANOVA. The post hoc tests of Bonferroni (equal variances), Tamhane-T2 (unequal variances) were carried out for sex, stroke and distance specific analysis to test which means are significantly different from each other. T-Tests for two independent samples were used to evaluate differences in sex in every stroke and every distance. For the athletes with more than one data set, the first data set was used for the ANOVAs, representing the initial status for the second part of the analysis. In the second part of the statistic analysis all 1902 data sets, which met the inclusion criteria, were involved. The number of data sets for each event and sex are in table 2. Most athletes had several data sets with different in-between time intervals, therefore a Multi-Level Analysis (MLA) for IAT (Model 1), % of PB on IAT (Model 2) and max. bLA (Model 3) as the dependent variable was conducted. The advantage of this method is, that information about the change over time of the dependent variables under consideration of the in-between time intervals between measure- ments is given. Also valid data does not have to be excluded. Therefore, this method is more flexible and formally correct for such data structures. The comparison of the basic models (without predictors; df = 1) (A) random intercept and (B) random intercept & random slope with the help of the information criteria proved a significantly (p,.001) different change of the dependent variables of the athletes over time. Random intercept (A) is based on the assumptions that the subjects have different base values, but the Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 4 October 2013 | Volume 8 | Issue 10 | e77185 same rate of change over time. Random intercept & random slope (B) assumes different base values and different change rates over time. Comparing these two basic models by using information criteria helps to fit the most economical model with the highest statistical power. ‘‘The smaller the values of these criteria, the better the fit of the model’’ [45] (p. 218). Values of the information criteria are difficult to interpret, but differences larger than 10 are substantial [46]. There is a decrease in the information criteria from Model (A) to (B) for Log-Likelihood (-2logL), Akaike information criterion (AIC), and Bayesian information criterion (BIC) for all dependent variables .24. In addition model (B) reduces the variance by 5.7% for IAT, by 23.48% for % of PB on IAT and by 5.6% for max. bLA (level 1-pseudo R2) compared to model (A), which is why further analysis are conducted with model (B). At first, the model was calculated in each case individually with the time invariant predictors (fixed) sex (S), swimming distance (D) and swimming stroke (St) to check the explained between-individual variation for each predictor on the initial status of the IAT (level 2-pseudo R2 C) and the individual variation ( = slopes) of the dependent variables over time (level 2-pseudo R2 S) [46,47]. According to guideline and suggested in Kwok et al. [34], statements about the effect size can be made with the help of R2 changes (.02; .13; .26 representing a small, medium and large effect). Finally the following model with all three predictors for each dependent variable separately was calculated. Level 1 Yi~b0izb1izeij ð1Þ Level 2 b0i~b0(S,D,St)izb2(S,D,St)izv0i ð2Þ b1i ~ b1(S,D,St)izv1i ð3Þ Integrated formula: Yij~b0(S,D,St)izb2(S,D,St)izv0i zb1(S,D,St)izv1izeij ð4Þ Level 1 represents athletes IAT/% of PB on IAT/max. bLA at different measuring times. Level 2 represents the different subjects (n = 398) taking the fixed predictors sex (S), distance (D) and stroke (St) into account. The alpha level of the tests was set to p,0.05. Figure 1. Graphical Determination of the IAT by Bunc et al. [11] with the Pansold-function y = a*e(b*x) [42,43]. doi:10.1371/journal.pone.0077185.g001 Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 5 October 2013 | Volume 8 | Issue 10 | e77185 Results ANOVAs In all three ANOVAs no significant interaction effects were found. IAT: The ANOVA with IAT [mmol*L21] as dependent variable showed significant effects for sex F(1, 382) = 7.88, p = .005, par. g2 = .020, distance F(2, 382) = 10.98, p,.001, par. g2 = .054 and stroke F(3, 382) = 6.76, p,.001, par. g2 = .050. Overall, 100 m differed significantly from 200 and 400 m (p,.001). Butterfly differed significantly from all other strokes (p,.001). The descriptive data showed lower lactate concentra- tions on IAT for women in all events, with only 200 m freestyle being significant (p = .05; cf. table 3). Sex specific analysis for stroke per distance showed significant differences for butterfly and breaststroke (p,.001), freestyle (p = .002) and tending to be significant for backstroke (p = .076) in the males 100 m events. The highest mean values were achieved in butterfly, followed by backstroke, freestyle and breaststroke. The sex specific analysis for stroke showed significant differences (p,.001) between 100 and 400 m, as well as between 200 and 400 m freestyle, each for men and women. % of PB on IAT/% of PB. The ANOVA with % of PB on IAT as dependent variable showed significant effects for sex F(1, 382) = 40.58, p,.001, par. g2 = .096, distance F(2, 382) = 88.36, p,.001, par. g2 = .316 and stroke F(3, 382) = 3.60, p = .014, par. g2 = .028. Overall, 100 m differed significantly from 200 and 400 m, as well as 200 from 400 m (p,.001) with higher values in longer distances. Significant differences were found between butterfly and backstroke/freestyle (p,.001), backstroke and breaststroke (p = .006), as well as between freestyle and breast- stroke (p,.001). Under consideration of the mean values and confidence intervals (cf. table 3) the lowest values were produced in butterfly, the highest for the freestyle events. Women showed significantly higher % of PB on IAT (t-test) with descriptive lower IAT in 7 of 8 events. There was also a significant difference (p = .039) for % of PB on IAT between 100 m butterfly and backstroke in women, with lower mean values in the butterfly event. For both sexes, the means are significantly higher (p,.001) with increasing distance for the freestyle events. The same effect occurred between 100 and 200 m breaststroke (p = .015) in women. The descriptive data showed (cf. table 4) that women achieve a higher % of PB with lower bLA compared to men in Table 2. Attribution of the 1902 data sets of n = 398 athletes for the multi level analysis (MLA) for each event and sex. 100 m Bu 100 m Ba 200 m Ba 100 m Br 200 m Br 100 m Fr 200 m Fr 400 m Fr men 66 97 120 124 78 194 347 119 women 19 66 67 75 76 145 232 77 Abbreviations: Ba = Backstroke, Br = Breaststroke, Bu = Butterfly, Fr = Freestyle. doi:10.1371/journal.pone.0077185.t002 Table 3. Differences in IAT of Bunc et al. [11] [mmol*L21] between sexes given as means (M), standard deviations (SD) and confidence intervals of the lactate step test of Pansold and the percentage of the personal best time on the IAT (% of PB on IAT) for male (M) and female (F) of n = 398 athletes. IAT [mmol*L21] % of PB on IAT 95% Confidence Interval 95% Confidence Interval Event sex n M ± SD Lower Bound Upper Bound p M ± SD Lower Bound Upper Bound p 100 m Bu M 15 7.5261.69 6.70 8.59 .35 86.3964.92 83.88 89.04 .87 F 6 6.7861.35 5.63 7.89 86.0661.92 84.18 87.40 100 m Ba M 15 6.4361.25 5.82 7.09 .22 87.2662.81 85.72 88.64 .02* F 13 5.8761.07 5.29 6.52 89.9662.81 88.54 91.57 200 m Ba M 19 5.6661.12 5.19 6.22 .26 88.2661.69 87.48 89.01 .001** F 16 5.1661.48 4.39 5.89 91.6063.48 90.19 93.67 100 m Br M 29 5.9661.09 5.58 6.39 .19 85.6763.40 84.42 86.88 .02* F 13 5.4461.31 4.76 6.14 88.3663.54 86.40 90.18 200 m Br M 20 5.6460.87 5.29 6.03 .06 86.9362.94 85.66 88.22 ,.001*** F 18 5.1360.74 4.81 5.48 91.2661.74 90.40 92.15 100 m Fr M 44 6.1960.99 5.92 6.49 .06 84.9362.83 84.14 85.80 ,.001*** F 35 5.7661.01 5.43 6.10 88.3262.61 87.46 89.17 200 m Fr M 58 5.7060.95 5.46 5.96 .05* 87.9662.61 87.32 88.70 ,.001*** F 50 5.3361.04 5.05 5.63 91.3962.26 87.46 89.17 400 m Fr M 28 5.1961.13 4.77 5.63 .96 92.5661.67 91.90 93.20 .001** F 19 5.2161.16 4.66 5.74 94.6162.41 93,52 95.73 p = .05*, p = .001**, p,.001***. Abbreviations: Ba = Backstroke, Br = Breaststroke, Bu = Butterfly, Fr = Freestyle. doi:10.1371/journal.pone.0077185.t003 Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 6 October 2013 | Volume 8 | Issue 10 | e77185 nearly all submaximal steps. In the last step, the % of PB values differ significant between men and women only for 200 m breaststroke (p = .041). Max. bLA. The ANOVA with max. bLA [mmol*L21] as dependent variable offered significant effects for sex F(1, 382) = 17.74, p,.001, par. g2 = .044, distance F(2, 382) = 49.43, p,.001, par. g2 = .206 and stroke F(3, 382) = 8.36, p,.001, par. g2 = .062. Overall, 100 m differed significantly from 200 and 400 m, as well as 200 from 400 m (p,.001) with lower values in longer distances (cf. table 4). For the strokes, there was a significant difference (p = .009) between breaststroke and backstroke, with higher mean values for the backstroke events. For the 100 m events, significant differences were found between freestyle and butterfly (p = .009), as well as between freestyle and breaststroke (p = .001), but only in men. Thereby the values in freestyle were the highest. For both sexes, the mean values were significantly lower (p = ,.001 to.03) with increasing distance for the freestyle events. The comparison of max. bLA for each stroke and sex showed significant differences between men and women for 200 m backstroke (p = .041), 200 m breaststroke (p = .001) and for all freestyle events (p = .001, .002 &.002), with constant higher values in men. Multi-Level Analysis (MLA) The Multi-Level Analysis (MLA) for the predictors sex, stroke and distance is significant in each case (p,.001). The calculation of the Level 2-pseudo-R2 C to check the individual within variation for each predictor showed a reduction of variance for sex [13.3% (IAT), 21.1% (% of PB on IAT) and 17.7% (max. bLA)], for distance [22.5% (IAT), 40.1% (% of PB on IAT) and 22.1% (max. bLA)] and stroke [6.4% (IAT), 1.1% (% of PB on IAT) and 0% (max. bLA)]. Consequently, according to Cohen [35], predomi- nantly medium effect sizes exist for the predictors sex and distance in initial status, whereas only a small effect exists for stroke on IAT. The Level 2-pseudo-R2 S showed a reduction of the variance of the slopes for sex [1.4% (IAT), 13% (% of PB on IAT) and 28% (max. bLA)], distance [13.1% (IAT), 52.4% (% of PB on IAT) and 32.6% (max. bLA)] and stroke [7.6% (IAT), 10.4% (% of PB on IAT) and 1% (max. bLA)]. The models with all three predictors each (S, D, St) proves a reduction of variance of in initial status of 36.6% for IAT, 60.3% for % of PB on IAT and 39.9% for max. bLA (Level 2-pseudo- R2 C). A reduction of variance was shown for the slopes of 13.6% for IAT, 62.5% for % of PB on IAT and 55.8% for max. bLA (Level 2-pseudo-R2 S). For IAT (Model 1, cf. table 5) the time intervals between measurements have a significant influence (p = .005) and the athletes showed different change rates over time (b = 7.95E28, p = .005), measuring a slight decrease (b = 1.19E24, p,.009). The IAT of the women (men set to 0) are substantially lower (b = 20.47, p,.001), confirming the descriptive results for the 398 data sets (cf. table 3). The factor distance also shows decreasing IAT with increasing length (b = 20.31, p,.001; 100 m set to 0). There is also a significant effect for IAT for stroke (b = 20.14, p,.001; butterfly set to 0), with the highest values for butterfly. The effect of stroke is difficult to interpret, because this variable is not ordinal scaled and consists of more than two strokes. For % of PB on IAT (Model 2, cf. table 5) there is a significant influence of the time intervals between measurements (p,.001) and the athletes showed different change rates over time (b = 1.638, p,.001) with a slight decrease (b = 1.08E-3, p,.001). The % of PB on IAT are significantly higher for women (b = 3.06, p,.001; men set to 0). For the factor distance the % of PB on IAT increases with increasing distance (b = 3.14, p,.001; 100 m set to 0). The effect of stroke is also significant (b = 20.23, p = .025; butterfly set to 0). For max. bLA (Model 3) the different time periods between measurements are not significant (p = .057), but the change rates Table 4. Blood lactate concentrations (bLA [mmol*L21]) and percentage of individual best time (% of PB) for each step. step 1 step 2 step 3 step 4 step 5 event sex bLA [mmol*L21] % of PB bLA [mmol*L21] % of PB bLA [mmol*L21] % of PB bLA [mmol*L21] % of PB bLA [mmol*L21] % of PB 100 vm Bu M 3.9862.21 72.8164.10 4.9862.61 76.9162.96 6.2562.60 82.0763.41 8.1462.63 86.9063.60 10.2162.11 93.6062.37 F 3.4561.49 73.6564.70 4.6061.52 78.9061.83 6.0361.67 83.8861.09 7.8061.86 88.7661.72 10.3562.23 93.4062.56 100 m Ba M 2.9961.29 76.6163.63 4.1861.52 80.5563.38 5.5061.75 84.9162.62 7.6661.64 89.3062.69 12.0162.41 95.6762.46 F 2.4660.96 80.6163.55 3.9161.60 84.5864.67 5.5962.51 86.1864.57 7.9863.39 88.2465.23 11.2463.41 95.6462.56 200 m Ba M 2.2361.08 80.1161.91 3.2161.46 83.3161.88 4.6761.84 86.6061.91 6.9662.10 89.6461.79 11.2462.21 93.7062.21 F 1.9961.14 83.4162.36 2.6961.45 85.9062.30 3.6461.75 88.6262.70 5.5362.40 91.4062.66 9.0763.74 94.5762.16 100 m Br M 2.5461.01 74.5863.72 3.7061.40 78.4062.60 5.0761.77 82.8562.45 7.3662.30 86.8262.96 10.3062.35 92.9062.72 F 2.1661.06 78.9962.75 2.9461.18 81.7262.58 4.3961.43 85.5962.90 6.6962.01 89.5663.54 9.0862.54 93.2163.82 200 m Br M 2.2460.77 79.0963.11 3.2161.10 82.1463.35 4.5961.65 85.2163.26 6.7862.07 88.0263.16 10.2061.76 91.7662.84 F 1.8460.57 84.8061.97 2.8660.86 87.3462.04 4.2061.15 89.7962.04 5.7861.34 91.6461.75 7.9962.10 93.8161.73 100 m Fr M 2.7061.01 72.4663.49 3.8961.34 77.2963.43 5.1561.78 82.6663.70 7.9962.55 87.7863.61 12.6562.58 95.1962.44 F 2.3561.01 75.2464.36 3.5261.29 81.7763.32 4.8461.64 85.9663.24 7.2361.93 90.0763.40 10.8262.26 95.4062.53 200 m Fr M 2.3060.85 79.3962.90 3.3161.10 82.8062.44 4.6861.36 86.4462.36 6.9362.14 89.5562.68 10.9762.46 94.5062.61 F 2.0360.85 83,.562.78 2.9861.27 86.2862.71 4.1361.63 89.3362.74 6.0162.10 91.6562.49 9.4362.61 95.2862.32 400 m Fr M 1.9560.93 85.5162.29 2.9361.16 88.5462.29 4.5561.28 91.2662.26 8.2061.92 95,5362,66 F 1.9660.89 87.3661.67 2.7661.06 90.1562.05 4.2061.79 94.9462.52 6.0862.19 94,9562,52 Abbreviations: Ba = Backstroke, Br = Breaststroke, Bu = Butterfly, Fr = Freestyle. doi:10.1371/journal.pone.0077185.t004 Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 7 October 2013 | Volume 8 | Issue 10 | e77185 over time are significant (b = 4.16E24; P,.001). The results confirm the descriptive results of the 398 data sets. Women (men set to 0) show significantly lower max. bLA (b = 1.71, p,.001). For distance (100 m set to 0) there are lower bLA values for longer distances (b = 1.45, p,.001). The differences of max. bLA for the variable stroke (butterfly set to 0) are significant (p = .039) but slight (b = 0.19). Discussion The present study examined the influence of sex, stroke and distance on the IAT, the % of PB on IAT and the max. bLA in high performance swimming. Furthermore the MLA was used as a method, which is able to analyze typical data structures in high performance sports. Compared to other studies [9,49] the calculated IAT in this study (range of means: 5.19– 7.52 mmol*L21) seem to be very high. Otherwise, Dekerle and Pelayo [50] described, that lactate thresholds in swimming occur at speeds up to 90% of the 200 m pace, which is the case in most events (cf. table 3) in this study. Furthermore, Faude et al. [44] reported higher running speeds/power output in watt at IAT than at MLSS using the threshold concept of Bunc et al. [11]. MLSS concentrations are reported for swimming up to 3–5 mmol*L21 [50]. Therefore the calculated IATs seem to be realistic with the used threshold concept. Furthermore, the differences show that the IAT is strongly dependent on the applied method [1,50]. Irrespective of the mean values of the IAT, it was not the aim of the study to give recommendations for training intensities on basis of the IAT, but to identify lactate-affecting factors. According to current research, the lack of evidence for the effect of threshold training [51], the positive findings of high intensity training (HIT) on the endurance capacity [52,53], conceptions like the polarized training model [54] lead to a critical perspective on the ‘‘classical’’ threshold training. The results for the three independent variables are discussed in the following paragraphs. Sex The descriptive data (cf. table 3) and estimations of the MLA (cf. table 5) show in average 0.4–0.6 [mmol*L21] (b = 20.47; p,.001) lower IAT and 1–2 [mmol*L21] lower max. bLA (b = 21.70; p,.001) for women compared to men. These effects are significant, both in ANOVA and MLA, with medium effect sizes. For the IAT these sex specific differences are only significant for 200 m freestyle, whereas the differences for max. bLA are significant for five events (200 m Ba & Br and the freestyle events). These findings are similar to the results of Vescovi et al. [20] where sex specific differences in post-race bLA were only found for the freestyle events. With women reaching lower lactate concentrations in swimming, being in agreement with the findings of for resistance training. Together with muscle mass, the area occupied by muscle fiber types and the size of the fibers types seem to be sex specific, described by for the m. vastus lateralis. This could provide an explanation on a physiological level. With the metabolic characteristics of the muscle fiber types, the connection to the lactate kinetic is given [5,7,13]. The overall lower lactate concentrations on IAT and max. bLA in women support the idea of a better developed aerobic metabolism in women compared to men [21,22]. For % of PB on IAT, both statistical methods showed significant influences for sex (p,.001) with higher values Table 5. Estimates of Fixed Effects of the multi level analysis (MLA, random intercept & random slope) for IAT [mmol*L21], % of PB on IAT and max. bLA [mmol*L21]. Model 1: IAT [mmol*L21] Parameter Estimate SE df t p 95% Confidence Interval Lower Bound Upper Bound Intercept 6.35 0.09 644.33 67.96 .000 6.16 6.54 Time* 21.19E-4 4.44E-5 71.62 22.67 .009 22.07E-4 23.02E-5 Sex 20.47 0.08 313.60 26.16 .000 20.62 20.32 Distance 20.31 0.05 841.46 26.25 .000 20.40 20.21 Stroke 20.14 0.04 728.85 23.80 .000 20.21 20.07 Model 2: % of PB on IAT Intercept 86.00 0.27 893.37 322.80 .000 85,47 86.52 Time* 21.07E-3 1.45E-4 79.86 27.38 .000 21,36E-3 27.83E-4 Sex 3.06 0.24 417.17 12.70 .000 2,58 3.53 Distance 3.14 0.13 1419.17 23.69 .000 2,88 3.40 Stroke 20.23 0.10 1221.69 22.25 .025 20,43 20.03 Model 3: max. bLA [mmol*L21] Intercept 11.49 0.23 787.76 49.41 .000 11.03 11.94 Time* 4.16E-4 8.60E-5 47.06 4.83 .000 2.42E-4 5.89E-4 Sex 21.70 0.20 368.90 28.60 .000 22.10 21.32 Distance 21.45 0.12 1115.11 212.22 .000 21.69 21.22 Stroke 20.19 0.09 973.45 2.07 .039 9.83E-3 0.37 *Time = number of days between the measuring times of the athlete. doi:10.1371/journal.pone.0077185.t005 Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 8 October 2013 | Volume 8 | Issue 10 | e77185 for women (b = 3.06). These differences are also significant between men and women in 7 of 8 events (cf. table 3). The results are supported by, reporting a more economic swimming of women, explaining the lower lactate concentrations and the higher % of PB values at submaximal intensities. Other reasons could be a greater proportion of fatty tissue and different distribution in women compared to men [18]. This can give women a higher net buoyancy [55]. Furthermore, Caspersen et al. [18] reported lower added mass in women as a result of differences of body shape, which seems to be positive for the drag. Stroke For both statistical methods the effects for stroke is significant for all three dependent variables. Although the influence of the stroke is significant in each Model of the MLA (cf. table 5), there are only little reductions of variance for initial status of 0–6,4% (Level 2-pseudo-R2 C) and for the slopes of 1–11,4% (Level 2- pseudo-R2 S), meaning no to small effect sizes for stroke. For the max. bLA, the post-hoc analysis (Bonferroni adjusted) showed only a significant difference (p = .009) between breaststroke and backstroke with higher values for backstroke. These findings support the results of other studies [25,26] which show no stroke differences/different orders for bLA depending on the distance after maximal swim. Other studies [20,27] reported the highest max. bLA for butterfly, but none of these studies analyzed the influence of stroke on the IAT. Overall the highest IAT were achieved in 100 m butterfly, which differs significantly from all other strokes (p,.001, Bonferroni adjusted). The MLA confirms this results with the highest IATs in butterfly (b = 20.14; p,.001). There are also the lowest % of PB on IAT for butterfly (b = 20.23; p,.025), which differs significantly from backstroke and freestyle (p,.001, Bonferroni adjusted). A possible explanation could be, that around the IAT swimming speed is not at maximum, the statement of Barbosa et al. [15] could be confirmed in terms of economy, with butterfly showing the slightest swimming economy at lower swimming speeds. However the economy improves with increasing speed [26]. The highest IAT was found for 100 m butterfly, presumably connected with the fact that it seems to be difficult to swim butterfly with low intensities (first steps) [26], because of the high demand of interaction between strength and coordination [30,56]. On the other hand, the butterfly technique entirely makes high demands for strength, thus low intensities (in %) could be a high individual exposure [30]. Therefore, the glycolytic muscle fibers could be recruited at an early stage [57,58], operating as main lactate producers when recruited [5,7], explained by the size principle of motor unit recruitment [58,59]. However these results have to be interpreted carefully because of the small data sets for butterfly. Anyhow, the statements of Barbosa et al. [15] are not confirmed by the descriptive results of the IAT and the max. bLA for the breaststroke events. The lactate concentrations of the breaststroke events are the lowest on average in direct comparison with the same distance for the other strokes. Regarding the 200 m events, the IATs are in a similar zone for both sexes in each case, although the energy consumption in freestyle and backstroke seems to be lower, because of the lower intracyclic variation of swimming velocity compared to butterfly and breaststroke [16,28]. Summarized, for the variable stroke the biggest effects occurred for butterfly on the IAT for the 100 m events, with the significantly highest values. Overall the stroke seems not to play a key role in terms of affecting lactate parameters. The knowledge about the economy [16,28] of the strokes is reflected only partly in our results. Distance For distance all ANOVAs and Models of the MLA were significant (p,.001) for the three independent variables IAT, % of PB on IAT and max. bLA. The MLA show reductions of variance for initial status between 22.1 to 40.1% (Level 2-pseudo-R2 C) and for the slopes between 13.1 to 52.4% (Level 2-pseudo-R2 S), meaning mostly large effect sizes. The descriptive data, the ANOVAs and MLA showed significant decreasing IAT (b = 20.31; p,.001), increasing % of PB on IAT (b = 3.14; p,.001) and decreasing max. bLA (b = 21.45; p,.001) with increasing distance. These general results were confirmed by sex and stroke specific analysis only for the freestyle events (Bonferroni adjusted) in men and women. The results are not surprising, because with increasing distance the aerobic capacity becomes more important [32]. From a practitioner perspective a certain versatility of 100 and 200 or 200 and 400 m seems to be attractive to qualify for the 4*100 m and 4*200 m freestyle relays in international competitions, because five to six places are awarded. For the individual events at most two athletes can qualify for each country, therefore the chances to qualify but also the performance density are much higher in the 100 m and 200 m freestyle events. Although Meckel et al. [39] described that swimming is taking a special position in order to achieve world-class performances in various events with different performance profiles, but priorities in the training content are necessary from current point of view [22], which are reflected in the significant differences in the freestyle events. According to actual knowledge it is known, that signaling pathways of endurance training mainly trigger transcriptional changes, while weight training adaptations are caused by changes in mRNA-translation [41]. Concerning this, it is unclear whether signaling pathways initiated by weight training or endurance training overlap or hinder each other [38,40,41]. Therefore it is not completely clarified at the moment what this specifically means for training periodization, thus e.g. to coordinate strength and endurance training over the season in a synergetic way [59] is important to achieve the best performance in the individual best events on the main competition. Therefore it can be supposed that on the one hand the constitutional conditions and anthropome- trical and morphological factors [60] influences the ‘‘choice’’ of the individual special event (sprints or longer distances) and on the other hand the individual specificity is tried to be maintained or optimized in the training process. Consequently a sprinter (50– 100 m events) will try to maintain the dominance of muscle fiber type II and train in terms of an ‘‘optimum’’ endurance capacity [54]. An ‘‘optimum’’ endurance capacity means that the quality of training, which implies the training intensity and the exercise tolerance, becomes the key factor and not the volume in kilometers [32]. This point of view confirms the positive effects of high intensity training on the endurance capacity [52,53] and the concept of the polarized-training model (about 75% of the training below threshold intensity, only 5–10% in threshold intensity and 15–20% above) versus the ‘‘classical’’ threshold-training model [54]. This implements avoiding a shift towards type I fibers by excessive endurance training stimuli, which seems to be only partly reversible [33,34,61]. Nevertheless, in swimming, current training regimes seem to be characterized by an aerobic predominance [39], which is possibly not up to date anymore for sprint and middle distances. Methodical Discussion The large number of data sets of athletes at highest performance level supports the explanatory power of the results. It is to be mentioned, that the threshold concept of Bunc et al. [11] is not swim-specific. However, the Pansold-test is characterized by a Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 9 October 2013 | Volume 8 | Issue 10 | e77185 step-shaped load protocol, which is carried out under field conditions. Therefore a reliable realization of the Pansold-test requires a good sense of time of the swimmers with the ability to swim evenly, particularly in the first steps. Short-term speed increases or to quick beginning speeds can lead to high first-step lactate concentrations. This lactate concentration influences the further course of the test and the lactate curve [44]. However this is accepted for the examination of the endurance capacity in the field accounting for the different structures of swimming disciplines. A methodical strength of this study is the statistical analysis with a MLA. This method allows analyzing different numbers of data sets for an athlete with different time intervals between measurements [46]. For elite sport, it seems to be important to consider the time periods between performance diagnostics to make statements about e.g. the effect of training input on the lactate kinetics without excluding data. For that reason it was possible to analyze all 1902 valid data sets in the second part. Furthermore it is common, that some athletes are part of the squad for longer time periods, experiencing more performance diagnostics than other athletes. A disadvantage of this method is, that detailed information about descriptive data cannot be provided because of the different numbers of data sets for each athlete. A limitation of this retrospective analysis of data between 1992–2010 is, that the equipment used for the lactate diagnostics and the exact process of acquisition is unknown. Conclusion In conclusion, we identified the influence, especially of sex and distance, on lactate parameters in swimming and tried to explain them with current physiological knowledge. Furthermore we showed the importance of considering the different time periods between measurements in a formally correct way by using the MLA for general statements on basis of large data sets in high performance sports. The slight but significant influences of the time periods between measurements show the dynamic and sensitivity of the lactate molecule. These findings may help interpreting results of lactate tests in context of e.g. strength parameters or as a consequence of specific training content explained by physiological adaptations (e.g. metabolism of fiber types) or sex specific factors like body density or a better developed aerobic/anaerobic metabolism. Men overall showed higher IAT and max. bLA lactate concentrations compared to women. Whereas in submaximal intensity women achieved higher % of PB with lower lactate concentrations compared to men, which confirms a more economical technique [15,16] and a better developed aerobic metabolism [22] in women. For the variable stroke, the MLA showed significant results but no or small effect sizes. Therefore, when comparing inter- and intraindividual results of lactate tests in swimming, especially sex and distance specific lactate parameters have to be considered for initial status measurements and the development over time. In particular, longitudinal comparisons under steady conditions, which means applying the same test protocol and threshold concept, could benefit from this [1,51]. Acknowledgments We would like to thank the Olympic Centre Hamburg/Schleswig-Holstein (Department of Training Science) for the data sets, in particular Janina Gerkens and Ronald Berndt. Special thanks to Jochen Schweikert and Martin Keh for the support on the mathematical conversion. Author Contributions Analyzed the data: BH NB. Contributed reagents/materials/analysis tools: BH NB DB. Wrote the paper: BH NB DB. References 1. Olbrecht J (2011) Lactate Production and Metabolism in Swimming. In: Seifert L, Chollet D, Mujika I, editors. World Book of Swimming. From Science to Performance. New York: Nova: pp. 255–276. 2. Kiely J (2012) Periodization paradigms in the 21st century: evidence-led or tradition-driven? Int J Sports Physiol Perform 7(3): 242–250. 3. Bosquet L, Le´ger L, Legros P (2001) Blood lactate response to overtraining in male endurance athletes. Eur J Appl Physiol 84(1–2): 107–114. 4. Crewther B, Cronin J, Keogh J (2006) Possible stimuli for strength and power adaptation: acute metabolic responses. Sports Med 36(1): 65–78. 5. Philp A, Macdonald AL, Watt PW (2005) Lactate–a signal coordinating cell and systemic function. J exp biol 208(Pt 24): 4561–4575. 6. Gladden LB (2008) A lactatic perspective on metabolism. Med Sci Sports Exerc 40(3): 477–485. 7. Gladden LB (2004) Lactate metabolism: a new paradigm for the third millennium. J Physiol 558(Pt 1): 5–30. 8. van Hall G (2010) Lactate kinetics in human tissues at rest and during exercise. Acta Physiol (Oxf) 199(4): 499–508. 9. Pyne DB, Lee H, Swanwick KM (2001) Monitoring the lactate threshold in world-ranked swimmers. Med Sci Sports Exerc 33(2): 291–297. 10. Di Michele R, Gatta G, Leo A Di, Cortesi M, Andina F, et al. (2012) Estimation of the anaerobic threshold from heart rate variability in an incremental swimming test. J Strength Cond Res 26(11): 3059–3066. 11. Bunc V, Heller J, Novak J, Leso J (1985) Determination of the individual anaerobic threshold. Acta Universitatis Carolinae Gymnica 21(1): 73–81. 12. Pansold B, Zinner J (1994) Die Laktat-Leistungs-Kurve - ein Analyse und Interpretationsmodell der Leistungsdiagnostik im Schwimmen. In: Clasing D, Weicker H, Bo¨ning D (editors). Stellenwert der Laktatbestimmung in der Leistungsdiagnostik. Stuttgart: Fischer: p. 47–64. 13. Holfelder B, Bubeck D (2012) Theoretische Betrachtungen u¨ber die Trainings- steuerung anhand des Laktatstoffwechsels und der Muskelfasertypisierung. Schweiz Z Sportmed Sporttraumatol 60(1): 32–39. 14. Thibault V, Guillaume M, Berthelot G, El Helou N, Schaal K, et al. (2010) Women and men in sport performance: The gender gap has not evolved since 1983. J Sport Sci Med 9: 214–223. 15. Barbosa TM, Fernandes R, Keskinen KL, Colac¸o P, Cardoso C, et al. (2006) Evaluation of the energy expenditure in competitive swimming strokes. Int J Sport Med 27(11): 894–899. 16. Barbosa TM, Fernandes RJ, Keskinen KL, Vilas-Boas JP (2008) The influence of stroke mechanics into energy cost of elite swimmers. Eur J Appl Physiol 103(2): 139–149. 17. Onodera S, Miyachi M, Yano H, Yano L (1999) Effects of buoyancy and body density on energy cost during swimming. In: Keskinen KL, Komi PV, Hollander AP, editors. Biomechanics and Medicine in Swimming. VIII. Jyvaskyla: Gummerus Printing: 355–358. 18. Caspersen C, Berthelsen PA, Eik M, Paˆkozdi C, Kjendlie P-L (2010) Added mass in human swimmers: Age and gender differences. J Biomech 43(12): 2369– 2373. 19. Barbosa TM, Costa MJ, Morais JE, Moreira M, Silva AJ, et al. (2012) How Informative are the Vertical Buoyancy and the Prone Gliding Tests to Assess Young Swimmers’ Hydrostatic and Hydrodynamic Profiles? J Hum Kinet 32: 21–32. 20. Vescovi JD, Falenchuk O, Wells GD (2011) Blood lactate concentration and clearance in elite swimmers during competition. Int J Sports Physiol Perform 6(1): 106–117. 21. Chatterjee S, Laudato M (1995) Gender and performance in athletics. Soc Biol 42(1–2): 124–132. 22. Stanula A, Maszczyk A, Roczniok R, Pietraszewski P, Ostrowski A, et al. (2012) The development and prediction of athletic performance in freestyle swimming. J Hum Kinet 32: 97–107. 23. Staron RS, Hagerman FC, Hikida RS, Murray TF, Hostler DP, et al. (2000) Fiber type composition of the vastus lateralis muscle of young men and women. J Histochem Cytochem 48(5): 623–629. 24. Brooks GA (2009) Cell-cell and intracellular lactate shuttles. J Physiol 587(Pt 23): 5591–5600. 25. Sawka MN, Knowlton RG, Miles DS, Critz JB (1979) Post-competition blood lactate concentrations in collegiate swimmers. Eur J Appl Physiol Occup Physiol 41(2): 93–99. 26. Capelli C, Pendergast D, Termin B (1998) Energetics of swimming at maximal speeds in humans. European Eur J Appl Physiol Occup Physiol 78: 385–393. 27. Issurin VB, Kaufman LE, Tenenbaum G (2001) Modeling of velocity regimens for anaerobic and aerobic power exercises in high-performance swimmers. J Sports Med Phys Fitness 41(4): 433–440. 28. Barbosa TM, Bragada JA, Reis VM, Marinho DA, Carvalho C, et al. (2010) Energetics and biomechanics as determining factors of swimming performance: updating the state of the art. J Sci Med Sport 13(2): 262–269. Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 10 October 2013 | Volume 8 | Issue 10 | e77185 29. Seifert L, Leblanc H, Chollet D, Sanders RH, Persyn U (2011) In: Seifert L, Chollet D, Mujika I, editors. World Book of Swimming. From Science to Performance. New York: Nova: pp. 135–152. 30. Sanders RH (2011) Rhythms in Butterfly Swimming. In: Seifert L, Chollet D, Mujika I, editors. World Book of Swimming. From Science to Performance. New York: Nova: pp. 191–202. 31. Rodriguez FA, Mader A (2011) Energy Systems in Swimming. In: Seifert L, Chollet D, Mujika I, editors. World Book of Swimming. From Science to Performance. New York: Nova: pp. 225–240. 32. Chatard J-C, Stewart AM (2011) Training Load and Performance in Swimming. In: Seifert L, Chollet D, Mujika I, editors. World Book of Swimming. From Science to Performance. New York: Nova: pp. 359–374. 33. Schiaffino S (2010) Fibre types in skeletal muscle: a personal account. Acta physiol (oxf) 2010: 199(4): 451–463. 34. Koulmann N, Bigard A-X (2006) Interaction between signalling pathways involved in skeletal muscle responses to endurance exercise. Pflugers Arch - Eur J Appl Physiol 452(2): 125–139. 35. Cormie P, McGuigan MR, Newton RU (2011a) Developing maximal neuromuscular power: Part 1–biological basis of maximal power production. Sports Med 41(1): 17–38. 36. Baar K (2010) Epigenetic control of skeletal muscle fibre type. Acta physiol 199(4): 477–487. 37. Guzun R, Saks V (2010) Application of the principles of systems biology and Wiener’s cybernetics for analysis of regulation of energy fluxes in muscle cells in vivo. Int J Mol Sci 11(3): 982–1019. 38. Van Wessel T, De Haan A, Van der Laarse WJ, Jaspers RT (2010) The muscle fiber type-fiber size paradox: hypertrophy or oxidative metabolism? Eur J Appl Physiol 110(4): 665–694. 39. Meckel Y, Bishop DJ, Rabinovich M, Kaufman L, Nemet D, et al. (2012) The relationship between short- and long-distance swimming performance and repeated sprint ability. J Strength Cond Res 26(12): 3426–3431. 40. Aagaard P, Andersen JL, Bennekou M, Larsson B, Olesen JL, et al. (2011) Effects of resistance training on endurance capacity and muscle fiber composition in young top-level cyclists. Scand J Med Sci Sports 21(6): e298–307. 41. Hoppeler H, Baum O, Lurman G, Mueller M (2011) Molecular Mechanisms of Muscle Plasticity with Exercise. Comprehensive Physiology. Wiley Online Library. 42. Deutscher Schwimmverband (2000) Komplexe Leistungsdiagnostik Testbes- chreibung. In: Freitag W, editor. Schwimmen Lernen und Optimieren. Band 17. p.168. 43. Rudolph K, Berbalk A (2000) Ausdauerdiagnostik im Rahmen der DSV-KLD von 1992–1997. In: Freitag W, editor. Schwimmen Lernen und Optimieren, Band 17. pp. 33–55. 44. Faude O, Kindermann W, Meyer T (2009) Lactate threshold concepts: how valid are they? Sports Med 39(6): 469–490. 45. West BT (2009) Analyzing longitudinal data with the linear mixed models procedure in SPSS. Eval Health Prof 32(3): 207–228. 46. Kwok O-M, Underhill AT, Berry JW, Luo W, Elliott TR, et al. (2008) Analyzing Longitudinal Data with Multilevel Models: An Example with Individuals Living with Lower Extremity Intra-articular Fractures. Rehabil Psychol 53(3): 370–386. 47. Singer J, Willett J (2003) Applied longitudinal data analysis: Modeling change and event occurrence. Oxford University Press. 48. Cohen J (1988) Statistical Power analysis for the behavioral sciences (2nd ed.). Mahwah, NJ: Erlbaum. 49. Skorski S, Faude O, Urhausen A, Kindermann W, Meyer T (2012) Intensity control in swim training by means of the individual anaerobic threshold. J Strength Cond Res 26(12): 3304–3311. 50. Dekerle J, Pelayo P (2011) Assessing Aerobic Endurance in Swimming In: Seifert L, Chollet D, Mujika I, editors. World Book of Swimming. From Science to Performance. New York: Nova: pp. 277–296. 51. Beneke R, Leitha¨user RM, Ochentel O (2011) Blood lactate diagnostics in exercise testing and training. Int J Sports Physiol Perform 6(1): 8–24. 52. Laursen PB (2010) Training for intense exercise performance: high-intensity or high-volume training? Scand J Med Sci Sports 20(2): 1–10. 53. Little JP, Safdar A, Bishop D, Tarnopolsky MA, Gibala MJ (2011) An acute bout of high-intensity interval training increases the nuclear abundance of PGC-1a and activates mitochondrial biogenesis in human skeletal muscle. Am J Physiol Regul Integr Comp Physiol 300(6): R1303–1310. 54. Seiler KS, Kjerland GØ (2006) Quantifying training intensity distribution in elite endurance athletes: is there evidence for an ‘‘optimal’’ distribution? Scand J Med Sci Sports 16(1): 49–56. 55. Zamparo P, Antonutto G, Capelli C, Francescato MP, Girardis M, et al. (1996) Effects of body size, body density, gender and growth on underwater torque. Scand J Med Sci Sports 6: 273–280. 56. Vilas-Boas JP, Fernandes RJ, Barbosa TM (2011) Intra-Cycle Velocity Variations, Swimming, Economy, Performance, and Training in Swimming. In: Seifert L, Chollet D, Mujika I, editors. World Book of Swimming. From Science to Performance. New York: Nova: pp. 119–134. 57. Cormie P, McGuigan MR, Newton RU (2011b) Developing maximal neuromuscular power: Part 2 - training considerations for improving maximal power production. Sports Med 41(2): 125–146. 58. Toigo M, Boutellier U (2006) New fundamental resistance exercise determinants of molecular and cellular muscle adaptations Eur J Appl Physiol 97(6): 643–63. 59. Rønnestad BR, Mujika I (2013) Optimizing strength training for running and cycling endurance performance: A review. Scand J Med Sci Sports [Epub ahead of print]. 60. Kjendlie PL, Stallman R (2011) Morphology and swimming performance. In: Seifert L, Chollet D, Mujika I, editors. World Book of Swimming. From Science to Performance. New York: Nova: pp. 203–222. 61. Harridge SDR (2007) Plasticity of human skeletal muscle: gene expression to in vivo function. Exp Physiol 92(5): 783–797. Lactate Characteristics in Swimming PLOS ONE | www.plosone.org 11 October 2013 | Volume 8 | Issue 10 | e77185

The influence of sex, stroke and distance on the lactate characteristics in high performance swimming.

10-22-2013

Holfelder, Benjamin,Brown, Niklas,Bubeck, Dieter

eng

PMC7961446

sensors Article Validity and Reliability of an Instrumented Treadmill with an Accelerometry System for Assessment of Spatio-Temporal Parameters and Impact Transmission Alberto Encarnación-Martínez 1,* , Pedro Pérez-Soriano 1 , Roberto Sanchis-Sanchis 1,2 , Antonio García-Gallart 3 and Rafael Berenguer-Vidal 4



 Citation: Encarnación-Martínez, A.; Pérez-Soriano, P.; Sanchis-Sanchis, R.; García-Gallart, A.; Berenguer-Vidal, R. Validity and Reliability of an Instrumented Treadmill with an Accelerometry System for Assessment of Spatio-Temporal Parameters and Impact Transmission. Sensors 2021, 21, 1758. https:// doi.org/10.3390/s21051758 Academic Editor: Ernesto De La Cruz-Sánchez Received: 6 February 2021 Accepted: 27 February 2021 Published: 4 March 2021 Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affil- iations. Copyright: © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/). 1 Research Group in Sports Biomechanics (GIBD), Department of Physical Education and Sports, University of Valencia, 46010 Valencia, Spain; pedro.perez-soriano@uv.es (P.P.-S.); roberto.sanchis@uv.es (R.S.-S.) 2 Physical Education and Sport, University of Alicante, 03690 San Vicente del Raspeig, Spain 3 The Civil Guard, Secretary of State for Security, Ministry of the Interior, 28010 Madrid, Spain; garciagallart@gmail.com 4 Grupo de Investigación en Telecomunicaciones Avanzadas (GRITA), Catholic University of Murcia, 30107 Guadalupe, Spain; rberenguer@ucam.edu * Correspondence: alberto.encarnacion@uv.es Abstract: Running retraining programs focused on concurrent feedback of acceleration impacts have been demonstrated to be a good strategy to reduce running-related injuries (RRI), as well as to improve running economy and reduce acceleration impacts and injury running incidence. Traditionally, impacts have been registered by mean of accelerometers attached directly to the athletes, which is inaccessible to the entire population, because it requires laboratory conditions. This study investigated the validity and reliability of a new device integrated directly into the treadmill, compared to a traditional acceleration impact system. Thirty healthy athletes with no history of RRI were tested on two separate days over the instrumented treadmill (AccTrea) and simultaneously with an acceleration impact system attached to the participant (AccAthl). AccTrea was demonstrated to be a valid and reliable tool for measuring spatio-temporal parameters like step length (validity intraclass correlation coefficient (ICC) = 0.94; reliability ICC = 0.92), step time (validity ICC = 0.95; reliability ICC = 0.96), and step frequency (validity ICC = 0.95; reliability ICC = 0.96) during running. Peak acceleration impact variables showed a high reliability for the left (reliability ICC = 0.88) and right leg (reliability ICC = 0.85), and peak impact asymmetry showed a modest validity (ICC = 0.55). These results indicated that the AccTrea system is a valid and reliable way to assess spatio-temporal variables, and a reliable tool for measuring acceleration impacts during running. Keywords: impact acceleration; spatio-temporal; instrumented treadmill; running; retraining 1. Introduction Running is one of the most popular recreational activities [1–3]. Its success may be because it is an aerobic activity that improves health and longevity, prevents diseases, and is very effective for getting fit [1–3]. Against the numerous benefits of running, injuries in this activity have a high incidence as almost half of runners are injured every year [1]. The annual incidence of lower-limbs injuries ranges from 19.4% to 79.3% [1], or even 92.4% [2] in long-distance runners. Most injuries are caused by the overuse of certain structures, [1,2,4] and the knee is the most common place of injury [2,4], ranging from 7.2% to 50% [2]. Thus, injuries can lead to a temporary or permanent interruption of exercise and even inability to work, leading to the need for medical treatment, where direct costs may exceed 1300 € [5]. Scientifically related to running injuries [6–8], an impact is generated with each foot contact with the floor that produces stress up to 1.5 to 2.5 times the body weight [8], and it is transmitted and absorbed by the whole body [9–11]. These impacts are attenuated Sensors 2021, 21, 1758. https://doi.org/10.3390/s21051758 https://www.mdpi.com/journal/sensors Sensors 2021, 21, 1758 2 of 16 internally by passive structures such as bones, cartilage, and ligaments, and by active movements such as joint angular displacements and eccentric muscle actions, in addition to external components such as footwear or surfaces [11]. The impacts during running have been broadly studied, and accelerometry is the technique most used to register this mechan- ical stress in sports activities [6–8,11]. This technique is based on the placement of low-mass accelerometers (uniaxial or triaxial), mostly on the tibia and front of the head to register in “g” or gravities (1 g = 9.8 m/s2) the acceleration/deceleration of body segments to calculate the magnitude and attenuation of impact [8–11]. It has been found that after a prolonged running fatigue protocol, while tibial accelerations increase [6–8], head accelerations re- main stable [7,8,11], which means that impact absorption also increases [6,8,12]. Therefore, it is necessary to adopt measures to reduce these stress levels on the musculoskeletal system during running and their negative effects [4]. The running surface, as an external component, plays a major role in impact atten- uation [11,13]. Although overground running is the surface preferred for recreational runners [14], running on a treadmill is a very popular activity in gyms, in therapeutic activities, rehabilitation, training, or athletic performance testing [13,15]. It has been shown that running on a treadmill can modify running biomechanics compared with overground running [16]. These kinematic modifications during treadmill running favors a running technique characterized by a higher level of security [17] as the magnitude of the impact is lower [13,17] and the risk of stress injuries is lower on treadmills in comparison with overground running [18]. Around 5000 impacts can occur during a typical 30 min running practice [8]. Thus, an excessively high impact level, due to a poor running technique or a reduction in attenuation ability as the fatigue progresses, has been related to an increased risk of injury [6–8]. Despite the potential benefits associated with running on a treadmill, the spatial and sensory constraints imposed by treadmills alter temporal and neuromuscular control in comparison with the overground condition [19]. Nevertheless, some research focused on analyzing the effects of biofeedback or auditory or visual information on some modifiable factors, such as running technique, that could reduce the severity of impacts [20–23]. These authors showed that by providing visual [21] or auditory [22] information through a screen about the impact levels received during treadmill running, athletes were able to make small modifications in their running technique autonomously to lower the impact peak [21,22], and their running technique became more efficient or economical [20]. Therefore, the implementation of biofeedback is an effective measure to reduce impacts and improve running economy [20–22]. It is important to highlight that all the studies that analyze impacts during running using accelerometry place the sensors directly on the athletes’ body, and the biofeedback system is used as an external element to the instruments used to carry out the activity. Similarly, to analyze spatio-temporal variables during running, other systems based on contact platforms [24] or optoelectronic technology [25] have been previously used. How- ever, these systems allow just a limited number of strides or require expensive technology, making them inaccessible to the general population. Some research works have used instrumented treadmills with force-plates or pressure sensors that allow measurement of the pressure produced by the runner on the lower board of the treadmill [26,27]. Force-plates present interesting advantages in motion and gait analysis [28], although the substantial cost of this instrumentation reduces the possibility of its use outside the laboratory on a large scale. On the other hand, despite the numerous advantages of using accelerometers for impact analysis described above, as far as we know to date [29–31], there are no treadmills that integrate acceleration sensors into their own system. Accelerometers are today a proven and low-cost technology used for displacement estimation [32] in a wide range of applications, such as electrohydraulic systems [33], architecture, civil engineering [34,35], seismology [36], or even astronomy [37]. In all these applications, the accelerometers are rigidly attached to the element whose displacement is Sensors 2021, 21, 1758 3 of 16 to be monitored, and with an analysis of the accelerometry signals, the motion and other parameters of interest can be estimated. For this reason, we proposed the placement of accelerometry sensors directly on a treadmill, which will allow us to similarly estimate the movement on the treadmill and thus analyze the movement of the runner when using the treadmill. Thus, our aims were: (a) To implement and validate an accelerometry system, placed directly in the treadmill and integrated into the software; (b) to compare the impact and space–time data during running obtained from the accelerometry system integrated in the treadmill with the data extracted from the accelerometry system placed directly on the athlete’s body. We hypothesize that: (a) The accelerometry system integrated in the treadmill is a valid and reliable tool for measuring impacts and space–time parameters during running; (b) the accelerometry system integrated in the treadmill offers similar data to those provided by an accelerometry system placed directly on the athlete’s body. 2. Materials and Methods 2.1. Participants This study was approved by the institution’s Human Research Ethics Committee (registry number: 6775). Thirty recreational athletes, ten women and twenty men, were recruited from local Athletics recreational teams, from March to April 2019, and were tested twice. Both tests were completed within 2 weeks and at least 24 h apart. Inclusion criteria were: To be physically active (to run a minimum of twice a week in the last year, do 2 h and 30 min a week of moderate-intensity, or 1 h and 15 min a week of vigorous-intensity aerobic physical activity), to have no history of lower body injuries within the last six months, to not be taking medication that hinders stability during the running, and to not suffer musculoskeletal disorders, heart failure, or neurological disorders that could affect normal locomotion. Athletes were excluded if they have had significant illness, injury, or surgery within the previous six months, and if they were overweight or obese (BMI < 24.9 kg/m2). All participants provided informed consent before their inclusion in the study. The baseline characteristics are shown in Table 1. Table 1. Baseline characteristics of the thirty participants, values are means ± SD. Characteristics (M ± SD) Female (n = 10) Male (n = 20) Age, y 24.4 ± 6.1 27.2 ± 7.5 Weight, kg 55.8 ± 4.0 73.3 ± 8.0 Height, cm 161.3 ± 4.3 175.6 ± 5.1 BMI, kg/m2 21.4 ± 1.3 23.7 ± 2.3 M = mean, SD = standard deviation, BMI: Body mass index. 2.2. Experimental Setups Acceleration impact data during running were recorded using a wireless triaxial accelerometry system (AcelSystem, Blautic, Spain; dimensions: 40 mm × 22 mm × 12 mm) adjusted to the athletes (AccAthl), at a sampling ratio of 415 Hz, a measuring range of up to ±16 g, and a total mass of 2.5 g. Simultaneously, a system consisting of a set of four triaxial MPU-9250 accelerometry sensors (TDK InvenSense, San José, CA, USA) embedded in the treadmill (AccTrea) was used. These four accelerometers were set at a sampling frequency of 250 Hz and with a range up to ±8 g, appropriate for the expected measurement values [29–31]. For every participant, a lightweight triaxial accelerometer was placed on the distal and anteromedial portion of each tibia with the vertical axis of each accelerometer aligned to be parallel to the long axis of the shank [38], as the location of the tibial accelerometer does influence the acceleration signal [38]. The skin was previously prepared and the accelerometers were adjusted with elastic belts as recommended by Encarnación-Martínez, García-Gallart, Gallardo, Sánchez-Sáez, and Sánchez-Sánchez [9] (Figure 1). The treadmill Sensors 2021, 21, 1758 4 of 16 accelerometry system was encased inside the treadmill (EVOT1, Bodytone International Sports, Murcia, Spain), and comprised three parts: A group of triple-axis Micro Electro- Mechanical System (MEMS) accelerometers, a data acquisition unit, and a processing unit. Appendix A details the operation and connection between these constituent elements of AccTrea. Both AccAthl and AccTrea systems were triggered simultaneously to collect the impact acceleration data. Sensors 2021, 21, x FOR PEER REVIEW 4 of 17 to be parallel to the long axis of the shank [38], as the location of the tibial accelerometer does influence the acceleration signal [38]. The skin was previously prepared and the ac- celerometers were adjusted with elastic belts as recommended by Encarnación-Martínez, García-Gallart, Gallardo, Sánchez-Sáez, and Sánchez-Sánchez [9] (Figure 1). The treadmill accelerometry system was encased inside the treadmill (EVOT1, Bodytone International Sports, Murcia, Spain), and comprised three parts: A group of triple-axis Micro Electro- Mechanical System (MEMS) accelerometers, a data acquisition unit, and a processing unit. Appendix A details the operation and connection between these constituent elements of AccTrea. Both AccAthl and AccTrea systems were triggered simultaneously to collect the impact acceleration data. Figure 1. Graphical schematic of the body-worn accelerometer fixed to athletes: (A) Accelerometry system (AcelSystem, Blautic); (B) accelerometer fixation on distal and anteromedial portion of each tibia; (C) final experimental setup; and (D) graphical representation of the accelerometer signal and the experimental setup. Participants performed two running tests on different days. The first session in- tended to assess the validity of the accelerometry system implemented in the treadmill (AccTrea) versus an accelerometry system adjusted to the athletes (AccAthl), and the sec- ond session intended to test the reliability. Both running accelerations’ measurement ses- sions were undertaken in the biomechanics lab at the same environmental conditions and at similar times of the day. All participants used the heel–toe running style and wore their own running shoes (the same for all two tests). After the informed consent, participants performed a free 5 min warm-up until they were familiar with the testing treadmill con- dition [39]. Next, the participants were instrumented with the accelerometers and the run- ning tests were performed. They ran for 5 min at 10 km/h and 0% slope in order not to affect the parameters evaluated [40], and acceleration impacts and spatio-temporal pa- rameters were collected by the AccTrea and the AccAthl systems in two sets of 10 s during the last minute taken in each measurement session. Rate of Perceived Exertion (RPE) [41] was also registered after the warm-up and after each of the running test. The vertical component (z-coordinate) of the accelerometry signals has been proven to be most important for the assessment of acceleration impacts and injury stroke inci- dence [42]. Therefore, in both AccAthl and AccTrea, the vertical component of all accel- erometers was gathered for analysis. Data from the AccAthl system were analyzed using the Matlab program (Math- Works, MA, USA), custom-made. The accelerometers were previously calibrated by the manufacturer. The acceleration signal from each of the sensors was first filtered (Butter- worth, second-order, low-pass, cut-off frequency = 50 Hz) [43]. The signal was then seg- mented by calculating the signal period (using the autocorrelation) and locating the points of interest (maximum, minimum, etc.) for each step. The positive peak tibial acceleration was measured for each leg in g (1 g = 9.82 m/s2), as well as the asymmetry between the Figure 1. Graphical schematic of the body-worn accelerometer fixed to athletes: (A) Accelerometry system (AcelSystem, Blautic); (B) accelerometer fixation on distal and anteromedial portion of each tibia; (C) final experimental setup; and (D) graphical representation of the accelerometer signal and the experimental setup. Participants performed two running tests on different days. The first session intended to assess the validity of the accelerometry system implemented in the treadmill (AccTrea) versus an accelerometry system adjusted to the athletes (AccAthl), and the second session intended to test the reliability. Both running accelerations’ measurement sessions were un- dertaken in the biomechanics lab at the same environmental conditions and at similar times of the day. All participants used the heel–toe running style and wore their own running shoes (the same for all two tests). After the informed consent, participants performed a free 5 min warm-up until they were familiar with the testing treadmill condition [39]. Next, the participants were instrumented with the accelerometers and the running tests were per- formed. They ran for 5 min at 10 km/h and 0% slope in order not to affect the parameters evaluated [40], and acceleration impacts and spatio-temporal parameters were collected by the AccTrea and the AccAthl systems in two sets of 10 s during the last minute taken in each measurement session. Rate of Perceived Exertion (RPE) [41] was also registered after the warm-up and after each of the running test. The vertical component (z-coordinate) of the accelerometry signals has been proven to be most important for the assessment of acceleration impacts and injury stroke inci- dence [42]. Therefore, in both AccAthl and AccTrea, the vertical component of all ac- celerometers was gathered for analysis. Data from the AccAthl system were analyzed using the Matlab program (MathWorks, MA, USA), custom-made. The accelerometers were previously calibrated by the manu- facturer. The acceleration signal from each of the sensors was first filtered (Butterworth, second-order, low-pass, cut-off frequency = 50 Hz) [43]. The signal was then segmented by calculating the signal period (using the autocorrelation) and locating the points of interest (maximum, minimum, etc.) for each step. The positive peak tibial acceleration was measured for each leg in g (1 g = 9.82 m/s2), as well as the asymmetry between the legs, calculated as the relative difference between both peaks (right leg impact minus left leg impact) expressed as a percentage (%). On the other hand, as detailed in Appendix A, AccTrea incorporated four MPU-9250 sensors. According to the manufacturer’s specifications, these devices included a motion processing unit with low-pass filters and an EEPROM for on-chip factory calibration of Sensors 2021, 21, 1758 5 of 16 the sensor. Thus, factory-trimmed scale factors eliminated the need for external active components and end-user calibration. Nevertheless, a calibration routine was performed at sensor initialization on the data acquisition unit to offset the bias of gravity [44]. The AccTrea system allowed us to measure the acceleration of the table of the treadmill due to the runner impacts at each sensor position (see Figure A1). The difference in amplitude and phase between the signals from the different sensors made it possible to automatically detect the accelerations produced by each leg. Then, like AccAhtl, the asymmetry between the legs was also calculated from the signals of these sensors. Finally, the accelerometry data from both AccAhtl and AccTrea approaches were analyzed using Matlab (R2015a with Signal Processing Toolbox, MathWorks Inc., Natick, MA, USA), providing a set of spatio-temporal parameters such as step time (ms), step length (m), and step frequency (spm), that allowed a comparison of the two approaches. Appendix B details the algorithms used for calculating these parameters. 2.3. Statistics Prior to the validity and reliability tests, a chi-square test was performed to determine whether there were differences between males and females. The agreement between the two systems was reviewed by a Bland–Altman plot for each of the variables analyzed. The differences between the two systems (AccTrea–AccAthl) in each variable were plotted against the mean results [45]. Reliability was contrasted by means of a two-way, random- effects, single-measure (median of the two trials) intraclass correlation coefficients (ICC(2,1)) model. In conjunction with the ICC values, standard error of measurement (SEM) and minimum detectable change (MDC) values were calculated to assess the concurrent validity between the AccTrea and the AccAthl, as well as the within-device test–retest reliability and measurement error over the two testing sessions for all outcome measures [46]. Point estimates of the ICCs were interpreted as follows: Excellent (0.75–1), modest (0.4–0.74), or poor (0–0.39) [47]. All statistical analyses were conducted using the Statistical Package for the Social Sciences (SPSS Inc. Version 26.0, Chicago, IL, U.S.A.). The MDC, which is otherwise known as the reliable change index score, was calculated using the equations reported previously by Jacobson and Truax [48]. It is expressed as the percentage test–retest change in impact acceleration or spatio-temporal parameter required to find a significant difference at an alpha level of 0.05 based on the Day 1 mean value. 3. Results 3.1. Gender Differences The results of the chi-square test showed no statistically significant differences (mean bilateral asymptotic significance 0.411) regarding gender for any of the variables analyzed. Therefore, during this study, all subsequent statistical analyses were conducted jointly, including men and women, as a single sample for each of the groups. 3.2. Perceived Exertion Regarding the perceived exertion, no differences were found between sessions for any of the study groups (Table 2). Table 2. Rate of perceived exertion (RPE) differences between sessions at warm-up and the run- ning test. Day 1 Day 2 p Value 1 Warm-up (M ± SD) 9.0 ± 1.9 8.8 ± 2.0 0.68 Running test (M ± SD) 9.8 ± 1.6 9.5 ± 1.8 0.46 1 RPE differences between days (t-test). Sensors 2021, 21, 1758 6 of 16 3.3. Bland–Altman Plots All participants successfully completed the two days’ sessions. The Bland–Altman plots for the step length, step time, step frequency, and peak acceleration impact asymmetry are provided in Figure 2. There was a small relationship between the difference and the mean for all the spatio-temporal variables. Specifically, step length and time were slightly lower in the AccAthl system compared to AccTrea, and as a result, the step frequency variable was higher in the AccAthl system. The acceleration impact asymmetry did not show any obvious relationship between systems. Table 2. Rate of perceived exertion (RPE) differences between sessions at warm-up and the run- ning test. Day 1 Day 2 p Value 1 Warm-up (M ± SD) 9.0 ± 1.9 8.8 ± 2.0 0.68 Running test (M ± SD) 9.8 ± 1.6 9.5 ± 1.8 0.46 1 RPE differences between days (t-test). 3.3. Bland–Altman Plots All participants successfully completed the two days’ sessions. The Bland–Altman plots for the step length, step time, step frequency, and peak acceleration impact asym- metry are provided in Figure 2. There was a small relationship between the difference and the mean for all the spatio-temporal variables. Specifically, step length and time were slightly lower in the AccAthl system compared to AccTrea, and as a result, the step fre- quency variable was higher in the AccAthl system. The acceleration impact asymmetry did not show any obvious relationship between systems. Figure 2. Bland–Altman plots representing comparisons between the AccTrea system and the AccAthl system for four of the variables analyzed: (A) Step length; (B) step time (duration); (C) step frequency; and (D) peak acceleration impact asymmetry. The mean line represents the mean difference between the devices, with the upper and lower dashed lines representing the 95% limits of agreement (LOAs). 3.4. Validity and Reliability The results for the step length, step time, step frequency, left leg peak acceleration impact, right leg peak acceleration impact, and peak acceleration impact asymmetry var- iables are provided in Table 3. The step length and step time were lower in the AccAthl system compared with AccTrea. Step frequency, left leg peak acceleration impact, and right leg peak acceleration impact variables showed a bias toward higher values in the tests performed on the AccAthl. Inconsistent results were found for peak acceleration im- pact asymmetry variables. In general, both systems showed excellent test–retest reliability (Table 3), with only the peak acceleration impact asymmetry values’ performance on the AccTrea (ICC = 0.36) failing to reach an ICC value of 0.75, considered as an excellent value. Concurrent validity Figure 2. Bland–Altman plots representing comparisons between the AccTrea system and the AccAthl system for four of the variables analyzed: (A) Step length; (B) step time (duration); (C) step frequency; and (D) peak acceleration impact asymmetry. The mean line represents the mean difference between the devices, with the upper and lower dashed lines representing the 95% limits of agreement (LOAs). 3.4. Validity and Reliability The results for the step length, step time, step frequency, left leg peak acceleration impact, right leg peak acceleration impact, and peak acceleration impact asymmetry variables are provided in Table 3. The step length and step time were lower in the AccAthl system compared with AccTrea. Step frequency, left leg peak acceleration impact, and right leg peak acceleration impact variables showed a bias toward higher values in the tests performed on the AccAthl. Inconsistent results were found for peak acceleration impact asymmetry variables. In general, both systems showed excellent test–retest reliability (Table 3), with only the peak acceleration impact asymmetry values’ performance on the AccTrea (ICC = 0.36) failing to reach an ICC value of 0.75, considered as an excellent value. Concurrent validity was shown to be consistently excellent across spaciotemporal variables and testing sessions (ICC = 0.94–0.98), but not in acceleration impact variables for every testing session (ICC = −0.01–0.55). The SEM for the spaciotemporal variables ranged from 0.92 to 1.31% in the AccTrea system, and from 1.19 to 1.29% in the AccAthl system. For impact variables, the SEM ranged from 10.1 to 358% in the AccTrea system and from 12.25 to 297% in the AccAthl system. Sensors 2021, 21, 1758 7 of 16 Table 3. Validity and reliability of an instrumented treadmill with an accelerometry system for assessment of spatio-temporal parameters and impact transmission. AccTrea AccAthl Mean Diff (95%CI) ICC (95%CI) Step Length (m) Day 1 (M ± SD) 1.04 ± 0.05 1.01 ± 0.04 0.04 (0.03/0.05) 0.94 (0.87/0.97) Day 2 (M ± SD) 1.03 ± 0.04 1.00 ± 0.05 0.04 (0.03/0.05) 0.95 (0.89/0.98) Mean Diff (95%CI) 0.002 (−0.005/0.010) 0.011 (−0.002/0.023) ICC (95%CI) 0.92 (0.82/0.96) 0.88 (0.73/0.95) SEM (% SEM) 0.01 (1.31) 0.01 (1.29) MDC (%) 0.04 0.04 Step Time (ms) Day 1 (M ± SD) 374.8 ± 16.8 363.1 ± 13.9 12.5 (9.7/15.3) 0.94 (0.87/0.97) Day 2 (M ± SD) 371.8 ± 15.2 359.7 ± 17.1 13.0 (10.1/15.9) 0.95 (0.89/0.98) Mean Diff (95%CI) 0.88 (−1.75/3.51) 3.85 (−0.60/8.29) ICC (95%CI) 0.96 (0.90/0.98) 0.89 (0.73/0.95) SEM (% SEM) 3.55 (0.95) 4.7 (1.29) MDC (%) 9.85 13.03 Step Frequency (spm) Day 1 (M ± SD) 160.5 ± 7.2 166.1 ± 6.6 −5.59 (−6.7/−4.5) 0.95 (0.90/0.97) Day 2 (M ± SD) 161.3 ± 6.7 167.1 ± 8.0 −5.94 (−7.2/−4.7) 0.95 (0.89/0.98) Mean Diff (95%CI) −0.17 (−1.24/0.89) −0.80 (−2.9/1.3) ICC (95%CI) 0.96 (0.91/0.98) 0.91 (0.82/0.93) SEM (% SEM) 1.48 (0.92) 1.97 (1.19) MDC (%) 4.12 5.47 Left Leg Peak Impact (g) Day 1 (M ± SD) 0.72 ± 0.21 3.76 ± 1.37 −3.04 (−3.53/−2.56) 0.09 (−0.86/0.56) Day 2 (M ± SD) 0.72 ± 0.22 3.93 ± 1.30 −3.21 (−3.69/−2.73) 0.08 (−0.93/0.56) Mean Diff (95%CI) 0.001 (−0.052/0.053) −0.181 (−0.514/0.152) ICC (95%CI) 0.88 (0.75/0.94) 0.88 (0.74/0.94) SEM (% SEM) 0.07 (10.05) 0.48 (12.25) MDC (%) 0.20 1.34 Right Leg Peak Impact (g) Day 1 (M ± SD) 0.73 ± 0.20 3.91 ± 1.62 −3.18 (−3.76/−2.59) 0.01 (−1.04/0.52) Day 2 (M ± SD) 0.76 ± 0.18 3.97 ± 1.71 −3.21 (−3.85/−2.56) −0.01 (−1.13/0.52) Mean Diff (95%CI) −0.03 (−0.08/0.02) −0.05 (−0.42/0.32) ICC (95%CI) 0.85 (0.69/0.93) 0.90 (0.80/0.95) SEM (% SEM) 0.08 (10.20) 0.50 (12.64) MDC (%) 0.21 1.39 Peak Impact Asymmetry (%) Day 1 (M ± SD) −2.80 ± 12.53 −1.29 ± 14.49 −1.51(15.11/2.67) 0.55 (0.07/0.78) Day 2 (M ± SD) −2.75 ± 9.79 2.44 ± 17.94 −6.16 (18.34/3.47) 0.28 (−0.55/0.67) Mean Diff (95%CI) 0.75 (14.16/2.63) −3.82 (14.94/2.77) ICC (95%CI) 0.36 (−0.37/0.70) 0.75 (0.46/0.88) SEM (% SEM) 10.04 (−358.66) 7.28 (297.729) MDC (%) 27.82 20.17 AccTrea: Treadmill system; AccAthl: Athlete system; M: Mean; SD: Standard deviation; CI: Confidence interval; ICC: Intraclass correlation coefficient; Diff: Difference; SEM: Standard error of the measurement; MDC: Minimum detectable change, expressed as a percentage of the Day 1 mean value. The MDC in all variables ranged from 0.04 to 27.8% for the AccTrea system and from 0.04 to 20.2% for the AccAthl system. The MDCs were reasonably high for both devices only in the peak acceleration impact asymmetry variable (27.8% at AccTrea and 20.2% at AccAthl). With respect to the other variables (spaciotemporal and impacts), the MDCs were lower for both systems, with the AccAthl MDC values higher than the AccTrea values in all values. Sensors 2021, 21, 1758 8 of 16 4. Discussion Validity and reliability of spatio-temporal and impact transmission variables during running are important in biomechanical analysis under laboratory conditions. Treadmills are becoming popular between recreational runners [15]. Oxygen uptake, heartrate, and perceived effort are similar between submaximal treadmill and overground running [15]. However, running on a treadmill provides greater control over environmental variables such as temperature, wind speed, or relative humidity [15]. Treadmills also offer control over running velocity and surface gradient [15], and generate changes in biomechanics parameters like step length, contact time, and stride frequency compared with overground running [17]. These kinematics modifications favor the reduction in impact acceleration magni- tude [13,17], axial compression strains in tibia [18], and plantar load [13,49] in comparison with overground running. It causes runners to adopt a safer running style [17]. In addition, running retraining programs, focused on reducing the severity of impacts that are related to running injuries, have demonstrated good results by means of intro- ducing biofeedback systems (auditory or visual information) during training sessions on the treadmill. Previous studies have shown that runners were able to reduce impacts and improve running economy thanks to the concurrent information about the severity of the impacts received from accelerometers placed directly on their body [20–22]. The control of environmental and performance factors, along with the kinematic modifications offered by the treadmills, can make it a safer activity than overground running. Introducing auditory or visual biofeedback information from the treadmill could allow the control of impact acceleration and make that system accessible for all types of runners, both professional and recreational. Our results partially confirmed the hypothesis raised in the study, that the Acc- Trea system integrated in the treadmill is a valid and reliable tool for measuring spatio- temporal parameters like step length (validity ICC 95%CI = 0.87/0.97; reliability ICC 95%CI = 0.82/0.96), step time (validity ICC 95%CI = 0.87/0.97; reliability ICC 95%CI = 0.90/0.98), and step frequency (validity ICC 95%CI = 0.90/0.97; reliability ICC 95%CI = 0.91/0.98) during running on a treadmill compared to the AccAthl system under the same speed condition. Nevertheless, peak acceleration impact variables measured during running showed a high reliability for the left leg (reliability ICC 95%CI = 0.75/0.94) and right leg (reliability ICC95%CI = 0.69/0.93), but not a high validity (Table 2). On the other hand, peak acceleration impact asymmetry showed a modest validity (ICC = 0.55) but a poor reliability (Table 2). Prior to our study, other systems that measure spatio-temporal variables during run- ning have been validated. These systems were initially based on contact platforms [24], but they allowed the analysis of just a limited number of strides, in addition to the possibility of altering the running gait. Other systems based on optoelectronics technology were also validated [25] to measure the spatio-temporal variables without altering the natural run- ning pattern [50], but the drawback of these systems was that they need to install different extremely sensitive instruments whose technology is relatively expensive compared to the technology of the AccTrea system, analyzed in the present study. The spatio-temporal variables analyzed in our study have shown intraclass correlation coefficients (ICC > 0.946) close to those obtained by Ogueta-Alday, Morante, Rodríguez- Marroyo, and García-López [50] when they validated a new method to measure contact time and flight time during treadmill running (SportJump System Pro, V2.0., León, Spain) (ICC > 0.993). It should be noted that in the present study, other variables have been analyzed than those evaluated in the SportJump System Pro, as the objective of the AccTrea system was to provide concurrent feedback to runners in order to modify step length, frequency, and time to improve their running economy. The excellent validity and reliability results for the spatio-temporal variables, together with the technology used, make the AccTrea system a low-cost and high-reliability system, nonexistent until now. Sensors 2021, 21, 1758 9 of 16 The results of peak acceleration impact asymmetries are considered modest (ICC = 0.55) for validity between systems and poor (ICC = 0.36) for the within-device test–retest reli- ability for the AccTrea system. Both acceleration impact peaks of the left leg (ICC = 0.88) and right leg (ICC = 0.85) obtained a high degree of reliability of the AccTrea system between days, which was not the case for validity between systems (AccTrea and AccAthl), considered as poor (ICC = 0.01). Symmetry/asymmetry in running is very difficult to keep within the same values between different sessions as there are many factors that affect running technique [51]. It is a personal technical factor subject to the variability in the dynamic complex systems, an aspect that makes the standardization of the results difficult [51]. The values obtained in the present study were relatively low (little asymmetry on impacts between legs), which could justify the poor reliability results of the system between days. The poor validity results between systems obtained in the acceleration impact peak variables could be related to the fact that the AccTrea system, compared to the AccAthl, presents elements that could favor the reduction or loss of acceleration and impact dis- sipation. These elements could be classified as elements typical of the runner, such as the cushioning of the shoes [52]; or elements of the system itself (AccTrea), such as the treadmill, the table, or the protection of the accelerometers, that avoid their displacement and make them register lower values [53]. The Bland–Altman plots demonstrated low mean differences and wide limits of agreement (LoAs) of 95%, except for the step frequency variable, with a mean difference between systems of ±5 ppm. These differences could be explained because step time is also slightly lower in the AccAthl system, possibly associated with acceleration losses of the system previously mentioned. Regarding the system, there are currently no studies with which the results obtained from the acceleration impact variables can be compared. There are also no systems on the market that can directly or indirectly measure the acceleration impact variables without instrumenting the athlete and with the technology used inserted directly into the treadmill. Previous studies that have analyzed the effect of immediate biofeedback, via audi- tory [22] or visual [21], during running have shown that the maximum impact peak was significantly reduced [21,22], improving running economy [20]. Recent studies have shown that the effects of an intervention, applying instant feedback, can last up to a year after the intervention, notably improving the reduction of impacts and reducing the percentage of injured athletes [54]. However, all these biofeedback systems used in previous studies have required ath- letes to be instrumented with expensive systems and under laboratory conditions, making the use of this type of system impractical on a recurring basis by the general population. The implementation of a biofeedback system, such as the one analyzed in this study, represents a step forward to make impact reductions and running economy improvements accessible to the entire population [23] thanks to its low cost and the unneeded instrumentation of athletes. The results of this study determined that the system has moderate validity for the scientific measurement of the acceleration impact variables (ICC = 0.55), but it can also be transferred to the sports world, being a valid approximation like the contributions that already exist in the market in the measurement of other variables. 5. Conclusions AccTrea is a reliable and valid tool for athletes to be informed, in a concurrent way, of their biomechanical responses in relation to spatio-temporal variables (step length, step time, and step frequency) during running on an instrumented treadmill. On the other hand, the limitations found in the placement of the accelerometers under the treadmill, which in turn, are great advantages of the system by not having to instrument the athletes, make AccTrea a reliable system measuring running impacts. While peak acceleration impact asymmetry variables presented a modest validity between systems. As a noninvasive Sensors 2021, 21, 1758 10 of 16 biofeedback system for running biomechanical response, AccTrea demonstrates potential as a commercial system of easy access to the general population, with high reliability in spatio-temporal variables and peak acceleration impacts. MEMS sensor technology coupled with a data acquisition unit and a processing unit connected with the treadmill can provide accurate and objective data to improve running mechanics or to allow personal trainers to select running exercises in order to change running mechanics. 6. Patents European patent application with reference EP3735900A1 and entitled “Treadmill for sport training” in May 2019. U.S. patent application with reference US20200353309A1 and entitled “Ergometric treadmill for sport training” in May 2019. Chinese patent application with reference CN111905333A and entitled “Force measur- ing running machine for sports training” in May 2019. Author Contributions: Conceptualization, A.E.-M., P.P.-S. and R.B.-V.; methodology, A.E.-M. and A.G.-G.; software, A.E.-M., R.B.-V., A.G.-G. and P.P.-S.; validation, A.E.-M., R.B.-V., A.G.-G., R.S.-S., and P.P.-S.; formal analysis, A.E.-M., A.G.-G., R.S.-S., and P.P.-S.; investigation, A.E.-M., R.B.-V., A.G.-G., R.S.-S., and P.P.-S.; resources, A.E.-M., R.B.-V., A.G.-G. and P.P.-S.; data curation, A.E.-M. and R.B.-V.; writing—original draft preparation, A.E.-M., R.B.-V., R.S.-S., and A.G.-G.; writing—review and editing, A.E.-M., R.B.-V., R.S.-S., and A.G.-G.; visualization, A.E.-M., R.B.-V., R.S.-S., and A.G.-G.; supervision, A.E.-M.; project administration, A.E.-M., R.B.-V., A.G.-G. and P.P.-S.; funding acquisition, A.E.-M. All authors have read and agreed to the published version of the manuscript. Funding: This research was funded by Bodytone International Sport, S.L., grant number CFE- BODYTONE-03-18. Institutional Review Board Statement: The study was conducted according to the guidelines of the Declaration of Helsinki and approved by the Institutional Review Board of the University of Valencia (protocol number: 6775, date 2018). Informed Consent Statement: Informed consent was obtained from all subjects involved in the study. Acknowledgments: Authors want to thank Inmaculada Aparicio Aparicio for her support in the study conceptualization and her help during data collection. Conflicts of Interest: The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results. Appendix A. Treadmill Accelerometry System (AccTrea) The proposed measurement system comprises three main parts: A set of triple-axis MEMS accelerometers (MEMS-As), a data acquisition unit (DAU), and a processing unit (PU). All components are encased inside the treadmill. Thus, the system does not require external instrumentation. The MEMS-As send the data set to the DAU, which performs the first stage of filtering and conditioning of the data. This pre-processed data set is then used by the PU to estimate the parameters under interest as detailed in Section 2. Two different MEMS-A settings can be used in the system. The first approach employs two accelerometers, which are located on the front of the running belt, near the landing zone of the runner. The second approach uses a four-accelerometers setting, where two more accelerometry sensors are placed at the back of the belt. In either approach, all accelerometers are firmly attached to the treadmill board by means of a specifically designed holder in order to maximize the capture of the vibrations produced by the runner. The V120 optical tracking system (Optitrack V120:Trio, NaturalPoint, Inc., Corvallis, OR, USA) has been used to determine the optimal position of the sensors to maximize the measurement of these oscillations. Figure A1 shows the placement of the sensors of the treadmill along with the coordinate system used in the device. Sensors 2021, 21, 1758 11 of 16 MPU-9250 accelerometry sensors (TDK InvenSense, San José, CA, USA) are used in both settings. These sensors provide digital-output triple-axis accelerometry data with a programmable full-scale range between ±2 g and ±16 g. This feature is particularly interesting as it allows both low-acceleration values, such as when walking on the treadmill, and large values, such as in fast running, to be measured accurately. MPU-9250 accelerometers use 16-bit analog-to-digital converters (ADCs), which pro- vide enough bit resolution for the subsequent parameter calculation. This allows us to set the scale range to ±8 g, providing a relatively small quantification error and, at the same time, a wide reading range to avoid data clipping. Note that this requirement is important because the acceleration values of the lateral (x-axis) and front-rear (y-axis) directions are much lower than the vertical one (z-axis). Although only the z-axis component is analyzed in this work, other accelerometry components may be used in future work. For this reason, the acquisition and storage of all accelerometry components are implemented in this approach. ors 2021, 21, x FOR PEER REVIEW 12 of 17 Figure A1. Location of accelerometers (MEMS-A), data acquisition unit (DAU), and processing unit (PU) on the treadmill (AccTrea): (a) Four-accelerometer approach; (b) two-accelerometer ap- proach. The MPU-9250 devices include user-programmable digital filters for noise reduction. Although a low-pass filter with a cut-off frequency of 50 Hz is commonly used in many similar applications [38], no in-board filtering has been set in the device for this approach. As shown in Appendix B, the critical information for the calculation of the parameters under analysis involves the time intervals between acceleration peaks. As any low-pass filter smooths these peaks [43], thereby reducing the accuracy of the parameter calcula- tion, the raw data from the sensors are used for further processing and analysis in this approach. Note, however, that prior to any measurement, a calibration is performed on the DAU to compensate for gravity bias [44]. For the connection between the MEMS-A and the DAU, the I2C bus is chosen [55]. This protocol allows the transmission of data of all sensors using a single bus, minimizing wiring and system complexity. Each sensor is set-up in a specific address, allowing all accelerometry signals to be transmitted using only two wires. The choice of sampling frequency of the analog-to-digital converter of the sensors is l l A l l bl h h h l l f h Figure A1. Location of accelerometers (MEMS-A), data acquisition unit (DAU), and processing unit (PU) on the treadmill (AccTrea): (a) Four-accelerometer approach; (b) two-accelerometer approach. The MPU-9250 devices include user-programmable digital filters for noise reduction. Although a low-pass filter with a cut-off frequency of 50 Hz is commonly used in many similar applications [38], no in-board filtering has been set in the device for this approach. As shown in Appendix B, the critical information for the calculation of the parameters under analysis involves the time intervals between acceleration peaks. As any low-pass filter smooths these peaks [43], thereby reducing the accuracy of the parameter calculation, the raw data from the sensors are used for further processing and analysis in this approach. Note, however, that prior to any measurement, a calibration is performed on the DAU to compensate for gravity bias [44]. For the connection between the MEMS-A and the DAU, the I2C bus is chosen [55]. This protocol allows the transmission of data of all sensors using a single bus, minimizing Sensors 2021, 21, 1758 12 of 16 wiring and system complexity. Each sensor is set-up in a specific address, allowing all accelerometry signals to be transmitted using only two wires. The choice of sampling frequency of the analog-to-digital converter of the sensors is also a critical issue. A large sampling rate enables a high accuracy in the calculation of the parameters. Nevertheless, it limits the use of multiple sensors with a single I2C bus due to its bandwidth restrictions. A sampling frequency of 250 Hz has been chosen as the compromise value, as it is large enough to obtain sufficient accuracy in the required parameters, while allowing real-time transmission of raw data from up to four sensors. The data are collected by the DAU, powered by a ATmega2560 microcontroller (Mi- crochip Technology Incorporated, Chandler, AZ, USA). This unit performs the following tasks: (1) Initialization and calibration of the sensors before each training; (2) time synchro- nization of the signals to be sent as a matrix to the subsequent PU; and (3) monitoring with automatic restart in case of reading or transmission failure. This data set is transmitted via a USB connection to the PU of the treadmill. This unit is responsible for calculating the parameters listed in Section 2, i.e., positive peak tibial acceleration, asymmetry, step time, step length, and step frequency. In addition, a frequency analysis is carried out, which will allow future work to carry out harmonic analysis, among others. Both Appendix B and patents [29–31] detail the process of calculating these parameters. Figure A2 depicts the connection diagram between the MEMS-A, DAU, and PU. Sensors 2021, 21, x FOR PEER REVIEW 13 of 17 tasks: (1) Initialization and calibration of the sensors before each training; (2) time syn- chronization of the signals to be sent as a matrix to the subsequent PU; and (3) monitoring with automatic restart in case of reading or transmission failure. This data set is transmitted via a USB connection to the PU of the treadmill. This unit is responsible for calculating the parameters listed in Section 2, i.e., positive peak tibial acceleration, asymmetry, step time, step length, and step frequency. In addition, a fre- quency analysis is carried out, which will allow future work to carry out harmonic analy- sis, among others. Both Appendix B and patents [29–31] detail the process of calculating these parameters. Figure A2 depicts the connection diagram between the MEMS-A, DAU, and PU. Figure A2. Block diagram of the AccTre system: MEMS-A connected to the DAU via an I2C bus; DAU linked to the PU via an USB connection; graphical user interface for biofeedback to the runner on the treadmill touch screen; transmission of processed data to a cloud server and access to the data set through a mobile device. Appendix B. Procedure for Calculating Spatio-Temporal Parameters The procedure for calculating these parameters is based on Pérez-Soriano and Encar- nación-Martínez [56], although it has been adapted for each of the approaches analyzed in this work. Note that in this study, two accelerometers have been used in both systems. In AccAthl, the sensors are placed on the distal and anteromedial portion of each tibia [38] while in AccTrea, they are located on the front of the running belt [38] (Figure A1). It is important to note that although both systems measure the acceleration produced by both legs, the way in which both sets of data are collected is completely different. Each AccAthl accelerometer is attached to one of the legs. Thus, the acceleration measured on each leg is clearly recorded on its corresponding accelerometer while the signal caused by the opposite leg is noticeably lower. By contrast, as the accelerometers included in Ac- cTrea are both attached to the rigid board of the treadmill, both accelerometers collect the vibration produced by both legs. Their signals vary only subtly in amplitude depending on the proximity of each sensor to the landing zone of each leg. Nevertheless, this slight difference is sufficient to determine the parameters of interest. Figure A3 illustrates the acceleration levels recorded by each system for the same measurement session. Figure A2. Block diagram of the AccTre system: MEMS-A connected to the DAU via an I2C bus; DAU linked to the PU via an USB connection; graphical user interface for biofeedback to the runner on the treadmill touch screen; transmission of processed data to a cloud server and access to the data set through a mobile device. Appendix B. Procedure for Calculating Spatio-Temporal Parameters The procedure for calculating these parameters is based on Pérez-Soriano and Encar- nación-Martínez [56], although it has been adapted for each of the approaches analyzed in this work. Note that in this study, two accelerometers have been used in both systems. In AccAthl, the sensors are placed on the distal and anteromedial portion of each tibia [38] while in AccTrea, they are located on the front of the running belt [38] (Figure A1). It is important to note that although both systems measure the acceleration produced by both legs, the way in which both sets of data are collected is completely different. Each AccAthl accelerometer is attached to one of the legs. Thus, the acceleration measured on each leg is clearly recorded on its corresponding accelerometer while the signal caused by the opposite leg is noticeably lower. By contrast, as the accelerometers included in AccTrea are both attached to the rigid board of the treadmill, both accelerometers collect the vibration produced by both legs. Their signals vary only subtly in amplitude depending Sensors 2021, 21, 1758 13 of 16 on the proximity of each sensor to the landing zone of each leg. Nevertheless, this slight difference is sufficient to determine the parameters of interest. Figure A3 illustrates the acceleration levels recorded by each system for the same measurement session. Sensors 2021, 21, x FOR PEER REVIEW 14 of 17 Figure A3. Comparison of the accelerometry signals from the system AccAthl (a) and the two-accelerometer approach AccTrea (b). The solid and dashed line display the accelerations corresponding to the left and right legs, respectively. The left- and right-pointing triangles mark the acceleration peaks of the left and right legs, respectively. The first step in the calculation of the parameters is to determine the accelerometry signal peaks for both legs in each approach, including their temporal location (s) and their peak amplitude value (g). To obtain the location ൛݈஺௧௛,௅ሾ݊௅ሿ, ݈஺௧௛,ோሾ݊ோሿൟ and amplitude ൛݌஺௧௛,௅ሾ݊௅ሿ, ݌஺௧௛,ோሾ݊ோሿൟ of the peaks in AccAthl, the maximum negative acceleration values are detected. The variables ݊௅ and ݊ோ denote the step number for the left and right leg, respectively. These values correspond to the time when the runner lands on each leg. To avoid false-negative detection, constraints on the minimum distance between peaks and on their prominence values are applied. From the data coming from AccTrea, all the peaks of both accelerometers are detected first. In contrast to the previous approach, the values of the maximum positive accelera- tion are considered here, as it has been empirically proven that they provide better results for the analysis due to the vibration of the table. Note that, as shown in Figure A3b, the contribution of both legs is recorded in both accelerometry signals. Therefore, the correspondence of the detected peaks to each leg must be estimated. This is done by averaging the values of the odd and even peaks of each signal and assigning the higher values to the closest sensor. Once this is accomplished, a procedure similar to AccAthl is applied to obtain the location ൛்݈௥௘௔,௅ሾ݊௅ሿ, ்݈௥௘௔,ோሾ݊ோሿൟ and amplitude ൛݌்௥௘௔,௅ሾ݊௅ሿ, ݌்௥௘௔,ோሾ݊ோሿൟ of the peaks as depicted in Figure A3b. From these vectors ൛݌஺௧௛,௅ሾ݊௅ሿ, ݌஺௧௛,ோሾ݊ோሿൟ and ൛݌்௥௘௔,௅ሾ݊௅ሿ, ݌்௥௘௔,ோሾ݊ோሿൟ , statistical values are calculated for the impacts of both legs, as well as for their asymmetry. Table 3 shows these results for both approaches. The step locations are used to compute the step times using backward differences, Figure A3. Comparison of the accelerometry signals from the system AccAthl (a) and the two-accelerometer approach AccTrea (b). The solid and dashed line display the accelerations corresponding to the left and right legs, respectively. The left- and right-pointing triangles mark the acceleration peaks of the left and right legs, respectively. The first step in the calculation of the parameters is to determine the accelerometry signal peaks for both legs in each approach, including their temporal location (s) and their peak amplitude value (g). To obtain the location  lAth,L[nL], lAth,R[nR] and amplitude  pAth,L[nL], pAth,R[nR] of the peaks in AccAthl, the maximum negative acceleration values are detected. The variables nL and nR denote the step number for the left and right leg, respectively. These values correspond to the time when the runner lands on each leg. To avoid false-negative detection, constraints on the minimum distance between peaks and on their prominence values are applied. From the data coming from AccTrea, all the peaks of both accelerometers are detected first. In contrast to the previous approach, the values of the maximum positive acceleration are considered here, as it has been empirically proven that they provide better results for the analysis due to the vibration of the table. Note that, as shown in Figure A3b, the contribution of both legs is recorded in both accelerometry signals. Therefore, the correspondence of the detected peaks to each leg must be estimated. This is done by averaging the values of the odd and even peaks of each signal and assigning the higher values to the closest sensor. Once this is accomplished, a procedure similar to AccAthl is applied to obtain the location {lTrea,L[nL], lTrea,R[nR]} and amplitude {pTrea,L[nL], pTrea,R[nR]} of the peaks as depicted in Figure A3b. From these vectors  pAth,L[nL], pAth,R[nR] and {pTrea,L[nL], pTrea,R[nR]}, statistical values are calculated for the impacts of both legs, as well as for their asymmetry. Table 3 shows these results for both approaches. Sensors 2021, 21, 1758 14 of 16 The step locations are used to compute the step times using backward differences, ∆tAthl,L[nL] = 103 fs,Athl (lAth,L[nL] − lAth,L[nL − 1]) ms, ∆tAth,R[nR] = 103 fs,Athl (lAth,R[nR] − lAth,R[nR − 1]) ms, ∆tTrea,L[nL] = 103 fs,Trea (lTrea,L[nL] − lTrea,L[nL − 1]) ms, ∆tTrea,R[nR] = 103 fs,Trea (lTrea,R[nR] − lTrea,R[nR − 1]) ms, (A1) where ∆t{,◦} h n{◦} i (ms) denotes the step time for approach {}; leg {◦} and step number, n{◦} and l{,◦}, respectively, are the peak locations (samples); fs,{} (samples per second) stands for the sampling frequency for each approach; and {nL, nR} represents the left and right step numbers, respectively. The step length is simply calculated by multiplying the step time by the linear speed of the belt, provided by the treadmill electronics, ∆l{,◦} h n{◦} i = v·10−3∆t{,◦} h n{◦} i m (A2) where ∆l{,◦} h n{◦} i (m) denote the step length for the step number n{◦}, ∆t{,◦} h n{◦} i are the step times calculated in Equation (A1), and v (m/s) represents the linear speed of the belt of the treadmill, which is identical in both approaches. To determine the step frequency, the number of steps detected by each sensor is counted and divided by the duration of the experiment, s f{,◦} = 60N{,◦}/T spm (A3) where s f{,◦} (spm) denotes the step frequency for approach {} and leg {◦}, and T is the duration of the experiment (T = 10 s). Once again, the statistical parameters of these variables are shown in Table 3. References 1. Fields, K.B.; Sykes, J.C.; Walker, K.M.; Jackson, J.C. Prevention of running injuries. Curr. Sports Med. Rep. 2010, 9, 176–182. [CrossRef] [PubMed] 2. Van Gent, R.N.; Siem, D.; van Middelkoop, M.; van Os, A.G.; Bierma-Zeinstra, S.M.; Koes, B.W. Incidence and determinants of lower extremity running injuries in long distance runners: A systematic review. Br. J. Sports Med. 2007, 41, 469–480. [CrossRef] [PubMed] 3. Cheung, R.T.; Wong, M.Y.; Ng, G.Y. Effects of motion control footwear on running: A systematic review. J. Sports Sci. 2011, 29, 1311–1319. [CrossRef] [PubMed] 4. Hreljac, A.; Ferber, R. A biomechanical perspective of predicting injury risk in running: Review article. Int. J. Sports Med. 2006, 7, 98–108. 5. Van der Worp, M.P.; ten Haaf, D.S.; van Cingel, R.; de Wijer, A.; Nijhuis-van der Sanden, M.W.; Staal, J.B. Injuries in runners; a systematic review on risk factors and sex differences. PLoS ONE 2015, 10, e0114937. [CrossRef] 6. Mizrahi, J.; Verbitsky, O.; Isakov, E.; Daily, D. Effect of fatigue on leg kinematics and impact acceleration in long distance running. Hum. Mov. Sci. 2000, 19, 139–151. [CrossRef] 7. Lucas-Cuevas, A.G.; Priego-Quesada, J.I.; Aparicio, I.; Giménez, J.V.; Llana-Belloch, S.; Pérez-Soriano, P. Effect of 3 Weeks Use of Compression Garments on Stride and Impact Shock during a Fatiguing Run. Int. J. Sports Med. 2015, 36, 826–831. [CrossRef] 8. Derrick, T.R.; Dereu, D.; McLean, S.P. Impacts and kinematic adjustments during an exhaustive run. Med. Sci. Sports Exerc. 2002, 34, 998–1002. [CrossRef] 9. Encarnación-Martínez, A.; García-Gallart, A.; Gallardo, A.M.; Sánchez-Sáez, J.A.; Sánchez-Sánchez, J. Effects of structural components of artificial turf on the transmission of impacts in football players. Sports Biomech. 2018, 17, 251–260. [CrossRef] 10. Gruber, A.H.; Boyer, K.A.; Derrick, T.R.; Hamill, J. Impact shock frequency components and attenuation in rearfoot and forefoot running. J. Sport. Health Sci. 2014, 3, 113–121. [CrossRef] 11. Mercer, J.A.; Bates, B.T.; Dufek, J.S.; Hreljac, A. Characteristics of shock attenuation during fatigued running. J. Sports Sci. 2003, 21, 911–919. [CrossRef] 12. Hamill, J.; Derrick, T.R.; Holt, K.G. Shock attenuation and stride frequency during running. Hum. Mov. Sci. 1995, 14, 45–60. [CrossRef] Sensors 2021, 21, 1758 15 of 16 13. García-Pérez, J.A.; Pérez-Soriano, P.; Llana Belloch, S.; Lucas-Cuevas, A.G.; Sánchez-Zuriaga, D. Effects of treadmill running and fatigue on impact acceleration in distance running. Sports Biomech. 2014, 13, 259–266. [CrossRef] 14. Tessutti, V.; Trombini-Souza, F.; Ribeiro, A.P.; Nunes, A.L.; Sacco Ide, C. In-shoe plantar pressure distribution during running on natural grass and asphalt in recreational runners. J. Sci. Med. Sport 2010, 13, 151–155. [CrossRef] [PubMed] 15. Miller, J.R.; Van Hooren, B.; Bishop, C.; Buckley, J.D.; Willy, R.W.; Fuller, J.T. A Systematic Review and Meta-Analysis of Crossover Studies Comparing Physiological, Perceptual and Performance Measures between Treadmill and Overground Running. Sports Med. 2019, 49, 763–782. [CrossRef] [PubMed] 16. García-Pérez, J.A.; Pérez-Soriano, P.; Llana, S.; Martínez-Nova, A.; Sánchez-Zuriaga, D. Effect of overground vs treadmill running on plantar pressure: Influence of fatigue. Gait Posture 2013, 38, 929–933. [CrossRef] [PubMed] 17. Wank, V.; Frick, U.; Schmidtbleicher, D. Kinematics and electromyography of lower limb muscles in overground and treadmill running. Int. J. Sports Med. 1998, 19, 455–461. [CrossRef] [PubMed] 18. Milgrom, C.; Finestone, A.; Segev, S.; Olin, C.; Arndt, T.; Ekenman, I. Are overground or treadmill runners more likely to sustain tibial stress fracture? Br. J. Sports Med. 2003, 37, 160–163. [CrossRef] 19. Mileti, I.; Serra, A.; Wolf, N.; Munoz-Martel, V.; Ekizos, A.; Palermo, E.; Arampatzis, A.; Santuz, A. Muscle Activation Patterns Are More Constrained and Regular in Treadmill than in Overground Human Locomotion. Front. Bioeng. Biotechnol. 2020, 8, 581619. [CrossRef] 20. Eriksson, M.; Halvorsen, K.A.; Gullstrand, L. Immediate effect of visual and auditory feedback to control the running mechanics of well-trained athletes. J. Sports Sci. 2011, 29, 253–262. [CrossRef] 21. Crowell, H.P.; Davis, I.S. Gait retraining to reduce lower extremity loading in runners. Clin. Biomech. 2011, 26, 78–83. [CrossRef] 22. Wood, C.M.; Kipp, K. Use of audio biofeedback to reduce tibial impact accelerations during running. J. Biomech. 2014, 47, 1739–1741. [CrossRef] [PubMed] 23. Clansey, A.C.; Hanlon, M.; Wallace, E.S.; Nevill, A.; Lake, M.J. Influence of tibial shock feedback training on impact loading and running economy. Med. Sci. Sports Exerc. 2014, 46, 973–981. [CrossRef] 24. Nummela, A.T.; Paavolainen, L.M.; Sharwood, K.A.; Lambert, M.I.; Noakes, T.D.; Rusko, H.K. Neuromuscular factors determining 5 km running performance and running economy in well-trained athletes. Eur. J. Appl. Physiol. 2006, 97, 1–8. [CrossRef] 25. Gullstrand, L.; Nilsson, J. A new method for recording the temporal pattern of stride during treadmill running. Sports Eng. 2009, 11, 195–200. [CrossRef] 26. Möckel, G.; Perka, C.; Labs, K.; Duda, G. The influence of walking speed on kinetic and kinematic parameters in patients with osteoarthritis of the hip using a force-instrumented treadmill and standardised gait speeds. Arch. Orthop. Trauma Surg. 2003, 123, 278–282. [CrossRef] 27. Donath, L.; Faude, O.; Lichtenstein, E.; Nüesch, C.; Mündermann, A. Validity and reliability of a portable gait analysis system for measuring spatiotemporal gait characteristics: Comparison to an instrumented treadmill. J. Neuroeng. Rehabil. 2016, 13, 6. [CrossRef] 28. Watkins, C.M.; Maunder, E.; Tillaar, R.V.D.; Oranchuk, D.J. Concurrent Validity and Reliability of Three Ultra-Portable Vertical Jump Assessment Technologies. Sensors 2020, 20, 7240. [CrossRef] 29. Encarnación-Martínez, A.; Berenguer-Vidal, R.; García-Gallart, A.; Rodríguez-Mayol, F.A.; Pernías-Reverte, J.J.; Pérez-Soriano, P. Ergometric Treadmill for Sport Training. US Patent US20200353309A1, 7 May 2019. 30. Encarnación-Martínez, A.; Berenguer-Vidal, R.; García-Gallart, A.; Rodríguez-Mayol, F.A.; Pernías-Reverte, J.J.; Pérez-Soriano, P. Force Measuring Running Machine for Sports Training. CN Patent CN111905333A, 7 May 2019. 31. Encarnación-Martínez, A.; Berenguer-Vidal, R.; García-Gallart, A.; Rodríguez-Mayol, F.A.; Pernías-Reverte, J.J.; Pérez-Soriano, P. Treadmill for Sport Training. EP Patent EP3735900A1, 7 May 2019. 32. Ursel, T.; Olinski, M. Displacement Estimation Based on Optical and Inertial Sensor Fusion. Sensors 2021, 21, 1390. [CrossRef] 33. Rybarczyk, D. Application of the MEMS Accelerometer as the Position Sensor in Linear Electrohydraulic Drive. Sensors 2021, 21, 1479. [CrossRef] 34. Zhang, J.; Huang, W.; Zhang, W.; Li, F.; Du, Y. Train-Induced Vibration Monitoring of Track Slab under Long-Term Temperature Load Using Fiber-Optic Accelerometers. Sensors 2021, 21, 787. [CrossRef] [PubMed] 35. Kim, K.; Sohn, H. Dynamic Displacement Estimation for Long-Span Bridges Using Acceleration and Heuristically Enhanced Displacement Measurements of Real-Time Kinematic Global Navigation System. Sensors 2020, 20, 5092. [CrossRef] [PubMed] 36. Won, J.; Park, J.; Park, J.W.; Kim, I. BLESeis: Low-Cost IoT Sensor for Smart Earthquake Detection and Notification. Sensors 2020, 20, 2963. [CrossRef] 37. Arranz-Martínez, F.; Martín-Ferrer, R.; Palacios-Navarro, G.; Ramos-Lorente, P. Study on the Vibration Characteristics of the Telescope T80 in the Javalambre Astrophysical Observatory (JAO) Aimed at Detecting Invalid Images. Sensors 2020, 20, 6523. [CrossRef] [PubMed] 38. Lucas-Cuevas, A.G.; Encarnación-Martínez, A.; Camacho-García, A.; Llana-Belloch, S.; Pérez-Soriano, P. The location of the tibial accelerometer does influence impact acceleration parameters during running. J. Sports Sci. 2017, 35, 1734–1738. [CrossRef] [PubMed] 39. Lavcanska, V.; Taylor, N.F.; Schache, A.G. Familiarization to treadmill running in young unimpaired adults. Hum. Mov. Sci. 2005, 24, 544–557. [CrossRef] Sensors 2021, 21, 1758 16 of 16 40. Lucas-Cuevas, A.G.; Pérez-Soriano, P.; Llana-Belloch, S.; Macián-Romero, C.; Sánchez-Zuriaga, D. Effect of custom-made and prefabricated insoles on plantar loading parameters during running with and without fatigue. J. Sports Sci. 2014, 32, 1712–1721. [CrossRef] 41. Borg, G.A. Psychophysical bases of perceived exertion. Med. Sci. Sports Exerc. 1982, 14, 377–381. [CrossRef] 42. Van den Berghe, P.; Six, J.; Gerlo, J.; Leman, M.; De Clercq, D. Validity and reliability of peak tibial accelerations as real-time measure of impact loading during over-ground rearfoot running at different speeds. J. Biomech. 2019, 86, 238–242. [CrossRef] 43. Oppenheim, A. Discrete-Time Signal. Processing: Pearson New International Edition; Pearson: London, UK, 2013. 44. Batista, P.; Silvestre, C.; Oliveira, P.; Cardeira, B. Accelerometer Calibration and Dynamic Bias and Gravity Estimation: Analysis, Design, and Experimental Evaluation. IEEE Trans. Control. Syst. Technol. 2011, 19, 1128–1137. [CrossRef] 45. Bland, J.M.; Altman, D.G. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986, 1, 307–310. [CrossRef] 46. Shrout, P.E.; Fleiss, J.L. Intraclass correlations: Uses in assessing rater reliability. Psychol. Bull. 1979, 86, 420–428. [CrossRef] 47. Fleiss, J.L. The Design and Analysis of Clinical Experiments; John Wiley & Sons: New York, NY, USA, 1986. 48. Jacobson, N.S.; Truax, P. Clinical significance: A statistical approach to defining meaningful change in psychotherapy research. J. Consult. Clin. Psychol. 1991, 59, 12–19. [CrossRef] [PubMed] 49. Hong, Y.; Wang, L.; Li, J.X.; Zhou, J.H. Comparison of plantar loads during treadmill and overground running. J. Sci. Med. Sport 2012, 15, 554–560. [CrossRef] [PubMed] 50. Ogueta-Alday, A.; Morante, J.C.; Rodríguez-Marroyo, J.A.; García-López, J. Validation of a new method to measure contact and flight times during treadmill running. J. Strength Cond. Res. 2013, 27, 1455–1462. [CrossRef] 51. Hanley, B.; Tucker, C.B. Gait variability and symmetry remain consistent during high-intensity 10,000 m treadmill running. J. Biomech. 2018, 79, 129–134. [CrossRef] [PubMed] 52. Horvais, N.; Samozino, P.; Chiementin, X.; Morin, J.B.; Giandolini, M. Cushioning perception is associated with both tibia acceleration peak and vibration magnitude in heel-toe running. Footwear Sci. 2019, 11, 35–44. [CrossRef] 53. Garofolini, A.; Taylor, S.; Lepine, J. Evaluating dynamic error of a treadmill and the effect on measured kinetic gait parameters: Implications and possible solutions. J. Biomech. 2019, 82, 156–163. [CrossRef] [PubMed] 54. Letafatkar, A.; Rabiei, P.; Farivar, N.; Alamouti, G. Long-term efficacy of conditioning training program combined with feedback on kinetics and kinematics in male runners. Scand. J. Med. Sci. Sports 2020, 30, 429–441. [CrossRef] 55. Leens, F. An introduction to I 2 C and SPI protocols. IEEE Instrum. Meas. Mag. 2009, 12, 8–13. [CrossRef] 56. Pérez-Soriano, P.; Encarnación-Martínez, A. Análisis de impactos mediante técnicas de acelerometría. In Metodología y Aplicación Práctica de la biomecánica Deportiva; Pérez-Soriano, P., Ed.; Paidotribo: Badalona, Spain, 2018; pp. 9–20.

Validity and Reliability of an Instrumented Treadmill with an Accelerometry System for Assessment of Spatio-Temporal Parameters and Impact Transmission.

03-04-2021

Encarnación-Martínez, Alberto,Pérez-Soriano, Pedro,Sanchis-Sanchis, Roberto,García-Gallart, Antonio,Berenguer-Vidal, Rafael

eng

PMC6528988

RESEARCH ARTICLE Ventilatory efficiency during constant-load test at lactate threshold intensity: Endurance versus resistance exercises Lluis Albesa-Albiol1, Noemı´ Serra-Paya´1, Marı´a Ana Garnacho-Castaño1, Lluis Guirao Cano1,2, Eulogio Pleguezuelos Cobo1,3, Jose´ Luis Mate´-Muñoz4, Manuel V. Garnacho- CastañoID1* 1 GRI-AFIRS, School of Health Sciences, TecnoCampus-Pompeu Fabra University, Mataro´, Barcelona, Spain, 2 Department of Rehabilitation, Hospital Asepeyo, Sant Cugat, Barcelona, Spain, 3 Department of Physical and Rehabilitation Medicine, Hospital de Mataro´, Mataro´, Barcelona, Spain, 4 Department of Physical Activity and Sports Science, Alfonso X El Sabio University, Villanueva de la Cañada, Madrid, Spain * mgarnacho@escs.tecnocampus.cat Abstract There is a lack of evidence about the ventilatory efficiency in resistance exercises despite the key role played in endurance exercises. This study aimed to compare the cardiorespira- tory, metabolic responses and ventilatory efficiency between half-squat (HS) and cycle ergometer exercises during a constant-load test at the lactate threshold (LT) intensity. Eigh- teen healthy male participants were randomly assigned in a crossover design to carry out HS or cycle ergometer tests. For the three HS tests, a one repetition maximum (1RM) test was performed first to determine the load (kg) corresponding to the 1RM percentages. In the second test, the incremental HS exercise was carried out to establish the load (kg) at the LT intensity. Finally, a constant-load HS test was performed at the LT intensity. The first cycle ergometer test was incremental loading to determine the intensity in watts correspond- ing to the LT, followed by a constant-load test at the LT intensity. A recovery time of 48 hours between each test was established. During both constant-load test, cardiorespiratory and metabolic responses were monitored. A significant exercise mode x time interaction effect was only detected in oxygen uptake (VO2), heart rate, and blood lactate (p < 0.001). No differences were found between the two types of exercise in ventilatory efficiency (p >0.05). Ventilation (VE) and carbon dioxide were highly correlated (p <0.001) in the cycle ergometer (r = 0.892) and HS (r = 0.915) exercises. In the VO2 efficiency slope (OUES), similarly significant and high correlations (p <0.001) were found between VO2 and log10 VE in the cycle ergometer (r = 0.875) and in the HS (r = 0.853) exercise. Although the cardioven- tilatory responses were greater in the cycle ergometer test as compared to HS exercise, ventilatory efficiency was very similar between the two exercise modalities in a predomi- nantly aerobic metabolism. PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 1 / 18 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Albesa-Albiol L, Serra-Paya´ N, Garnacho- Castaño MA, Guirao Cano L, Pleguezuelos Cobo E, Mate´-Muñoz JL, et al. (2019) Ventilatory efficiency during constant-load test at lactate threshold intensity: Endurance versus resistance exercises. PLoS ONE 14(5): e0216824. https://doi.org/ 10.1371/journal.pone.0216824 Editor: Daniel Boullosa, James Cook University College of Healthcare Sciences, BRAZIL Received: January 18, 2019 Accepted: April 29, 2019 Published: May 21, 2019 Copyright: © 2019 Albesa-Albiol et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Funding: The author(s) received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. Introduction Recent studies have used the lactate threshold (LT) or the ventilatory threshold as parameters to monitor and assess cardiorespiratory responses [1, 2, 3], slow component of oxygen uptake, and gross mechanical efficiency [4] in unusual resistance exercises using a cardiopulmonary exercise tests (CPET), as also occurs in endurance exercise [5–9]. During constant-load test at a load intensity equivalent to the LT, it was observed a greater cardiorespiratory response to cycle ergometer exercise compared to the half-squat (HS). The cardiorespiratory and meta- bolic response was stable in both types of exercise; greater muscular fatigue was observed after completion of the HS test [2]. As could be expected, resistance exercises increased local muscu- lar fatigue in the lower limbs, while endurance exercises increased cardiorespiratory response. Cardiorespiratory fitness is frequently evaluated by means of ventilatory efficiency [10, 11]. The fundamental cause of ventilatory efficiency is the matching of ventilation (VE) and perfu- sion in the lungs. The mismatching of perfusion and VE diminishes the efficiency of lung gas exchange, demanding an increase in VE for a given CO2 output and arterial PCO2. This mis- matching phenomenon contributes essentially to hyperpnea and dyspnea [12] affecting venti- latory performance. It is common to assess ventilatory efficiency in endurance exercises mostly in different types of diseases or pathologies [13–14], in sports performance [11], and in healthy subjects [10, 15], establishing the slope of the linear relationship between VE and car- bon dioxide (VE/VCO2 slope) during an incremental test up to the anaerobic [16] or ventila- tory threshold [17] and the ventilatory compensation point [10]. Another option to quantify ventilatory efficiency in endurance exercises is to determine the oxygen uptake efficiency slope (OUES). The OUES indicates how effectively oxygen is extracted and taken into the body dur- ing incremental exercise [17]. The OUES is considered a very appropriate tool in the evalua- tion of cardiovascular fitness in overweight adolescents [18], the severity of heart disease [19], the effects of physical training or treatment [20, 21], and the risk of a serious or fatal event [22]. Although many studies have analyzed the slope of VE/VCO2 and OUES by age, sex, fitness level, and diseases in endurance exercises [10, 11, 14, 17, 19], it is unusual to observe studies comparing VE/VCO2 slope and OUES between different exercise modalities [23, 24]. Sun et al. [10] demonstrated that VE/VCO2 slope is not exercise mode-dependent, however, Davis et al. detected that VE/VCO2 slope was lower on the cycle ergometer than the treadmill in women but not in men [25]. For OUES, treadmill values were higher than cycle ergometer [24]. Recently, Salazar-Martinez et al. demonstrated that ventilatory efficiency was unaffected by ergometer type [26]. The assumption that ventilatory efficiency could be similar between different exercise modalities is controversial and more research is needed to compare several exercise modes. Despite the importance that has been given in the scientific literature to the assessment of ventilatory efficiency in endurance exercises in healthy people and especially in the clinical settings, it is a field of knowledge that needs to be explored in resistance exercises. There are no previous data regarding VE/VCO2 slope and OUES in HS exercise and, to the best of our knowledge, ventilatory efficiency has not been compared between resistance and endurance exercises. In cardiorespiratory fitness assessment, this knowledge could have an added value in select- ing the type of exercise to improve ventilation efficiency. If resistance exercises demonstrate adequate ventilatory efficiency, professionals in the health field could use resistance training to increase local muscle endurance while maintaining good ventilatory efficiency. It is common to assess ventilatory efficiency during incremental endurance tests, however, prolonged constant-load endurance tests can be recommended as a good option in the clinical health setting to determine the VE/VCO2 slope [27, 28] or OUES because they do not subject Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 2 / 18 the participants to significant cardiorespiratory, metabolic, and muscular stress. A constant- load test at LT intensity might be an interesting alternative for applying to healthy people in both endurance and resistance exercises to assess ventilatory efficiency without inducing a strenuous cardiorespiratory and metabolic stress. The main objective of this study was to compare the ventilatory efficiency, measured by OUES and VE/VCO2 slope, and cardioventilatory responses of HS and cycle ergometer exer- cise in a constant-load test at LT intensity. A secondary goal was to determine the relationship between the OUES and the VE/VCO2 slope in both exercise modalities in each participant. Material and methods Participants Eighteen healthy male participants were recruited among the students of the Department of Physical Activity and Sports Sciences (age: 21.8 ± 1.5 years, height: 180.3 ± 5.7 cm, weight: 82.6 ± 9.0 kg, body mass index: 25.4 ± 2.0). All participants had at least 6 months of resistance training experience and were completely familiar with the HS exercise and the cycle ergometer. Four exclusion criteria were established: 1) any cardiovascular, metabolic, neurological, pulmonary, or orthopedic disorder that could limit exercise performance, 2) the use of any medication, supplements, or substance that could improve performance, 3) 1RM 150 kg in the exercise of the HS, 4) elite athlete status. Eligible participants were informed of the tests to be performed and those who agreed with the study protocols signed their written consent to participate. The subjects were instructed to abstain from other exercise or training during the two-week study period. The study protocol adhered to the principles of the Declaration of Helsinki for studies with human beings and was approved by the Ethics Committee of the Alfonso X El Sabio University (Villanueva de la Cañada, Madrid, Spain). Experimental design The participants visited the Exercise Physiology Laboratory five times during the two-week study period, at the same time of day (± 2 hours) and in similar environmental conditions (room temperature 21–25˚C, atmospheric pressure 715–730 mm Hg, relative humidity ~ 45%). Participants were randomly assigned in a crossover design to perform HS or cycle ergometer tests. A rest period of 48 hours was established between each of the five tests. The protocols were implemented according to procedures previously established by our research group [2]. For the three HS tests, a one repetition maximum (1RM) test was performed first to deter- mine the load (kg) corresponding to the 1RM percentages to be used during the second test, the incremental HS exercise to establish the load (kg) at the intensity corresponding to the LT. Finally, a constant-load HS test was performed at the LT intensity established during the incre- mental exercise test. The first cycle ergometer test was incremental loading to determine the intensity in watts (W) corresponding to the LT, followed by a constant load test at the LT intensity. During both constant-load test, acute cardiorespiratory and metabolic responses were monitored. The timing of the blood lactate sampling was the same for both the HS and cycle ergometer testing. Half squat tests In the HS tests, a Smith machine (Matrix Fitness, Johnson Health Tech, Cottage Grove, MN, USA) was used to ensure safe and controlled movements. HS technique was determined as in Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 3 / 18 previous studies [3, 29]. The variation in range of motion (ROM) during HS exercise was accu- rately determined during a familiarization session and in all the tests. Participants positioned themselves under the barbell in an upright position with the knees and hips fully extended and legs spread approximately at the shoulders’ width. The barbell was placed on the upper back (trapezius muscle), approximately at the level of the acromion. During the descent of the bar, participants flexed the knees and hips (eccentric action) to lower the barbell in a controlled manner, until 900 flexion of the knees [30]. From this position, the concentric muscle action was started until fully extending the knees and hips. The body position was individually adjusted and exactly replicated on each HS test. Each HS test started with a 5-minute low-intensity general run and 5-minute general joint mobility warm-up, followed by a specific warm-up of a series of 3–5 repetitions (HS) at a rela- tive intensity of 40–60% of 1RM. 1RM test. After a 2-minute rest, the HS test protocols began. To determine 1RM, 3–5 series were carried out, using an increasing weight each time. The 1RM was defined as the last load lifted by the subject, completing a knee extension to the required position. The rest period between each attempt was 4 minutes. Incremental HS test. The incremental HS test was carried out in 6 one-minute series, at relative intensities of 10%, 20%, 25%, 30%, 35%, and 40% 1RM as described in previous studies [2, 4, 29]. In each series, 30 repetitions of 2 seconds each were performed (1 second for eccen- tric muscle action and 1 second for concentric action), using a metronome to establish the rhythm; a member of the research team provided visual and verbal cues to maintain an ade- quate rate. A passive rest period of 2 minutes between series was established. During this period, blood samples were collected by an experienced researcher and the corresponding load was increased. The test ended when the repetitions were no longer executed correctly or was voluntarily terminated by the participant when he could not perform the repetitions at the established cadence. Blood samples (5 μL) were obtained by pricking the finger 30 seconds after the end of each series. Lactate levels were measured using a portable analyzer (Lactate Pro LT-1710, Arkray Factory Inc., KDK Corporation, Siga, Japan). Based on the algorithmic adjustment method described by Orr et al. [31], LT was defined as the load intensity at which blood lactate concentrations begin to increase exponentially [32]. LT was detected by two-segment linear regression, placing the 2 emergent linear regression equations for each segment at the point of intersection between a plot of blood lactate concen- tration and relative intensity [33]. Data analysis was done using Matlab version 7.4 (Math- Works, Natick, MA, USA). Constant-load HS test. In the constant load HS test, 21 sets of 15 repetitions were per- formed. The duration of each set was 30 seconds (1 second each for the eccentric and concen- tric phases, guided by a metronome and visual and verbal signals), with 1-minute rest between sets. The entire constant-load test lasted 31 minutes. Respiratory exchange data were recorded during the constant-load test using a breath-by-breath open-loop gas analyzer (Vmax spectra 29, Sensormedics Corp., Yorba Linda, California, USA), previously calibrated. VO2, VE, VCO2, and respiratory exchange ratio (RER) were monitored. The heart rate was quantified every 5 seconds by telemetry (RS-800CX, Polar Electro OY, Finland). To determine lactate concentrations, finger-prick blood samples were obtained, as described for the incremental test, at rest and 30 seconds after the end of 7 HS sets (S): S3, S6, S9, S12, S15, S18 and S21. Cycle ergometer tests The incremental and constant-load tests on a cycle ergometer (Monark ergomedic 828E, Vansbro, Sweden) included a 5-minute warm-up at a pedaling rate of 50 rev.min-1 and a load Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 4 / 18 of 50 W, followed by 5 minutes of dynamic joint mobility and stretching exercises. The load during the incremental and constant-load tests was defined according to the characteristics of the cycle ergometer, as previously described [34]. Briefly, pedaling at an intensity of 50 W is the same as pedaling at a rate of 50 rev.min-1 at a load of 1-kilogram force (kgf). To increase the load by 25 W during an incremental protocol, pedaling cadence at a rate of 50 rev.min-1 should be performed at a load equivalent to 0.5 kfg. After 2 minutes of rest, the specific tests on the cycle ergometer began. Incremental cycle ergometer test. The incremental test using a ramp protocol that started with a load of 50 W (50 rev.min-1 at a load of 1kgf), increased in steps of 25 W.min-1 until completion of 8 min at a pedaling rate of 50 rev.min-1 at a load of 0.5 kgf. Blood samples (5 μL) were obtained by finger pricking at rest and every 2 minutes during the incremental test. The LT was detected by inspecting the plot of blood lactate concentrations against the workload, according to the protocol described by Weltman et al. [35]. LT was defined as the highest exercise load completed when an increase of 0.5 mmol.L-1 was detected over baseline concentrations in at least 2 consecutive samples. Constant-load cycle ergometer test. The constant load cycle ergometer test was per- formed with continuous pedaling at a rate between 70–80 rev.min-1 at an intensity (W) equiva- lent to the LT, previously determined in the incremental test. The load in kfg was individually adjusted to each subject at 70–80 rev.min-1 to develop the W corresponding to the LT inten- sity. Total duration of the test was 31 minutes. The blood lactate samples were obtained with the same portable analyzer as in the HS test, at the beginning of the test and (coinciding with the timing in the HS test) at the following minutes (M) thereafter: M4, M8.5, M13, M17.5, M22, M26.5, M31. During the constant-load test, respiratory exchange and heart rate data were recorded as described in the HS constant-load test. Ventilatory efficiency The ventilatory efficiency of each participant was determined in two ways: 1) the slope of the relationship between VE and VCO2 during each constant-load test; 2) the OUES slope, calcu- lated as the relationship between VO2 and the logarithm of the VE during the constant-load test: VO2 = a log10 VE + b). Statistical analysis The Shapiro-Wilk test was used to verify the normal distribution of the data, reported as mean, standard deviation (SD), and confidence intervals (95% CI). To identify significant dif- ferences between the HS and cycle ergometer exercises in the cardioventilatory and lactate var- iables, a general linear model was performed with a two-way analysis of variance (ANOVA) for repeated measurements. The two factors were the exercise mode (HS or cycle ergometer) and time point (corresponding to 7 control points in both exercise modes). When appropriate, a post-hoc Bonferroni adjustment was implemented for multiple comparisons. To determine the differences between the two exercise modes in the VE/VCO2 and OUES slopes, Student-t was applied for related samples. The slope of VE/VCO2 and OUES was calculated by linear regression between VE and VCO2 and between VO2 and log10 VE, respectively. The Pearson product-moment correlation coefficients were calculated to determine significant relation- ships between the VE and the VCO2 and between the VO2 and the log10 VE, and to establish the possible relationship between the OUES and the VE/VCO2 slope. Partial eta square (ηp 2) was calculated to determine the magnitude of the response in ANOVA analysis. Cohen´s d for the planned comparisons was used to determine effect sizes. A large effect size was defined as ηp 2  0.26, d  0.80; moderate ηp 2  0.13, d  0.40; and Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 5 / 18 small ηp 2 < 0.02, d < 0.40 [36]. Statistical power (SP) was also determined. The intraclass cor- relation coefficients and the percentage of variation coefficients were calculated to determine the relative and absolute reliability. The level of significance was set at p <0.05. All statistical methods were performed using the SPSS Statistics software package version 23.0 for Macintosh (SPSS, Chicago, IL, USA). The graphics were made in the Microsoft Excel version 16.20 for Mac. Results Anthropometric characteristics and incremental test data for the HS and cycle ergometer exer- cises are shown in Table 1. Differences in cardioventilatory and lactate responses are shown in Table 2. The mean of the intraclass correlation coefficients and the coefficients of variation for cardioventilatory var- iables and lactate were 0.970 (0.942–0.987) and 6.7% ± 3.4%, respectively. For absolute VO2, a significant exercise mode x time interaction effect was observed (p = 0.007, F(6, 102) = 3.18). Bonferroni test confirmed that VO2 was significantly higher in cycle ergometer than HS exercise at the 7 established control points (p <0.001; large effect d  1.64). In cycle ergometer, a significant lower VO2 was detected in M4 regarding the rest of the control points (p  0.002; moderate effect d  0.46 and  0.60). However, a VO2 stabiliza- tion was observed after M8.5 (p > 0.05). In HS exercise, a significant increase in VO2 was observed (p < 0.05) in S3 with respect to S6 (moderate effect, d = 0.46), S18 (large effect, d = 0.95), and S21 (large effect, d = 0.86) (Fig 1A). No significant exercise mode x time interaction effect was found for the relative VO2 (p > 0.05) and VE variable (p > 0.05). For heart rate, a significant effect (p <0.001) was observed for exercise mode x time interac- tion (F(6, 102) = 5.85). The Bonferroni test determined that heart rate was significantly lower in HS exercise than in cycle ergometer test at the 7 established control points (p <0.05; in M4/S3, moderate effect d = 0.49; rest of control points large effect d  0.94). In cycle ergometer exer- cise, a significant increase in heart rate was confirmed in M4 regarding all control points (p < 0.01; moderate effect versus M8.5 and M13, d  0.62 and  0.74; large effect versus M17.5, M22, M26.5, M31, d  0.83) (Fig 1B). Blood lactate concentrations indicated a significant exercise mode x time interaction (p < 0.001, F(7, 119) = 6.93). The Bonferroni adjustments showed a significant increase from rest period in both exercise modes (p 0.005; large effect d  1.71). Significant higher blood Table 1. Descriptive data related to anthropometric characteristics, 1RM- and incremental-load tests. Variables Mean (SD) Participants N = 18 Age (years) 21.2 (1.5) Height (cm) 180.3 (5.7) Weight (kg) 82.6 (9.0) BMI (kg.m-2) 25.4 (2.1) 1RM in HS (kg) 206.3 (36.4) HS load at LT (kg) 51.2 (9.0) HS relative intensity at LT (%) 25.8 (4.6) CYC load at LT (W) 130.8 (24.8) Abbreviations: 1RM = one-repetition maximum; BMI: body mass index; CYC: cycle-ergometer; HS = half-squat; LT = lactate threshold; SD = standard deviation. https://doi.org/10.1371/journal.pone.0216824.t001 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 6 / 18 lactate levels were found in HS exercise regarding cycle ergometer (p <0.05) at control points M22/S15 (moderate effect, d = 0.72), M26.5/S18 (large effect, d = 0.82), and M31/S21 (large effect, d = 1.19) (Fig 2). In the RER, no significant interaction effect was observed for exercise mode x time (p > 0.05). Regarding VE/VCO2 slope and OUES, no differences were found between the two types of exercise (p >0.05) (Fig 3). In the VE/VCO2 slope, VE and VCO2 were highly correlated (p <0.001), both in the cycle ergometer (r = 0.892) and HS (r = 0.915) modalities (Fig 4A and 4B, respectively). In the OUES, similarly high correlations (p <0.001) were found between VO2 and log10 VE in the cycle ergometer (r = 0.875) and in the HS (r = 0.853) (Fig 5A and 5B, respectively). No significant correlation was found between the OUES and the slope of the VE/VCO2 in the HS (r = -0.345, p = 0.160) nor in the cycle ergometer (r = 0.315, p = 0.203). Also, no signifi- cant correlation was observed between the HS and cycle ergometer modes in OUES (r = 0.356, p = 0.147) or VE/VCO2 slope (r = 0.422, p = 0.081). Discussion To the best of our knowledge, this study applied two novelties in methodological approach, with respect to previous research. First, it determined ventilatory efficiency in HS exercise by two distinct methods (VE/VCO2 slope and OUES); second, it compared HS and cycle ergome- ter ventilatory efficiency in constant-load tests conducted at an intensity equivalent to the LT. Although the cardioventilatory responses were greater in the cycle ergometer test as compared to HS, ventilatory efficiency was very similar between the two exercise modalities. In addition, the blood lactate concentrations were similar between both exercise modes although these val- ues were slightly higher in HS exercise than in the cycle ergometer exercise at the end of the constant-load tests. Table 2. Differences in cardioventilatory and lactate responses between half-squat vs cycle-ergometer during constant-load test at lactate threshold intensity. CYC (95% CI) HS (95% CI) P1 ES/SP P2 ES/SP P3 ES/SP VO2 (L.min-1) 2.2 1.6 0.007 < 0.001 < 0.001 (2.0–2.5) (1.5–1.7) 0.2/0.9 0.6/1.0 0.6/1.0 VO2 (mL.kg-1.min-1) 19.8 27.8 0.517 < 0.001 < 0.001 (18.6–20.9) (24.8–30.8) (0.1–0.3) (0.2–1.0) (0.6–1.0) VCO2 (L.min-1) 2.1 1.5 0.062 < 0.001 < 0.001 (1.8–2.3) (1.4–1.6) 0.1/0.7 0.6/1.0 0.6/1.0 VE (L.min-1) 53.7 43.1 0.510 < 0.001 0.002 (48.2–59.2) (40.1–46.1) 0.1/0.3 0.4/1.0 0.5/0.9 RER 0.9 0.9 0.923 < 0.001 0.084 (0.9–0.9) (0.9–1.0) 0.0/0.1 0.5/1.0 0.2/0.4 HR (beat.min-1) 139.6 123.8 < 0.001 < 0.001 < 0.001 (131.2–148.0) (116.7–130.8) 0.3/1.0 0.6/1.0 0.6/1.0 Lactate (mmol.L-1) 2.6 2.8 < 0.001 < 0.001 0.148 (2.2–2.9) (2.6–3.1) 0.3/1.0 0.8/1.0 0.1/0.3 Abbreviations used: CYC: cycle-ergometer; ES: effect size; HR: heart rate; HS: half-squat; L: liter; min: minute; RER: respiratory exchange ratio; SP: statistical power; VCO2: carbon dioxide production; VE: minute ventilation; VO2: oxygen uptake. P1 Significant differences for exercise mode x time interaction effect. P2 Significant differences for time effect. P3 Significant differences for exercise mode effect. Data are provided as mean and 95% confidence intervals (95% CI). https://doi.org/10.1371/journal.pone.0216824.t002 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 7 / 18 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 8 / 18 These findings replicated the results obtained in previous investigations, in which the cardiorespiratory responses were higher in the cycle ergometer test than in HS exercise [2]. The constant-load HS test at LT intensity likely induced a lower cardioventilatory response because a rest time was implemented between sets. To date, it has been unfeasible to perform a continuous protocol in the HS exercise at the LT intensity. In theory, a continuous HS proto- col would increase intramuscular pressure leading to augmented muscle tension and progres- sive fatigue. These physiological mechanisms would produce the collapse of capillaries and diminish the oxygen available into the muscle and thus increasing the blood lactate levels [37]. Although it is usual to find a different cardiorespiratory response between several endurance exercise modalities at the same relative intensity [38], the available studies comparing resis- tance versus endurance exercises during constant-load test at LT intensity are currently insuf- ficient to draw more precise conclusions. The VE/VCO2 slope and OUES results obtained in both exercises are considered normal and comparable to other studies with healthy adults (19–30 in VE/VCO2 slope, 2.55 ± 1.01 n = 417 in OUES) [10, 16, 24]. In elite youth cyclists [11], the slope of the VE/VCO2 was simi- lar (about 28) to our study, but the OUES was higher: 3.8 vs. 2.5 in our study. The difference could be due to the novel methodology used in our study and the greater cardiorespiratory fit- ness of elite youth cyclists. No studies are available for comparison of the VE/VCO2 slope, cycle ergometer values, or HS data in a constant-load test at LT intensity. However, our results on ventilatory efficiency were very similar to those obtained in other studies in endurance exercises (cycling) at the intensity of the anaerobic threshold [11], perhaps because both Fig 1. Multiple comparisons between cycle ergometer (CYC) and half-squat (HS): (A) Oxygen uptake (VO2). (B) Heart rate (HR). δ Significant differences p < 0.05 between cycle ergometer and half-squat at each checkpoint. † Significantly different from M8.5, M13, M17.5, M22, M26.5, M31 in cycle ergometer, p < 0.01. ⍵ Significantly different from M8.5, M17.5 in cycle ergometer, p = 0.017. ⏚ Significantly different from S6, S18, S21 in HS exercise, p < 0.05. ⏆ Significantly different from S21 in HS exercise, p = 0.026. https://doi.org/10.1371/journal.pone.0216824.g001 Fig 2. Multiple comparisons between cycle ergometer (CYC) and half-squat (HS) in blood lactate. ⏚ Significantly different from S3, S6, S9, S12, S15, S18, S21 in HS exercise, p < 0.001. † Significantly different from M4, M8.5, M13, M17.5, M22, M26.5, M31 in cycle ergometer, p < 0.01. δ Significantly different from cycle ergometer in M22/S15, M26.5/S18, M31/S21, p < 0.05. ⍵ Significantly different from M4 in cycle ergometer, p = 0.028. ⏆ Significantly different from S3 and S6 in HS exercise, p < 0.05. https://doi.org/10.1371/journal.pone.0216824.g002 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 9 / 18 Fig 3. Differences between cycle ergometer (CYC) and half-squat (HS) in the VE/VCO2 slope and OUES. No significant differences between both exercise modalities. https://doi.org/10.1371/journal.pone.0216824.g003 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 10 / 18 Fig 4. Linear relationship between ventilation (VE) and carbon dioxide (VE/VCO2 slope): (A) Cycle ergometer (CYC). (B) Half-squat (HS). https://doi.org/10.1371/journal.pone.0216824.g004 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 11 / 18 Fig 5. Relationship between oxygen uptake (VO2) and log10 VE (OUES): (A) Cycle ergometer (CYC). (B) Half-squat (HS). https://doi.org/10.1371/journal.pone.0216824.g005 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 12 / 18 intensities (LT and anaerobic thresholds) reflect a similar metabolic moment, beyond which lactate concentrations begin to increase. Our data from both exercise modes in healthy young adults verify that this protocol could be another option to evaluate the slope of VE/VCO2 and OUES in a mostly aerobic metabolism, controlling the acidosis of the body, without having to reach the high intensities and avoiding a higher cardiorespiratory stress that could become problematic in some pathologies. Probably, during a constant-load test at moderate intensity (LT) the relationship between VE and both VCO2 and VO2 is normally stable and uniform before the onset of ventilatory compensation for the exercise-induced lactic acidosis [10], justi- fying, at least in part, the similarities detected in VE/VCO2 slope and OUES between both exercise modalities at the same metabolic state. A surprising aspect of this study was the almost identical values in the ventilatory efficiency observed in the HS and cycle ergometer tests. Studies comparing the VE/VCO2 slope and the OUES in different types of exercises are rare; therefore, there is a significant lack of informa- tion about which exercise modality could induce a higher ventilatory efficiency. Our findings show that two types of exercise with different cardioventilatory responses induce the same ven- tilatory efficiency at similar metabolic intensity. During incremental exercise tests [23], no sig- nificant changes were found between treadmill and cycle ergometer trials, although both exercise modalities showed a lower VE/VCO2 slope (higher efficiency) compared to a robot- ics-assisted tilt table. A study compared OUES in 17 healthy subjects in two exercise modali- ties, observing higher values in the treadmill test compared to the cycle ergometer [24]. Although further evidence is needed, ventilatory efficiency could be dependent on the type of exercise, test protocol, and mode of assessing ventilatory efficiency (OUES vs VE/VCO2 slope). It was expected that subjects with a lower VE/VCO2 slope (greater efficiency) throughout each of the tests would increase their OUES. The lack of significant correlation between the OUES and the slope of the VE/VCO2 in the two exercise modalities analyzed indicates that those subjects who showed greater ventilatory efficiency in the HS did not achieve greater ven- tilatory efficiency in the cycle ergometer. The OUES has been accepted as a valid submaximal measure of the function and prognosis of disease [39], and the slope of VE/VCO2 is a reliable assessment in healthy adults [40] and in those with pathologies [41]. However, their usefulness in healthy and athletically trained people is dubious. It is not yet clear which factors contribute to modify ventilatory efficiency during exercise, but the established postulates through this dis- cussion may be more relevant in the clinical field than in fitness and sports performance because it seems that the VE/VCO2 slope did not change in elite cyclists after 16 weeks of train- ing [11] and, regardless of gender, in children [42] and healthy adults [10] engaging in exer- cise. Training did not improve the OUES in healthy subjects [43] and could have a limited effect in athletes [11]. As a practical application, these findings could be an interesting alternative for the processes of physical rehabilitation and recovery from diseases associated with a loss of strength and muscle mass. For example, patients with heart failure are characterized by a significant loss of muscle mass, and these same physiological mechanisms are closely related to dyspnea and ven- tilatory fatigue [44]. Therefore, ventilatory efficiency is related to the severity of heart failure with reduced ejection fraction [45, 46] and, as a corollary, poor ventilatory efficiency is related to increased morbidity and mortality. In addition, it is common to diagnose strength and mus- cle loss (sarcopenia) in older adults. Sarcopenia is a prevalent syndrome associated with pre- mature mortality in elderly [47]. Resistance exercises at LT intensity could increase local muscular endurance avoiding the losses of strength and muscular mass and, in addition, with the same ventilatory efficiency that could produce the cycle ergometer exercise. Unfortunately, our arguments cannot be consolidated with previous studies in different pathologies, as data Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 13 / 18 clarifying the effect of the resistance exercises to LT intensity in patients with heart failure or sarcopenia are not available; therefore, these observations remain purely intuitive and specula- tive. It is clear that the combination of both resistance and endurance training has improved exercise capacity and diastolic function in patients with heart failure with reduced ejection fraction [48]. Accordingly, the combination of resistance exercises and endurance exercises could be an adequate methodology to increase cardiorespiratory response (endurance exer- cises) on the one hand and strength and muscular resistance (resistance exercises) on the other hand. There are some limitations in this study with regard to the HS exercise protocol, which should be considered. The experimental procedures established in both incremental tests prompt controversial debate with regards to the location of the LT. Consequently, the relative intensity or external load prescribed in each exercise could have been different during both constant-load tests. In this case, an important bias would occur when comparing ventilatory efficiency, cardiorespiratory and metabolic responses between both exercises. However, the results reported by our research group in a recent study [49] revealed that the detection of the LT in both exercises using this same methodology could occur at a similar metabolic instant and relative intensity according to the criteria defined by Binder et al. [50]. In both incremen- tal tests, an equivalent load intensity was produced at the LT, however, cardiorespiratory response was higher in cycle ergometer than in HS exercise during constant-load tests. It is habitual to observe an unequal cardioventilatory response when several exercise modes are compared at the same relative intensity or external load [51]. An identical trend was found in other studies that compared blood lactate, RER and cardiorespiratory responses in various exercises at lower and moderate intensities [38, 52]. Probably, cardiorespiratory responses are exercise mode-dependent at the same metabolic intensity and, therefore, these differences seem larger and more important to considerer at lighter and moderate intensities [38]. We cannot fail to mention that the recovery time established between each series is a key factor in maintaining low and stable levels of blood lactate in a primarily aerobic metabolism. It is assumed that this rest period would mainly affect the mechanisms of cardioventilatory recovery. However, our research group has observed in preliminary trials (unpublished data) that the combination of resistance exercises (in the form of circuit training), without rest between exercises, could keep blood lactate concentrations low and stable. The results stated in this study have important implications for our understanding of the load intensity and the recovery time that regulate ventilatory efficiency in a predominantly aerobic metabolism in HS exercise. Probably, a discontinuous constant-load HS test might induce a similar metabolic intensity and ventilatory efficiency as occurred during continuous constant-load cycle ergom- eter test. Further studies are needed to determine if the hypothetical increase in VO2 and venti- lation associated with a continuous protocol without recovery time between series would increase ventilatory efficiency in the resistance exercises to LT intensity. Conclusions Our findings showed that: 1. Cardioventilatory response was lower in HS exercise than in cycle ergometer during a con- stant-load test at LT intensity. 2. Ventilatory efficiency was equally efficient in the HS resistance exercise and in cycle ergom- eter exercise in a predominantly aerobic metabolism, which could have a significant impact in healthy people. Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 14 / 18 3. There was no correlation between the OUES and the slope of the VE/VCO2 in the two exer- cise modalities analyzed. Those subjects who showed greater ventilatory efficiency in the HS did not achieve higher ventilatory efficiency in the cycle ergometer. 4. Performing a constant-load HS protocol at LT intensity does not generate significant cardiorespiratory stress, while ventilatory efficiency is maintained and muscle strength and local muscular endurance, as well as gross mechanical efficiency, may improve according to previous findings of our research group. Further research is needed to analyze ventilatory efficiency for better understanding of ven- tilatory mechanisms that conditioning resistance exercises performance in a predominantly aerobic metabolism. Supporting information S1 File. Statistical analysis performed with the data obtained during constant-load test. (DOC) S1 Fig. Results for the preparation of the figures. (XLSX) Acknowledgments We thank all our participants who volunteered to take part in this study. Author Contributions Conceptualization: Jose´ Luis Mate´-Muñoz, Manuel V. Garnacho-Castaño. Data curation: Lluis Albesa-Albiol, Manuel V. Garnacho-Castaño. Formal analysis: Jose´ Luis Mate´-Muñoz, Manuel V. Garnacho-Castaño. Investigation: Lluis Albesa-Albiol, Noemı´ Serra-Paya´, Marı´a Ana Garnacho-Castaño, Lluis Guirao Cano, Eulogio Pleguezuelos Cobo, Jose´ Luis Mate´-Muñoz, Manuel V. Garnacho- Castaño. Methodology: Lluis Albesa-Albiol, Noemı´ Serra-Paya´, Marı´a Ana Garnacho-Castaño, Lluis Guirao Cano, Eulogio Pleguezuelos Cobo, Jose´ Luis Mate´-Muñoz, Manuel V. Garnacho- Castaño. Supervision: Lluis Albesa-Albiol, Noemı´ Serra-Paya´, Marı´a Ana Garnacho-Castaño, Lluis Guirao Cano, Eulogio Pleguezuelos Cobo, Jose´ Luis Mate´-Muñoz, Manuel V. Garnacho- Castaño. Validation: Lluis Albesa-Albiol, Noemı´ Serra-Paya´, Marı´a Ana Garnacho-Castaño, Lluis Guirao Cano, Eulogio Pleguezuelos Cobo, Jose´ Luis Mate´-Muñoz, Manuel V. Garnacho- Castaño. Visualization: Lluis Albesa-Albiol, Noemı´ Serra-Paya´, Marı´a Ana Garnacho-Castaño, Lluis Guirao Cano, Eulogio Pleguezuelos Cobo, Jose´ Luis Mate´-Muñoz, Manuel V. Garnacho- Castaño. Writing – original draft: Lluis Albesa-Albiol, Jose´ Luis Mate´-Muñoz, Manuel V. Garnacho- Castaño. Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 15 / 18 Writing – review & editing: Lluis Albesa-Albiol, Noemı´ Serra-Paya´, Marı´a Ana Garnacho- Castaño, Lluis Guirao Cano, Eulogio Pleguezuelos Cobo, Jose´ Luis Mate´-Muñoz, Manuel V. Garnacho-Castaño. References 1. de Sousa NM, Magosso RF, Pereira GB, Souza MV, Vieira A, Marine DA, et al. Acute cardiorespiratory and metabolic responses during resistance exercise in the lactate threshold intensity. Int J Sports Med. 2012; 33: 108–113. https://doi.org/10.1055/s-0031-1286315 PMID: 22127560 2. Garnacho-Castaño MV, Domı´nguez R, Mate´-Muñoz JL. Understanding the meaning of lactate thresh- old in resistance exercises. Int J Sports Med. 2015; 36: 1–7. https://doi.org/10.1055/s-0034-1384545 3. Mate´-Muñoz JL, Domı´nguez R, Lougedo JH, Garnacho-Castaño MV. The lactate and ventilatory thresholds in resistance training. Clin Physiol Funct Imaging. 2016: 1–7. https://doi.org/10.1111/cpf. 12327 PMID: 26749178 4. Garnacho-Castaño MV, Albesa-Albiol L, Serra-Paya N, Gomis Bataller M, Pleguezuelos Cobo E, Guirao Cano L, et al. Oxygen uptake slow component and the efficiency of resistance exercises. J Strength Cond Res. 2018. https://doi.org/10.1519/JSC.0000000000002905 PMID: 30335719 5. Bacon L, Kern M. Evaluating a test protocol for predicting maximum lactate steady state. J Sports Med Phys Fitness. 1999; 39(4): 300–8. PMID: 10726430 6. Coen B, Urhausen A, Kindermann W. Individual anaerobic threshold: methodological aspects of its assessment in running. Int J Sports Med. 2001; 22: 8–16. https://doi.org/10.1055/s-2001-11332 PMID: 11258646 7. MacIntosh BR, Esau S, Svedahl K. The lactate minimum test for cycling: estimation of the maximal lac- tate steady state. Can J Appl Physiol. 2002; 27(3): 232–49. PMID: 12180316 8. Ribeiro LFP, Balakian Jr P, Malachias P, Baldissera V. Stage length, spline function and lactate mini- mum swimming speed. J Sports Med Phys Fitness. 2003; 43(3): 312–8. PMID: 14625512 9. Garnacho-Castaño MV, Palau-Salvà G, Cuenca E, Muñoz-Gonza´lez A, Garcı´a-Ferna´ndez P, Del Car- men Lozano-Estevan M et al. Effects of a single dose of beetroot juice on cycling time trial performance at ventilatory thresholds intensity in male triathletes. 2018; 15(1):49. https://doi.org/10.1186/s12970- 018-0255-6 PMID: 30286760 10. Sun XG, Hansen JE, Garatachea N, Storer TW, Wasserman K. Ventilatory efficiency during exercise in healthy subjects. Am J Respir Crit Care Med. 2002; 166(11): 1443–1448. https://doi.org/10.1164/rccm. 2202033 PMID: 12450934 11. Brown SJ, Raman A, Schlader Z, Stannard SR. Ventilatory efficiency in juvenile elite cyclists. J Sci Med Sport. 2013; 16(3): 266–270. https://doi.org/10.1016/j.jsams.2012.06.010 PMID: 22840997 12. Reindl I, Kleber FX. Exertional hyperpnea in patients with chronic heart failure is a reversible cause of exercise intolerance. Basic Res Cardiol. 1996; 91: 37–43 PMID: 8896742 13. Arena R, Myers J, Guazzi M. The clinical and research applications of aerobic capacity and ventilatory efficiency in heart failure: An evidence-based review. Heart Fail Rev. 2008; 13(2): 245–269. https://doi. org/10.1007/s10741-007-9067-5 PMID: 17987381 14. Chlif M, Chaouachi A, Ahmaidi S. Effect of aerobic exercise training on ventilatory efficiency and respi- ratory drive in obese subjects. Respir Care. 2017; 62(7): 936–946. https://doi.org/10.4187/respcare. 04923 PMID: 28442632 15. Koch B, Scha¨per C, Ittermann T, Spielhagen T, Do¨rr M, Vo¨lzke H, et al. Reference values for cardiopul- monary exercise testing in healthy volunteers: The SHIP study. Eur Respir J. 2009; 33(2): 389–397. https://doi.org/10.1183/09031936.00074208 PMID: 18768575 16. Brown SJ, Brown JA. Heart rate variability and ventilatory efficiency. Int J Sports Med. 2009; 30(7): 496–502. https://doi.org/10.1055/s-0028-1112146 PMID: 19301223 17. Baba R, Nagashima M, Goto M, Nagano Y, Yokota M, Tauchi N, et al. Oxygen uptake efficiency slope: a new index of cardiorespiratory functional reserve derived from the relationship between oxygen con- sumption and minute ventilation during incremental exercise. J Am Coll Cardiol. 1996; 59: 55–62. 18. Drinkard B, Roberts MD, Ranzenhofer LM, Han JC, Yanoff LB, Merke DP, et al. Oxygen uptake effi- ciency slope as a determinant of fitness in overweight adolescents. Med Sci Sports Exerc. 2007; 39: 1811–1816. https://doi.org/10.1249/mss.0b013e31812e52b3 PMID: 17909409 19. Arena R, Brubaker P, Moore B, Kitzman D. The oxygen uptake efficiency slope is reduced in older patients with heart failure and a normal ejection fraction. Int J Cardiol. 2010; 144: 101–102. https://doi. org/10.1016/j.ijcard.2008.12.143 PMID: 19174312 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 16 / 18 20. Van Laethem C, Goethals M, Verstreken S, Walravens M, Wellens F, De Proft M, et al. Response of the oxygen uptake efficiency slope to orthotopic heart transplantation: lack of correlation with changes in central hemodynamic parameters and resting lung function. J Heart Lung Transplant. 2007; 26(9): 921–926. https://doi.org/10.1016/j.healun.2007.06.001 PMID: 17845931 21. Van Laethem C, Van De Veire N, De Backer G, Bihija S, Seghers T, Cambier D, el al. Response of the oxygen uptake efficiency slope to exercise training in patients with chronic heart failure. Eur J Heart Fail. 2007; 9: 625–629. https://doi.org/10.1016/j.ejheart.2007.01.007 PMID: 17347033 22. Myers J, Arena R, Dewey F, Bensimhon D, Abella J, Hsu L, et al. A cardiopulmonary exercise testing score for predicting outcomes in patients with heart failure. Am Heart J. 2008; 156: 1177–1183. https:// doi.org/10.1016/j.ahj.2008.07.010 PMID: 19033016 23. Saengsuwan J, Nef T, Laubacher M, Hunt KJ. Submaximal cardiopulmonary thresholds on a robotics- assisted tilt table, a cycle and a treadmill: a comparative analysis. Biomed Eng Online. 2015; 14(1): 104. https://doi.org/10.1186/s12938-015-0099-0 PMID: 26555762 24. Sun XG, Hansen JE, Stringer WW, Ward SA. Oxygen uptake efficiency plateau: Physiology and refer- ence values. Eur J Appl Physiol. 2012; 112(3): 919–928. https://doi.org/10.1007/s00421-011-2030-0 PMID: 21695524 25. Davis JA, Tyminski TA, Soriano AC, Dorado S, Costello KB, Sorrentino KM, Pham PH. Exercise test mode dependency for ventilatory efficiency in women but not men. Clin Physiol Funct Imaging. 2006; 26(2): 72–78. https://doi.org/10.1111/j.1475-097X.2006.00657.x PMID: 16494595 26. Salazar-Martı´nez E, de Matos TR, Arrans P, Santalla A, Orellana JN. Ventilatory efficiency response is unaffected by fitness level, ergometer type, age or body mass index in male athletes. Biol Sport. 2018; 35(4): 393–398 https://doi.org/10.5114/biolsport.2018.78060 PMID: 30765925 27. Hoshimoto-Iwamoto M, Koike A, Nagayama O, Tajima A, Uejima T, Adachi H, et al. Determination of the VE/VCO2 slope from a constant work-rate exercise test in cardiac patients. J Physiol Sci. 2008; 58 (4): 291–295. https://doi.org/10.2170/physiolsci.RP006108 PMID: 18647443 28. Form P, Reviewing P, Sheet C. Comparative determination of ventilatory efficiency from constant load and incremental exercise testing. Cell Mol Biol Res. 1995; 41(3): 207–216. 29. Garnacho-Castaño MV, Domı´nguez Herrera R, Ruiz Solano P, Mate´-Muñoz JL. Acute physiological and mechanical responses during resistance exercise executed at the lactate threshold workload. J Strength Cond Res. 2015; 29(10): 2867–2873. https://doi.org/10.1519/JSC.0000000000000956 PMID: 25844868 30. Hartmann H, Wirth K, Klusemann M. Analysis of the load on the knee joint and vertebral column with changes in squatting depth and weight load. Sports Med. 2013; 43(10): 993–1008. https://doi.org/10. 1007/s40279-013-0073-6 PMID: 23821469 31. Orr GW, Green HJ, Hughson RL, Bennett GW. A computer linear regression model to determine venti- latory anaerobic threshold. J Appl Physiol 1982; 52: 1349–1352 https://doi.org/10.1152/jappl.1982.52. 5.1349 PMID: 7096157 32. Wasserman K, Mcllroy MB. Detecting the threshold of anaerobic metabolism in cardiac patients during exercise. Am J Cardiol 1964; 14: 844–852 PMID: 14232808 33. de Sousa NMF, Magosso RF, Pereira GB, Leite RD, Arakelian VM, Montagnolli AN, Andrade S, Baldis- sera V. The measurement of lactate threshold in resistance exercise: a comparison of methods. Clin Physiol Funct Imaging. 2011; 31(5): 376–381. https://doi.org/10.1111/j.1475-097X.2011.01027.x PMID: 21771257 34. Domı´nguez R, Garnacho-Castaño MV, Cuenca E, Garcı´a-Ferna´ndez P, Muñoz-Gonza´lez A, de Jesu´s F, et al. Effects of Beetroot Juice Supplementation on a 30-s High-Intensity Inertial Cycle Ergometer Test. Nutrients. 2017; 9(12):1360. 35. Weltman A, Snead D, Stein P, Seip R, Schurrer R, Rutt R, et al. Reliability and validity of a continuous incremental treadmill protocol for the determination of lactate threshold, fixed blood lactate concentra- tions, and VO2max. Int J Sports Med. 1990; 11: 26–32. https://doi.org/10.1055/s-2007-1024757 PMID: 2318561 36. Cohen J. A power primer. Psychol Bull. 1992; 112(1):155–9 PMID: 19565683 37. Petrofsky JS, Phillips CA, Sawka MN, Hanpeter D, Stafford D. Blood flow and metabolism during iso- metric contractions in cat skeletal muscle. J Appl Physiol 1981; 50: 493–502. https://doi.org/10.1152/ jappl.1981.50.3.493 PMID: 7251439 38. Abrantes C, Sampaio J, Reis V, Sousa N, Duarte J. Physiological responses to treadmill and cycle exer- cise. Int J Sports Med. 2012; 33: 26–30. https://doi.org/10.1055/s-0031-1285928 PMID: 22052028 39. Akkerman M, van Brussel M, Hulzebos E, Vanhees L, Helders PJ, Takken T. The oxygen uptake effi- ciency slope: what do we know?. J Cardiopulm Rehabil Prev. 2010; 30(6): 357–373. https://doi.org/10. 1097/HCR.0b013e3181ebf316 PMID: 20724931 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 17 / 18 40. Davis JA, Sorrentino KM, Ninness EM, Pham PH, Dorado S, Costello KB. Test–retest reliability for two indices of ventilatory efficiency measured during cardiopulmonary exercise testing in healthy men and women. Clin Physiol Funct Imaging. 2006; 26(3): 191–196. https://doi.org/10.1111/j.1475-097X.2006. 00674.x PMID: 16640516 41. Chua TP, Ponikowski P, Harrington D, Anker S, Webb-Peploe K, Clark AL, et al. Clinical correlates and prognostic significance of the ventilatory response to exercise in chronic heart failure. J Am Coll Cardiol. 1997; 29: 1585–1590. PMID: 9180123 42. Guerrero L, Naranjo J, Carranza MD. Influence of gender on ventilatory efficiency during exercise in young children. J Sports Sci. 2008; 26: 1455–1457. https://doi.org/10.1080/02640410802255771 PMID: 18923953 43. Mourot L, Perrey S, Tordi N, Rouillon JD. Evaluation of fitness level by the oxygen uptake efficiency slope after a short-term intermittent endurance training. Int J Sports Med. 2004; 25(2): 85–91. https:// doi.org/10.1055/s-2004-819943 PMID: 14986189 44. Keller-Ross ML, Johnson BD, Carter RE, Joyner MJ, Eisenach JH, Curry TB, et al. Improved ventilatory efficiency with locomotor muscle afferent inhibition is strongly associated with leg composition in heart failure. Int J Cardiol. 2016; 202: 159–166. https://doi.org/10.1016/j.ijcard.2015.08.212 PMID: 26397403 45. Arena R, Myers J, Aslam SS, Varughese EB, Peberdy MA. Peak VO2 and VE/VCO2 slope in patients with heart failure: a prognostic comparison. Am Heart J. 2004; 147: 354–60. https://doi.org/10.1016/j. ahj.2003.07.014 PMID: 14760336 46. Cohn JN, Rector TS. Prognosis of congestive heart failure and predictors of mortality. Am J Cardiol. 1988; 62: 25A–30A. PMID: 3389302 47. Brown JC, Harhay MO, Harhay MN. Sarcopenia and mortality among a population-based sample of community-dwelling older adults. J Cachexia Sarcopenia Muscle 2016; 7(3): 290–298. https://doi.org/ 10.1002/jcsm.12073 PMID: 27239410 48. Nolte K, Schwarz S, Gelbrich G, Mensching S, Siegmund F, Wachter R, et al. Effects of long-term endurance and resistance training on diastolic function, exercise capacity, and quality of life in asymp- tomatic diastolic dysfunction vs. heart failure with preserved ejection fraction. ESC heart failure. 2014; 1(1): 59–74. https://doi.org/10.1002/ehf2.12007 PMID: 28834666 49. Garnacho-Castaño MV, Albesa-Albiol L, Serra-Paya´ N, Gomis Bataller M, Felı´u-Ruano R, Guirao Cano L, Pleguezuelos Cobo E, Mate´-Muñoz JL. The slow component of oxygen uptake and efficiency in resis- tance exercises: A comparison with endurance exercises. Front Physiol. 2019; 10: 357. https://doi.org/ 10.3389/fphys.2019.00357 PMID: 31019469 50. Binder RK, Wonisch M, Corra U, Cohen-Solal A, Vanhees L, Saner H, Schmid JP. Methodological approach to the first and second lactate threshold in incremental cardiopulmonary exercise testing. Eur J Cardiovasc Prev Rehabil. 2008; 15(6): 726–734. https://doi.org/10.1097/HJR.0b013e328304fed4 PMID: 19050438 51. Orr JL, Williamson P, Anderson W, Ross R, McCafferty S, Fettes P. Cardiopulmonary exercise testing: arm crank vs cycle ergometry. Anesth. 2013: 68(5): 497–501. 52. Thomas TR, Ziogas G, Smith T, Zhang Q, Londeree BR. Physiological and perceived exertion responses to six modes of submaximal exercise. Res Q Exerc Sport. 1995; 66: 239–246. https://doi. org/10.1080/02701367.1995.10608838 PMID: 7481085 Ventilatory efficiency in resistance exercises at lactate threshold intensity PLOS ONE | https://doi.org/10.1371/journal.pone.0216824 May 21, 2019 18 / 18

Ventilatory efficiency during constant-load test at lactate threshold intensity: Endurance versus resistance exercises.

05-21-2019

Albesa-Albiol, Lluis,Serra-Payá, Noemí,Garnacho-Castaño, María Ana,Guirao Cano, Lluis,Pleguezuelos Cobo, Eulogio,Maté-Muñoz, José Luis,Garnacho-Castaño, Manuel V

eng

PMC6720997

International Journal of Environmental Research and Public Health Article Variations of Internal and External Load Variables between Intermittent Small-Sided Soccer Game Training Regimens Filipe Manuel Clemente 1,2 , Pantelis Theodoros Nikolaidis 3 , Thomas Rosemann 4 and Beat Knechtle 4,5,* 1 School of Sport and Leisure, Polytechnic Institute of Viana do Castelo, 4960-320 Melgaço, Portugal 2 Instituto de Telecomunicações, Delegação da Covilhã, 6200-001 Covilha, Portugal 3 Exercise Physiology Laboratory, 18450 Nikaia, Greece 4 Institute of Primary Care, University of Zurich, 8091 Zurich, Switzerland 5 Medbase St. Gallen Am Vadianplatz, 9001 St. Gallen, Switzerland * Correspondence: beat.knechtle@hispeed.ch; Tel.: +41-(0)-71-226-9300 Received: 24 July 2019; Accepted: 13 August 2019; Published: 15 August 2019



 Abstract: The purpose of this study was twofold: (i) analyze the variations of internal and external load between intermittent regimens (6 × 3’ and 3 × 6’) during a small-sided game (SSG); and (ii) analyze the variations of internal and external load within-intermittent regimens (between sets). Ten male amateur soccer players (age: 21.7 ± 2.1 years) participated in this study. Almost certain large decreases in total distance (−8.6%, [−12.3; −4.8], Effect Size (ES): −1.51, [−2.20; −0.82]) and running distance (−34.0%, [47.0; −17.8], ES: −2.23, [−3.40; −1.05]) were observed when comparing the 3 × 6’ and 6 × 3’. Very likely moderate and large decreases in total accelerations (−24.0%, [−35.1; −10.9]; ES: −1.11, [−1.75; −0.47]) and total of decelerations (−26.7%, [−38.8; −12.1]; ES:−1.49, [−2.36; −0.62]), respectively, were found when comparing the 3 × 6’ and 6 × 3’. Very likely increases in rated of perceived exertion in the set 3 in comparison to the 1st during the 3 × 6’ SSG (34.5%, [12.4; 61.0], ES: 1.35, [0.53; 2.16]) and the 6 × 3’ (29.9%, [11.6; 51.2]; ES: 1.17, [0.49; 1.85]). Longer sets increase the perception of effort and contribute to a large decrease in total and running distances, and total of accelerations and decelerations. Meaningful decreases in time-motion demands occur between sets 2 and 3 while perceived effort increases. Keywords: association football; drill-based tasks; intermittent exercises; physiological; physical; performance 1. Introduction Small-sided games (SSGs) are very popular exercise drills designed by coaches to replicate official match dynamics and increase the intensity and individual participation of players during soccer training sessions [1,2]. In SSGs, the format of play (number of players involved), pitch size, and some rules can be manipulated to adjust the exertion required by players to meet the coach’s proposed objective [3,4]. One of the advantages of SSGs is that, if properly designed, they may represent an effective strategy for multicomponent training [5], allowing for the development of both physical/physiological and technical/tactical skills at the same time [6]. SSGs are often used to promote new affordances and to adjust the tactical complexity to the main goal of the coach, improving the decision making of players [7]. In fact, small variations in these games may promote a significant change in the player’s behavior, thus resulting in consequences for the overall intensity of exercise [8]. Despite these games being promoted for improving the collective Int. J. Environ. Res. Public Health 2019, 16, 2923; doi:10.3390/ijerph16162923 www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2019, 16, 2923 2 of 11 and individual behavior of players, there is also a relationship with the physical and physiological demands of players. Among the common topics researched in SSGs and its physiological effects is the effects of training regimens on players’ performance and acute responses [9]. The load imposed on players and the way in which this load occurs must be understood to ensure that training stimuli are adjusted to promote high-intensity exertion without compromising external load or technical/tactical performance [10]. The load is closely related to the following training prescription that should be taken in account [2]: (a) work intensity and duration; (b) recovery type (rest/active recovery) and duration; and (c) total work duration (work interval number × work duration). In the specific case of SSGs, comparisons between continuous and intermittent training regimens are commonly conducted [9,11]. Most results do not reveal meaningful or clear changes between these types of regimens in terms of heart rate responses and blood lactate concentrations [9,12]. However, when comparing different intermittent regimens (short, moderate, and long) with continuous regimens, the evidence suggests that continuous regimens result in higher values of maximal heart rate, blood lactate concentration, and perceived exertion [13]. Moreover, continuous regimens seem to increase the distance covered at low running speeds and decrease moderate-to-intense running distances [13]. As high-intensity drills, SSGs seem to fit in the category of interval training. Rather than compare the acute effects of continuous versus intermittent regimens, it is important to analyze the effects of different intermittent regimens. In a study conducted using a 3 × 3 SSG format, it was found that long bouts (sets) (3 × 6 min/2 min rest) decreased heart rate responses in comparison to medium (3 × 4 min/2 min rest) and small (3 × 2 min/2 min rest) bouts [14] if the first minute of data is excluded. However, no meaningful changes were observed in perceived exertion or technical actions [14]. In another study, it was found that, compared to long bouts, shorter bouts elicited lower maximal heart rate, shorter total distance covered at low running speed, and greater distances covered at medium and high running speeds [13]. The proper adjustment of SSGs to suit the purpose of training may help coaches optimize the amount of exertion imposed on players and may improve their performance. However, while a couple of experiments have compared different intermittent soccer training regimens [13,14], the information presented lacks a clear demonstration of the effects of different training regimens on internal (acute physiological responses) and external (physical demands) load variables. A comparison between bout durations should help researchers understand the patterns of exertion and provide meaningful information to coaches to help them choose the most effective regimens for their players. Moreover, coaches will also be able to analyze the effects of different intermittent regimens on time-motion performance during SSGs. Based on this rationale, the aim of the present study was two-fold: (i) to analyze variations in rate of perceived exertion, heart rate responses, and time-motion demands between two intermittent training regimens (6 × 3’ and 3 × 6’) during an SSG (5 × 5 format); and (ii) to test the variations of the above-mentioned variables within training regimens (between sets). 2. Materials and Methods 2.1. Participants Ten male amateur soccer players (age: 23.7 ± 1.1 years; experience: 10.3 ± 3.1 years; height: 178.2 ± 5.3 cm; weight: 72.1 ± 4.9 kg) competing at the regional level participated in this study. The participants usually trained three times a week and played one match every week. Participants were informed about the study design and the potential implications, risks, and benefits of participating. After that, participants freely signed an informed consent. The experiment followed the ethical standards of the Declaration of Helsinki for the study in humans. The study was approved by the local ethical committee (School of Sport and Leisure) with the code number IPVC-ESDL180503. Int. J. Environ. Res. Public Health 2019, 16, 2923 3 of 11 2.2. Experimental Approach This study used a counterbalanced repeated-measures design to compare the rate of perceived exertion (RPE), mean heart rate (HRmean), total distance (TD), running distance (RD), sprinting distance (SD), total accelerations (TAc), total decelerations (TDc) and player load (PL) of the participants in two different SSG regimens: 6 sets of 3 min with 2 min of rest (6 × 3’ regimen) and 3 sets of 6 min with 2 min of rest (3 × 6’ regimen). The study was conducted for 2 weeks immediately after the last official match of the season. The study occurred in the same pitch and the 10 players participated in all data collection sessions. The 5 × 5 format was employed in both training regimens. Each regimen was implemented twice, interspaced by a period of one week to ensure each session was performed under similar conditions. In each session, only one SSG regimen was implemented. In the first week, the 6 × 3’ regimen was implemented first and the 3 × 6’ regimen was implemented 48 h afterward, without any other training sessions in between. In the second week, the inverse sequence was employed. The mean of results obtained in each condition (mean of the two days of data collection with the same regimen) was used between regimens comparisons. The players were distributed into two teams based on skill level and playing position to homogenize the competitive level. The teams did not change during the study. Players wore vests equipped with a GPS and HR sensors during the SSGs. The SSGs were played on synthetic turf at 6:00 p.m. at an average temperature of 23 ◦C and a relative humidity of 57% no SSGs played under rainy weather conditions. All SSGs were preceded by a standardized warm-up consisting of 5 min of jogging, 5 min of lower-limb dynamic stretching and mobility exercises, 5 min of agility and speed drills, and 5 min of a ball possession game. 2.3. Small-Sided Game The 5 × 5 format was implemented with small goals (2 × 1 m) without goalkeepers. The size of the pitch was 42 × 22 m (924 m2). The individual playing area (area divided by the total number of players) followed previous recommendations to promote SSGs based on the real game situations (area of play) in attacking processes [15]. Two training regimens were used: 6 × 3’/2’ rest and 3 × 6’/2’ rest. The teams were composed of 2 defenders, 2 midfielders and 1 forward with similar skill levels as based on a preliminary observational test. No specific verbal instructions were provided before, during, or after the SSGs. However, verbal encouragement was provided to keep players committed and to help them to maintain a high exertion level. Six balls were placed around the pitch to ensure a quick repositioning if the ball in play went out of bounds. The SSGs followed official soccer rules with exception of offside. 2.4. Rating of Perceived Exertion (RPE) Players were instructed to rate their perceived effort immediately after each SSG. The CR-10 point scale [16] was used to classify the effort; on this scale, 1 means “very light activity” and 10 means “maximal exertion.” Players rated the effort individually as to not hear or be influenced by other teammates’ responses. All the players were previously instructed of the use of the scale to optimize the accuracy of their ratings during the experiments. 2.5. Heart Rate (HR) and Global Positioning System (GPS) Players wore an HR sensor and a chest belt (Polar H7, Polar Electro, OY, Kempele, Finland) which recorded data every second during the SSGs. Data were imported into the Polar Team application. The HRmean (bpm) per each set was used to measure this variable. The players also wore a vest with a geolocation tracker (JOHAN Sports, Noordwijk, The Netherlands) consisting of a GPS sensor (10 Hz, including EGNOS correction), accelerometer, gyroscope, and magnetometer (100 Hz, 3 axes). The validity and reliability values of the devices can be found in a Int. J. Environ. Res. Public Health 2019, 16, 2923 4 of 11 previous study [17]. The tracker was placed in a bag of the vest located at the dorsal region. The data was exported and treated immediately after each session. The following variables were collected from the tracker devices: (a) total distance (meters per minute); (b) running distance at 14–19.9 km/h (meters per minute); (c) sprinting distance at >20.0 km/h (meters per minute); (d) total accelerations >2 m/s2 (number per minute); (e) total decelerations >2 m/s2 (number per minute); and (f) player’s load (g per min), being calculated by estimating the total acceleration difference between two consecutive time steps being the length of the three-dimensional vector of accelerations in the anteroposterior, mediolateral, and craniocaudal axes between time step = 0 and time step = 1. Players were familiarized with the use of the HR belts and trackers vests before the study began. 2.6. Statistical Procedures The results are presented as either means and standard deviations (SD) or percentage differences and 90% confidence intervals (90% CI). The confidence intervals were defined following the recommendations for this kind of sample [18]. Normality and homogeneity of the data were firstly tested and verified before the inference analyses. Between-training regimens and within-training regimens differences were analyzed using the standardized differences of the effect size (ES) [19], with a 90% CI. ES was classified as trivial (<0.2), small (0.2–0.6), moderate (0.6–1.2), or large (>1.2) [18]. Probabilities were calculated considering the smallest worthwhile changes (SWC, 0.2 × between-subjects SD) [20]. Qualitative probabilistic mechanistic inferences of the true effects were made using these probabilities [20]. The scale for qualitative probabilities was as follows: 25–75% = possible; 75–95% = likely; 95–99% = very likely; >99% = almost certain [20]. 3. Results Descriptive statistics of internal and external load variables in both SSG training regimens can be found in Table 1 (values represent the average of the two sessions per type of intermittent protocol). Descriptive analyses reveal that RPE were higher in the last set in both 6 × 3’ (6.1 ± 1.9 arbitrary units (A.U.)) and 3 × 6’ (6.7 ± 1.6 A.U.) training regimens. The HRmean was higher in the second set of 6 × 3’ regimen (171.6 ± 10.0 bpm) and in the third set of 3 × 6’ regimen (171.1 ± 10.9). TD was greater in the third set of 6 × 3’ regimen (112.5 ± 11.1 m/min) and in the first set of 3 × 6’ regimen (103.3 ± 7.6 m/min). RD was greater in the second set of 6 × 3’ regimen (10.0 ± 5.1 m/min) and in the second set of 3 × 6’regimen (9.4 ± 5.6 m/min). SD was greater in the third set of 6 × 3’ regimen (1.5 ± 1.9 m/min) and in the second set of 3 × 6’ regimen (0.7 ± 1.3 m/min). TAc and TDc were greater in the second set of 6 × 3’ regimen (2.9 ± 0.8 and 2.7 ± 0.9 n/min, respectively) and in the first set of 3 × 6’ regimen (2.1 ± 1.0 and 1.9 ± 1.0 n/min, respectively). Finally, PL was greater in the second and third sets of 6 × 3’ regimen (7.3 ± 1.3 g/min) and in the first set of 3 × 6’ regimen (1.9 ± 1.0 g/min). Between-training regimen variations can be found in Table 2 (values represent the average of the two sessions per type of intermittent protocol). Almost certain large decreases of TD (−8.6%, [−12.3; −4.8], ES: −1.51, [−2.20; −0.82]) and RD (−34.0%, [47.0; −17.8], ES: −2.23, [−3.40; −1.05]) were observed when comparing 3 × 6’ versus 6 × 3’ regimens. Very likely moderate and large decreases of TAc (−24.0%, [−35.1; −10.9]; ES: −1.11, [−1.75; −0.47]) and TDc (−26.7%, [−38.8; −12.1]; ES: −1.49, [−2.36; −0.62]), respectively, were found when comparing 3 × 6’ versus 6 × 3’ regimens. Int. J. Environ. Res. Public Health 2019, 16, 2923 5 of 11 Table 1. Descriptive statistics (Mean (Standard Deviation)) of internal and external load during the SSGs. Variable 6 × 3’ Regimen 3 × 6’Regimen S1 S2 S3 S4 S5 S6 S1 S2 S3 RPE (A.U.) 4.2 (1.5) 4.9 (1.3) 5.6 (1.4) 6.0 (1.3) 5.9 (1.3) 6.1 (1.9) 5.0 (1.2) 6.3 (1.1) 6.7 (1.6) HRmean (bpm) 170.9 (11.1) 171.6 (10.0) 169.8 (10.9) 170.2 (12.5) 166.3 (10.8) 165.2 (12.4) 169.1 (11.7) 168.8 (10.5) 171.1 (10.9) TD (m/min) 107.7 (10.2) 111.7 (9.1) 112.5 (11.1) 106.8 (9.8) 104.1 (8.8) 101.6 (10.9) 103.3 (7.6) 99.8 (6.9) 90.9 (15.6) RD (m/min) 10.0 (5.1) 14.1 (5.7) 12.8(4.7) 10.6 (5.2) 11.1 (5.6) 9.1 (6.6) 8.7 (5.5) 9.4 (5.6) 6.0 (3.8) SD (m/min) 0.5 (1.1) 1.0 (1.6) 1.5 (1.9) 0.7 (1.2) 1.0 (1.6) 0.3 (0.7) 0.6 (0.9) 0.7 (1.3) 0.5 (0.8) TAc (n/min) 2.7 (0.9) 2.8 (0.6) 2.9 (0.8) 2.2 (1.0) 2.2 (1.0) 2.4 (1.2) 2.1 (1.0) 1.8 (0.9) 1.9 (1.1) TDc (n/min) 2.4 (1.1) 2.5 (0.6) 2.7 (0.9) 1.7 (0.9) 2.1 (1.0) 2.1 (1.2) 1.9 (1.0) 1.7 (0.9) 1.6 (0.9) PL (g/min) 7.2 (1.2) 7.3 (1.3) 7.3 (1.3) 6.8 (1.3) 6.4 (1.1) 6.5 (1.1) 6.9 (0.9) 6.3 (0.8) 5.9 (1.3) SSGs: small-sided games. RPE: rated of perceived exertion (CR-10 scale); HRmean: mean heart rate; TD: total distance; RD: running distance; SD: sprinting distance; TAc: total accelerations; TDc: total decelerations; PL: player’s load; S: set; A.U.: arbitrary units; bpm: beats per minute; m/min: meters per minute; n/min: number per minute: g/min: g per minute. Table 2. Comparison of internal and external load variables between training regimens in terms of percentage and standardized differences and the probabilities of each standardized difference. Variable M (SD) 3 × 6’ Reg. M (SD) 6 × 3’ Reg. % Difference (3 × 6’ Reg.–6 × 3’ Reg.) Standardized Difference (3 × 6’ Reg.–6 × 3’ Reg.) % Greater/Similar/Lower Values for 3 × 6’ Reg. vs. 6 × 3’ Reg. Value [90% CI] Value (Magnitude) 90% CI RPE (A.U.) 5.97 (0.80) 5.43 (0.91) 10.7 [2.9; 19.2] 0.49 small [0.14; 0.84] 92/8/0 Likely HRmean (bpm) 169.67 (10.05) 169.52 (9.75) 0.1 [−1.8; 2.0] 0.01 trivial [−0.28; 0.31] 14/75/11 Unclear TD (m/min) 98.39 (7.49) 107.56 (5.99) −8.6 [−12.3; −4.8] −1.51 large [−2.20; −0.82] 0/0/100 Almost certain RD (m/min) 8.04 (3.31) 11.28 (1.87) −34.0 [−47.0; −17.8] −2.23 large [−3.40; −1.05] 0/0/100 Almost certain SD (m/min) 0.62 (0.42) 0.83 (0.49) −13.9 [−46.7; 39.3] −0.13 trivial [−0.54; 0.29] 9/53/38 Unclear TAc (n/min) 1.95 (0.55) 2.53 (0.54) −24.0 [−35.1; −10.9] −1.11 moderate [−1.75; −0.47] 0/1/99 Very likely TDc (n/min) 1.70 (0.53) 2.24 (0.39) −26.7 [−38.8; −12.1] −1.49 large [−2.36; −0.62] 0/1/99 Very likely PL (g/min) 6.37 (0.79) 6.92 (0.99) −7.8 [−12.6; −2.7] −0.58 small [−0.97; −0.20] 0/5/95 Likely RPE: rated of perceived exertion (CR-10 scale); HRmean: mean heart rate; TD: total distance; RD: running distance; SD: sprinting distance; TAc: total accelerations (>2 m/s2); TDc: total decelerations (>2 m/s2); PL: g/min; S: set; Reg.: training regimen; A.U.: arbitrary units; bpm: beats per minute; m/min: meters per minute; n/min: number per minute: g/min: g per minute. Int. J. Environ. Res. Public Health 2019, 16, 2923 6 of 11 Within-training regimen differences can be found in Figures 1 and 2. To simplify the analysis, the six sets within 6 × 3’ SSG regimen were grouped in three sets consisting the first set in the average of sets 1 and 2, the second set in the average of sets 3 and 4 and the third set in the average of 5 and 6. Very likely increases of RPE were found in the 3rd set in comparison to the 1st during the 3 × 6’ SSG regimen (34.5%, [12.4; 61.0], ES: 1.35, [0.53; 2.16], large magnitude) and 6 × 3’ SSG regimen (29.9%, [11.6; 51.2]; ES: 1.17, [0.49; 1.85], moderate magnitude). Trivial-to-small changes of HR were found between sets in both training regimens. Likely decreases of total distance were found from set 3 to set 1 (−11.6%, [−20.8; −1.3], ES: −1.78, [−3.38; −0.19], large magnitude) and from set 2 to set 3 (−9.2%, [−18.6; 1.3], ES: −1.33, [−2.85; 0.18], large magnitude) in 3 × 6’ SSG regimen. Very likely decreases of total distance were found from set 3 to set 1 (−6.2%, [−8.8; −3.5], ES: −0.96, [−1.39; −0.53], moderate magnitude) and almost certain decreases were found from set 3 to set 2 (−6.0%, [−8.0; −3.9], ES:−0.87, [−1.18; −0.56], moderate magnitude) in 6 × 3’ SSG regimen. Very likely decreases of player load were found from set 3 to set 1 (−13.1%, [−23.1; −2.9], ES: −1.37, [−2.46; −0.27], large magnitude] during 3 × 6’ SSG regimen. Int. J. Environ. Res. Public Health 2019, 16, x 7 of 11 (a) (b) (c) (d) Figure 1. Standardized difference (Cohen) between sets in (a) RPE; (b) HRmean; (c) Total Distance; and (d) Player Load. The six sets of the 6 × 3’ regimen were grouped in three sets (S1: mean of set 1 and 2; S2: mean of set 3 and 4; S3: mean of set 5 and 6) to a better visualization and analysis. Standardized value direction depends on the relationship A-B. Figure 1. Standardized difference (Cohen) between sets in (a) RPE; (b) HRmean; (c) Total Distance; and (d) Player Load. The six sets of the 6 × 3’ regimen were grouped in three sets (S1: mean of set 1 and 2; S2: mean of set 3 and 4; S3: mean of set 5 and 6) to a better visualization and analysis. Standardized value direction depends on the relationship A-B. Int. J. Environ. Res. Public Health 2019, 16, 2923 7 of 11 Figure 1. Standardized difference (Cohen) between sets in (a) RPE; (b) HRmean; (c) Total Distance; and (d) Player Load. The six sets of the 6 × 3’ regimen were grouped in three sets (S1: mean of set 1 and 2; S2: mean of set 3 and 4; S3: mean of set 5 and 6) to a better visualization and analysis. Standardized value direction depends on the relationship A-B. (a) (b) Int. J. Environ. Res. Public Health 2019, 16, x 8 of 11 (c) (d) Figure 2. Standardized difference (Cohen) between sets in (a) running distance; (b) sprinting distance; (c) total accelerations; and (d) total decelerations. The six sets of the 6 × 3’ regimen were grouped in three sets (S1: mean of set 1 and 2; S2: mean of set 3 and 4; S3: mean of set 5 and 6) to a better visualization and analysis. Standardized value direction depends on the relationship A-B. Within-training regimens differences in running distance were trivial-to-small in the majority of comparisons and the moderate magnitudes were unclear. Considering the sprinting distances, the variations were also trivial-to-small. Unclear large (−30.6%, [−58.3; 15.5], ES: −1.96, [−4.69; 0.77], large magnitude) and unclear moderate (−24.9%, [−52.3; 18.7], ES: −1.18, [−3.08; 0.71], moderate magnitude) decreases of total accelerations were found from set 3 to set 1 and set 3 to set 2 during 6 × 3’ SSG regimen, respectively. Trivial-to-small changes of total decelerations and unclear moderate differences were found in both training regimens. 4. Discussion Between-SSG training regimens, changes were observed in the present study. Almost certain large increases in total and running distances were observed in the shorter sets (6 × 3’), and very likely moderate and large increases in total accelerations and decelerations, respectively, during shorter sets were found. RPE showed likely small increases during longer sets (6 × 3’). Likely small increases in player load during shorter sets were also found. Briefly, this evidence suggests that shorter sets contribute to an increase in terms of moderate-running speed distances and high-intensity actions associated with accelerations and decelerations higher than 2 m/s2 while resulting in a lower RPE than in longer sets. Although no meaningful differences were found in a study that compared time-motion variables between 4 × 4’ and 2 × 8’ 5 × 5 SSG regimens [11], our results are partially in line with the findings of a study that compared short (6 × 2’), medium (3 × 4’), and long (2 × 6’) regimens [13]. Heart rate responses were trivially different between regimens, suggesting that this variable is not sensitive to variations in the intermittent regimens tested in our study. These results are in line with previous studies that compared heart rate responses after different intermittent regimens [11 14] Ho e e the alte ati e i te al load a ke u ed i ou tudy (RPE) e ealed a likely all Figure 2. Standardized difference (Cohen) between sets in (a) running distance; (b) sprinting distance; (c) total accelerations; and (d) total decelerations. The six sets of the 6 × 3’ regimen were grouped in three sets (S1: mean of set 1 and 2; S2: mean of set 3 and 4; S3: mean of set 5 and 6) to a better visualization and analysis. Standardized value direction depends on the relationship A-B. Within-training regimens differences in running distance were trivial-to-small in the majority of comparisons and the moderate magnitudes were unclear. Considering the sprinting distances, the variations were also trivial-to-small. Unclear large (−30.6%, [−58.3; 15.5], ES: −1.96, [−4.69; 0.77], large magnitude) and unclear moderate (−24.9%, [−52.3; 18.7], ES: −1.18, [−3.08; 0.71], moderate magnitude) decreases of total accelerations were found from set 3 to set 1 and set 3 to set 2 during 6 × 3’ SSG regimen, respectively. Trivial-to-small changes of total decelerations and unclear moderate differences were found in both training regimens. 4. Discussion Between-SSG training regimens, changes were observed in the present study. Almost certain large increases in total and running distances were observed in the shorter sets (6 × 3’), and very likely moderate and large increases in total accelerations and decelerations, respectively, during shorter sets were found. RPE showed likely small increases during longer sets (6 × 3’). Likely small increases in player load during shorter sets were also found. Briefly, this evidence suggests that shorter sets contribute to an increase in terms of moderate-running speed distances and high-intensity actions associated with accelerations and decelerations higher than 2 m/s2 while resulting in a lower RPE than in longer sets. Although no meaningful differences were found in a study that compared time-motion Int. J. Environ. Res. Public Health 2019, 16, 2923 8 of 11 variables between 4 × 4’ and 2 × 8’ 5 × 5 SSG regimens [11], our results are partially in line with the findings of a study that compared short (6 × 2’), medium (3 × 4’), and long (2 × 6’) regimens [13]. Heart rate responses were trivially different between regimens, suggesting that this variable is not sensitive to variations in the intermittent regimens tested in our study. These results are in line with previous studies that compared heart rate responses after different intermittent regimens [11,14]. However, the alternative internal load marker used in our study (RPE) revealed a likely small increase during longer sets, suggesting that raising RPE may compromise the consistency of physical demands across sets. Comparisons within SSG regimens were also performed to analyze the variations between sets. Main variations were found in RPE, total and running distances, and player load. Intriguingly, higher-intensity physical demands (sprinting distance and total accelerations and decelerations) were relatively constant across sets. Moderate and progressive increases in RPE were found across sets, mainly during the 3 × 6’ regimen. The progressive increase in RPE scores was also found during shorter bouts. Trivial-to-small differences were found between the third and second sets. Eventually, longer sets may contribute to a perceptual increase of effort [21]. However, the fatigue effect may be the cause of such RPE increases, especially considering that decreases in total and running distances and player load occurred across the sets in both regimens. Despite these progressive decreases, it was observed that shorter sets are probably more beneficial to delaying the impact of fatigue on total distance because large decreases were found between sets 1 and 3 and between sets 2 and 3 during the 3 × 6’ regimens, and only moderate effects were found during the 6 × 3’ regimens. Similar evidence was found regarding running distance. Moderate decreases were found across the sets during longer sets, and only small differences were observed during shorter sets. In both regimens and for both variables (total and running distance) trivial-to-small decreases were found from set 1 to set 2, suggesting that the main effects of fatigue emerge from the continuity of exertion [22]. The greater constancy of results in terms of sprinting distance and total of accelerations and decelerations across the sets may suggest that the resting period of 2 min was enough to maintain intensity levels and to maximize energy phosphates as the primary energy source [23]. However, in the specific case of sprinting distance, the values per minute did not reach 1 m on average; thus, the constancy can also be justified by the low frequency of these demands across the 5 × 5 format. Interestingly, knowledge about the period of time may also constrain the pace of players during the games. In fact, a previous study that compared different intermittent regimens and their effects on pacing revealed that high-speed distances progressively and largely decreased across shorter sets (1 min) based on the ‘all-out’ pacing strategy used by rugby players [22]. Conversely, during longer sets, a more constant high-speed pace across the sets was observed [22]. The results of the present are in line with these previous findings. However, this constancy in the most intense activities that occurs in longer sets also resulted in smaller values when compared with shorter sets. Our study had some limitations. The number and the competitive level of the players may constrain the inferences of this study. Theoretically, professional players present greater aerobic and anaerobic capacities, thus making it possible to reduce the magnitude of decreases throughout the bouts. However, the main results are in line with those of previous studies conducted in elite youth players [13], amateurs [11], and professionals [14]. In addition, our study analyzed the effects in only one SSG format, and for that reason, the occurrences of sprinting distances are scarce. Bigger formats should be considered for testing variations between regimens. Different work-to-rest ratios should be considered in future study designs in order to analyze the effects of rest on physical demands and the internal effect of the exercise. In fact, in our study the difference between training regimen also constrained the time of recovery and the work-to-rest ratios were different between regimens, thus this may interfere with some generalizations that can be made from our study. The small number of participants should also be considered as a limitation for a possible generalization of the evidence. Int. J. Environ. Res. Public Health 2019, 16, 2923 9 of 11 Other conditions should also be understood as contextual and for that reason may affect comparisons with future studies, namely: (i) the synthetic grass was not wet; (ii) the same study conducted in another period of the season may conduct to different evidence; (iii) possibly more sessions would lead to a better copy with the decrements throughout the bouts; and (iv) different task conditions that affect the player’s behavior and, consequently, the intensity of exercise [24]. Despite these limitations, this study revealed that different training regimens led to different effects on internal and external load variables. The main highlights of the present study are that shorter sets (3’) seem to be more beneficial than longer sets (6’) in keeping total and running distances and total accelerations and decelerations constant while decreasing the RPE. Additionally, the greater decrements seem to mostly occur in the last sets. Despite the somewhat predictable results, the small amount of previous research into such specific issues highlights the innovative character of the present study to reveal that shorter sets contribute to a greater external load stimulus despite that no meaningful changes were found between training regimens in the internal load. This may suggest that the internal load measures cannot be the unique criteria to choose between bout periods. In fact, in the present study it is possible to observe that despite the similarity of internal load between intermittent regimens, meaningful increases of some important external load measures were found (e.g., total distance, running distance, accelerations and decelerations). As practical implications, we may hypothesize that smaller periods of exertion with a greater number of sets (bouts) can be recommended to ensure a more constant high level of physical demand and to contribute to an optimization of the high-energy systems that support highly demanding actions. Possibly, longer bouts should be used in situations in which coaches want to develop aerobic capacity while players experience acute fatigue effects and consequent decrements in the external load. Moreover, shorter sets enhance a meaningful greater stimulus in terms of distance covered, running distance and total number of accelerations and decelerations. However, for a better adjustment of the training regimen with the reality of the match, it will possibly be interesting in the future to compare the regular periods of high-intensity effort in the match and adjust such periods in SSGs, thus making more real the period of high-exertion and the work-to-rest ratios. 5. Conclusions Between-SSG training regimens revealed that shorter sets (6 × 3’) almost certainly largely increased total and running distances and very likely moderately and largely increased total accelerations and decelerations, respectively, in comparison to longer sets (3 × 6’) while likely small increases in RPE were found in longer sets. The within-regimen analyses revealed that longer sets contributed to increases in RPE across sets and to large and progressive decreases in total distance and player load across sets. Additionally, moderate decreases in running distance and total decelerations were found progressively across shorter sets, while these variables were more stable between longer sets. The overall conclusions should be faced carefully, considering that the recovery periods were also longer after the longer periods of exertion. The results may suggest that shorter sets can be beneficial to maintain external load demands without resulting in large increases in the perceived effort or heart rate responses, while ensuring a greater stimulus in total distance, running distance and number of accelerations and decelerations. However, in terms of acceleration and deceleration profiles, it may also be appropriate to choose longer sets as these contribute to a smaller decrease across sets (decrements are not so meaningful between sets as in the shorter intervals). Author Contributions: F.M.C. conceived the study. F.M.C., P.T.N., T.R. and B.K. designed the study. F.M.C. collected data. F.M.C. analyzed and interpreted the data and drafted the manuscript. F.M.C., P.T.N., T.R. and B.K. revised the manuscript and approved the final version. Funding: This research received no external funding. Acknowledgments: The authors would like to thank to Diogo Peixoto, Mónica Gomes, Leandro Silva and Miguel Moreira for the help in data collection and to JOHAN Sports for providing the GPS units. Int. J. Environ. Res. Public Health 2019, 16, 2923 10 of 11 Conflicts of Interest: The authors declare no conflict of interest. References 1. Clemente, F.M.; Martins, F.M.; Mendes, R.S. Developing Aerobic and Anaerobic Fitness Using Small-Sided Soccer Games: Methodological Proposals. Strength Cond. J. 2014, 36, 76–87. 2. Halouani, J.; Chtourou, H.; Gabbett, T.; Chaouachi, A.; Chamari, K. Small-Sided Games in Team Sports Training. J. Strength Cond. Res. 2014, 28, 3594–3618. [CrossRef] 3. Davids, K.; Araújo, D.; Correia, V.; Vilar, L. How small-sided and conditioned games enhance acquisition of movement and decision-making skills. Exerc. Sport Sci. Rev. 2013, 41, 154–161. 4. Casamichana, D.; Bradley, P.S.; Castellano, J. Influence of the Varied Pitch Shape on Soccer Players Physiological Responses and Time-Motion Characteristics During Small-Sided Games. J. Hum. Kinet. 2018, 64, 171–180. [CrossRef] 5. Hammami, A.; Gabbett, T.J.; Slimani, M.; Bouhlel, E. Does small-sided games training improve physical-fitness and specific skills for team sports? A systematic review with meta-analysis. J. Sports Med. Phys. Fitness 2018, 58, 1446–1455. [CrossRef] 6. Clemente, F.M. Small-Sided and Conditioned Games in Soccer Training: The Science and Practical Applications; SpringerBriefs in Applied Sciences and Technology; Springer: Singapore, 2016; ISBN 978-981-10-0880-1. 7. Aguilar, M.P.; Navarro-Adelantado, V.; Jonsson, G.K. Detection of Ludic Patterns in Two Triadic Motor Games and Differences in Decision Complexity. Front. Psychol. 2018, 8. [CrossRef] 8. Méndez-Domínguez, C.; Gómez-Ruano, M.A.; Rúiz-Pérez, L.M.; Travassos, B. Goals scored and received in 5vs4 GK game strategy are constrained by critical moment and situational variables in elite futsal. J. Sports Sci. 2019, 37. [CrossRef] 9. Köklü, Y. A Comparison Of Physiological Responses To Various Intermittent And Continuous Small-Sided Games In Young Soccer Players. J. Hum. Kinet. 2012, 31, 89–96. [CrossRef] 10. Radziminski, L.; Rompa, P.; Barnat, W.; Dargiewicz, R.; Jastrzebski, Z. A Comparison of the Physiological and Technical Effects of High-Intensity Running and Small-Sided Games in Young Soccer Players. Int. J. Sports Sci. Coach. 2013, 8, 455–465. [CrossRef] 11. Casamichana, D.; Castellano, J.; Dellal, A. Influence of Different Training Regimes on Physical and Physiological Demands During Small-Sided Soccer Games. J. Strength Cond. Res. 2013, 27, 690–697. [CrossRef] 12. Hill-Haas, S.V.; Coutts, A.J.; Rowsell, G.J.; Dawson, B.T. Generic versus small-sided game training in soccer. Int. J. Sports Med. 2009, 30, 636–642. [CrossRef] 13. Köklü, Y.; Alemdaro˘glu, U.; Cihan, H.; Wong, D.P. Effects of Bout Duration on Players’ Internal and External Loads During Small-Sided Games in Young Soccer Players. Int. J. Sports Physiol. Perform. 2017, 12, 1370–1377. [CrossRef] 14. Fanchini, M.; Azzalin, A.; Castagna, C.; Schena, F.; McCall, A.; Impellizzeri, F.M. Effect of bout duration on exercise intensity and technical performance of small-sided games in soccer. J. Strength Cond. Res. 2011, 25, 453–458. [CrossRef] 15. Fradua, L.; Zubillaga, A.; Caro, O.; Iván Fernández-García, A.; Ruiz-Ruiz, C.; Tenga, A. Designing small-sided games for training tactical aspects in soccer: Extrapolating pitch sizes from full-size professional matches. J. Sports Sci. 2013, 31, 573–581. [CrossRef] 16. Borg, G. Perceived Exertion and Pain Scales; Human Kinetics: Champaign, IL, USA, 1998. 17. Nikolaidis, P.T.; Clemente, F.M.; van der Linden, C.M.I.; Rosemann, T.; Knechtle, B. Validity and Reliability of 10-Hz Global Positioning System to Assess In-line Movement and Change of Direction. Front. Physiol. 2018, 9. [CrossRef] 18. Batterham, A.M.; Hopkins, W.G. Making Meaningful Inferences about Magnitudes. Int. J. Sports Physiol. Perform. 2006, 1, 50–57. [CrossRef] 19. Cohen, J. Statistical Power Analysis for the Behavioral Sciences; Lawrence Erlbaum Associates: Hillsdale, NJ, USA, 1988. 20. Hopkins, W.G.; Marshall, S.W.; Batterham, A.M.; Hanin, J. Progressive Statistics for Studies in Sports Medicine and Exercise Science. Med. Sci. Sport. Exerc. 2009, 41, 3–13. [CrossRef] Int. J. Environ. Res. Public Health 2019, 16, 2923 11 of 11 21. Smits, B.L.M.; Pepping, G.-J.; Hettinga, F.J. Pacing and Decision Making in Sport and Exercise: The Roles of Perception and Action in the Regulation of Exercise Intensity. Sport. Med. 2014, 44, 763–775. [CrossRef] 22. Sampson, J.A.; Fullagar, H.H.K.; Gabbett, T. Knowledge of bout duration influences pacing strategies during small-sided games. J. Sports Sci. 2015, 33, 85–98. [CrossRef] 23. Billaut, F.; Bishop, D.J.; Schaerz, S.; Noakes, T.D. Influence of Knowledge of Sprint Number on Pacing during Repeated-Sprint Exercise. Med. Sci. Sport. Exerc. 2011, 43, 665–672. [CrossRef] 24. Castellano, J.; Silva, P.; Usabiaga, O.; Barreira, D. The influence of scoring targets and outer-floaters on attacking and defending team dispersion, shape and creation of space during small-sided soccer games. J. Hum. Kinet. 2016, 51, 153–163. [CrossRef] © 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Variations of Internal and External Load Variables between Intermittent Small-Sided Soccer Game Training Regimens.

08-15-2019

Clemente, Filipe Manuel,Nikolaidis, Pantelis Theodoros,Rosemann, Thomas,Knechtle, Beat

eng

PMC10681188

RESEARCH ARTICLE Rat and mouse cardiomyocytes show subtle differences in creatine kinase expression and compartmentalization Jelena Branovets, Ka¨rol Soodla, Marko Vendelin, Rikke Birkedal* Laboratory of Systems Biology, Department of Cybernetics, Tallinn University of Technology, Tallinn, Estonia * rikke@sysbio.ioc.ee Abstract Creatine kinase (CK) and adenylate kinase (AK) are energy transfer systems. Different stud- ies on permeabilized cardiomyocytes suggest that ADP-channelling from mitochondrial CK alone stimulates respiration to its maximum, VO2_max, in rat but not mouse cardiomyocytes. Results are ambiguous on ADP-channelling from AK to mitochondria. This study was under- taken to directly compare the CK and AK systems in rat and mouse hearts. In homogenates, we assessed CK- and AK-activities, and the CK isoform distribution. In permeabilized cardi- omyocytes, we assessed mitochondrial respiration stimulated by ADP from CK and AK, VO2_CK and VO2_AK, respectively. The ADP-channelling from CK or AK to mitochondria was assessed by adding PEP and PK to competitively inhibit the respiration rate. We found that rat compared to mouse hearts had a lower aerobic capacity, higher VO2_CK/VO2_max, and dif- ferent CK-isoform distribution. Although rat hearts had a larger fraction of mitochondrial CK, less ADP was channeled from CK to the mitochondria. This suggests different intracellular compartmentalization in rat and mouse cardiomyocytes. VO2_AK/VO2_max was similar in mouse and rat cardiomyocytes, and AK did not channel ADP to the mitochondria. In the absence of intracellular compartmentalization, the AK- and CK-activities in homogenate should have been similar to the ADP-phosphorylation rates estimated from VO2_AK and VO2_CK in permeabilized cardiomyocytes. Instead, we found that the ADP-phosphorylation rates estimated from permeabilized cardiomyocytes were 2 and 9 times lower than the activ- ities recorded in homogenate for CK and AK, respectively. Our results highlight the impor- tance of energetic compartmentalization in cardiac metabolic regulation and signalling. Introduction Creatine kinase (CK) is thought to play a crucial role in storage and spatial transport of energy-rich phosphates in tissues with high and fluctuating energy demands [1, 2]. In the heart, there are three cytosolic CK isoforms (MM-, MB-, and BB-CK) and one mitochondrial CK isoform (Mi-CK). Several studies on rat heart have reported that phosphotransfer via CK is several times faster than ATP synthesis by oxidative phosphorylation in mitochondria [3–5], and Mi-CK in the intermembrane space is coupled to the adenine nucleotide translocase (ANT), which exchanges ATP for ADP, providing ADP for mitochondrial oxidative PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 1 / 25 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Branovets J, Soodla K, Vendelin M, Birkedal R (2023) Rat and mouse cardiomyocytes show subtle differences in creatine kinase expression and compartmentalization. PLoS ONE 18(11): e0294718. https://doi.org/10.1371/journal. pone.0294718 Editor: Luis Eduardo M Quintas, Universidade Federal do Rio de Janeiro, BRAZIL Received: June 6, 2023 Accepted: November 6, 2023 Published: November 27, 2023 Peer Review History: PLOS recognizes the benefits of transparency in the peer review process; therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. The editorial history of this article is available here: https://doi.org/10.1371/journal.pone.0294718 Copyright: © 2023 Branovets et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting information files. phosphorylation [6, 7]. For many years, the CK system was considered to have a pivotal role in the regulation of cardiac energy metabolism [1, 2, 8–10]. Surprisingly, studies on genetically manipulated mouse models having modifications in various components of the CK system like expression of different CK isoforms, creatine syn- thesis or uptake have been equivocal [11]. Notably, the baseline cardiac function of mice with compromised CK system is minimally disturbed [12–16] and compromised CK function does not exacerbate heart failure [17, 18]. Thus, studies on mice have not corroborated the theory, based mainly on rat experiments, that CK is crucial for energy transfer and regulation of oxida- tive phosphorylation. Our recent study on permeabilized cardiomyocytes from creatine-deficient AGAT [19] and GAMT [20] mice also pointed to a difference between mouse and rat cardiomyocytes. We per- formed an assay, where we recorded the respiration rate of permeabilized cardiomyocytes. In addition to substrates for oxidative phosphorylation (glutamate and malate, GM) and creatine, we added ATP to the solution to initiate endogenous ADP-generation by CK (creatine + ATP ! phosphocreatine + ADP + H+). Endogenous ADP then stimulated oxidative phosphoryla- tion. We found that the rate of ADP-generation by CK sustained a respiration rate that was ~80% of the maximal respiration rate [21]. The channelling of ADP from CK to the mitochon- dria was assessed by addition of phosphoenolpyruvate (PEP) and pyruvate kinase (PK) in excess, which compete with the mitochondria for ADP. This is an experimental strategy to assess how much of the CK is so closely associated with the mitochondria that ADP is chan- nelled directly to the mitochondria without being released to the bulk phase, where it would be consumed by PK. The addition of PEP and PK lowered the rate of oxidative phosphorylation by ~75%, demonstrating that some ADP generated by CK is channelled to the mitochondria and inaccessible to PK [21]. However, our results on mouse cardiomyocytes were in sharp contrast to experiments on rat cardiomyocytes, where even in the presence of PEP and PK, the rate of ADP-generation by CK sustained the respiration rate at its maximum [22, 23]. This dif- ference raised the question whether rat and mouse hearts differ in terms of their CK activity, either total and/or relative to the maximal respiration rate in the absence and presence of PEP and PK. Adenylate kinase (AK) is another well-known alternative phosphotransfer system in the heart [9]. Its role in facilitating energy transfer and in the compartmentalization of adenine nucleotides has also been investigated [24–29]. Although the contribution of AK-mediated phosphoryl transfer to the total ATP turnover is only ~10%, some studies observed an increased importance of AK under stress conditions [29, 30]. In our recent study on mouse cardiomyocytes [21], we also assessed the respiration rate stimulated by endogenous ADP gen- erated by AK. In the presence of substrates (GM) and ATP, the addition of AMP initiated the endogenous ADP-generation by AK (AMP + ATP ! 2 ADP). The rate of ADP-generation by AK sustained the respiration rate near its maximal rate, but the addition of PEP and PK abol- ished the effect of adding AMP, suggesting that all ADP generated by AK was accessible to PK. On the one hand, this is in agreement with the finding of a minimal AK activity in rat and mouse heart mitochondria [31–33]. On the other hand, it contradicts results demonstrating a stronger functional coupling between AK and mitochondrial respiration in rats [27]. Thus, there is an inexplicable mismatch between different studies regarding the importance of mito- chondrial AK in regulation of mitochondrial oxidative phosphorylation, and we speculated whether species differences might be adding to the confusion. The aim of the present study was to assess whether rat and mouse hearts are different in terms of 1) the overall CK and AK activities in whole heart homogenates, and 2) how much endogenous ADP generated by CK or AK stimulates oxidative phosphorylation in the absence and presence of PEP and PK competing with the mitochondria for the consumption of ADP. PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 2 / 25 Funding: This work was supported by the Estonian Research Council (www.etag.ee/en/), grant number PRG1127. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The authors have declared that no competing interests exist. Abbreviations: AGAT, Arginine:glycine amidinotransferase (EC 2.1.4.1); AK, Adenylate kinase; AKADP_GMPS, Name of the protocol used to record adenylate kinase-stimulated respiration with glutamate, malate, pyruvate, and succinate as substrates, followed by stimulation of maximal respiration with 2mM ADP; AKf, Adenylate kinase activity measured in the forward direction: ATP + AMP ! ADP + ADP; AKPEP/PK_GMPS, Name of the protocol used to record adenylate kinase- stimulated respiration with glutamate, malate, pyruvate, and succinate as substrates, followed by inhibition of respiration with phosphoenolpyruvate and pyruvate kinase; AKr, Adenylate kinase activity measured in the reverse direction: ADP + ADP ! ATP + AMP; AMPK, AMP-activated protein kinase; ANT, adenine nucleotide translocase; BB-CK, Homodimeric brain isoform of creatine kinase; CK, Creatine kinase; CKADP_GM, Name of the protocol used to record creatine kinase-stimulated respiration with glutamate and malate as substrates, followed by stimulation of maximal respiration with 2mM ADP; CKADP_GMPS, Name of the protocol used to record creatine kinase- stimulated respiration with glutamate, malate, pyruvate, and succinate as substrates, followed by stimulation of maximal respiration with 2mM ADP; CKf, Creatine kinase activity measured in the forward direction:ATP + creatine ! ADP + phosphocreatine + H+; CKPEP/PK_GM, Name of the protocol used to record creatine kinase-stimulated respiration with glutamate and malate as substrates, followed by inhibition of respiration with phosphoenolpyruvate and pyruvate kinase; CKPEP/PK_GMPS, Name of the protocol used to record creatine kinase-stimulated respiration with glutamate, malate, pyruvate, and succinate as substrates, followed by inhibition of respiration with phosphoenolpyruvate and pyruvate kinase; CKr, Creatine kinase activity measured in the reverse direction:ADP + phosphocreatine + H+ ! ATP + creatine; CM, Cardiomyocytes; CO, Cytochrome oxidase; CS, Citrate synthase; Cyt aa3, Cytochrome aa3; Cyt c, Cytochrome c; DTNB, dithiobis(2-nitrobenzoic acid); FCCP, Carbonyl cyanide 4-trifluoromethoxyphenylhydrazone; GAMT, Guanidinoacetate N-methyltransferase (EC 2.1. 1.2); GM, Glutamate and malate; GMPS, Glutamate, malate, pyruvate and succinate; MB-CK, In whole heart homogenates, we recorded the activities of CK and AK in the direction of both ADP- and ATP-generation. In addition, we recorded the activities of citrate synthase (CS, a marker of mitochondrial density) and cytochrome oxidase (CO, a marker of oxidative capacity, i.e. maximal O2 consumption rate per muscle mass) [34]. In permeabilized cardio- myocytes, we recorded the respiration rate stimulated by endogenous ADP generated by either CK or AK. In one set of experiments, we assessed how much CK or AK stimulated respiration relative to its maximal rate by subsequent addition of ADP. In parallel, the ADP channelling between kinases and mitochondria was assessed by subsequent additions of PEP and PK. The maximal respiration rate is substrate-dependent [35, 36]. Most studies have used only GM, which lead to NADH-linked electron flux only through complex I of the respiratory chain. More substrates are required to obtain the maximal respiration rate. For example, a combina- tion of glutamate, malate, pyruvate and succinate (GMPS) leads to NADH and FAD-linked elec- tron flux through both complexes I and II of the respiratory chain [36, 37]. Here, the role of CK was assessed with GM as in previous experiments [21–23], and with GMPS. Due to a limited number of chambers in the respirometer, the role of AK was only assessed with GMPS. To the best of our knowledge, this is the first time these recordings have been performed with GMPS. Throughout this work, we compare whole heart homogenate with isolated cardiomyocytes. However, whole heart homogenates are prepared using pieces of tissue, which in addition to cardiomyocytes also contain extracellular matrix, endothelial cells, and fibroblasts. Cardio- myocytes, endothelial cells and fibroblasts take up 70–80%, 3.2–5.3%, and 1.4–1.9% of the vol- ume, respectively [38], and cardiomyocytes have a much larger mitochondrial volume (31– 40%) than the other cell types (5% for endothelial cells) [39–41]. Therefore, in order to com- pare whole heart homogenate and isolated cardiomyocytes, we assumed that the CS activity in non-cardiomyocytes was negligible and compared the data normalized to the CS activity. Results Animals and cell preparations The morphological characteristics of the animals used in this study are given in Table 1. Table 2 shows several characteristics of the cell suspensions. The viability of the cell Table 1. Characteristics of the mice and rats used in the experiments. n BW, g HW, mg HW/BW, mg/g Mice 15 (7) 27.3 ± 0.8 123.8 ± 2.3 4.7 ± 0.1 Rats 14 (7) 316.6 ± 8 920.4 ± 26 2.9 ± 0.1 Values are shown as mean ± SEM. BW, body weight; HW, absolute heart weight; HW/BW, relative heart weight. The total number of animals is shown in column n. The absolute and relative heart weight (HW and HW/BW) are reported for a smaller number of animals, indicated in parenthesis, because the hearts used for cardiomyocyte isolation could not be weighed. https://doi.org/10.1371/journal.pone.0294718.t001 Table 2. Characteristics of the isolated cardiomyocyte suspensions from mice and rats. Viability % Protein mg/ml CS activity μmol/min/g protein Cyt aa3 μmol/g protein Mouse cardiomyocytes 75.0 ± 1.1 15.52 ± 0.80 806 ± 37 Rat cardiomyocytes 73.7 ± 1.9 20.80 ± 2.41 609 ± 9 *** 0.16 ± 0.01 Cardiomyocytes were isolated from 8 mice and 7 rats. Values are shown as mean ± SEM. Viability is the fraction of rod-shaped to the total number of cells. CS is the citrate synthase activity, and cyt aa3 is the cytochrome aa3 content. *** denotes p < 0.001, significant difference between species. https://doi.org/10.1371/journal.pone.0294718.t002 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 3 / 25 Heterodimeric muscle-brain isoform of creatine kinase; Mi-CK, Sarcomeric mitochondrial creatine kinase; MM-CK, Homodimeric muscle isoform of creatine kinase; P/O2 ratio, The phosphate to oxygen ratio, describes how much ADP the mitochondria phosphorylate to ATP per O2 consumed; PEP, Phosphoenolpyruvate; PK, Pyruvate kinase; SERCA, sarcoendoplasmic reticulum Ca2+-ATPase; TNB, thionitrobenzoate; VADP_AK, Estimated rate of ADP-phosphorylation, when mitochondria are stimulated by ADP generated by adenylate kinase; VADP_CK, Estimated rate of ADP-phosphorylation, when mitochondria are stimulated by ADP generated by creatine kinase; VADP_max, Estimated rate of ADP- phosphorylation, when mitochondria are stimulated by 2 mM exogenous ADP; VDAC, Voltage dependent anion channel; VFCCP, Respiration rate during maximal uncoupled respiration rate; VO2_AK, Respiration rate stimulated by adenylate kinase after addition of ATP and AMP; VO2_CK, Respiration rate stimulated by creatine kinase after addition of ATP with creatine in the medium; VO2_max, Maximal coupled respiration rate stimulated by 2 mM ADP; VO2_max_GM, Maximal coupled respiration rate with glutamate and malate as substrates; VO2_max_GMPS, Maximal coupled respiration rate with glutamate, malate, pyruvate and succinate as substrates. suspensions did not differ between mice and rats. Although there was a tendency to a higher protein content in rats than in mice, this was not statistically significant. CS activity, when nor- malized to the protein content, was significantly higher in mouse than in rat cardiomyocytes. The cytochrome aa3 (Cyt aa3) content was assessed only for cell suspensions from rat hearts due to the large volume required for these measurements. Enzyme activities and CK isoform distribution The overall activities of CS, CK, AK and CO, measured in cardiac whole tissue homoge- nates, are shown in Fig 1. In Fig 1A, the enzymatic activities normalized to the heart wet weight are shown for comparison with the literature (see Discussion). In Fig 1B, the activi- ties of CO, AK and CK normalized to the CS activity are shown for comparison with the res- piration rates recorded in permeabilized cardiomyocytes (Figs 3 and 4). Since both CK- and AK-catalysed phosphotransfer reactions are reversible, the activities of these enzymes were Fig 1. The activities of citrate synthase (CS), adenylate kinase (AK), creatine kinase (CK) and cytochrome oxidase (CO) in mouse and rat hearts. CS, AK and CK activities are represented as rates (μmol
min-1
g ww-1) and CO activity as a rate constant (min-1
g ww-1). AKf and CKf, and AKr and CKr, are the forward (f) and reverse (r) reaction rates of AK and CK, respectively. A: When normalized to the wet weight of the tissue, CS and AK activities were higher in mouse than in rat heart. B: When normalized to the CS activity, the CK and CO activities (μmol
min-1
IU CS-1) were higher in rat than in mouse heart. The number of animals was n = 7 for mice and n = 7 for rats. * denotes p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001 significant difference between species. https://doi.org/10.1371/journal.pone.0294718.g001 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 4 / 25 measured in both directions. In the forward direction, CK and AK generate ADP, i.e. CK catalyses the reaction: creatine + ATP ! phosphocreatine + H+ + ADP, and AK catalyses the reaction: ATP + AMP ! 2 ADP. In the reverse direction, CK catalyses the reaction: phosphocreatine + H+ + ADP ! creatine + ATP, and AK catalyses the reaction: 2 ADP ! ATP + AMP. When enzyme activities were normalized to the wet weight (Fig 1A), the CS activity was ~35% higher in mice than in rats (139.1 ± 4.7 vs 101.4 ± 1.6 μmol
min-1
g ww-1 respectively; p < 0.001). The AK activity was ~30% higher in mice than in rats in both directions (p < 0.001). The CKr activity in the direction of ATP-production tended to be higher in rat than in mouse heart (p = 0.0815), and the CKf activity measured in the direction of ADP-pro- duction was 11% lower in mice than in rats (p < 0.05). The CO activity was similar for mice and rats. As the CS activity was different in mouse and rat hearts, normalizing hereto changed the pattern (Fig 1B). The AK/CS activity (measured in both directions) was similar in mice and rats, whereas the CK/CS activity was ~35% lower in mice than in rats in both directions (Fig 1B; p < 0.0001), and CO/CS activity was ~30% lower in mice than in rats (Fig 1B; p < 0.01). The CK isoform distribution in the hearts of mice and rats was assessed by gel electrophore- sis. A representative picture is shown in Fig 2. For each lane, the intensity of the four major bands corresponding to Mi-, MM-, MB and BB-CK was quantified, and the fractional intensity of each band was calculated. The averaged data are given in Table 3. Compared to mouse hearts, rat hearts had a different CK isoform distribution with a smaller fraction of MM-CK, and larger fractions of Mi-CK, MB-CK and BB-CK. Stimulation of respiration by CK and AK In permeabilized cardiomyocytes, the stimulation of respiration by CK or AK was assessed relative to the maximal coupled respiration rate (VO2_max) (Fig 3). CK-stimulated Fig 2. The CK isoform distribution in rat and mouse hearts. The CK isoform distribution in rat and mouse hearts was assessed by agarose gel electrophoresis. A: Representative picture showing on the left the raw image, and, on the right the same image, in pseudocolour to highlight the bands. B. The intensity profiles from the rat and mouse lanes shown in A. https://doi.org/10.1371/journal.pone.0294718.g002 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 5 / 25 respiration rate was recorded with either substrates for complex I alone (GM; protocol named CKADP_GM) or substrates for complexes I and II (GMPS; protocol named CKADP_GMPS). AK-stimulated respiration was recorded only with substrates for complexes I and II (GMPS; protocol named AKADP_GMPS). A representative experimental trace of Table 3. Distribution of CK isoforms in the hearts of mice and rats. Mi-CK, % MM-CK, % MB-CK, % BB-CK, % Mouse heart 31.6 ± 1.8 63.8 ± 1.6 3.7 ± 0.3 0.83 ± 0.10 Rat heart 37.1 ± 1.1* 50.5 ± 0.5 *** 11.1 ± 0.8 *** 1.20 ± 0.03* The mitochondrial Mi-CK and the three cytosolic MM-, MB-, and BB-CK isoforms in cardiac homogenates from 4 mice and 4 rats were separated by electrophoresis, and the fraction of each isoform is shown as mean ± SEM. * denotes p < 0.05 and *** p < 0.001, significant difference between species. https://doi.org/10.1371/journal.pone.0294718.t003 Fig 3. Stimulation of respiration by CK and AK assessed relative to the maximal respiration rate in permeabilized mouse and rat cardiomyocytes. When normalized to the CS activity, CK stimulated the respiration rate (μmol O2
min-1
IU CS-1) more in rat than in mouse cardiomyocytes. A: Representative example of a respirometer recording of CKADP_GMPS with permeabilized rat cardiomyocytes after addition of: CM–cardiomyocytes; ATP– 2 mM ATP; ADP– 2 mM ADP; Cyt c– 8 μM cytochrome c; FCCP–FCCP was gradually increased, the number behind shows the final concentration in μM. B: The averaged results of the experiments in the presence of GM, where respiration was stimulated by CK. C and D: The averaged results of the experiments in the presence of GMPS, where respiration was stimulated by AK (C) and CK (D). V0 was subtracted from the subsequent rates. In B and D, as creatine was already present in the respiration chamber, ATP is the CK-stimulated respiration rate. In C, ATP is the rate stimulated by non-specific ATPases, and AMP is the AK-stimulated respiration rate. ADP is the maximal respiration rate recorded with 2 mM ADP. The number of animals was n = 8 and n = 5–7 for mice and rats, respectively. * denotes p < 0.05, ** p < 0.01, *** p < 0.001 significant difference between species. https://doi.org/10.1371/journal.pone.0294718.g003 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 6 / 25 CKADP_GMPS is shown in Fig 3A. Due to the low solubility of creatine, it was present in the medium before the start of the experiments at a concentration of 20 mM. After addition of cardiomyocytes, and recording of the basal respiration rate (V0), the CK reaction was stimu- lated by the addition of 2 mM ATP. Then, 2 mM ADP was added to stimulate the respiration rate to VO2_max. Finally, cytochrome c (Cyt c) was added to test the intactness of the outer mitochondrial membrane, and the ionophore carbonyl cyanide-p-trifluoromethoxyphenyl- hydrazone (FCCP) was added in steps to determine the maximal electron flux capacity, i.e. the maximal uncoupled respiration rate. For AK-stimulated respiration measurements, AMP was added (instead of creatine) after addition of ATP. Fig 3B–3D show the averaged results and the statistical analyses from the respiration measurements of CKADP_GM, AKADP_GMPS, and CKADP_GMPS, respectively, normalized to the CS activity in cell suspen- sions from mice and rats. Table 4 shows the respiration rates stimulated by CK (VO2_CK) and AK (VO2_AK), and the maximal coupled respiration rate (VO2_max) normalized to the CS activity of the cell suspension. In addition, it shows VO2_CK and VO2_AK relative to VO2_max. Note that in Fig 3 and Table 4, all respiration rates recorded after V0 had V0 subtracted before analysis. The same notation was used throughout the study. Mitochondrial respiration has a lower phosphate to oxygen (P/O2) ratio with GMPS than with GM, i.e. with GMPS as substrates, fewer ADP molecules are phosphorylated to ATP for a given O2 consumption. This is because complex I substrates translocate 10 H+/O2, whereas complex II substrates translocate 6 H+/O2 across the inner mitochondrial membrane [42]. As the proton gradient across the inner mitochondrial membrane is the driving force for oxida- tive phosphorylation by the F1F0 ATPase, the higher H+/O2 with GM than with GMPS II leads to the theoretical P/O2 ratios of 6 and 4 for GM and GMPS, respectively [43]. To determine whether differences in respiration rates were associated with differences in ADP-phosphoryla- tion rates, we multiplied the respiration rates in Table 4 with these P/O2 ratios to calculate the rates of ADP-phosphorylation, when respiration was stimulated by CK (VADP_CK), AK (VAD- P_AK), or 2 mM ADP (VADP_max). These values are shown in Table 5. Table 4. Respiration rates of permeabilized mouse and rat cardiomyocytes stimulated by CK, AK, or 2 mM ADP. VO2_CK VO2_AK VO2_max VO2_CK /VO2_max VO2_AK/VO2_max nmol O2 /min/IU CS % Mouse GM 55 ± 2 ## 61 ± 3 91 ± 2 GMPS 93 ± 4 #### 89 ± 9 #### 124 ± 4 75 ± 1 79 ± 2 Rat GM 77 ± 4 72 ± 4 107 ± 4 GMPS 113 ± 4 ## 79 ± 7 ### 128 ± 7 89 ± 3 71 ± 3 Substrate **** **** *** Species *** *** * The respiration of permeabilized cardiomyocytes was recorded in the presence of either GM or GMPS as substrates. Under these conditions, the respiration rate is limited by the availability of ADP. The respiration was stimulated by endogenous ADP generated by CK (VO2_CK), or AK (VO2_AK). The maximal respiration rate (VO2_max), was recorded in the presence of 2 mM exogenous ADP. In addition, the fractional stimulation of respiration by CK and AK relative to VO2_max was determined (VO2_CK /VO2_max and VO2_AK/VO2_max, respectively). Values from 8 mice and 5–7 rats are shown as mean ± SEM. ## denotes p < 0.01, ### p < 0.001, #### p < 0.0001, significantly different from VO2_max. * denotes p < 0.05, *** p < 0.001, **** p < 0.0001, significant effect of substrate or species. We found no interaction between substrates and species. https://doi.org/10.1371/journal.pone.0294718.t004 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 7 / 25 The CK-stimulated respiration rates, VO2_CK, were species dependent and significantly higher in rat than in mouse cardiomyocytes in both CKADP_GM and CKADP_GMPS (Fig 3B and 3D, p < 0.01 and p < 0.001, respectively). However, after addition of ADP, Cyt c and FCCP, there was no significant difference in respiration rates between mice and rats (Fig 3B and 3D). As a result, the CK-stimulated respiration rate relative to the maximal respiration rate, VO2_CK/VO2_max, was higher in rats than in mice (Table 4; p < 0.001). The CK-stimulated respiration rates were also substrate dependent (Table 4, compare Fig 3B and 3D), but the estimated rates of ADP-phosphorylation (VADP_CK) were only affected by species and not substrate (Table 5). This suggests that the rate with which CK generated ADP to stimulate respiration was not affected by substrates but resulted in different VO2_CK rates because of the substrate dependent P/O2 ratios. The respiration rate stimulated by AK, VO2_AK, did not differ between mouse and rat cardi- omyocytes (AKADP_GMPS; Fig 3C). However, when assessed relative to the maximal respira- tion, VO2_AK/VO2_max was slightly higher in mouse than in rat cardiomyocytes (Table 4; p < 0.05). When VO2_max was normalized to the CS activity of the cell suspension, there was no dif- ference between mouse and rat cardiomyocytes (Fig 3 and Table 4). However, the CS activity was higher in mouse than in rat cardiomyocytes (Table 2). When VO2_max was normalized to the protein content of the cell suspension (see S1 Table), it was significantly higher in mouse than in rat cardiomyocytes. Thus, VO2_max correlated with the CS activity of the cell suspension. VO2_max was significantly higher than VO2_CK except in rat cardiomyocytes with only com- plex I substrates (GM). VO2_max was also significantly higher than VO2_AK in both mouse and rat cardiomyocytes (Table 4). VO2_max was significantly affected by the substrates. In the presence of GM, VO2_max was lower than in the presence of GMPS (Table 4). VADP_max was also significantly affected by the substrates (Table 5). Thus, in the presence of 2 mM ADP, the rate of ADP-phosphorylation was lower with GM alone than with GMPS. The quality of the permeabilized cardiomyocytes was assessed through the coupling effi- ciency of respiration and the Cyt c test. The coupling efficiency of respiration, calculated as 1-V0/VO2_max, indicates the proportion of oxygen used for ATP synthesis. In permeabilized cardiomyocytes, a high coupling efficiency indicates that they are 1) adequately permeabilized Table 5. Estimated rates of ADP-phosphorylation by mitochondria stimulated by CK, AK or 2 mM ADP. VADP_CK VADP_AK VADP_max nmol ADP/min/IU CS Mouse GM 332 ± 13 367 ± 19 GMPS 373 ± 14 354 ± 41 498 ± 18 Rat GM 461 ± 23 433 ± 27 GMPS 451 ± 17 315 ± 26 512 ± 28 Substrate **** Species *** We used the data from Table 4 to estimate the rates of ADP-phosphorylation by mitochondria, when respiration was stimulated by endogenous ADP from CK (VADP_CK) or AK (VADP_AK), or 2 mM exogenous ADP (VADP_max). The respiration rates were multiplied by a P/O2 ratio of 6 or 4, for GM and GMPS, respectively, as explained in the main text. Values from 8 mice and 5–7 rats are shown as mean ± SEM. *** denotes p < 0.001, **** p < 0.0001, significant effect of substrate or species. We found no interaction between substrates and species. https://doi.org/10.1371/journal.pone.0294718.t005 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 8 / 25 and respond to ADP addition, and 2) not damaged, as this would cause an elevated V0. According to Chance and Williams, tightly coupled mitochondria show a substrate-dependent 4- to 10-fold increase in respiration rate upon addition of ADP [44]. This corresponds to a coupling efficiency of 0.75–0.90. In the present experiment, the coupling efficiency was similar in mouse and rat cardiomyocytes but substrate dependent and lower with GMPS than with GM (81.6 ± 0.5% and 89.7 ± 0.7%, respectively; p < 0.0001; data were pooled for rats and mice). The addition of Cyt c is commonly used to test the intactness of the outer mitochondrial membrane. If the outer mitochondrial membrane is damaged, electron transfer will be com- promised as Cyt c leaks out of the mitochondria. The associated decline in respiration rate will be rescued upon addition of Cyt c. In our experiments, the addition of Cyt c did not signifi- cantly affect the respiration rate in any of the measurements (Fig 3), indicating that in the car- diomyocytes in the present study, the outer mitochondrial membrane was intact. The addition of FCCP, an uncoupler of respiration, did not increase the respiration rate with GM only. But with GMPS, FCCP increased the respiration rate by ~40% (Fig 3). The phosphorylation control ratio, calculated as VO2_max/VFCCP, was 106 ± 2.1 and 58.3 ± 0.6% with GM and GMPS, respectively (pooled data from rats and mice). Channelling of ADP from CK or AK to the mitochondria The ADP channelling between CK or AK and the mitochondria was assessed in parallel experiments, where the kinase stimulated respiration was subsequently inhibited by addition of PEP and PK. Using the same substrate-kinase combinations as before, CK-stimulated res- piration rate was recorded with either GM (CKPEP/PK_GM) or GMPS (CKPEP/PK_GMPS), and AK-stimulated respiration was recorded only with GMPS (AKPEP/PK_GMPS). Fig 4A shows a representative experimental trace from a recording of CKPEP/PK_GMPS. After addition of cardi- omyocytes (CM) to the respiration chamber, addition of ATP to the solution (already con- taining 20 mM creatine) stimulated CK to generate endogenous ADP. This ADP distributed in the solution and stimulated respiration. Then, endogenous PK was stimulated by addition of PEP and competed with mitochondrial respiration for some of the ADP. Subsequently, exogenous PK in excess was added to the chamber. Thus, PK converted all accessible ADP in the solution to ATP, leaving only ADP that was directly channelled from CK to mitochondria to stimulate respiration. For AK-stimulated respiration measurements, AMP was added (instead of creatine) after addition of ATP. Fig 4B–4D, shows the averaged results and the sta- tistical analysis from the respiration measurements of CKPEP/PK_GM, AKPEP/PK_GMPS, and CKPEP/PK_GMPS, respectively, normalized to the CS activity in cell suspensions from mouse and rat hearts. CK-stimulated respiration was significantly higher in rats than in mice irrespective of the substrates (Fig 4B; p < 0.0001; Fig 4D; p < 0.01), in agreement with the results in Figs 1B and 3. In CKPEP/PK_GM, the addition of PEP and PK lowered respiration rate more in rat than in mouse cardiomyocytes (by 75 ± 1% and 58 ± 1%, respectively; p < 0.001), and the respiration rate in the presence of PEP and PK was lower in rat than in mouse (Fig 4B). In CKPEP/PK_GMPS, the addition of PEP and PK also lowered respiration rate more in rat than in mouse cardio- myocytes (by 62 ± 1% and 48 ± 1%, respectively; p < 0.001), as the initially higher respiration rate in rat cardiomyocytes was lowered to the same level as in mice in the presence of PEP and PK (Fig 4D). In AKPEP/PK_GMPS, there was no difference between mice and rats, and the respiration rate was lowered to the same level as before addition of AMP (Fig 4C; compare rates at ATP and PK). PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 9 / 25 Discussion This is the first study to show that there are differences between rat and mouse hearts in terms of both aerobic capacity, CK isoform distribution, and intracellular compartmentalization. When comparing the measurements in homogenate and permeabilized cardiomyocytes, we were surprised to find that CK and, to an even greater extent, AK activities assessed in heart homogenates were much higher than the ADP-phosphorylation rates estimated from the CK- and AK-stimulated respiration rates of permeabilized cardiomyocytes. This is a consequence of intracellular compartmentalization and demonstrates that results from whole-heart homog- enates cannot be directly extrapolated to the situation in permeabilized cardiomyocytes. Rat hearts have a lower oxidative capacity than mouse hearts The CS activities reported in the present study were close to activities reported in other studies [35, 45–48]. The CS activity was higher in mouse than in rat in both whole heart homogenates and cell suspensions (Fig 1A and Table 2). Furthermore, when normalized to protein content, VO2_max was higher in mouse than in rat cardiomyocytes (S1 Table). Thus, all three measure- ments indicate that rat hearts have a lower oxidative capacity than mouse hearts. Fig 4. Channeling of ADP from creatine kinase (CK) and adenylate kinase (AK) to mitochondria in permeabilized mouse and rat cardiomyocytes. As in Fig 3, when normalized to the CS activity, CK stimulated the respiration rate (μmol O2
min-1
IU CS-1) more in rat than in mouse cardiomyocytes. However, this species difference was reversed (B) or lost (D) after addition of PEP and PK to compete with the mitochondria for endogenous ADP from CK. A: Representative example of a respirometer recording of CKPEP/PK_GMPS with permeabilized rat cardiomyocytes after addition of: CM–cardiomyocytes; ATP– 2 mM ATP; PEP– 5 mM PEP; PK– 20 U/ml PK. B: The averaged results of the experiments in the presence of GM, where respiration was stimulated by CK. C and D: The averaged results of the respiration experiments in the presence of GMPS, where respiration was stimulated by AK (C) and CK (D). V0 was subtracted from the subsequent rates. For further explanation, see the legend of Fig 3 and the main text. The number of animals was n = 7–8 and n = 5–7 for mice and rats, respectively. * denotes p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001 significant difference between species. https://doi.org/10.1371/journal.pone.0294718.g004 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 10 / 25 In skeletal muscle, the oxidative capacity correlates with the CO activity [34]. However, in the present study, the CO activity was similar in mouse and rat hearts (Fig 1A), and VO2_max correlated with the CS activity (Fig 3). This suggests that, in heart muscle, the CS activity is a better marker of oxidative capacity. As the CS activity correlated with the mitochondrial oxidative capacity, and one of our aims was to assess the rate of ADP-generation by AK and CK relative to the mitochondrial oxi- dative capacity, we normalized the respiration data to the CS activity. As described in the Introduction, we assumed that CS is mainly from cardiomyocytes, and that the CS in other cell types is negligible. Furthermore, we assumed that this distribution of CS is similar in rat and mouse heart, so that normalizing to CS allowed us to compare the reaction rates in whole heart homogenates and isolated cardiomyocytes across the species. Rat hearts have more Mi-CK and B-CK, and mouse hearts have more M-CK The CK isoform distribution was also different in mouse and rat hearts (Fig 2 and Table 3). Mouse hearts had a larger fraction of MM-CK, whereas rat hearts had larger fractions of Mi-, MB- and BB-CK. The large fraction of MB-CK in rat heart is consistent with results in the liter- ature [14, 17, 49]. We are uncertain whether rat heart benefits functionally from having a greater expression of B-CK. B-CK has a higher creatine affinity than M-CK [50], but the total creatine content seems to be similar in rat and mouse ventricles (~80 nmol/mg protein) [49, 51]. MM-CK and BB-CK are both mainly soluble, but a fraction associates to cellular struc- tures, and they seem to do so differently. Their N-terminal regions differ, and due to four con- served lysine residues, MM-CK binds to the M-band, whereas BB-CK does not [52, 53]. Instead, B-CK binds to the I-band of the myofibrils [52, 54, 55]. MM-CK is also found at the I- band of the myofibrils, but here, it is bound loosely and through phosphofructokinase of the glycolytic pathway [56]. Both MM-CK and BB-CK are known to associate with membranes. In muscle tissue, MM-CK associates near the sarcoendoplasmic reticulum Ca2+-ATPase (SERCA) and the Na+/K+-ATPase [57, 58]. More recent evidence suggests that BB-CK also locates near membrane structures, in some cases in a manner that is regulated through phos- phorylation by AMP-activated protein kinase (AMPK), suggesting that this association may be weaker and transient depending on the state of the cell [59]. We were unable to find informa- tion regarding the heterodimeric MB-CK. At present, we speculate whether the differences in CK isoform composition in rat and mouse heart relate to the different binding properties of M- and B-CK, but this warrants further studies. Rat hearts have a higher CK activity and larger fraction of Mi-CK, but less ADP from CK is channelled to the mitochondria Traditionally, the reverse CK activity is recorded in homogenates. When taking into account temperature differences, the CK activities in rat and mouse hearts (Fig 1A) were close to and a little higher, respectively, than reported in other studies [45, 46, 48, 60]. When normalized to the wet weight, CKr tended to be higher and CKf activity was slightly, but significantly higher in rat than in mouse heart (Fig 1A). When normalized to the CS activity, the CK activities were clearly higher in rat than in mouse heart (Fig 1B). This difference was also reflected in the experiments on permeabilized cardiomyocytes, where stimulation of CK led to higher VO2_CK in rat than in mouse cardio- myocytes (Fig 3B and 3D and Table 4). As expected, VO2_max was substrate dependent with VO2_max_GM being lower than VO2_max_GMPS (compare Fig 3B and 3D; Table 4). This has also been shown before [35, 36]. As a result, VO2_CK/VO2_max was also substrate dependent PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 11 / 25 (Table 4; effect of substrate), and consistently higher in rat than in mouse heart (Table 4; effect of species). Although rat cardiomyocytes had a higher VO2_CK/VO2_max, the subsequent addition of PEP and PK to competitively inhibit the flux of ADP from CK to mitochondria lowered the respiration rate more in rat than in mouse cardiomyocytes. In the presence of PEP and PK, the respiration rate was similar (CKPEP/PK_GMPS) or slightly lower (CKPEP/PK_GM) in rats than in mice (Fig 4B and 4D). This suggests that in mouse hearts, a larger fraction of ADP generated by CK was channelled to the mitochondria. However, the fraction of Mi-CK was lower in mouse than in rat hearts (Fig 2 and Table 3). It may seem contradictory that mouse cardio- myocytes have less Mi-CK and greater ADP channelling to the mitochondria. However, this can be explained by differences in the intracellular compartmentalization. The greater channelling of ADP from CK to mitochondria in mouse heart could be due to a tighter cou- pling of cytosolic CK as well as Mi-CK to the respiration. As to Mi-CK, in isolated mitochondria, approximately half of the ADP generated by Mi-CK is translocated by the ANT into the matrix, while the rest diffuses out through the voltage dependent anion channel (VDAC) in the outer mitochondrial membrane [6, 61]. In vivo and in permeabilized cardiomyocytes, the mitochondrial outer membrane permeability is more restricted [62]. This, in turn, is expected to increase the channeling between Mi-CK and ANT. Thus, it is possible that there is more direct transfer between Mi-CK and ANT or the outer mitochondrial membrane is less permeable to ADP in mouse cardiomyocytes, so more ADP from Mi-CK cycles within the mitochondria and is inaccessible to PK. As to cytosolic CK, it was shown in Mi-CK knockout mice that cytosolic CK can also be coupled to respiration [63] possibly due to intracellular diffusion barriers in the cytosol, which can group CK and mitochondria [39, 64–67]. The extent of channelling we observe between cytosolic CK and the mitochondria depends on the interplay between cytosolic CK, PK, and diffusion barriers. It is possible that in mouse cardiomyocytes, a larger fraction of cytosolic CK is on the mitochondrial side of the diffusion barriers, or the diffusion barriers are less perme- able, so more ADP cycles between CK and the mitochondria and is inaccessible to PK. If the cytosolic diffusion barriers are less permeable to ADP, then they are presumably also less permeable to the diffusion of proteins. This is a relevant point for the present experiments on permeabilized cardiomyocytes, which do not necessarily reflect the situation in vivo, because some cytosolic CK and PK may diffuse out of the cells, and also some exogenous PK diffuses into the cell. The diffusion of CK out of the cardiomyocytes does not on its own inhibit respiration, because CK continues to generate ADP in the solution. However, it may lead to an underestimation of the overall coupling between all CK isoforms and respiration. Thus, if the cytosolic diffusion barriers are less permeable in mouse cardiomyocytes, it is also possible that the species differences in the PEP-PK assay are caused in part by the differences in CK and PK diffusion between solution and cardiomyocytes. Clearly, further studies are needed to pinpoint the exact mechanism behind the different outcomes of the PEP-PK assay. Nevertheless, our results suggest that compartmentalization and/or energy transfer is different in rat and mouse cardiomyocytes. The lowering of respiration rate by PEP and PK in mouse cardiomyocytes is in agreement with our previous finding [21]. With the present study, we extend this finding to rat cardio- myocytes. However, our results contradict the findings from another group, who found on rat cardiomyocytes that with GM as substrates CK-stimulated respiration rate was similar to VO2_max even in the presence of PEP and PK [22, 23]. We speculated whether this difference between studies could be because the cardiomyocytes in the other study had damaged mito- chondria. If the mitochondria are damaged, VO2_max will be very low and VO2_CK/VO2_max will be high. In order to compare our data with those of others, we also normalized our maximal PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 12 / 25 respiration rate in rat cardiomyocytes to the Cyt aa3 content (S1 Table, VO2_max/cyt aa3). In the present study, VO2_max_GM was ~35% higher than reported by others (273 ± 19 versus 178 ± 34 nmol O2
min-1nmol
cyt aa3 -1) [68]. However, in rat cardiomyocytes in the present study, the addition of PEP and PK lowered the respiration rate by 75% (Fig 4B). As the difference in VO2_max between studies was smaller than the lowering of respiration rate by PEP and PK, the higher VO2_max in the present study can only partially explain the difference between the stud- ies. In the present study, the cell viability was acceptable (Table 2), the addition of Cyt c had no effect on respiration rate (Fig 3B–3D), and the coupling efficiency of respiration was high. Therefore, we are confident that our preparation was sound. Taken together, our study confirms that intracellular compartmentalization leads to a local pool of phosphates circulating between mitochondria and CK and suggests that this is more predominant in mouse than in rat cardiomyocytes. It must be noted that these experiments were performed on isolated, non-contracting cardiomyocytes, and it is uncertain how defor- mation of the cells during contraction affects the compartmentalization. The rate of ADP-generation by CK is lower than the maximal rate of ADP- consumption by the mitochondria As noted above, rat hearts had a higher VO2_CK/VO2_max than mouse hearts. According to our data, mitochondrial and cytosolic CK stimulated respiration rate to 90% of the maximal rate with GMPS in rat (Fig 3D and Table 4). This is far from the finding on isolated rat hearts that the CK reaction rate is 10 times higher than the maximal respiration rate [4, 5], but it is in agreement with a previous study [69]. This may explain why the CK shuttle is bypassed under extreme workloads [69], and why studies on transgenic mice with disturbances in the CK sys- tem have been equivocal regarding the importance of the CK system [11]. Rat and mouse hearts have similar VO2_AK/VO2_max In homogenates, the AK activity normalized to the wet weight was ~30% higher in mouse than in rat hearts, measured in both directions (Fig 1A). In mouse hearts, the AK activity was ~1.5– 2 times higher than reported previously [21, 45, 47], whereas in rat hearts, it was only slightly higher than in another study [46]. In contrast to our previous study on creatine-deficient mice, where AK had the highest activity of the kinases [21], the present results showed that the AK activity was lower than the CK activity in both mouse and rat hearts, when measured in the direction of ATP-production. The AK activities were similar in the forward and reverse direc- tions, which is compatible with another study [70], showing that the AKf/AKr ratio is ~1.3. When the AK activity was normalized to the CS activity in homogenates, it was similar in mouse and rat heart (Fig 1B). In agreement with this, there was no difference between rat and mouse in the respiration experiments (Fig 3C). However, relative to the maximal respiration rate, VO2_AK/VO2_max was slightly higher in mouse than in rat hearts (Table 4), but below 80%. In our previous experiments, AK stimulated respiration to the maximum, but they were per- formed with only GM as substrates [21]. In the present study, AK-stimulated respiration mea- surements were performed only with GMPS, and as VO2_max is higher with GMPS than with GM, it was not surprising, that VO2_AK/VO2_max was lower than recorded with GM, as was also the case for CK (Fig 3B and 3D, Table 4). No channelling of ADP from AK to the mitochondria When the flux of ADP from AK to the mitochondria was inhibited by the subsequent addition of PEP and PK, the respiration rate was lowered to the same level as before the addition of AMP in both rat and mouse cardiomyocytes (Fig 4C). This was similar to our previous results PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 13 / 25 on mouse cardiomyocytes [21], and in agreement with the low expression of AK2 in the mouse heart [32], but in contrast to another study on isolated rat heart mitochondria [27]. We hypothesize that methodological differences could cause this discrepancy between studies. The present study suggests that the majority of AK is cytosolic in both rat and mouse hearts. AK and CK activities in homogenate are higher than estimated VADP_AK and VADP_CK in permeabilized cardiomyocytes We estimated the rates of ADP-generation by CK and AK in permeabilized cardiomyocytes (VADP_CK and VADP_AK, Table 5) by multiplying the respiration rates with the P/O2 ratios (4 for GMPS and 6 for GM). This represents their forward reaction rates, and we had expected that they would be similar to the forward reaction rates recorded in homogenate (CKf and AKf, Fig 1B). Surprisingly, we found that the rates estimated from permeabilized cardiomyo- cytes were much lower than in homogenate. CKf was ~2 times higher and AKf was ~9 times higher than VADP_CK and VADP_AK, respectively. This difference is highlighted in Fig 5. Our finding suggests that in permeabilized cardiomyocytes with the intracellular structures left intact, local substrate and product concentrations in the vicinity of an enzyme can be very dif- ferent from the concentrations in solution. One factor is that diffusion of substrates into the permeabilized cardiomyocytes is restricted so the substrate concentrations could be smaller Fig 5. Intracellular compartmentalization shapes energy transfer in cardiomyocytes. In permeabilized cardiomyocytes (left panel; data from Table 5, units converted to μmol ADP
min-1
IU CS-1), diffusion of substrates to the center of the cell is restricted. Thus, kinases in the center of the cell may be exposed to smaller concentrations of substrates than are present in the surrounding solution. More importantly, as diffusion out of the cell is restricted, the products accumulate near the kinase and inhibit the reaction rate. In contrast, when recording the kinase activity in homogenate (right panel; data from Fig 1), the kinases are in solution, where diffusion is much faster. Thus, there is no build-up of products, and the reaction takes place without inhibition. https://doi.org/10.1371/journal.pone.0294718.g005 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 14 / 25 than in solution. More importantly, the diffusion of products out of the permeabilized cardio- myocytes is restricted, so the products accumulate near the kinases, inhibiting the reaction rate. We have illustrated this in Fig 5. The restriction of diffusion in cardiomyocytes has been studied for several decades. In car- diomyocytes, intracellular membranes such as the transverse tubules, the sarcoplasmic reticu- lum, and the mitochondria, together with protein dense parts of the sarcomeres constitute barriers that form modules [64, 65]. These modules influence energy transfer within the cell [39]. Local environments are also known to play a crucial role in cAMP signalling [71], excita- tion-contraction coupling [72] and mitochondrial calcium uptake [73, 74]. In the present study, we were surprised to see that compartmentalization had such a large effect on the AK activity. This, in turn, is likely to affect not only ATPase function, but also energetic signalling through, for example, AMPK, which is a cellular energy sensor implicated in both acute signal- ling and regulation of gene expression [75]. Our results highlight the challenge for computa- tional models, which should take into account the intracellular heterogeneity of substrate and product concentrations and the spatial limitations of their diffusion inside the cell [62, 64, 65]. Conclusions In the present study, we found species differences between mouse and rat hearts. Rat hearts had a lower oxidative capacity than mouse hearts. As a result, CK/CS and VO2_CK/VO2_max were higher in rat than in mouse, and the distribution of CK isoforms was different. In rat heart, although VO2_CK/VO2_max and the fraction of Mi-CK was higher than in mouse heart, less ADP was channelled from CK to the mitochondria. This suggests differences in the compartmentalization of mouse and rat cardiomyocytes. An interesting finding of this study was that AK/CS activity in whole tissue homogenates was several times higher than the VADP_AK estimated from the respiration rate in isolated per- meabilized cardiomyocytes. This difference is a consequence of intracellular compartmentali- zation. Our results highlight how intracellular structural organization shapes energetic compartmentalization, which plays a pivotal role in energy homeostasis, signalling, and regula- tion of cardiac metabolism. Materials and methods All experiments and animal procedures complied with directive 2010/63/EU of the European Parliament for the protection of animals used for scientific purposes and were approved by the Project Authorisation Committee for Animal Experiments in the Estonian Ministry of Rural Affairs. All methods are reported in accordance with ARRIVE guidelines. Animals The animals used in this study were 7–10 months for mice and 10–12 months for rats. Due to the low quality of cardiomyocytes isolated from male rat hearts, only females were used in this study. Sprague-Dawley rats were a gift from the Laboratory of Neurobiology at Tallinn Univer- sity of Technology. C57BL/6J Ola Hsd mice were originally from Envigo RMS B.V. (The Neth- erlands). The animals were kept in the animal facility of Tallinn University of Technology at an ambient temperature of 22–22.8˚C and a 12:12 hours light:dark cycle. They had free access to water and food (V1534-000 Rat/mouse maintenance from Ssniff Spezialdia¨ten GmbH, Germany). PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 15 / 25 Isolation of cardiomyocytes Cardiomyocytes from mouse [15] and rat [76] hearts were isolated using a slightly modified version of a method described previously. The mice were anesthetized with a mixture of keta- mine/dexmedetomidine (150 mg/kg and 0.5mg/kg, respectively) and received an injection of 250U of heparin to prevent blood coagulation. When the toe-pinch reflex was absent, the ani- mal was euthanized by cervical dislocation. The rats were anesthetized with 2% isoflurane using Open-Drop system (or Drop Jar method) (https://animal.research.uiowa.edu/iacuc- guidelines-anesthesia) [77]. Briefly, the rats were placed in a closed container of known volume (~5l) with tightly fitting lid, a gauze pad soaked with appropriate volume (~0.5 ml) of isoflur- ane was placed in the bottom of the container and the animals were left to inhale isoflurane vapours. When they lost the righting reflex and breathing had slowed but was regular, 2500U of heparin was injected intraperitoneally and they were allowed to sleep a little bit more. Under deep isoflurane anaesthesia, the rats were euthanized by decapitation. The hearts of both rat and mouse were excised and immediately placed in ice-cold wash solution consisting of the following (mM): 117 NaCl, 5.7 KCl, 1.5 KH2PO4, 4.4 NaHCO3, 1.7 MgCl2, 21 HEPES, 20 taurine, 11.7 glucose, and 10 2,3-butanedione monoxime (pH was adjusted to 7.4 with NaOH). It was cannulated via the aorta on a Langendorff perfusion system. The heart was first perfused with wash solution at 38.5˚C at a constant pressure of 80 cm H2O. When the heart was washed free of blood, the perfusion was switched to a constant flow with digestion solution containing 0.37–0.435 mg/ml Liberase DL (Roche) and 1.36 mg/ml of dispase II (Roche). The pressure was observed for 10–15 minutes or ~30 min (mouse and rat heart, respectively) until the pressure had decreased to 40–50% of the initial. When the heart was soft, the perfusion was stopped. The ventricles were cut into smaller pieces, transferred to a beaker with digestion solution and incubated further at 38.5˚C with gentle shaking until the tissue started falling apart. Cells were harvested with a Pasteur pipette several times and filtered through a 100μm cell strainer (EASYstrainerTM Cell Strainer, Greiner Bio-One) into a vial with sedimentation solution consisting of wash solution (without 2,3-butanedione monoxime) containing addi- tional 2mM pyruvate, 10 μM leupeptin, 2 μM soybean trypsin inhibitor, and 3 mg
mL-1 BSA. The viable cells were separated by sedimentation or by centrifugation for 2 min at 300 rpm/ 12g. During the first washes, extracellular Ca2+ was gradually increased to 2 mM to ensure Ca2+ tolerance of the cells. Then, extracellular Ca2+ was washed out again by washing the cells three times with 5–8 ml of sedimentation solution. The isolated cells were stored in this solu- tion at room temperature until use within 3 hours. To assess the quality of the cell preparation, the yield of the cell suspension was measured with a 1000 or 5000 μl pipette (Eppendorf), and a 1:10 dilution of the cells was counted in a chamber to estimate the total number of cells as well as the viability (the percentage of live cells relative to the total number of cells). Respiration measurements For the respiration experiments, we used protocols similar to those described previously [21, 35]. In brief, we used a Strathkelvin RC 650 Respirometer equipped with six 1302 O2-elec- trodes connected via a 929 Oxygen System interface (all from Strathkelvin Instruments Lim- ited, UK). The respirometer was thermostatted to 25˚C (Julabo F12-ED, JULABO Labortechnik GmbH). The respiration measurements in cardiomyocytes were performed in 2 ml of respiration solution consisting of 110 mM sucrose, 60 mM K-lactobionic acid, 3 mM KH2PO4, 3 mM MgCl2, 20 mM HEPES, 20 mM taurine, 0.5 mM EGTA, 0.5 mM dithiothreitol (DTT) (pH was adjusted to 7.1 with KOH). 5 mg
mL-1 BSA and 25 μg
mL-1 saponin were added just before use. Saponin was present in the respiration chamber throughout the PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 16 / 25 measurements. It interacts with cholesterol to form pores in the sarcolemma, which contains 90% of the cellular cholesterol [78]. At this concentration, saponin does not damage the mito- chondrial membranes, which have a very low cholesterol content [79]. CK experiments were performed in the presence of 2.5 mM glutamate and 2 mM malate only (CKGM) and in the presence of 2.5 mM glutamate, 2 mM malate, 5 mM pyruvate and 15 mM succinate (CKGMPS). AK experiments were carried out only in the presence of 2.5 mM glutamate, 2 mM malate, 5 mM pyruvate and 15 mM succinate (AKGMPS). We used CK and AK protocols similar to those described in [21]. First, 5 ul of cell suspen- sion, for CKGMPS and AKGMPS, and 10 ul for CKGM were added to the respiration chambers. Cells were allowed at least 5 min to permeabilize before the steady-state basal respiration rate, Vo, was recorded. After that, we stimulated AK by adding 2 mM ATP and 1 mM AMP, or CK by adding 2 mM ATP while 20 mM creatine was already present in the respirometer chamber before the cells were added. In the chambers, where we recorded CKPEP/PK_GM, CKPEP/PK_GMPS, and AKPEP/PK_GMPS, this was followed by the addition of 5 mM PEP and 20 U/ml exogenous PK. In the parallel chambers, where we recorded CKADP_GM, CKADP_GMPS, and AKADP_GMPS, this was followed by the addition of 2 mM ADP, 10 μM Cyt c and stepwise titration of FCCP in 2.5 μM steps until the maximum uncoupled respiration rate, VFCCP, was reached. As a reference, respiration rate measured at 2 mM ADP was taken as the maximal coupled respiration rate, VO2_max. This measurement was used to estimate AK and CK stimulated res- piration rates (VO2_AK and VO2_CK, respectively) relative to VO2_max (VO2_CK/VO2_max and VO2_AK/VO2_max), and to calculate the coupling efficiency of respiration as follows: (VO2_max−- V0)/VO2_max [37]. This allowed us to determine the oxidative phosphorylation control ratio, defined as: VO2_max/VFCCP [37]. Homogenization Cardiac homogenates were prepared as in Barsunova et al. [80]. The mice and rats were anes- thetized and killed as described above for the isolation of cardiomyocytes. The heart was quickly removed from animals and immediately transferred to a glass beaker with ice-cold iso- lation solution. The heart was trimmed of any obvious fat and connective tissue, gently blotted to remove excess fluids, weighed, cut into several pieces if needed (for rat heart), and then stored in cryovials at -80˚C until further experiments. All subsequent homogenization proce- dures were carried out on ice. The heart tissue was minced with scissors into small pieces, transferred to a glass homogenizer, and ice-cold homogenization buffer was added to a con- centration of 50 mg tissue/ml buffer. The buffer consisted of 5 mM HEPES, 1mM EGTA, 0.1% Triton X-100, 1 mM DL-Dithiothreitol, and 1 tablet of cOmplete Mini Protease inhibitors per 10 ml buffer (Roche, Merck) (pH 8.7). Next, the heart tissue was ground with a pestle attached to a drill until the solution was homogenous. The homogenized samples were incubated on ice for one hour before use. Fresh, non-diluted homogenates were use to measure CO activity. The remaining homogenates were kept at -80˚C until activities of CS, CK and AK were measured. Enzyme activities Enzyme activities were recorded using Evolution 600 spectrophotometer (Thermo Fisher Sci- entific) equipped with a Peltier water-cooled cell changer (SPE 8 W, Thermo Fisher Scientific) to maintain temperature at 25˚C. CO and CS activities were determined as described earlier [80]. CO activity was deter- mined by measuring the decrease in absorbance, caused by oxidation of Cyt c by cytochrome PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 17 / 25 oxidase, at 550 nm [81]. The reaction took place in 1 ml 13 mM sodium phosphate buffer (pH 7.4) containing 0.4 mg/ml Cyt c, which had been reduced with Na-dithionite. After recording the initial absorbance for 10–20 seconds, the reaction was initiated by the addition of 10 μl of undiluted homogenate. The reaction of Cyt c oxidation represents a first-order reaction with respect to reduced Cyt c and is observed as a logarithmical decline in absorp- tion (as a function of time). The rate constant, obtained by fitting to the equation δ(Absorp- tion) / δ(time) = k (Absorption), was normalized to the tissue wet weight, min−1
g−1. The IOCBIO Kinetics software for fitting is open source and available at https://iocbio.gitlab.io/ kinetics. CS activity was recorded using a coupled enzyme assay in a total volume of 1 mL CS buffer containing the following (in mM): 100 Tris
HCl buffer (pH 8.1), 0.1 5,5’ -dithiobis(2-nitroben- zoic acid) (DTNB), and 0.3 acetyl-CoA. The assay was started by the addition of 10 μL or 15– 20 μL of diluted (1:10) cell suspension or heart homogenate, respectively. The change in absor- bance was recorded for 2 min at 412 nm before (for reference) and after addition of 0.5 mM oxaloacetate. The enzyme activity was calculated using the extinction coefficient for thionitro- benzoate (TNB), which is 14150 M−1 cm−1 at 25˚C [82]. The activities of CK and AK in the reverse direction (ATP-production) were measured in a coupled enzyme assay in a total volume of 1ml respiration buffer (without BSA and saponin) (see composition in Respiration measurements section) containing the following: 10 mM glu- cose, 0.6 mM NADP, 2 mM ADP, 5 U
ml-1 hexokinase and 5 U
ml-1 glucose-6-phosphate dehydrogenase. The reaction was initiated by the addition of 3–5 ul of diluted (1:10) heart homogenate. The increase in absorbance was measured for 3 min at 340 nm before (AK activ- ity) and after addition of 10 mM creatine phosphate (CK activity + AK activity). In the second run, 50 uM P1,P5-Di(adenosine-5’)pentaphosphate was added to the buffer to inhibit AK. The absorbance was measured for 3 min at 340 nm before (to verify AK inhibition) and after addi- tion of 10 mM creatine phosphate (CK activity). The activities of CK and AK in the forward direction (ATP-consumption) were measured in a total volume of 1ml respiration buffer containing the following: 5 mM PEP, 2 mM ATP, 30 mM NADH, 5 U/ml PK, 2.5 U/ml lactate dehydrogenase. First, the absorbance was mea- sured for 3 min at 340 nm with buffer to check whether the response is stable (to record the absorbance shift). Then 10 ul of diluted (1:10) heart homogenate was added (non-specific ATPase activity). Finally, the decrease in absorbance was measured after addition of 20 mM creatine (CK activity) or 1 mM AMP (AK activity). The activities of CK and AK in both directions were measured with conditions similar to those used in respiration measurements and were calculated using the extinction coefficient for NADH/NADPH (ε340 = 6.220 mM−1 cm−1). All enzyme activity measurements were performed in triplicate (in quadruplicate for CO) and the results averaged. Data was analysed using IOCBIO Kinetics [83]. Determination of CK isoforms The rat and mouse homogenates were also used to determine the CK isoform distribution as previously described [84]. The CK isoforms were separated by native agarose gel electrophore- sis on a 1% agarose gel of approximately 1 mm thickness. The gel was transferred to a ceramic plate, which in turn was placed upon a cooling pad maintained at 15˚C by a thermostat (Julabo F12-ED, JULABO Labortechnik GmbH). A small strip of filter paper was used to mark the row of loading spots about 1/3 from the anode. Drops of 1 μl homogenate (corresponding to tissue extract from 50 μg heart wet weight) were put on each loading spot with rat and mouse samples put one after another. Electrophoresis buffer was added to the anodic and cathodic PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 18 / 25 chamber, and wicks consisting of 4 layers of filter paper were used to connect the gel to the buffer in the two chambers. The gel was run at 250 V for 120 min. The gel was transferred to the imaging chamber of an ImageQuant 400 (GE Healthcare Life Sciences). A filter paper the size of the gel was soaked in 5 ml visualization buffer and put on top of the gel. As the solution from the filter paper entered the gel, CK in the gel catalyzed the first of a chain of reactions: phosphocreatine + ADP + H+ ! creatine + ATP. ATP from this reaction was used by hexokinase in the reaction: ATP + glucose ! ADP + glucose-6-phos- phate. Finally, glucose-6-phosphate was used by glucose-6-phosphate-dehydrogenase in the reaction: glucose-6-phosphate + NADP+ ! 6-phospho-D-glucono-1,5-lactone + H+ + NADPH. The increase in NADPH from the assay coupled to CK was followed by transillu- mination with UV-light and image capture with a SYBR Green filter. After ~20 min, the filter paper was carefully removed to image the NADPH signal in the gel. The gel pictures were ana- lyzed using ImageJ software. Each lane was marked with a rectangle, and the area of each intensity profile peak corresponding to Mi-CK, MM-CK, MB-CK and BB-CK was noted. The relative intensity of each isoform was calculated. The gel electrophoresis buffer consisted of (in mM): Tris 60, Tricine 60, EGTA 1, dithiotrei- tol 1, Triton X-100 0.1%, pH 8.6. The visualization buffer consisted of (in mM): N-Acetyl-L-cysteine 120, phosphocreatine 120, glucose 70, MgAcetate 50, MES 22, ADP 9, β-Nicotinamide adenine dinucleotide phos- phate 9, P1P5-Di(adenosine-50) pentaphosphate pentasodium salt 0.2, pH 7.4. Immediately before use, hexokinase and glucose-6-phosphate dehydrogenase were both added to the visual- ization buffer to a final concentration of 5 IU/ml. Determination of Cyt aa3 content The Cyt aa3 content in isolated rat cardiomyocytes was determined using a spectroscopic method independent of myoglobin contamination. This assay relies on the selective reduction of mitochondrial cytochromes by the action of sodium cyanide [85]. The rat cardiomyocytes were solubilized with 5% TritonX-100 in 0.1M potassium phosphate buffer (pH 7.5). The first graph (oxidized cytochromes) was obtained by scanning from 500 to 650 nm using an Evolu- tion 600 spectrophotometer (Thermo Fisher Scientific Inc.) equipped with a Peltier water cooled cell changer (SPE 8 W; Thermo Fisher Scientific Inc.) to maintain temperature at 25˚C. The second graph (reduced cytochromes) was obtained the same way after reduction with 2 mM sodium cyanide in the presence of 15 mM ascorbic acid. The differential absorbance (reduced versus oxidized cytochromes) at 605 and 630 nm was used for quantification of respi- ratory chain Cyt aa3 content (cytochrome c oxidase), using the extinction coefficient ε605 = 18.6 mM-1 cm-1 [86]. Normalization For better comparison of our results to the findings of earlier studies, the respiration rates from permeabilized cardiomyocytes were normalized to protein content, Cyt aa3 content (rat cardiomyocytes only), and CS activity. Enzyme activities in homogenates were normalized to wet weight and CS activity. The protein content in cardiomyocytes was measured spectropho- tometrically in a BioSpec-nano (Shimadzu Scientific Instruments Inc., Columbia, MD) as pre- viously described [21]. Statistics The values are shown as mean ± standard error of the mean (SEM). Statistical analysis of most data was performed using unpaired Student’s t-test using R. However, the difference between PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 19 / 25 VO2_CK and VO2_max, and VO2_AK and VO2_max was assessed using a paired t-test. Furthermore, for the comparison of VO2 rates and VADP rates (Tables 4 and 5, respectively), the effect of spe- cies and substrate was analysed by a mixed type ANOVA. p < 0.05 was considered statistically significant. The raw data are given in the S2 Table. Supporting information S1 Table. Maximal respiration rate, VO2_max, and maximal ADP-phosphorylation rate, VADP_max, normalized to the protein content of the cell suspensions, and VO2_max, normal- ized to the cytochrome aa3 content. The maximal respiration rate, VO2_max, was recorded in the presence of either GM or GMPS, and 2 mM ADP (see representative recording in Fig 3A). The rate was normalized to the protein content and, for rat cardiomyocytes only, the cyto- chrome aa3 content. For statistical purposes, only the results from CK recordings are given. The corresponding maximal ADP-phosphorylation rate, VADP_max, was calculated assuming P/O2 ratios of 6 and 4 for GM and GMPS, respectively (see main text). Values from 8 mice and 7 rats are shown as mean ± SEM. * denotes p < 0.05, ** p < 0.01, *** p < 0.001, **** p < 0.0001, significant effect of substrate, species, or interaction between substrates and spe- cies. (DOCX) S2 Table. Raw data from the experiments. (ODS) S1 Raw images. (PDF) Author Contributions Conceptualization: Rikke Birkedal. Formal analysis: Jelena Branovets, Marko Vendelin, Rikke Birkedal. Funding acquisition: Marko Vendelin, Rikke Birkedal. Investigation: Jelena Branovets, Ka¨rol Soodla, Rikke Birkedal. Methodology: Jelena Branovets, Rikke Birkedal. Supervision: Rikke Birkedal. Visualization: Jelena Branovets, Marko Vendelin. Writing – original draft: Jelena Branovets, Rikke Birkedal. Writing – review & editing: Jelena Branovets, Marko Vendelin, Rikke Birkedal. References 1. Bessman SP, Carpenter CL. The creatine-creatine phosphate energy shuttle. AnnuRevBiochem. 1985; 54:831–62. https://doi.org/10.1146/annurev.bi.54.070185.004151 PMID: 3896131 2. Wallimann T, Wyss M, Brdiczka D, Nicolay K, Eppenberger HM. Intracellular compartmentation, struc- ture and function of creatine kinase isoenzymes in tissues with high and fluctuating energy demands: the “phosphocreatine circuit” for cellular energy homeostasis. BiochemJ. 1992 Jan; 281 (Pt 1):21–40. https://doi.org/10.1042/bj2810021 PMID: 1731757 3. Ingwall JS. ATP and the Heart: An Overview. In: Ingwall JS, editor. ATP and the Heart. Boston, MA: Springer US; 2002. p. 3–6. (Basic Science for the Cardiologist). https://doi.org/10.1007/978-1-4615- 1093-2_1 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 20 / 25 4. Bittl JA, Ingwall JS. Reaction rates of creatine kinase and ATP synthesis in the isolated rat heart. A 31P NMR magnetization transfer study. J Biol Chem. 1985 Mar; 260(6):3512–7. PMID: 3972835 5. Katz LA, Swain JA, Portman MA, Balaban RS. Relation between phosphate metabolites and oxygen consumption of heart in vivo. Am J Physiol. 1989; 256(1 Pt 2):H265–74. https://doi.org/10.1152/ ajpheart.1989.256.1.H265 PMID: 2912189 6. Vendelin M, Lemba M, Saks VA. Analysis of Functional Coupling: Mitochondrial Creatine Kinase and Adenine Nucleotide Translocase. Biophys J. 2004 Jul; 87(1):696–713. https://doi.org/10.1529/biophysj. 103.036210 PMID: 15240503 7. Gellerich FN, Schlame M, Bohnensack R, Kunz W. Dynamic compartmentation of adenine nucleotides in the mitochondrial intermembrane space of rat-heart mitochondria. Biochimica et Biophysica Acta (BBA)—Bioenergetics. 1987 Feb 11; 890(2):117–26. https://doi.org/10.1016/0005-2728(87)90012-0 PMID: 3801462 8. Saks VA, Khuchua ZA, Vasilyeva EV, Belikova OY, Kuznetsov AV. Metabolic compartmentation and substrate channelling in muscle cells. Role of coupled creatine kinases in in vivo regulation of cellular respiration–a synthesis. MolCell Biochem. 1994 Apr; 133–134:155–92. https://doi.org/10.1007/ BF01267954 PMID: 7808453 9. Dzeja PP, Terzic A. Phosphotransfer networks and cellular energetics. JExpBiol. 2003 Jun; 206(Pt 12):2039–47. https://doi.org/10.1242/jeb.00426 PMID: 12756286 10. Joubert F, Mazet JL, Mateo P, Hoerter JA. 31P {NMR} detection of subcellular creatine kinase fluxes in the perfused rat heart: contractility modifies energy transfer pathways. J Biol Chem. 2002 May 24; 277 (21):18469–76. https://doi.org/10.1074/jbc.M200792200 PMID: 11886866 11. Lygate CA, Neubauer S. The Myocardial Creatine Kinase System in the Normal, Ischaemic and Failing Heart. In: Lopaschuk GD, Dhalla NS, editors. Cardiac Energy Metabolism in Health and Disease. New York, NY: Springer; 2014. p. 155–68. (Advances in Biochemistry in Health and Disease). https://doi. org/10.1007/978-1-4939-1227-8_10 12. Hove M ten Lygate CA, Fischer A Schneider JE, Sang AE, Hulbert K, et al. Reduced inotropic reserve and increased susceptibility to cardiac ischemia/reperfusion injury in phosphocreatine-deficient guanidi- noacetate-N-methyltransferase-knockout mice. Circulation. 2005 May; 111(19):2477–85. https://doi. org/10.1161/01.CIR.0000165147.99592.01 PMID: 15883212 13. Nahrendorf M, Streif JU, Hiller KH, Hu K, Nordbeck P, Ritter O, et al. Multimodal functional cardiac MRI in creatine kinase-deficient mice reveals subtle abnormalities in myocardial perfusion and mechanics. Am J Physiol Heart Circ Physiol. 2006 Jun 1; 290(6):H2516–21. https://doi.org/10.1152/ajpheart. 01038.2005 PMID: 16415075 14. Faller KME, Atzler D, McAndrew DJ, Zervou S, Whittington HJ, Simon JN, et al. Impaired cardiac con- tractile function in arginine:glycine amidinotransferase knockout mice devoid of creatine is rescued by homoarginine but not creatine. Cardiovasc Res. 2018 Mar 1; 114(3):417–30. https://doi.org/10.1093/ cvr/cvx242 PMID: 29236952 15. Branovets J, Sepp M, Kotlyarova S, Jepihhina N, Sokolova N, Aksentijevic D, et al. Unchanged mito- chondrial organization and compartmentation of high-energy phosphates in creatine deficient GAMT-/- mouse heart. Am J Physiol Heart Circ Physiol. 2013 Jun 21; Available from: http://ajpheart.physiology. org/content/early/2013/06/17/ajpheart.00919.2012 16. Aksentijević D, Zervou S, Eykyn TR, McAndrew DJ, Wallis J, Schneider JE, et al. Age-Dependent Decline in Cardiac Function in Guanidinoacetate-N-Methyltransferase Knockout Mice. Front Physiol. 2020 Jan 21; 10. Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6985570/ PMID: 32038270 17. Lygate CA, Aksentijevic D, Dawson D, Hove M ten, Phillips D, de Bono JP, et al. Living Without Creatine Unchanged Exercise Capacity and Response to Chronic Myocardial Infarction in Creatine-Deficient Mice. Circulation Research. 2013 Mar 15; 112(6):945–55. https://doi.org/10.1161/CIRCRESAHA.112. 300725 PMID: 23325497 18. Nahrendorf M, Spindler M, Hu K, Bauer L, Ritter O, Nordbeck P, et al. Creatine kinase knockout mice show left ventricular hypertrophy and dilatation, but unaltered remodeling post-myocardial infarction. Cardiovascular Research. 2005 Feb 1; 65(2):419–27. https://doi.org/10.1016/j.cardiores.2004.10.006 PMID: 15639481 19. Choe C un, Nabuurs C, Stockebrand MC, Neu A, Nunes P, Morellini F, et al. l-arginine:glycine amidino- transferase deficiency protects from metabolic syndrome. Human Molecular Genetics. 2013 Jan 1; 22 (1):110–23. https://doi.org/10.1093/hmg/dds407 PMID: 23026748 20. Schmidt A, Marescau B, Boehm EA, Renema WKJ, Peco R, Das A, et al. Severely altered guanidino compound levels, disturbed body weight homeostasis and impaired fertility in a mouse model of guanidi- noacetate N-methyltransferase (GAMT) deficiency. Hum Mol Genet. 2004 May 1; 13(9):905–21. https://doi.org/10.1093/hmg/ddh112 PMID: 15028668 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 21 / 25 21. Branovets J, Karro N, Barsunova K, Laasmaa M, Lygate CA, Vendelin M, et al. Cardiac expression and location of hexokinase changes in a mouse model of pure creatine deficiency. American Journal of Physiology-Heart and Circulatory Physiology. 2020 Dec 18; 320(2):H613–29. https://doi.org/10.1152/ ajpheart.00188.2020 PMID: 33337958 22. Guzun R, Timohhina N, Tepp K, Monge C, Kaambre T, Sikk P, et al. Regulation of respiration controlled by mitochondrial creatine kinase in permeabilized cardiac cells in situ: Importance of system level prop- erties. Biochimica et Biophysica Acta (BBA)—Bioenergetics. 2009 Sep 1; 1787(9):1089–105. 23. Tepp K, Shevchuk I, Chekulayev V, Timohhina N, Kuznetsov AV, Guzun R, et al. High efficiency of energy flux controls within mitochondrial interactosome in cardiac intracellular energetic units. Biochi- mica et Biophysica Acta (BBA)—Bioenergetics. 2011 Dec; 1807(12):1549–61. https://doi.org/10.1016/j. bbabio.2011.08.005 PMID: 21872567 24. Laterveer FD, Nicolay K, Gellerich FN. Experimental evidence for dynamic compartmentation of ADP at the mitochondrial periphery: coupling of mitochondrial adenylate kinase and mitochondrial hexokinase with oxidative phosphorylation under conditions mimicking the intracellular colloid osmotic pressure. Mol Cell Biochem. 1997 Sep; 174(1–2):43–51. PMID: 9309664 25. Dzeja PP, Zeleznikar RJ, Goldberg ND. Adenylate kinase: kinetic behavior in intact cells indicates it is integral to multiple cellular processes. Mol Cell Biochem. 1998 Jul; 184(1–2):169–82. PMID: 9746320 26. Zeleznikar RJ, Goldberg ND. Adenylate Kinase-catalyzed Phosphoryl Transfer Couples ATP Utilization with Its Generation by Glycolysis in Intact Muscle. J Biol Chem. 1995 Mar 31; 270(13):7311–9. https:// doi.org/10.1074/jbc.270.13.7311 PMID: 7706272 27. Gellerich FN. The role of adenylate kinase in dynamic compartmentation of adenine nucleotides in the mitochondrial intermembrane space. FEBS Lett. 1992 Feb 3; 297(1–2):55–8. https://doi.org/10.1016/ 0014-5793(92)80326-c PMID: 1551437 28. Dzeja P, Terzic A. Adenylate Kinase and AMP Signaling Networks: Metabolic Monitoring, Signal Com- munication and Body Energy Sensing. Int J Mol Sci. 2009 Apr 17; 10(4):1729–72. https://doi.org/10. 3390/ijms10041729 PMID: 19468337 29. Aksentijević D, Lygate CA, Makinen K, Zervou S, Sebag-Montefiore L, Medway D, et al. High- energy phosphotransfer in the failing mouse heart: role of adenylate kinase and glycolytic enzymes. Eur J Heart Fail. 2010 Dec; 12(12):1282–9. https://doi.org/10.1093/eurjhf/hfq174 PMID: 20940173 30. Dzeja PP, Vitkevicius KT, Redfield MM, Burnett JC, Terzic A. Adenylate kinase-catalyzed phospho- transfer in the myocardium: Increased contribution in heart failure. Circ Res. 1999 May 28; 84 (10):1137–43. https://doi.org/10.1161/01.res.84.10.1137 PMID: 10347088 31. Jacobus WE, Lehninger AL. Creatine kinase of rat heart mitochondria. Coupling of creatine phosphory- lation to electron transport. J Biol Chem. 1973; 248(13):4803–10. PMID: 4718746 32. Noma T. Dynamics of nucleotide metabolism as a supporter of life phenomena. The Journal of Medical Investigation. 2005; 52(3,4):127–36. https://doi.org/10.2152/jmi.52.127 PMID: 16167529 33. Pucar D, Bast P, Gumina RJ, Lim L, Drahl C, Juranic N, et al. Adenylate kinase AK1 knockout heart: energetics and functional performance under ischemia-reperfusion. Am J Physiol Heart Circ Physiol. 2002 Aug 1; 283(2):H776–782. https://doi.org/10.1152/ajpheart.00116.2002 PMID: 12124227 34. Larsen S, Nielsen J, Hansen CN, Nielsen LB, Wibrand F, Stride N, et al. Biomarkers of mitochondrial content in skeletal muscle of healthy young human subjects. J Physiol. 2012 Jul 15; 590(Pt 14):3349– 60. https://doi.org/10.1113/jphysiol.2012.230185 PMID: 22586215 35. Karro N, Laasmaa M, Vendelin M, Birkedal R. Respiration of permeabilized cardiomyocytes from mice: no sex differences, but substrate-dependent changes in the apparent ADP-affinity. Sci Rep. 2019 Aug 29; 9(1):12592. https://doi.org/10.1038/s41598-019-48964-x PMID: 31467353 36. Lemieux H, Blier PU, Gnaiger E. Remodeling pathway control of mitochondrial respiratory capacity by temperature in mouse heart: electron flow through the Q-junction in permeabilized fibers. Scien- tific Reports. 2017 Jun 6; 7(1):2840. https://doi.org/10.1038/s41598-017-02789-8 PMID: 28588260 37. Gnaiger E. Mitochondrial pathways and respiratory control. An introduction to OXPHOS analysis. 4th ed. Innsbruck, Austria: Steiger Druck GmbH, Axams, Austria; 2014. 81 p. www.bioblast.at/index.php/ Gnaiger_2014_MitoPathways 38. Zhou P, Pu WT. Recounting cardiac cellular composition. Circ Res. 2016 Feb 5; 118(3):368–70. https:// doi.org/10.1161/CIRCRESAHA.116.308139 PMID: 26846633 39. Birkedal R, Laasmaa M, Branovets J, Vendelin M. Ontogeny of cardiomyocytes: ultrastructure optimiza- tion to meet the demand for tight communication in excitation–contraction coupling and energy transfer. Philosophical Transactions of the Royal Society B. 2022 Nov 21; https://royalsocietypublishing.org/doi/ 10.1098/rstb.2021.0321 PMID: 36189816 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 22 / 25 40. Blouin A, Bolender RP, Weibel ER. Distribution of organelles and membranes between hepatocytes and nonhepatocytes in the rat liver parenchyma. A stereological study. J Cell Biol. 1977 Feb; 72 (2):441–55. https://doi.org/10.1083/jcb.72.2.441 PMID: 833203 41. Davidson SM. Endothelial mitochondria and heart disease. Cardiovascular Research. 2010 Oct 1; 88 (1):58–66. https://doi.org/10.1093/cvr/cvq195 PMID: 20558442 42. Brand MD, Chien LF, Ainscow EK, Rolfe DF, Porter RK. The causes and functions of mitochondrial pro- ton leak. BiochimBiophysActa. 1994 Aug; 1187(2):132–9. https://doi.org/10.1016/0005-2728(94) 90099-x PMID: 8075107 43. Nicholls DG, Ferguson, S. J. Bioenergetics, 4th edition. 4th ed. Elsevier; 2013. 434 p. 44. CHANCE B, WILLIAMS GR. Respiratory enzymes in oxidative phosphorylation. I. Kinetics of oxygen utilization. JBiolChem. 1955 Nov; 217(1):383–93. PMID: 13271402 45. Momken I, Lechêne P, Koulmann N, Fortin D, Mateo P, Doan BT, et al. Impaired voluntary running capacity of creatine kinase-deficient mice. J Physiol (Lond). 2005 Jun 15; 565(3):951–64. https://doi. org/10.1113/jphysiol.2005.086397 PMID: 15831533 46. De Sousa E, Veksler V, Minajeva A, Kaasik A, Mateo P, Mayoux E, et al. Subcellular creatine kinase alterations. Implications in heart failure. CircRes. 1999 Jul; 85(1):68–76. https://doi.org/10.1161/01.res. 85.1.68 PMID: 10400912 47. Kaasik A, Veksler V, Boehm E, Novotova M, Minajeva A, Ventura-Clapier R. Energetic crosstalk between organelles: architectural integration of energy production and utilization. Circ Res. 2001; 89 (2):153–9. https://doi.org/10.1161/hh1401.093440 PMID: 11463722 48. Veksler VI, Kuznetsov AV, Anflous K, Mateo P, van Deursen J, Wieringa B, et al. Muscle Creatine Kinase-deficient Mice II. CARDIAC AND SKELETAL MUSCLES EXHIBIT TISSUE-SPECIFIC ADAPTATION OF THE MITOCHONDRIAL FUNCTION. J Biol Chem. 1995 Aug 25; 270 (34):19921–9. 49. Boehm E, Chan S, Monfared M, Wallimann T, Clarke K, Neubauer S. Creatine transporter activity and content in the rat heart supplemented by and depleted of creatine. American Journal of Physiology- Endocrinology and Metabolism. 2003 Feb 1; 284(2):E399–406. https://doi.org/10.1152/ajpendo.00259. 2002 PMID: 12531746 50. Matsushima K, Uda K, Ishida K, Kokufuta C, Iwasaki N, Suzuki T. Comparison of kinetic constants of creatine kinase isoforms. International Journal of Biological Macromolecules. 2006 Mar 30; 38(2):83–8. https://doi.org/10.1016/j.ijbiomac.2005.12.023 PMID: 16451808 51. Wallis J, Lygate CA, Fischer A, ten Hove M, Schneider JE, Sebag-Montefiore L, et al. Supranormal myocardial creatine and phosphocreatine concentrations lead to cardiac hypertrophy and heart failure: insights from creatine transporter-overexpressing transgenic mice. Circulation. 2005 Nov 15; 112 (20):3131–9. https://doi.org/10.1161/CIRCULATIONAHA.105.572990 PMID: 16286605 52. Stolz M, Wallimann T. Myofibrillar interaction of cytosolic creatine kinase (CK) isoenzymes: allocation of N-terminal binding epitope in MM-CK and BB-CK. JCell Sci. 1998 May; 111 (Pt 9):1207–16. https://doi. org/10.1242/jcs.111.9.1207 PMID: 9547297 53. Hornemann T, Stolz M, Wallimann T. Isoenzyme-specific interaction of muscle-type creatine kinase with the sarcomeric M-line is mediated by NH(2)-terminal lysine charge-clamps. JCell Biol. 2000 Jun; 149(6):1225–34. https://doi.org/10.1083/jcb.149.6.1225 PMID: 10851020 54. Wallimann T, Turner DC, Eppenberger HM. Localization of creatine kinase isoenzymes in myofibrils. I. Chicken skeletal muscle. J Cell Biol. 1977 Nov; 75(2 Pt 1):297–317. https://doi.org/10.1083/jcb.75.2. 297 PMID: 264112 55. Wallimann T, Kuhn HJ, Pelloni G, Turner DC, Eppenberger HM. Localization of creatine kinase isoen- zymes in myofibrils. II. Chicken heart muscle. J Cell Biol. 1977 Nov; 75(2 Pt 1):318–25. https://doi.org/ 10.1083/jcb.75.2.318 PMID: 264113 56. Kraft T, Hornemann T, Stolz M, Nier V, Wallimann T. Coupling of creatine kinase to glycolytic enzymes at the sarcomeric I-band of skeletal muscle: a biochemical study in situ. J Muscle Res Cell Motil. 2000; 21(7):691–703. https://doi.org/10.1023/a:1005623002979 PMID: 11227796 57. Grosse R, Spitzer E, Kupriyanov VV, Saks VA, Repke KR. Coordinate interplay between (Na+ + K +)-ATPase and creatine phosphokinase optimizes (Na+/K+)-antiport across the membrane of vesicles formed from the plasma membrane of cardiac muscle cell. BiochimBiophysActa. 1980 Dec; 603 (1):142–56. 58. Rossi AM, Eppenberger HM, Volpe P, Cotrufo R, Wallimann T. Muscle-type MM creatine kinase is spe- cifically bound to sarcoplasmic reticulum and can support Ca2+ uptake and regulate local ATP/ADP ratios. J Biol Chem. 1990 Mar 25; 265(9):5258–66. PMID: 2318892 59. Schlattner U, Klaus A, Ramirez Rios S, Guzun R, Kay L, Tokarska-Schlattner M. Cellular compartmen- tation of energy metabolism: creatine kinase microcompartments and recruitment of B-type creatine PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 23 / 25 kinase to specific subcellular sites. Amino Acids. 2016 Aug 1; 48(8):1751–74. https://doi.org/10.1007/ s00726-016-2267-3 PMID: 27318991 60. Ventura-Clapier R, Kuznetsov A, Veksler V, Boehm E, Anflous K. Functional coupling of creatine kinases in muscles: species and tissue specificity. Mol Cell Biochem. 1998; 184(1–2):231–47. PMID: 9746324 61. Gellerich F, Saks VA. Control of heart mitochondrial oxygen consumption by creatine kinase: the impor- tance of enzyme localization. Biochem Biophys Res Commun. 1982 Apr; 105(4):1473–81. https://doi. org/10.1016/0006-291x(82)90954-8 PMID: 7103968 62. Simson P, Jepihhina N, Laasmaa M, Peterson P, Birkedal R, Vendelin M. Restricted ADP movement in cardiomyocytes: Cytosolic diffusion obstacles are complemented with a small number of open mito- chondrial voltage-dependent anion channels. Journal of Molecular and Cellular Cardiology. 2016 Aug; 97:197–203. https://doi.org/10.1016/j.yjmcc.2016.04.012 PMID: 27261153 63. Boehm E. Maintained Coupling of Oxidative Phosphorylation to Creatine Kinase Activity in Sarcomeric Mitochondrial Creatine Kinase-deficient Mice. Journal of Molecular and Cellular Cardiology. 1998 May; 30(5):901–12. https://doi.org/10.1006/jmcc.1998.0692 PMID: 9618231 64. Ramay HR, Vendelin M. Diffusion Restrictions Surrounding Mitochondria: A Mathematical Model of Heart Muscle Fibers. Biophysical Journal. 2009 Jul 22; 97(2):443–52. https://doi.org/10.1016/j.bpj. 2009.04.062 PMID: 19619458 65. Illaste A, Laasmaa M, Peterson P, Vendelin M. Analysis of Molecular Movement Reveals Latticelike Obstructions to Diffusion in Heart Muscle Cells. Biophysical Journal. 2012 Feb 22; 102(4):739–48. https://doi.org/10.1016/j.bpj.2012.01.012 PMID: 22385844 66. Birkedal R, Laasmaa M, Vendelin M. The location of energetic compartments affects energetic commu- nication in cardiomyocytes. Front Physiol. 2014; 5: 376. https://doi.org/10.3389/fphys.2014.00376 PMID: 25324784 67. Jepihhina N, Beraud N, Sepp M, Birkedal R, Vendelin M. Permeabilized Rat Cardiomyocyte Response Demonstrates Intracellular Origin of Diffusion Obstacles. Biophysical Journal. 2011 Nov 2; 101 (9):2112–21. https://doi.org/10.1016/j.bpj.2011.09.025 PMID: 22067148 68. Timohhina N, Guzun R, Tepp K, Monge C, Varikmaa M, Vija H, et al. Direct measurement of energy fluxes from mitochondria into cytoplasm in permeabilized cardiac cells in situ: some evidence for Mito- chondrial Interactosome. J Bioenerg Biomembr. 2009 Jun; 41(3):259–75. https://doi.org/10.1007/ s10863-009-9224-8 PMID: 19597977 69. Vendelin M, Hoerter JA, Mateo P, Soboll S, Gillet B, Mazet JL. Modulation of Energy Transfer Pathways between Mitochondria and Myofibrils by Changes in Performance of Perfused Heart. J Biol Chem. 2010 Nov 26; 285(48):37240–50. https://doi.org/10.1074/jbc.M110.147116 PMID: 20847056 70. Noda L. ADENOSINE TRIPHOSPHATE-ADENOSINE MONOPHOSPHATE TRANSPHOSPHORY- LASE: III. KINETIC STUDIES. Journal of Biological Chemistry. 1958 May 1; 232(1):237–50. 71. Mika D, Leroy J, Vandecasteele G, Fischmeister R. PDEs create local domains of cAMP signaling. Journal of Molecular and Cellular Cardiology. 2012 Feb; 52(2):323–9. https://doi.org/10.1016/j.yjmcc. 2011.08.016 PMID: 21888909 72. Weber CR, Piacentino V, Ginsburg KS, Houser SR, Bers DM. Na-Ca(2+) exchange current and sub- membrane [Ca(2+)] during the cardiac action potential. CircRes. 2002 Feb; 90(2):182–9. 73. Franzini-Armstrong C. ER-mitochondria communication. How privileged? Physiology (Bethesda). 2007 Aug; 22: 261–8. 74. Williams GSB, Boyman L, Chikando AC, Khairallah RJ, Lederer WJ. Mitochondrial calcium uptake. PNAS. 2013 Jun 25; 110(26):10479–86. https://doi.org/10.1073/pnas.1300410110 PMID: 23759742 75. Hardie DG. AMP-activated protein kinase: a key regulator of energy balance with many roles in human disease. J Intern Med. 2014 Dec; 276(6):543–59. https://doi.org/10.1111/joim.12268 PMID: 24824502 76. Sepp M, Vendelin M, Vija H, Birkedal R. ADP Compartmentation Analysis Reveals Coupling between Pyruvate Kinase and ATPases in Heart Muscle. Biophysical Journal. 2010 Jun; 98: 2785–93. https:// doi.org/10.1016/j.bpj.2010.03.025 PMID: 20550890 77. Itah R, Gitelman I, Davis C. A replacement for methoxyflurane (Metofane) in open-circuit anaesthesia. Lab Anim. 2004 Jul; 38(3):280–5. https://doi.org/10.1258/002367704323133664 PMID: 15207039 78. Lange Y. Disposition of intracellular cholesterol in human fibroblasts. Journal of Lipid Research. 1991 Feb 1; 32(2):329–39. PMID: 2066666 79. Saks VA, Veksler VI, Kuznetsov AV, Kay L, Sikk P, Tiivel T, et al. Permeabilized cell and skinned fiber techniques in studies of mitochondrial function in vivo. Mol Cell Biochem. 1998 Jul; 184(1–2):81–100. PMID: 9746314 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 24 / 25 80. Barsunova K, Vendelin M, Birkedal R. Marker enzyme activities in hindleg from creatine-deficient AGAT and GAMT KO mice–differences between models, muscles, and sexes. Sci Rep. 2020 May 14; 10 (1):1–9. 81. Christensen M, Hartmund T, Gesser H. Creatine kinase, energy-rich phosphates and energy metabo- lism in heart muscle of different vertebrates. J Comp Physiol [B]. 1994; 164(2):118–23. https://doi.org/ 10.1007/BF00301652 PMID: 8056878 82. Eyer P, Worek F, Kiderlen D, Sinko G, Stuglin A, Simeon-Rudolf V, et al. Molar absorption coefficients for the reduced Ellman reagent: reassessment. Analytical Biochemistry. 2003 Jan 15; 312(2):224–7. https://doi.org/10.1016/s0003-2697(02)00506-7 PMID: 12531209 83. Vendelin M, Laasmaa M, Kalda M, Branovets J, Karro N, Barsunova K, et al. IOCBIO Kinetics: An open-source software solution for analysis of data traces. PLOS Computational Biology. 2020 Dec 22; 16(12):e1008475. https://doi.org/10.1371/journal.pcbi.1008475 PMID: 33351800 84. Karro N, Sepp M, Jugai S, Laasmaa M, Vendelin M, Birkedal R. Metabolic compartmentation in rainbow trout cardiomyocytes: coupling of hexokinase but not creatine kinase to mitochondrial respiration. J Comp Physiol B, Biochem Syst Environ Physiol. 2016 Aug 13; 85. Balaban RS, Mootha VK, Arai A. Spectroscopic Determination of Cytochrome c Oxidase Content in Tis- sues Containing Myoglobin or Hemoglobin. Analytical Biochemistry. 1996 Jun 1; 237(2):274–8. https:// doi.org/10.1006/abio.1996.0239 PMID: 8660576 86. Liao GL, Palmer G. The reduced minus oxidized difference spectra of cytochromes a and a3. Biochi- mica et Biophysica Acta (BBA)—Bioenergetics. 1996 Jun 13; 1274(3):109–11. https://doi.org/10.1016/ 0005-2728(96)00014-x PMID: 8664303 PLOS ONE Creatine kinase compartmentalization in rat and mouse cardiomyocytes PLOS ONE | https://doi.org/10.1371/journal.pone.0294718 November 27, 2023 25 / 25

Rat and mouse cardiomyocytes show subtle differences in creatine kinase expression and compartmentalization.

11-27-2023

Branovets, Jelena,Soodla, Kärol,Vendelin, Marko,Birkedal, Rikke

eng

PMC10293173

1 Vol.:(0123456789) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports Estimation of maximal lactate steady state using the sweat lactate sensor Yuki Muramoto 1, Daisuke Nakashima 2, Tsubasa Amano 3, Tomota Harita 3, Kazuhisa Sugai 4, Kyohei Daigo 5, Yuji Iwasawa 5, Genki Ichihara 5, Hiroki Okawara 2, Tomonori Sawada 2, Akira Kinoda 1, Yuichi Yamada 1, Takeshi Kimura 1, Kazuki Sato 1 & Yoshinori Katsumata 1,5* A simple, non-invasive algorithm for maximal lactate steady state (MLSS) assessment has not been developed. We examined whether MLSS can be estimated from the sweat lactate threshold (sLT) using a novel sweat lactate sensor for healthy adults, with consideration of their exercise habits. Fifteen adults representing diverse fitness levels were recruited. Participants with/without exercise habits were defined as trained/untrained, respectively. Constant-load testing for 30 min at 110%, 115%, 120%, and 125% of sLT intensity was performed to determine MLSS. The tissue oxygenation index (TOI) of the thigh was also monitored. MLSS was not fully estimated from sLT, with 110%, 115%, 120%, and 125% of sLT in one, four, three, and seven participants, respectively. The MLSS based on sLT was higher in the trained group as compared to the untrained group. A total of 80% of trained participants had an MLSS of 120% or higher, while 75% of untrained participants had an MLSS of 115% or lower based on sLT. Furthermore, compared to untrained participants, trained participants continued constant-load exercise even if their TOI decreased below the resting baseline (P < 0.01). MLSS was successfully estimated using sLT, with 120% or more in trained participants and 115% or less in untrained participants. This suggests that trained individuals can continue exercising despite decreases in oxygen saturation in lower extremity skeletal muscles. Exercise with appropriate frequency and intensity is paramount to maintaining good health in all generations and improving exercise performance in athletes. Maximal lactate steady state (MLSS) is the intensity at which constant-workload exercise can be performed for 40–60 min without lactate accumulation1,2. Above the MLSS intensity, blood lactate shows an identifiable increase during constant-workload exercise, with a concomitant decrease in oxygen saturation in the vastus lateralis in the thigh3. MLSS has been utilized as a measure of train- ing intensity in endurance sports, such as track and field4, cycling5, and swimming6,7. MLSS assessment requires several constant submaximal load tests performed on separate days and frequent blood lactate measurements during exercise, which are complicated and physically strenuous for the participants8. Therefore, the lactate threshold (LT) and blood lactate accumulation onset time (OBLA) are frequently used instead of MLSS as measures of training intensity8. More specifically, LT and OBLA are thought to reflect low9 and high10 intensity, respectively, relative to MLSS. However, there are limitations to the use of LT and OBLA for MLSS. Recently, several algorithms for MLSS estimation from LT or OBLA have been developed, mainly for use with athletes4,11,12. Despite being simple indices for determining training intensity, LT and OBLA require frequent blood lactate measurements and exercise cessation to collect blood samples. Since these methods are somewhat invasive, they are impractical, particularly for non-athletes or those with no exercise habits. To overcome this limitation, we developed a sweat lactate sensor for real-time, non-invasive measurement of sweat lactate. The sweat lactate threshold (sLT) is reportedly consistent with the anaerobic metabolic threshold13. Therefore, we expected that sLT could be utilized to estimate MLSS with minimal stress on participants. Additionally, this simple and non- invasive algorithm may be applicable to non-athletes as well as athletes. OPEN 1Institute for Integrated Sports Medicine, Keio University School of Medicine, Tokyo, Japan. 2Department of Orthopaedic Surgery, Keio University School of Medicine, Tokyo, Japan. 3Keio University School of Medicine, Tokyo, Japan. 4School of Veterinary Nursing and Technology, Faculty of Veterinary Science, Nippon Veterinary and Life Science University, Tokyo, Japan. 5Department of Cardiology, Keio University School of Medicine, Tokyo, Japan. *email: goodcentury21@keio.jp 2 Vol:.(1234567890) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports/ This study aimed to verify whether MLSS could be estimated from sLT using a sweat lactate sensor for healthy adults, with consideration of their exercise habits, and to investigate whether a decrease in oxygen saturation in the vastus lateralis is a factor contributing to the difference in MLSS based on sLT. Results In-vitro characterization of the lactate sensor under simulated sweat environments. Figure 1 showed the amperometric response of the lactate sensor to increasing lactate concentrations in 0, 2.5, 5, 10, and 20 mM. In our sensors, a significant difference was observed in sensor responses to several lactate solutions (2.5, 5, 10, and 20 mM) with different pH (5, 6, 7, and 8) and different temperatures (25, 31, and 36 °C) (Fig. 1A and Online Figs. 1–4). Secondly, a significant response in the sensor to 10 mM lactate solution was observed even in the presence of NaCl (10, 25, 50, 100 mM) and KCl (2.5, 5, 10 mM) (Fig. 1B, C). These findings indicated that the lactate values (current values) of sweat obtained from this sensor could show a significant enough difference to determine the inflection point under various sweat environments. Characteristics of participants. The fifteen participants (14 male, 1 female) had a mean age of 26 ± 6 years, mean height of 173 ± 9 cm, mean weight of 67 ± 10 kg, mean skeletal muscle mass of 52 ± 18 kg, mean body fat percentage of 18 ± 7%, mean peak VO2 of 41 ± 5 mL/min/kg, and mean HRmax of 175 ± 9 beats per min (bpm). Eleven participants engaged in regular exercise (Table 1). MLSS based on sLT. The sLT was correlated with the VT (r = 0.70) (Online Fig. 5). The Bland–Altman plot described no bias between the mean values (mean differences: −3.0 W, respectively) (Online Fig. 6). Constant- load exercises at 125%, 120%, 115%, and 110% of sLT load were performed in that order, and completed by 8, 10, 14, and 15 participants, respectively (Table 2, Online Fig. 7). Each result shows the values of participants who were able to perform 30 min of exercise. Blood lactate levels at the end of exercise at each load in the aforemen- tioned order were 6.5 ± 3.7 mM, 4.4 ± 1.1 mM, 4.1 ± 1.2 mM, and 3.7 ± 1.7 mM (Table 2, Fig. 2). VO2 values (% peak VO2) at the end of exercise were 36.3 ± 4.0 mL/min/kg (84.5 ± 7.2%), 32.8 ± 5.5 mL/min/kg (76.9 ± 8.0%), 30.7 ± 6.7 mL/min/kg (71.1 ± 11.1%), and 29.0 ± 6.6 mL/min/kg (68.8 ± 13.1%). HR values (% HRmax) at the end of exercise were 168.6 ± 14.1 bpm (84.8 ± 6.8%), 158.1 ± 14.8 bpm (79.9 ± 7.4%), 152.6 ± 16.2 bpm (77.5 ± 7.8%), and 144.4 ± 15.2 bpm (73.2% ± 7.3%). Among the participants who completed the constant-load exercise at 125% of sLT load, one had an increase in blood lactate > 1 mM at the end of the exercise (30 min). Therefore, MLSS was 125% of sLT in seven participants, 120% in three, 115% in four, and 110% in one, suggesting that estimating MLSS based on sLT was difficult (Table 2). MLSS data obtained from 15 participants were calculated (Table 3, 35 30 25 20 15 10 5 0 (C) KCL 0 2 4 Current (uA) Lactate + NaCL 6 Time (min) 10mM 50mM 25mM 100mM 25 20 15 10 5 0 PBS aCL N (B) NaCL Current (uA) 0 5 10 15 20 25 30 35 0 2.5 5 10 20 (mM) 36٦ 31٦ 25٦ (A) PBS 0 2 4 Lactate + KCL 6 Time (min) PBS KCL PBS 2.5mM 10mM 5.0mM Figure 1. In-vitro characterization of the lactate sensor under imitated sweat environments. (A) The graph shows the corresponding calibration plots of the sensor with pH 7 under different temperature (25, 31, and 36 °C) conditions. The interference study for individual lactate (B,C). The presence of non-target electrolytes; Na, K, and Cl cause negligible interference to the response of our lactate sensors. Applied voltage = 0.16 V versus Ag/AgCl. The data were obtained from three samples. 3 Vol.:(0123456789) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports/ Table 1. Characteristics of participants (mean ± standard deviation). P-values refer to the significance of differences between the trained and untrained groups. MLSS > 120%: number and percentage of participants whose MLSS was 120% or higher. bpm beats per min, HRmax maximal heart rate, peak VO2/kg peak oxygen uptake/weight, MLSS maximal lactate steady state, sLT sweat lactate threshold. *p < 0.05. Measures All (n = 15) Trained (n = 11) Untrained (n = 4) Mean difference p-value 95% CI Cohen’d Age (year) 26 ± 6 22 ± 4 33 ± 4 −11.3 0.01* −18.6 to −3.9 2.5 Height (cm) 173 ± 9 173 ± 10 172 ± 7 1.4 0.79 −10.0 to 12.7 0.14 Weight (kg) 67 ± 10 65 ± 8 72 ± 16 −7.6 0.43 −32.1 to 16.9 0.70 Skeletal muscle mass (kg) 52 ± 18 52 ± 8 50 ± 10 1.8 0.76 −12.9 to 16.6 0.21 Body fat percentage (%) 18 ± 7 15 ± 6 25 ± 4 −10.0 0.01* −16.6 to −3.2 1.59 PeakVO2/kg (mL/min/ kg) 41 ± 5 43 ± 3 36 ± 6 6.9 0.13 −3.5 to 17.2 1.72 HRmax (bmp) 175 ± 9 174 ± 8 177 ± 7 −2.3 0.61 −12.3 to 7.7 0.27 MLSS (% of sLT) 120 ± 5 122 ± 4 115 ± 4 7.3 0.03* 1.3 to 13.3 1.78 MLSS > 120% (n,%) 10.67% 9.82% 1.25% 0.03* Table 2. Exercise data of participants at each load (mean ± standard deviation). Exercise completion: number of participants who were able to achieve 30 min of constant-load exercise. MLSS: number of participants who were greatest load among the loads in which blood lactate values at the end of exercise (30 min) increased within 1 mM, compared to those at 10 min after exercise initiation. The value of “measures at the end of the exercise” is exercise completion. % HRmax heart rate/maximal heart rate, % peak VO2 oxygen uptake/ peak oxygen uptake, bpm beats per min, HR heart rate, MLSS maximal lactate steady state, sLT sweat lactate threshold, VO2/kg oxygen uptake/weight. Relative load based on sLT (%) 125% 120% 115% 110% Exercise completion (n, %) 8 (53%) 10 (66%) 14 (93%) 15 (100%) MLSS (n, %) 7 (47%) 3 (20%) 4 (27%) 1 (6%) Measures at the end of the exercise  Blood lactate (mM) 6.5 ± 4.4 4.4 ± 1.1 4.1 ± 1.2 3.6 ± 1.7   VO2/kg (mL/min/kg) 36.3 ± 4.0 32.8 ± 5.5 30.7 ± 6.7 29.0 ± 6.6  % peak VO2 (%) 84.5 ± 7.2 76.9 ± 8.0 71.1 ± 11.1 68.8 ± 13.1  HR (bpm) 168.6 ± 14.1 158.1 ± 14.8 152.6 ± 16.2 144.4 ± 15.2  % HRmax (%) 84.8 ± 6.8 79.9 ± 7.4 77.5 ± 7.8 73.2 ± 7.3 Figure 2. Blood lactate for participants who were able to exercise at each load. 4 Vol:.(1234567890) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports/ Fig. 3). Blood lactate, % peak VO2, and % HRmax at the end of exercise at the MLSS load in the full sample were 4.6 ± 1.3 mM, 77.2 ± 12.7%, and 83.3 ± 6.9%, respectively. MLSS based on sLT in participants with daily exercise. Next, the effect of daily exercise on MLSS based on sLT was investigated. The MLSS based on sLT in trained participants was higher than in untrained participants (Table 1). The MLSS for the trained group accounted for more than 120% of the sLT, while the untrained group accounted for less than 115% of the sLT (P = 0.03, φ = 0.55). These findings suggested that MLSS was 120% or more of sLT in regularly trained participants and 115% or less of sLT in untrained participants. To determine a physiological contributor to the difference in MLSS based on sLT between regularly trained and untrained participants, we investigated the relationship between blood lactate accumulation during constant loading tests and a decrease in oxygen saturation in intra-skeletal muscles. ΔTOI and blood lactate levels in participants with daily exercise. ΔTOI strongly correlated with blood lactate level at the end of exercise (trained: r = −0.7, untrained: r = −0.8, Fig. 4 and Online Fig. 8). The plot revealed that the constant-load exercise was discontinued in the untrained group only when ΔTOI was lower than the resting value (Fig. 4, red triangle). In contrast, in regularly trained participants, the exercise continued until up to a 15% decrease in ΔTOI, as compared to the resting value (Fig. 4, black circle). Moreover, a steeper increase in blood lactate was associated with a decrease in ΔTOI in untrained participants as compared to the trained group, suggesting that a slight decrease in ΔTOI immediately contributed to the increase in blood lactate (regression line: trained = −0.286, untrained = −0.479). Further, trained participants continued constant-load exercise even if their ΔTOI decreased (Fig. 5) (mean difference: −7.4, 95% confidence interval [CI]: −11.4 to −3.4, P < 0.01). The optimal cut-off value for completion of the constant-load exercise was estimated to occur at ΔTOI of −17% (sensitivity: 0.97, specificity: 1.00) and −2.6% (sensitivity: 0.88, specificity: 1.00) in trained and untrained participants, respectively, by the ROC curve analysis (Online Fig. 9). Table 3. Exercise data of participants at MLSS (mean ± standard deviation). % HRmax heart rate/maximal heart rate, % peak VO2 oxygen uptake/peak oxygen uptake, bpm beats per min, HR heart rate, MLSS maximal lactate steady state, VO2/kg oxygen uptake/weight. Measures at the end of exercise (n = 15)  Blood lactate (mM) 4.63 ± 1.21   VO2/kg (mL/min/kg) 32.6 ± 7.1  % peak VO2 (%) 77.2 ± 12.7  HR (bpm) 164.3 ± 15.4  % HRmax (%) 83.3 ± 6.9  125%:120%:115%:110% (n) 7:3:4:1 Figure 3. Blood lactate, heart rate, VO2/kg at the MLSS in all participants (n = 15). Blood lactate (red), heart rate (gray), and oxygen uptake-adjusted weight (blue) at the MLSS in each participant. MLSS maximal lactate steady state, VO2/kg oxygen uptake/weight. 5 Vol.:(0123456789) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports/ Figure 4. Correlation between change in tissue oxygenation index (TOI) and blood lactate. The black circle represents trained participants, who showed a good correlation between ΔTOI and blood lactate level (y = −0.2859x + 3.035, r = −0.7, P < 0.01). The red triangle represents untrained participants, who showed a good correlation between ΔTOI and blood lactate (y = −0.479x + 5.2349, r = −0.8, P < 0.01). A steeper increase in blood lactate level was associated with a decrease in ΔTOI in untrained participants as compared to trained participants. ΔTOI TOI (pre-post). Figure 5. Difference in ΔTOI between trained and untrained participants in the completed exercise. Trained participants continued constant-load exercise even if their ΔTOI decreased (mean difference: −7.4, 95% confidence interval: −11.4 to −3.4, P < 0.01). ΔTOI TOI (pre-post); *: P < 0.05. Figure 6. Flowchart of the study protocol. NIRS near-infrared spectrometer, VO2 oxygen uptake. 6 Vol:.(1234567890) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports/ Discussion This prospective study provided novel evidence of successful MLSS estimation via sLT calculation by a wearable and non-invasive sweat lactate sensor, with consideration of daily exercise. sLT can be determined independ- ent of the amount of sweating using a sweat lactate sensor on the upper arm13. The device also determines the inflection point but not the absolute sweat lactate value13–15. The most significant result was that MLSS approxi- mated 120–125% of sLT in regularly trained participants and 115% or less of sLT in untrained participants. The difference in the physiological response to the decrease in oxygen saturation in lower limb skeletal muscle may contribute to this relationship between MLSS and sLT. Determining MLSS requires multiple constant-load exercise tests. MLSS was defined as the greatest load among the loads in which blood lactate values at the end of exercise (30 min) increased within 1 mM, compared to those at 10 min after exercise initiation4,11,12 Therefore, methods have been developed to estimate MLSS using LT and OBLA, with MLSS of 124–127% of LT load4,11 and 90% of OBLA load12, mainly for athletes4,11,12. Additionally, the Functional Threshold Power (FTP) used by cyclists has been determined through several constant submaximal load tests performed on separate days as well as MLSS16. FTP is a non-invasive method for measuring training intensity, which correlates well with MLSS17. However, LT and OBLA require frequent blood lactate measurements and exercise cessation to collect blood samples. FTP is an index specific to cyclists that requires several constant submaximal load tests performed on separate days and frequent blood lactate measurements during exercise. Therefore, although exercise with appropriate dosage and intensity is essential for maintaining good health in all generations, MLSS measurements are impractical, particularly in non-athletes or those without exercise habits. A sweat sensor was developed to monitor sweat lactate values in real-time during progressive exercise in a clinical setting and for sports use. Our sensor is highly flexible and can be smoothly adjusted to curved surfaces using PET substrates. The upper arm and forehead are appropriate sites to monitor the lactate levels in sweat due to a high-sweat rate during exercise, smooth skin surfaces for sensor placement, and noninterference during pedaling tasks13–15. Especially in healthy subjects, the upper arm has been used because of its simplicity of attach- ment and minimal interference. sLT defined as the first significant increase in sweat lactate concentration above baseline based on graphical plots, is consistent with LT calculated from blood samples and ventilatory threshold assessed with exhaled gas analysis13. In this study, MLSS was successfully estimated via sLT, with 120–125% of sLT in regularly trained participants and 115% or less in untrained participants. Blood lactate, % peak VO2, and % HRmax at the end of exercise at MLSS load were consistent with data from previous reports1,2,8,17,18. Report- edly, 124–127% of blood LT intensity at the running speed was the MLSS intensity in track and field athletes or cyclists4,11. These previous findings were consistent with MLSS load based on sLT in participants who regularly exercised. In untrained participants, MLSS approximated 115% or less of sLT. Assessment of appropriate exercise dosage and intensity should be further targeted for the well-being of non-athletes. MLSS, estimated in a simple and non-invasive manner using a sweat lactate sensor, could be used for health maintenance in non-athletes. To determine a physiological contributor to the difference in MLSS based on sLT between regularly trained and untrained participants, we investigated the relationship between blood lactate accumulation during constant loading tests and a decrease in oxygen saturation in intra-skeletal muscles. The constant-load exercise was com- pleted for 30 min in trained participants without blood lactate accumulation, even with substantial decreases in oxygen saturation in lower limb skeletal muscles. This finding suggests that training enables constant-load exercise for long periods, even at loads relatively greater than an anaerobic threshold, at which oxygen saturation in intra-skeletal muscles can be preserved. In contrast, a steeper increase in blood lactate was associated with a decrease in ΔTOI in the untrained group as compared to the trained group, suggesting that a slight decrease in ΔTOI immediately contributes to blood lactate accumulation. Exercise tolerance improves through biological responses, such as increased blood flow in skeletal muscles19, improved mitochondrial function20, and a shift from IIb to IIa in skeletal muscle subsets21. These biological responses are induced by the activation of hypoxic response signals following oxygen saturation reduction in skeletal muscles during exercise22–27. Therefore, the extent and variability of oxygen saturation reduction during exercise may be related to training effectiveness. Training results in the acquisition of hypoxic tolerance in skeletal muscles, causing increases in exercise endur- ance and enabling exercise with stronger intensity. Positive feedback between the decrease in oxygen saturation in skeletal muscles and improvement in exercise tolerance could maximize training benefits. Limitations. Our findings should be interpreted with consideration of the following limitations. First, because of the observational study design, we cannot exclude the influence of selection bias. Second, our study included a relatively small number of cases, particularly for the untrained group, and primarily healthy college- age male individuals. Further research should include untrained participants and women. Third, constant-load exercises at 130% of sLT load were not performed in this study. Finally, there was a possibility of non-response in the sweat lactate sensor owing to a lack of sweat during exercise. Particularly, older adults and women sweat less28. Therefore, in such cases, adjusting exercise parameters to promote sweating is necessary. However, sLT could be clearly determined in all participants in this study. Conclusions By dividing the participants into trained and untrained groups, MLSS was successfully estimated using sLT, with 120% or more of the sLT load in trained participants and 115% or less in untrained participants. This finding may involve the ability of an individual to continue exercising despite a decrease in oxygen saturation in the lower extremity skeletal muscles. This novel actualized measurement of sLT is expected to enable non-invasive MLSS estimation. This simple and non-invasive algorithm can be used as a convenient indicator of good health maintenance for non-athletes and a potential guide for training athletes. 7 Vol.:(0123456789) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports/ Methods Participants. Fifteen healthy adults representing a broad spectrum of fitness levels, regardless of exercise habits, were recruited between May and September 2022. Participants with/without exercise habits were defined as “trained”, and “untrained,” respectively. Exercise habit was defined as > 75 min per week of exercise at vigor- ous intensity29. The inclusion criteria were as follows: no underlying or pre-existing cardiovascular, respiratory, or metabolic diseases; no athletic injuries; non-smokers; and no dietary supplements or medication habits of any type. The study protocol was approved by the Institutional Review Board of the Keio University School of Medicine (approval number: 20190229) and conducted in accordance with the principles of the Declaration of Helsinki. All participants provided informed consent because the Institutional Review Board approved the use of oral consent, in accordance with the Japanese guidelines for clinical research. Experimental procedure A flowchart of the study protocol is shown in Fig. 6. First, the Ramp stress test was performed using an electro- magnetically braked ergometer (StrengthErgo8 V2; Fukuda Denshi Co., Ltd., Tokyo, Japan) with a sweat lactate sensor (Grace Imaging Inc., Tokyo, Japan), an exhaled gas analyzer (Aeromonitor AE-301S; Minato Medical Science Co., Ltd., Osaka, Japan), and a heart rate (HR) monitor (POLAR H10 N; Polar Electro Japan, Tokyo, Japan). Subsequently, constant-load exercise was performed for 30 min at 125%, 120%, 115%, and 110% of sLT intensity in this order. An electromagnetically-braked ergometer was used during the exercise to determine MLSS12. At least 24 h were allowed between each test (mean: 7.0 ± 2.9 days)5. During constant-load exercise, an exhaled gas analyzer, HR monitor, and near-infrared spectroscopy (NIRS) monitor (NIRO-200NX; Hamamatsu Photonics K.K., Hamamatsu, Japan) were attached. Blood lactate values were obtained via auricular pricking and gentle squeezing of the ear lobe using a blood lactate analyzer (Lactate Pro 2, ARKRAY Inc., Kyoto, Japan). Blood lactate levels were measured before exercise and every 5 min during exercise. Exercise test protocol. Participants avoided caffeine and alcohol consumption, which would cause fatigue, the day before testing. After measuring resting data for 2 min, participants performed a warm-up exercise for 2 min at a 50-W load and then exercised at increasing intensities until they could no longer maintain the pedal- ing rate (volitional exhaustion). The resistance was increased in 25-W increments from 50-W at 1-min intervals. Rotational speed was maintained at 70 rotations per min (rpm). sLT determination. A sweat lactate sensor quantifies sweat lactate concentration as a value of current because it reacts with sweat lactate and generates an electric current. The value of current can be obtained as continuous data within 0.1–80 μA in 0.1-μA increments13. Further, we investigated whether the lactate values (current values) of sweat obtained from this sensor could show a relative difference significant enough to deter- mine this inflection point under various sweat environments (pH, temperature, and ionic conductivity) with several solutions that were close in composition to actual sweat. Regarding the pH and temperature of human sweat, it has been reported that sweat has a pH of 5–7 and a skin temperature of 25–37 °C30–32. Therefore, the electrochemical characterization of the lactate sensor chip was performed using L-lactic acid solutions in 0, 2.5, 5, 10, and 20 mM prepared in 0.1 mol/L phosphate buffer solution (PBS) under different temperatures (25, 31, 36 °C) and pH (5, 6, 7, and 8). Then, the three lactate sensor tips were evaluated in each condition using chronoamperometry at an applied voltage of 0.16 V (versus Ag/AgCl). Next, the major electrolytes in sweat are Na, K, and Cl. Generally, NaCl varies from 10 to 90 mM and KCl from 2 to 8 mM during exercise30. Therefore, the sensor evaluated a significant response to l-lactic acid solution in 10 mM even in the presence of NaCl (10, 25, 50, 100 mM) and KCl (2.5, 5, 10 mM). After calibration using saline for 2 or 3 min, the sensor chip connected to the sensor device was attached to the superior right upper limb of the participant13,14, which was cleaned with an alcohol-free cloth. The upper arm has a high-sweat rate during physical excursions33. In addition, it is a site that does not interfere with exercise during pedaling tasks. Additionally, data were recorded at a 1-Hz sampling frequency for mobile applications with a Bluetooth connection. Recorded data were converted to moving average values over 13-s intervals and individually underwent zero correction using the baseline value. sLT was defined as the first significant increase in sweat lactate concentration above baseline based on a graphical plot (Fig. 7)13–15,34. MLSS determination. Blood lactate was measured before exercise and every 5 min during constant-load exercise for 30 min at 110%, 115%, 120%, and 125% of sLT intensity. The rotational speed was set at 70 rpm. The criteria that did not achieve the exercise and exceeded the MLSS included participants who could not finish the trial due to fatigue, but could not maintain bicycle pedaling at 70 rpm, as well as participants who could finish 30 min of exercise but had an increase in blood lactate of more than 1 mM from 10 min after exercise initiation to the end of the exercise. MLSS was defined as the greatest load among the loads in which blood lactate values at the end of exercise (30 min) increased within 1 mM, compared to those at 10 min after exercise initiation12 (Fig. 8). Measurement data. On the first day of measurement, body weight, body fat, and skeletal muscle mass were measured using In-Body (InBody470; InBody Japan Inc., Tokyo, Japan). Expired gas flow was measured using a breath-by-breath automated system. Three calibration processes were performed on the system: flow volume sensor, gas analyzer, and delay time calibration. Parameters of respiratory gas exchange, including ventilation (VE), oxygen uptake (VO2), and carbon dioxide production (VCO2), were continuously monitored and meas- ured using a 10-s average. Skeletal muscle oxygenation in the right thigh was measured using NIRS spectroscopy. 8 Vol:.(1234567890) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports/ The monitor consists of a light-sending probe and a light-receiving probe. Near-infrared light emitted from the light-sending unit is absorbed by skeletal muscle tissue, and changes in the intensity of the light returned to the light-receiving unit enable tissue oxygenation measurement35. A pair of probes was attached 4 cm apart on the skin over the vastus lateralis muscle in the distal third of the thigh36,37 and then covered and secured with tape38. In this study, tissue hemoglobin oxygen saturation (tissue oxygenation index [TOI]), calculated using the spa- tially resolved spectroscopy method, was assessed39,40. Statistical analyses. All data are presented as means and standard deviations. The obtained HR and VO2 were calculated as a percentage of the maximal HR (% HRmax) and peak VO2 (% peak VO2). The relationships between the sLT and ventilatory threshold (VT) were investigated using Pearson’s correlations. Additionally, the Bland and Altman technique was applied to verify the similarities among the different methods. This compari- son is a graphical representation of the difference between the methods and the average of these methods. As previous reports have shown that MLSS is 120% or more of LT intensity, we divided our cohort into two groups using the cut-off of 120% of sLT intensity4,11,12. Unpaired t-tests and Chi-squared tests were used to compare participant characteristics between the two groups. The correlation value was used to determine the relationship between the relative change in TOI from baseline (ΔTOI) and blood lactate at the end of the exercise. Unpaired t-tests were used to compare ΔTOI across trained and untrained participants. Figure 7. Sweat lactate levels during ramp exercise. HR heart rate, VO2/W oxygen uptake/weight. Figure 8. Imaging of the constant-load exercise. HR heart rate, VO2/W oxygen uptake/weight, BLt blood lactate. 9 Vol.:(0123456789) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports/ Receiver operating characteristic (ROC) curve analysis was used to determine the ΔTOI cut-off value for the completed constant exercise test. All analyses were performed using SPSS version 28 software (IBM Japan Ltd., Tokyo, Japan). Statistical significance was set at P < 0.05. Data availability All data from these studies are contained within this manuscript or are available from the corresponding author upon reasonable request. Source data are provided in this paper. Received: 27 February 2023; Accepted: 13 June 2023 References 1. Faude, O., Kindermann, W. & Meyer, T. Lactate threshold concepts: How valid are they?. Sports Med. 39, 469–490 (2009). 2. Beneke, R. Methodological aspects of maximal lactate steady state-implications for performance testing. Eur. J. Appl. Physiol. 89, 95–99 (2003). 3. Azevedo, R. A., Forot, J., Millet, G. Y. & Murias, J. M. Comparing of muscle V̇O2 from near-infrared spectroscopy desaturation rate to pulmonary V̇O2 during cycling below, at, and above the maximal lactate steady state. J. Appl. Physiol. 132, 641–652 (2022). 4. Garcia-Tabar, I. & Gorostiaga, E. M. A “blood relationship” between the overlooked minimum lactate equivalent and maximal lactate steady state in trained runners. Back to the old days?. Front. Physiol. 9, 1034 (2018). 5. Greco, C. C., Barbosa, L. F., Caritá, R. A. & Denadai, B. S. Is maximal lactate steady state during intermittent cycling different for active compared with passive recovery?. Appl. Physiol. Nutr. Metab. 37, 1147–1152 (2012). 6. Pelarigo, J. G., Machado, L., Fernandes, R. J., Greco, C. C. & Vilas-Boas, J. P. Oxygen uptake kinetics and energy system’s contribu- tion around maximal lactate steady state swimming intensity. PLoS ONE 12, e0167263 (2017). 7. Espada, M. C. et al. Ventilatory and physiological responses in swimmers below and above their maximal lactate steady state. J. Strength. Cond. Res. 29, 2836–2843 (2015). 8. Jones, A. M., Burnley, M., Black, M. I., Poole, D. C. & Vanhatalo, A. The maximal metabolic steady state: Redefining the “gold standard”. Physiol. Rep. 7, e14098 (2019). 9. Kilding, A. E. & Jones, A. M. Validity of a single-visit protocol to estimate the maximum lactate steady state. Med. Sci. Sports Exerc. 37, 1734–1740 (2005). 10. Baldari, C. & Guidetti, L. A simple method for individual anaerobic threshold as predictor of max lactate steady state. Med. Sci. Sports Exerc. 32, 1798–1802 (2000). 11. Garcia-Tabar, I., Rampinini, E. & Gorostiaga, E. M. Lactate equivalent for maximal lactate steady state determination in soccer. Res. Q. Exerc. Sport 90, 678–689 (2019). 12. Urhausen, A., Coen, B., Weiler, B. & Kindermann, W. Individual anaerobic threshold and maximum lactate steady state. Int. J. Sports Med. 14, 134–139 (1993). 13. Seki, Y. et al. A novel device for detecting anaerobic threshold using sweat lactate during exercise. Sci. Rep. 11, 4929 (2021). 14. Maeda, Y. et al. Implications of the onset of sweating on the sweat lactate threshold. Sensors 23, 3378 (2023). 15. Katsumata, Y. et al. Laminar flow ventilation system to prevent airborne infection during exercise in the COVID-19 crisis: A single-center observational study. PLoS ONE 16, e0257549 (2021). 16. Bräuer, E. K. & Smekal, G. VO2 steady state at and just above the maximum lactate steady state intensity. Int. J. Sports Med. 41, 574–581 (2020). 17. Iannetta, D. et al. A critical evaluation of current methods for exercise prescription in women and men. Med. Sci. Sports Exerc. 52, 466–473 (2020). 18. Vobejda, C., Fromme, K., Samson, W. & Zimmermann, E. Maximal constant heart rate—A heart rate based method to estimate maximal lactate steady state in running. Int. J. Sports Med. 27, 368–372 (2006). 19. Egan, B. & Zierath, J. R. Exercise metabolism and the molecular regulation of skeletal muscle adaptation. Cell Metab. 17, 162–184 (2013). 20. Hood, D. A., Memme, J. M., Oliveira, A. N. & Triolo, M. Maintenance of skeletal muscle mitochondria in health, exercise, and aging. Annu. Rev. Physiol. 81, 19–41 (2019). 21. Wilson, J. M. et al. The effects of endurance, strength, and power training on muscle fiber type shifting. J. Strength Cond. Res. 26, 1724–1729 (2012). 22. Gatterer, H. E. et al. Exercise performance, muscle oxygen extraction and blood cell mitochondrial respiration after repeated-sprint and sprint interval training in hypoxia: A pilot study. J. Sports Sci. Med. 17, 339–347 (2018). 23. Lindholm, M. E. & Rundqvist, H. Skeletal muscle hypoxia-inducible factor-1 and exercise. Exp. Physiol. 101, 28–32 (2016). 24. Marshall, H. C. et al. Effects of intermittent hypoxia on SaO2, cerebral and muscle oxygenation during maximal exercise in athletes with exercise-induced hypoxemia. Eur. J. Appl. Physiol. 104, 383–393 (2008). 25. Nagahisa, H., Mukai, K., Ohmura, H., Takahashi, T. & Miyata, H. Effect of high-intensity training in normobaric hypoxia on thoroughbred skeletal muscle. Oxid. Med. Cell Longev. 2016, 1535367 (2016). 26. Pramkratok, W., Songsupap, T. & Yimlamai, T. Repeated sprint training under hypoxia improves aerobic performance and repeated sprint ability by enhancing muscle deoxygenation and markers of angiogenesis in rugby sevens. Eur. J. Appl. Physiol. 122, 611–622 (2022). 27. Suzuki, J. Short-duration intermittent hypoxia enhances endurance capacity by improving muscle fatty acid metabolism in mice. Physiol. Rep. 4, e12744 (2016). 28. D’Souza, A. W., Notley, S. R. & Kenny, G. P. The relation between age and sex on whole-body heat loss during exercise-heat stress. Med. Sci. Sports Exerc. 52, 2242–2249 (2020). 29. Bull, F. C. et al. World Health Organization 2020 guidelines on physical activity and sedentary behaviour. Br. J. Sports Med. 54, 1451–1462 (2020). 30. Baker, L. B. & Wolfe, A. S. Physiological mechanisms determining eccrine sweat composition. Eur. J. Appl. Physiol. 120, 719–752 (2020). 31. Mehnert, P. et al. Prediction of the average skin temperature in warm and hot environments. Eur. J. Appl. Physiol. 82, 52–60 (2000). 32. Torii, M., Yamasaki, M., Sasaki, T. & Nakayama, H. Fall in skin temperature of exercising man. Br. J. Sports Med. 26, 29–32 (1992). 33. Havenith, G., Fogarty, A., Bartlett, R., Smith, C. J. & Ventenat, V. Male and female upper body sweat distribution during running measured with technical absorbents. Eur. J. Appl. Physiol. 104, 245–255 (2008). 34. Okawara, H. Kinetic changes in sweat lactate following fatigue during constant workload exercise. Physiol. Rep. 10, e15169 (2022). 35. Grassi, B. et al. Muscle oxygenation and pulmonary gas exchange kinetics during cycling exercise on-transitions in humans. J. Appl. Physiol. 95, 149–158 (2003). 36. Ishii, K. et al. Central command contributes to increased blood flow in the noncontracting muscle at the start of one-legged dynamic exercise in humans. J. Appl. Physiol. 112, 1961–1974 (2012). 10 Vol:.(1234567890) Scientific Reports | (2023) 13:10366 | https://doi.org/10.1038/s41598-023-36983-8 www.nature.com/scientificreports/ 37. Ishii, K. et al. Central command generated prior to arbitrary motor execution induces muscle vasodilatation at the beginning of dynamic exercise. J. Appl. Physiol. 120, 1424–1433 (2016). 38. Kowalchuk, J. M., Rossiter, H. B., Ward, S. A. & Whipp, B. J. The effect of resistive breathing on leg muscle oxygenation using near-infrared spectroscopy during exercise in men. Exp. Physiol. 87, 601–611 (2002). 39. Grassi, B. & Quaresima, V. Near-infrared spectroscopy and skeletal muscle oxidative function in vivo in health and disease: A review from an exercise physiology perspective. J. Biomed. Opt. 21, 091313 (2016). 40. Kurihara, K., Kikukawa, A., Kobayashi, A. & Nakadate, T. Frontal cortical oxygenation changes during gravity-induced loss of consciousness in humans: A near-infrared spatially resolved spectroscopic study. J. Appl. Physiol. 103, 1326–1331 (2007). Acknowledgements We are grateful to Editage for editing this manuscript. Author contributions The author contributions are stated as follows; Y.M. and Y.K. drew the manuscript. Y.M., D.N., T.S., H.O., and Y.K. prepared the images. Y.M., D.N., T.S., H.O., T.A., T.H., K.S., D.K., Y.I., G.I., and Y.K. collected the patient information. A.K., Y.Y., T.K., K.S., and Y.K. provided a critical revision of the manuscript for the key intellectual content and supervision. All of the authors have approved all aspects of our work, and have read and approved the manuscript. Competing interests No funding was received to conduct this study. Daisuke Nakashima is the shareholder and CEO of Grace Imaging Inc., which provided the lactate sensor equipment. The other authors declare no competing interests. Additional information Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1038/ s41598- 023- 36983-8. Correspondence and requests for materials should be addressed to Y.K. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2023

Estimation of maximal lactate steady state using the sweat lactate sensor.

06-26-2023

Muramoto, Yuki,Nakashima, Daisuke,Amano, Tsubasa,Harita, Tomota,Sugai, Kazuhisa,Daigo, Kyohei,Iwasawa, Yuji,Ichihara, Genki,Okawara, Hiroki,Sawada, Tomonori,Kinoda, Akira,Yamada, Yuichi,Kimura, Takeshi,Sato, Kazuki,Katsumata, Yoshinori

eng

PMC5345849

RESEARCH ARTICLE Comparison of vitality states of finishers and withdrawers in trail running: An enactive and phenomenological perspective Nadège Rochat1,2,3*, Denis Hauw1, Roberta Antonini Philippe1, Fabienne Crettaz von Roten1, Ludovic Seifert3 1 Institute of Sport Sciences, University of Lausanne, Lausanne, Switzerland, 2 Raidlight-Vertical SAS Outdoor Lab, Saint-Pierre-de-Chartreuse, France, 3 CETAPS Laboratory—EA 3832, Faculty of Sports Sciences, University of Rouen, Rouen, France * nadege.rochat@unil.ch Abstract Studies on ultra-endurance suggest that during the races, athletes typically experience three vitality states (i.e., preservation, loss, and revival) at the phenomenological level. Nev- ertheless, how these states contribute to the management and outcome of performance remains unclear. The aim of this study was to determine whether and how the vitality states experienced by runners and their evolution during a trail race can be used to distinguish finishers from withdrawers. From an enactive and phenomenological framework, we pro- cessed enactive interviews and blog posts of race narratives. We distinguished units of meaning, which were grouped into sequences of experience; each sequence was then cate- gorized as one of the three vitality states: state of vitality preservation (SVP), state of vitality loss (SVL) or state of vitality revival (SVR). We analyzed the distribution of these vitality states and their temporal organization at the beginning, in the second and third quarters, and at the end of the races, and we qualitatively characterized runners’ adaptations to SVL. Results showed that finishers completed the race in SVP, with overall significantly more sequences in SVP and significantly fewer sequences in SVL than withdrawers. SVR did not discriminate finishers from withdrawers. The temporal organization of the vitality states showed a significant difference in the emergence of SVP from the second quarter of the race, as well as a significant difference in the emergence of SVL from the third quarter of the race. The analysis of adaptations to SVL confirmed that finishers were more capable of exit- ing SVL by enacting a preservation world when they felt physical or psychological alerts, whereas withdrawers remained in SVL. Our results showed that finishers and withdrawers did not enact the same phenomenological worlds in the race situation, especially in the orga- nization of vitality adaptations and their relationships to difficulties; the cumulative effect of the succession of experienced vitality states differed, as well. PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 1 / 24 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Rochat N, Hauw D, Antonini Philippe R, Crettaz von Roten F, Seifert L (2017) Comparison of vitality states of finishers and withdrawers in trail running: An enactive and phenomenological perspective. PLoS ONE 12(3): e0173667. doi:10.1371/journal.pone.0173667 Editor: Luca Paolo Ardigò, Universita degli Studi di Verona, ITALY Received: May 2, 2016 Accepted: February 26, 2017 Published: March 10, 2017 Copyright: © 2017 Rochat et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: Data are available from the institutional data access of University of Lausanne (contact corresponding author) for researchers who meet the criteria for access to confidential data. Funding: Grants from ANRT (National association of research and technology) (http://www.anrt.asso. fr/fr/espace_cifre/accueil.jsp#.VyTPoXAjFO19) under the CIFRE agreement (Industrial Convention of Learning by Research) with the Raidlight compagny, http://www.raidlight.com/fr/ and Introduction Trail running has become a popular sport over the last 40 years, as noted by Hoffman et al. [1], who observed an increasing number of races per year and more participants per race [2]. Trail running is no longer practiced by a minority of elite runners, but has also become accessible to non-professional runners [3] despite the need for high investment and compromise in terms of training, work schedule and personal life [4]. The races consist of semi-autonomous run- ning along marked trails in natural environments and impose considerable constraints that the runners must adapt to, raising effort, logistic and safety issues. The distances vary between 20 kilometers to more than 300 kilometers, with ultra-trail races generally more than 80 kilo- meters. During these races, runners risk extreme fatigue and/or exhaustion and at times exceed their personal limits [4]. Ultra-trail racing has therefore been considered an extreme sport or even a dangerous activity [5]. There are multiple ways to analyze this type of ultra-endurance performance. A current trend is the “third-person approach” to identify the determinants of performance. In this case, studies are based on the assumption that performance is dependent on two types of factors: (a) before-the-race factors, which include, for example, training habits [6], the impact of training characteristics on running-related injuries [7], and physical, mental and tactical preparation [4], and (b) during-the-race factors, which include sleep-deprivation effects [8] and neuromus- cular damage [9]. Other determinants of performance have been examined by isolating specific characteristics, without distinguishing between these two types of timed factors; these include mood states [10], cognitive functioning [11], personality traits [12], emotions [13], sarcomere disruption [14], and alteration of jump height mechanics after a mountain footrace [15]. How- ever, these approaches partition the unity of runners’ activity into specific processes, which precludes the possibility of understanding runners from a holistic perspective – that is, as capa- ble of compensating a performance deficit in one phase of the race by heightened performance in another phase, or compensating one process by another, such as psychological coping with physiological problems [16]. Another trend in cognitive science has developed in this direction and consists of investigating the way people integrate physiological and psychological factors into a mental unity that emerges at the psycho-phenomenological level [17,18]. From this per- spective, human activity is (a) Embedded in the whole dynamics of the changing situation [19], (b) Extended by tools or cultural artifacts (e.g., [20]), (c) Embodied as recurrent sensori- motor patterns of perception and action [21,22], and (d) Enacted by bringing forth a cognitive being’s world with specific asymmetrical relationships between the person and his/her envi- ronment [17,23]. This last element of the four-E approach assumes that the cognitive being’s world–whatever that being is able to experience, know, or practically handle–is a constitutive part of human activity that should be investigated in a rigorous phenomenological manner [21]. Phenomenology benefits from a philosophical background and has shown its practical applications in research in the cognitive sciences [24]. The objectives of the phenomenological approach are to analyze experience and describe a phenomenon in terms of how it emerges at the level of consciousness [25]. This approach thus operates at the interface of conscious and uncon- scious processes, or at the “fringe of consciousness” [26]. Furthermore, by analyzing experience in context, phenomenology roots its analysis in the pre-reflectively experienced lived body [27]. At this level of analysis, accounting for how a person feels and acts in a given situation requires a “first-person approach” (e.g., [24,25]) to identify how experience includes physical events and the synthesis of sensorial events. The heterogeneity of these events thus emerges at the level of mental or cognitive unity or the “feeling of what is happening” [21,28,29]. This experience is the sense (including bodily, emotional, cognitive, action and situational dimensions) that corresponds to the manner by which people make worlds emerge, in which they have being and act [30]. Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 2 / 24 swissuniversities (NR). We have read and understood PLOS ONE policies on Financial Disclosure and Conflicts of Interest and declare the employment of one author (NR) in Raidlight- Vertical SAS Outdoorlab Company, Saint-Pierre-de- Chartreuse, France. The funder provided support in the form of salaries for the author (NR), but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The specific roles of this author are articulated in the “author contributions” section. Competing interests: We have read and understood PLOS ONE policies on Financial Disclosure and Conflicts of Interest and declare the employment of one author (NR) in Raidlight- Vertical SAS Outdoorlab Company, Saint-Pierre-de- Chartreuse, France. This commercial affiliation does not alter our adherence to all PLOS ONE policies on sharing data and materials. Therefore, Di Paolo et al. [31] suggested that a cognitive being’s world consists of sense-making instead of information processing, which is key to the computationalist view. Cognition can thus be conceived as a form of embodied action in which a person creates meaning by interacting autonomously with the environment, leading to fluctuations in experience. By rooting the analy- sis of these fluctuations in the enactive paradigm, which encompasses the dimensions of experi- ence, meaning and asymmetrical interactions, we argue that it is possible to obtain the temporal organization of the salient phenomenological states that characterize experience. A multiplicity of phenomenological states in trail running Based on the four-E approach, we postulated that trail running in a competitive context would shape runners’ activity and experience in such a way that we would be able to identify the emer- gence of typical phenomenological states. Our challenge was to define the nature of these states, which are singular, context-dependent and peculiar to each runner; therefore, we assumed there would be many conceivable phenomenological states. However, experiences in trail run- ning also have many similarities, as shown in a study about the story of withdrawals [32], which identified the categories that characterized 20 trail runners’ enacted worlds from the start of the race until the moment they decided to quit it. The findings opened new possibilities for analyz- ing trail running experience by suggesting that the runners went through various phenomeno- logical states that could be clearly labeled in categories. In the same vein, qualitative studies of ultra-endurance runners identified three major states in finishers related to effort management. The first concerned the stressors that impact their experience (i.e., cramping and injuries, gas- trointestinal problems, and thoughts about quitting) [16] and characterize a state of suffering. The second concerned the protective processes aiming at preserving oneself, such as psychologi- cal coping strategies like setting short-term goals, pace monitoring, hydration and nutrition, and social support [16]. The third state concerned positive emotions, such as group cohesive- ness during a part of the race, self-awareness or mental stamina emerging during effort [33], and positive self-talk [4] to revive vitality during the race. Interestingly, these states correspond to the findings of other targeted studies in physiology that had to do with the state of suffering. For example, sleep deprivation was reported to nega- tively impact runners’ cognitive performances (i.e., decreased psychomotor vigilance, increased reaction time lapses, inability to stay awake, and sometimes visual hallucinations) [8]. More- over, long-distance effort was found to lead to emotional disturbances and negative energetic balances [13]. In contrast, physiological evidence also indicated that runners can prevent this state by using anticipatory processes that ensure a form of self-preservation: indeed, less neuro- muscular fatigue, muscle damage and inflammation were reported in a 330-kilometer race than in shorter races [9]. According to the authors, the runners adopted a protective pacing strategy during the first half of the race, which reduced muscle damage. Hence, these studies have all shown that the experience of running a trail race corresponds to the various dimensions of the runners’ activity [34]. These dimensions document the pro- cesses involved when runners prepare for a race or confront difficulties in the race, and during key moments of performance. Moreover, the notion of vitality embeds all these dimensions. Subjective vitality is “a conscious experience of possessing energy and aliveness” ([35], p.530), which can also be absent in other contexts. These authors emphasized that vitality is a psycho- logical experience that depends on a physiological state (e.g., fatigue, illness) and psychological states. Therefore, based on the evidence that vitality has phenomenological anchorage, we assume that it (2) embeds physiological and psychological processes that constitute heteroge- neous information synthesized into mental unity and (2) can fluctuate according to the context. Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 3 / 24 There are several ways to analyze vitality, which has been conceived as a global state [35] and has been used to investigate well-being and mindfulness [36,37] with a validated scale to measure it [38]. However, although this scale provides information about a subject’s general well-being in relation to other factors (i.e., friendship, intrinsic values), the temporal fluctua- tions of vitality still remain unclear. We postulate that trail runners’ experiences and meaning are timely and structured [39], reflecting an emergent process of experiencing the race situa- tion. Indeed, as previously emphasized, runners experience phenomenological vitality states, and analysis of the temporal organization of these states might provide an understanding of how they emerge. This in turn might provide insight into the runners’ adaptations to the phenomenological vitality states and thereby explain race outcomes (finish vs. withdraw). We sought to connect this conception of vitality with the existing literature about trail run- ning and endurance running and identified three vitality states. The first was a “state of vitality loss” (i.e., SVL) that runners may experience during a trail race, prompting them to withdraw because of the constraints typical of this sport [40]. Conversely, runners may also live a “state of vitality revival” (i.e., SVR) in which they feel good sensations and positive moments. Third, the protective processes emerging during these races suggest that runners experience a “state of vitality preservation” (i.e., SVP), which suggests that withdrawers might not be exclusively those who experience SVL, but also those who cannot preserve themselves sufficiently. They therefore experience more vitality loss during the race and progressively become unable to fin- ish [32]. Comparing how these phenomenological vitality states evolve in finishers and with- drawers may provide insight into the processes that lead to withdrawal from or completion of the race. By providing a qualitative description of the race outcome, the temporal organization of SVL, SVP and SVR may be important not only to understand how an immediate state can impact the following states, but also to further explore how the positive mood states that run- ners experience during long races are able to re-launch them completely after a period of suf- fering or vitality loss. We hypothesized that (1) performance outcomes (finish vs. withdraw) would result from the temporal organization and interactions of these phenomenological states and (2) the lack of a vitality revival during a race would prompt a runner’s decision to withdraw. To summarize, few studies have investigated long-distance running outcomes (finish vs. withdraw) and the phenomenological vitality states that have been observed during the races, which include physiological, emotional and cognitive processes. We report here the results of an investigation of the phenomenological vitality states during multiples races to understand why runners finish or withdraw from a race. We used an enactive and phenomenological per- spective to characterize the distribution and temporal organization of these states and runners’ adaptations to them. Material and methods Research design For the analysis of experience, the sport and psychological sciences have mainly used two approaches to consider its temporal organization [41,42]. Within an enactive framework, the “course of experience” analyzes experience through the succession of enactments at the level of what agents are able to perceive, feel, know and do [43,44]. Here, sense-making is studied by identifying the succession of linkages between action and situation considered at the level of what is meaningful for an agent, using a semiotic approach to cognition and action inspired by Peirce (1931–1935) [45–48]. The course of experience reflects the world enacted by genuine agents in situation through the characterization of elementary units of meaning (EUMs) that mainly emerge from the association of the agent’s intentional state (i.e., the field of possible Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 4 / 24 actions he or she can undertake) and situation-related judgments of a proprioceptive, percep- tive or memory-based nature (i.e., representamens) [43,49]. The course of experience is thus the chaining of these EUMs during a period of an agent’s activity characterized by emergent or higher-order structures of meaning, such as sequences (i.e., the succession of EUMs that corre- sponds to a similar agent’s concern). A key element that distinguishes the course of experience from the narrative approach is the use of second-person reports coupled with third-person descriptions, such as video recordings or biomechanical data [47]. Second-person reports are collected through a process by which traces of past activity are presented to the agent to stimu- late a re-enactment [50–52]. Here, agents are invited to re-experience and describe the stream of past experience in relation to these traces by adopting a stance that consists of reliving their own past experience, although deliberately ignoring the outcome [52]. In doing so, they re- enact meaningful parts of their past experience. The confrontation with each trace of the peri- ods and shifts in an agent’s experience is considered as a new situation, although the new meaningful experience that is built has many similarities with the one that the agent lived in the past [46,49]. As this approach has been successfully used in studies analyzing performance outcomes in various sports (e.g., [51,53–55]), we assumed it would be suitable for trail running. The second approach, narrative inquiry, has received a great deal of attention in relation to the report of the stream of activity. Bruner [56], for example, suggested that our personal knowledge and experience are organized through narratives of sequences of events that corre- spond to a psycho-phenomenological level. According to Gibbs [22], agents describe those events that are meaningful for them through narration. The narrative structure provides land- marks that meaningfully and in timely fashion discretize the stream of the events that form the structure of a story of a person’s experience [57]. For Bruner [56], Propp [58] and Greimas [59], the analysis of narratives shows that their structure possesses properties that represent how the meaning of sequences of events is experienced (e.g., the existence of a plot, obstacles to overcome, problems to solve, presence of allies, the necessity of having continuity in one’s identity, the possibility of expressing mood or the search for meaning). Relatedly, researchers have shown increasing interest in experience-sharing blogs and forums as an innovative tool for analyzing narratives. Blogs provide a space for personal expression, and Bortree [60] observed that teenage girls were more likely to express their thoughts and report on their daily activity by recounting their experiences on their blogs. Blogs are thus suited for gathering experiential data because these are expression spaces in which people feel comfortable to talk about their personal experiences [61]. The accounts posted on blogs can thus help us obtain valuable narratives of the personal experience that has marked runners [62,63], as a participa- tory sense-making process emerges from the interaction between the teller and the reader [64]. Sport sciences have also shown great interest in these narratives in the fields of health (e.g., [65]), physical activity and leisure [66], physical education (e.g., [67]), adapted physical activity [68] and elite sport (e.g., [69]). By putting together the course-of-experience and narrative approaches, runners’ reports of experience have provided insight into the way temporal organization reflects the segmentation of separate phenomenological states [70–73], including the key shifts in the vitality states dur- ing a trail running race. These successions of t-time vitality states enacted by the athletes emerged at the phenomenological level as “perceptual packets” [39,74] forming sequences (e.g., SVP, SVL, SVR) with unpredictable duration and chaining. They characterized runners’ step-by-step experiences that depicted their singular story of vitality states during a race. Thus, our research was designed to process two types of data: (a) recorded and transcribed commentaries elicited by researchers during enactive interviews (EIs) with athletes who were confronted with traces of their own past activity and asked to rebuild their experience (i.e., EI Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 5 / 24 data) and (b) freely written accounts of races retrieved from blogs published on the Raidlight community website (i.e., blog data). EI data processing EI participants. Thirteen French runners (nine males and four females) who participated in the “Grand Raid de la Re´union” volunteered to participate in the study. All were between 26 and 52 years old. They were recruited (a) through a message posted on the Raidlight community forum or (b) by responding to a notice before the race. Snowball sampling enabled us to obtain these participants. The protocol was carried out during a trail running event on Reunion Island, which comprised three races in which our sample participated, according to the following repartition (Table 1): During this event, the race statistics indicated a 48.1% rate of withdrawal. Participants were between 24 and 74 years old (mean age = 43 years old) [75] and 90% were men and 10% were women. The six finishers in our sample were ranked between 8th place and 1130th place. There- fore, the proportion of finishers in our sample matched the race statistics, and the demographic characteristics suggest that the runners in our sample were representative of the diversity among the participants. EI data collection. EI data were collected using traces of the runners’ past activity. The traces were two maps of the race: the first provided information about aid stations and geo- graphic landmarks (depicting the view of the route from above) and the second showed the elevation changes along the route. EIs took place shortly after the race, lasted between 60 to 120 minutes, and were recorded. During the EIs, the runners were confronted with these traces. The interviews were designed to provoke the re-emergence of elements of past experience when the participant was bodily face to face with traces of his/her own activity. The runners were asked to show, tell about and comment on their experience. In doing so, they revealed how they handled it online by build- ing new meanings (i.e., re-enactment) or activating pre-existing ones (i.e., remembering) [47,48,53,76]. The researchers took steps to prevent the runners from retrospectively recalling their experience. First, they asked the runners to avoid judging their activity (i.e., judgment suspension) and to concentrate on explaining the experience, as suggested for phenomenologi- cal research [77]. Second, the traces of past activity presented during the EI were aimed at stimulating the runners to re-enact the stream of their experience in situation while deliber- ately ignoring the outcome [51]. Third, to ensure that the runners were not retrospectively recalling their race experience, the researcher took careful note during the interviews that all runners, whatever their outcome, related positive and/or negative experiences, such as pain, joy, ease, etc. In addition, if a runner emphasized a positive account during the interview, the principle of in-depth qualitative research dictated that the researchers looked for a more accu- rate and authentic report of experience, always in relation to the unfolding situation. Verbal prompts were used to elicit further information about the meaning of each runner’s activity, including sense-making from an enactive perspective, this being the actions insepara- bly coupled with their experience and following their own story of the race: involvements, Table 1. Repartition of the EI participants (N = 13). Race name Length (km) Positive elevation gain (m) Finishers (n) Withdrawers (n) “Diagonale des Fous” 173 9996 3 5 “The Bourbon Trail” 97 5655 2 2 “The Mascareignes Trail” 65 3922 1 0 doi:10.1371/journal.pone.0173667.t001 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 6 / 24 units of meaning (i.e., action) and representamens, as has been done in previous research using the course-of-experience approach [44,45,47]. The involvement (I) refers to the possibilities that are conceivable by the actor in the situa- tion; it expresses how he/she enters into activity (e.g. “What were you concerned about at this moment?”). The unit of meaning (UM) depicts the breaks of the runner’s race story that corre- spond to the fraction of activity that is meaningful for him/her at each moment (e.g. “What were you doing?”). The representamen (R) corresponds to what the runner is feeling in rela- tion to the UM or the feeling of what is happening in the unfolding situation (e.g. “What was significant for you at this moment?”). EI data coding. The data treatment was grounded in a phenomenological method that articulates the inductive and deductive approaches by identifying the descriptions of a phe- nomenon that can be clustered into discrete categories and then put together to identify the core and the structure of the experience (see Starks & Brown Trinidad, p. 1373 [77]). Hence, the phenomenological data were inductively coded into units of meaning, then deductively classified using first categories of meaning (i.e., involvements, representamens, and units of meaning), and last classified as one of the three vitality states identified in our lit- erature review (i.e., SVL, SVP and SVR). This approach has been used in various studies in sport science, ergonomic and educational research [78–80]. It required a succession of four data coding steps. First, the enactive interviews were transcribed verbatim. Second, a general coding system for describing the settings of activity was established for each runner (Table 2). The system put together all the information collected in the EI with traces of past activity; this allowed us to rebuild the story of the race as experienced by each runner. Example: Extract from an EI “Researcher: Please comment on your race as you lived it, tell me when there were changes. You have the race maps to help you. Runner: The atmosphere was great this year! There were people everywhere, enthusiastic at the start and this enthusiasm lasted a really long time. The weather was very good, warm, so I started with a T-shirt; just a T-shirt so it was great, encouraged by the crowd, by people on the sides of the trails. Then we started, I didn’t want to start too fast, because I was recently injured, so I started to run at a good pace but I moderated it in order to preserve myself. Researcher: Did you feel pain? Runner: A little bit. But. . . It was not. . . Actually I injured my adductors in July. I had more or less recovered with physiotherapy and I knew that I shouldn’t start too fast to avoid getting hurt. However, in the meantime, I had sciatica so these last few days, it was painful and I told myself: the sciatica will pass because often before the start I feel pain everywhere.. . I’m experienced with that but I started confident anyway, so I’ll see, I’ll go as far as I can and it’ll be a tough race as usual, so at the first aid station, it was perfect, I drank water, I needed nothing and I continued in the direction of Berive, many people were outside and it was very motivating to see people everywhere. I really enjoyed it.” Table 2. Example of UM coding system from EI data. Unit of meaning (UM) Starts the race in St-Pierre wearing just a T-shirt Runs at a good pace but moderates her speed Drinks water at the first aid station Continues to Berive Involvement (I) Shouldn’t start too fast and is confident Shouldn’t start too fast - Motivated Representamen (R) Great atmosphere with the crowd at the start/ sciatica pain People on the sides encouraging It’s perfect, no need of anything Many people encouraging doi:10.1371/journal.pone.0173667.t002 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 7 / 24 Third, these UMs were grouped into sequences that referred to the same story during a part of the race. Two UMs belong to the sequence “if one is partly determined by the outcome of the other or if they both refer to the same theme” [81]; their formulation synthesized the con- tent of the UM (Table 3). Fourth, each sequence was classified into one of the three types of vitality states identified in the coding system, according to the following classification (Table 4). To identify the states, we collected all the sequences and examined their content (i.e., involvements, representamens and units of meaning). By doing so, we were able to classify them in their corresponding vital- ity state. For greater clarity, we normalized the formulation of the content of each category suc- cinctly to portray all the typical dimensions of the trail runners’ experience. Thus, for each athlete we obtained a succession of the vitality states identified in their course of experience (Fig 1). As shown in this figure, the same vitality state could be distinguished in two successive sequences when, for example, the involvement and the representamen changed focus while the general theme stayed the same. To ensure the validity of the data coding, the steps of data treatment were carried out independently by three researchers who then com- pared their respective codings in order to find common agreement. Blog data processing Blog data selection. Thirty-three blog posts on the community website of the Raidlight brand were selected from among several types of online contents. All were post-race accounts of experience. We collected 17 blog posts reporting finishing the race and 16 reporting with- drawal. The data collection complied with the terms and services of the Raidlight website. Table 3. Example of sequences identified from the UM coding. Unit of meaning (UM) Runs on a unknown trail segment Crosses a village Leaves the village Keeps on crossing villages Finds a known path Arrives at the aid station Involvement (I) Destabilized because he is on a unknown path Angry Tired, bad mood, less concentration Angry Focused on the race again Can stick to his plans again Representamen (R) Feeling of loss of control, negative emotion Technical difficulties, negative emotion - It’s hard Many people are encouraging Members of his support team, feeling good Sequences Runs angry and less concentrated on an unknown trail segment Runs relaxed, with good sensations on a known path doi:10.1371/journal.pone.0173667.t003 Table 4. Criteria for coding the sequences as phenomenological vitality states. State of vitality revival (SVR) State of vitality preservation (SVP) States of vitality loss (SVL) Involvement (I) Lead the race, get ahead of competitors, motivated to overcome, gain time or increase advance Be careful with the pace, preserve oneself, energy, keep physical integrity, do not get hurt Hold on, struggling to go on Unit of meaning (UM) Run/walk fast, accelerate, decide not to stop at an aid station, pass other runners Slow down, do medical procedure, use logistical supports, force oneself to stay at a perceived slow pace, deliberately do not pass a competitor, take breaks, hydrate, eat, sleep Constrained activity such as slow down, walk slowly, lose the route Representamen (R) Other runners’ activity, feeling of having much energy, speed is higher than expected Feeling of ease, pleasure Bad sensations, difficulty, pain, tiredness, cold, negative emotions, bad sleep, hallucinations, concerns about not being able to finish the race, feeling of going slower than expected, people passing, thoughts about abandoning doi:10.1371/journal.pone.0173667.t004 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 8 / 24 Blog data collection. In order to make the narrative contents compatible with the course- of-experience analysis, we searched each of them for the same information on meaning as for the EI data collection. When the blog data were not suitable for this coding system, they were deleted from our database (n = 5, 2 finishers and 3 withdrawers). The selection targeted trail running experience in various races (M = 94.90 km, SD = 39.92) according to the following repartition (Table 5). In total the data set was composed of 28 blog posts (15 finishers and 13 withdrawers). Blog data coding. As we obtained the same type of data as in the EIs, we applied the same coding procedures. We successively established the general coding system for each blog and the corresponding vitality states coding system (Table 6). “At 8 p.m. at St-Pierre, I’m among the first ten runners to enter the start area. I feel stress and stamina, mixed with the feeling of living something exceptional. I feel a bit nervous because I arrived by plane yesterday and I’m afraid I lack sleep. Anyway, I’m here with one single idea: finish. At 11 p.m., I’m literally transported by the stream of 2182 runners behind me. With D. we’re starting fast as planned. Too fast, sometimes at 14 kilometers per hour in the first 7 kilometers, and we passed the first checkpoint in 40th place. We start the ascent. D. slows down and around the 12th kilometer, I start chatting with A., a runner I met in another race. This makes make realize I should not be here, and even if I feel good, I slow down.” Ensuring data validity. Several measures were taken to ensure the comparability of the data. First, two investigators, each experienced at conducting qualitative research indepen- dently, coded the 41 data transcripts according to the criteria for the general and vitality states coding systems. An agreement rate of 90% was obtained between the two coders. A third cod- ing session was conducted to reach consensus for the 10% disagreement. Second, the number of UMs collected with EIs and blogs were compared to statistically assess whether they were of the same order of size. We hypothesized that a non-significant Fig 1. Succession of the vitality states in sequences identified from the runners’ courses of experience. doi:10.1371/journal.pone.0173667.g001 Table 5. Repartition of the blog data participants (N = 28). Race name Length (km) Positive elevation gain (m) Finishers (n) Withdrawers (n) “CCC” 106 6100 8 1 “UTMB” 170 10000 0 3 “Nicolet-Revard” 51 2700 6 0 “TransjuraTrail” 72 3200 1 1 “UTPMA” 105 5600 0 2 “GRP” 80 5090 0 1 “Infernal des Vosges” 160 7300 0 1 “TVS” 110 8375 0 1 “Ecotrail” 50 3681 0 1 “TGV” 73 3800 0 1 “Sainte´lyon” 72 1950 0 1 doi:10.1371/journal.pone.0173667.t005 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 9 / 24 difference would reflect a comparable segmentation of the courses of experience. This would confirm the agreement across reports on the criteria that the runners used to indicate mean- ingful breakpoints in their experience for a comparable segmentation of narratives. A chi- square test compared the repartition of the three vitality states (i.e., SVR, SVP, SVL) between the two datasets (i.e., EI and blog coding) and indicated a non-significant difference (χ2(2) = 0.301, p = 0.860), suggesting that the number of sequences in SVR, SVP and SVL did not sig- nificantly differ when coding blog and EI data. In addition, we compared the number of sequences between the two datasets and found no significant difference (Welch’s test of two independent samples with inequality of variances: t (16.257) = 1.233, p = 0.235), and we compared the length of the races between the two datasets and found no significant difference (Welch test: t(14.455) = -1.877, p = 0.081). Statistical results on the assessment of the comparability of the EI and blog data authorized us to gather them into a single dataset for the next step of data processing. Ethics statement The protocol was approved by the ethics committees of both the University of Rouen and the University of Lausanne (joint agreement) and followed the guidelines of the Declaration of Helsinki. Procedures were explained to the participants, who then gave their written informed consent to participate. EI and blog data processing We performed a logistic regression to explain the dichotomous outcome (finish vs. withdraw) with two independent variables: the number of kilometers of the race and the number of sequences. The results indicated that the number of kilometers was significant (exp(B) = 1.048, p = 0.013: when the number of kilometers of the race increased, the chances of finishing it increased as well) but not the number of sequences (exp(B) = 0.942, p = 0.540). Therefore, we used percentages instead of counts in the subsequent analyses. The data were processed in four steps to determine whether the race outcome (finish vs. withdraw) could be characterized by: (a) the distribution of the vitality states, (b) their tempo- ral organization, (c) the runners’ immediate adaptation to the experienced state of vitality loss and (d) the contents of the runners’ adaptations. Distribution of the vitality states. The distribution of the vitality states in relation to the race outcome (i.e., finish vs. withdraw) was determined by comparing the means and standard deviations of the percentages of each vitality state for finishers and withdrawers. T-tests com- pared the repartition of SVR, SVP and SVL in finishers and withdrawers; when variances dif- fered between the groups, we used the Welch test. Normality was tested with the Shapiro-Wilk test. All tests were performed using the significance level of 5% (p0.05) with SPSS statistical Table 6. Example of coding system for blog data. Unit of meaning (UM) Enters the start area Runs the first 7 kilometers fast with D Passes the first checkpoint Starts the ascent chatting with A Realizes he is running too fast Slows down Involvement (I) Wants to finish the race Planned to start fast - - - Should slow down in spite of his good sensations Representamen (R) Feels stress, nervousness and fear of lacking sleep Reaches a speed of 14 kilometers per hour Holding the 40th place His friend slows down Feels good Feels good doi:10.1371/journal.pone.0173667.t006 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 10 / 24 software. Furthermore, for each course of experience, we quantified the total number of sequences of experience and their repartition in the vitality states of four periods: the begin- ning, the second quarter, the third quarter and the end of the race. Temporal organization of the vitality states. The temporal organization of the vitality states in relation to the race outcome was determined using measures or cumulative mea- sures. First, we analyzed the number of SVL and SVP sequences in the four race periods to detect a difference in pattern between finishers and withdrawers. Moreover, we performed a logistic regression of eight independent variables (four measures of SVL and SVP), which were important to predict race outcome via an iterative method (forward, likelihood ratio). Second, we divided the cumulative number of sequences per category of vitality state by the cumulated number of sequences in the four race periods as explained above. To do so, we identified the relative accumulation of each vitality state for each sequence of experience (Table 7). Therefore, we calculated the ratio of each vitality state for each sequence: for instance, the ratio of SVP at the 6th sequence was 4/6 = 0.66. For each vitality state and for each race, we cal- culated the ratio at the one-third and two-third points and the end of the race. Thus, each course of experience was split into four periods in which we obtained the per- centage of the state of vitality experienced for each state in each sequence and the percentage of each cumulated state each period. T-tests compared the percentages of cumulated states in the four periods by controlling type I error, i.e., using a p-value of 0.0125 (i.e., 0.05/4) as the level of significance (Bonferroni approach). As before, we checked the homogeneity of vari- ances and normality. Runners’ immediate adaptation to the state of vitality loss. The runners’ adaptations after experiencing a state of vitality loss during the race were assessed to determine whether finishers and withdrawers could be distinguished by their ability to reorganize their activity when they went through various vitality states. To do so, we calculated the types and frequency of vitality states at t+1 after a sequence in SVL for finishers and withdrawers. A chi-square test compared the number of sequences in SVR, SVP or SVL after a sequence of SVL. Content of the runners’ adaptations. The analysis of the content of the runners’ adapta- tions compared the representamens and involvements between finishers and withdrawers in the sequences in SVL and SVP, based on the following assumptions: (a) the more time runners spend in SVL, the more probable it is that they will withdraw, (b) the more time runners spend in SVP, the more probable it is that they will finish, (c) attempts to cope with SVL will help runners exit from this state of vitality loss, and (d) being able to maintain SVP will help run- ners to experience less SVL. We clustered these two elements of meaning contrasting finishers and withdrawers into types using thematic analysis, as suggested by Vaismoradi [82]. Our two- fold intent was to determine whether during the SVL sequences runners were only in a state of suffering or were also trying to enact a new experience in response to difficulties, and whether during SVP sequences they were only running without being aware of this preservation state or were actively trying to maintain this state. Table 7. Example of emergence of vitality states for each sequence. Sequences 1 2 3 4 5 6 7 8 9 10 Course of vitality states SVP SVP SVR SVP SVL SVP SVL SVL SVP SVL Cumulative number of SVR 0 0 1 1 1 1 1 1 1 1 Cumulative number of SVP 1 2 2 3 3 4 4 4 5 5 Cumulative number of SVL 0 0 0 0 1 2 3 3 3 4 doi:10.1371/journal.pone.0173667.t007 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 11 / 24 Results Distribution of the vitality states The repartition of the sequences in each vitality state revealed that finishers had significantly more sequences in SVP than withdrawers (i.e., 59.5% units of preservation for finishers whereas withdrawers had 39.8% units of preservation, t(39) = 6.782, p = 0.000). Moreover, fin- ishers had significantly fewer units of SVL than withdrawers (18.7% for finishers and 42.2% for withdrawers, t(39) = -7.853, p = 0.000). There was no difference in the units of vitality revival (SVR) between finishers and withdrawers (t(39) = 1.279, p = 0.208) (Table 8). Temporal organization of the vitality states The evolution of sequences in SVP in the four periods of the race for finishers and withdrawers is represented in Fig 2. Throughout the race, the two groups increasingly diverged, although both followed a similar pattern of decrease. For the SVL category (Fig 3), the evolution was also different for finishers and withdrawers, following a similar pattern of increase for the first three periods, but different trajectories for the last period. The cumulated frequency of the states in each period (i.e., beginning, second quarter, third quarter, end of the race) while controlling type I error (i.e., Bonferroni approach with level of significance of 0.05/4 = 0.0125) showed the following: • For SVP: There was no significant difference in the beginning between the two groups (t(39) = 0.852 and p = 0.400). Then in the second quarter a significant difference was Table 8. Percentages of the three categories of vitality states in blogs and EIs (N = 41). SVR SVP SVL Finishers Withdrawers Finishers Withdrawers Finishers Withdrawers M 21.74 17.98 59.51 39.81 18.75 42.21 SD 8.61 10.19 8.20 10.33 8.20 10.80 doi:10.1371/journal.pone.0173667.t008 Fig 2. Estimated means of sequences in SVP in finishers and withdrawers in the four periods. doi:10.1371/journal.pone.0173667.g002 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 12 / 24 observed (t(27.453) = 2.783 and p = 0.011), which remained in the third quarter and the end of the race (resp. t(39) = 4.756 and p = 0.000, t(39) = 6.782 and p = 0.000). • For SVL: There was no significant difference in the beginning (t(39) = -0.631 and p = 0.532) and in the second quarter (t(30.076) = -2.597 and p = 0.014). A significant difference was observed in the third quarter (t(39) = -5.050 and p = 0.000) and the end of the race (t(39) = -7.853 and p = 0.000). A logistic regression assessed the temporal organization of the SVL and SVP states simulta- neously. The iterative procedure of the method designated the following as the two most important measures to explain the race outcome: the measure of SVL at the end and the mea- sure of SVP in the second quarter (Table 9). With these two measures, we could accurately pre- dict 95.1% of the runners’ outcomes. A higher number of SVL sequences at the end decreased the likelihood of finishing the race, whereas a higher number of SVP sequences in the second quarter increased the likelihood of finishing the race. Runners’ immediate adaptations to the state of vitality loss Finishers more often experienced an SVL-SVP transition than withdrawers (66.12% against 40%, Table 10). Withdrawers more often experienced two consecutive sequences of SVL than finishers (24.76% against 6.45%). Last, 25.8% of SVL-SVR was observed among finishers against 18.09% among withdrawers. The chi-square test showed a significant difference (χ2(2) = 12.21, Fig 3. Estimated means of sequences in SVL in finishers and withdrawers in the four periods. doi:10.1371/journal.pone.0173667.g003 Table 9. Results of the logistic regression to explain the race outcome (i.e., finish or withdraw). A Wald statistics p SVL end -15.15 6.619 0.010 SVP second quarter 9.52 6.341 0.012 doi:10.1371/journal.pone.0173667.t009 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 13 / 24 p = 0.002) in finishers’ and withdrawers’ states at t+1. Note in addition that 90% of withdrawers’ last sequences are in SVL versus 4.76% in finishers. Content of the runners’ adaptations. The thematic analysis of the coding showed that both finishers and withdrawers experienced negative physical sensations such as pain, cold and cramps at an immediate level, as identified in the representamens during SVL sequences (Table 11). However, the analysis of involvements showed that finishers attempted to cope with these difficulties by (a) attempting to get back to a preservation state, as shown earlier by the type and frequency of vitality states after a sequence of SVL (see Table 10), and (b) directly and locally reorganizing after an immediate sensation. In contrast, withdrawers (a) more fre- quently experienced negative physical sensations and (b) had more difficulty getting into a preservation mode and thus tended to remain in SVL. In the same vein, the finishers’ involve- ments during SVP appeared more focused not only on this experience but also on concerns about maintaining it during the race (Table 12). The two examples depicted in Figs 4 and 5 show that both the finishers and withdrawers experienced various difficulties expressed in the representamens that impacted their involve- ments. The finishers’ involvements indicated an overriding concern with preserving oneself in order to finish the race, specifically by refusing to focus on the performance itself after an expe- rienced SVL (Fig 4). The withdrawers’ involvements indicated various concerns about vitality issues and perceptions of being in difficulty and not being able to stay in preservation (Fig 5). Discussion The aim of this study was to characterize the distribution and temporal organization of the vitality states experienced by runners in a trail race and their adaptations to them, in order to distinguish finishers and withdrawers. Our results showed that the three vitality states emerged in all of them; however, the temporal organization of these experiences suggests that a situated vitality adaptation is a central point in determining whether a runner will finish or withdraw. We must remember that these three vitality states were considered as emerging at the level of Table 10. Types and frequency of vitality states after a sequence of SVL among finishers and withdrawers. SVR SVP SVL Finishers Withdrawers Finishers Withdrawers Finishers Withdrawers % 25.8 18.09 66.12 40 6.45 24.76 doi:10.1371/journal.pone.0173667.t010 Table 11. Types of representamens and involvements in finishers and withdrawers in a state of vitality loss. Finishers Withdrawers Representamens Involvements Representamens Involvements Gastric pains Being careful with pace and food Gastric pains Hoping it will pass Muscle cramps and pain Seeking preservation Muscle cramps and pain Trying to hold on, overcoming this state Fatigue Having a break Cold Trying to warm up Stress Trying to relax Hunger Should supply Feeling of difficulty Not focusing on the performance, just on finishing Fatigue Hoping to get better Foot pain Adapting the stride Foot pain Adapting the stride Bad mood Trying to stay positive Bad mood, negative emotions Hoping for a better moment to come or thoughts of abandoning Difficulties of the environmental conditions Trying to cope and hold on Difficulties of the environmental conditions Suffering, thinking of withdrawing doi:10.1371/journal.pone.0173667.t011 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 14 / 24 the athletes’ experience in relation to what they enacted during their races. They were mean- ingful parts of the stream of their sense of what happened when they were in the situations and, by performing such actions as accelerating, pacing, or sleeping, they enacted these worlds of feelings, appraisals and thoughts at their pre-reflective level of consciousness [28,29,83]. Having said that, our results clearly showed that finishers and withdrawers did not enact the same world in the race situation, and our in-depth discussion focuses on three points: (a) fin- ishing the race in a vitality preservation state, (b) adapting to a vitality loss state, and (c) the temporal organization of the vitality states. Finishing the race in a vitality preservation state Finishers privileged a world that corresponded to vitality preservation, whereas withdrawers spent less time in preservation and more in a state of vitality loss. These runners thus had to deal with the problem of not only creating a world of preservation but also maintaining it over the entire race in order to finish. We were able to observe how the succession of asymmetrical interactions between runners’ organization and the perturbations emerging from constraints in the race environment distinguished finishers from withdrawers. Finishers were able to pre- serve their own organization during a meaningful and significant part of the race [9], while withdrawers enacted a new organization in relation to these perturbations that protected them from the troublesome consequences but progressively excluded them from the race. Table 12. Types of representamens and involvements in finishers and withdrawers in a state of vitality preservation. Finishers Withdrawers Representamens Involvements Representamens Involvements Impression of running at a slow pace In spite of wanting to accelerate, set oneself to slow down/keep his pace Concerns about past injuries Preservation of physical integrity Good mood Enjoy each moment Too much time spent at the aid stations Careful with food and drink, reserves Beautiful landscapes Attempting to finish the race without getting hurt or too exhausted Medical procedures Getting healed Feeling relaxed Looking for recuperation Time barriers Following the pace of another racer Absence of stress or anxiety Split the race into smaller stages People encouraging Hoping to feel better Feeling of having a sustainable pace Should manage the entire race Being overtaken, others getting ahead Adapting the stride Carefulness Anticipate each potential difficulty Difficulty to have a regular pace Avoid getting into physical difficulty doi:10.1371/journal.pone.0173667.t012 Fig 4. Example of thematic analysis in a finisher. doi:10.1371/journal.pone.0173667.g004 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 15 / 24 Our results thus showed that preservation should be the privileged enacted world as it leads to finishing, and switches to loss or revival worlds during the race can have negative impacts on the outcome. Indeed, the finishers might have been able to maintain the enactment of a preservation vitality state because they were able to block the possibility of switching to one of these other worlds. Enacting a world of preservation also meant taking into account the poten- tial risk of overwhelming vitality loss by carefully and continuously monitoring to ensure that a new but ineffective enactment would not emerge. Previous publications [8,9,13] have docu- mented the specific management required for long-distance sports, including recovery time, and sleeping might be one of the performance factors. Hurdiel et al. [8] reported that runners made compromises between racing and resting by taking short naps, which let them complete the race in spite of the cognitive deficits that they observed. During a 4,856-kilometer cycling race, Lahart et al. [13] examined the consequences of sleep deprivation and energy deficiency on four cyclists’ emotions. They found that the cyclists managed less than one hour of continu- ous sleep per sleep episode: in addition to short sleep duration, inadequate energy intake led to unpleasant emotions and difficulty in regulating them. Moreover, the authors showed that actual sleep and sleep efficiency were better maintained during longer rest periods, highlight- ing the importance of a race strategy that optimizes the balance between average cycling veloc- ity and sleep time. They suggested that cyclists should: (a) have a plan prepared in advance to ensure sufficient sleep and recovery, (b) develop nutritional strategies to maintain energy intake and thus reduce energy deficits, and (c) anticipate the deleterious effects of sleep depri- vation to be able to appropriately respond to unexpected stressors [16]. Our results are in line with these suggestions, which address a broad preservation issue in endurance sports, and expand on them by showing how finishers enacted a world of preservation that also curbed the emergence of the revival option. Here, our thematic analysis of runners’ adaptations suggests that finishers are able to control their propensity to accelerate, even though their immediate feelings might be good, in contrast to withdrawers. Hence, these findings provide further insight into the organization of vitality adaptations and underscore the key role of preserva- tion, a critical factor in finishing ultra-long races. It seems less important to be able to enact a new world after a difficult period than to be able to maintain a preservation state once it is enacted. Last, our results also showed that the differences between finishers and withdrawers in rela- tion to a preservation strategy are particularly important when the race is shorter. This result appears counterintuitive at first view if we assume that race difficulty is directly linked to dis- tance. However, although no causal link between personality factors and ultra-race participa- tion was found [12], we might interpret this result as indicating that runners in ultra-races are more skillful and pay more attention to their pace and the supply procedures that are vital for finishing the race. This assumption of pace management is in line with the results of Lambert Fig 5. Example of thematic analysis in a withdrawer. doi:10.1371/journal.pone.0173667.g005 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 16 / 24 et al. [84], who demonstrated that the faster runners in a 100-kilometer race were able to main- tain their initial pace for a longer time compared with the slower runners, who showed greater variation in their pace, which also decreased more rapidly. In contrast, other types of runners, including beginners, might run shorter races but still fall into the traps that would be avoided with sufficient self-awareness. Highly self-aware runners are very careful and pay attention to their physical alerts, fitness and how they are feeling during the run [33]. Interestingly, a study on nutrient intake showed that most amateur runners did not meet their energy intake and nutritional requirements during a mountain marathon [85]. Moreover, we can assume that preservation concerns are stronger when the race is long or known for its difficulty, and the need for preserving oneself during a short race might be underestimated, especially for inexpe- rienced or inadequately trained runners, in line with the finding of a study on effort regulation in rowing between elites and sub-elites [86]. Therefore, runners’ preparation should also be explored and considered as one of the factors of race completion. For example, Krouse et al. [6] investigated female ultra-runners’ training practices and found empirical evidence of self- regulated training practices, such as using their own experience, blogs and websites to com- plete their training knowledge. Yet, such self-regulated practices might generate incomplete knowledge about the types of preparation needed for these races. Adaptation to a vitality loss state Nevertheless, because withdrawers more frequently enacted longer vitality loss worlds than those of finishers, it is also possible that they became stuck in this state as they were unable to enact an exit. From this perspective, the capacity to enact a new world may be a determi- nant of outcome. Our thematic analysis of runners’ adaptations confirmed this difference between finishers and withdrawers: Finishers rapidly modified their mode of involvement when this world emerged, whereas withdrawers appeared to focus on their feelings of dis- comfort. In short, withdrawers contemplated their difficulties, whereas finishers tried to find a better world by enacting local adaptions in response to perturbations. Thus, although both finishers and withdrawers felt vitality loss during the race, as already shown in previ- ous research on ultra-marathons [10,16], finishers enacted new meanings and put aside their difficulties by immediately trying to find solutions to change the world of feelings they were in. Our results therefore also confirm the interpretation that the capacity to immediately enact a new world when feelings of difficulties appear helps to overcome the temptation of race withdrawal. This agrees with the findings of a study on the variation in emotions throughout a multi-stage race regarding the importance of adaptive psychologi- cal states [87]: put differently, runners should pay attention to and interpret their emo- tions, using them as a guide for adapting their activity and thereby ensuring the emergence of a better state for carrying on in the race. How do withdrawers enact a world of vitality loss? Our results showed that repeated experiences of states of vitality loss were associated with withdrawal, contrasting with the repeated experiences of states of vitality preservation observed for finishers. One interpre- tation is that the more often an individual enacts a type of world, the easier it becomes to maintain that world, despite any perturbations that may arise. When finishers enacted a more continuous world of preservation, they ensured and reinforced satisfactory levels of relative comfort and economical organization compatible with the race duration [9,33]. In contrast, repeatedly enacting a world of vitality loss reinforced its impact, increased its degree, and progressively led to a pressing need to stop this world from developing further: withdrawers then enacted a new world in an attempt to preserve their long-term viability as ultra-runners. Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 17 / 24 Temporal organization of the vitality states Our results revealed that very early on, in about the first quarter of the race, the differences in the states of preservation and loss between future finishers and withdrawers were significant. This suggests that the race outcome (i.e., finish or withdraw) began to take shape relatively early on and was sensitive to the runners’ initial vitality states. Thus, we can argue that these results can be interpreted as relying on the temporal chain of these states, and particularly on the cumulative effect of the succession of vitality states. When runners began a race and soon after experienced a state of vitality loss, this early experience had a more powerful impact on the following states of vitality loss, which all the runners encountered. Here, a cumulative effect increased the differences we observed between finishers and withdrawers over the course of the race. Each successive state of vitality loss was immediately experienced more powerfully, impacting more negatively on the runners’ overall experience and affecting them more pro- foundly. In contrast, when finishers, who experienced fewer states of vitality loss, encountered this type of difficulty, they were still feeling sufficiently well and thus had the psychological resources to enact a new state. Experienced states of vitality preservation played the same role but in an opposite direction: the feeling of preservation kept the runners in a state of regular rhythmicity/pacing, and because they had found a comfortable way to run, the kilometers seemed to pass easier and the distance to run seemed less daunting. This phenomenon has already been reported in long-distance walking, during which a cumulative effect of walkers’ positive feelings and emotion increased throughout the duration of the walk [88]. This phenomenon does not rely only on pacing, however, because it is part of a more global experience of running that is made up of many different feelings (e.g., [4,16]). None of the withdrawers found a stable state of preservation, but instead moved from one state to another. This irregularity also explained why they did not find a stable state of relative ease, which would have helped them to continue the race. Instead, the differences with finishers increased throughout the race. Methodological issues and limitations Some methodological aspects of this study should be underlined. The difference in the number of sequences resulting from our coding of the data from the blogs and EIs was not significant, suggesting that we used comparable narratives to document the experience of vitality states. This result is in accordance with Bargh et al. [62] and Jones & Alony [63], who claimed that accounts posted on blogs could be used to obtain valuable data on the personal experience that marked people’s minds. Furthermore, the data extracted from the blogs were considered as primary data, which by definition are not influenced by the researcher’s intervention. The rela- tive anonymity of the blog posts (e.g., use of pseudonyms) is thought to facilitate the expres- sion of what the authors called the “true self” [62]. Therefore, researchers may well be able to access real lived experience. Of course, for this study, we selected specific blog post narratives that rendered this type of analysis possible. Indeed, not all the narratives were adapted for this kind of analysis, because some of them contained inaccurate information, some were humor- ous narrations, and others were reports about other runners’ activity. A key strength of our data is that we were able to distinguish the states of vitality revival, preservation and loss that were then restored in the chronological logic of the trail runners’ experience. We were also able to document the contents of these vitality states in finishers and withdrawers: thanks to their courses of experience depicted in their narratives, we were able to understand more deeply how they continuously organized their activity. This perspective is not completely new, but it provides a way to link experiences in other domains of human activity, such as effort, pain, and feelings of ease, in a succession of states, Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 18 / 24 which is innovative in exploring human activity as it is experienced. This research also suggests the need for further reflection on Bruner’s question about the purpose of narrative analysis and whether it should be focused only on specific and singular events in the precise situation in which they occur or whether there are realities common to all narrations [56]. Indeed, in the example of the two races (i.e., the “CCC” and the “Nivolet Revard” races) in which several run- ners participated (Table 3), we sometimes observed common representamens but, although the runners all had singular experiences, these common representamens did not necessarily have the same impact on their experience. This study had some methodological consequences. We were able to validate a method of data processing that was less time-consuming than EIs by using high-value narratives directly available on the Internet that portrayed a significant part of trail running experience. Indeed, social media provide a platform on which the sense of the trail running community is impor- tant in terms of experience-sharing and race preparation [4]. A study limitation that bears mentioning concerns the qualitative approach: some of our data were collected during post-race interviews, which inherently raises the issue of retrospec- tive recall [89]. However, as noted, we took great care to minimize the effects of retrospective recall by systematically keeping the runners in a re-enactment process. Also, the question of post-race judgment should be addressed: one might argue that the finishers displayed better judgment due to their successful completion of the race. However, our methodological design aimed to reduce this risk because the runners were asked to avoid judgment and to focus on the stream of situated experience. Furthermore, the direct relationship between race outcome and positive/negative judgment about the race is in itself debatable; indeed, some of the finishers were not satisfied with their race performance, whereas some of the withdrawers minimized negative judgments by stating that the decision to quit was the right one [32]. In addition, despite the difficult moments, the withdrawers also mentioned very positive moments with good sensations. Another limitation has to do with the characterization of the vitality states as discrete. They were presented as a temporal chain, with clear distinctions between them, as this was a neces- sary step in constructing valid quantitative and qualitative analyses. However, it is quite likely that in a real race vitality states are far less clearly delineated, with states emerging and being experienced more progressively. In this respect, although coding rendered our data clearer, the discretization was also somewhat artificial to highlight the shifts in the runners’ experience through the changes in the representamens and involvements identified in the coding. Yet it is important to note that this procedure is current in research that analyzes the stream of experi- ence using the Experience Sampling Method (e.g., [90]) or the Day Reconstruction Method (e.g., [91]). Also, although we did our best to ensure the accuracy of the experience shifts, we assume that during the race, these changes in the representamens and involvements that the runners were able to report emerged sufficiently strongly in their experience, reducing the fine grain analysis of the shifts. Conclusion This study showed that the notions of (a) seeking preservation, (b) making a good start, (c) delaying the emergence of a state of vitality loss, and (d) being able to exit a state of vitality loss may enrich our understanding of the factors that determine a runner’s ability to finish an extreme race (generally perceived as the capacity for self-surpassing). The notion of self-sur- passing might be real when runners remain in a state of vitality loss, especially when they expe- rience suffering without trying to enact a new world. Last, our results suggested that the main Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 19 / 24 challenge for runners is to avoid entering into this state: the more they are able to remain in a state of preservation, the more likely they are to finish. Acknowledgments We would like to thank Mickael Vauthier for his contribution to the collection of EI data. Author Contributions Conceptualization: NR DH LS. Data curation: NR DH LS. Formal analysis: FCR NR DH. Funding acquisition: NR DH LS. Investigation: NR DH LS RAP. Methodology: NR DH LS. Project administration: NR DH LS. Supervision: NR DH LS. Validation: NR DH LS RAP FCR. Visualization: NR DH LS. Writing – original draft: NR DH LS. Writing – review & editing: NR DH LS. References 1. Hoffman MD, Ong JC, Wang G. Historical Analysis of participation in 161 km Ultramarathons in North America. Int J Hist Sport. 2010; 27:1877–1891. doi: 10.1080/09523367.2010.494385 PMID: 20684085 2. Cejka N, Ru¨st CA, Lepers R, Onywera V, Rosemann T, Knechtle B. Participation and performance trends in 100-km ultra-marathons worldwide. J Sports Sci. 2014; 32(4):354–366. doi: 10.1080/ 02640414.2013.825729 PMID: 24015856 3. Hoffman MD, Fogard K. Demographic characteristics of 161-km ultramarathon runners. Res Sports Med Int J. 2012; 20(1):59–69. 4. Simpson D, Young G, Jensen PR. “It’s not about taking the easy road”: The experiences of ultramara- thon runners. Sport Psychol. 2014; 28:176–185. 5. Le Breton D. Conduites à risque. Presses Universitaires de France. Paris; 2002. French. 6. Krouse RZ, Ransdell LB, Lucas SM, Pritchard ME. Motivation, goal orientation, coaching, and training habits of women ultrarunners: J Strength Cond Res. 2011; 25(10):2835–2842. doi: 10.1519/JSC. 0b013e318204caa0 PMID: 21946910 7. Hespanhol LC Junior, Pena Costa LO, Lopes AD. Previous injuries and some training characteristics predict running-related injuries in recreational runners: a prospective cohort study. J Physiother. 2013; 59:263–269. doi: 10.1016/S1836-9553(13)70203-0 PMID: 24287220 8. Hurdiel R, Peze´ T, Daugherty J, Girard J, Poussel M, Poletti L, et al. Combined effects of sleep depriva- tion and strenuous exercise on cognitive performances during The North Face® Ultra Trail du Mont Blanc® (UTMB®). J Sports Sci. 2015; 33(7):670–674. doi: 10.1080/02640414.2014.960883 PMID: 25333827 9. Saugy J, Place N, Millet GY, Degache F, Schena F, Millet GP. Alterations of neuromuscular function after the world’s most challenging mountain ultra-marathon. Hug F, editor. PLoS ONE. 2013; 8:e65596. doi: 10.1371/journal.pone.0065596 PMID: 23840345 10. Tharion WJ, Strowman SR, Rauch TM. Profile and changes in moods of ultramarathoners. J Sport Exerc Psychol. 1988; 10(2):229–235. Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 20 / 24 11. Cona G, Cavazzana A, Paoli A, Marcolin G, Grainer A, Bisiacchi PS. It’s a matter of mind! Cognitive functioning predicts the athletic performance in ultra-marathon runners. Di Pellegrino G, editor. PLoS ONE. 2015; 10:e0132943. doi: 10.1371/journal.pone.0132943 PMID: 26172546 12. Hughes S, Case HS, Stuempfle K, Evans D. Personality profiles of Iditasport Ultra-marathon partici- pants. J Appl Sport Psychol. 2003; 15(3):256–261. 13. Lahart IM, Lane AM, Hulton A, Williams K, Godfrey R, Pedlar C, et al. Challenges in maintaining emo- tion regulation in a sleep and energy deprived state induced by the 4800Km ultra-endurance bicycle race; The Race Across AMerica (RAAM). J Sports Sci Med. 2013; 12:481–488. PMID: 24149155 14. Carmona G, Roca E, Guerrero M, Cusso´ R, Irurtia A, Nescolarde L, et al. Sarcomere disruptions of slow fiber resulting from mountain ultramarathon. Int J Sports Physiol Perform. 2015; 10:1041–1047. doi: 10. 1123/ijspp.2014-0267 PMID: 25848839 15. Rousanoglou EN, Noutsos K, Pappas A, Bogdanis G, Vagenas G, Bayios IA, et al. Alterations of vertical jump mechanics after a half-marathon mountain running race. J Sports Sci Med. 2016; 15:277–286. PMID: 27274665 16. Holt NL, Lee H, Kim Y, Klein K. Exploring experiences of running an ultramarathon. Sport Psychol. 2014; 28:22–35. 17. Varela F, Thompson E, Rosch E. The embodied mind: Cognitive mind and human experience. MIT Press: Cambridge; 1991. 18. Gallagher S. Mutual enlightenment: recent phenomenology in cognitive science. J Conscious Stud. 1997; 4:195–214. 19. McGann M, De Jaegher H, Di Paolo EA. Enaction and psychology. Rev Gen Psychol. 2013; 17:203– 209. 20. Norman DA. Psychology of everyday things. New York: Basic Books; 1988. 21. Thompson E. Sensorimotor subjectivity and the enactive approach to experience. Phenomenol Cogn Sci. 2005; 4:407–427. 22. Gibbs RW. Embodiment and Cognitive Science. Cambridge University Press; 2005. 23. Froese T, Di Paolo EA. The enactive approach: Theoretical sketches from cell to society. Pragmat Cogn. 2011; 19:1–36. 24. Langdridge D. Phenomenological Psychology: Theory, research and method. Pearson Education; 2007. 25. Davidsen AS. Phenomenological approaches in psychology and health sciences. Qual Res Psychol. 2013; 10:318–339. doi: 10.1080/14780887.2011.608466 PMID: 23606810 26. Mangan B. Taking phenomenology seriously: The “Fringe” and its implications for Cognitive Research. Conscious Cogn. 1993; 2:89–108. 27. Froese T, Fuchs T. The extended body: A case study in the neurophenomenology of social interaction. Phenomenol Cogn Sci. 2012; 11:205–235. 28. Legrand D. The bodily self: the sensori-motor roots of pre-reflective self-consciousness. Phenomenol Cogn Sci. 2006; 5(1):89–118. 29. Damasio A. The feeling of what happens: Body and emotion in the making of consciousness. Fort Worth, TX, US: Harcourt College Publishers; 1999. 30. Stewart JR, Gapenne O, Di Paolo EA. Enaction: Toward a new paradigm for cognitive ccience. MIT Press; 2010. 31. Di Paolo EA, Rohde M, De Jaegher H. Horizons for the enactive mind: Values, social interaction, and play. Enaction: Toward a new paradigm for cognitive science. MIT Press. London: Stewart J, Gapenne O, Di Paolo Ea; 2011. pp. 33–88. 32. Antonini Philippe R, Rochat N, Vauthier M, Hauw D. The story of withdrawals during an ultra-trail run- ning race: A qualitative investigation of runners’ courses of experience. Sport Psychol. 2016;1–43. 33. Johnson U, Kentta¨ G, Ivarsson A, Alvmyren I, Karlsson M. An ultra-runner’s experience of physical and emotional challenges during a 10-week continental run. Int J Sport Exerc Psychol. 2015; 14(1):72–84. 34. Hauw D, Rochat N, Gesbert V, Astolfi T, Philippe RA, Mariani B. Putting together first- and third-person approaches for sport activity analysis: The case of ultra-trail runners’ performance analysis. In: Salmon P, Macquet A-C, editors. Advances in Human Factors in Sports and Outdoor Recreation. Cham: Springer International Publishing; 2017. pp. 49–58. 35. Ryan RM, Frederick C. On energy, personality, and health: Subjective vitality as a dynamic reflection of well-being. J Pers. 1997; 65:529–565. PMID: 9327588 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 21 / 24 36. Lekes N, Houlfort N, Milyavskaya M, Hope NH, Koestner R. The role of intrinsic values for self-growth and community contribution at different life stages: Differentially predicting the vitality of university stu- dents and teachers over one year. Personal Individ Differ. 2016; 98:48–52. 37. Akin U, Akin A, Uğur E. Mediating role of mindfulness on the associations of friendship quality and sub- jective vitality. Psychol Rep. 2016; 119:516–526. doi: 10.1177/0033294116661273 PMID: 27469364 38. Bostic TJ, Rubio DM, Hood M. A validation of the subjective vitality scale using structural equation modeling. Soc Indic Res. 2000; 52:313–324. 39. Tversky B, Zacks JM, Hard BM. Understanding events: From perception to action. Oxford University Press, USA; 2008. 40. Hoffman MD, Fogard K. Factors related to successful completion of a 161-km ultramarathon. Int J Sports Physiol Perform. 2011; 6:25–37. PMID: 21487147 41. Froese T, Gould C, Seth AK. Validating and calibrating first- and second-person methods in the science of consciousness. J Conscious Stud. 2011; 18(2):38–64. 42. Olivares FA, Vargas E, Fuentes C, Martinez-Pernea D, Canales-Johnson A. Neurophenomenology revisited: Second-person methods for the study of human consciousness. Front Psychol. 2015; 6:673. doi: 10.3389/fpsyg.2015.00673 PMID: 26074839 43. Theureau J. Le cours d’action: l’enaction & l’expe´rience. Toulouse: Octares; 2015. French. 44. Theureau J. Le cours d’action: me´thode de´veloppe´e. Toulouse: Octares Editions; 2006. French. 45. Hauw D, Durand M. Temporal dynamics of acrobatic activity: An approach of elite athletes specious present. J Sports Sci Med. 2008; 7(1):8–14. PMID: 24150128 46. Theureau J. Le cours d’action: me´thode e´le´mentaire. 2nd ed. Toulouse: Octares Editions; 2004. French. 47. Sève C, Nordez A, Poizat G, Saury J. Performance analysis in sport: Contributions from a joint analysis of athletes’ experience and biomechanical indicators. Scand J Med Sci Sports. 2013; 23(5):576–584. doi: 10.1111/j.1600-0838.2011.01421.x PMID: 22150999 48. Seifert L, Wattebled L, Herault R, Poizat G, Ade´ D, Gal-Petitfaux N, et al. Neurobiological degeneracy and affordance perception support functional intra-individual variability of inter-Limb coordination during ice climbing. Alway SE, editor. PLoS ONE. 2014; 9:e89865. doi: 10.1371/journal.pone.0089865 PMID: 24587084 49. Theureau J. Course-of-action analysis and course-of-action centered design. In: Hollnagel E, editor. Handbook of cognitive task design. Mahwah, NJ: Lawrence Erlbaum Associates; 2003. p. 55–81. 50. Hauw D. L’entre´e «activite´» pour l’analyse des techniques et des performances sportives des athlètes de haut niveau. Bull Psychol. 2009; 502:365–372. French. 51. Hauw D, Bilard J. Situated activity analysis of elite track and field athletes’ use of prohibited perfor- mance-enhancing substances. J Subst Use. 2012; 17(2):183–197. 52. Theureau J. Les entretiens d’autoconfrontation et de remise en situation par les traces mate´rielles et le programme de recherche «cours d’action». Rev Anthropol Connaiss. 2010; 4(2):287–322. French. 53. Mohamed S, Favrod V, Philippe RA, Hauw D. The situated management of safety during risky Sport: Learning from skydivers’ courses of experience. J Sports Sci Med. 2015; 14:340–346. PMID: 25983583 54. Hauw D, Durand M. Situated analysis of elite trampolinists’ problems in competition using retrospective interviews. J Sports Sci. 2007; 25(2):173–183. doi: 10.1080/02640410600624269 PMID: 17127592 55. Baker J, Coˆte´ J, Deakin J. Cognitive characteristics of expert, middle of the pack, and back of the pack ultra-endurance triathletes. Psychol Sport Exerc. 2005; 6:551–558. 56. Bruner J. Acts of meaning. Cambridge, Mass.: Harvard University Press; 2000. 57. Adler JM, McAdams DP. Time, culture, and stories of the self. Psychological Inq. 2007; 18 (22):97–99. 58. Propp V, Meletinski E, Derrida M, Todorov T, Kahn C. Morphologie du conte. Paris: Le Seuil; 1973. French. 59. Greimas AJ. La description de la signification et la mythologie compare´e. L’Homme. 1963; 3(3):51–66. French. 60. Bortree DS. Presentation of self on the Web: An ethnographic study of teenage girls’ weblogs. Educ Commun Inf. 2005; 5(1):25–39. 61. Ducate L, Lomicka L. Adventures in the blogosphere: from blog readers to blog writers. Comput Assist Lang Learn. 2008; 21(1):9–28. 62. Bargh JA, McKenna KYA, Fitzsimons GM. Can you see the real me? Activation and expression of the “True Self” on the internet. J Soc Issues. 2002; 58(1):33–48. Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 22 / 24 63. Jones M, Alony I. Blogs–The new source of data analysis. Issues Informing Sci Inf Technol. 2008; 5:433–446. 64. Popova YB. Narrativity and enaction: the social nature of literary narrative understanding. Front Psy- chol. 2014; 5. 65. Owton H, Allen-Collinson J, Siriwardena N. Using a narrative approach to enhance clinical care for patients with asthma. Chest. 2015; 148(1):288–293. doi: 10.1378/chest.14-2630 PMID: 25789925 66. Evans AB, Sleap M. Older adults’ lifelong embodied experiences of leisure time aquatic physical activity in the United Kingdom. Leis Stud. 2015; 34(3):335–353. 67. McMahon J, Penney D. Body pedagogies, coaching and culture: three Australian swimmers’ lived expe- riences. Phys Educ Sport Pedagogy. 2013; 18(3):317–335. 68. Block BA, Weatherford GM. Narrative research methodologies: Learning lessons from disabilities research. Quest. 2013; 65(4):498–514. 69. Carless D, Douglas K. “In the boat” but “selling myself short”: Stories, narratives, and identity develop- ment in elite sport. Sport Psychol. 2013; 27:27–39. 70. Ezzyat Y, Davachi L. What constitutes an episode in episodic memory? Psychol Sci. 2011; 22(2):243– 252. doi: 10.1177/0956797610393742 PMID: 21178116 71. Magliano JP, Radvansky GA, Forsythe JC, Copeland DE. Event segmentation during first-person con- tinuous events. J Cogn Psychol. 2014; 26(6):649–661. 72. Reimer JF, Radvansky GA, Lorsbach TC, Armendarez JJ. Event structure and cognitive control. J Exp Psychol Learn Mem Cogn. 2015; 41(5):1374–1387. doi: 10.1037/xlm0000105 PMID: 25603168 73. Zacks JM, Swallow KM. Event segmentation. Curr Dir Psychol Sci. 2007; 16(2):80–84. doi: 10.1111/j. 1467-8721.2007.00480.x PMID: 22468032 74. Rosch E. Principles of categorization. In Rosch E, Lloyd B, editors. Cognition and categorization. Hills- dale NJ: Lawrence Erlbaum Associates; 1978. p. 27–48. 75. Diagonale des Fous—Trail de Bourbon—Mascareignes [Internet]. [cited 20 Oct 2016]. Available: 76. Sève C, Saury J, Leblanc S, Durand M. Course-of-action theory in table tennis: a qualitative analysis of the knowledge used by three elite players during matches. European Rev Appl Psychol. 2005; 55 (3):45–155. 77. Starks H, Brown Trinidad S. Choose your method: A comparison of phenomenology, discourse analy- sis, and grounded Theory. Qual Health Res. 2007; 17:1372–1380. doi: 10.1177/1049732307307031 PMID: 18000076 78. Berteloot A, Trohel J, Sève C. Analyse se´miologique de l’activite´ d’un coureur de demi-fond en situation compe´titive. Staps. 2010; 4:7–23. French. 79. Leblanc S, Saury J, Sève C, Durand M, Theureau J. An analysis of a user’s exploration and learning of a multimedia instruction system. Comput Educ. 2001; 36:59–82. 80. Saury J, Ade´ D, Gal-Petitfaux N, Huet B, Sève C, Trohel J. Actions, significations et apprentissages en EPS. Une approche centre´e sur les cours d’expe´riences des e´lèves et des enseignants. Paris: Editions Revue E.P.S; 2013. French. Available: 81. Hauw D, Durand M. Elite athletes’ differentiated action in trampolining: A qualitative and situated analy- sis of different levels of performance using retrospective interviews. Percept Mot Skills. 2004; 98 (3c):1139–1152. 82. Vaismoradi M, Turunen H, Bondas T. Content analysis and thematic analysis: Implications for conduct- ing a qualitative descriptive study: Qualitative descriptive study. Nurs Health Sci. 2013; 15:398–405. doi: 10.1111/nhs.12048 PMID: 23480423 83. Legrand D. Pre-reflective self-as-subject from experiential and empirical perspectives. Conscious Cogn. 2007; 16(3):583–599. doi: 10.1016/j.concog.2007.04.002 PMID: 17533140 84. Lambert MI, Dugas JP, Kirkman MC, Mokone GG, Waldeck MR. Changes in running speeds in a 100 KM ultra-marathon race. J Sports Sci Med. 2004; 3:167–173. PMID: 24482594 85. Kruseman M, Bucher S, Bovard M, Kayser B, Bovier PA. Nutrient intake and performance during a mountain marathon: an observational study. Eur J Appl Physiol. 2005; 94:151–157. doi: 10.1007/ s00421-004-1234-y PMID: 15714291 86. Brown MR, Delau S, Desgorces FD. Effort regulation in rowing races depends on performance level and exercise mode. J Sci Med Sport. 2010; 13:613–617. doi: 10.1016/j.jsams.2010.01.002 PMID: 20227342 87. Lane AM, Wilson M. Emotions and trait emotional intelligence among ultra-endurance runners. J Sci Med Sport. 2011; 14(4):358–362. doi: 10.1016/j.jsams.2011.03.001 PMID: 21440500 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 23 / 24 88. Crust L, Keegan R, Piggott D, Swann C. Walking the walk: A phenomenological study of long distance walking. J Appl Sport Psychol. 2011; 23(3):243–262. 89. Tenenbaum G, Elran E. Congruence between actual and retrospective reports of emotions for pre- and postcompetition states. J Sport Exerc Psychol. 2003; 25:323–340. 90. Cerin E, Barnett A. A processual analysis of basic emotions and sources of concerns as they are lived before and after a competition. Psychol Sport Exerc. 2006; 7:287–307. 91. Kahneman D. A survey method for characterizing daily life experience: The day reconstruction method. Science. 2004; 306:1776–1780. doi: 10.1126/science.1103572 PMID: 15576620 Enaction of vitality states and performance in trail running PLOS ONE | DOI:10.1371/journal.pone.0173667 March 10, 2017 24 / 24

Comparison of vitality states of finishers and withdrawers in trail running: An enactive and phenomenological perspective.

03-10-2017

Rochat, Nadège,Hauw, Denis,Antonini Philippe, Roberta,Crettaz von Roten, Fabienne,Seifert, Ludovic

eng

PMC3874308

Hindawi Publishing Corporation The Scientific World Journal Volume 2013, Article ID 670217, 5 pages http://dx.doi.org/10.1155/2013/670217 Research Article Demographic Characteristics of World Class Jamaican Sprinters Rachael Irving,1 Vilma Charlton,2 Errol Morrison,3 Aldeam Facey,1 and Oral Buchanan1 1 Department of Basic Medical Sciences, Faculty of Medical Sciences, University of the West Indies, Mona, Kingston 6, Jamaica 2 Institute of Education, University of the West Indies, Kingston 6, Jamaica 3 University of Technology, Kingston 7, Jamaica Correspondence should be addressed to Rachael Irving; rachael.irving@uwimona.edu.jm Received 1 September 2013; Accepted 8 October 2013 Academic Editors: C. Y. Guezennec and T. Noakes Copyright © 2013 Rachael Irving et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. The dominance of Jamaican sprinters in international meets remains largely unexplained. Proposed explanations include demographics and favorable physiological characteristics. The aim of this study was to analyze the demographic characteristics of world class Jamaican sprinters. Questionnaires administered to 120 members of the Jamaican national team and 125 controls elicited information on place of birth, language, ethnicity, and distance and method of travel to school. Athletes were divided into three groups based on athletic disciplines: sprint (s: 100–400 m; 𝑛 = 80), jump and throw (j/t: jump and throw; 𝑛 = 25) and, middle distance (md: 800–3000 m; 𝑛 = 15). Frequency differences between groups were assessed using chi-square tests. Regional or county distribution of sprint differed from that of middle distance (𝑃 < 0.001) but not from that of jump and throw athletes (𝑃 = 0.24) and that of controls (𝑃 = 0.59). Sprint athletes predominately originated from the Surrey county (s = 46%, j/t = 37%, md = 17, C = 53%), whilst middle distance athletes exhibited excess from the Middlesex county (md = 60%). The language distribution of all groups showed uniformity with a predominance of English. A higher proportion of middle distance and jump and throw athletes walked to school (md = 80%, j/t = 52%, s = 10%, and C = 12%) and travelled greater distances to school. In conclusion, Jamaica’s success in sprinting may be related to environmental and social factors. 1. Introduction The success of Jamaicans in the sprint events during the decades of Olympic participation from 1948 to 2012 reached a crescendo at the Olympic Games in London in 2012. Jamaicans won 12 sprint medals and had a 1-2 finish in the men’s 100 m final, an 1-2-3 finish in the men’s 200 m final, and a gold medal in the women’s 100 m final. Jamaica has three of the world’s four fastest men at 100 m. Several studies have tried to explain the success of Jamaicans in the sprint events. Proposed mechanisms include favorable physiological characteristics that could be environmentally, regionally, or genetically determined [1]. Psychological pro- gramming also helps in molding sprinting talent; there is a culture of running in Jamaica with young children being actively involved in sprinting competitions [2]. Studies have compared the physiological characteristics of “black” and “white” athletes, reporting that the former have lower levels of blood and muscle lactate at a given exercise intensity [3] and a greater ability to tolerate higher fractional utiliza- tion of maximal oxygen uptake (VO2 max) [4]. Athletes of African descent have a higher percentage of fast-twitch muscle fibres, greater activity in the glycolytic, phosphagenic, and lactate dehydrogenase metabolic pathways, and greater rate of ventilation [1]. The limitation of extrapolating the findings of these studies is that these studies have classified groups based primarily on skin color without accounting for the fact that there are sometimes more differences within races than between; also these findings are not exclusive to athletes [5]. The influence of the compensatory sickle cell gene on oxygen transport and availability to the tissues is reported to give black athletes an advantage in sprinting [1]. It is postulated that the reduced availability coupled with the reduced myoglobin in the preponderant fast-twitch muscle fibers which are adapted for rapid energy (ATP) regeneration, all give a net outcome of muscle anatomi- cal and biochemical advantages which proffer a superior performance [1]. Performance-related genes, biomechanics 2 The Scientific World Journal and the environment have been implicated in elite sporting performance [6, 7]; however, no study has been done that specifically looks at the demographics of Jamaica’s world class sprinters. Jamaica was selected as the model for the present study as the country’s athletes have an unparalleled record of success at the international level dating as far back as 1952 when the men of the 4 × 400 meters relay team set a world record at the Olympics in Helsinski. Jamaica has a population of approximately 2.7 million people distributed in 3 counties consisting of 14 parishes. The country is English speaking. Jamaica was colonized by the English in 1634 and until 1962 it was under direct British rule. The country’s population according to a United Nations report [8] consists of predominate blacks whose ancestors originated from West and West-Central Africa [9]. To our knowledge, no study has attempted to trace the ethnic or environmental background of world class Jamaican sprinters and by demographics determine the possibility that they might share a common ethnic or environmental origin. The aim of this study therefore was to determine the demographic characteristics of world class Jamaican sprinters. The findings were then compared with those of the general (nonathletic) Jamaican population to determine whether the sprinters differ in demographics from the ordi- nary population. 2. Methods The study was approved by the University Hospital of the West Indies, Kingston, Jamaica Ethics Committee. Written informed consents were obtained from the 245 participants. The experimental procedures were in accordance with the policy statement of the American College of Sports Medicine. Participants comprised 120 elite athletes, many of whom were world and Olympic record holders, and 125 control participants. All the athletes had represented Jamaica at inter- national games. The control participants, 125 students from the G.C. Foster College and the University of the West Indies, Mona, were intended to be representative of the general Jamaican population (C: Controls, 𝑛 = 125). This group does not actively participate in sports at the professional or amateur level. The athletes were divided into three groups based on athletic disciplines: sprint (s: 100–400 m, 𝑛 = 80), jump and throw (j/t: jump and throw, 𝑛 = 25), and middle distance (md: 800–3000 m, 𝑛 = 15). Athletes in the sprints were truly elite athletes, regularly dominating in international sprint events; many were current or former world, Olympic, and Commonwealth record holders. Although Jamaica is not usually successful internationally in middle distance events (800–3000 m), these athletes were included in the study to investigate the possibility of disproportionate number of athletes originating from a particular geographical region being the result of an abundant prominence of athletics in that region. The questionnaires used were written in English and modeled off those used in two similar demographic studies done on world class athletes from Kenya and Ethiopia [10, 11]. Questions were simple and were explained to those who Key Counties Cornwall (Hanover, St. Elizabeth, Saint James, and Westmore land) Middlesex (Clarendon, Manchester, St. Ann, St. Catherine, and St. Mary) Surrey (Kingston, Portland, St. Andrew, and St. Thomas) Boundaries 0 Miles Parishes Counties 20 Figure 1: Parishes of Jamaica divided in the three counties: Corn- wall, Middlesex and Surrey. could not easily understand. The questions were designed to obtain the following information. Place of Birth. This was classified according to the 14 parishes (Figure 1) and three counties of Jamaica [12]. The intention was to identify particular regions with a disproportionate high number of athletes in response to reports that the majority of Jamaica’s most successful sprinters are from the county of Cornwall and in particular the parish of Trelawny. Spoken Language and That of Parents and Grandparents. This serves to provide information on ethnicity. A common language is often indicative of common origin, and a related language or a language of the same family indicates a common origin dating further back in time [13]. At present only two languages are used by most Jamaicans: English and Patois (Creole-English). Mode (Walk, Run, and Transport) and Distance Travelled to School (2 Km, 2–5 Km, 5–10 Km, 10–15 Km, and >15 Km). This was used to access the link between distance travelled to school and running success. 3. Data Analysis Contingency chi-squares using IBM SPSS Statistics 20 were performed using the Yates, correction factor in all occasions to identify frequency differences between groups given the low subject numbers in each field (place of birth, languages, ethnicity, mode, and distance travelled to school). Individual chi-squares were then performed to identify between which groups the differences lay (place of birth: df = 12, language: df = 1, ethnicity: df = 4, distance travell to school: df = 4 and method of travelled to school: df = 2). Statistical significance was defined as 𝑃 ≤ 0.05. The 14 parishes were collapsed into the three counties of Jamaica to allow for statistical analysis using contingency chi-squares. The Scientific World Journal 3 100 90 80 70 60 50 (%) 40 30 20 10 0 C j/t md s Groups Cornwall Middlesex Surrey Figure 2: Place of birth of athletes and controls. County distribution of controls did not differ from the sprint and jump and throw athletes (𝑃 = 0.59 and 0.23) but differed from middle distance athletes (𝑃 < 0.0001). The county distribution of middle distance athletes differed significantly from that of sprint athletes (𝑃 < 0.001). 4. Results 4.1. Place of Birth. County or regional distribution of sprint (s) and jump and throw athletes (j/t) did not differ from that of the controls (𝜒2 = 3.4, 𝑃 = 0.59 and 𝜒2 = 7.5 and 𝑃 = 0.23, resp.) but county distribution of controls differed significantly from that of middle distance athletes (𝜒2 = 40 and 𝑃 < 0.0001). The county distribution of sprint and middle distance athletes differed significantly (𝜒2 = 30, 𝑃 < 0.0001). County distribution of jump and throw athletes did not differ significantly from that of sprint athletes (𝜒2 = 6.1, 𝑃 = 0.24). Most of the sprint athletes were from the county of Surrey (46% versus 37% from Middlesex and 17% from Cornwall). The controls were mainly from Surrey (53%). There was a marked overrepresentation of middle distance athletes in the county of Middlesex (60% versus 40% from Cornwall and 0% from Surrey, see Figure 2). Only 12% of the control participants were from Cornwall. Jump and throw athletes were distributed across the three counties (Surrey: 30%, Middlesex: 40%, and Cornwall: 30%). 4.2. Language. The language spoken did not differ signifi- cantly (𝑃 > 0.05) amongst any of the groups (C versus s, C versus j/t, C versus md, s versus j/t, s versus md, and j/t versus md). 4.3. Mode of Travel to School. The mode of travelling to school did not differ between the controls and sprint athletes but differed slightly between the controls and jump and throw athletes (𝜒2 = 0.4, 𝑃 = 0.8 and 𝜒2 = 10.4, 𝑃 < 0.05, resp.); however, there was a significant difference between the controls and middle distance athletes (𝜒2 = 29.6, 𝑃 < 0.001). A significant difference was seen between the sprint Table 1: Mode of travel by groups. Groups Mode of travel to school Walk Bicycle Transport (bus, car) Jump and throw 10 (40%) 2 (8%) 13 (52%) Middle distance 12 (80%) 3 (20%) 0 Sprint 12 (10%) 0 108 (90%) Control 15 (12%) 0 110 (88%) The mode of travel to school did not differ between the controls and sprint athletes but differs slightly between the controls and jump and throw athletes (𝜒2 = 0.4, 𝑃 = 0.8 and 𝜒2 = 10.4 and 𝑃 < 0.05, resp.). There was a significant difference between the controls and middle distance athletes (𝜒2 = 29.6, 𝑃 = 0.001). There was also a significant difference between the sprint and middle distance athletes (𝜒2 = 32.1 and 𝑃 = 0.0001) and between the jump and throw and middle distance athletes (𝜒2 = 22.1, 𝑃 = 0.001). and middle distance athletes (𝜒2 = 32.1, 𝑃 < 0.0001) and between the jump and throw and middle distance athletes (𝜒2 = 22.1, 𝑃 < 0.001, see Table 1). 4.4. Distance Travelled to School. The distance travelled to school did not differ between the controls and sprint athletes (𝜒2 = 5.2, 𝑃 = 0.058) but differed between the controls and the jump and throw athletes (𝜒2 = 13.1, 𝑃 > 0.001). Seventy percent of the controls Travelled 2 kilometers or less to school (see Figure 3). Fifty five percent of the sprint athletes and 20% of the jump and throw athletes travelled ≤ 2 kilometres to school. The control and the middle distance athletes differed significantly in the distance travelled to school (𝜒2 = 23.4, 𝑃 < 0.001). Forty percent (40%) of the middle distance athletes travelled between 10–15 km to school and approximately 26.7% travelled >15 km to school. All the athletic groups differed significantly in the distance travelled to school (s versus j/t: 𝜒2 = 14.1, 𝑃 < 0.01; s versus md, 𝜒2 = 20.1, 𝑃 < 0.001; j/t versus md: 𝜒2 = 15.2, 𝑃 < 0.001). 5. Discussion The study showed that Jamaican sprinters are of similar environmental and ethnic background as ordinary Jamaicans or controls. The sprint athletes were distributed across the island with 46% from Surrey, 37% from Middlesex, and 17% from Cornwall. The controls originated from across the island with 53% originating from the Surrey county. There were no differences between the sprint athletes and the controls in the distance and the mode they travelled to school. Most use a private automobile or the public bus. No population stratification was identified between controls and sprint athletes, as seen in the areas where both groups reside. The jump and throw athletes showed a slight dominance in the county of Middlesex (40%); however, they were equally represented in Cornwall and Surrey (30% and 30%). There was a significant difference in the distance travelled to school between the sprint and jump and throw athletes (𝜒2 = 14.1, 𝑃 < 0.001). Approximately 60% of the jump 4 The Scientific World Journal 24% 4% 72% Control <2 km 2–5 km 5–10 km 10–15 km >15 km (a) 15% 55% 30% Sprint <2 km 2–5 km 5–10 km 10–15 km >15 km (b) Jump and throw 20% 20% 60% <2 km 2–5 km 5–10 km 10–15 km >15 km (c) 40% 27% 26% 7% Middle distance <2 km 2–5 km 5–10 km 10–15 km >15 km (d) Figure 3: Distances travelled to school. Charts showing percentage of participants and distances traveled to school daily. The jump and throw and middle distance athletes differed significantly from the sprint athletes and controls. and throw athletes travelled 5–10 kilometres to school and approximately 55% of the sprint athletes travelled ≤2 km to school. There was a slight difference between the jump and throw athletes and the controls in the mode travelling to school (𝑃 < 0.05). The middle distance runners seemed to be of a distinct environment or county background. Most of the middle distance runners were from the county of Middlesex (60%) which consists of the parishes of Clarendon, Manchester, St. Ann, St. Catherine, and St. Mary. Many of these parishes are deep rural with unreliable means of travel. Roads are often dirty and distances between schools are much greater than those in the county of Surrey which is more developed in terms of modern amenities. Another 40% of the middle distance athletes originated from Cornwall which has more mountainous parishes than Middlesex. Middlesex however forms part of the Blue Mountain range with highest point at about 2,250 meters [14]. No middle distance athletes originated from the urban Surrey but 53% of controls were from Surrey. Surrey also forms part of the Blue Mountain range but the controls and sprinters tend to originate from the valleys in the Surrey county. Environment and not ethnicity seemed to be the factor that differentiated the middle distance runners from the sprinters as both groups mainly spoke Cre- ole English and English. A common language is suggestive of same origin and a related language may also indicate a com- mon origin, but one that is older [13]. The middle distance runners were over represented in Middlesex and Cornwall which consist of mountainous areas. High altitude may be benefital in distance running [7]. This finding supports the theory that the success of the Kenyans in distance running in some way may be linked to the proximity of the Rift Valley as many of Kenyans distance runners are from the high altitude, Rift Valley region. Jamaican middle distance runners seemed to be a special set from the mountainous areas. The middle distance runners travelled longer distance to school (93% travelled > 5 km to school versus 60% of the jump and throw athletes and 15% of the sprint athletes) and were more likely and jump and throw athletes to travel to school via bicycle or walking than the sprinters. The significant differences between controls and middle distance athletes in regards to place of birth, distance travelled to school, and mode of The Scientific World Journal 5 travel suggested some link between place of birth and middle distance athletic ability. The finding that the middle distance athletes seem to be clustered in the deep rural parishes or counties of Jamaica may support the hypothesis of a link between mountain and endurance. The athletes from clusters in the deep rural parishes showed more African haplotypes than the general population [2]. When this study is compared to the findings that Kenyans who walked or run to school had VO2 max values that are 30% higher than those who did not walk or run [10] there is an implication that childhood endurance activity might be a determinant of middle distance athletic selection. 6. Conclusion The results showed that world class Jamaican sprinters have the same demographic profile as the general population. The middle distance athletes seem to have a distinct demographic profile for all variables except language. References [1] E. Y. S. A. Morrison and P. D. Cooper, “Some bio-medical mechanisms in athletic prowess,” West Indian Medical Journal, vol. 55, no. 3, pp. 205–209, 2006. [2] R. Irving and V. Charlton, Jamaican Gold: Jamaican Sprinters, University Press, Kingston, Jamaica, 2010. [3] A. R. Weston, O. Karamizrak, A. Smith, T. D. Noakes, and K. H. Myburgh, “African runners exhibit greater fatigue resistance, lower lactate accumulation, and higher oxidative enzyme activ- ity,” Journal of Applied Physiology, vol. 86, no. 3, pp. 915–923, 1999. [4] A. N. Bosch, B. R. Goslin, T. D. Noakes, and S. C. Dennis, “Physiological differences between black and white runners during a treadmill marathon,” European Journal of Applied Physiology and Occupational Physiology, vol. 61, no. 1-2, pp. 68– 72, 1990. [5] N. Yu, F.-C. Chen, S. Ota et al., “Larger genetic differences within Africans than between Africans and Eurasians,” Genetics, vol. 161, no. 1, pp. 269–274, 2002. [6] N. Yang, D. G. MacArthur, J. P. Gulbin et al., “ACTN3 genotype is associated with human elite athletic performance,” American Journal of Human Genetics, vol. 73, no. 3, pp. 627–631, 2003. [7] W. Schmidt, K. Heinicke, J. Rojas et al., “Blood volume and hemoglobin mass in endurance athletes from moderate alti- tude,” Medicine and Science in Sports and Exercise, vol. 34, no. 12, pp. 1934–1940, 2002. [8] United Nations, “International Convention on the elimination of all form of racial discrimination,” CERD/ C/ 383 /Add. 1, 2001. [9] R. A. Scott, R. Irving, L. Irwin et al., “ACTN3 and ACE genotypes in elite Jamaican and US sprinters,” Medicine and Science in Sports and Exercise, vol. 42, no. 1, pp. 107–112, 2010. [10] R. A. Scott, E. Georgiades, R. H. Wilson, W. H. Goodwin, B. Wolde, and Y. P. Pitsiladis, “Demographic characteristics of elite Ethiopian endurance runners,” Medicine and Science in Sports and Exercise, vol. 35, no. 10, pp. 1727–1732, 2003. [11] V. O. Onywera, R. A. Scott, M. K. Boit, and Y. P. Pitsiladis, “Demographic characteristics of elite Kenyan endurance run- ners,” Journal of Sport Sciences, vol. 22, no. 4, pp. 415–422, 2006. [12] STATIN, “Demographic Statistics: Population,” 2013, http:// www.statinja.gov.jm/population.aspx. [13] P. Pkalya and M. Aden, Conflicts in Kenya, ITDG, Nairobi, Kenya, 2003. [14] Gleaner, Geography & History of Jamaica, 2013, http:// www.discoverjamaica.com/gleaner/discover/geography/fea- tures.htm.

Demographic characteristics of world class Jamaican sprinters.

12-10-2013

Irving, Rachael,Charlton, Vilma,Morrison, Errol,Facey, Aldeam,Buchanan, Oral

eng

PMC9264511

Karam et al. BMC Anesthesiology (2022) 22:211 https://doi.org/10.1186/s12871-022-01757-8 RESEARCH Assessing the discriminative ability of the respiratory exchange ratio to detect hyperlactatemia during intermediate-to-high risk abdominal surgery Lydia Karam1, Olivier Desebbe2, Sean Coeckelenbergh1,3, Brenton Alexander4, Nicolas Colombo3, Edita Laukaityte1, Hung Pham1, Marc Lanteri Minet1, Leila Toubal1, Maya Moussa1, Salima Naili1, Jacques Duranteau1, Jean‑Louis Vincent5, Philippe Van der Linden6 and Alexandre Joosten1,3* Abstract Background: A mismatch between oxygen delivery (DO2) and consumption (VO2) is associated with increased perioperative morbidity and mortality. Hyperlactatemia is often used as an early screening tool, but this non‑continu‑ ous measurement requires intermittent arterial line sampling. Having a non‑invasive tool to rapidly detect inadequate DO2 is of great clinical relevance. The respiratory exchange ratio (RER) can be easily measured in all intubated patients and has been shown to predict postoperative complications. We therefore aimed to assess the discriminative ability of the RER to detect an inadequate DO2 as reflected by hyperlactatemia in patients having intermediate‑to‑high risk abdominal surgery. Methods: This historical cohort study included all consecutive patients who underwent intermediate‑to‑high risk surgery from January 1st, 2014, to April 30th, 2019 except those who did not have RER and/or arterial lactate meas‑ ured. Blood lactate levels were measured routinely at the beginning and end of surgery and RER was calculated at the same moment as the blood gas sampling. The present study tested the hypothesis that RER measured at the end of surgery could detect hyperlactatemia at that time. A receiver operating characteristic (ROC) curve was constructed to assess if RER calculated at the end of the surgery could detect hyperlactatemia. The chosen RER threshold corre‑ sponded to the highest value of the sum of the specificity and the sensitivity (Youden Index). Results: Among the 996 patients available in our study cohort, 941 were included and analyzed. The area under the ROC curve was 0.73 (95% CI: 0.70 to 0.76; p < 0.001), with a RER threshold of 0.75, allowing to discriminate a lac‑ tate > 1.5 mmol/L with a sensitivity of 87.5% and a specificity of 49.5%. Conclusion: In mechanically ventilated patients undergoing intermediate to high‑risk abdominal surgery, the RER had moderate discriminative abilities to detect hyperlactatemia. Increased values should prompt clinicians to inves‑ tigate for the presence of hyperlactatemia and treat any potential causes of DO2/VO2 mismatch as suggested by the subsequent presence of hyperlactatemia. © The Author(s) 2022. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. The Creative Commons Public Domain Dedication waiver (http:// creat iveco mmons. org/ publi cdoma in/ zero/1. 0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data. Open Access *Correspondence: joosten‑alexandre@hotmail.com 3 Department of Anesthesiology, Paul Brousse Hospital, 12 Avenue Paul Vaillant Couturier, 94800 Villejuif, France Full list of author information is available at the end of the article Page 2 of 8 Karam et al. BMC Anesthesiology (2022) 22:211 Background Patients undergoing abdominal surgery are at risk of an oxygen delivery (DO2) and oxygen consumption (VO2) mismatch, which can lead to postoperative complications [1]. Many studies have attempted to identify DO2/VO2 mismatch using various methods [2].  Although meas- urement of arterial lactate is considered standard of care for this purpose, this approach is intermittent and opti- mally requires an arterial catheter [3]. Although other measures, including the central venous oxygen saturation (ScvO2) [4], the veno-arterial CO2 gradient, [5] and the ratio of veno-arterial CO2 to arterio-venous oxygen dif- ference [6] have shown promise, they share the same lim- itations as lactate measurements (i.e., intermittent and invasive blood sampling) [7]. Monitoring of the respiratory exchange ratio (RER), can reflect the presence of anaerobic metabolism in the mechanically ventilated patient [1, 8]. This ratio increases in the presence of inadequate DO2 for a given VO2 and is computed using a standard anesthesia machine gas ana- lyzer that continuously measures inspiratory and expira- tory concentrations of O2 and CO2. The RER is calculated by dividing the difference in inspiratory and expiratory CO2 by that of inspiratory and expiratory O2 values (i.e., (FeCO2—FiCO2) / (FiO2—FeO2)), and has been shown to predict postoperative complications in abdominal sur- gery patients. In a swine model and in abdominal surgery patients, increased RER has been associated with intra- operative hyperlactatemia [1, 9]. However, the sensitivity and specificity of the RER to detect hyperlactatemia in humans remains to be determined. If an abnormal RER can detect hyperlactatemia, it could be used as a non- invasive indicator for more aggressive hemodynamic optimization (e.g., fluid, vasopressors, and blood admin- istration) and additional monitoring, especially as this variable can be easily calculated at the bedside during surgery. The present study tested the hypothesis that RER calculated at the end of surgery could help detect hyper- lactatemia in patients undergoing intermediate-to-high risk abdominal surgery. Methods The Academic Erasme University Hospital (Université Libre de Bruxelles) ethical committee approved this study December 4th, 2020 (SRB2020_654) and waived the need for informed consent. All methods were performed in accordance with the relevant guidelines and regulations. We report this study using STROBE guidelines. Patients were identified using TrackPro (UltraGenda, Belgium), their medical records checked with MediView (IMMJ Systems, United Kingdom), and intraoperative data retrieved from the anesthetic electronic records software (Innovian, Drager, Germany). All patients from our insti- tution aged 18 years or older were included if they: 1) Underwent elective intermediate-to-high-risk open abdominal surgery (including hepatobiliary surgery, pancreatectomy, gastrectomy, oesophagectomy, can- cer debulking, cystectomy, colectomy, nephrectomy, aorto-bifemoral bypass, abdominal aortic aneurysm surgery or other major abdominal surgery requir- ing a laparotomy) under general anesthesia between January 1st, 2014, and April 30th, 2019. Patients who had major vascular surgery were also included if the surgery involved an abdominal incision (e.g., aortobifemoral bypass and abdominal aortic aneu- rysm surgery). All of these surgeries were classified as intermediate- or high-risk surgeries according to Kristensen et al. [10] 2) Had routinely arterial blood gas, arterial lactate, FiO2, FiCO2, FeO2, and FeCO2 measurements done simul- taneously at the start and end of surgery. Impor- tantly, RER values are not displayed in real time on our intraoperative monitors but were calculated ret- rospectively using inhaled and exhaled O2 and CO2 values (electronic medical records). Exclusion criteria included pregnancy, lack of arterial blood gas analysis during surgery, emergency surgery, and laparoscopy. As arterial lactate values above 1.5  meq/L have been associated with increased mortality in multiple stud- ies examining surgical and critically ill patients and we wanted to establish a clinical cutoff value for optimal util- ity by most practicing physicians, [11–13] patients were split into two groups depending on the arterial lactate value recorded at the end of surgery: above 1.5  mEq/l (i.e., high lactate) or less than or equal to 1.5meEq/l (i.e., low lactate). While splitting patients into two groups is not ideal for a continuous variable, we felt this was important for increasing clinical utility and was therefore done for exploratory purposes. Anaesthesia protocol All patients had at least one large peripheral venous catheter and were monitored with standard American Society of Anesthesiology (ASA) monitoring (i.e., pulse Keywords: Tissue hypoxia, Anaerobic metabolism, Shock, Goal‑directed hemodynamic therapy Page 3 of 8 Karam et al. BMC Anesthesiology (2022) 22:211 oximetry, non-invasive blood pressure, 3 or 5 lead EKG, inhaled and expired gases, and temperature monitor- ing), and invasive blood pressure monitoring through a radial, femoral, or brachial artery catheter. Frontal elec- troencephalogram monitoring with the Bispectral index, hemodynamic pulse-contour analysis, central venous pressure, and other supplemental monitoring tools were used at the discretion of the attending anesthetist. Anesthesia administration was not standardized. Induction agents included propofol, etomidate, and ketamine. Opioids consisted of either remifentanil or sufentanil administration. Neuromuscular blockade was administered with succinylcholine, rocuronium, or cisa- tracurium. Maintenance of anesthesia was done with either sevoflurane, desflurane, or propofol. Adjuvant antinociception with spinal morphine, locoregional local anesthetics (e.g., epidural, transverse abdominal plane block, etc.), and opioid sparing agents (e.g., ketamine, dexmedetomidine, dexamethasone, lidocaine, paraceta- mol, diclofenac) were administered at the discretion of the attending anesthetist. Although no strict hemodynamic protocol was applied during surgical cases, anesthesiologists from our insti- tution traditionally use vasopressors to maintain mean arterial blood pressure (MAP) above 65  mmHg. The vasopressor of choice was norepinephrine, but both phe- nylephrine and ephedrine were less commonly used. Data collection and outcomes Patient baseline characteristics, intraoperative variables, postoperative major and minor complications, 30-day mortality, and 1-year mortality were collected. FiO2, FiCO2, FeO2, and FeCO2 were collected within the first hour and during the last hour of surgery at a moment of ventilator stability (i.e., no modifications greater than 50  ml in tidal volume) that coincided with an arterial blood gas analysis. The primary objective was to test the hypothesis that RER measured at the end of surgery could detect hyperlactatemia at that time. As there was also an interest to explore other possible associations between patients having low vs a high lactate values, we presented these data but they should be considered purely exploratory. Statistical analysis The Kolmogorov Smirnov test determined that distribu- tion was not normal (skewness of the different variables) and continuous variables were thus reported as median with quartiles [25th -75th percentile]. Discrete variables were reported as frequencies (with proportions). Our primary analysis was the estimation of the area under the receiver operating characteristics (AUROC) curve to establish discriminative properties of the RER to detect hyperlactatemia. We first fit a logistic regression model to estimate the ROC curve. We then estimate the AUROC according to the Delong et al. methodology and its 95% confidence intervals with the calculation of an exact Binomial Confidence Interval [14].  From the ROC curves, the optimal cut-off value yielding the greatest combined sensitivity and specificity was measured. We defined values within the 95% CI of the obtained thresh- old  value as an inconclusive (gray zone)  according to  Cannesson  et al. [15]. This gray zone approach is a zone of uncertainty which explores the clinical usefulness of the RER to detect hyperlactatemia. Statistical analysis was conducted with MedCalc® Statistical Software ver- sion 19.6.4 (MedCalc Software Ltd, Ostend, Belgium; https:// www. medca lc. org; 2021). A total of 996 patients undergoing intermediate- to- high risk abdominal surgery were eligible, of whom 55 were excluded due to missing data. Consequently, 941 patients were included, 622 being in the low lactate group and 319 in the high lactate group (Fig. 1). A RER of > 0.75 (Youden index) at the end of surgery detected a lactate value above 1.5  mEq/L with a sensi- tivity of 87.5% and a specificity of 49.5%. The area under the receiver operating characteristic curve was 0.730 (95% CI: 0.70 to 0.76; p < 0.001) (Fig. 2). Using a sensitivity of 90% and a specificity of 90%, four hundred and seven patients (43%) were in the grey zone defined from a RER of 0.72 to a RER of 0.98 (283 patients had a RER < 0.72 and 251 patients a RER > 0.98, thus having respectively a false negative rate and a false positive rate of 10% or below (relative high certainty)). RER was significantly higher in the high lactate group at the beginning and at the end of surgery (Table 2). Over- all, baseline characteristics were not different between groups except for body mass index, which was higher in the high lactate group; preoperative hemoglobin, which was lower in the high lactate group; preoperative aspirin use, which was more frequent in the low lactate group; and the type of surgery (Table 1). Intraoperative variables were significantly different between the two groups with respect to anesthesia and surgery times, fluids infused, blood products administered, net fluid balance, vasopres- sors, lactate, and RER, indicating more challenging intra- operative conditions in the high lactate group (Table 2). While these differences may indicate potential confound- ing factors, our intention was simply to determine if postoperative RER values could be useful to help detect increased arterial lactate levels, irrespective of differences in patients’ baseline clinical characteristics. Exploratory secondary objectives were not different between groups (Table 3). In mechanically ventilated patients undergoing inter- mediate to high-risk abdominal surgery, the RER had Page 4 of 8 Karam et al. BMC Anesthesiology (2022) 22:211 moderate discriminative properties to detect hyper- lactatemia. Based on a grey zone approach, 43% of the patients lied in an uncertainty zone with limited clinical usefulness. However, a RER value above 0.75 can detect hyperlactatemia with a relatively high sensitivity (88%). Hence, a normal RER makes hyperlactatemia relatively unlikely. Since its specificity is rather poor (50%) at this level, an increased RER cannot definitely rule in the presence of hyperlactatemia. Bar and colleagues recently demonstrated the poten- tial of RER to predict postoperative complications following both open abdominal high-risk and laparo- scopic surgeries [1, 8]. During high-risk open abdomi- nal surgery, both lactate and RER were found to predict postoperative complications, with an AUC of 0.77 and 0.67, respectively [1]. This confirms the importance of these measurements. Although this team did demon- strate an association between these two measurements, the sensitivity of specificity of RER to detect hyperlac- tatemia was not investigated [1]. Fig. 1 Flow Chart Fig. 2 Receiver operating characteristic curve to examine if the RER at the end of the surgery can detect hyperlactatemia Page 5 of 8 Karam et al. BMC Anesthesiology (2022) 22:211 In our study, patients in the high lactate group had more challenging intraoperative conditions as surgery time, fluid requirements, vasopressor infusions, and blood loss were all greater than in the low lactate group. The major postoperative complications of sepsis, peri- tonitis, and renal injury confirm that adverse postop- erative outcomes are associated with hyperlactatemia. Other variables have been related to tissue hypoperfu- sion, such as ScvO2, the veno-arterial CO2 gradient, the ratio of veno-arterial CO2 to arterio-venous oxy- gen. Unfortunately, these measures, similar to arterial lactate, require access to arterial, central venous, or even pulmonary artery blood sampling. RER has the advan- tage of being easy to calculate, non-invasive, free and its components continuously displayed on all modern anesthesia machines. It is easy to imagine a clinical sce- nario in which an elevated RER following surgery results in changes in post-operative clinical management. For example, the patient with an elevated RER may require a higher level of care, additional monitoring, supple- mentary laboratory tests or any other treatment that can correct DO2/VO2 mismatch. Such interventions could Table 1 Baseline characteristics Values are presented as medians [interquartiles ranges] or numbers (percentages %) a Included other laparotomies such as surrenalectomy or prostatectomy Variables Lactate ≤ 1.5 mEq/l (N = 622) Lactate > 1.5 mEq/l (N = 319) P-value Age (years) 66 [56–73] 64 [55—72] 0.078 Sex, Female (%) 220 (35.4%) 125 (39.2%) 0.258 BMI (kg/m2) 25.17 [22.5–28.86] 25.8 [23.31–29.4] 0.031 ASA score (1–2 / 3–4) 361 / 261 193 / 126 0.125 Preoperative hemoglobin (g/dL) 13.1 [11.8–14.3] 13.5 [12.1–14.5] 0.013 Preoperative creatinine (mg/dL) 0.9 [0.7–1.1] 0.9 [0.71–1] 0.850 Comorbidities; N (%) Ischemic heart disease Coronary artery bypass graft Hypertension Hyperlipidemia Stroke Atrial fibrillation Heart failure Diabetes mellitus 1 Diabetes mellitus 2 COPD Cirrhosis Asthma 63 (10.1%) 34 (5.5%) 319 (51.3%) 179 (28.8%) 25 (4.0%) 49 (7.9%) 10 (1.6%) 2 (0.3%) 138 (22.2%) 84 (13.5%) 43 (6.9%) 30 (4.8%) 24 (7.5%) 11(3.5%) 144 (45.1%) 86 (27.0%) 15 (4.7%) 20 (6.3%) 4 (1.3%) 2 (0.6%) 66 (20.7%) 30 (9.4%) 30 (9.4%) 19 (6.0%) 0.192 0.170 0.074 0.557 0.623 0.370 0.669 0.495 0.598 0.068 0.176 0.459 Medications; N (%) Aspirin Clopidogrel ẞ blocker ACEI ARB Calcium channel blocker Diuretics Statin Oral hypoglycaemic drugs Insulin Oral anticoagulation 224 (36.0%) 30 (4.8%) 172 (27.7%) 136 (21.9%) 50 (8.0%) 119 (19.1%) 65 (10.5%) 181 (29.1%) 94 (15.1%) 49 (7.9%) 64 (10.3%) 85 (26.7%) 8 (2.51%) 81 (25.4%) 57 (17.9%) 22 (6.9%) 49 (15.4%) 25 (7.8%) 91 (28.5%) 47 (14.7%) 27 (8.5%) 26 (8.2%) 0.004 0.088 0.459 0.151 0.533 0.153 0.197 0.854 0.877 0.755 0.291 Type of Surgery (N) < 0.001 Pancreatectomy Hepatobiliary Oesophagectomy Cystectomy Cancer debulking Major vascular surgery Gastrectomy Colectomy Nephrectomy Other surgical procedurea 104 (16.7%) 161 (25.9%) 82 (13.2%) 72 (11.6%) 25 (4.0%) 141 (22.7%) 7 (1.1%) 17 (2.7%) 7 (1.1%) 6 (0.1%) 58 (18.2%) 152 (47.7%) 37 (11.6%) 18 (5.6%) 10 (3.1%) 32 (10.0%) 1 (0.3%) 6 (1.9%) 2 (0.6%) 3 (0.9%) Page 6 of 8 Karam et al. BMC Anesthesiology (2022) 22:211 Table 2 Intraoperative Variables Values are presented as medians [interquartiles ranges] or numbers (percentages %) a use of any vasopressor (ephedrine, phenylephrine, noradrenaline) b total colloid included 3% gelatin and 6% tetrastarch Variables Lactate ≤ 1.5 mEq/l (N = 622) Lactate > 1.5 mEq/l (N = 319) P-value Anaesthesia duration ( min) 347 [254–450] 372 [287–472] 0.001 Surgery duration (min) 262 [180–360] 289 [209–381] 0.002 Total crystalloid (ml) 2000 [1200–3000] 2400 [1500–3500] < 0.001 Total colloid (ml)b 500 [500–1000] 1000 [500–1500] < 0.001 Total blood product (ml) 498 [261–580] 540 [288–1475] 0.009 Total IN (ml) 2500 [1500–3500] 3000 [2000–4500] < 0.001 Estimated blood loss (ml) 400 [200–900] 700 [300–1500] < 0.001 Diuresis (ml) 300 [150–500] 300 [150–510] 0.258 Total out (ml) 870 [500–1400] 1110 [680–2000] < 0.001 Net fluid balance (ml) 1505 [788–2350] 1750 [1000–2800] < 0.001 Use of vasopressors, N (%)a 484 (77.9%) 269 (84.3%) 0.02 Lactate beginning of surgery (mEq/L) 0.7 [0.6–0.9] 0.9 [0.7–1.2] < 0.001 RER beginning of surgery 0.80 [0.67–0.80] 0.80 [0.71–0.83] < 0.001 Lactate end of surgery (mEq/L) 0.9 [0.7–1.1] 2.3 [1.8–3.0] < 0.001 RER end of surgery 0.80 [0.67–0.80] 0.83 [0.80–1.0] < 0.001 Table 3 Postoperative outcome Values are presented as medians [interquartiles ranges] or numbers (percentages %) Variables Lactate ≤ 1.5 mEq/l (N = 622) Lactate > 1.5 mEq/l (N = 319) P-value Secondary outcomes 1) Length of stay in hospital (days) 9 [6—14] 9 [9—28] 0.426 2) Minor complications; N (%) 147 (23.6%) 66 (20.7%) 0.307 ➢ Superficial wound infection 22 (3.5%) 7 (2.2%) 0.259 ➢ Urinary infection 37 (5.9%) 13 (4.1%) 0.225 ➢ Paralytic ileus 22 (3.5%) 12 (3.8%) 0.861 ➢ Pneumonia 19 (3.1%) 8 (2.5%) 0.634 ➢ Postoperative confusion 23 (3.7%) 10 (3.1%) 0.657 ➢ Other infection 65 (10.5%) 42 (13.2%) 0.214 3) Major complications; N (%) 123 (19.8%) 72 (22.6%) 0.317 ➢ Anastomotic leakage 19 (3.1%) 14 (4.4%) 0.292 ➢ Peritonitis 3 (0.5%) 6 (1.9%) 0.037 ➢ Sepsis 32 (5.1%) 28 (8.8%) 0.031 ➢ Necrosis stoma 10 (1.6%) 2 (0.6%) 0.206 ➢ Wound dehiscence 10 (1.6%) 5 (1.6%) 0.963 ➢ Bleeding requiring a redo surgery 19 (3.1%) 17 (5.3%) 0.085 ➢ Pulmonary embolism 5 (0.8%) 2 (0.6%) 0.765 ➢ Pulmonary edema 9 (1.5%) 5 (1.6%) 0.881 ➢ Acute coronary syndrome 2 (0.3%) 0 (0%) 0.678 ➢ Atrial fibrillation / arrhythmia 15 (2.4%) 8 (2.5%) 0.928 ➢ Acute kidney injury 34 (5.5%) 29 (9.1%) 0.035 ➢ Reoperation 49 (7.9%) 17(5.3%) 0.147 ➢ 30‑day mortality 6 (0.1%) 4 (1.3%) 0.682 Page 7 of 8 Karam et al. BMC Anesthesiology (2022) 22:211 include a fluid challenge, administration of vasoactive agents to increase cardiac output or to target a higher blood pressure, or a red blood cell transfusion to increase the hemoglobin level. Conversely, a normal RER value could potentially accelerate the postoperative care and limit unnecessary tests and excessive length of stay in a high dependency unit. In addition, in patients equipped with an arterial line, RER calculation could justify less frequent arterial blood gas measurements to check lac- tate levels. Further clinical exploration and the eventual implementation into goal-directed protocols may help further clarify the DO2/VO2 relationship and establish the best clinical use for RER. Additional data should be soon available as a French team completed recently a large randomized controlled trial (N = 350) comparing an individualized hemodynamic optimization strategy guided by indirect measurement of the RER to a routine care in major surgery [16].  This retrospective study had both strengths and limi- tations. It was not possible to couple lactate and RER measurements more frequently intraoperatively due to the lack of a protocolized approach to sampling blood. Consequently, only the first and last arterial lactate val- ues were compared to concomitant RER values. Patients without any arterial blood gas values were excluded which represent 5.5% of the study collective. Imputa- tion methods for missing data were not performed as it is never recommended to impute a missing outcome since it would only improve predictive properties. Moreover, missingness around 5% is usually considered as only creating limited bias [17, 18]. Anesthesia practice was not standardized, but this reflects typical clinical prac- tice. Likewise, the population heterogeneity reflects real life practices. Types of surgeries (e.g., one-lung venti- lation or hepatic resection) and patient comorbidities (e.g., chronic obstructive pulmonary disease), could have effects on either DO2 or lactate metabolism and may alter the relation between RER and lactate. Future studies should investigate the impact of these conditions on the relationship between RER and hyperlactatemia. In conclusion, the RER had moderate discriminative abilities to detect hyperlactatemia. A RER value above 0.75 can detect hyperlactatemia with a moderately high sensitivity but with a poor specificity. Increased values should prompt clinicians to investigate for the pres- ence of hyperlactatemia and treat any potential causes of DO2/VO2 mismatch as suggested by the subsequent presence of hyperlactatemia. Abbreviations RER: Respiratory exchange ratio; DO2: Oxygen delivery; VO2: Oxygen consump‑ tion; ScvO2: Central venous oxygen saturation; AUC : Area under curve; ROC: Receiver operatingcharacteristics. Acknowledgements All the clinicians who helped in data collection from the current abdominal surgery database. More specifically, we want to thank Dr. Francois Martin Car‑ rier from the University of Montreal for the help in the statistics of the paper. Authors’ contributions L.K: Analyzed the data and drafted the manuscript. O.D.: Analyzed the data and edited the manuscript. B.A: Analyzed the data and edited the manuscript. B.A: Analyzed the data and edited the manuscript. N.C: Collected the data and edited the final manuscript. E.L: Collected the data and edited the final manu‑ script. H.P: Collected the data and edited the final manuscript. M.LM: Collected the data and edited the final manuscript. L.T: Collected the data and edited the final manuscript. M.M: Collected the data and edited the final manuscript. S.N: Analyzed the data and edited the final manuscript. J.D: Analyzed the data and edited the final manuscript. JLV: Analyzed the data and edited the final manuscript. P.VdL: Statistical analysis of the data and edited the final manu‑ script. A.J: Designed the study, collected and analyzed the data and drafted the manuscript. All authors read and approved the final manuscript. Funding The authors received no funding for this work. Availability of data and materials The database is closed and there is no public access. However, permission to access and use the database can be obtained if necessary by request to the corresponding author. Declarations Ethics approval and consent for publication The Erasme University Hospital (Université Libre de Bruxelles) ethical commit‑ tee approved this study December 4th, 2020 (SRB2020_654) and waived the need for informed consent. All methods were performed in accordance with the relevant guidelines and regulations. This study adheres to the STROBE guidelines. Consent for publication Not applicable. Competing interests AJ is a consultant for Edwards Lifesciences (Irvine, California, USA), Aguettant Laboratoire (Lyon, France) and Fresenius Kabi (Bad Homburg, Germany) OD is consultant for Medtronic (Trévoux, FRANCE) and received honoraria for giving lectures for Medtronic (Trévoux, FRANCE) and Livanova (Châtillon, France). Jean‑Louis Vincent is Editor‑in‑Chief of Critical Care. He has no other conflicts related to this article. The other authors have no conflicts of interest related to this article. Author details 1 Department of Anesthesiology and Intensive Care, Université Paris‑Saclay, Paul Brousse Hospital, Assistance Publique ‑ Hôpitaux de Paris (APHP), Villejuif, France. 2 Department of Anesthesiology and Perioperative Medicine Sauve‑ garde Clinic, Ramsay Santé, Lyon, France. 3 Department of Anesthesiology, Paul Brousse Hospital, 12 Avenue Paul Vaillant Couturier, 94800 Villejuif, France. 4 Department of Anesthesiology, University of California San Diego, La Jolla, CA, USA. 5 Department of Intensive Care, Erasme Hospital, Université Libre de Bruxelles, Brussels, Belgium. 6 Department of Anesthesiology, Brugmann Hospi‑ tal, Université Libre de Bruxelles, Brussels, Belgium. Received: 13 February 2022 Accepted: 23 June 2022 References 1. Bar S, Grenez C, Nguyen M, de Broca B, Bernard E, Abou‑Arab O, Bouhe‑ mad B, Lorne E, Guinot PG. Predicting postoperative complications with the respiratory exchange ratio after high‑risk noncardiac surgery: A prospective cohort study. Eur J Anaesthesiol. 2020;37(11):1050–7. Page 8 of 8 Karam et al. BMC Anesthesiology (2022) 22:211 • fast, convenient online submission • thorough peer review by experienced researchers in your field • rapid publication on acceptance • support for research data, including large and complex data types • gold Open Access which fosters wider collaboration and increased citations maximum visibility for your research: over 100M website views per year • At BMC, research is always in progress. Learn more biomedcentral.com/submissions Ready to submit your research Ready to submit your research ? Choose BMC and benefit from: ? Choose BMC and benefit from: 2. Schumacker PT, Cain SM. The concept of a critical oxygen delivery. Inten‑ sive Care Med. 1987;13(4):223–229. 3. Pino RM, Singh J. Appropriate Clinical Use of Lactate Measurements. Anesthesiology. 2021;134(4):637–644. 4. Krafft P, Steltzer H, Hiesmayr M, Klimscha W, Hammerle AF. Mixed venous oxygen saturation in critically ill septic shock patients. The role of defined events Chest. 1993;103(3):900–906. 5. Futier E, Robin E, Jabaudon M, Guerin R, Petit A, Bazin JE, Constantin JM, Vallet B. Central venous saturation and venous‑to‑arterial CO2 differ‑ ence as complementary tools for goal‑directed therapy during high‑risk surgery. Crit Care. 2010;14(5):R193. 6. Ospina‑Tascón GA, Umaña M, Bermúdez W, Bautista‑Rincón DF, Her‑ nandez G, Bruhn A, Granados M, Salazar B, Arango‑Dávila C, De Backer D. Combination of arterial lactate levels and venous‑arterial CO2 to arterial‑venous O 2 content difference ratio as markers of resuscitation in patients with septic shock. Intensive Care Med. 2015;41(5):796–805. 7. Kraut JA, Madias NE. Lactic acidosis. N Engl J Med. 2014;371(24):2309–2319. 8. Bar S, Santarelli D, de Broca B, Abou Arab O, Leviel F, Miclo M, Dupont H, Guinot PG, Lorne E. Predictive value of the respiratory exchange ratio for the occurrence of postoperative complications in laparoscopic surgery: a prospective and observational study. J Clin Monit Comput. 2021;35(4):849–858. 9. Cohen IL, Sheikh FM, Perkins RJ, Feustel PJ, Foster ED. Effect of hemor‑ rhagic shock and reperfusion on the respiratory quotient in swine. Crit Care Med. 1995;23(3):545–552. 10. Kristensen SD, Knuuti J, Saraste A, Anker S, Bøtker HE, Hert SD, Ford I, Gonzalez‑Juanatey JR, Gorenek B, Heyndrickx GR, et al. 2014 ESC/ESA Guidelines on non‑cardiac surgery: cardiovascular assessment and management: The Joint Task Force on non‑cardiac surgery: cardiovascu‑ lar assessment and management of the European Society of Cardiology (ESC) and the European Society of Anaesthesiology (ESA). Eur Heart J. 2014;35(35):2383–2431. 11. Bolondi G, Mocchegiani F, Montalti R, Nicolini D, Vivarelli M, De Pietri L. Predictive factors of short term outcome after liver transplantation: A review. World J Gastroenterol. 2016;22(26):5936–5949. 12. Nichol AD, Egi M, Pettila V, Bellomo R, French C, Hart G, Davies A, Stachowski E, Reade MC, Bailey M, et al. Relative hyperlactatemia and hospital mortality in critically ill patients: a retrospective multi‑centre study. Crit Care. 2010;14(1):R25. 13. Kim DG, Lee JY, Jung YB, Song SH, Lee JG, Han DH, Joo DJ, Ju MK, Choi GH, Choi JS, et al. Clinical significance of lactate clearance for the development of early allograft dysfunction and short‑term prognosis in deceased donor liver transplantation. Clin Transplant. 2017;31(12). 14. DeLong ER, DeLong DM, Clarke‑Pearson DL. Comparing the areas under two or more correlated receiver operating characteristic curves: a non‑ parametric approach. Biometrics. 1988;44(3):837–845. 15. Cannesson M, Le Manach Y, Hofer CK, Goarin JP, Lehot JJ, Vallet B, Taver‑ nier B. Assessing the diagnostic accuracy of pulse pressure variations for the prediction of fluid responsiveness: a “gray zone” approach. Anesthesi‑ ology. 2011;115(2):231–241. 16. Bar S, Boivin P, El Amine Y, Descamps R, Moussa M, Abou Arab O, Fischer MO, Dupont H, Lorne E, Guinot PG. Individualized hemodynamic optimi‑ zation guided by indirect measurement of the respiratory exchange ratio in major surgery: study protocol for a randomized controlled trial (the OPHIQUE study). Trials. 2020;21(1):958. 17. Jakobsen JC, Gluud C, Wetterslev J, Winkel P. When and how should multiple imputation be used for handling missing data in randomised clinical trials ‑ a practical guide with flowcharts. BMC Med Res Methodol. 2017;17(1):162. 18. White IR, Royston P, Wood AM. Multiple imputation using chained equa‑ tions: Issues and guidance for practice. Stat Med. 2011;30(4):377–399. Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in pub‑ lished maps and institutional affiliations.

Assessing the discriminative ability of the respiratory exchange ratio to detect hyperlactatemia during intermediate-to-high risk abdominal surgery.

07-08-2022

Karam, Lydia,Desebbe, Olivier,Coeckelenbergh, Sean,Alexander, Brenton,Colombo, Nicolas,Laukaityte, Edita,Pham, Hung,Lanteri Minet, Marc,Toubal, Leila,Moussa, Maya,Naili, Salima,Duranteau, Jacques,Vincent, Jean-Louis,Van der Linden, Philippe,Joosten, Alexandre

eng

PMC6308955

sensors Article Comparison of Different Algorithms for Calculating Velocity and Stride Length in Running Using Inertial Measurement Units Markus Zrenner 1,*, Stefan Gradl 1 , Ulf Jensen 2, Martin Ullrich 1 and Bjoern M. Eskofier 1 1 Machine Learning and Data Analytics Lab, Department of Computer Science, Friedrich-Alexander-Universität Erlangen-Nürnberg (FAU), 91052 Erlangen, Germany; stefan.gradl@fau.de (S.G.); martin.ullrich@fau.de (M.U.); bjoern.eskofier@fau.de (B.M.E.) 2 Finance & IT—IT Innovation, Adidas AG, 91074 Herzogenaurach, Germany; ulf.jensen@adidas.com * Correspondence: markus.zrenner@fau.de; Tel.: +49-9131-85-20162 Received: 30 August 2018; Accepted: 22 November 2018; Published: 30 November 2018



 Abstract: Running has a positive impact on human health and is an accessible sport for most people. There is high demand for tracking running performance and progress for amateurs and professionals alike. The parameters velocity and distance are thereby of main interest. In this work, we evaluate the accuracy of four algorithms, which calculate the stride velocity and stride length during running using data of an inertial measurement unit (IMU) placed in the midsole of a running shoe. The four algorithms are based on stride time, foot acceleration, foot trajectory estimation, and deep learning, respectively. They are compared using two studies: a laboratory-based study comprising 2377 strides from 27 subjects with 3D motion tracking as a reference and a field study comprising 12 subjects performing a 3.2-km run in a real-world setup. The results show that the foot trajectory estimation algorithm performs best, achieving a mean error of 0.032 ± 0.274 m/s for the velocity estimation and 0.022 ± 0.157 m for the stride length. An interesting alternative for systems with a low energy budget is the acceleration-based approach. Our results support the implementation decision for running velocity and distance tracking using IMUs embedded in the sole of a running shoe. Keywords: wearable sensors; inertial measurement unit; gait; running; stride length; velocity; smart shoe 1. Introduction Distance running is a very popular sport. Two main reasons for this popularity are simplicity and the health benefit, as running can be done in small and restricted time-frames and does not require a specific location. Besides sports gear, no equipment is needed. Moreover, running improves health. Studies have shown that aerobic endurance training like running can reduce blood pressure [1] and that moderate running twice a week (>51 min or six miles) reduces overall mortality risk and the occurrence of cardiovascular diseases [2]. However, overtraining can also lead to a higher risk of injury of the lower extremities for distance runners [3]. Tracking running performance over time can prevent overtraining and greatly support a healthy and effective training. A training diary helps to maintain the right training intensity and volume, which are essential for both performance and health enhancement. Training records are also motivating, as they highlight both effort and progress. However, a precise, objective, and easy measurement of relevant parameters is needed. Two common parameters that both professional and amateur runners use to track their performance is the average velocity and total distance. With these parameters, the running workout can be easily categorized, rated, and compared. In the past, runners estimated the distance of a predefined running track and took time with a stopwatch to calculate the average Sensors 2018, 18, 4194; doi:10.3390/s18124194 www.mdpi.com/journal/sensors Sensors 2018, 18, 4194 2 of 22 velocity of a distance run. With the rise of wearable technology in recent years, easier and more precise methods have become available. 1.1. Literature Review The predominant approach to tracking average velocity and total distance during running is the global positioning system (GPS). Smartphones or even smartwatches comprise a GPS chip, which allows a satellite-based localization of a runner. By tracking the runner’s absolute position over a complete run and using a solution to the second geodetic problem [4], the distance of a run can be measured. By incorporating the sampling frequency of the GPS module, a continuous time series of velocity values for the run can be computed. Thus, GPS delivers a time series of velocity, the cumulative distance, and the localization of the running track. From these data, the average velocity and the total distance can be extracted. The drawbacks of GPS are the additional gear (smartwatch, smartphone), the high energy demand, and the restriction to outdoor use. Integrating sensors directly into running shoes can solve these issues. One type of sensor that can be integrated into a shoe is an inertial measurement unit (IMU). It is a small, lightweight, and inexpensive sensor, which is capable of measuring triaxial accelerations and triaxial angular rates. A shoe setup with integrated IMUs overcomes the described GPS issues: runners only need a running shoe with integrated IMU; IMUs are energy efficient and work both indoors and outdoors. Using IMU data, it is possible to compute a stride length and an average velocity value per stride. The underlying assumption for the velocity computation is that the average velocity of the foot per stride matches the running velocity. By collecting stride velocity values and accumulating the stride length values over time, a distance measure and a continuous velocity recording of a complete run can be provided. The following paragraphs describe the state-of-the-art of four approaches for IMU data processing for calculating these metrics. In biomechanics, the relationships between stride frequency, stride length, running velocity, and body height was investigated [5]. The results indicated that with increasing running velocity, stride frequency and stride length increase. Thus, increasing running velocity is an interaction of increasing stride length and stride frequency [5]. Stride length itself depends on body height and can be expressed as a relative stride length. From these relationships, a generic model relating running velocity and stride length on the basis of the stride frequency can be deduced. The general idea behind this approach is the inverse correlation between velocity and stride time (the higher the velocity, the shorter the stride time). Thus, in order to estimate the stride length, only the stride time has to be distinguished by segmenting the data into single strides. An average velocity of the stride can then be calculated using the stride length and the measured stride duration. Recently, Gradl et al. [6] proposed an algorithm that uses quadratic regression to compute the velocity of movements. The velocity was evaluated during running, as well as other movements and showed a relative error of 6.9 ± 5.5%. The proposed algorithm is solely based on foot acceleration. Single strides are segmented from the data stream. Afterwards, the acceleration signal of all axes is integrated prior to the initial ground contact. Finally, the resulting integral value is converted to a velocity value using a quadratic regression model. Another method to compute velocity and stride length values from IMU signals is to reconstruct the trajectory of the sensor in the course of a stride. This method is heavily used for gait analysis for geriatric patients [7–9] or in inertial navigation scenarios [10,11]. For trajectory reconstruction, sensor fusion techniques must be applied to both the accelerometer and the gyroscope. Several fusion algorithms to cope with this task exist. Bailey et al. [12] and Foxlin et al. [13] used extended Kalman filters to compute the trajectory from the acceleration and angular rate signals, while Rampp et al. [7] applied a linear dedrifting technique. Both algorithms rely on a zero-velocity update during the stance phase for the initialization of the orientation. The literature shows that this approach works well while analyzing walking [7], but it was not evaluated for free running. Bailey et al. [12] applied Sensors 2018, 18, 4194 3 of 22 their approach to treadmill running and showed a good accuracy of 0.03 ± 0.2 m/s. However, they evaluated neither the velocity nor the stride length in a free running scenario. Deep learning techniques also show good results in IMU-based classification and regression tasks [14,15]. Hannink et al. [16] showed that deep convolutional neural network regression outperforms traditional stride length estimation in geriatric gait analysis. They trained a network with two convolutional layers, which was fed with the 6D IMU raw data of a stride. The output layer had a single node and provided an estimate of stride length. 1.2. Contribution Most of the described algorithms were evaluated either for walking or for running on a treadmill. However, both of these conditions yield different signal characteristics to those of free running. In running, different strike patterns, such as rearfoot or forefoot strike, exist and affect the performance of these algorithms. Besides, the movement is also more dynamic, which yields higher accelerations, angular rates, and impacts. Therefore, our contribution is the comparison of different algorithmic approaches for computing average velocity and stride length during overground running using an IMU embedded into the sole of a running shoe. We evaluate these algorithms on a large database including high variation of the input data. Additionally, we run a field study to assess the performance in a real-world scenario. Based on the results, we give implementation recommendations for specific use cases. 2. Methods 2.1. Data Collection We conducted two data collection studies for algorithm comparison, a lab study and a field study. The lab study was conducted in a sports research lab to evaluate the performance of the algorithms against ground-truth stride length and velocity labels on a per stride basis. A 3D motion tracking system was used as a reference. The field study was conducted on a 400-m outdoor running track to evaluate the performance regarding the total distance on a continuous 3.2-km run in realistic free-running conditions. The track length was used as a reference. 2.1.1. Lab Study In the lab study, data from 27 amateur runners (21 male, 6 female) were recorded. The dataset included runners with different strike types. Six of the subjects were forefoot/midfoot runners, and 21 subjects were rearfoot runners. The classification of the strike type was based on the definitions of Altman et al. [17]. Further anthropometric data can be found in Table 1. Before data acquisition, all subjects were informed about the related risks and gave written consent to participate in the study and for the collected data to be published. Table 1. Anthropometric data of subjects participating in the lab study. Parameter Mean ± Standard Deviation Age (years) 24.9 ± 2.4 Shoe size (U.S.) 9.3 ± 1.4 Height (cm) 178.6 ± 8.0 The subjects were equipped with running shoes in matching sizes (Response Cushion 21, Adidas AG, Herzogenaurach, Germany), as depicted in Figure 1a. This model had a cavity in the right shoe midsole for the placement of a sensor. We cut another cavity of the same size at the same location into the left shoe midsole to be able to acquire data from both the left and the right shoe in order to record more data for the training and evaluation of the algorithms. The specific IMU we used was Sensors 2018, 18, 4194 4 of 22 the miPod sensor [18]. The accelerometer of the sensor was configured with a range of ±16 g and the gyroscope with a range of ±2000 ◦s, and data were sampled with a frequency of fs = 200 Hz and a resolution of 16 bit. Before each data acquisition, the IMUs were calibrated using the calibration procedure introduced by Ferraris et al. [19]. Figure 1a depicts the orientation of the sensor in the sole of the running shoe: x points in the lateral direction, y in the dorsoventral direction, and z in the craniocaudal direction. As the gold standard for velocity and stride length, we used a motion capture system (Vicon Motion Systems Inc., Oxford, UK) with 16 infrared cameras and recorded data with a sampling rate of fs = 200 Hz. A submodel of the marker setup introduced by Michel et al. [20] containing six markers on each shoe (see Figure 1a) was used. The marker on the heel (for rearfoot runners) and the lateral sided toe marker (for forefoot/midfoot runners) were used to extract strides. Depending on the strike type, minima in the trajectory of the corresponding markers were used to label initial ground contacts [21]. The IMUs and the motion tracking system were synchronized using a wireless trigger [22], which was connected to light barriers (S40 Series, Datalogic, Bologna, Italy). The light barriers triggered the start and the end of the recording for each trial in both systems. Using the described synchronization technique, we were able to match strides in the motion capture gold standard data to strides in the IMU signal. The subjects were asked to run various trials with different velocities in the range of 2–6 m/s. We defined these velocity ranges to cover a wide range of relevant running velocities. As the capture volume was restricted to 6 m, and the stride length varied depending on the running velocity, different numbers of strides were recorded for the different running velocities. We recorded five additional trials for the two high velocity ranges to increase the number of captured strides. The velocity ranges and number of trials recorded can be found in Table 2. The subjects were asked to accelerate before and keep the pace within the capture volume. We measured the velocity at the beginning of the motion capture system volume using the above-mentioned light barriers used for synchronization. The velocity measured by the light barriers was used to ensure that a sufficient number of trials were recorded within each velocity range, for each subject. If necessary, the subjects were instructed to run faster or slower in order to ensure the defined number of trials in each velocity range. The ground truth value for each stride’s velocity vre f was computed from the motion capture reference as: vre f = dre f tre f = dre f · fs Nstride (1) where dre f is the stride length obtained by the difference of the positional data obtained by the motion capture system between two consecutive initial ground contacts, tre f the corresponding reference stride time, Nstride the number of samples in between two consecutive initial ground contacts, and fs the sampling rate. Figure 1b illustrates the setup and running path of the subjects during the lab data recording. Overall, 2377 strides were recorded during the lab study for the evaluation of the algorithms. Table 2. Number of trials and recorded strides per velocity range in the lab study. Velocity Range # of Trials # of Strides 2–3 m/s 10 921 3–4 m/s 10 558 4–5 m/s 15 544 5–6 m/s 15 354 Sensors 2018, 18, 4194 5 of 22 miPod x y z (a) Shoe (b) Setup runway Figure 1. (a) Shoe equipped with a miPod sensor and the marker setup. The IMU is located within the sole of the running shoe. The marker setup allowed for a computation of velocity and stride length. (b) Illustration of reference system setup. The subjects ran through the capture volume of the motion tracking system, created by 16 infrared cameras, and looped back around. 2.1.2. Field Study The goal of the field study was to evaluate the algorithm performance regarding total distance in a real-world scenario. We recorded twelve subjects who performed a self-paced 3.2-km run by completing eight rounds on a 400-m tartan track. We used this setup to be able to obtain a reference distance accurately. The equipment (IMUs, shoes) and settings were the same as described in the lab study (Section 2.1.1) to enable a direct comparison of the results. The subjects participating in the field study were not part of the lab study. Additionally, we recorded GPS data using a smartphone (Galaxy S8, Samsung Inc., Seoul, South Korea) and the fitness application Strava (Strava, Strava Inc., San Francisco, CA, USA), which is, with 136 million uploaded runs in 2017, one of the most popular fitness apps worldwide [23]. It also has the capability to export the GPS track in the GPX-format [24] allowing for a computation of the distance of the running track. The accuracy of the implemented algorithms can be compared to the GPS data for the total distance of the run. We used the great circle distance to compute the total distance of the GPS measurements [25]. Our computed total distance from the exported GPX file matched the distance that Strava provided via its services. Thus, we could compare the accuracy of the different algorithms to state-of-the-art running platform distance measurements. 2.2. Algorithms In this section, the algorithms will be described in detail. The section starts with the stride segmentation algorithm, which is required for all algorithms, except the acceleration-based algorithm, which includes a different approach to segment steps. Afterwards, the algorithms are described in the following order: Stride time, (foot) Acceleration, (foot) Trajectory estimation, and Deep Learning. 2.2.1. Stride Segmentation The first step in the IMU signal processing for velocity and distance calculation was the stride segmentation. In this step, single strides were extracted from the continuous IMU data stream with a threshold-based algorithm. Common algorithms use the distinct peaks in the acceleration signal in the dorsoventral direction ay[n] during initial ground contact to mark the beginning of a stride [26]. We enhanced this idea and used the beginning of the distinctive peak to mark the beginning of a stride. Sensors 2018, 18, 4194 6 of 22 This procedure is valid, as the ground already exerts a force to the IMU at the time instance of the peak in the acceleration signal. Using the peak itself would mean to mark a point in time that is part of the ground contact. To find the sample before the acceleration peak, we first differentiated the acceleration signal in the dorsoventral direction ay[n] and consecutively squared the resulting value to obtain a signal H[n] with amplified peak values. H[n] = (ay[n] − ay[n − 1])2 (2) In the signal H[n], the maxima were detected by comparing them to an empirical threshold (empirical threshold: H[n] > 1000 ( m s2 )2 ). For every detected maximum, the onset of the rise of H[n] was determined by setting all values below the threshold to zero and looking for the index of the last non-zero value in H[n] before the detected maximum. This index nIC was a potential candidate for an initial ground contact. To eliminate false detections, we added a detector for the swing phase prior to the peak in the acceleration signal. The swing phase detector computed an integral of ay[n] backwards from the first detected non-zero value until the first zero-crossing. This is the point in time where the foot starts decelerating during the swing phase. The integral value S[nIC] for the initial ground contact candidate nIC was computed as: S[nIC] = nIC ∑ n=nZC 1 fs ay[n] (3) In this equation, nZC corresponds to the index of the zero crossing marking the start of the deceleration. If the integral value S[nIC] exceeds an empirically-set threshold (empirical threshold: S[nIC] < −3[ m s ]), a swing phase is detected, and thus, the index of the first non-zero value before the acceleration peak is labeled as an initial ground contact. The described stride segmentation is depicted in Figure 2. -150 -100 -50 0 ay [ ] i n m/s2) Data Initial Contact Zero Crossing nZC nIC n n+1 ZC ( ) n+1 IC ( ) ( Figure 2. Example for the stride segmentation. The plot shows the acceleration signal in the dorsoventral direction ay[n], the detected initial ground contact nIC, and the beginning of the swing phase (zero crossing nZC) to confirm the stride candidate. The marked area depicts the integration area for the swing phase detection. 2.2.2. Stride Time Cavanagh et al. described the relationship between running velocity, stride length, and stride frequency [5]. Stride frequency is an inverse measure of the stride time and describes the number of strides per minute. They showed that runners can increase their running velocity either by increasing their stride length or by increasing their stride frequency, thus decreasing the stride time. For lower velocities, runners tend to increase the stride length, while for higher velocities, they tend to increase the stride frequency. Thus, both the stride time and stride length have no linear dependency on running velocity. Furthermore, it has to be noted that runners control their velocity individually. The stride length and therefore the velocity also depend on other parameters like the gender and the height of the runner. Male runners show greater stride lengths compared to female runners. The stride length increases with the body height [5]. Sensors 2018, 18, 4194 7 of 22 We used these biomechanical relations to build an algorithm that estimates stride length and velocity. Cavanagh et al. [5] provided averaged values for the non-linear correlation between the stride time tstride and a relative stride length dstride,rel, which is calculated by dividing the absolute stride length dstride by the runner’s height h. We looked for further publications describing this relationship and came up with two step functions for males and females that discretized the underlying non-linear relationship between stride time and stride length for each gender. The definition of the functions for both male and female runners can be found in Table 3. Table 3. Definition of step functions for the relative stride length dstride,rel[tstride] for (a) male and (b) female runners for the Stride time algorithm. (a) Male (b) Female tstride (s) dstride,rel Reference tstride (s) dstride,rel Reference 0.800 < tstride 0.830 [5,27] 0.800 < tstride 0.826 [27,28] 0.748 < tstride ≤ 0.800 1.080 [27,29] 0.735 < tstride ≤ 0.800 1.110 [5,27,29] 0.720 < tstride ≤ 0.748 1.260 [5] 0.720 < tstride ≤ 0.735 1.260 [5] 0.713 < tstride ≤ 0.720 1.330 [5] 0.704 < tstride ≤ 0.720 1.400 [5] 0.706 < tstride ≤ 0.713 1.410 [5] 0.667 < tstride ≤ 0.704 1.500 [5] 0.698 < tstride ≤ 0.706 1.490 [5] 0.607 < tstride ≤ 0.667 1.720 [5] 0.694 < tstride ≤ 0.698 1.590 [5] 0.578 < tstride ≤ 0.607 1.920 [5] 0.687 < tstride ≤ 0.694 1.740 [5] 0.500 < tstride ≤ 0.578 2.080 [5] 0.678 < tstride ≤ 0.687 1.880 [5] tstride ≤ 0.500 2.170 [30] 0.664 < tstride ≤ 0.678 1.960 [5] 0.649 < tstride ≤ 0.664 2.015 [5] 0.500 < tstride ≤ 0.649 2.060 [5] tstride ≤ 0.500 2.170 [30] The stride time tstride was obtained by dividing the number of samples of one stride Nstride by the sampling frequency fs. tstride = Nstride fs = (n + 1)IC − nIC fs (4) Nstride was computed by subtracting the indices of two consecutive initial ground contacts (n + 1)IC and nIC obtained from the stride segmentation algorithm. After obtaining the relative stride length dstride,rel of the runner based on the gender and the stride time, the absolute stride length dstride was computed by multiplying dstride,rel from the table and the runner’s height h in meters. dstride = h · dstride,rel (5) The running velocity vstride was then calculated using the stride time and the stride length. vstride = dstride tstride (6) Thus, the Stride time algorithm is solely based on the stride time. Gender and body height are usually known in all applications. 2.2.3. Acceleration The Acceleration method introduced in [6] uses only acceleration data for step segmentation and the computation of stride length and stride velocity. The method correlates the velocity of the foot (and thus, the subject) with the acceleration during the swing phase of the foot. It consists of three different algorithmic steps: (1) a continuous calculation of an integration value with a strong correlation to the movement velocity, (2) a stride segmentation based on initial ground contacts to determine the swing phase of the foot, and (3) a regression model to translate the continuous integration value to the velocity value. Sensors 2018, 18, 4194 8 of 22 We used the step segmentation algorithm from the cited publication due its applicability to movements other than running [6]. The inputs to the processing pipeline were the sampled triaxial acceleration signals from the foot sensor ax, ay, and az. After smoothing the input signals using a sliding window filter, the integration value ι was calculated as a multi-step absolute averaging across all directional components with: ι[n] = 1 L + 1 L ∑ i=0 ∑ d=x,y,z |sd[n − i]| , (7) where L is the window length, which is expected to be the same as the duration of the foot swing phase. Individual strides were determined in the smoothed dorsoventral acceleration signal sy using a peak detection process in combination with two knowledge-based parameter thresholds. The goal was to detect and isolate the high impact response of initial ground contact in the smoothed signal [6]. Each time a valid stride was detected by the stride segmentation algorithm, the average velocity per stride vstride was determined based on a second degree polynomial regression function: vstride = A + B · ι[nIC] + C · ι[nIC]2, (8) where the constants A, B, and C are derived during a regression model training phase where known reference velocity observations are matched to velocity integration values using parametric regression analysis. A trained regression model can be observed in Figure 3. Figure 3. Polynomial function of second degree (red line) that relates the velocity integration value ι to the reference velocity values vstride (grey dots). 2.2.4. Trajectory Based on the foot trajectory, the stride length and stride velocity can be deduced. The trajectory of the sensor during running can be computed using an extended Kalman filter approach or using dedrifting techniques. We applied dedrifting techniques due to two reasons: Firstly, Bailey et al. [12] showed that the results for the mean step velocity of the two techniques did not differ significantly with respect to accuracy (extended Kalman filter: 0.03 ± 0.02 m/s, linear dedrifting: 0.0 ± 0.03 m/s). Secondly, the same authors showed in a different article that a sampling rate of more than 250 Hz is required for an extended Kalman filter approach [31]. For embedded use cases (e.g., a smart shoe scenario), low sampling rates are beneficial from an energy perspective. In gait analysis, the linear dedrifting technique showed promising results for a lower sampling rate of 200 Hz [7]. Trajectory reconstruction algorithms based on linear dedrifting consist of four steps, as depicted in Figure 4, and have both the triaxial accelerometer and the triaxial gyroscope measurements as an input. In the following paragraphs, the four algorithmic steps will be explained in detail. Orientation Sensors 2018, 18, 4194 9 of 22 is computed by integrating the gyroscope measurements, and the position is obtained by integrating the accelerometer measurements. Midstance detection Orientation estimation Gravity removal Dedrifted integration Figure 4. The four steps of the algorithm for the trajectory reconstruction based on linear dedrifting. Midstance detection: A common problem with computing the trajectory from IMU measurements is the drift of the sensors introduced by noise in the acceleration and the angular rate measurements. This drift is limited by using zero velocity updates [32]. The idea behind these updates is to reinitialize the position and the orientation of the sensor for every stride. By applying that technique, absolute position in space is lost; however, the individual stride parameters can be computed more accurately. The reason for the higher accuracy lies in the integration of shorter durations and thus a smaller accumulated error. The point in time for the reinitialization of the stride values originates from gait analysis and is the midstance phase during a stride cycle. At this point in time, the foot has its lowest velocity, and the orientation of the foot is known, because during midstance in gait, the foot is expected to be flat on the ground. Thus, it can be assumed that the orientation of the sensor can be computed statically using the acceleration measurement. This allows the initialization of the position and velocity to zero and the orientation with respect to gravity. To find midstance, we computed the minimum gyroscopic energy after initial ground contact [32] in a 250-ms time interval. The duration of this time interval is the average time of the stance phase for velocities up to 6 m/s [33]. Hereafter, the trajectory reconstruction will be performed on strides segmented from midstance to midstance. Orientation estimation: After initializing the orientation based on the accelerometer measurement during midstance, the orientation of the sensor was computed using the gyroscope measurements. This step is necessary to calculate the orientation of the sensor so that gravity can be removed, which is an essential step for the computation of the position in space from the acceleration signal. For the orientation computation, we used the same quaternion integration approach as described by Rampp et al. [7]. Gravity removal: After the orientation estimation, gravity was removed. Without this removal, the gravitational acceleration of 9.81 m/s2 would be integrated additionally into the acceleration caused by running, which would lead to a large error over the duration of a stride. To remove gravity, we used the orientation of the sensor obtained by the gyroscope integration to rotate the acceleration measured in the sensor coordinate system to the world coordinate system. In the world coordinate system, we subtracted the gravitational acceleration from the measured acceleration. Dedrifted integration: The last step to come up with the full trajectory of the stride was to compute the position of the sensor by a double integration of the gravity removed acceleration. The first integration computed the velocity of the sensor over time, followed by the second integration, which resulted in the position of the sensor over time. Despite the gravity removal, there was still noise in the acceleration signal, causing drift in the results. This drift was reduced by dedrifting the velocity signal obtained after the first integration. The core idea behind dedrifting is the fact that we assume the velocity to be zero during midstance. For every stride, we fit a linear function in the velocity signal for all three directions, which was determined by the first and last velocity value of the stride. To dedrift the velocity signal, we subtracted the linear function from the integrated velocity signal, which enforced the velocity to be zero for both the first and the second midstance. This process is depicted in Figure 5. Sensors 2018, 18, 4194 10 of 22 (a) Velocity before dedrifting 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 t (s) −1 0 1 2 3 4 5 6 Velocity (m/s) X Y Z (b) Dedrifted velocity Figure 5. Visualization of the dedrifting method that ensures that the velocity during the second midstance is zero. (a) Velocity signal before dedrifting. The grey doted linear function is fit between the first and last point of the stride (midstance). (b) Velocity signal after dedrifting. Calculation of stride length and velocity: From the position of the sensor in space obtained after integrating the dedrifted velocity signal, the stride length dstride and the average stride velocity vstride were computed. The stride length was calculated as the L2-norm of the position in space at the index of the second midstance. Velocity was calculated by dividing stride length by stride time. 2.2.5. Deep Learning After outperforming conventional methods in various other fields like speech recognition, visual object recognition, and object detection [34], the methodology of deep learning started to become more and more popular for IMU data processing. Hannink et al. [16] introduced a deep convolutional regression network for calculating the stride length from raw IMU data in geriatric patients. The network learned a model for stride length regression based on raw IMU data without any domain knowledge. In this work, we used an adapted architecture for the stride length computation in running gait, which is depicted in Figure 6. It consisted of two convolutional layers, two max pooling layers, one flattening layer, and two fully-connected layers. For the implementation of the architecture, we used Keras [35] with a TensorFlow backend [36]. Sensors 2018, 18, 4194 11 of 22 Input data K 1 N 1 Convolutional layer 1 4 Max pooling layer N 2 K 2 Convolutional layer 1 4 Max pooling layer Fla
ening layer Fully- connected layer M1 Fully- connected layer M 2 Stride length Figure 6. Architecture of the convolutional neural network for stride length regression based on the raw 6D-IMU signal. For the first convolutional layer, we used N1 = 32 filter kernels of kernel length K1 = 30. The second convolutional layer consisted of N2 = 16 filter kernels of kernel length K2 = 15. The first fully-connected layer had M1 = 128 outputs that served as input to the second fully-connected layer, which had only a M2 = 1 output. This output represented the computed stride length. Before feeding data into the network, the segmented 6D-IMU data of a single stride were zero padded to 200 samples to assure a constant number of samples as an input to the network. One convolutional layer consisted of N convolution filters. The N outputs of a convolutional layer O(j) with j = 1 . . . N are called feature maps and were computed by the convolution of the six IMU input channels xc with c = 1...6 with the filter kernel φ(j) c of length K, adding biases b(j) c and finally applying a ReLU activation function: O(j) = ReLU  6 ∑ c=0 (Φ(j) c × xc + b(j) c )  (9) This formula has to be applied for all j = 1 . . . N filters to produce N feature maps O(j) after each convolutional layer. Thus, the two tunable parameters in the convolutional layers are the number of kernel coefficients K and the number of filters N. In the first convolutional layer, the kernel size was K1 = 30 and the number of filters N1 = 32. In the second convolutional layer, the kernel size was K2 = 15 and the number of filters N2 = 16 filters. After each convolutional layer, the resulting feature map was fed into a max pooling layer, which downsampled the resulting feature map by a downsampling factor of two by taking the maximum in non-overlapping windows of size two. After the second max pooling layer, the feature map was flattened to produce a one-dimensional feature list that can be fed into the fully-connected layers. Thus, the flattening layer appended the N2-dimensional output of the second max pooling layer after each other into one feature list. The two fully-connected layers at the end of the architecture computed a weighted sum of all k = 1 . . . Nf input features ϕk of the one-dimensional feature vector with weights wk,j and added biases bk. A ReLU function again activated the positive features. Fj = ReLU  Nf −1 ∑ k=0 (wk,j · ϕk + bk,j)  (10) Sensors 2018, 18, 4194 12 of 22 The outputs of the fully-connected layers were feature lists Fj with j = 1 . . . M, where M describes the number features. In our architecture, the first fully-connected layer had M1 = 128 output features. The second fully-connected layer had only M2 = 1 output feature, which was the resulting target value. In our implementation, the regressed target value was the stride length. To prevent overfitting, we also added a dropout layer to our network [37]. The dropout layer was stacked between the two fully-connected layers and dropped 30% of the neurons. During training, we fed the data into the network in five epochs with a batch size of 16. We trained the network both for the stride length and for the velocity. The network with the stride length as the output outperformed the velocity approach and was therefore used for the evaluation in this publication. Thus, the velocity vstride for the Deep Learning approach was computed by dividing the stride length dstride obtained from the neural network by the stride time tstride obtained from the stride segmentation. 2.3. Evaluation 2.3.1. Lab Study The results of the lab study dataset will be evaluated using the mean error (ME) and standard deviation (Std), the mean absolute percentage error (MAPE), and the mean absolute error (MAE). We provide all these measures to make our results comparable to prior studies. For the evaluation of the Acceleration and Deep Learning algorithms, we used leave-one-subject-out cross-validation to prevent overfitted results. We also show Bland–Altman plots [38] to visualize the results. 2.3.2. Field Study For the evaluation of the 3.2-km field study dataset, we used the MAE to evaluate the total distance of the runs. After segmenting the strides and calculating the stride lengths for each stride, we accumulated the single stride lengths and compared them to the ground truth value of 3200 m. The reason for choosing the MAE for this evaluation was the fact that the absolute deviation of the ground truth value is of great importance to runners. For the Acceleration and Deep Learning algorithms, we computed the regression models based on the lab study dataset. Due to having different subjects participating in the lab and the field study, the results were not overfitted. The GPS measurements of the total distance of the individual runs were also evaluated by comparing them to the gold standard value of 3.2 km. 3. Results 3.1. Lab Study Table 4 depicts the mean errors and standard deviations for both stride velocity and stride length of the four different algorithms for the lab study dataset. The results were averaged over all strides in the lab study dataset. The results show that the Trajectory algorithm performed best considering both the ME ± Std and the MAE. Sensors 2018, 18, 4194 13 of 22 Table 4. Mean error (ME) and standard deviations (Std), mean percentage error (MAPE), and mean absolute error (MAE) of stride length and average velocity per stride of the four algorithms for the lab study dataset. Parameter Error Measure Stride Time Acceleration Trajectory Deep Learning ME ± Std (m/s) 0.209 ± 0.782 0.005 ± 0.350 0.028 ± 0.252 0.055 ± 0.285 Velocity MAPE (%) 17.2 7.7 3.5 5.9 MAE (m/s) 0.622 0.272 0.133 0.216 ME ± Std (cm) 17.7 ± 57.3 −0.5 ± 25.6 2.00 ± 14.1 2.5 ± 20.1 Stride length MAPE (%) 17.1 7.9 2.8 5.9 MAE (cm) 45.2 19.9 7.6 15.3 Figure 7 shows the results of the stride length for the different velocity ranges. The MEs of the Deep Learning algorithm increased with higher velocities. The Trajectory showed lower MEs for the three slower velocity ranges than for the highest velocity range. The Acceleration algorithm showed small errors from 3–5 m/s. Its performance dropped for the outer velocity ranges from 2–3 m/s and from 5–6 m/s. The Stride time algorithm worked well for the velocity range of 2–3 m/s and 5–6 m/s; however, it showed large errors of more than 40 cm for the other velocity ranges. 2-3 3-4 4-5 5-6 Speed bins (m/s) 0 10 20 30 40 Mean error stride length (m) Stride time Acceleration Trajectory Deep Learning Figure 7. Mean error of the stride length of the four different algorithms for the different velocity ranges the subjects ran in the lab study. Figure 8 shows the Bland–Altman plots for both the stride length and the average velocity per stride for the lab study dataset. The results are color coded into the velocity ranges presented in Table 2. The Trajectory algorithm performed well for velocities up to 5 m/s. For the high velocity range, larger errors could be observed. The Stride time algorithm performed worst and showed a linear error distribution in the Bland–Altman plots. In the Acceleration, Trajectory, and Deep Learning plots for stride length, we see the samples of the different velocity ranges overlapping. This overlap is not visible in the velocity plots. Sensors 2018, 18, 4194 14 of 22 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Mean (m) −2 −1 0 1 2 Difference (m) mean 1.98⋅Std (a) Algorithm: Stride time; metric: stride length 2 3 4 5 6 Mean (m/s) −2 −1 0 1 2 Difference (m/s) mean 1.98⋅Std (b) Algorithm: Stride time; metric: velocity 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Mean (m) −2 −1 0 1 2 Difference (m) mean 1.98⋅Std (c) Algorithm: Acceleration; metric: stride length 2 3 4 5 6 Mean (m/s) −2 −1 0 1 2 Difference (m/s) mean 1.98⋅Std (d) Algorithm: Acceleration; metric: velocity 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Mean (m) −2 −1 0 1 2 Difference (m) mean 1.98⋅Std (e) Algorithm: Trajectory; metric: stride length 2 3 4 5 6 Mean (m/s) −2 −1 0 1 2 Difference (m/s) mean 1.98⋅Std (f) Algorithm: Trajectory; metric: velocity 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 Mean (m) −2 −1 0 1 2 Difference (m) mean 1.98⋅Std (g) Algorithm: Deep Learning; metric: stride length 2 3 4 5 6 Mean (m/s) −2 −1 0 1 2 Difference (m/s) mean 1.98⋅Std (h) Algorithm: Deep Learning; metric: velocity Figure 8. Bland–Altman plots for stride length and velocity for the four algorithms. Each row contains the metrics for one algorithm. The individual samples are color coded depending on the velocity bin of the sample: 2–3 m/s blue, 3–4 m/s red, 4–5 m/s green, 5–6 m/s purple. The dotted-dashed horizontal lines depict the mean error and the dotted horizontal line the 95% confidence interval. Sensors 2018, 18, 4194 15 of 22 3.2. Field Study Figure 9 shows the MAE of the total running distance for the field study dataset, both for the algorithms and GPS-based estimation. The figure indicates that the Trajectory algorithm performed best with a MAE of 94.0 m. The error was comparable to that of the GPS-based estimate (82.1 m). Stride time Acceleration Trajectory Deep Learning GPS 0 100 200 300 400 500 600 Mean absolute error (m) Figure 9. Mean absolute error of the 3.2-km run for the four different algorithms and GPS. 4. Discussion Firstly, we will compare our results to existing literature. Afterwards, we will discuss the results of the lab study including a detailed evaluation of the individual algorithms with respect to their accuracy and their advantages and disadvantages in a smart shoe scenario. Special emphasis will be placed on the number of sensors that are needed to run the algorithms and the underlying power consumption of these sensors. Finally, the results of the field study on the tartan track will be discussed. 4.1. Comparison to Existing Literature Different papers already evaluated the stride length or the velocity of single strides. Three of these papers are listed in Table 5. These three publications used similar approaches to ours: Bailey et al. [12] used a trajectory approach using a linear dedrifting technique; Gradl et al. [6] used the described acceleration approach; and Hannink et al. [16] a DCNN approach. Table 5. Results of other publications related to stride length and velocity calculation. Gait Type # Subjects # Strides Parameter Error Measure Result Bailey et al. [12] Running 5 1800 Velocity ME 0.04 ± 0.03 m/s Gradl et al. [6] Running 9 795 Velocity MAPE 6.9 ± 5.5% Hannink et al. [16] Walking 101 ∼1392 Stride length ME 0.01 ± 5.37 cm With respect to the standard deviation, the results of the trajectory implementation of Bailey et al. [12] are better than our results (Table 4). They also evaluated 1800 strides; however, these strides only originated from running velocities ranging from 2.3–3.4 m/s. We also evaluated our results for this velocity range and obtained an error of 0.004 ± 0.107 m/s. We observe that our standard deviation is still higher than the standard deviation reported from Bailey et al. One reason for that might be the higher number of different runners with different running styles who participated in our study. Furthermore, their study was conducted on a treadmill. On a treadmill, the variability of different strides at a given velocity is lower and does not reproduce overground running kinematics [39]. The errors reported by Gradl et al. [6] were obtained on a smaller database than the one presented in this paper. Thus, our worse results are due to the higher variability in our dataset, which the second degree polynomial could not appropriately approximate. Sensors 2018, 18, 4194 16 of 22 The results of Hannink et al. [16] were evaluated for gait in geriatric patients. Hence, there is a general difference in the stride patterns, causing differences in the results. Further differences between the setup of our network architecture and study population are listed and discussed in the following section. 4.2. Lab Study In this section, we will discuss the results of the lab study for each algorithm in detail with respect to their advantages/disadvantages and the number of sensors needed for their implementation. Stride time: The Stride time algorithm leads to the lowest accuracy for the lab study dataset. Even though stride time and stride length relative to the subject’s height correlate non-linearly, the correlation does not seem to be high enough to compute velocity and stride length accurately. The low correlation is also visible in Figure 10. The gray dots are the relative stride length values obtained from the lab study dataset, and the red line is the step function for male subjects defined in Table 3a. We see that the step function does not approximate the underlying data accurately. The standard deviation of the relative stride length within a certain stride time range (e.g., 0.748 < tstride ≤ 0.800) of the step function is high. This is due to the fact that velocity is controlled by stride frequency and stride length. The Stride time algorithm cannot handle that fact, as it only depends on stride frequency. Figure 10. Visualization of the correlation between the stride time tstride and the relative stride length dstride,rel for male subjects. The light gray dots depict the data obtained from the field study, whereas the red curve and the black dashed lines visualize the step function obtained from literature and implemented in the Stride time algorithm. In the Bland–Altman plots for the stride length metric (Figure 8), the other three algorithms showed overlapping sample clouds. This indicates that people increased their velocity both by increasing their stride length and by decreasing their stride time in higher velocities. The other algorithms are capable of dealing with this effect due to the fact that the sample clouds are separated in the Bland–Altman plots of the velocity metric. This is not observable in the plots for the Stride time algorithm. Thus, the other algorithms can deal better with the velocity control via stride frequency and stride length. Furthermore, we want to discuss the shape of the Stride time algorithm’s Bland–Altman plots briefly. The long diagonal lines in the plots (Figure 8b) originate from the steps in the step function introduced in Table 3. One line belongs to one stride time range. The small deviations within the diagonals originate from the different body heights. We observed that for some stride time ranges, the gold standard velocity ranged from 2–6 m/s (color coded within one diagonal), showing that the stride time ranges of the step function obtained from the literature do not generalize well. Furthermore, Sensors 2018, 18, 4194 17 of 22 the relative stride lengths presented in Table 3 are averaged over specific study populations. Even if a subject controls its stride frequency in the exact same manner as encoded by the stride time ranges of the step function, the resulting stride length could be incorrect due to an incorrect relative stride length. Despite the algorithm’s low accuracy, an advantage of the stride time algorithm is that it can be implemented very energy efficiently. In the case of an IMU scenario, only a stride segmentation is necessary to compute the stride time. The stride segmentation presented in this paper only needs the sampling of the acceleration in the dorsoventral direction; thus, a 1D-accelerometer would be sufficient. In fact, strides could be segmented without an IMU using sensors such as piezo-electric switches to detect the ground contact [40]. Acceleration: The plot with the ME for the different velocity ranges in Figure 7 shows that the Acceleration algorithm works better for the the two velocity ranges from 3–4 m/s and 4–5 m/s. In addition, the Bland–Altman plots in Figure 8 show outliers especially for the highest velocity range for both the stride length and the average velocity. The reason for that can be observed in Figure 3, where we see that the second degree polynomial used to map the velocity integration value ι to the velocity value approximates the reference data better for the velocity range from 3–5 m/s and especially not well for the highest velocity range. This can be explained by the spread of the underlying data being too large to be represented by the polynomial. However, the Acceleration algorithm outperforms the Stride time algorithm and shows comparable performance to the Deep Learning algorithm for the velocity range of 3–4 m/s. The advantage of the Acceleration algorithm over the better performing Trajectory and slightly better performing Deep Learning algorithm is its energy efficiency. For the computation of the stride length and the velocity, only a triaxial accelerometer needs to be sampled. Sampling only an accelerometer consumes less energy than sampling the gyroscope or sampling both sensors. For example, for the MPU9250 from InvenSense, the supply current needed for sampling only the accelerometer is less than 15% of the current needed for sampling both the accelerometer and the gyroscope [41]. Furthermore, the sampling rate can be further reduced for the Acceleration algorithm [6]. We also tested the reduction of the sampling rate for the lab study dataset and observed that a reduction to 60 Hz does not affect the accuracy of the algorithm. With such a low sampling rate, the energy consumption can be further reduced. Another advantage of the algorithm is its generalizability and its applicability to other movements like side stepping [6]. Foot trajectory: The Trajectory algorithm performs best for the lab study dataset. Especially for velocities up to 5 m/s, the algorithm achieves a ME of less than 0.012 m for the stride length and 0.014 m/s for the average velocity. For velocities higher than 5 m/s, the accuracy drops. In the Bland–Altman plots (Figure 8e,f), outliers for this velocity range are visible. The zero-velocity update based on the detection of the minimum energy in the gyroscope signal is error prone for such high velocities. The foot has no real zero-velocity phase and is always in motion. Thus, the underlying zero-velocity assumption does not hold. One way how to improve this algorithm is to propose a better solution for the initial condition when applying it to higher running velocities. Future work could evaluate whether a regression model based on the velocity during the swing phase would be a better initial condition. For the Trajectory algorithm, we were also interested in the applicability of the zero velocity update for the different strike types due to the foot never being flat on the ground for forefoot runners. Hence, we also evaluated the accuracy of the Trajectory algorithm for the different strike types. The violin plots for forefoot and rearfoot runners are depicted in Figure 11. The plots show that the MEs do not differ significantly for the two strike types. However, the standard deviation is higher for forefoot runners. The low ME both for the forefoot and the rearfoot strike type can be explained by the fact that we align the foot during the zero velocity phase with gravity. The higher standard deviation originates in the more dynamic nature of the forefoot running style. Thus, the zero velocity phase cannot be detected accurately, which results in higher errors. Sensors 2018, 18, 4194 18 of 22 Figure 11. Violin plots of the error (sre f − sstride) in the velocity computation for forefoot strikers and rearfoot strikers. An advantage of the Trajectory algorithm is that it provides more information about the stride than the velocity and the stride length. During the computation of these parameters, the orientation of the shoe in space is also calculated, which allows for a determination of other parameters like the sole angle, which defines the strike pattern or the range of motion in the frontal plane that is associated with pronation [42]. Furthermore, the algorithm uses solely signal processing and has no training phase, which makes it well applicable to unseen data. This holds for lower velocities and the transition to walking. In terms of an embedded implementation and energy efficiency, the Trajectory algorithm needs both accelerometer and gyroscope data. Thus, it needs more energy than the Stride time and the Acceleration algorithm for acquiring 6D-IMU data. Deep learning: The Deep Learning algorithm produced an ME of less than 0.095 m/s for the velocity and 0.104 m for the stride length for all velocity ranges in the lab study dataset. Compared to Hannink et al. [16], we reduced both the number of filters in the second convolutional layer and the number of outputs in the fully-connected layer, because the results using the identical structure yielded worse results for our use case. The differences in the architecture are listed in Table 6. Generally, the performance of the DCNN network is worse compared to the results reported in [16]. Table 6. Differences in the study setup and architecture presented in [16] from our DCNN implementation. # Parameters Range Stride # Training N2 M1 Trained ME ± Std Length Data Set Samples Hannink et al. [16] 64 1024 2,332,385 0.01 ± 5.37 cm 0.14–1.30 m ∼1392 Our approach 16 128 85,425 1.3 ± 19.4 cm 1.22–4.84 cm 2377 We see that our approach needs less parameters due to the reduction of filters in the second convolutional layer and the smaller output number of the fully-connected layer. However, our results show a larger error. The reason for that might be a larger variation in our training data and the different strike pattern in running. The range of the target parameter of stride length is 3.62 m in the lab study of this work and 1.16 m in the dataset for geriatric patients of Hannink et al. [16]. The strike patterns in running differ significantly for forefoot and rearfoot runners, which also introduces more variation in the input data. Besides, we observed that during training, the training errors and validation errors still varied after the five training epochs, even though we had more training samples than Hannink et al. [16]. Increasing the number of epochs or batches did not change the varying validation errors. This indicates that the DCNN does not generalize well. Thus, the results might be further improved by incorporating more data samples in the training process of the network. Sensors 2018, 18, 4194 19 of 22 The embedded implementation of the presented method is a challenge as the DCNN model comprises 85,425 parameters. However, it is still in a range where it can be implemented on a microcontroller. For this method, the acceleration and the gyroscope have to be sampled. This further increases the energy demand compared to the Acceleration approach. Taking computational effort and performance into account, the Acceleration method would be a better trade-off for an embedded implementation. 4.3. Field Study The aim of the field study dataset was the evaluation of the estimation of the overall distance of a run in an outside and real-world scenario. The Trajectory algorithm also worked best for this dataset. With an MAE of 94.0 m, it is comparable to the results of GPS, which also produced an MAE of 82.1 m, and is used in state-of-the-art running platforms tracking athlete performances. Besides, the IMU technology has the advantage that it allows velocity and distance computations indoors or in scenarios where no satellite connection for GPS is available. Based on the presented results, we argue that although the Trajectory algorithm has high standard deviations in the lab study for the stride length calculation, these have no major impact on the computation for longer distances based on stride length. We believe this is due to errors canceling out over time. As the subjects’ average velocity was 3.48 m/s during the data acquisition, the high velocity range of 5–6 m/s was not reached for the amateur runners that participated in this study. We expect the results to be worse for the high velocity range, which can be reached by professional athletes. The Stride time algorithm showed the worst performance for the field study dataset (MAE of 599.7 m). Despite its best energy efficiency, our results indicate that its accuracy is too low to use for tracking velocity and distance. The Deep Learning approach (MAE 194.5 m) performs better than the Acceleration approach (MAE 333.1 m). Due to the fact that the the neural network also needs the 6D-IMU data as an input, it has no benefit compared to the Trajectory approach, which performs better. The Acceleration approach only requires the sampling of the triaxial accelerometer, which makes it more energy efficient. Despite its decreased accuracy, we propose to use this algorithm in use cases where very strict energy limitations occur. 5. Conclusions and Future Work In this study, we compared four different algorithms with respect to their performance on stride length and mean average velocity per stride calculation for running. We conducted two studies to evaluate the accuracy of the algorithms: one study in a laboratory environment with a motion capture system as the ground truth, in which we acquired 2377 strides of 27 subjects, and one field study in a real-world scenario. We showed that the Trajectory algorithm performs best and especially well for velocities up to 5 m/s. The results of the field study showed that this algorithm does not only work on single strides, but also on longer outdoor runs in a real-world scenario. The MAEs for this scenario showed that the trajectory is comparable to GPS measurements, which is the common method for total distance tracking in amateur running. However, the Trajectory algorithm is more costly energy wise due to the fact that both the acceleration and the gyroscope have to be acquired with a sampling rate of 200 Hz. When it comes to an energy-efficient use case, the Acceleration algorithm is a good choice, as it only needs to sample the accelerometer, and the sampling rate can be decreased to 60 Hz. We therefore propose the implementation of the Trajectory algorithm for use cases with no energy limitations and the implementation of the Acceleration algorithm for use cases with energy restrictions. In future work, we want to address further parameters that can be computed using inertial measurement units and other sensors located in the sole of a running shoe. Using data acquired by sensors on both feet, it is possible to perform bi-lateral analysis by combining the information of both sensors. Thus, the contribution of the individual lower limbs to the running movement can be further evaluated. Using only data from IMUs within the sole of a running shoe and the Trajectory algorithm, analysis regarding imbalances in stride length, stride time or orientation of the two feet Sensors 2018, 18, 4194 20 of 22 can be conducted. Furthermore, other temporal parameters like flight time or stance time could be computed by adding a toe-off detection. Due to inaccuracies with the toe-off detection in running using only one IMU per foot [43], we plan to also incorporate pressure sensors for toe-off detection into the soles of a running shoe. Author Contributions: M.Z. conceived of and designed the experiments; M.Z. and M.U. performed the experiments; M.Z., S.G., and U.J. analyzed the data; B.M.E. contributed reagents/materials/analysis tools; M.Z. wrote the paper. Funding: This work was conducted during the Servicefactory research project supported by the German Federal Ministry for Economic Affairs and Energy. Bjoern Eskofier gratefully acknowledges the support of the German Research Foundation (DFG) within the framework of the Heisenberg professorship program (Grant Number ES 434/8-1). Acknowledgments: The authors also thank Christine Martindale for revising the script as a native English speaker. Conflicts of Interest: The authors declare no conflict of interest. References 1. Cornelissen, V.A.; Fagard, R.H. Effects of endurance training on blood pressure, blood pressure—Regulating mechanisms, and cardiovascular risk factors. Hypertension 2005, 46, 667–675. [CrossRef] [PubMed] 2. Lee, D.C.; Pate, R.R.; Lavie, C.J.; Sui, X.; Church, T.S.; Blair, S.N. Leisure-time running reduces all-cause and cardiovascular mortality risk. J. Am. Coll. Cardiol. 2014, 64, 472–481. [CrossRef] [PubMed] 3. Gallo, R.A.; Plakke, M.; Silvis, M.L. Common leg injuries of long-distance runners: Anatomical and biomechanical approach. Sports Health 2012, 4, 485–495. [CrossRef] [PubMed] 4. Mitchell, J.S.; Mount, D.M.; Papadimitriou, C.H. The discrete geodesic problem. SIAM J. Comput. 1987, 16, 647–668. [CrossRef] 5. Cavanagh, P.R. Biomechanics of Distance Running; Human Kinetics: Champaign, IL, USA, 1990. 6. Gradl, S.; Zrenner, M.; Schuldhaus, D.; Wirth, M.; Cegielny, T.; Zwick, C. Movement Speed Estimation Based on Foot Acceleration Patterns. In Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Honolulu, HI, USA, 17–21 July 2018; pp. 3506–3508. 7. Rampp, A.; Barth, J.; Schuelein, S.; Gassmann, K.G.; Klucken, J.; Eskofier, B.M. Inertial sensor-based stride parameter calculation from gait sequences in geriatric patients. IEEE Trans. Biomed. Eng. 2015, 62, 1089–1097. [CrossRef] [PubMed] 8. Mariani, B.; Hoskovec, C.; Rochat, S.; Büla, C.; Penders, J.; Aminian, K. 3D gait assessment in young and elderly subjects using foot-worn inertial sensors. J. Biomech. 2010, 43, 2999–3006. [CrossRef] [PubMed] 9. Ferrari, A.; Ginis, P.; Hardegger, M.; Casamassima, F.; Rocchi, L.; Chiari, L. A mobile Kalman-filter based solution for the real-time estimation of spatio-temporal gait parameters. IEEE Trans. Neural Syst. Rehabil. Eng. 2016, 24, 764–773. [CrossRef] [PubMed] 10. Bird, J.; Arden, D. Indoor navigation with foot-mounted strapdown inertial navigation and magnetic sensors [emerging opportunities for localization and tracking]. IEEE Wirel. Commun. 2011, 18, 28–35. [CrossRef] 11. Stroembaeck, P.; Rantakokko, J.; Wirkander, S.L.; Alexandersson, M.; Fors, K.; Skog, I.; Händel, P. Foot-mounted inertial navigation and cooperative sensor fusion for indoor positioning. In Proceedings of the ION International Technical Meeting (ITM), San Diego, CA, USA, 25–27 January 2010; pp. 89–98. 12. Bailey, G.; Harle, R. Assessment of foot kinematics during steady state running using a foot-mounted IMU. Procedia Eng. 2014, 72, 32–37. [CrossRef] 13. Foxlin, E. Pedestrian tracking with shoe-mounted inertial sensors. IEEE Comput. Gr. Appl. 2005, 25, 38–46. [CrossRef] 14. Kautz, T.; Groh, B.H.; Hannink, J.; Jensen, U.; Strubberg, H.; Eskofier, B.M. Activity recognition in beach volleyball using a Deep Convolutional Neural Network. Data Min. Knowl. Discov. 2017, 31, 1678–1705. [CrossRef] 15. Ordóñez, F.J.; Roggen, D. Deep Convolutional and LSTM Recurrent Neural Networks for Multimodal Wearable Activity Recognition. Sensors 2016, 16, 115. [CrossRef] [PubMed] 16. Hannink, J.; Kautz, T.; Pasluosta, C.; Barth, J.; Schulein, S.; Gassmann, K.G.; Klucken, J.; Eskofier, B. Mobile Stride Length Estimation with Deep Convolutional Neural Networks. IEEE J. Biomed. Health Inf. 2017, 21, 85–93. [CrossRef] [PubMed] Sensors 2018, 18, 4194 21 of 22 17. Altman, A.R.; Davis, I.S. A kinematic method for footstrike pattern detection in barefoot and shod runners. Gait Posture 2012, 35, 298–300. [CrossRef] [PubMed] 18. Blank, P.; Kugler, P.; Schlarb, H.; Eskofier, B.M. A Wearable Sensor System for Sports and Fitness Applications. In Proceedings of the 19th Annual Congress of the European College of Sport Science, Amsterdam, The Netherlands, 2–5 July 2014; p. 703 19. Ferraris, F.; Grimaldi, U.; Parvis, M. Procedure for effortless in-field calibration of three-axial rate gyro and accelerometers. Sens. Mater. 1995, 7, 311–330. 20. Michel, K.J.; Kleindienst, F.I.; Krabbe, B. Development of a lower extremity model for sport shoe research. In Proceedings of the 13rd Annual Meeting of ESMAC, Warsaw, Poland, 23–25 March 2004; p. 80. 21. Maiwald, C.; Sterzing, T.; Mayer, T.; Milani, T. Detecting foot-to-ground contact from kinematic data in running. Footwear Sci. 2009, 1, 111–118. [CrossRef] 22. Kugler, P.; Schlarb, H.; Blinn, J.; Picard, A.; Eskofier, B. A wireless trigger for synchronization of wearable sensors to external systems during recording of human gait. In Proceedings of the Annual International Conference of the IEEE Engineering in Medicine and Biology Society, San Diego, CA, USA, 28 August–1 September 2012; pp. 4537–4540. 23. Strava Stories—2017 in Stats. 2018. Available online: https://blog.strava.com/2017-in-stats/ (accessed on 22 August 2018). 24. GPX—The GPS Exchange Format. Available online: http://www.topografix.com/gpx.asp (accessed on 28 August 2018). 25. Karney, C.F. Algorithms for geodesics. J. Geod. 2013, 87, 43–55. [CrossRef] 26. Strohrmann, C.; Harms, H.; Tröster, G.; Hensler, S.; Müller, R. Out of the lab and into the woods: kinematic analysis in running using wearable sensors. In Proceedings of the 13rd International Conference on Ubiquitous Computing, Beijing, China, 17–21 September 2011; pp. 119–122. 27. Danion, F.; Varraine, E.; Bonnard, M.; Pailhous, J. Stride variability in human gait: The effect of stride frequency and stride length. Gait Posture 2003, 18, 69–77. [CrossRef] 28. Perry, J.; Davids, J.R. Gait analysis: Normal and pathological function. J. Pediat. Orthop. 1992, 12, 815. [CrossRef] 29. Elliott, B.; Blanksby, B. Optimal stride length considerations for male and female recreational runners. Br. J. Sports Med. 1979, 13, 15. [CrossRef] [PubMed] 30. Hunter, J.P.; Marshall, R.N.; McNair, P.J. Interaction of step length and step rate during sprint running. Med. Sci. Sports Exerc. 2004, 36, 261–271. [CrossRef] [PubMed] 31. Bailey, G.; Harle, R. Sampling Rates and Sensor Requirements for Kinematic Assessment During Running Using Foot Mounted IMUs. In International Congress on Sports Science Research and Technology Support; Springer: Berlin, Germany, 2014; pp. 42–56. 32. Skog, I.; Handel, P.; Nilsson, J.O.; Rantakokko, J. Zero-velocity detection—An algorithm evaluation. IEEE Trans. Biomed. Eng. 2010, 57, 2657–2666. [CrossRef] [PubMed] 33. De Wit, B.; De Clercq, D.; Aerts, P. Biomechanical analysis of the stance phase during barefoot and shod running. J. Biomech. 2000, 33, 269–278. [CrossRef] 34. LeCun, Y.; Bengio, Y.; Hinton, G. Deep learning. Nature 2015, 521, 436–444. [CrossRef] [PubMed] 35. Keras. 2015. Available online: https://keras.io (accessed on 22 August 2018). 36. Abadi, M.; Agarwal, A.; Barham, P.; Brevdo, E.; Chen, Z.; Citro, C.; Corrado, G.S.; Davis, A.; Dean, J.; Devin, M.; et al. TensorFlow: Large-Scale Machine Learning on Heterogeneous Systems, 2015. Available online: tensorflow.org (accessed on 29 November 2018) 37. Srivastava, N.; Hinton, G.E.; Krizhevsky, A.; Sutskever, I.; Salakhutdinov, R. Dropout: A simple way to prevent neural networks from overfitting. J. Mach. Learn. Res. 2014, 15, 1929–1958. 38. Bland, J.M.; Altman, D. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986, 327, 307–310. [CrossRef] 39. Lindsay, T.R.; Noakes, T.D.; McGregor, S.J. Effect of treadmill versus overground running on the structure of variability of stride timing. Percept. Mot. Skills 2014, 118, 331–346. [CrossRef] [PubMed] 40. Abdul Razak, A.H.; Zayegh, A.; Begg, R.K.; Wahab, Y. Foot plantar pressure measurement system: A review. Sensors 2012, 12, 9884–9912. [CrossRef] [PubMed] 41. MPU-9250 Datasheet. Available online: https://www.invensense.com/download-pdf/mpu-9250- datasheet/ (accessed on 22 August 2018). Sensors 2018, 18, 4194 22 of 22 42. Zrenner, M.; Ullrich, M.; Zobel, P.; Jensen, U.; Laser, F.; Groh, B.; Dümler, B.; Eskofier, B. Kinematic parameter evaluation for the purpose of a wearable running shoe recommendation. In Proceedings of the 15th IEEE International Conference on Wearable and Implantable Body Sensor Networks (BSN), Las Vegas, NV, USA, 4–7 March 2018; pp. 106–109. 43. Mo, S.; Chow, D.H. Accuracy of three methods in gait event detection during overground running. Gait Posture 2018, 59, 93–98. [CrossRef] [PubMed] c⃝ 2018 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

Comparison of Different Algorithms for Calculating Velocity and Stride Length in Running Using Inertial Measurement Units.

11-30-2018

Zrenner, Markus,Gradl, Stefan,Jensen, Ulf,Ullrich, Martin,Eskofier, Bjoern M

eng

PMC7505455

RESEARCH ARTICLE Player load in male elite soccer: Comparisons of patterns between matches and positions Terje DalenID1*, Tore Kristian Aune1, Geir Håvard Hjelde2, Gertjan EttemaID3, Øyvind SandbakkID3, David McGhie3 1 Department of Physical Education and Sport Science, Nord University, Levanger, Norway, 2 Rosenborg FC, Trondheim, Norway, 3 Centre for Elite Sports Research, Department of Neuromedicine and Movement Science, Norwegian University of Science and Technology, Trondheim, Norway * terje.dalen@nord.no Abstract Our primary aim was to explore the development of player load throughout match time (i.e., the pattern) using moving 5-min windows in an elite soccer team and our secondary aim was to compare player load patterns between different positions within the same team. The dataset included domestic home matches (n = 34) over three seasons for a Norwegian Elite League team. Player movements (mean ± SD age 25.5 ± 4.2 years, height 183.6 ± 6.6 cm, body mass 78.9 ± 7.4 kg) were recorded at 20 Hz using body-worn sensors. Data for each variable (player load, player load per meter, total distance, accelerations, decelerations, sprint distance, high-intensity running distance) were averaged within positions in each match, converted to z-scores and averaged across all matches, yielding one time series for each variable for each position. Pattern similarity between positions was assessed with cross-correlations. Overall, we observed a distinct pattern in player load throughout match time, which also occurred in the majority of individual matches. The pattern shows peaks at regular intervals (~15 min), each followed by a period of lower load, declining until the next peak. The same pattern was evident in player load per meter. The cross-correlation analy- ses support the visual evidence, with correlations ranging 0.88–0.97 (p < .001) in all position pairs. In contrast, no specific patterns were discernible in total distance, accelerations, decelerations, sprint distance and high-intensity running distance, with cross-correlations ranging 0.65–0.89 (p < .001), 0.32–0.64 (p < .005), 0.18–0.65 (p < .005 in nine position pairs), 0.02–0.38 (p < .05 in three pairs) and 0.01–0.52 (p < .05 in three pairs), respectively. This study demonstrated similarity in player load patterns between both matches and posi- tions in elite soccer competition, which could indicate a physical “pacing pattern” employed by the team. Introduction For optimal performance in team sports like soccer (association football), players are required to maximize their technical, tactical, and physical abilities. The physical demands of soccer matches are characterized by a constant variation between low- (e.g., standing and walking), PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 1 / 13 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Dalen T, Aune TK, Hjelde GH, Ettema G, Sandbakk Ø, McGhie D (2020) Player load in male elite soccer: Comparisons of patterns between matches and positions. PLoS ONE 15(9): e0239162. https://doi.org/10.1371/journal. pone.0239162 Editor: Laurent Mourot, University of Bourgogne France Comte´, FRANCE Received: June 2, 2020 Accepted: August 31, 2020 Published: September 21, 2020 Peer Review History: PLOS recognizes the benefits of transparency in the peer review process; therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. The editorial history of this article is available here: https://doi.org/10.1371/journal.pone.0239162 Copyright: © 2020 Dalen et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. high- (e.g., running), and very high-intensity (e.g., accelerations, decelerations and sprinting) activities [1–3]. Along with additional sport-specific activities (e.g., tackles, turns, headers, dribbles), these locomotor activities constitute the total physical load of a player during train- ing and matches [4]. However, the total physical load of the players is determined by a combi- nation of direct involvement in play, responding to movements of attacking players, tactical restrictions, and willingness to support team-mates [5]. These variations are likely to result in a relatively large match to match variability in physical performance [6, 7]. Time-motion analyses have provided accurate and objective quantification of the players’ activities, and therefore improved our understanding of the physical demands in soccer [8– 11]. However, measurements of different locomotor classifications or speed zones may be insensitive to the totality of mechanical stresses common to team sports. Tri-axial accelerome- ters provide complementary information to time-motion analysis for understanding player load during matches and training [12, 13] as they record the acceleration of body movement in three dimensions, which better estimates the players’ physical exertion. Therefore, manufac- turers of global positioning systems (GPS) and local positioning measurements (LPM) have incorporated high-resolution triaxial accelerometers as a measure of player load. Such analyses are shown useful for validly quantifying the physical demands in soccer [12, 14–16], in which various estimations of player load are regarded as acceptable measures of external load and largely correlated to players’ physiological and perceptual responses to training [17, 18] To date, monitoring external training and match load measures in soccer has tended to rely on results based on locomotor activities. In previous analyses of soccer matches, considerable heterogeneity has been observed in the within-match development of locomotor activities (total distance, HiR, sprint, accelerations and decelerations) throughout match time (i.e., the pattern) across studies [6, 9, 19–25]. Some studies report a reduction in total and high-inten- sity running (HiR) distances toward the end of each half [9, 26], whereas others do not find such changes [20, 21]. These contradictory results are likely caused by different measurement systems, different tactical elements, opponents’ playing style, pacing strategies, score line, and team formation, which would all affect the players’ ability to regulate and maintain their physi- cal effort [22]. However, previous studies show high variability in high-speed activities within matches and that individual players show inconsistency in high-speed activity (i.e., HiR and sprinting) across matches [6, 23]. A component of soccer matches that has received relatively less attention is the players’ number of accelerations and decelerations [19], although some previous studies suggest that inter- and intra-individual variability is smaller for accelerations compared to distance-related measures [6, 24]. Additionally, a recent study found a continu- ous reductional pattern in accelerations over the course of a match and after peak working periods of a match, which was consistent across positions [25]. In the existing literature, the within-match player load based on three-dimensional move- ment analyses has been investigated using a standardised soccer simulation with 15-min stan- dardised activity blocks [27]. Here, the authors found that player load increased over time in each half, likely due to a change in movement strategy and/or a reduced locomotor efficiency [27]. In contrast to this, reductions in player load were identified in the latter stages of each half in the analyses of 86 matches in U-21 English Championship teams [14]. However, in the same 15-min time periods, the player load per total distance covered increased, suggesting an increased loading for every given meter covered on the pitch [14]. These investigations have allowed a general determination of player load patterns during soccer matches and soccer-spe- cific intermittent exercises. However, to understand more in detail how teams and individual players distribute their player load and related locomotor activities throughout soccer matches, the same factors need to be analyzed over shorter time-periods than 15-min blocks. More instantaneous analyses of player load and the corresponding activities during soccer matches PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 2 / 13 Funding: The funder (Rosenborg FC) provided support in the form of salaries for author [G.H.H.], but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. Competing interests: The commercial affiliation (Rosenborg FC) does not alter our adherence to all PLOS ONE policies on sharing data and materials by including the following statement: "This does not alter our adherence to PLOS ONE policies on sharing data and materials.” would logically show a variable “pacing” influenced by e.g., tactical elements, player position, and the level of the opponents. In order to quantify this across e.g., positions, the similarity of patterns throughout the duration of matches must be analyzed. Long term analyses of such data and the relationships to changes in tactics, different opponents, and match outcome have the potential to provide imperative understanding of how the team and the players in different positions distribute the load (i.e., “pacing strategies”) during different types of matches. Since analyses based on predefined periods cannot provide information about the “real” peaks and valleys in the analysis of patterns throughout a match, moving windows is a poten- tial solution, providing more accurate information about player load and locomotor variables (total distance, acceleration, deceleration, HiR and sprint). Our primary aim was to explore the patterns of player load, as well as locomotor variables for comparison, with analyses from moving 5-min windows in an elite soccer team. Our secondary aim was to compare these pat- terns between different positions within the same team. Methods Participants The dataset includes domestic home matches (n = 34) over three full seasons for a team in the Norwegian Elite League. In one of the seasons, the team participated in the Europe League group stages. All matches were played on a grass surface. Movements of all players (mean ± SD age 25.5 ± 4.2 years, height 183.6 ± 6.6 cm, body mass 78.9 ± 7.4 kg) were observed, and only data from the 39 players completing an entire match were used (n = 212: complete match data of players, goalkeepers excluded). The sample included eight central defenders (CD, n = 47), six external defenders (ED, n = 52), six central midfielders (CM, n = 46), 11 external midfield- ers (EM, n = 40), and eight attackers (ATT, n = 27). Some players participated in different posi- tions across, but not within, the matches included in the data material. Following an explanation of the procedures, all participants gave verbal and written informed consent to participate in the study. The study was conducted according to the Declaration of Helsinki and has been approved by the Norwegian Social Science Data Services (reference number 468065). Study design and methodology This study used a fully automatic sport tracking system to evaluate match performances of pro- fessional soccer players at the elite level over three full seasons. Player movement was captured by small, body-worn sensors located at the lumbar region, continuously recording the players’ actions. Data were transferred by microwave radio channel to 10 RadioEyeTM sensors (ZXY SportTracking, ChyronHego, Trondheim, Norway) mounted in the team’s home arena. Player movement was registered at 20 Hz. Accelerations and decelerations were recorded when they reached limits of 2 m.s-2 and -2 m.s-2, respectively, and a HiR category of >19.8 km.h-1 and sprint category of >25.2 km.h-1 were selected for this study. The thresholds for accelerations, HiR, and sprint were similar to those reported in previous studies [12, 28]. In this study, the player load is calculated as a downscaled (by a factor of 800) value of the sum of the squared, high pass-filtered accelerometer values for the respective axes (X, Y, and Z): (X2 + Y2 + Z2) / 800 [12]. Test-retest reliability of the sport tracking system is reported earlier, indicating good reliability [12, 28]. Evaluation of 5-minute periods throughout match time. To construct an analysis cap- turing the immediate, dynamic nature of a match for all players, mean values were calculated over consecutive (i.e., moving) 5-min periods for player load and player load per meter, as well as time-motion variables (total distance, accelerations, decelerations, sprint distance, HiR PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 3 / 13 distance) for comparison, beginning with the first five minutes of the match [25, 29]. The sec- ond 5-min period lasts from the second to the sixth minute, and so on. This method is argued to provide a more accurate representation of the distances covered by players [29]. These 5-min periods were used to investigate patterns of player load and locomotor variables throughout match time. The similarities of patterns were then quantified between positions and patterns were evaluated across variables. Statistical analysis All data processing and statistical analysis was performed in Matlab R2019b version 9.7.0.1190202 (Mathworks, Natick, MA, USA). For each match, data for each variable (player load, player load per meter, total distance, accelerations, decelerations, sprint distance, HiR distance) were averaged within positions if there was data from multiple players at the same position, yielding a single time series per variable for each position measured in that match. These data were then converted to z-scores, to facilitate the direct comparison of patterns, dis- regarding absolute magnitudes. Finally, the z-scores were averaged across all matches for each position, resulting in one time series for each position for each variable. The degree of similar- ity of patterns between positions was assessed with cross-correlations. For statistical purposes, the break in the time series caused by halftime was disregarded (i.e., the data were treated as continuous for the duration of playing time). Linearity was assessed visually using scatter plots. Cross-correlations were calculated for every position pair for n-1 lags at either side of zero, where n = 82, the number of moving 5-min windows in a 90-min match (41 5-min win- dows in each 45-min half). To best represent the development of player load and time-motion variables across positions throughout match time, the correlation at zero lag (with 95% confi- dence interval and p-value) is presented. For comparison, maximum correlations and corre- sponding lags are also reported. A negative lag means that the first time series (player position in table columns) shifts to the left relative to the second time series (player position in table rows). The level of statistical significance was set at α = .05. Correlation values were interpreted categorically as trivial (0–0.1), low (0.1–0.3), moderate (0.3–0.5), high (0.5–0.7), very high (0.7–0.9), or nearly perfect (0.9–1) using the scale presented by Hopkins et al. [30]. Results Overall, we observed a distinct pattern in player load throughout match time (Fig 1, black line). The pattern shows peaks at seemingly regular intervals (~15 min), each followed by a period of lower load, typically declining until the next peak. This pattern was clear in all posi- tions (Fig 1A, colored lines), and could also generally be observed in the majority of individual matches (Fig 2). The cross-correlation analysis (Table 1) supports the visual evidence, indicat- ing very high to nearly perfect correlations (range 0.88–0.95, all p < .001) in all position pairs, all having the highest correlation at zero lag. The same pattern was evident for player load per meter, both overall (Fig 1B, black line) and in all positions (Fig 1B, colored lines), with nearly perfect correlation values (range 0.93–0.97, all p < .001; Table 1) in all position pairs, all having the highest correlation at zero lag. For total distance, no distinct pattern throughout match time was evident (Fig 3A, black line). However, the patterns for all positions appear to follow each other reasonably well (Fig 3A, colored lines), which is reflected in high to very high correlation values (range 0.65–0.89, all p < .001; Table 1), with all position pairs again having the highest correlation at zero lag. For accelerations, no specific pattern was evident throughout match time (Fig 3B, black line), but the different positions appear to follow roughly similar patterns (Fig 3B, colored lines). Further, correlation values were moderate to high (range 0.32–0.64, all p  .005; S1 PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 4 / 13 Table), with all but one position pair having the highest correlation at zero lag (EM vs. CD highest absolute correlation 0.56, lag -27; S2 Table). For decelerations, again no specific pattern was evident throughout match time (Fig 3C, black line), but the different positions sporadically follow roughly similar patterns (Fig 3C, colored lines). Correlation values were low to high (range 0.18–0.65, all but one p  .005; S1 Table), with more than half of all position pairs hav- ing the highest correlation at zero lag (highest correlation absolute range 0.32–0.65, lag -4–44; S2 Table). For sprint distance and HiR distance, no specific pattern throughout match time could be discerned in either variable (Fig 3D and 3E, black lines). Further, the patterns for the different positions do not follow each other well (Fig 3 and 3E, colored lines). In line with this, trivial to moderate correlation values were found for sprint distance (absolute range 0.02–0.38, p < .05 in three position pairs, two having the highest correlation at zero lag; highest correlation abso- lute range 0.29–0.53, lag -29–54 [S1 and S2 Tables]), whereas trivial to moderate (one high) correlation values were found for HiR distance (absolute range 0.01–0.52, p < .05 in three position pairs, two having the highest correlation at zero lag; highest correlation absolute range 0.32–0.52, lag -28–32 [S1 and S2 Tables]). Discussion The primary aim of this study was to explore the patterns of player load with analyses from moving 5-min windows in an elite soccer team. Further, the secondary aim was to compare the player load patterns between different positions within the same team. The main finding was the distinct player load pattern with three “high-load periods” in each half, separated by “lower-load periods”. The player load patterns were relatively similar between positions and Fig 1. Mean values (z-scores) of player load and player load per meter in 5-min moving windows throughout match time across all matches (n = 34) for each position (colored lines) and for all positions combined (black line). A: player load; B: player load per meter. https://doi.org/10.1371/journal.pone.0239162.g001 PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 5 / 13 occurred at approximately the same time points during the majority of matches. These novel findings will be discussed with two points of departure: the team’s pacing strategy from a phys- ical viewpoint and from a perspective based on interpersonal coordination between player positions. Player load patterns and pacing The use of 5-min moving averages to analyze within-match player load patterns in this study allowed us to study the players’ “pacing strategies” (i.e., distribution of player load and related locomotor activities) in more detail than in previous studies evaluating simulated soccer matches [27] and English championship matches [16] by dissection into 15-min periods. The present results show distinct player load patterns with three “high-load periods” in each half of the match (Fig 1), separated by “lower-load periods”, in most of the matches (Fig 2), which dif- fers from patterns found in research on English championship players [16]. Although the new methodology for analyzing player load used in the present study provides novel information about high- and lower-load periods of the soccer matches, these distinct patterns found in almost all matches were rather surprising since differences between the opponents’ level and tactics should rationally have influenced player load patterns between matches. In addition, the player load would also largely be determined by the players’ decision-making about oppor- tunities to become engaged in play. One likely explanation of this apparent player load pattern is that this study investigated one of the top-ranked clubs in the Norwegian top division at Fig 2. Mean player load (z-scores) in 5-min moving windows throughout match time for all measured positions per match. M: match number. https://doi.org/10.1371/journal.pone.0239162.g002 PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 6 / 13 their home arena, where they had the opportunity to “control the match” in most of the matches. Thus, it seems reasonable to ask whether these similar positional fluctuations in player load are typical for this team at their home arena matches where they normally were the dominant team. Therefore, an interesting approach for future studies would be to investigate these patterns with the same moving average-method in teams at different performance levels (i.e., if the investigated team or the opposition controls the match or in teams with different overall tactical dispositions). Since locomotor actions in soccer are not performed in isolation, consideration of player load as a proxy for “overall external load” might be useful. A previous investigation of player load found high to very high associations between player load and measures of internal train- ing load (TRIMP and sRPE) [18], with internal load being especially related to the volume of accelerations. Barrett et al. [18] found nearly perfect within-subject correlation between player load and heart rate/VO2, but trivial to moderate association for the between-subject correla- tion on the same variable [19]. Overall, this suggests that the fluctuations in player load found in the present study are also associated with fluctuations in internal load, thereby indicating a physical “pacing pattern” (pattern in distribution of load) employed by the investigated team (Fig 1). These “pacing patterns” were relatively similar between positions and occurred at the same time point during the matches (Fig 2), even though the different positions have different roles during attacks and defense; one single attack gives higher intensities on attacking players, but not for the defending players, and vice versa. However, the time scale with 5-min moving averages is too long to differentiate between high-intensity periods based on one single attack or one defensive stand and normally contain several attacking and defensive actions. More- over, player load patterns based on moving 5-min windows will give more information about Table 1. Cross-correlations [95% CI] of mean position values (z-scores) across all matches (n = 34) at zero lag for player load, player load per meter, and total distance. CD ED CM EM ATT Player load CD --- ED 0.95 [0.92, 0.96] --- CM 0.93 [0.89, 0.95] 0.94 [0.91, 0.96] --- EM 0.93 [0.89, 0.95] 0.94 [0.91, 0.96] 0.93 [0.90, 0.96] --- ATT 0.91 [0.87, 0.94] 0.95 [0.93, 0.97] 0.88 [0.83, 0.92] 0.89 [0.84, 0.93] --- Player load per meter CD --- ED 0.93 [0.89, 0.95] --- CM 0.95 [0.93, 0.97] 0.95 [0.92, 0.97] --- EM 0.95 [0.92, 0.97] 0.94 [0.91, 0.96] 0.95 [0.92, 0.97] --- ATT 0.93 [0.90, 0.96] 0.97 [0.96, 0.98] 0.96 [0.94, 0.97] 0.95 [0.92, 0.97] --- Total distance CD --- ED 0.88 [0.81, 0.92] --- CM 0.80 [0.71, 0.87] 0.84 [0.76, 0.89] --- EM 0.89 [0.84, 0.93] 0.87 [0.81, 0.92] 0.86 [0.79, 0.91] --- ATT 0.76 [0.65, 0.84] 0.71 [0.59, 0.81] 0.65 [0.51, 0.76] 0.72 [0.59, 0.81] --- CD = central defender; ED = external defender; CM = central midfielder; EM = external midfielder; ATT = attacker. All correlations p < .001. For all correlations, the maximum value occurred at zero lag. https://doi.org/10.1371/journal.pone.0239162.t001 PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 7 / 13 the overall load of the match, instead of detailed information about when the team is attacking (more load on offensive players) or defending (more load on defensive players). The present study shows very high to nearly perfect associations between positional pat- terns of player load and player load per meter (Table 1). Hence, the periods of high player load are associated with movements on the field that increases the player load per meter, which is shown to be associated with unorthodox movements such as jumping, tackling, collisions, passing, accelerations, decelerations etc., movement which are common for soccer and detected when triaxial accelerometers are employed [12, 13]. Although this study found differ- ences in the absolute values of the highest and lowest player load periods in the presented results, there were no positional differences in the pattern of increase and decrease of player load throughout the matches (Fig 1). Thus, the present study is the first to report similarity Fig 3. Mean values (z-scores) of time-motion variables in 5-min moving windows throughout match time across all matches (n = 34) for each position (colored lines) and for all positions combined (black lines). A: total distance; B: accelerations; C: decelerations; D: sprint distance; E: high-intensity running distance (HiR). https://doi.org/10.1371/journal.pone.0239162.g003 PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 8 / 13 across playing positions in the player load patterns throughout matches in male elite soccer players. The use of this approach and the findings from this study may contribute to new hypotheses concerning the patterns of player load and intensity throughout a soccer match. Therefore, before one can conceptualize more in-field applications, different aspects of player load patterns should be investigated further. Whereas other investigations show a considerable heterogeneity in the within-match pat- tern of total distance, HiR and sprint across studies [9, 20, 21], in this study, total distance was the variable besides player load and player load per meter which displayed the highest correla- tion between positions, with no lag between positional patterns (Table 1). Regardless of this, the patterns of total distance for the different positions do not follow the same distinct pattern as the player load variables. For accelerations and deceleration, no specific pattern was evident throughout match time (Fig 3B and 3C), but the different positions appear to follow roughly similar patterns with correlations ranging from low to high (S1 Table). For sprint and HiR dis- tance, the present study shows no meaningful similarities between positions, with negligible to moderate cross-correlations (S1 Table). These findings are similar to those from other studies investigating high-intensity patterns [6, 7]. In the present results, the patterns of the different HiR and sprint distance throughout match time show heterogeneity; patterns of sprint and HiR distance show that high-intensity periods occur at different times both between matches and between positions. These differences could be caused by different tactical elements, oppo- nents playing style, pacing strategies, score line, and team formation, which would all affect the players’ ability to regulate their physical effort and maintain work rates at appropriate levels [22]. Player load and interpersonal coordination patterns The observed in-phase pattern for player load in this study is also interesting from perspectives of interpersonal coordination patterns, and it demonstrates that the interaction in player load between the team’s subunits probably is more complex than the behavior of each individual player considered separately [31, 32]. Specifically related to soccer, the actions of one player or a player subunit (e.g., attackers, midfielders, defenders) cause re-actions and adjustments from other players or player subunits to stabilize performance, and these adjustments interact and influence player load collectively. The emergence of the synchronized player load patterns between subunits is likely self-organized to improve team performance and is a result of the interactions of a player’s constraints and information exchange within their own team and those imposed by the opponent. What type of constraints and information that evolves in spontaneous self-organization and synchronization of player load is not easy to identify, but might be easily understood intuitively. Examples of such constraints in soccer could be other players’ positions and movements, position and speed of the ball, tactical decisions, fatigue, etc. According to the rationale by Haken and Portugali [33], if the meaning of a player’s action is understood (information exchange), it triggers action and changes the structure or behavior (player load) in the whole team. E.g., the reaction of players on the action of another depends on the success or failure (information) of that action. The interesting finding of the present study is that, even though each action’s success or failure may occur randomly, the player load pattern that evolves seems very stable. Thus, the interpersonal patterns of coordination of player load in a soccer team might be modelled as an open complex dynamical system at a behavioral level of analysis, as suggested in evolutionary game theory [34]. Given the stable player load pattern over various matches, even though a soccer match is the complex combina- tion of actions by individuals, no individual player (or subunit) seems to initiate or control the behavior of the match. In other words, each player is enslaved in a self-organized system that PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 9 / 13 at the same time consists of all these same players. This self-organized system could be affected by the fact that this study investigated one of the top-ranked clubs at their home arena, which could have produced a more consistent player load pattern due to typically being the dominant team. Limitations Since this study investigated one of the top-ranked clubs at their home arena, it is possible a more consistent player load pattern was produced due to typically being the dominant team. It is unclear to what extent the results will replicate across teams or if they are particular to either the investigated team or e.g., teams sharing certain characteristics. This study did not investigate differences between various tactical elements, opponents’ playing styles, ball in versus out of play, score line, or team formations, which could all affect the players’ ability to regulate their physical effort and maintain work rate profiles. Differences in measurement technology makes it difficult to compare player load variable between different tracking systems (or even different versions of the same system), since differences in measurement technology could partly account for eventual discrepancies between the values registered in this study and other studies. Hence caution is required when comparing analyses of football match activities across studies. Conclusion This study demonstrated similarity in player load patterns between positions in elite soccer matches. The novelty is the clear pattern which consists of three high-load periods in both halves, where these “high load” periods are followed by periods with reduced load. The present study did not find similar unambiguous patterns on any of the locomotor variables. The evi- dent pattern in player load indicates a physical “pacing pattern” employed by the team. These “pacing patterns” were relatively similar between positions and occurred at the same time points during the matches over three successive seasons. From the perspective of interpersonal coordination patterns, these synchronized player load patterns between positions are likely self-organized to improve team performance and are a result of the interactions of the players’ constraints and information exchange within their own team and those imposed by the oppo- nents. It should be noted that a more consistent player load pattern might have been produced due to the investigated team being a top-ranked club playing home matches. Practical applications Since this study is the first to report this distinct pattern of player load it is important that more studies of player load patterns are conducted, in teams at different performance levels before in-field applications can be firmly conceptualized. Considering the previously reported high association between player load and internal training load, it could be argued that coaches might want to regulate player load in training for an overreaching effect. This could eventually allow for a more aggressive pacing strategy, shortening the lower-load periods and hence put- ting more pressure on the opposition. However, an approach like this must be cautious against overloading. During matches, coaches can also use the method proposed here in real-time to monitor if certain players or position groups appear to be “out of sync” with the rest of the team. Supporting information S1 Table. Cross-correlations [95% CI] of mean position values (z-scores) across all matches (n = 34) at zero lag for accelerations, decelerations, sprint distance, and high-intensity PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 10 / 13 running distance. CD: central defender; ED: external defender; CM: central midfielder; EM: external midfielder; ATT: attacker. For p-values, bold text indicates significance at α = .05. (DOCX) S2 Table. Maximum cross-correlations (corresponding lag) of mean position values (z- scores) across all matches (n = 34) for accelerations, decelerations, sprint distance, and high-intensity running distance. CD: central defender; ED: external defender; CM: central midfielder; EM: external midfielder; ATT: attacker. A negative lag means that the first time series (player position in table columns) shifts to the left relative to the second time series (player position in table rows). (DOCX) S1 Dataset. (XLSX) Acknowledgments We thank the players for their efforts throughout the period. Author Contributions Conceptualization: Terje Dalen, Tore Kristian Aune, Geir Håvard Hjelde, Øyvind Sandbakk, David McGhie. Data curation: Terje Dalen, David McGhie. Formal analysis: Terje Dalen, Øyvind Sandbakk, David McGhie. Investigation: Terje Dalen, Geir Håvard Hjelde, David McGhie. Methodology: Terje Dalen, Øyvind Sandbakk, David McGhie. Project administration: Terje Dalen, David McGhie. Resources: Terje Dalen. Supervision: Terje Dalen. Validation: Terje Dalen. Visualization: Terje Dalen, David McGhie. Writing – original draft: Terje Dalen, Tore Kristian Aune, Øyvind Sandbakk, David McGhie. Writing – review & editing: Terje Dalen, Tore Kristian Aune, Gertjan Ettema, Øyvind Sand- bakk, David McGhie. References 1. Ingebrigtsen J.; Bendiksen M.; Randers M.B.; Castagna C.; Krustrup P.; Holtermann A. Yo-Yo IR2 test- ing of elite and sub-elite soccer players: Performance, heart rate response and correlations to other interval tests. J Sports Sci 2012, 30, 1337–1345 https://doi.org/10.1080/02640414.2012.711484 PMID: 22867048 2. Bloomfield J.; Polman R.; O’Donoghue P. Physical demands of different positions in FA Premier League soccer. J Sports Sci Med 2007, 6, 63–70. PMID: 24149226 3. Krustrup P.; Andersson H.; Mohr M.; Randers M.B.; Jensen J.M.; Zebis M. et al. Match activities and fatigue development of elite female soccer players at different levels of competition. Science and Foot- ball VI: the proceedings of the Sixth World Congress on Science and Football 2009, 205–211. 4. Mallo J.; Navarro E. Physical load imposed on soccer players during small-sided training games. J Sports Med Phys Fitness 2008, 48, 166–171. PMID: 18427410 PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 11 / 13 5. Drust B.; Atkinson G.; Reilly T. Future Perspectives in the Evaluation of the Physiological Demands of Soccer. Sports Med 2007, 37, 783–805. https://doi.org/10.2165/00007256-200737090-00003 PMID: 17722949 6. Gregson W.; Drust B.; Atkinson G.; Salvo V.D. Match-to-match variability of high-speed activities in pre- mier league soccer. Int J Sports Med 2010, 31, 237–242. https://doi.org/10.1055/s-0030-1247546 PMID: 20157871 7. Carling C.; Bradley P.; McCall A.; Dupont G. Match-to-match variability in high-speed running activity in a professional soccer team. J Sports Sci 2016, 34, 2215–2223. https://doi.org/10.1080/02640414. 2016.1176228 PMID: 27144879 8. Rebelo A.; Brito J.; Seabra A.; Oliveira J.; Drust B.; Krustrup P. A New Tool to Measure Training Load in Soccer Training and Match Play. Int J Sports Med 2012, 297–304. https://doi.org/10.1055/s-0031- 1297952 PMID: 22290322 9. Bradley P.S.; Di Mascio M.; Peart D.; Olsen P.; Sheldon B. High-intensity activity profiles of elite soccer players at different performance levels. J Strength Cond Res 2010, 24, 2343–2351. https://doi.org/10. 1519/JSC.0b013e3181aeb1b3 PMID: 19918194 10. Mohr M.; Krustrup P.; Bangsbo J. Fatigue in soccer: a brief review. J Sports Sci 2005, 23, 593–599. https://doi.org/10.1080/02640410400021286 PMID: 16195008 11. Carling C.; Bloomfield J.; Nelsen L.; Reilly T. The Role of Motion Analysis in Elite Soccer: Contemporary Performance Measurement Techniques and Work Rate Data. Sports Med 2008, 38, 839–862. https:// doi.org/10.2165/00007256-200838100-00004 PMID: 18803436 12. Dalen T.; Ingebrigtsen J.; Ettema G.; Hjelde G.H.; Wisløff U. Player load, acceleration, and deceleration during forty-five competitive matches of elite soccer. J Strength Cond Res 2016, 30, 351–359. https:// doi.org/10.1519/JSC.0000000000001063 PMID: 26057190 13. Dalen T.; Overas O.; van den Tillaar R.; Welde B.; von Heimburg E.D. Influence of different soccer-spe- cific maximal actions on physiological, perceptual and accelerometer measurement loads. Open Access J Sports Med 2018, 9, 107–114, https://doi.org/10.2147/OAJSM.S167347 PMID: 29942166 14. Barrett S.; Midgley A.; Reeves M.; Joel T.; Franklin E.; Heyworth R.et al. The within-match patterns of locomotor efficiency during professional soccer match play: Implications for injury risk? J Sci Med Sport 2016, 19, 810–815. https://doi.org/10.1016/j.jsams.2015.12.514 PMID: 26787341 15. Scott B.R.; Lockie R.G.; Knight T.J.; Clark A.C.; De Jonge X.A.K.J. A Comparison of Methods to Quan- tify the In-Season Training Load of Professional Soccer Players. Int J Sports Physiol Perform 2013, 8, 195–202. https://doi.org/10.1123/ijspp.8.2.195 PMID: 23428492 16. Gaudino P.; Iaia F.M.; Strudwick A.J.; Hawkins R.D.; Alberti G.; Atkinson G. et al. Factors Influencing Perception of Effort (Session Rating of Perceived Exertion) During Elite Soccer Training. Int J Sports Physiol Perform 2015, 10, 860–864. https://doi.org/10.1123/ijspp.2014-0518 PMID: 25671338 17. Casamichana D.; Castellano J.; Calleja-Gonzalez J.; San Roman J.; Castagna C. Relationship between indicators of training load in soccer players. J Strength Cond Res 2013, 27, 369–374. https://doi.org/10. 1519/JSC.0b013e3182548af1 PMID: 22465992 18. Barrett S.; Midgley A.; Lovell R. PlayerLoad™: Reliability, Convergent Validity, and Influence of Unit Position During Treadmill Running. Int J Sports Physiol Perform 2014, 9, 945–952. https://doi.org/10. 1123/ijspp.2013-0418 PMID: 24622625 19. Harper D. J.; Carling C.; Kiely J. (2019). High-Intensity Acceleration and Deceleration Demands in Elite Team Sports Competitive Match Play: A Systematic Review and Meta-Analysis of Observational Stud- ies. Sports Med 2019 49(12): 1923–1947. https://doi.org/10.1007/s40279-019-01170-1 PMID: 31506901 20. Wehbe G.M.; Hartwig T.B.; Duncan C.S. Movement Analysis of Australian National League Soccer Players Using Global Positioning System Technology. J Strength Cond Res 2014, 28, 834–842. https://doi.org/10.1519/JSC.0b013e3182a35dd1 PMID: 23897019 21. Rampinini E.; Couts A.J.; Castagna C.; Sassi R.; Impellizzeri F.M. Variation in Top Level Soccer Match Performance. Int J Sports Med 2007, 28, 1018–1024. https://doi.org/10.1055/s-2007-965158 PMID: 17497575 22. Waldron M.; Highton J. Fatigue and Pacing in High-Intensity Intermittent Team Sport: An Update. Sports Med 2014, 44, 1645–1658. https://doi.org/10.1007/s40279-014-0230-6 PMID: 25047854 23. Akenhead R.; Hayes P.R.; Thompson K.G.; French D. Diminutions of acceleration and deceleration out- put during professional football match play. J Sci Med Sport 2013, 16, 556–561. https://doi.org/10. 1016/j.jsams.2012.12.005 PMID: 23333009 24. Carling C. Interpreting Physical Performance in Professional Soccer Match-Play: Should We be More Pragmatic in Our Approach? Sports Med 2013, 43, 655–663. https://doi.org/10.1007/s40279-013- 0055-8 PMID: 23661303 PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 12 / 13 25. Dalen T.; Lorås H.; Hjelde G.H.; Kjøsnes T.N.; Wisløff U. Accelerations–a new approach to quantify physical performance decline in male elite soccer? Eur J Sport Sci 2019, 19(8), 1015–1023. https://doi. org/10.1080/17461391.2019.1566403 PMID: 30632940 26. Bradley P. S.; Sheldon W.; Wooster B.; Olsen P.; Boanas P.; Krustrup P. (2009). High-intensity running in English FA Premier League soccer matches. J Sports Sci 2009, 27(2): 159–168. 27. Barrett S.; Midgley A.W.; Towlson C.; Garrett A.; Portas M.; Lovell R. Within-Match PlayerLoad™ Pat- terns During a Simulated Soccer Match: Potential Implications for Unit Positioning and Fatigue Manage- ment. Int J Sports Physiol Perform 2016, 11, 135–140. https://doi.org/10.1123/ijspp.2014-0582 PMID: 26114855 28. Ingebrigtsen J, Dalen T, Hjelde GH, Drust B, Wisloff U. Acceleration and sprint profiles of a professional elite football team in match play. Eur J Sport Sci 2015, 15(2), 101–110. https://doi.org/10.1080/ 17461391.2014.933879 PMID: 25005777 29. Varley M.C.; Elias G.P.; Aughey R.J. Current Match-Analysis Techniques’ Underestimation of Intense Periods of High-Velocity Running. Int J Sports Physiol Perform 2012, 7, 183–185. https://doi.org/10. 1123/ijspp.7.2.183 PMID: 22634968 30. Hopkins W.; Marshall S.; Batterham A.; Hanin J. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc 2009, 41, 3–13. https://doi.org/10.1249/MSS. 0b013e31818cb278 PMID: 19092709 31. Schmidt R.C.; Fitzpatrick P. Dynamical perspective on motor learning. In In, Zelaznik H.N. (ed.), Advances in motor learning and control, Champaign, Ill., Human Kinetics Publishers, 1996, p. 195– 223; 1996. 32. Turvey M.T. Coordination. Am Psychol 1990, 45, 938. https://doi.org/10.1037//0003-066x.45.8.938 PMID: 2221565 33. Haken H.; Portigali J. Information and self-organization. Entropy 2017, 19, 18. 34. Santos F.; Encarnacão S.; Santos F.; Portugali J.; Pacheco J. An evolutionary game theoretic approach to multi-sector coordination and self-organization. Entropy 2016, 18, 152. PLOS ONE Player load pattern in soccer matches PLOS ONE | https://doi.org/10.1371/journal.pone.0239162 September 21, 2020 13 / 13

Player load in male elite soccer: Comparisons of patterns between matches and positions.

09-21-2020

Dalen, Terje,Aune, Tore Kristian,Hjelde, Geir Håvard,Ettema, Gertjan,Sandbakk, Øyvind,McGhie, David

eng

PMC9273616

1 Vol.:(0123456789) Scientific Reports | (2022) 12:11762 | https://doi.org/10.1038/s41598-022-15344-x www.nature.com/scientificreports Variation in human 3D trunk shape and its functional implications in hominin evolution Markus Bastir1*, José María González Ruíz1, Javier Rueda2, Gonzalo Garrido López2, Marta Gómez‑Recio1, Benoit Beyer3, Alejandro F. San Juan2,4 & Enrique Navarro2,4 This study investigates the contribution of external trunk morphology and posture to running performance in an evolutionary framework. It has been proposed that the evolution from primitive to derived features of torso shape involved changes from a mediolaterally wider into a narrower, and antero‑posteriorly deeper into a shallower, more lightly built external trunk configuration, possibly in relation to habitat‑related changes in locomotor and running behaviour. In this context we produced experimental data to address the hypothesis that medio‑laterally narrow and antero‑ posteriorly shallow torso morphologies favour endurance running capacities. We used 3D geometric morphometrics to relate external 3D trunk shape of trained, young male volunteers (N = 27) to variation in running velocities during different workloads determined at 45–50%, 70% and 85% of heart rate reserve (HRR) and maximum velocity. Below 85% HRR no relationship existed between torso shape and running velocity. However, at 85% HRR and, more clearly, at maximum velocity, we found highly statistically significant relations between external torso shape and running performance. Among all trained subjects those with a relatively narrow, flat torso, a small thoracic kyphosis and a more pronounced lumbar lordosis achieved significantly higher running velocities. These results support the hypothesis that external trunk morphology relates to running performance. Low thoracic kyphosis with a flatter ribcage may affect positively respiratory biomechanics, while increased lordosis affects trunk posture and may be beneficial for lower limb biomechanics related to leg return. Assuming that running workload at 45–50% HRR occurs within aerobic metabolism, our results may imply that external torso shape is unrelated to the evolution of endurance running performance. Evolutionary anatomical changes. The trunk consists of the ribcage, the spine and the pelvis. During human body shape evolution, each of these elements experienced specific morphological changes. For example, the ribcages of Homo erectus and Neandertals were not only wider at the level of the central and lower thorax, but also antero-posteriorly deeper than most modern human populations1–4. Also the pelvis shows a systemic evolutionary trend towards reduction of its bi-iliac width, when comparing modern humans with H. erectus and members of the Neandertal lineage5–8. Evolutionary changes in the spine of the genus Homo show changes in overall height, it’s position within the ribcage and possibly spine curvatures. Within Homo, the overall spine length has increased, as a consequence of larger body size9. Greater dorsal orientation of the transverse processes in non-modern humans likely positioned the thoracic vertebral bodies more within the ribcage, producing a greater spine invagination10,11. Also, in Neandertals a smaller lumbar lordosis (hypolordosis) is discussed and could be particularly relevant with respect to trunk morphology as it directly affects the position and orientation of the sacrum and, thus, the pelvis12–14. The potential adaptive significance and functional implications of these features in hominin trunk evolution are not well understood and have been discussed in the context of thermo- regulatory15, digestive16, respiratory3, and locomotor functions17. Here, we focus on the latter two aspects. Trunks with a narrow lower thorax and a narrow, tall waist have been associated with emerging endurance running capacities, possibly appearing with African H. erectus and together with elongated lower limbs18. Yet, a recent reconstruction of the KNM-WT 15,000 African H. erectus ribcage seems more similar to Neandertals in OPEN 1Paleoanthropology Group, Museo Nacional de Ciencias Naturales, CSIC, J.G. Abascal 2, 28006 Madrid, Spain. 2Department of Health and Human Performance, Faculty of Physical Activity and Sports Sciences-INEF, Universidad Politécnica de Madrid, 28040 Madrid, Spain. 3Laboratory of Functional Anatomy (LAF), Faculty of Motor Skills Sciences, Université Libre de Bruxelles, Brussels, Belgium. 4These authors jointly supervised this work: Alejandro F. San Juan and Enrique Navarro. *email: mbastir@mncn.csic.es 2 Vol:.(1234567890) Scientific Reports | (2022) 12:11762 | https://doi.org/10.1038/s41598-022-15344-x www.nature.com/scientificreports/ terms of width and depth than to modern human populations3. Nevertheless, Neandertals are thought to show adaptations for sprinting based on the anatomy of their foot skeleton19, and for power locomotion, as paleo- ecological and genetic evidence indicates, which is interpreted in the context of ambush hunting in a forested ecosystem20. Thus, given the new evidence for greater similarities of trunk shape in primitive Homo and Neandertals3, together with known differences in the lower limb anatomy,—i.e. longer limbs in H. erectus adapted to endurance running17,18,21,22, and shorter limbs with specialized feet in Neandertals adapted to sprinting19,20,23—it is inter- esting to investigate the implications of variation in trunk morphology in the context of locomotor capacities. Trunk anatomy and running capacities. The trunk contributes to locomotor performance and energet- ics in two different ways: (1) the effect of trunk morphology on limb biomechanics, and (2) the effect of thorax morphology on breathing mechanics. Grossly speaking, sprinting and endurance running differ at energetic and locomotor (limb) biomechanics in the context of stride lengths, frequency and energetics. It has been shown that runners with relative longer lower limbs have lower locomotor costs24. Effective sprinting requires greater stride length25 and powerful lumbar muscles, specifically erector spinae and quadratus lumborum26. Endurance run- ning, nevertheless, does not require longer strides. Higher frequency is more important to running performance during long distances and time, especially in longer trails, where the loss of stride length typically appeared due to fatigue27,28. Generally, a more upright trunk posture is observed among runners who perform efficiently in comparison with those less efficient, whose trunks were increasingly flexed during endurance running29. Besides a positive effect of overall trunk muscularity26,30 on running performance, it has been shown that several other specific trunk morphological aspects relate to running performance, including the width of the pelvis8,31, the trunk flexion angle32, lumbar lordosis33,34 and associated hip flexion31, and thorax breathing mechanics35,36. Within modern humans, the relationship between the widths of the thorax and the pelvis are important parameters of human variability in form and function37. The narrower pelvis relative to the wider thorax in males is associated with a gait pattern that differs biomechanically from that of females, who are characterized by a wider pelvis and narrower thorax dimensions38. The width of the pelvis influences the biomechanics of the psoas major affecting its hip rotator and flexor capacities31. Trunk flexion also affects significantly stride kinematics and kinetics. Although the factors of trunk flexion are unclear, higher trunk flexion angle correlates with shorter stride length, higher stride frequency, greater reaction forces and increased locomotion costs32. Lumbar lordosis varies considerably in human populations12,39–41 and affects locomotor capacities. It has been shown that greater lordosis facilitates shock absorption, for example, when running34, while weaker lor- dosis produces a more forwards orientation of the pelvis, which is beneficial for leg return during sprinting31. Weaker lordosis is also related to greater trunk muscle strength33. Overall trunk muscularity (e.g. erector spinae, quadratus lumborum, psoas major, transverse abdominal, etc.…) has been correlated positively with sprinting capacities26,30. Differences in the tonus of the erector spinae and quadratus lumborum muscles have been related to greater lumbar lordosis42. The contribution of thorax shape to trunk morphology is further interesting in the context of respiratory biomechanics35,36. It has been suggested that morphological features of the rib joints are relevant for ventilatory capacity during running43. These authors showed that H. erectus has similar rib joint morphologies as modern humans that differed from Australopithecus and chimpanzees. But also overall thorax morphology is important: antero-posteriorly flatter ribcages with more inferiorly declined ribs were suggested to show different thoraco- diaphragmatic and abdominal muscle recruitment during ventilatory movement than antero-posteriorly deeper thoraces with more horizontally aligned ribs44–46. Although pump- and bucket-handle patterns of rib motion seem more uniformly distributed along the ribs than originally assumed47,48, variation in thorax-shape related breathing biomechanics indirectly affect the locomotor capacities due to energetic competition and demands between the locomotor and the respiratory systems49,50. Thus, several studies have so far addressed the implica- tion of specific elements of trunk morphology in isolation on locomotor performance. This study explores the relationship between entire 3D trunk shape and running performance based on virtual and geometric morpho- metric methods51,52. In the light of the functional anatomical evidence reviewed above, we address the hypothesis that trunks with an antero-posteriorly flat ribcage, a medio-laterally narrow pelvis and a lower lumbar lordosis are associated to a better running performance. Materials and methods Functional analyses, variables and experimental set up, ethics. Twenty-seven healthy trained young male students of the Degree in Sciences of Physical Activity and Sports (Table  1) were voluntarily recruited. Twelve of them were trained in endurance (ER) disciplines and fifteen were team sport players (non- ER). The inclusion criteria were the following: (1) Age between 18 and 30 years; (2) volunteers athletes had to be either long distance runners or team sports players (e.g., rugby, soccer, basketball); (3) not having suffered a musculoskeletal injury one month prior to the date of the protocol (i.e., checked through a previous exclusion questionnaire). And exclusion criteria were: (1) Age younger than 18 years; (2) having consumed any narcotic and/or psychotropic agents or drugs during the test; (3) any cardiovascular, metabolic, neurologic, pulmonary, or orthopaedic disorder that could limit performance in the different tests. Informed consent was obtained by all volunteers. The study protocol adhered to the declaration of Helsinki and was approved by the Ethics Committee of the Technical University of Madrid (Spain). All the participants performed a physiological (ramp) protocol on a treadmill (Telju JT4100-Liton -035, Toledo, Spain) in three different phases of exercise intensities: 45–50%, 70%, and 85% of the heart rate reserve (HRR). These three intensities correspond with the cardiorespiratory phase 1 [i.e., Light intensity, below the 3 Vol.:(0123456789) Scientific Reports | (2022) 12:11762 | https://doi.org/10.1038/s41598-022-15344-x www.nature.com/scientificreports/ ventilatory threshold (VT)], phase 2 [i.e., Moderate intensity, between the VT and the respiratory compensation threshold (RCT)], and phase 3 [i.e., High intensity, above the RCT)]27. The rate of perceived exertion (RPE) was introduced as a complement of the HRR to help the control of the adequate intensity in each of the three submaxi- mal workloads (i.e., For a HRR of 45–50% the RPE should be 2–4/10, HRR of 70% the RPE 5–6/10, and for HRR 85% the RPE ≥ 8/10). Before warm-up, rest heart rate was measured in sitting position until it was stable. After a general warm-up, the test started between 6 and 7 km  h−1 and 1% of slope to mimic effects of air resistance53,54. Then, running velocity was increased by 0.5 km  h−1 every 30 s until the achievement of HRR ≥ 85%, RPE ≥ 8 and volitional exhaustion. The following variables were recorded at these instances: time, running velocity, and RPE. Heart rate (beats·min−1) was continuously monitored during the test using a telemeter (Polar Ceinture H10+; Polar Electro OY, Kempele, Finland). Changes in velocity during the different work load phases were analysed by repeated measures ANOVA carried out in PAST55. Anthropometrical and running performance data were collected and summarized in Table 1 and Table 2. 3D shape data collection and geometric morphometric analyses. 3D body surface data were man- ually recorded by an Artec MHT 3D (www. artec 3d. com) surface scanner in standardized positions, standing upright on a turning table, with quiet breathing and the arms slightly raised over the head to leave the 360° of Table 1. Descriptives of the sample showing age, body size, and weight. Age (yr) Stature (m) Body weight (kg) BMI N 27 27 27 27 Min 18 1.62 53.50 19.97 Max 29 1.90 83.00 25.76 Mean 20.78 1.77 69.06 22.06 SD 2.53 0.07 7.20 1.34 Figure 1. Frontal, lateral and posterior views of the 3D landmarks on the trunk surface. Red dots are fixed landmarks (Supplementary Table 1) and anatomically homologous between subjects, blue dots are curve semilandmarks, and green dots are surface semilandmarks. After resliding the semilandmarks are mathematically homologous among subjects. Table 2. Descriptive statistics of the running velocities at different experimental steps. Vinitial (km/h) V1 (km/h) V2 (km/h) V3 (km/h) Vmax (km/h) N 27 26 26 27 27 Min 6 6 8 12 12 Max 8 9 14 20 20 Mean 6.85 7.35 10.33 14.42 15.07 SD 0.43 0.75 1.35 2.09 1.81 4 Vol:.(1234567890) Scientific Reports | (2022) 12:11762 | https://doi.org/10.1038/s41598-022-15344-x www.nature.com/scientificreports/ the trunk contour free for image caption. 160 landmarks and semilandmarks (Fig. 1) were digitized according to the template described in González-Ruiz et al.52 (Supplementary Table 1), and postprocessed following standard methods56. Trunk landmark data were then analysed and visualized following standard methods of virtual, geo- metric morphometric analyses51. Specifically, generalized procrustes analysis (GPA) was carried out to obtain 3D shape data and multivariate regression analyses were carried out on the 3D shape data on running veloc- ity. In order to account for different influence of muscularity on torso shape in endurance and non-endurance athletes, we performed a pooled-within group regression. We also tested the hypotheses with a reduced torso landmark set (N = 142 lms), where those landmarks that covered the skin surface related to the latissimus dorsi and pectoralis major muscles were removed. Finally, we explored the data for a possible impact of stature and weight on running performance using GLM. We set the significance level for the regression analyses on p < 0.05. The analyses were carried out using MorphoJ software57, PAST v3.2555, STATISTICA v.858, geomorph package for R59,60 and Evan toolkit61 following the workflows outlined in Bastir et al.51 (Table 2). Results Repeated-measures ANOVA (Table 3) shows that mean velocity increased significantly during incremental HRR phases (Fig. 2). The regression analyses of torso shape on velocity phases indicated no significant relations during the first two stages (V1, V2) of workload. However, statistically significant relations were found between torso shape and velocity during phase 3 (V3) and maximum velocity (Vmax) (Table 4). Comparison of slopes in the ER and non-ER groups in the pooled within group regression models revealed no evidence for differences in the full torso shape data (P = 0.19; F = 1.833), nor in the non-muscular torso shape data (P = 0.18; F = 0.187) in relation Vmax. The GLM model revealed a significant influence of both, full and non-muscular torso shapes on running performance but no such effect of stature or weight (Table 5). The associated 3D shapes (Fig. 2) show that the following morphological features of the trunk are positively associated with increased running performance: smaller antero-posterior diameter at the central-lower rib cage (flat thorax), narrower lower trunk (narrow pelvis), taller trunk, reduced thoracic kyphosis and more pronounced lumbar lordosis. Discussion Modern humans are characterized by a relatively flat and narrow ribcage and pelvis when compared to fossil representatives of the genus Homo that are characterised by more stocky, wider and antero-posteriorly deeper torso configurations2–6,15,62,63. While more and more evidence seems to document this morphological trend, possible functional implications of reduced widths and depths of the trunk remain poorly understood. Because the trunk comprises elements of the respiratory and locomotor systems, the interaction of trunk shape with respiratory and locomotor performance is of specific interest. In the present study, we address possible relations between torso shape and locomotor function in an experi- mental setting relating 3D external trunk surface shape with running velocity at different levels of intensity. The results showed no relationship between trunk shape and running performance at lower levels of exercise (V1, V2) below the anaerobic (respiratory) threshold, and just above it, indicating no relations between external torso shape and endurance running speeds between 7 and 10 km/h. However, at higher intensities and velocities above the anaerobic (respiratory) threshold (V3; average 14.4 km/h) a statistical relation between torso shape and run- ning speed emerged. According to our results, subjects with a flatter and slightly narrower thorax, lower thoracic kyphosis, more pronounced lumbar lordosis, and slightly narrower pelvis can achieve such higher velocities such as indicated by the higher variances of 3D trunk shape shown at V3 and maximum velocity. It has been suggested that an endurance running velocity of about (5  ms−1 = 18 km/h) can be sustained by many amateurs without special training18, which is considerably faster than in our sample. At moderate intensity (V2), presumably within the aerobic metabolic domain, the average speed was about 10 km/h (Table 2). This may be related to the slight inclination of the treadmill (1%) during the incremental experiment (and the thereby simulated air resistance), but it could also reflect the fact that not all the volunteers were specialized endurance runners. Likewise, the average speed of 14 km/h at V3, which is likely already beyond the anaerobic threshold, is still lower than the published one and, again, could be related to the factors mentioned before. However, at and beyond this velocity, 3D torso shape was statistically related to running capacity. The most visible features related to higher running capacities were a low degree of thoracic kyphosis, with a flatter, slightly narrower central thorax and a greater degree of lumbar spine curvature with a relatively slightly narrower pelvis. Covariation in depths was more clearly recognisable than in widths (Fig. 2). While the thoracic part suggests interpretation within a respiratory biomechanical perspective, the lumbo-pelvic part of the torso Table 3. ANOVA of velocities during the three different phases (V1, V2, V3). Sum of sqrs df Mean square F p (same) Between groups 1526.05 4 381.51 393.5 < 0.001 Within groups 251.161 125 2.01 Error 96.954 100 0.96 Between subjects 154.207 25 6.17 Total 1777.21 129 5 Vol.:(0123456789) Scientific Reports | (2022) 12:11762 | https://doi.org/10.1038/s41598-022-15344-x www.nature.com/scientificreports/ also requires consideration within functions of the locomotor system, although both are clearly related with each other. For example, the role of the posterior lumbar muscles is essential, as they act keeping an upright posture of the lower trunk during running and giving stability to the diaphragm and psoas major lumbar insertions. So, trunk extensors have the ability to reduce the kyphosis angle64,65. Links between breathing biomechanics and lumbar stability have been found in Kang et al.66 who showed that spinal posture was improved by specific breathing exercises in a clinical context. The combination of a reduced thoracic kyphosis and a flat ribcage, with anteriorly declined ribs, in which the anterior rib ends are more caudally located than the posterior rib ends, could point to the importance of ventilatory biomechanics in higher intensity running. Bellemare et al.44,45 suggested that declined ribs can be elevated more during inspiration than horizontally aligned ones accentuating potentially the costal contribu- tion to thorax movement during lung ventilation. Also, anteriorly declined ribs may have better biomechanical leverage during forced expiration, which crucially increases the tidal volume during heavy exercise breathing35. Because the declination of the ribs is morphologically related to a flatter rib cage configuration, the hypothesis that a flat thorax is positively related to running performance finds support. Physiologically, a less curved tho- racic spine increases further the vertical space potentially available for lung expansion through enhancing of rib mobility. For example, negative consequences for lung ventilation due to kyphotic thoracic spine deformations, which compress thoracic space and affect rib biomechanics, have been reported46,67. The implication of lumbar lordosis for locomotor biomechanics consists of its effect on the forwards orienta- tion of the anterior superior iliac spine, which is an advantageous position for efficient leg return31. However, while these authors have not found a significant relation between lumbar lordosis angle and hip flexion capacity, Figure 2. Torso shapes (160 lms) and thin-plate splines warped to the highest and lowest velocity at maximum intensity and running speed. (a) Non-muscular torso shape (142 lms) on maximum velocity. (b) Full torso shape (160 lms) on maximum velocity (c) Full torso shape warped to the configuration of lowest (left) and highest (right) maximum velocities. Upper panel left lateral view, lower panel frontal view. Note that flatter ribcages, narrower trunks with low thoracic kyphosis and more pronounced lumbar lordosis correlate significantly with higher velocities at maximum intensity. (Magnification factor from left to right: − 7.5; − 5; + 5, + 7.5, for better visualization). 6 Vol:.(1234567890) Scientific Reports | (2022) 12:11762 | https://doi.org/10.1038/s41598-022-15344-x www.nature.com/scientificreports/ our results in Fig. 2 clearly show that more pronounced lumbar curvature, to which also the lower thoracic kyphosis contributes, produces forwards tilt of the pelvis. Warrener et al.32 have found a significant reduction of length and an increment of frequency of strides associ- ated with higher trunk flexion posture during running. This finding is supported by Castillo and Liebermann34, who pointed out that higher lumbar lordosis (trunk extension) is linked with longer stride length in runners, a key factor in speed running as we have observed in our sample. Additionally, upright posture have been associ- ated with better economy and running performance in the context mechanically interactions between trunk kinetics, reaction forces and spatiotemporal patterns of strides29. Table 4. Multivariate regressions of full torso shape (160lms) and non-muscular torso shape (142 lms) on running performance at different workloads (V1, V2, V3 and Vmax). (Note that sample size is N = 27 for Vmax, but N = 26 for V1, V2 and V3). Significant values are in bold. Df SS MS R2 F Z p Value 160 lms V1 1 0.003868 0.003868 0.04589 1.1543 0.54784 0.3 Residuals 24 0.080422 0.003351 0.95411 Total 25 0.08429 V2 1 0.00358 0.00358 0.04248 1.0646 0.30142 0.387 Residuals 24 0.08071 0.003363 0.95752 Total 25 0.08429 V3 1 0.005871 0.005871 0.06965 1.7968 1.8401 0.034 Residuals 24 0.078419 0.003267 0.93035 Total 25 0.08429 Vmax 1 0.007102 0.007102 0.08071 2.1948 2.4053 0.009 Residuals 25 0.08089 0.003236 0.91929 Total 26 0.087991 142 lms V1 1 0.003644 0.003645 0.04369 1.0965 0.39524 0.359 Residuals 24 0.079767 0.003324 0.95631 Total 25 0.083412 V2 1 0.003291 0.003291 0.03945 0.9857 0.082949 0.458 Residuals 24 0.080121 0.003338 0.96055 Total 25 0.083412 V3 1 0.005547 0.005547 0.0665 1.7096 1.728 0.044 Residuals 24 0.077865 0.003244 0.9335 Total 25 0.083412 Vmax 1 0.006828 0.006828 0.07836 2.1256 2.2905 0.01 Residuals 25 0.080303 0.003212 0.92164 Total 26 0.08713 Table 5. Generalized Linear Models assessing the effects of stature, weight, torso shape (160 lms, 142 lms) on running performance. Significant values are in bold. SS df MS F p 160 lms Intercept 7.93 1 7.93 5.38 0.030 Stature 0.51 1 0.51 0.35 0.562 Weight 0.88 1 0.88 0.60 0.447 Torso shape 46.85 1 46.85 31.79 0.000 Error 33.90 23 1.47 142 lms Intercept 6.43 1.00 6.43 4.33 0.048 Stature 0.19 1.00 0.19 0.13 0.720 Weight 46.59 1.00 46.59 31.36 0.519 Torso shape 0.64 1.00 0.64 0.43 0.000 Error 34.16 23.00 1.49 7 Vol.:(0123456789) Scientific Reports | (2022) 12:11762 | https://doi.org/10.1038/s41598-022-15344-x www.nature.com/scientificreports/ Therefore, the empirical evidence reported in the present study seems to indicate that trunk evolution as a whole may have brought about the appearance of some features that are more clearly related to long distance running, along with others that are more related to power locomotion with higher workloads. However, these features lead to a mosaic notion, which reflects a complex picture of potential adaptations to running economy. In Neandertals, some adaptations to power locomotion were proposed on anatomical, genetic, and ecological grounds19,20. Our results suggest that the relatively straight thoracic column along with their high level of trunk muscularity, possibly reflected by wide, deep thorax shape and associated high body mass estimates, would fit with the power locomotion hypothesis2,68,69. On the other hand, their supposed hypo-lordosis would argue against such interpretation as the relatively uncurved reconstruction of the thoracic and lumbar spine in the Kebara 2 Neandertal13,69 would indicate reduced pelvic tilt and thus a reduced capacity of leg return, hip flexion and sprinting capacity. Yet, the most recent reconstruction of the La Chapelle aux Saints Neandertal suggests vertebral curvatures similar to modern humans14 and this indicates that a better fossil documentation of lumbar spine anatomy in Neandertals is needed. Importantly, a recent study accounting for a wide range of population variability in modern humans, identified consistently and significantly more pronounced lordotic wedging in Neandertal L5 of Kebara 2, Shanidar 3, and La Chapelle aux Saints41 together with a more hypo-lordotic wedg- ing in upper lumbar vertebra. Accordingly, this could suggest a completely different position of the lumbar spine within the trunk, with yet unclear biomechanical implications. Therefore, further fossil reconstructions of Neandertal torso skeletons together with experimental testing are necessary. In African H. erectus, as reconstructed on the remains of KNM-WT 15,000, the straight thoracic3 and curved lumbar spine morphology70 would be more in line with effective power-locomotion. This, together with greater torso width and depth would be also compatible with higher muscularity and body mass3,15,63,71,72. However, clearly, the elongated limbs favour an interpretation of long-distance locomotion and, possibly, running17,21. Altogether, the present evidence and reviews suggest that our interpretations relate to a great extent on the reli- ability of the fossil body reconstructions. However, it is important to bear in mind the limitations of our experimental evidence in the evolution- ary context of endurance running. Obviously, the fossil record does not contain information about soft tissue anatomy, while the present data was exclusively collected on the external surface of the torso and so the relations between skeletal and soft tissue anatomy are unknown. Yet, bony features are considered. The curvature of the spine is assessed by the tips of the spinous processes which are variable in terms of sagittal orientations and thus do not directly inform about the curvature as assessable on the basis of the vertebral bodies. Also, the ribcage anatomy is only indirectly reflected by the skin surface landmarks and closer to skeletal thorax shape only at the central and lower parts of the rib cage. These data can thus only give a general idea about thorax shape. The pelvic landmarks are clearer in this respect as the iliac spines can be identified without problems. However, the reduced landmark set, which excluded shape information related to the latissimus dorsi and major pectoralis muscles may be less influenced by muscularity, and the fact that the results of the full and the reduced data are similar suggests little soft tissue effects on the results. Further limitations are related to the proper running experiment. Endurance running in the evolutionary context appeared in the context of specific climatic conditions that were not considered in the present experi- ment. Also, actual endurance running is defined as running at intermediate velocities and aerobic conditions for longer time than considered in our experiment, where we only tested for potential relations between veloc- ity and aerobic running conditions during the early stages of the incremental exercise. In this perspective, our data are only informative about shape-function relation during higher intensity running. Future studies should relate torso shape to running performance data on velocity and distance during longer trails and in hot weather conditions. Such analysis will provide further insight into the important relationships between torso shape, body shape and locomotor performance relevant for human evolution. Received: 22 December 2021; Accepted: 22 June 2022 References 1. Franciscus, R. G. & Churchill, S. E. The costal skeleton of Shanidar 3 and a reappraisal of Neandertal thoracic morphology. J. Hum. Evol. 42, 303–356 (2002). 2. Gómez-Olivencia, A. et al. 3D virtual reconstruction of the Kebara 2 Neandertal thorax. Nat. Commun. 9, 4387 (2018). 3. Bastir, M. et al. Rib cage anatomy in Homo erectus suggests a recent evolutionary origin of modern human body shape. Nat. Ecol. Evol. 4, 1178–1187 (2020). 4. García-Martínez, D. et al. Early development of the Neanderthal ribcage reveals a different body shape at birth compared to modern humans. Sci. Adv. 6, eabb4377 (2020). 5. Arsuaga, J. L. et al. A complete human pelvis from the middle pleistocene of Spain. Nature 399, 255–258 (1999). 6. Simpson, S. W. et al. A female Homo erectus pelvis from Gona, Ethiopia. Science 322, 1089–1092 (2008). 7. Torres-Tamayo, N. et al. Three-dimensional geometric morphometrics of thorax-pelvis covariation and its potential for predicting the thorax morphology: A case study on Kebara 2 Neandertal. J. Hum. Evol. 147, 102854 (2020). 8. Stansfield, E., Fischer, B., Grunstra, N. D. S., Pouca, M. V. & Mitteroecker, P. The evolution of pelvic canal shape and rotational birth in humans. BMC Biol. 19, 1–11 (2021). 9. Meyer, M. R. & Williams, S. A. The Spine of Early Pleistocene Homo. In The Human Spine (eds Been, E. et al.) 153–1845 (Springer, 2019). 10. Gómez-Olivencia, A., Couture-Veschambre, C., Madelaine, S. & Maureille, B. The vertebral column of the Regourdou 1 Neandertal. J. Hum. Evol. 64, 582–607 (2013). 11. Bastir, M. et al. Three-dimensional morphometrics of thoracic vertebrae in Neandertals and the fossil evidence from El Sidrón (Asturias, Northern Spain). J. Hum. Evol. 108, 47–61 (2017). 12. Been, E., Gómez-Olivencia, A. & Kramer, P. A. Lumbar lordosis of extinct hominins. Am. J. Phys. Anthropol. 147, 64–77 (2012). 8 Vol:.(1234567890) Scientific Reports | (2022) 12:11762 | https://doi.org/10.1038/s41598-022-15344-x www.nature.com/scientificreports/ 13. Been, E. et al. Evolution of spinopelvic alignment in hominins. Anat. Rec. 300, 900–911 (2017). 14. Haeusler, M. et al. Morphology, pathology, and the vertebral posture of the La Chapelle-aux-Saints Neandertal. Proc. Natl. Acad. Sci. USA 116, 4923–4927 (2019). 15. Ruff, C. Body size and body shape in early hominins—implications of the Gona Pelvis. J. Hum. Evol. 58, 166–178 (2010). 16. Aiello, A. & Wheeler, P. The expensive tissue hypothesis: The brain and digestive system in human and primate evolution. Curr. Anthropol. 36, 199–221 (1995). 17. Bramble, D. M. & Lieberman, D. E. Endurance running and the evolution of Homo. Nature 432, 345–352 (2004). 18. Lieberman, D. E., Bramble, D. M., Raichlen, D. A. & Shea, J. J. Brains, brawn and the volution of human endurance running capabilities. in The first humans: origin and early evolution of the genus Homo. 77–92 (Springer, 2009). 19. Raichlen, D. A., Armstrong, H. & Lieberman, D. E. Calcaneus length determines running economy: Implications for endurance running performance in modern humans and Neandertals. J. Hum. Evol. 60, 299–308 (2011). 20. Stewart, J. R. et al. Palaeoecological and genetic evidence for Neanderthal power locomotion as an adaptation to a woodland environment. Quat. Sci. Rev. 217, 310–315 (2019). 21. Lordkipanidze, D. et al. Postcranial evidence from early Homo from Dmanisi, Georgia. Nature 449, 305–310 (2007). 22. Pontzer, H. Economy and endurance in human evolution. Curr. Biol. 27, R613–R621 (2017). 23. Weaver, T. D. The meaning of Neandertal skeletal morphology. Proc. Natl. Acad. Sci. USA 106, 16028–16033 (2009). 24. Steudel-Numbers, K. L., Weaver, T. D. & Wall-Scheffler, C. M. The evolution of human running: Effects of changes in lower-limb length on locomotor economy. J. Hum. Evol. 53, 191–196 (2007). 25. Mattes, K., Schaffert, N., Habermann, N. & Mühlbach, T. A longitudinal study of kinematic stride characteristics in maximal sprint running. J. Hum. Sport 9, 686–699 (2014). 26. Kubo, T. Contribution of trunk muscularity on sprint run. Int. J. Sports Med. 32, 223–228 (2011). 27. Esteve-Lanao, J., Sanjuan, A. F., Earnest, C., Foster, C. & Lucia, A. How do endurance runners actually train? Relationship with competition performance. Med. Sci. Sports Exerc. 37, 496–504 (2005). 28. Esteve-Lanao, J., Rhea, M. R., Fleck, S. J. & Lucia, A. Running-specific, periodized strength training attenuates loss of stride length during intense endurance running. J. Strength Cond. Res. 22, 1176–1183 (2008). 29. van Oeveren, B. T., de Ruiter, C. J., Beek, P. J. & van Dieën, J. H. The biomechanics of running and running styles: a synthesis. Sports Biomech. 1–39 (2021). 30. Fujita, S. A 100-m sprint time is associated with deep trunk muscle thickness in collegiate male sprinters. Front. Sports Act. Living 1, 32 (2019). 31. Copaver, K., Hertogh, C. & Hue, O. The effects of psoas major and lumbar lordosis on hip flexion and sprint performance. Res. Q. Exerc. Sport 83, 160–167 (2012). 32. Warrener, A., Tamai, R. & Lieberman, D. E. The effect of trunk flexion angle on lower limb mechanics during running. Hum. Mov. Sci. 78, 102817 (2021). 33. Castillo, E. R., Hsu, C., Mair, R. W. & Lieberman, D. E. Testing biomechanical models of human lumbar lordosis variability. Am. J. Phys. Anthropol. 163, 110–121 (2017). 34. Castillo, E. R. & Lieberman, D. E. Shock attenuation in the human lumbar spine during walking and running. J. Exp. Biol. 221, jeb177949 (2018). 35. Aliverti, A. et al. Human respiratory muscle actions and control during exercise. J Appl. Physiol. 83, 1256–1269 (1997). 36. Kenyon, C. M. et al. Rib cage mechanics during quiet breathing and exercise in humans. J. Appl. Physiol. 83, 1242–1255 (1997). 37. Torres-Tamayo, N. The torso integration hypothesis revisited in Homo sapiens: Contributions to the understanding of hominin body shape evolution. Am. J. Phys. Anthropol. 167, 777–790 (2018). 38. Waldock, C., Milne, N., Rubenson, J. & Donnelly, C. J. The use of geometric morphometric techniques to identify sexual dimor- phism in Gait. J. Appl. Biomech. 32, 441–448 (2016). 39. Vialle, R. et al. Radiographic analysis of the sagittal alignment and balance of the spine in asymptomatic subjects. J. Bone Jt. Surg. 87, 260–267 (2005). 40. Lois Zlolniski, S. et al. 3D geometric morphometric analysis of variation in the human lumbar spine. Am. J. Phys. Anthropol. 170, 361–372 (2019). 41. García-Martínez, D. et al. Sexual dimorphism in the vertebral wedging of the human lumbar vertebrae and its importance as a comparative framework for understanding the wedging pattern of Neanderthals. Quat. Int. 566–567, 224–232 (2020). 42. Hsu, C., Castillo, E. & Lieberman, D. The relationship between trunk muscle strength and flexibility, intervertebral disc wedging, and human lumbar lordosis. Harvard Undergrad. Res. J. 8, 35–41 (2015). 43. Callison, W. É., Holowka, N. B. & Lieberman, D. E. Thoracic adaptations for ventilation during locomotion in humans and other mammals. J. Exp. Biol. 222, jeb189357 (2019). 44. Bellemare, F., Jeanneret, A. & Couture, J. Sex differences in thoracic dimensions and configuration. Am. J. Respir. Crit. Care Med. 168, 305–312 (2003). 45. Bellemare, F., Fuamba, T. & Bourgeault, A. Sexual dimorphism of human ribs. Respir. Physiol. Neurobiol. 150, 233–239 (2006). 46. Sanchis-Gimeno, J. A. et al. Association between ribs shape and pulmonary function in patients with Osteogenesis Imperfecta. J. Adv. Res. 21, 177–185 (2020). 47. Beyer, B. et al. In vivo thorax 3D modelling from costovertebral joint complex kinematics. Clin. Biomech. 29, 434–438 (2014). 48. Beyer, B., Van Sint Jan, S., Chèze, L., Sholukha, V. & Feipel, V. Relationship between costovertebral joint kinematics and lung volume in supine humans. Respir. Physiol. Neurobiol. 232, 57–65 (2016). 49. Dominelli, P. B. et al. Effects of respiratory muscle work on respiratory and locomotor blood flow during exercise. Exp. Physiol. 102, 1535–1547 (2017). 50. Horiuchi, M., Kirihara, Y., Fukuoka, Y. & Pontzer, H. Sex differences in respiratory and circulatory cost during hypoxic walking: potential impact on oxygen saturation. Sci. Rep. 9, 9550 (2019). 51. Bastir, M. et al. Workflows in a Virtual Morphology Lab: 3D scanning, measuring, and printing. J. Anthropol. Sci. 97, 1–28 (2019). 52. González-Ruiz, J. M., Pérez-Núñez, M. I., García-Alfaro, M. D. & Bastir, M. Geometric morphometrics of adolescent idiopathic scoliosis: a prospective observational study. Eur. Spine J. 30, 612–619 (2021). 53. Pugh, L. G. Oxygen intake in track and treadmill running with observations on the effect of air resistance. J. Physiol. 207, 823–835 (1970). 54. Lucia, A. et al. Physiological characteristics of the best Eritrean runners-exceptional running economy. Appl. Physiol. Nutr. Metab. 31, 530–540 (2006). 55. Hammer, Ø. PAST: Palaeontological Statistics, version 3.25. (2019) 56. Gunz, P. & Mitteroecker, P. Semilandmarks: A method for quantifying curves and surfaces. Hystrix 24, 103–109 (2013). 57. Klingenberg, C. P. MorphoJ: An integrated software package for geometric morphometrics. Mol. Ecol. Resour. 11, 353–357 (2011). 58. StatSoft. Statistica 8.0. StatSoft, Inc., Tulsa, OK, USA. (1984–2007). 59. Adams, D. C. & Otárola-Castillo, E. Geomorph: An R package for the collection and analysis of geometric morphometric shape data. Methods Ecol. Evol. 4, 393–399 (2013). 60. R. Core Team. R: A language and environment for statistical computing (2017). 61. EVAN-Society. ET, Toolkit for geometric morphometric analysis (2010). 9 Vol.:(0123456789) Scientific Reports | (2022) 12:11762 | https://doi.org/10.1038/s41598-022-15344-x www.nature.com/scientificreports/ 62. Rosenberg, K. R., Zuné, L. & Ruff, C. B. Body size, body proportions, and encephalization in a Middle Pleistocene archaic human from northern China. Proc. Natl. Acad. Sci. USA 103, 3552–3556 (2006). 63. Ruff, C. B., Burgess, M. L., Squyres, N., Junno, J.-A. & Trinkaus, E. Lower limb articular scaling and body mass estimation in Pliocene and Pleistocene hominins. J. Hum. Evol. 115, 85–111 (2018). 64. Ball, J. M., Cagle, P., Johnson, B. E., Lucasey, C. & Lukert, B. P. Spinal extension exercises prevent natural progression of kyphosis. Osteoporos. Int. 20, 481–489 (2009). 65. Feng, Q., Wang, M., Zhang, Y. & Zhou, Y. The effect of a corrective functional exercise program on postural thoracic kyphosis in teenagers: A randomized controlled trial. Clin. Rehabil. 32, 48–56 (2018). 66. Kang, J.-Y., Seo, D.-K., Cho, J.-C. & Lee, B.-K. Effectiveness of breathing exercises on spinal posture, mobility and stabilization in patients with lumbar instability. J. Korean Soc. Phys. Med. 13, 81–89 (2018). 67. LoMauro, A. et al. Rib cage deformities alter respiratory muscle action and chest wall function in patients with severe osteogenesis imperfecta. PLoS ONE 7, e35965 (2012). 68. Ruff, C. B., Trinkaus, E. & Holliday, T. W. Body mass and encephalization in Pleistocene Homo. Nature 387, 173–176 (1997). 69. Been, E., Gómez-Olivencia, A., Kramer, P. A. & Barash, A. 3D Reconstruction of the spinal posture in the Kebara 2 Neanderthal. In Human Paleontology and Prehistory (eds Marom, A. & Hovers, E.) 239–251 (Springer, 2017). 70. Latimer, B. & Ward, C.V. The thoracic and lumbar vertebrae, in The Nariokotome Homo erectus Skeleton. 266–293 (Harvard University Press, 1993). 71. Ruff, C. B. & Burgess, M. L. How much more would KNM-WT 15000 have grown?. J. Hum. Evol. 80, 74–82 (2015). 72. Torres-Tamayo, N. et al. New reconstruction of the pelvis of KNM-WT 15000 supports a wide body shape for Early African H. erectus. Proc. Eur. Soc. Stud. Hum. Evol. 120 (2021). Acknowledgements We thank the volunteers for participating in this study. Funding: Grant PID2020-115854GB-I00 to MB is funded by MCIN/AEI/10.13039/501100011033 of the Spanish Ministry of Science and Innovation and the European Union. We thank Prof. Mitteroecker and one anonymous reviewer for their helpful comments. Author contributions M.B. designed the associated research project, collected 3D surface data, analysed the geometric morphometrics data, wrote the main manuscript and pepared the Figures. J.M.G.R. collected and postprocessed the 3D-landmark data, preprared Figures and wrote parts of the paper. M.G.R programmed code and analysed additional data. M.B., B.B., A.S.J. and E.N worked on the development of the experimental procedures. M.B., J.R., G.G.L., A.S.J. and E.N. carried out the experiments and A.S.J and E.N supervised all steps of the experiment at the laboratory. All authors reviewed the manuscript. Competing interests The authors declare no competing interests. Additional information Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1038/ s41598- 022- 15344-x. Correspondence and requests for materials should be addressed to M.B. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2022

Variation in human 3D trunk shape and its functional implications in hominin evolution.

07-11-2022

Bastir, Markus,González Ruíz, José María,Rueda, Javier,Garrido López, Gonzalo,Gómez-Recio, Marta,Beyer, Benoit,San Juan, Alejandro F,Navarro, Enrique

eng

PMC6209185

RESEARCH ARTICLE Examination of gas exchange and blood lactate thresholds in Paralympic athletes during upper-body poling Julia Kathrin BaumgartID1*, Maaike Moes2, Knut Skovereng1, Gertjan EttemaID1, Øyvind Sandbakk1 1 Centre for Elite Sports Research, Department of Neuroscience and Movement Science, Faculty of Medicine and Health Sciences, Norwegian University of Science and Technology, Trondheim, Norway, 2 Department of Human Movement Sciences, Faculty of Health, Medicine and Life Sciences, Maastricht University, Maastricht, The Netherlands * jk.baumgart@gmail.com Abstract Objectives The primary aim was to compare physiological and perceptual outcome parameters identi- fied at common gas exchange and blood lactate (BLa) thresholds in Paralympic athletes while upper-body poling. The secondary aim was to compare the fit of the breakpoint models used to identify thresholds in the gas exchange thresholds data versus continuous linear and curvilinear (no-breakpoint) models. Methods Fifteen elite Para ice hockey players performed seven to eight 5-min stages at increasing workload until exhaustion during upper-body poling. Two regression lines were fitted to the oxygen uptake (VO2)-carbon dioxide (VCO2) and minute ventilation (VE)/VO2 data to deter- mine the ventilatory threshold (VT), and to the VCO2-VE and VE/VCO2 data to determine the respiratory compensation threshold (RCT). The first lactate threshold (LT1) was deter- mined by the first rise in BLa (+0.4mmol
L-1 and +1.0mmol
L-1) and a breakpoint in the log- log transformed VO2-BLa data, and the second lactate threshold (LT2) by a fixed rise in BLa above 4mmol
L-1 and by employing the modified Dmax method. Paired-samples t-tests were used to compare the outcome parameters within and between the different threshold meth- ods. The fit of the two regression lines (breakpoint model) used to identify thresholds in the gas exchange data was compared to that of a single regression line, an exponential and a 3rd order polynomial curve (no-breakpoint models) by Akaike weights. Results All outcome parameters identified with the VT (i.e., breakpoints in the VO2-VCO2 or VE/VO2 data) were significantly higher than the ones identified with a fixed rise in BLa (+0.4 or +1.0mmol
L-1) at the LT1 (e.g. BLa: 5.1±2.2 or 4.9±1.8 vs 1.9±0.6 or 2.3±0.5mmol
L-1, p<0.001), but were not significantly different from the log-log transformed VO2-BLa data PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 1 / 18 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Baumgart JK, Moes M, Skovereng K, Ettema G, Sandbakk Ø (2018) Examination of gas exchange and blood lactate thresholds in Paralympic athletes during upper-body poling. PLoS ONE 13(10): e0205588. https://doi.org/ 10.1371/journal.pone.0205588 Editor: Tiago M Barbosa, Nanyang Technological University, SINGAPORE Received: February 19, 2018 Accepted: September 17, 2018 Published: October 31, 2018 Copyright: © 2018 Baumgart et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper and its Supporting Information files. Funding: The laboratory equipment was provided by NeXt Move, Norwegian University of Science and Technology (NTNU). NeXt Move is funded by the Faculty of Medicine at NTNU and Central Norway Regional Health Authority. The funder had no role in study design, how the data collection and analysis was performed, decision to publish, or preparation of the manuscript. (4.3±1.6mmol
L-1,p>0.06). The outcome parameters identified with breakpoints in the VCO2-VE data to determine the RCT (e.g. BLa: 5.5±1.4mmol
L-1) were not different from the ones identified with the modified Dmax method at the LT2 (5.5±1.1mmol
L-1) (all p>0.53), but were higher compared to parameters identified with VE/VCO2 method (4.9±1.5mmol
L-1) and a fixed BLa value of 4mmol
L-1 (all p<0.03). Although we were able to determine the VT and RCT via different gas exchange threshold methods with good fit in all 15 participants (mean R2>0.931), the continuous no-breakpoint models had the highest probability (>68%) of being the best models for the VO2-VCO2 and the VCO2-VE data. Conclusions In Paralympic athletes who exercise in the upper-body poling mode, the outcome parame- ters identified at the VT and the ones identified with fixed methods at the LT1 showed large differences, demonstrating that these cannot be used interchangeably to estimate the aero- bic threshold. In addition, the close location of the VT, RCT and LT2 does not allow us to dis- tinguish the aerobic and anaerobic threshold, indicating the presence of only one threshold in athletes with a disability exercising in an upper-body mode. Furthermore, the better fit of continuous no-breakpoint models indicates no presence of clear breakpoints in the gas exchange data for most participants. This makes us question if breakpoints in the gas exchange data really exist in an upper-body exercise mode in athletes with disabilities. Introduction In able-bodied endurance athletes performing lower-body or whole-body exercise, gas exchange and blood lactate (BLa) threshold concepts are well-established in the diagnosis of endurance performance as well as in the prescription of systematic training with different exer- cise intensity zones [1]. Two thresholds are commonly described in the literature: 1) The aero- bic threshold (AT)–determined by the ventilatory threshold (VT) or the first lactate threshold (LT1)–separates low- from moderate-intensity exercise [2, 3]. 2) The anaerobic threshold (ANT)–determined by the respiratory compensation threshold (RCT) or the second lactate threshold (LT2)–separates moderate- from high-intensity exercise [2, 3]. However, to what extent the outcome parameters identified at the VT and LT1 as well as the RCT and LT2 coin- cide in Paralympic sitting sport athletes who exercise in an upper-body mode remains to be investigated. Various methods have been employed to determine the VT and the RCT, as well as the LT1 and the LT2 [3–6]. The VT is based on a disproportionate increase (i.e. a breakpoint) in carbon dioxide production (VCO2) and minute ventilation (VE) in relation to oxygen uptake (VO2) [3, 7], and the LT1 on an onset in BLa concentration above resting levels that marks the begin- ning of exercise [5] or on a breakpoint in the log-log transformed VO2-BLa data [4]. Even though these physiological changes occur above the VT and LT1, the body is still able to main- tain equilibrium at intensities up to the ANT, and aerobic metabolism (indicated by measure- ments of oxygen uptake and the corresponding energy equivalent) reflects overall energy expenditure [2]. The ANT marks the point beyond which any attempt of the body to maintain metabolic equilibrium at a constant rate of work fails [6]. The RCT is based on a dispropor- tionate increase (i.e. a breakpoint) of VE in relation to VCO2 [3], a mechanism that has been suggested to correspond with the point where BLa starts to accumulate with constant workload Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 2 / 18 Competing interests: The authors have declared that no competing interests exist. [6]. In contrast, it has been argued that the changes in gas exchange with increasing work rate are continuous transitions where fatigue gradually accumulates rather than clear breakpoints [8]. The assumption that the VT corresponds with the LT1, and the RCT with the LT2, are based on the initial studies by Beaver et al. [3] and Wassermann et al. [6, 9, 10] from the 1980’s. However, there has been a continuous debate around the existence of and the physio- logical link between these different thresholds [2, 11–14]. Although physiological parameters identified at the VT and LT1, and at the RCT and LT2 have shown high correlations in able- bodied participants during cycling and running in some studies [15, 16], others find low corre- lations [17]. In wheelchair basketball and wheelchair rugby athletes with a spinal cord injury, the % of VO2peak was lower at the LT1 compared to the VT, whereas it did not significantly dif- fer at the LT2 and RCT [18]. In contrast, in able-bodied swimmers, there were no significant differences in physiological outcome parameters at the LT1 and the VT [19]. Whereas a range of studies have investigated the VT during upper-body exercise in able- bodied participants and participants with a disability [20–29], knowledge is limited on whether gas exchange and BLa threshold concepts can be used interchangeably in athletes with disabili- ties who exercise in an upper-body mode, or whether breakpoints exist in the gas exchange data of these athletes. Therefore, the primary aim of this study was to compare physiological and perceptual outcome parameters at the gas exchange and BLa thresholds in the data obtained from Paralympic athletes while upper-body poling. The secondary aim was to com- pare the fit of breakpoint models used to identify gas exchange thresholds with continuous lin- ear or curvilinear (no-breakpoint) models. Methods Participants Fourteen male and one female endurance-trained Norwegian Para ice hockey players partici- pated in this study. Anthropometrics and training hours per month of the participants are depicted in Table 1. All participants were healthy and free of injuries at the time of testing. The study was approved by the Norwegian Data Protection Authority and conducted in accor- dance with the Declaration of Helsinki. All participants signed an informed consent form prior to voluntarily take part in the study, and were made aware that they could withdraw from the study at any point without providing an explanation. Experimental design The testing consisted of two consecutive test days at similar test times, during which partici- pants performed an incremental test to exhaustion on day one, followed by seven to eight 5-min stages at gradually increasing effort for each stage until exhaustion on day two. All tests were performed in upper-body poling on a Concept2 ski ergometer 1 (Concept2, Inc., Morris- ville, USA, http://www.concept2.com/service/skierg/skierg-1), while sitting in an ice sledge hockey seat. Test set-up After being equipped with an oro-nasal mask (Hans Rudolph Inc, Kansas City, MO, USA) and a heart rate monitor (Polar Electro Inc., Port Washington, NY, USA), the participants were tightly strapped around the thighs and hips into an ice sledge hockey seat that was mounted on a wooden platform (Fig 1). The distance of the seat to the Concept2 ski ergometer and the position of the feet depended on personal preference but was the same for test day one and Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 3 / 18 two. The ski ergometer uses wind resistance, which is generated by the spinning flywheel. The ski ergometer has a spiral damper with settings from one to ten, which works like a gearing system. We had this damper set at “eight” for all participants. Power output was measured with the ergometer’s software, which was previously validated with force and velocity measure- ments using a force cell (Noraxon USA inc., Scottsdal, AZ, USA) and the Oqus cameras of the Qualisys motion capture system (Qualisys AB, Gothenburg, Sweden) as described by Hegge et al. [30]. The Metamax II ergospirometer CORTEX Biophysik GmbH, Leipzig, Germany) was calibrated against a known mixture of gases (16% O2 and 4% CO2) and ambient air prior to the testing procedure of every second participant. Before each athlete was tested, the flow transducer was calibrated with a 3 L syringe and then connected to the oro-nasal mask, which allowed for the measurement of breath-by-breath respiratory parameters. Test protocol The participants were instructed to refrain from heavy training and alcohol consumption 24 hours before, caffeine intake the day of, and food intake two hours before testing. Additionally, the participants were instructed to void their bladder directly before arriving at the laboratory. A questionnaire was filled out on each of the two test days to monitor if the participants fol- lowed these instructions, as well as to exclude any prior illness or injury that might have inter- fered with the testing. Test day one. A standardized warm-up of five 5-min submaximal stages with a 2- to 3-min break between stages was performed in the upper-body poling mode at an overall rating of perceived exertion (RPE) of 7 (very light), 9 (very light), 11 (light), 13 (somewhat hard) and 15 (hard). Next to serving as a warm-up, the submaximal stages were used to familiarize the participants with the use of the Borg scale [31] to indicate RPE after the incremental test and Table 1. Sex, age, anthropometric and disability characteristics as well as monthly training hours of the 15 Norwegian national team Para ice hockey players partici- pating in this study. Sex Age (years) Body mass (kg) Height (cm) Disability (level of injury) Training hrs/month 1 Male 53 83.3 186 Paraplegia (Th12-L1) 25 2 Male 18 75.7 160 Spina bifida (L5) 49 3 Male 27 61.0 160 Athrogryposis multiplex congenita 63 4 Male 31 69.4 184 Hereditary spastic paraplegia 45 5 Male 28 90.0 173 Paraplegia (Th10) 26 6 Male 21 70.4 164 Spina bifida (ns) 59 7 Male 33 70.5 160 Spina bifida (Th12) 67 8 Male 34 75.3 173 Paraplegia (Th11-12) 48 9 Female 22 70.0 167 Spina bifida (L3-S1) 33 10 Male 22 63.4 164 Paraplegia (Th11-12) 28 11 Male 18 64.2 154 Spina bifida (ns) 54 12 Male 20 68.0 186 Paraplegia (Th12) 40 13 Male 20 77.0 163 Cerebral Palsy (motor only) 23 14 Male 28 66.5 173 Amputation (single leg above the knee) 80 15 Male 32 63.2 165 Paraplegia (ns) 56 Mean ± SD 27.1±8.9 71.2±8.0 170±10 - 47±18  Players are from the Norwegian national B-team All other players are from the Norwegian national A-team. Thoracic (Th), lumbar (L), sacral (S), not specified (ns) https://doi.org/10.1371/journal.pone.0205588.t001 Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 4 / 18 Fig 1. Test set-up. The participants were strapped in around the hips and thighs in an ice sledge hockey seat mounted on a platform in front of the Concept2 ski-ergometer. https://doi.org/10.1371/journal.pone.0205588.g001 Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 5 / 18 each of the 5-min stages on day two. After a 5-min break, the incremental test started at the individual power output of the third submaximal stage (rounded to the nearest 10-point value), and participants were instructed to continuously increase power output by 10 W every 30 s. The test was terminated when the participant, despite strong verbal encouragement, could no longer maintain the required power output of the 30-s stage and the VO2 values either plateaued or decreased (a drop of more than 2 mL
kg-1
min-1). After the incremental test, participants recovered passively for five min and actively for three min (at the power out- put of the first submaximal stage). They then performed a verification stage at a 10% higher power output than the peak power output of the incremental test (rounded to the nearest 10-point value) to verify the attainment of a true VO2peak [32]. The verification stage was ter- minated when the participant dropped more than 10% of target power output for more than five s. Test day 2. Seven to eight 5-min stages were performed with a 2- to 3-min break between stages and in the same upper-body poling mode. The first stage started at 20% of the individual peak power output obtained during the incremental test on day one, with increases of 10% (of the individual peak power output) for each consecutive stage. The last stage was terminated when the participant, despite strong verbal encouragement, could no longer maintain the power output of that stage and dropped more than 10% in the target power output for longer than five s. The intermittent exercise protocol was chosen to take a BLa sample from the fin- gertip in between stages. The duration of five min per stage was chosen, since in an upper- body mode two to three min are needed to achieve steady-state of physiological outcome parameters [33]. Outcome measurements Heart rate was measured every second with a Polar heart rate monitor, and respiratory param- eters (i.e., VO2, VCO2, VE, and respiratory exchange ratio (RER)) were measured breath-by- breath and averaged over 10 s by the in-built software of a Metamax II. A blood sample was taken from the fingertip and BLa analysed with a Lactate Pro device (Arkray Inc., Japan) at rest and directly after each of the submaximal stages on day one and day two, and one and three min after the incremental test and the verification stage on day one as well as the last stage of day two. Overall RPE was recorded after each of the submaximal stages on day one and two, as well as after the incremental test on day one and the last stage on day two. Power output was displayed per stroke and saved as 20-s averages during the submaximal stages on day one and day two by the in-built Concept2 software (Concept2, Morrisville, VT, USA). Peak power output during the incremental test and during the verification stage was registered as the highest 30-s average. Data analysis Data processing. Peak power output and gas exchange outcome parameters were calcu- lated as the highest 30-s moving average and peak heart rate (HRpeak) as the highest 3-s moving average of the incremental test performed on test day one. The gas exchange, heart rate and power output data of the last two min (12 x 10-s averages) of each complete 5-min stage conducted on test day two was included for data analysis in MATLAB (R2016a; Mathworks Inc., Natick, MA). The analyses in the following were based on the concatenated 2-min gas exchange data for the VT and RCT and on the BLa values after each 5-min stages for the LT1 and LT2. Different methods were used to determine both the VT and the RCT, as well as the LT1 and the LT2. For the determination of the VT, VO2 was plotted against VCO2 (V-slope method) Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 6 / 18 [3] as well as time against VE/VO2 and VE/VCO2 (ventilatory equivalent method) [7] and two regression lines fit to the data. For a valid detection of the VT with the ventilatory equivalent method, the VE/VO2 had to increase before an increase in VE/VCO2 [15, 34]. For the detec- tion of the RCT, VCO2 was plotted against VE [3] and two regression lines fit to the data. The LT1 was determined in two different ways: the first fixed rise in BLa concentration by 0.4 and 1 mmol
L-1 above the lowest individual BLa value [5, 35]. Additionally, the LT1 was deter- mined by breakpoints in the log-log transformed VO2-BLa relationship [4]. The LT2 was determined by a fixed BLa concentration of 4 mmol
L-1 [36]. Additionally, the LT2 was deter- mined by the modified Dmax method, which identifies the point on the 3rd order polynomial curve fitted to the BLa values that yields the maximal perpendicular distance to the straight line formed by the first stage with an increase of 0.4 mmol
L-1 and the BLa measured after the last stage [5]. Outcome parameters (% of peak power output, % of VO2peak, % of HRpeak, as well as BLa and RPE) were interpolated at the thresholds identified with each of the above described methods used to determine the VT, LT1, RCT and LT2. Statistical analyses. Paired-samples t-tests were used to compare the physiological and perceptual outcome parameters within the VT, LT1, RCT and LT2, and between all four differ- ent thresholds. Pearson’s r was used to investigate relationships between the outcome parame- ters identified with the different methods used to determine VT, LT1, RCT and LT2. Ranges of 0.26–0.49, 0.50–0.69, 0.70–0.89 and 0.90–1.0 were used to indicate low, moderate, high and very high correlations according to Munro’s criteria [37]. An α level of 0.05 was used to indi- cate statistical significance. To compare the fit of breakpoint models versus continuous linear or curvilinear (no-break- point) models to the gas exchange data, two regression lines (Eq 1) versus a single linear regression line (Eq 2), an exponential curve (Eq 3), and a 3rd order polynomial curve (Eq 4) were fitted to the VO2-VCO2, VE/VO2, VE/VCO2, and VCO2-VE data by linear least squares fitting. y ¼ a1 þ b1 x; t < k a2 þ b2 x; t  k ( ð1Þ y ¼ a þ bx ð2Þ y ¼ a þ c
exp x þ g d   ð3Þ y ¼ a þ b1 x þ b2 x2 þ b3 x3 ð4Þ y is the variable of interest, a the y-axis offset, b the slope coefficients, c and d spreading coeffi- cients, g the x-axis offset and k the point where the first and the second regression line of the piecewise function cross. To compare the fit of the four models, the Akaike information crite- rion (AIC) (Eq 5) [38] and the Akaike weights (wi) (Eq 7) for each model i relative to the set of R candidate models were calculated based on the delta AIC (Δi) (Eq 6) [39, 40]. AIC ¼ n
log SSer n   þ 2
K ð5Þ Delta AIC ¼ Di ¼ AICi AIC weight ¼ wi ¼ exp or LT1 (+1.0) (exception: power output and BLa at LT1 (+0.4): r>0.55, p<0.04; all other out- come parameters: r<0.38, p>0.16) (S1 File, sheet “correlations”). All outcome parameters at LT1 (+0.4) and LT1 (+1.0) were highly or very highly correlated (all r>0.83, p<0.001). In addi- tion, some of the outcome parameters identified with breakpoints in the log-log transformed VO2-BLa moderately correlated with the outcome parameters identified by the V-slope method (HR: r = 0.64, p = 0.01; BLa: r = 0.54, p = 0.04) and the breakpoints in the VE/VO2 data of the ventilatory equivalent method (HR: r = 0.54, p = 0.04). The outcome parameters identified with breakpoints in the VCO2-VE data at the RCT (e.g. BLa: 5.5±1.4 mmol
L-1) were not significantly different from the ones identified with the modi- fied Dmax method at the LT2 (5.5±1.1 mmol
L-1) (all p>0.53), but were higher compared to parameters identified with VE/VCO2 method (4.9±1.5 mmol
L-1) and a fixed BLa value of 4 mmol
L-1 (all p<0.03). Furthermore, there was no significant difference between the outcome parameters identified with V-slope method used to determine the VT and the ones identified with breakpoints in the VE/VCO2 and VCO2-VE data used to determine the RCT (p>0.22). However, most outcome parameters identified at the breakpoints in the VE/VO2 and VE/ VCO2 data (ventilatory equivalent method) were highly or very highly correlated with those identified at the breakpoints in the VCO2-VE data (RCT) (exception: % of VO2peak r = 0.67, p = 0.006; all other outcome parameters: r>0.73, p<0.01) (Fig 2). In addition, most outcome parameters identified at the thresholds in the VE/VCO2 data were moderately to highly corre- lated with the same outcome parameters at the thresholds identified with the modified Dmax method (exception: % of peak power output: r = 0.43, p = 0.11; all other outcome parameters: r>0.57, p<0.03). Furthermore, there was no significant difference between the outcome parameters identified with the log-log transformed VO2-BLa method used to determine the LT1 and at a fixed BLa concentration of 4 mmol
L-1 used to determine the LT2 and (all p>0.43). For the gas exchange data, all fitting procedures for the VO2-VCO2 and the VCO2-VE plots, including the single linear regression line, showed very good fit on the data for all 15 participants (mean r2>0.97) (Table 3). However, the fit of the breakpoint model compared to the continuous no-breakpoint models on the VO2-VCO2 and the VCO2-VE data was only bet- ter among five participants. Accordingly, the continuous no-breakpoint models had 71% and 68% probability of being the best models for the VO2-VCO2 and the VCO2-VE data, respec- tively (Table 4). Exemplary VO2-VCO2 and VCO2-VE plots are illustrated in Figs 3 and 4, respectively. In the gas exchange data displayed in the VE/VO2 plots and the VE/VCO2 plots, the break- point model fitted better than the continuous no-breakpoint models in six and seven of the athletes, respectively (Fig 5). Accordingly, it is unclear if in general the breakpoint (41 and Table 2. Mean ± SD (95% CI) peak power output and peak physiological and perceptual outcome parameters. Peak values Peak power output (W) 144 ± 37 (125–163) VO2peak (mL
kg-1
min-1) 36 ± 7 (32–39) HRpeak (beats
min-1) 188 ± 12 (182–194) Blood lactate (mmol
L-1) 14.4 ± 1.5 (13.7–15.2) RPE (6–20) 19.7 ± 0.5 (19.4–19.9) The data was collected during an incremental test to exhaustion while upper-body poling of 15 Norwegian Para ice hockey players. Peak oxygen uptake (VO2peak), peak heart rate (HRpeak), rating of perceived exertion (RPE) https://doi.org/10.1371/journal.pone.0205588.t002 Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 9 / 18 47%, respectively) or continuous no-breakpoint (59 and 53%, respectively) models fit the VE/ VO2 and the VE/VCO2 data best (Table 4). The rise in VE/VO2 occurred earlier than the VE/ VCO2 only in four athletes (S1 Fig). The VT detection by the VE/VO2 relationship was, there- fore, only valid in these four athletes. In none of these four athletes, did the breakpoint model fit the VE/VO2 data better than the continuous no-breakpoint models. Discussion The main aim of this study was to compare physiological and perceptual outcome parameters identified with common gas exchange and BLa thresholds methods used to determine the VT, LT1, RCT and LT2 in Paralympic athletes while upper-body poling. Furthermore, we com- pared the fit of breakpoint models used to determine gas exchange thresholds to the fit of con- tinuous linear or curvilinear (i.e., no-breakpoint) models. The LT1 occurred at much lower exercise intensity than the VT although both are used as indicators of AT, whereas there were no or minor differences between the methods used to identify the RCT and LT2 that deter- mine the ANT. Furthermore, the RCT and LT2 did not differ from the VT. In addition, the outcome parameters corresponding to the LT1 and LT2 using the log-log transformed VO2- BLa data and the modified Dmax method, respectively, were significantly higher than ones identified with fixed BLa values at the LT1 and LT2 (i.e., rise in BLa of +0.4/1.0 at LT1 or BLa concentration of 4 mmol
L-1 at LT2). We were able to determine breakpoints at the VT and RCT with different gas exchange methods with good fit in all 15 participants, although contin- uous no-breakpoint models showed even better fit for the majority of participants. The physiological and perceptual outcome parameters identified with a fixed rise in BLa at the LT1 were significantly lower than the ones at the VT, and the outcome parameters using these methods only low or moderately correlated with each other. Overall, this indicates that these two thresholds cannot be used interchangeably to determine the AT. In addition, Table 4. Akaike weights (wi) representing a measure of strength of evidence for probability of best fit of the two regression lines (breakpoint model) and the single regression line, exponential and 3rd order polynomial curve (continuous no-breakpoint models) (mean wi ± SD (95% CI) [# of participants with better fit of the respective model compared to the two regression lines]) fitted to the gas exchange data of 15 elite Para ice hockey players following a protocol with stepwise increases in workload every 5 min while upper-body poling. Two regression lines Single regression line Exponential curve 3rd order polynomial curve VO2-VCO2 plots 0.29 ± 0.35 (0.11–0.46) 0.07 ± 0.18 (-0.03–0.16) [#0] 0.03 ± 0.10 (-0.01–0.08) [#0] 0.61 ± 0.37 (0.42–0.79) [#10] VE/VO2 plots 0.41 ± 0.45 (0.18–0.64) 0.00 ± 0.00 (0.00–0.00) [#0] 0.13 ± 0.24 (0.01–0.26) [#2] 0.46 ± 0.40 (0.25–0.66) [#7] VE/VCO2 plots 0.47 ± 0.49 (0.22–0.72) 0.00 ± 0.00 (0.00–0.00) [#0] 0.09 ± 0.19 (-0.01–0.19) [#2] 0.44 ± 0.43 (0.22–0.66) [#6] VCO2-VE plots 0.31 ± 0.44 (0.09–0.54) 0.00 ± 0.00 (0.00–0.00) [#0] 0.00 ± 0.01 (0.00–0.01) [#0] 0.68 ± 0.44 (0.46–0.91) [#10] Oxygen uptake (VO2), carbon dioxide production (VCO2), minute ventilation (VE) https://doi.org/10.1371/journal.pone.0205588.t004 Table 3. The coefficient of determination (mean r2 ± SD (range) for the two regression lines (breakpoint model) and the single regression line, exponential and 3rd order polynomial curve (continuous no-breakpoint models) fitted to the gas exchange data of 15 elite Para ice hockey players following a protocol with stepwise increases in workload every 5 min while upper-body poling. Two regression lines Single regression line Exponential curve 3rd order polynomial curve VO2-VCO2 plots 0.995 ± 0.005 (0.993–0.998) 0.994 ± 0.005 (0.991–0.996) 0.993 ± 0.005 (0.991–0.996) 0.996 ± 0.005 (0.993–0.998) VE/VO2 plots 0.931 ± 0.069 (0.896–0.966) 0.764 ± 0.094 (0.716–0.811) 0.919 ± 0.064 (0.886–0.951) 0.932 ± 0.064 (0.900–0.964) VE/VCO2 plots 0.940 ± 0.044 (0.918–0.962) 0.700 ± 0.142 (0.628–0.772) 0.920 ± 0.044 (0.898–0.942) 0.940 ± 0.041 (0.919–0.961) VCO2-VE plots 0.995 ± 0.003 (0.994–0.997) 0.968 ± 0.015 (0.960–0.976) 0.992 ± 0.006 (0.989–0.994) 0.996 ± 0.003 (0.994–0.998) Oxygen uptake (VO2), carbon dioxide production (VCO2), minute ventilation (VE) https://doi.org/10.1371/journal.pone.0205588.t003 Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 10 / 18 Fig 3. Exemplary VO2-VCO2 plots. The VO2-VCO2 data was fitted with a single regression line, a bilinear regression line, an exponential curve, and a 3rd order polynomial curve for an athlete without breakpoint (the four plots to the left) and with suggested breakpoint presence (the four plots to the right). (Note that the plots of the five athletes with a suggested breakpoint also show a rather linear increase in the VO2-VCO2 relationship). Oxygen uptake (VO2), carbon dioxide production (VCO2). https://doi.org/10.1371/journal.pone.0205588.g003 Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 11 / 18 Fig 4. Exemplary VCO2-VE plots. The VCO2-VE data was fitted with a single regression line, a bilinear regression line, an exponential curve, and a third order polynomial curve for an athlete without breakpoint (the four plots to the left) and with suggested breakpoint presence (the four plots to the right). (Note that the plots of the five athletes with a suggested breakpoint show a rather curvilinear increase in the VCO2-VE relationship). Carbon dioxide production (VCO2), minute ventilation (VE). https://doi.org/10.1371/journal.pone.0205588.g004 Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 12 / 18 Fig 5. Exemplary VE/VO2 and VE/VCO2 plots. Exemplary VE/VO2 data fitted with a bilinear regression line and a 3rd order polynomial curve for an athlete without breakpoint (upper two plots to the left) and with suggested breakpoint presence (upper two plots to the right). Exemplary VE/VCO2 data fitted with a bilinear regression line and a 3rd order polynomial curve for an athlete without breakpoint (lower two plots to the left) and with suggested breakpoint presence (upper two plots to the right). Oxygen uptake (VO2), carbon dioxide production (VCO2), minute ventilation (VE). https://doi.org/10.1371/journal.pone.0205588.g005 Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 13 / 18 thresholds identified by a fixed BLa increase at the LT1 were significantly lower compared with the breakpoints identified in the log-log transformed VO2-BLa data, showing that indi- vidually adjustable BLa methods did not correspond with fixed methods in determining the LT1. The early occurrence of a rise in BLa in upper-body exercise is in accordance with Beneke et al. [41], who found BLa to be higher at a given workload in activities involving smaller mus- cle mass, where power output per kg of active muscle mass and, thus, local metabolic stress is increased compared to lower body exercise. In addition, BLa accumulation after cessation of exercise was shown to be faster in individuals with a spinal cord injury as compared to able- bodied individuals [42]. However, although outcome parameters identified with breakpoints in the log-log transformed VO2-BLa data are not significantly lower than the ones identified at the VT, outcome parameters identified with methods using fixed BLa values to identify the LT1 are much lower than the VT. As estimates of the ANT, the outcome parameters identified with the Dmax method to deter- mine LT2 did not significantly differ from the ones identified with breakpoints in the VCO2- VE data at the RCT, whereas most of the outcome parameters identified with breakpoints in the VE/VCO2 data were significantly lower than these. However, the outcome parameters identified by the latter method differ only marginally from the two other ANT methods (VCO2-VE, Dmax), indicating that the exercise intensity where a disproportionate increase in BLa and in VE occurs is relatively similar. Note that we decided to not correct for multiple comparisons and rather present the uncorrected p-values from paired samples t-tests instead. Although we are aware of the subsequent increased chances of making a type 1 errors, the decreased chances of making a type a type 2 errors were regarded more important, which is in accordance with Rothman [43]. However, if Bonferroni corrections would have been used in this specific case, there would have been no significant differences between the outcome mea- sures identified at with these three methods. Furthermore, most of the outcome parameters identified with the different methods at the LT2 and RCT are low to moderately correlated, coinciding with high individual variation in the outcome parameters within each of the methods used to identify the LT2 and RCT. This indicates that an individual with a high LT2 does not necessarily display a high RCT. The high individual variation may be explained by disability-related differences in the cardio-respiratory system that might affect physiological responses to upper-body exercise. For example, athletes with a spinal cord injury exercising in an upper-body mode were shown to vary considerably in their VO2peak depending on their level of injury [44], which might also reflect differences in the % of VO2peak that can be sustained during exercise. In addition, the inclusion of one par- ticipant that was much older than the rest and one female participant may have contributed to the high variation. Furthermore, individual variation in physiological responses may be higher in upper-body exercise compared to lower-body exercise. Altogether, it is questionable whether the similar outcome parameters identified at the LT2 and the RCT on a group basis, result in similar outcome parameters at the LT2 and RCT for the individual sitting athlete when training in an upper-body mode. The thresholds identified by the breakpoints in the VE/VO2 at the VT and in VE/VCO2 at the RCT did not significantly differ and were highly correlated. This, together with the rather linear increase in the VO2-VCO2 relationship suggests that it is solely the disproportionate rise in VE that leads to a rather rapid increase in the data of the VE/VO2 and the VE/VCO2 plots, and to discernible breakpoints in approximately half of the participants. Together with the close location of the breakpoints identified in the VCO2-VO2 data at the VT and the VCO2- VE data at the RCT, this indicates that a two-phase (low-high) rather than a three-phase (low- moderate-high) intensity zone model could be applicable in athletes with a disability who exer- cise in an upper-body mode. This is in contrast to significant differences between the VT and Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 14 / 18 the RCT in Dekerle et al. [45], who test able-bodied participants in the arm crank ergometry mode, and Leicht et al. [18], who tested wheelchair athletes in the wheelchair treadmill mode. However, our findings are in line with a study of Pires et al. [46], who also found one rather than two thresholds in the gas exchange data in upper-body trained able-bodied participants during exercise in the arm crank ergometry mode. Whether the discrepancies between studies are related to employment of e.g. different populations, protocols or exercise modes needs to be examined further in other experimental designs. All gas exchange threshold methods have in common that there is an a priori assumption of the presence of a breakpoint, defined as “a place where an interruption or change occurs” [47]. However, the presence or absence of breakpoints in the gas exchange data is a debated topic [8, 12]. Thus, in addition to the breakpoint models used to identify the VT and the RCT in the present study, we fitted continuous no-breakpoint models to the data to investigate if there are clear breakpoints in our data. Here, we found good fit for the breakpoint model used to iden- tify the gas exchange thresholds, but better fit for the curvilinear no-breakpoint models in most cases. We, hence, question if clear breakpoints really exist in the gas exchange data of ath- letes with disabilities in an upper-body exercise mode. Conclusion In Paralympic athletes who exercise in upper-body poling, the physiological and perceptual outcome parameters identified at the VT and the LT1 showed large differences, which demon- strates that these cannot be used interchangeably to identify the AT. In addition, the close loca- tion of the VT, RCT and LT2 does not allow us to distinguish the AT and ANT, indicating that there might only be one threshold in athletes with a disability exercising in an upper-body mode. Furthermore, continuous no-breakpoint models fit the gas exchange data better than breakpoint models in most participants. We, hence, question if clear breakpoints in the gas exchange data really exist in an upper-body exercise mode in athletes with disabilities. Supporting information S1 File. Data. Data and analyses conducted in this study of gas exchange and blood lactate threshold in Paralympic sitting athletes. (XLSX) S1 Fig. VE/VO2 and VE/VCO2 plots fitted with two regression lines. The data is of the six or seven completed stages of each of the 15 athletes. Breakpoint presence is indicated above each individual plot. Furthermore, it is indicated in the second row above the figures whether the two thresholds occur at the same time, or the VE/VO2 occurs before or after the VE/VCO2 threshold. Oxygen uptake (VO2), carbon dioxide production (VCO2), minute ventilation (VE). (TIF) Acknowledgments The laboratory equipment was provided by NeXt Move, Norwegian University of Science and Technology (NTNU). NeXt Move is funded by the Faculty of Medicine at NTNU and Central Norway Regional Health Authority. The funder had no role in study design, how the data collection and analysis was performed, decision to publish, or preparation of the manuscript. The authors acknowledge the support of the Olympic and Paralympic Centre in Oslo and the Centre for Elite Sports Research in Trondheim in conducting this research. The eager partici- pation of the athletes is deeply appreciated. None of the authors have any conflicts of interest to declare. Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 15 / 18 Author Contributions Conceptualization: Julia Kathrin Baumgart, Maaike Moes, Knut Skovereng, Gertjan Ettema, Øyvind Sandbakk. Data curation: Julia Kathrin Baumgart, Maaike Moes. Formal analysis: Julia Kathrin Baumgart, Knut Skovereng, Gertjan Ettema, Øyvind Sandbakk. Funding acquisition: Øyvind Sandbakk. Investigation: Julia Kathrin Baumgart, Øyvind Sandbakk. Methodology: Julia Kathrin Baumgart, Maaike Moes, Knut Skovereng, Gertjan Ettema, Øyvind Sandbakk. Project administration: Julia Kathrin Baumgart, Øyvind Sandbakk. Supervision: Gertjan Ettema, Øyvind Sandbakk. Validation: Julia Kathrin Baumgart, Gertjan Ettema, Øyvind Sandbakk. Visualization: Julia Kathrin Baumgart, Gertjan Ettema. Writing – original draft: Julia Kathrin Baumgart. Writing – review & editing: Maaike Moes, Knut Skovereng, Gertjan Ettema, Øyvind Sandbakk. References 1. Seiler S, Toennessen E. Intervals, Thresholds, and Long Slow Distance: the Role of Intensity and Dura- tion in Endurance Training. Sportscience. 2009; 13:32–53. 2. Binder RK, Wonisch M, Corra U, Cohen-Solal A, Vanhees L, Saner H, et al. Methodological approach to the first and second lactate threshold in incremental cardiopulmonary exercise testing. Eur J Cardio- vasc Prev Rehabil. 2008; 15(6):726–34. https://doi.org/10.1097/HJR.0b013e328304fed4 PMID: 19050438 3. Beaver WL, Wasserman K, Whipp BJ. A new method for detecting anaerobic threshold by gas exchange. J Appl Physiol (1985). 1986; 60(6):2020–7. 4. Beaver WL, Wasserman K, Whipp BJ. Improved detection of lactate threshold during exercise using a log-log transformation. J Appl Physiol (1985). 1985; 59(6):1936–40. 5. Bishop D, Jenkins DG, Mackinnon LT. The relationship between plasma lactate parameters, Wpeak and 1-h cycling performance in women. Med Sci Sports Exerc. 1998; 30(8):1270–5. PMID: 9710868 6. Wasserman K. The anaerobic threshold: definition, physiological significance and identification. Adv Cardiol. 1986; 35:1–23. 7. Reinhard U, Muller PH, Schmulling RM. Determination of anaerobic threshold by the ventilation equiva- lent in normal individuals. Respiration. 1979; 38(1):36–42. https://doi.org/10.1159/000194056 PMID: 493728 8. Myers J, Ashley E. Dangerous curves. A perspective on exercise, lactate, and the anaerobic threshold. Chest. 1997; 111(3):787–95. PMID: 9118720 9. Wasserman K. The anaerobic threshold measurement to evaluate exercise performance. Am Rev Respir Dis. 1984; 129(2 Pt 2):S35–40. https://doi.org/10.1164/arrd.1984.129.2P2.S35 PMID: 6421216 10. Wasserman K. Determinants and detection of anaerobic threshold and consequences of exercise above it. Circulation. 1987; 76(6 Pt 2):Vi29–39. PMID: 3315297 11. Faude O, Kindermann W, Meyer T. Lactate threshold concepts: how valid are they? Sports Med. 2009; 39(6):469–90. https://doi.org/10.2165/00007256-200939060-00003 PMID: 19453206 12. Hopker JG, Jobson SA, Pandit J. Controversies in the physiological basis of the ‘anaerobic threshold’- and their implications for clinical cardiopulmonary exercise testing. Anaesthesia. 2011; 66(2):111–23. https://doi.org/10.1111/j.1365-2044.2010.06604.x PMID: 21254986 Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 16 / 18 13. Peronnet F, Aguilaniu B. Lactic acid buffering, nonmetabolic CO2 and exercise hyperventilation: a criti- cal reappraisal. Respir Physiol Neurobiol. 2006; 150(1):4–18. https://doi.org/10.1016/j.resp.2005.04. 005 PMID: 15890562 14. Brooks GA. Anaerobic threshold: review of the concept and directions for future research. Med Sci Sports Exerc. 1985; 17(1):22–34. PMID: 3884959 15. Gaskill SE, Ruby BC, Walker AJ, Sanchez OA, Serfass RC, Leon AS. Validity and reliability of combin- ing three methods to determine ventilatory threshold. Med Sci Sports Exerc. 2001; 33(11):1841–8. PMID: 11689733 16. Maffulli N, Testa V, Capasso G. Anaerobic threshold determination in master endurance runners. J Sports Med Phys Fitness. 1994; 34(3):242–9. PMID: 7830387 17. Chicharro JL, Perez M, Vaquero AF, Lucia A, Legido JC. Lactic threshold vs ventilatory threshold during a ramp test on a cycle ergometer. J Sports Med Phys Fitness. 1997; 37(2):117–21. PMID: 9239989 18. Leicht CA, Griggs KE, Lavin J, Tolfrey K, Goosey-Tolfrey VL. Blood lactate and ventilatory thresholds in wheelchair athletes with tetraplegia and paraplegia. Eur J Appl Physiol. 2014; 114(8):1635–43. https:// doi.org/10.1007/s00421-014-2886-x PMID: 24781928 19. Ribeiro J, Figueiredo P, Sousa M, De Jesus K, Keskinen K, Vilas-Boas JP, et al. Metabolic and ventila- tory thresholds assessment in front crawl swimming. J Sports Med Phys Fitness. 2015; 55(7–8):701–7. PMID: 25069963 20. Bernardi M, Guerra E, Di Giacinto B, Di Cesare A, Castellano V, Bhambhani Y. Field evaluation of para- lympic athletes in selected sports: implications for training. Med Sci Sports Exerc. 2010; 42(6):1200–8. PMID: 19997027 21. Bhambhani YN, Holland LJ, Steadward RD. Anaerobic threshold in wheelchair athletes with cerebral palsy: validity and reliability. Arch Phys Med Rehabil. 1993; 74(3):305–11. PMID: 8439260 22. Coutts KD, McKenzie DC. Ventilatory thresholds during wheelchair exercise in individuals with spinal cord injuries. Paraplegia. 1995; 33(7):419–22. PMID: 7478733 23. Davis JA, Vodak P, Wilmore JH, Vodak J, Kurtz P. Anaerobic threshold and maximal aerobic power for three modes of exercise. J Appl Physiol. 1976; 41(4):544–50. https://doi.org/10.1152/jappl.1976.41.4. 544 PMID: 985399 24. Keyser RE, Mor D, Andres FF. Cardiovascular responses and anaerobic threshold for bicycle and arm ergometer exercise. Arch Phys Med Rehabil. 1989; 70(9):687–91. PMID: 2774887 25. Lin KH, Lai JS, Kao MJ, Lien IN. Anaerobic threshold and maximal oxygen consumption during arm cranking exercise in paraplegia. Arch Phys Med Rehabil. 1993; 74(5):515–20. PMID: 8489362 26. Orr JL, Williamson P, Anderson W, Ross R, McCafferty S, Fettes P. Cardiopulmonary exercise testing: arm crank vs cycle ergometry. Anaesthesia. 2013; 68(5):497–501. https://doi.org/10.1111/anae.12195 PMID: 23573845 27. Schneider DA, Sedlock DA, Gass E, Gass G. VO2peak and the gas-exchange anaerobic threshold dur- ing incremental arm cranking in able-bodied and paraplegic men. Eur J Appl Physiol Occup Physiol. 1999; 80(4):292–7. PMID: 10483798 28. Vinet A, Le Gallais D, Bernard PL, Poulain M, Varray A, Mercier J, et al. Aerobic metabolism and cardioventilatory responses in paraplegic athletes during an incremental wheelchair exercise. Eur J Appl Physiol Occup Physiol. 1997; 76(5):455–61. https://doi.org/10.1007/s004210050275 PMID: 9367286 29. Yasuda N, Gaskill SE, Ruby BC. No gender-specific differences in mechanical efficiency during arm or leg exercise relative to ventilatory threshold. Scand J Med Sci Sports. 2008; 18(2):205–12. https://doi. org/10.1111/j.1600-0838.2007.00637.x PMID: 17490463 30. Hegge AM, Bucher E, Ettema G, Faude O, Holmberg HC, Sandbakk O. Gender differences in power production, energetic capacity and efficiency of elite cross-country skiers during whole-body, upper- body, and arm poling. Eur J Appl Physiol. 2015. 31. Borg GA. Psychophysical bases of perceived exertion. Med Sci Sports Exerc. 1982; 14(5):377–81. PMID: 7154893 32. Leicht CA, Tolfrey K, Lenton JP, Bishop NC, Goosey-Tolfrey VL. The verification phase and reliability of physiological parameters in peak testing of elite wheelchair athletes. Eur J Appl Physiol. 2013; 113 (2):337–45. https://doi.org/10.1007/s00421-012-2441-6 PMID: 22718268 33. Inbar O, Faina M, Demarie S, Whipp BJ. VO 2 Kinetics during Moderate Effort in Muscles of Different Masses and Training Level. ISRN Physiology. 2012; 2013. 34. Powers SK, Dodd S, Garner R. Precision of ventilatory and gas exchange alterations as a predictor of the anaerobic threshold. Eur J Appl Physiol Occup Physiol. 1984; 52(2):173–7. PMID: 6538832 Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 17 / 18 35. Buckley JD, Bourdon PC, Woolford SM. Effect of measuring blood lactate concentrations using different automated lactate analysers on blood lactate transition thresholds. J Sci Med Sport. 2003; 6(4):408–21. PMID: 14723391 36. Stegmann H, Kindermann W. Comparison of prolonged exercise tests at the individual anaerobic threshold and the fixed anaerobic threshold of 4 mmol.l(-1) lactate. Int J Sports Med. 1982; 3(2):105– 10. https://doi.org/10.1055/s-2008-1026072 PMID: 7107102 37. Plichta SB, Kelvin EA, Munro BH. Munro’s statistical methods for health care research: Wolters Kluwer Health/Lippincott Williams & Wilkins; 2013. 38. Bozdogan H. Model selection and Akaike’s information criterion (AIC): The general theory and its ana- lytical extensions. Psychometrika. 1987; 52(3):345–70. 39. Burnham KP, Anderson DR. Model selection and multimodel inference: a practical information-theoretic approach: Springer Science & Business Media; 2003. 40. Wagenmakers E-J, Farrell S. AIC model selection using Akaike weights. Psychonomic bulletin & review. 2004; 11(1):192–6. 41. Beneke R, Leitha¨user R, Hu¨tler M. Dependence of the maximal lactate steady state on the motor pat- tern of exercise. Br J Sports Med. 2001; 35(3):192–6. https://doi.org/10.1136/bjsm.35.3.192 PMID: 11375880 42. Leicht C, Perret C. Comparison of blood lactate elimination in individuals with paraplegia and able-bod- ied individuals during active recovery from exhaustive exercise. J Spinal Cord Med. 2008; 31(1):60–4. PMID: 18533413 43. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology. 1990; 1(1):43–6. PMID: 2081237 44. Bhambhani Y. Physiology of wheelchair racing in athletes with spinal cord injury. Sports Med. 2002; 32 (1):23–51. PMID: 11772160 45. Dekerle J, Dupont L, Caby I, Marais G, Vanvelcenaher J, Lavoie JM, et al. Ventilatory thresholds in arm and leg exercises with spontaneously chosen crank and pedal rates. Percept Mot Skills. 2002; 95(3 Pt 2):1035–46. https://doi.org/10.2466/pms.2002.95.3f.1035 PMID: 12578244 46. Pires FO, Hammond J, Lima-Silva AE, Bertuzzi RC, Kiss MA. Ventilation behavior during upper-body incremental exercise. J Strength Cond Res. 2011; 25(1):225–30. https://doi.org/10.1519/JSC. 0b013e3181b2b895 PMID: 20093972 47. Definition of a break point in English [Internet]. Oxford University Press;2017 [cited 2017 May 19]. https://en.oxforddictionaries.com/definition/break_point. Gas exchange and blood lactate thresholds in upper-body poling PLOS ONE | https://doi.org/10.1371/journal.pone.0205588 October 31, 2018 18 / 18

Examination of gas exchange and blood lactate thresholds in Paralympic athletes during upper-body poling.

10-31-2018

Baumgart, Julia Kathrin,Moes, Maaike,Skovereng, Knut,Ettema, Gertjan,Sandbakk, Øyvind

eng

PMC4377953

Int. J. Environ. Res. Public Health 2015, 12, 3077-3090; doi:10.3390/ijerph120303077 International Journal of Environmental Research and Public Health ISSN 1660-4601 www.mdpi.com/journal/ijerph Article Attentional Distraction during Exercise in Overweight and Normal-Weight Boys Benedicte Deforche 1,2,* and Ilse De Bourdeaudhuij 3 1 Department of Public Health, Ghent University, De Pintelaan 185, 9000 Gent, Belgium 2 Department of Human Biometrics and Biomechanics, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussel, Belgium 3 Department of Movement and Sports Sciences, Ghent University, Watersportlaan 2, 9000 Gent, Belgium; E-Mail: Ilse.Debourdeaudhuij@UGent.be * Author to whom correspondence should be addressed; E-Mail: Benedicte.Deforche@UGent.be. Academic Editor: Andrew P. Hills Received: 19 October 2014 / Accepted: 4 March 2015 / Published: 13 March 2015 Abstract: The purpose of this study was to investigate the effect of attentional distraction on field running distance and activity intensity during an exercise session in normal-weight and overweight youngsters and to investigate potential mediators. Fifty-three 12–14 yr-old boys participated twice in a 12-min running test and a 20-min exercise session, once with attentional distraction (by listerning to music) and once without distraction (counterbalanced randomised controlled design). At the end of the endurance test running distance was recorded. During the exercise session activity intensity was assessed by accelerometers. After each experiment, rate of perceived exertion (RPE) was estimated and seven questions were asked about how participants experienced the experiment. Both overweight and normal-weight boys ran further during the running test with music (p < 0.05) and this effect was mediated by a decrease in feelings of annoyance. During the exercise session with music, both overweight and normal-weight boys exercised less at low and high intensity and more at moderate and very high intensity (p < 0.01) and this effect was mediated by a decrease in RPE. We can conclude that attentional distraction has a positive effect on running distance on a field endurance test and on activity intensity during an exercise session through different mechanisms in both overweight and normal-weight boys. OPEN ACCESS Int. J. Environ. Res. Public Health 2015, 12 3078 Keywords: adolescents; music; obesity; physical activity; running performance 1. Introduction Overweight youngsters are found to be less active or to be active at a lower intensity than normal-weight counterparts [1,2]. As regular physical activity of high enough intensity is essential in the prevention of obesity [3,4], efforts should be made to increase physical activity adherence in overweight youngsters. Enjoyment is the most important predictor of physical activity in children [5–7]. Unfortunately, overweight youngsters generally do not choose to be active because they like it, but rather because they hope to lose weight or look better [8]. These extrinsic motives may not encourage continued participation in physical activity, since weight-loss directly attributable to increased physical activity is usually small [9]. The lack of direct effects of physical activity may disappoint overweight youngsters and cause drop-out. In addition, overweight youngsters perceive more barriers towards physical activity [8,10,11]. They find it more exhausting and report more physical complaints such as side stiches, knee pain, suffocating feeling, excessive sweating, etc. Moreover, a higher perception of barriers in overweight youngsters is related to a lower participation in physical activities [11]. Since physical activity plays an important role in the prevention of overweight, it is important to find solutions to overcome these barriers and to intrinsically motivate overweight youngsters to be physically active. A key element is that overweight youngsters should perform enough activity to substantially increase total energy expenditure. This can be accomplished either by prolonging the duration of the activity or by raising its intensity. Generally, the importance of exercise duration, rather than intensity, is emphasized to promote a significant fat oxidation and to prevent drop-out [12]. The intensity of physical activity has been found to be negatively associated with exercise adherence in overweight children and adults [13,14]. This could be due to the fact that higher intensity activities entail more physical complaints and are therefore experienced as less pleasant. However, especially activities of moderate to high intensity have shown to contribute to the prevention of weight gain in adults [15,16]. It might be interesting to investigate ways to increase exercise intensity in overweight youngsters without increasing physical complaints or annoyance. It is possible that attentional distraction may increase adherence to higher intensity activities in overweight youngsters. Previous experiments in athletes showed that by focusing attention to external stimuli (such as music) instead of internal sensory information (such as heart rate, breathing, bodily symptoms) running performances increased [17]. Although the effect of attentional distraction is well established in athletes, only one study has investigated this issue in overweight people. A previous experiment in obese children and adolescents [18] showed that attentional distraction by music has a positive effect on perseverance during a treadmill test. This study was performed in a controlled laboratory environment. As findings from a laboratory setting might not be transferable to field settings and the effect of attentional distraction on activity intensity during an exercise session has not been studied yet, further research is needed to investigate whether attentional distraction is also useful to increase performance in field conditions and during exercise programs in overweight youngsters. Int. J. Environ. Res. Public Health 2015, 12 3079 The purpose of this study is to investigate the effect of attentional distraction in overweight versus normal-weight youngsters on: (1) running distance in the field and on intensity of activity during an exercise session (primary outcomes) and (2) rate of perceived exertion (RPE), degree of annoyance, attention given to bodily sensations or thoughts about being able to carry on (secondary outcomes). We hypothesize that attentional distraction will have a positive effect on primary and secondary outcomes in both exercise conditions. As overweight youngsters report more physical complaints while exercising [8], we further hypothesize that this effect will be stronger in overweight compared to normal-weight youngsters. In addition, we want to investigate whether the effect of attentional distraction on exercise performance is mediated by changes in rate of perceived exertion (RPE), degree of annoyance, attention given to bodily sensations or thoughts about being able to carry on. 2. Methods 2.1. Participants Four classes of 12 to 14 year old boys (N = 53) were recruited in a boys’ technical/vocational school with high overweight prevalence and were grouped into normal-weight or overweight according to international cutoffs for overweight in children [19]. Based on differences in running performance with and without distraction found in a previous laboratory study in obese children and adolescents [18] an a priori power analysis was conducted. This analysis showed that to study 2 × 2 within-between interactions with a power of 0.80 (given a 0.05 level of significance), a total sample size of minimum 38 subjects was needed. All participants were orally informed about the purpose of this study and received an information letter for their parents. Written informed consent was obtained from all boys and their parents before participating in the study. The study was conducted in accordance with the Declaration of Helsinki, and the protocol was approved by the ethical committee of the Ghent University Hospital. 2.2. Procedure Participants performed twice a field running test and participated twice in an exercise session, once with attentional distraction and once without distraction. To control for order effects, in both the running test and the exercise session, half of the classes started with distraction and half of the classes without distraction (counterbalanced design). As randomization of order of the conditions was at the class level and not individual level, this was a quasi-experimental randomized controlled design. Since tests were performed during physical education classes, they took place each time at the same day of the week and the same time of the day. There were two weeks between each test. No encouragement was given during the tests. 2.3. Measurements 2.3.1. Anthropometric Measurements Height was measured to the nearest 0.1 cm using a stadiometer (Holtain Ltd, Crymmych, Pembs, UK). Body mass was measured to the nearest 0.1 kg on a digital balance scale (Seca, max 200 kg, Int. J. Environ. Res. Public Health 2015, 12 3080 Hamburg, Germany) with the participant wearing lightweight clothing and no shoes. BMI was calculated from height and weight measures (weight in kg/height in m2). 2.3.2. Physical Activity Physical activity was estimated using a modified version of the Baecke Questionnaire [20], previously used to assess physical activity in 12–18 year old youngsters [8,21]. Responses to 13 questions were scored on a 5-point scale and resulted in two indices reflecting physical activity during sport (sport index) and during leisure time excluding sport (leisure time index). Items regarding physical activity during work were omitted for this study. A sport score was calculated from a combination of the intensity of the (organised or non-organised) sport which was played, the amount of time per week playing that sport, and the proportion of the year in which the sport was played regularly. The sport index was calculated from the sport score, level of activity in comparison with friends and frequency of sweating during leisure time physical activities. The leisure time index was based on the amount of television watching and the frequency and daily amount of walking or cycling as a means of transportation. The validity of the Baecke Questionnaire for the assessment of physical activity has been previously reported [22,23]. 2.3.3. Field Running Performance Field running performance was assessed by the Cooper Test [24] which is a 12 min running test. Different investigators found high correlations (0.82 < r < 0.94) between VO2max and performance on the 12 min running test in young adults [25,26]. The tests were performed on an outdoor athletics track. Weather conditions were similar on the test days with and without attentional distraction. In the condition with attentional distraction, participants were wearing a portable audio player (Sony D-EJ 750, G-protection). As the largest benefits to RPE were found with music that is preferred [27], each participant could bring his own favourite music. Music volume was standardised, however tempo was not controlled. At the end of the endurance tests running distance was recorded. 2.3.4. Activity during Exercise Session The exercise session consisted of a 20 min exercise circuit with focus on movements with vertical displacement of the body. Participants were not wearing a portable audio player, but music was played on a CD-player (AZ 2030, Philips, Eindhoven, The Netherlands). It was a mix of popular hits with a fast tempo and a strong rhythm. During the exercise sessions physical activity was assessed by accelerometry (model 7164, Computer Science Application, Inc., Shalimar, FL, USA). The accelerometers were set to measure activity counts in an epoch time of one minute. Activity counts are the summation of the accelerations measured over the epoch and are used to determine the intensities of activities performed. A distinction was made between minutes of less than 1952 counts (less than 3 METS), 1952 to 5724 counts (3.0 to 5.99 METS), 5725 to 9498 counts (6.0 to 8.99 METS) and more than 9498 counts (more than 8.99 METS), corresponding respectively to activity of light, moderate, high and very high intensity [28]. Participating boys were imposed to wear the accelerometer above the right hipbone, underneath the clothes. Accelerometers were held in place with an elastic belt and Int. J. Environ. Res. Public Health 2015, 12 3081 adjustable buckle. The accelerometer has been shown to be a valid and reliable tool for the assessment of physical activity in children [29–31]. 2.3.5. Rate of Perceived Exertion After each test or exercise session, rate of perceived exertion (RPE) was obtained using the Borg 15-point category scale [32]. RPE is defined as the subjective intensity of effort, strain, discomfort and/or fatigue experienced during exercise. The Borg 15-point category scale consists of numbered categories, 6–20, and verbal cues from “very, very light” to “very, very hard”. This scale is commonly used to measure RPE during exercise in normal-weight and overweight youth [33–35]. 2.3.6. Questionnaire After each test or exercise session seven questions were asked about how the participants experienced the test or the session using a 5-point scale (1 = not at all, 5 = very much). Participants reported (1) how annoying they experienced the test (or exercise session), (2) how much attention was given to bodily sensations during the test (or exercise session), (3) how often they had thoughts about being able to carry on with the test (or exercise session), (4) to what extent they liked the music, (5) to what extent they could listen to the music during the test (or exercise session), (6) how pleasant they found the test (or exercise session) while listening to music, and (7) to what extent they believed they could run further (or exercise more intensively) while listening to music. The latter four questions were only assessed in the conditions with attentional distraction. The first three questions were assessed after the tests with and without attentional distraction. 2.4. Statistical Analyses Data were analysed using SPSS software (version 21.0). Values of p < 0.05 were considered statistically significant. Effect of attentional distraction was studied using a 2 (condition: distraction versus no distraction) × 2 (group: normal-weight versus overweight) repeated measures analyses of variance. Results from items only obtained during the tests with distraction were analysed with independent samples t-tests. Mediation of attentional distraction effects in primary outcomes (running distance and activity intensity) by attentional distraction effects in secondary outcomes (RPE, degree of annoyance, attention given to bodily sensations or thoughts about being able to carry on) was tested using a within-subject method suggested by Judd et al. [36]. In order to do this analysis, there must first be a distraction effect for both the primary outcomes (dependent variable) and secondary outcomes (mediator variables). Mediation analysis was performed by estimating a regression model where the difference in running distance/activity intensity (with music minus without music) was regressed onto the sum of the mediator variables and the difference of the mediator variables. If the regression coefficient for the difference predictor is significant, this indicates that differences in running distance/activity intensity are mediated by differences in the mediator variable. If the sum predictor, but not difference predictor, is mean-centered (each participant’s score subtracted from the mean score of the sample), complete mediation is indicated by a non-significant intercept [36]. In order to be able Int. J. Environ. Res. Public Health 2015, 12 3082 to assess complete mediation mean-centered sums of the mediator variables were included in the regression analyses. 3. Results 3.1. Descriptive Charateristics of Participants Table 1 presents descriptive charateristics of the participants. There were no differences in age, height and leisure time index between overweight and normal-weight boys, but overweight boys had a higher weight and BMI and lower sport index compared to normal-weight peers. Table 1. Descriptive characteristics of participants. Characteristics Normal-Weight (n = 33) Overweight (n = 20) t p age (yrs) 12.8 ± 0.6 12.8 ± 0.8 0.42 0.67 height (cm) 160.0 ± 9.4 163.2 ± 8.4 −1.63 0.11 weight (kg) 45.4 ± 7.3 70.3 ± 13.7 −7.5 <0.001 BMI (kg/m2) 17.9 ± 1.5 26.2 ± 3.7 −9.5 <0.001 leisure time index * 3.0 ± 0.6 3.1 ± 0.7 −0.52 0.61 sport index * 3.3 ± 0.7 2.9 ± 0.6 2.04 0.05 Note: * 5 point scale. 3.2. Field Running Performance Primary outcomes: Running distances during the field tests are shown in Figure 1. There were no significant distraction by group interaction effects (F = 1.3, n.s.). Both overweight and normal-weight boys ran further in 12 min with music than without music (F = 5.0, p < 0.05). Overweight youngsters showed poorer performances compared to their normal-weight counterparts (F = 40.5, p < 0.001). 1445 2207,9 1617,2 2265,4 0 500 1000 1500 2000 2500 3000 normal weight overweight metre no music music Figure 1. Running distance on the Cooper test with and without attentional distraction by music in normal-weight and overweight youngsters. Int. J. Environ. Res. Public Health 2015, 12 3083 Secondary outcomes: There were no significant distraction by group interaction effects for RPE (F = 0.6, n.s). With music (11.8 ± 2.9), participants reported lower RPE compared to without music (13.1 ± 3.7) (F = 5.9, p < 0.05). Overweight youngsters reported higher rates of perceived exertion after the running tests compared to their normal-weight counterparts (14.0 ± 3.5 versus 11.2 ± 2.8) (F = 11.9, p < 0.001). When asking how participants experienced the running tests, both overweight and normal-weight youngsters found the test less annoying with music (2.1 ± 1.1) compared to without music (2.8 ± 1.4) (F = 10.1, p < 0.01). They also reported to pay less attention to bodily sensations during the test with music (2.3 ± 1.0) compared to without music (2.7 ± 1.4) (F = 4.6, p < 0.05). Overall, degree of annoyance was higher in overweight (2.3 ± 1.1) compared to normal-weight youngsters (2.8 ± 1.5) (F = 5.3, p < 0.05). There were no significant differences between conditions or groups, nor interaction effects (F < 3.8, n.s.) regarding how often they had thoughts about being able to carry on with the test. Experiences during the test with music did not differ between overweight and normal-weight boys (t < 1.6, n.s). Both groups reported that they liked the self-selected music a lot (3.9 ± 1.4), that they could listen quite good to the music during the test (3.4 ± 1.3), that it was quite pleasant to run while listening to the music (3.3 ± 1.5) and that they believed they could run further while listening to music (3.6 ± 1.5). Mediation: As attentional distraction had a significant effect on (1) RPE, (2) how annoying they experienced the running test and (3) attention given to bodily sensations during the running test, mediation analyses were performed for these three potential mediators. Only the difference in degree of annoyance mediated the difference in running distance between both conditions (β = −0.46, t = −3.34, p < 0.01). The intercept of the regression was non-significant (t = 0.7, n.s.) indicating a complete mediation. 3.3. Activity during Exercise Session Primary outcomes: In Figure 2 the total duration of the exercise session is equalled to 100%. The proportion of time spent in activities of different intensity is represented by the different coloured bars. There were no significant distraction by group interaction effects (F < 0.8, n.s). Both overweight and normal-weight boys exercised less at low and high intensity and more at moderate and very high intensity with music compared to without music (F = 6.0, p < 0.01). Overweight boys exercised a higher proportion of time at low and moderate intensity and a lower proportion of time at high and very high intensity, compared to the normal-weight boys (F = 4.0, p < 0.05). Secondary outcomes: In both overweight and normal-weight boys, RPE was lower in the condition with music (11.7 ± 3.0) than without music (12.9 ± 2.5) (F = 9.3, p < 0.01). Overweight boys reported higher RPE after the exercise session (13.0 ± 2.9) compared to their normal-weight peers (11.2 ± 3.3) (F = 5.9, p = 0.01). When asking how participants experienced the exercise sessions, both overweight and normal-weight youngsters found the session less annoying with music (2.5 ± 1.1) compared to without music (3.1 ± 1.5) (F = 7.1, p = 0.01). Both groups also reported to pay less attention to bodily sensations during the session with music (2.1 ± 1.1) compared to without music (2.4 ± 1.1) and to think less about being able to carry on with the exercise during the session with music (1.9 ± 1.2) compared to without music (2.2 ± 1.3), but these differences were not statistically significant (F = 2.1 and F = 3.2, n.s.). Although scores were generally somewhat higher in overweight compared to normal-weight boys, there were no significant group effects (F < 1.2, n.s.), nor distraction by group Int. J. Environ. Res. Public Health 2015, 12 3084 effects (F < 0.3, n.s.). Experiences during the exercise session with music did not differ between overweight and normal-weight participants (t < 1.9, n.s). Both groups reported that they quite liked the music (3.0 ± 1.4), that they could listen quite good to the music during the exercise session (2.9 ± 1.3), that it was quite pleasant to exercise while listening to the music (3.2 ± 1.5) and that they believed they could exercise more intensively while listening to music (3.4 ± 1.3). 2 4.2 3.7 26.6 38.2 44.5 51.8 38 43.7 36.1 24.3 35.2 13.9 15.8 0.2 22 0% 25% 50% 75% 100% no music music no music music normal weight overweight low moderate high very high Figure 2. Proportion of time spent in activities of different intensities during the exercise session with and without attentional distraction by music in normal-weight and overweight participants. Mediation: As attentional distraction had a significant effect on (1) RPE and (2) how annoying they experienced the exercise session, mediation analyses were performed for these two potential mediators. Only the difference in rate of perceived exertion mediated the difference in mean activity intensity during the exercise session between both conditions (β = −0.37, t = −2.6, p = 0.01). The intercept of the regression was non-significant (t = 1.8, n.s.) indicating a complete mediation. 4. Discussion The purpose of this study was to investigate the effect of attentional distraction by music on field running distances and on intensity of activity during an exercise session. We also wanted to investigate whether the effect of attentional distraction was moderated by overweight status and mediated by RPE, degree of annoyance, attention given to bodily sensations or thoughts about being able to carry on. Results clearly indicated that both overweight and normal-weight boys ran further in 12 min while listening to their favourite piece of music compared to without music. This finding is in line with previous studies in students and athletes [17] and with the previous laboratory experiment in obese youngsters [18]. Further, we also found that both overweight and normal-weight youngsters were exercising at a higher intensity during the exercise session with music compared to without music. With music, some proportion of time spent in low intensity activity was probably replaced by moderate intensity activity and some proportion of time spent in high intensity activity was replaced by very high intensity activity. To our knowledge, this is the first study to investigate the effect of attentional distraction within the context of an exercise session. Since participants were exercising at a higher Int. J. Environ. Res. Public Health 2015, 12 3085 intensity while listening to music, more energy was expended, which is the main aim of any exercise program for overweight youngsters. Previous studies suggest different ways in which music may enhance exercise performance. Szmedra and Bacharach [37] suggest that music might allow participants to relax and reduce muscle tension, thereby increasing blood flow and lactate clearance while decreasing lactate production in the working muscles and consequently having a psychobiological impact on exercise. Further, it is assumed that the exerciser has a limited attentional capacity [38,39]. During exercise individuals have access to internal sensory information (such as heart rate, breathing, pain,…) and external environmental cues (such as noise, music, other exercisers, scenery) that compete for attentional focus [40]. So, turning participants’ attention away from internal cues resulting from physiological stimuli through some distracter (such as music) during exercise, will prevent them from focusing on feelings of discomfort associated with exercise and will reduce perceived exertion. This hypothesis has been confirmed by several previous studies in adults [37,41–49] and is also in agreement with the findings of this study. With music, participants reported lower RPE and paid less attention to bodily sensations compared to without music. However, only RPE and not attention paid to bodily sensations was found to be a mediator of the effect of attentional distraction and a decrease in RPE only mediated the effect of attentional distraction on activity intensity during the exercise session, but not on field running performance. Some studies also suggest that music enhances enjoyment levels during exercise [40,43,49,50]. Music may influence emotions and mask unpleasant feelings during exercise [51]. In this study, participants found the running test and the exercise session with music less annoying compared to without music. They may associate the music with positive past experiences, they may indulge in pleasant fantasizing or may focus attention on pleasant future events which may improve emotional or affective state during exercise [51]. Synchronisation of music with exercise may have a psyching-up effect [41]. However, a decrease in feelings of annoyance only mediated the effect of attentional distraction on field running performance, but not on activity intensity during the exercise session. From the mediation analyses, we can conclude that the effect of attentional distraction works through different mechanisms depending on the type of activity. Effect of music on the running test, which is a very monotonous and less pleasant activity, is mediated by feelings of annoyance, while the effect of music on activity intensity during the exercise session, which is a more diverse and pleasant activity, is mediated by RPE. Previous research showed that the effect of attentional distraction is different in trained versus untrained athletes [52]. Elite and novice athletes employ different cognitive coping strategies to meet exercise demands [53]. Trained athletes direct their focus to internal cues during exercise in order to adapt pace and intensity to the functional information of bodily sensations. Brownley et al. [54] demonstrated that listening to fast, upbeat music during exercise is beneficial for untrained runners but counterproductive for trained runners. We hypothesized that the effect of attentional distraction would also be different in overweight compared to normal-weight youngsters. Overweight youngsters report more physical complaints while exercising [8]. Therefore we expected the effect of attentional distraction to be stronger in overweight compared to normal-weight youngsters. However, this hypothesis was not confirmed, the effect of attentional distraction was similar in both groups. Although this was not the main purpose of this study, we also found that overweight youngsters showed poorer performances on the field running tests. This is in agreement with findings of previous Int. J. Environ. Res. Public Health 2015, 12 3086 studies [21,55,56]. This poorer performance in this weight-bearing activity in overweight youngsters is probably mainly due to the fact that excess body fat adds to the mass of the body without contributing to its force producing capability, thus becoming an inert load to be moved during running. Another explanation could be that overweight youngsters avoid running because of the greater energy cost required to move the total body. In this case the poorer performance could be the consequence of a lack of experience in running. Overweight youngsters also exercised at a lower activity intensity during the exercise sessions compared to normal-weight counterparts. Although performances were generally lower in overweight youngsters, they reported higher rates of perceived exertion and they found the exercise session more annoying compared to normal-weight youngsters. Obese children generally rate perceived exertion higher than normal-weight counterparts when subjected to standardised workload on a treadmill [57,58]. These higher rates of perceived exertion and annoyance are in line with previous findings that overweight youngsters report more physical complaints and perceive less enjoyment while exercising [8]. Since this study demonstrated that attentional distraction by music increases exercise performance and intensity without increasing physical complaints or annoyance, this might be a useful strategy to increase enjoyment and exercise adherence in overweight youngsters. Future research needs to investigate whether listening to music also has a positive effect on motivation to exercise. As exercising while listening to music is perceived as less annoying or more pleasant, this might increase intrinsic motivation [59]. This was the first study to investigate the effect of attentional distraction in overweight and normal-weight youngsters in field settings. Previous studies in this field were conducted in normal-weight adults, only one laboratory study was conducted in obese youngsters [18]. However, this study has some limitations. First, the results of this study are limited to overweight boys and cannot be generalised to overweight girls. Secondly, it is possible that there were differences in sexual or skeletal maturity between normal-weight and overweight boys. Unfortunately we were not able to assess this. Thirdly, it is unknown which features of the music were critical in obtaining the distraction effect. Previous studies showed that the effect of distraction may depend on type of music, music tempo, music loudness, synchronisation with exercise or emotional significance of the music [17]. Next, the effect of distraction on exercise intensity was limited to a standardised 20 min exercise session. Further research is needed to investigate whether attentional distraction also works within a more comprehensive exercise program consisting of a variety of activities. Finally, in this study attention was distracted by music, the usefulness of other forms of distractions such as watching a video or environmental distraction (f.i. running on the beach or in a forest) needs further investigation. Our experience in overweight 6 to 12 year old children is that making activities part of an exciting adventure works well as a way of distraction. 5. Conclusions The present study showed that attentional distraction by music has a positive effect on running distances on a field endurance test and on activity intensity during an exercise session. The effect on the field endurance test was mediated by feelings of annoyance, while the effect on activity intensity during the exercise session was mediated by RPE. This indicates that the effect of attentional distraction is working through different mechanisms depending on the type of activity. Despite our hypothesis that the effect of attentional distraction would be stronger in overweight compared to Int. J. Environ. Res. Public Health 2015, 12 3087 normal-weight youngsters, the effect was similar in both groups. Motivating overweight youngsters to exercise at high enough intensity is a big challenge. Music may help overweight adolescents to enjoy physical activity and adhere to higher intensity physical activity. Further research is needed to investigate whether attentional distraction is a useful technique to increase exercise adoption and adherence in obesity prevention and treatment. Acknowledgements The authors are grateful to Cindy Stevens and Vanessa Vanhooren for their assistance in collecting the data. Author Contributions Benedicte Deforche and Ilse De Bourdeaudhuij conceived and designed the study. Benedicte Deforche coordinated the experiments analysed the data and wrote the paper. Ilse De Bourdeaudhuij critically reviewed the paper for writing and intellectual content. Conflicts of Interest The authors declare no conflict of interest. References 1. Trost, S.G.; Kerr, L.M.; Ward, D.S.; Pate, R.R. Physical activity and determinants of physical activity in obese and non-obese children. Int. J. Obes. Relat. Metab. Disord. 2001, 25, 822–829. 2. Page, A.; Cooper, A.R.; Stamatakis, E.; Foster, L.J.; Crowne, E.C.; Sabin, M.; Shield, J.P. Physical activity patterns in nonobese and obese children assessed using minute-by-minute accelerometry. Int. J. Obes. Relat. Metabol. Disord. 2005, 29, 1070–1076. 3. Fulton, J.E.; McGuire, M.T.; Caspersen, C.J.; Dietz, W.H. Interventions for weight loss and weight gain prevention among youth: Current issues. Sports Med. 2001, 31, 153–165. 4. Saris, W.H.; Blair, S.N.; van Baak, M.A.; Eaton, S.B.; Davies, P.S.; Di Pietro, L.; Fogelholm, M.; Rissanen, A.; Schoeller, D.; Swinburn, B.; et al. How much physical activity is enough to prevent unhealthy weight gain? Outcome of the IASO 1st Stock Conference and consensus statement. Obes. Rev. 2003, 4, 101–114. 5. Deci, E.L.; Ryan, R.M. Intrinsic Motivation and Self-Determination in Human Behaviour; Plenum: New York, NY, USA, 1985. 6. Stucky-Ropp, R.; DiLorenzo, T. Determinants of exercise in children. Prev. Med. 1993, 22, 880–889. 7. DiLorenzo, T.M.; Stucky-Ropp, R.C.; Vander Wal, J.S.; Gotham, H.J. Determinants of exercise among children. II. A longitudinal analysis. Prev. Med. 1998, 27, 470–477. 8. Deforche, B.I.; De Bourdeaudhuij, I.M.; Tanghe, A.P. Attitude toward physical activity in normal-weight, overweight and obese adolescents. J. Adolesc. Health 2006, 38, 560–568. 9. Shaw, K.; Gennat, H.; O’Rourke, P.; Del Mar, C. Exercise for overweight or obesity. Cochrane Database Syst. Rev. 2006, 18, doi:10.1002/14651858.CD003817.pub3. Int. J. Environ. Res. Public Health 2015, 12 3088 10. Zabinski, M.F.; Saelens, B.E.; Stein, R.I.; Hayden-Wade, H.A.; Wilfley, D.E. Overweight children’s barriers to and support for physical activity. Obes. Res. 2003, 11, 238–246. 11. De Bourdeaudhuij, I.; Lefevre, J.; Deforche, B.; Wijndaele, K.; Matton, L.; Philippaerts, R. Physical activity and psychosocial correlates in normal weight and overweight 11 to 19 year olds. Obes. Res. 2005, 13, 1097–1105. 12. Parízková, J.; Hills, A. Practical programs for weight management during the growing years. In Childhood Obesity: Prevention and Treatment; Parízková, J., Hills, A., Eds.; CRC Press: Boca Raton, FL, USA, 2001; pp. 271–305. 13. Epstein, L.H.; Koeske, R.; Wing, R.R. Adherence to exercise in obese children. J. Cardiac. Rehabil. 1984, 4, 185–195. 14. Jakicic, J.M.; Marcus, B.H.; Gallagher, K.I.; Napolitano, M.; Lang, W. Effect of exercise duration and intensity on weight loss in overweight, sedentary women: A randomized trial. JAMA 2003, 290, 1323–1330. 15. Jakicic, J.M. The role of physical activity in prevention and treatment of body weight gain in adults. J. Nutr. 2002, 132, 3826S–3829S. 16. Jeffery, R.W.; Wing, R.R.; Sherwood, N.E.; Tate, D.F. Physical activity and weight loss: Does prescribing higher physical activity goals improve outcome? Am. J. Clin. Nutr. 2003, 78, 684–689. 17. Karageorghis, C.I.; Priest, D.L. Music in the exercise domain: A review and synthesis (Part I and Part II). Int. Rev. Sport Exerc. Psychol. 2012, 5, 44–84. 18. De Bourdeaudhuij, I.; Crombez, G.; Deforche, B.; Vinaimont, F.; Debode, P.; Bouckaert, J. Effects of distraction on treadmill running time in severely obese children and adolescents. Int. J. Obes. Relat. Metab. Disord. 2002, 26, 1023–1029. 19. Cole, T.J.; Bellizzi, M.C.; Flegal, K.M.; Dietz, W.H. Establishing a standard definition for child overweight and obesity worldwide: International survey. BMJ 2000, 320, 1240–1243. 20. Baecke, J.A.; Burema, J.; Frijters, J.E. A short questionnaire for the measurement of habitual physical activity in epidemiological studies. Am. J. Clin. Nutr. 1982, 36, 936–942. 21. Deforche, B.; Lefevre, J.; De Bourdeaudhuij, I.; Hills, A.P.; Duquet, W.; Bouckaert, J. Physical fitness and physical activity in obese and nonobese Flemish youth. Obes. Res. 2003, 11, 434–441. 22. Pols, M.A.; Peeters, P.H.; Buenodemesquita, H.B.; Ocké, M.C.; Wentink, C.A.; Kemper, H.C.; Collette, H.J. Validity and repeatability of a modified Baecke questionnaire on physical activity. Int. J. Epidemiol. 1995, 24, 381–388. 23. Phillippaerts, R.M.; Westerterp, K.R.; Lefevre, J. Doubly labelled water validation of three physical activity questionnaires. Int. J. Sports Med. 1999, 20, 284–289. 24. Cooper, K. A means of assessing maximal oxygen intake: Correlation between field testing and treadmill testing. JAMA 1968, 203, 201–204. 25. Mccutcheon, M.C.; Sticha, S.A.; Giese, M.D.; Nagle, F.J. A further analysis of the 12-minute run prediction of maximal aerobic power. Res. Q. 1990, 61, 280–283. 26. McNaughton, L.; Hall, P.; Cooley, D. Validation of several methods estimating maximal oxygen uptake in young men. Percept. Mot. Skills 1998, 87, 575–584. 27. Dyrlund, A.K.; Wininger, S.R. The effects of music preference and exercise intensity on psychological variables. J. Music Ther. 2008, 45, 114–134. Int. J. Environ. Res. Public Health 2015, 12 3089 28. Freedson, P.S.; Melanson, E.; Sirard, J. Calibration of the Computer Science and Applications, Inc. accelerometer. Med. Sci. Sports Exerc. 1998, 30, 777–781. 29. Janz, K.F. Validation of the CSA accelerometer for assessing children’s physical activity. Med. Sci. Sports Exerc. 1994, 26, 369–375. 30. Trost, S.G.; Ward, D.S.; Moorehead, S.M.; Watson, P.D.; Riner, W.; Burke, J.R. Validity of the computer science and applications (CSA) activity monitor in children. Med. Sci. Sports Exerc. 1998, 30, 629–633. 31. Puyau, M.R.; Adolph, A.L.; Vohra, F.A.; Butte, N.F. Validation and calibration of physical activity monitors in children. Obes. Res. 2002, 10, 150–157. 32. Borg, G.A. Psychophysical bases of perceived exertion. Med. Sci. Sports Exerc. 1982, 14, 377–381. 33. Lamb, K.L. Children’s ratings of effort during cycle ergometry: An examination of the validity of two effort scales. Pediatr. Exerc. Sci. 1995, 7, 407–421. 34. Lamb, K.L. Exercise regulation during cycle ergometry using the children’s effort rating table (CERT) and rating of perceived exertion (RPE) scales. Pediatr. Exerc. Sci. 1996, 8, 337–350. 35. Ward, D.S.; Bar-Or, O. Use of the Borg scale in exercise prescription for overweight youth. Can. J. Sport Sci. 1990, 15, 120–125. 36. Judd, C.M.; Kenny, D.A.; McClelland, G.H. Estimating and testing mediation and moderation in within-subject designs. Psychol. Methods 2001, 6, 115–134. 37. Szmedra, L.; Bacharach, D.W. Effect of music on perceived exertion, plasma lactate, norepinephrine and cardiovascular hemodynamics during treadmill running. Int. J. Sports Med. 1998, 19, 32–37. 38. Pennebaker, J.W. The Psychology of Physical Symptoms; Springer: New York, NY, USA, 1982. 39. Abernethy, B. Attention. In Handbook of Sport Psychology; Singer, R.N., Hausenblas, H.A., Janelle, C.M., Eds.; Wiley: New York, NY, USA, 2001; pp. 53–85. 40. Rejeski, W.J. Perceived exertion: An active or passive process? J. Sport Psychol. 1985, 7, 371–378. 41. Anshel, M.H.; Marisi, D.Q. Effect of music and rhythm on physical performance. Res. Q. 1978, 49, 109–113. 42. Pennebaker, J.W.; Lightner, J.M. Competition of internal and external information in an exercise setting. J. Pers. Soc. Psychol. 1980, 39, 165–174. 43. Karageorghis, C.I.; Terry, P.C. The psychophysical effects of music in sport and exercise: A review. J. Sport Behav. 1997, 20, 54–68. 44. Potteiger, J.A.; Schroeder, J.M.; Goff, K.L. Influence of music on ratings of perceived exertion during 20 minutes of moderate intensity exercise. Percept. Mot. Skills 2000, 91, 848–854. 45. Nethery, V.M. Competition between internal and external sources of information during exercise: Influence on RPE and the impact of the exercise load. J. Sports Med. Phys. Fitness 2002, 42, 172–178. 46. Bharani, A.; Sahu, A.; Mathew, V. Effect of passive distraction on treadmill exercise test performance in healthy males using music. Int. J. Cardiol. 2004, 97, 305–306. 47. Yamashita, S.; Iwai, K.; Akimoto, T.; Sugawara, J.; Kono, I. Effects of music during exercise on RPE, heart rate and the autonomic nervous system. J. Sports Med. Phys. Fitness 2006, 46, 425–430. Int. J. Environ. Res. Public Health 2015, 12 3090 48. Karageorghis, C.I.; Mouzourides, D.; Priest, D.L.; Sasso, T.; Morrish, D.; Whalley, C. Psychophysical and ergogenic effects of synchronous music during treadmill walking. J. Sport Exerc. Psychol. 2009, 31, 18–36. 49. Miller, T.; Swank, A.M.; Manire, J.T.; Robertson, R.J.; Wheeler, B. Effect of music and dialog on perception of exertion, enjoyment, and metabolic responses during exercise. Int. J. Fitness 2010, 6, 45–52. 50. Edworthy, J.; Waring, H. The effects of music tempo and loudness level on treadmill exercise. Ergonomics 2006, 49, 1597–1610. 51. Boutcher, S.H.; Trenske, M. The effects of sensory deprivation and music on perceived exertion and affect during exercise. J. Sport Exerc. Psychol. 1990, 12, 167–176. 52. Mohammadzadeh, H.; Tartibiyan, B.; Ahmadai, A. The effects of music on the perceived exertion rate and performances of trained and untrained individuals during progressive exercise. Facta. Universit. Ser. Phys. Educ. Sport 2008, 6, 67–74. 53. Morgan, W.P. Psychological characterizations of the elite distance runner. Ann. NY Acad. Sci. 1977, 301, 383–403. 54. Brownley, K.A.; McMurray, R.G.; Hackney, A.C. Effects of music on physiological and affective responses to graded treadmill exercise in trained and untrained runners. Int. J. Psycholphysiol. 1995, 19, 193–201. 55. Beunen, G.; Malina, R.M.; Ostyn, M.; Renson, R.; Simons, J.; Van Gerven, D. Fatness, growth and motor fitness of Belgian boys 12 through 20 years of age. Hum. Biol. 1983, 55, 599–613. 56. Minck, M.R.; Ruiter, L.M.; Van Mechelen, W.; Kemper, H.C.G.; Twisk, J.W.R. Physical fitness, body fatness, and physical activity: The Amsterdam Growth Study. Am. J. Hum. Biol. 2000, 12, 593–599. 57. Marinov, B.; Kostianev, S.; Turnovska, T. Ventilatory efficiency and rate of perceived exertion in obese and non-obese children performing standardized exercise. Clin. Physiol. Funct. Imaging 2002, 22, 254–260. 58. Marinov, B.; Kostianev, S. Exercise performance and oxygen uptake efficiency slope in obese children performing standardized exercise. Acta Physiol. Pharmacol. Bulg. 2003, 27, 59–64. 59. Deforche, B.; Haerens, L.; De Bourdeaudhuij, I. How to make overweight children exercise and follow the recommendations. Int. J. Pediatr. Obes. 2011, 6 (Suppl. 1), 35–41. © 2015 by the authors; licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution license (http://creativecommons.org/licenses/by/4.0/).

Attentional distraction during exercise in overweight and normal-weight boys.

03-13-2015

Deforche, Benedicte,De Bourdeaudhuij, Ilse

eng

PMC10086059

1 Vol.:(0123456789) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports Exogenous lactate augments exercise‑induced improvement in memory but not in hippocampal neurogenesis Deunsol Hwang 1,2,4, Jisu Kim 1,2,4, Sunghwan Kyun 1,2, Inkwon Jang 1,2, Taeho Kim 1,2, Hun‑Young Park 1,2 & Kiwon Lim 1,2,3* Adult hippocampal neurogenesis (AHN), the lifelong process of formation of new neurons in the mammalian brain, plays an important role in learning and memory. Exercise is an effective enhancer of AHN; however, the molecular mediators of exercise‑induced AHN are unknown. Recently, lactate was considered as an important mediator of exercise‑induced AHN. Therefore, we hypothesized that exercise with lactate intake could augment exercise‑induced AHN. This study was conducted for 5 weeks with 7‑week‑old ICR male mice that performed mild‑intensity exercise (just below lactate threshold, 55–60%VO2max) with or without oral administration of lactate 5 days/week. Cell proliferation, neuronal differentiation, neurogenesis‑relevant factors, reference and retention memory, and spatial working memory were evaluated at the end of the experiment. The results showed that AHN was enhanced by lactate intake, but exercise‑induced AHN was not augmented by exercise with lactate intake. Nevertheless, exercise‑induced improvement in reference and retention memory was augmented by exercise with lactate intake. And spatial working memory was promoted by the co‑treatment, also protein expression of hippocampal FNDC5, BDNF, PGC1α, and MCT2 were elevated by the co‑treatment. Therefore, our findings suggest that lactate has a potential to be developed as a novel supplement that improves the positive effects of exercise on the hippocampus and its cognitive function. Adult hippocampal neurogenesis (AHN) refers to the lifelong process of formation of new neurons in the den- tate gyrus (DG) of the adult mammalian brain. It plays an important role in learning and memory; thus, it is associated with cognitive deficits in neurodegenerative conditions. An impairment of AHN has been shown to cause cognitive decline1. Therefore, promoting AHN is considered a substantial way to prevent or ameliorate cognitive deficits, which is important for improving the quality of life because of global acceleration of risk factors for neurodegenerative conditions, such as aging, Alzheimer’s disease (AD), obesity, and physical inactivity2–4. Exercise effectively enhances AHN. Studies have shown that regular exercise rescues AHN impairment caused by aging5 and chronic stress6. Deteriorated AHN in an AD model was also ameliorated by long-term exercise7. In addition to these studies on rodents with neurodegenerative conditions, AHN was enhanced by endurance exercise training, even in normal adult rodents8–10. Although the effect of exercise on AHN is well-demonstrated, what molecule primarily regulates exercise-induced AHN and triggers neurogenesis-relevant factors such as brain derived neurotrophic factor (BDNF) and fibronectin type III domain containing 5 (FNDC5), important mediators of exercise-induced AHN11–13, are not yet known. Lactate, which are known as just byproduct of glycolysis and a primary factor in fatigue during exercise, has been highlighted as a signaling molecule in the brain14. In mice, pharmacological disruption of lactate transport to hippocampal neurons impaired memory formation, and this impairment was fully reversed by intrahippocam- pal injection of lactate, but not glucose15. This result indicates that lactate acts as a signaling molecule rather than an energy substrate in the brain during memory formation. OPEN 1Laboratory of Exercise and Nutrition, Department of Sports Medicine and Science in Graduate School, Konkuk University, Seoul, Republic of Korea. 2Physical Activity and Performance Institute (PAPI), Konkuk University, Seoul, Republic of Korea. 3Department of Physical Education, Konkuk University, Seoul, Republic of Korea. 4These authors contributed equally: Deunsol Hwang and Jisu Kim. *email: exercise@konkuk.ac.kr 2 Vol:.(1234567890) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ Recently, lactate has been considered a primary mediator for exercise-induced AHN16,17 because the effect of lactate parallels the beneficial effects of exercise on the hippocampus18,19. Previous in vitro studies showed that lactate treatment upregulated the protein expression of BDNF in hippocampal cells20 and enhanced the prolifera- tion of neural precursor cells obtained from the hippocampus of adult mice by shortening the generation time of neural precursor cells21. Similarly, in vivo studies using an inhibitor of lactate transporters (monocarboxylate transporters; MCTs) increased mRNA expression levels of hippocampal BDNF through exercise, but this increase was abolished by inhibition of MCT1/2 (MCT1: transporter for efflux of lactate, MCT2: transporter for influx of lactate only in neurons)22. Further, the inhibition of MCT1/2 impeded the beneficial effect of exercise on learn- ing and memory. These results indicate that lactate is involved in AHN. Indeed, long-term injection of lactate significantly elevated AHN, but the elevation was blocked by co-treatment with an MCT2 inhibitor23, and lactate injection also enhanced the protein expression of hippocampal BDNF and FNDC522 in rodents. Collectively, these observations indicate that lactate promotes AHN and may be a primary mediator of exercise-induced AHN. Therefore, we hypothesized that exercise with lactate intake can augment exercise- induced AHN. Specifically, exogenous lactate can cross the blood–brain barrier via MCT124 and subsequently elevate both the hippocampal extracellular lactate concentration25 and the hippocampal lactate concentration via MCT222. To our knowledge, this is the first study to investigate the effect of co-treatment with exercise and lactate on AHN as well as to show the effect of lactate on AHN via oral administration. Results Blood lactate concentration was sufficiently elevated by oral administration of lactate and not by mild exercise. First, we conducted a pilot experiment to know whether either lactate or exercise intervention could satisfy our criterion with 6-week-old male ICR mice. Oral administration of 3 g/kg lactate significantly elevated blood lactate concentration to 7.66 ± 1.43 mM, 15 min after administration [Fig. 1A; base- line vs. 15 min in the lactate intake group (LAC): p = 0.001, vehicle group (VEH) vs. LAC at 15 min: p = 0.001]. Circulating lactate transports into the central nervous system26,27 via MCTs28, and MCTs are particularly abun- dant in the hippocampus compared to other regions of brain28. Therefore, we assert that oral administration of 3 g/kg lactate, an applied treatment, can affect the hippocampus. We set the exercise intensity as “mild” (just below lactate threshold, 55%–60% VO2max)29–31, considering that an extremely low or high exercise intensity would be ineffective on the AHN8,25,32. To ensure that the exercise intensity is equally maintained throughout the entire experiment (equalization of relative exercise intensity), we had gradually increased the treadmill speed and/or exercise duration over time (increasing absolute exercise intensity; Fig. 1B), based on our previous studies33–35. The increase in absolute exercise intensity is required because chronic exercise enhances the exercise ability (occurrence of an exercise adaptation), and the enhanced exercise ability subsequently leads to decrease in the relative exercise intensity if the absolute exercise intensity is not increased. Figure 1. Blood lactate concentration was sufficiently elevated by oral administration of lactate and not by mild exercise. (A) Blood lactate concentration after oral administration of 3 g/kg lactate over time (VEH, saline administration group, n = 4; LAC, lactate administration group, n = 6). Data were analyzed using two- way repeated ANOVA with Student’s t-test between groups for post hoc test and paired t-test for comparison in baseline and a timepoint within group. (B) Blood lactate concentration immediately after mild-intensity exercise training over period (n = 8 per group; SED, sedentary group; EXE, exercise training group). For comparison within EXE, one-way repeated ANOVA was used, and comparison between SED and a timepoint of EXE was performed using independent Student’s t-test. D1, experimental day 1; D15, experimental day 15; D29, experimental day 29; #p < 0.05, baseline vs time point within group; *p < 0.05, VEH vs LAC at the same time point except baseline; $p = 0.01, VEH vs LAC analyzed by Mann–Whitney test. Data are presented as the mean ± standard deviation. 3 Vol.:(0123456789) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ In exercise training group, blood lactate levels measured immediately after exercise on experimental days 1 (D1), 15 (D15), and 29 (D29) did not differ from the baseline lactate level (Fig. 1B). Therefore, our exercise protocol was verified to have an intensity below the lactate threshold. Of note, the unchanged level of blood lactate does not mean that lactates are not produced during the exercise; this results from an increase in both production and consumption of lactate in balance when exercise intensity is set below the lactate threshold. Exercise with lactate administration did not augment exercise‑induced AHN, but lactate administration promoted AHN. To validate the effect of exercise with lactate intake on AHN, we evalu- ated proliferation and neuronal differentiation of the neurogenic pool in the DG. After 5 weeks of exercise training and/or lactate administration five times per week in 7-week-old male ICR mice (Fig. 2A), the number of Ki67+ (cell proliferation marker) and doublecortin (DCX, immature neuronal marker)+ cells in the exercise without lactate group (EXE + VEH) was significantly higher than that in the sedentary without lactate group (VEH) (Fig. 3A–C; p = 0.001 and p = 0.009, respectively). The number of Ki67+ cells in the sedentary with lactate intake group (LAC) indicated a slight increase compared to that in the VEH group (Fig. 3A; p = 0.064), and the number of DCX+ cells in the LAC significantly increased compared to VEH (Fig. 3B; p = 0.03). However, there was no difference between EXE + VEH and the exercise with lactate intake group (EXE + LAC) in the number of Ki67+ and DCX+ cells (Fig. 3A–C). Therefore, these results indicate that exercise with lactate intake did not aug- ment exercise-induced AHN, although both exercise and lactate promoted AHN independently. Exercise with lactate administration augmented exercise‑induced improvement in reference and retention memory. To examine the effect of exercise with lactate intake on reference and retention memory, another cohort of mice was used (Fig. 2B) and performed an eight-arm radial arm maze (RAM). Of note, the entire results are presented in Fig. 4A, and for readability of indications of post hoc test we subdi- vide the results and presented in Fig. 4Ba–d (comparison of changes over time within the same group) and in Fig. 4Ca–f (comparison of difference among groups within the same day). In the learning phase (day 1–5) of the task, the appearance of learning curves was different among groups. On day 1, the performance of RAM task did not differ among groups [Fig. 4Ca; two-way ANOVA: lactate (p = 0.192), exercise (p = 0.590) and interaction (p = 0.319)]. On day 2, however, errors ratio was reduced only in EXE + LAC compared to day 1 (Fig. 4Bd; p = 0.001). EXE + LAC showed significant improvement in reference memory compared to other groups also (Fig. 4Cb; EXE + VEH vs. EXE + LAC: p = 0.002; LAC vs. EXE + LAC: p = 0.001). In turn, the group showed the second-best performance in reference memory was EXE + VEH (Fig. 4Bc; day 1 vs. day 4: p = 0.023) and LAC (Fig. 4Bb; day 1 vs. day 4: p = 0.065, day 1 vs. day 5: p = 0.056) (Fig. 4Cd; VEH vs. LAC: p = 0.001, VEH vs. EXE + VEH: p = 0.001), and the next was VEH (Fig. 4Ba; day 1 vs. day 5: p = 0.022). Finally, on day 5, there was no difference in reference memory among groups [Fig. 4Ce; two-way ANOVA: lactate (p = 0.712), exercise (p = 0.612) and interaction (p = 0.619)]. Figure 2. The schematic representation of experimental procedure. The experiments started with 7-week-old male ICR mice. In case of EXE + LAC, mice were administrated lactate immediately after exercise. This study comprised two independent experiments except the pilot test: (A) one mainly for biochemical analysis (n = 9 per group) and (B) the other for behavioral analysis (n = 8 per group). VEH sedentary without lactate, LAC sedentary with lactate, EXE + VEH exercise training without lactate, EXE + LAC exercise training with lactate, IHC immunohistochemistry, IB immunoblotting, RAM radial arm maze. The mouse icon is “Created with BioRender.com”. 4 Vol:.(1234567890) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ In the retention memory trial of the task (day 11), the improved reference memory was significantly retained only in EXE + LAC (Fig. 4Bd; day 1 vs. day 11: p = 0.049). However, there was no difference in retention memory among groups although significant main effect of exercise was confirmed [Fig. 4Cf; two-way ANOVA: lactate (p = 0.483), exercise (p = 0.02) and interaction (p = 0.426)]. Collectively, these results indicate that exercise with lactate administration augmented exercise-induced improvement in reference and retention memory. Exercise with lactate administration promoted spatial working memory. Also, we performed Y-maze test to measure spatial working memory. The ability to alternate between two arms of the maze requires mice to know which arms have already been visited. Therefore, alternation behavior can be regarded as a meas- ure of spatial working memory, which is a hippocampus-related cognitive function. The total number of arm entries did not differ among the groups (Fig. 5A), which indicated that there was likely no bias in the alternation that could exist when the total number of arms entered was unequal. Nevertheless, alternation in EXE + LAC was significantly higher than that in EXE + VEH and LAC (p = 0.004 and p = 0.009, respectively; Fig. 5B), and there was no difference between VEH and either LAC or EXE + VEH (Fig. 5B). As a result, exercise with lactate administration promoted spatial working memory. Exercise with lactate administration effectively enhanced hippocampal FNDC5, BDNF, PGC1α, and MCT2 protein expression. To understand the molecular changes resulting from exercise with lactate intake, we investigated the expression of the following proteins relevant to AHN in the context of exercise and lactate effects: FNDC5, BDNF, peroxisome proliferator-activated receptor gamma, coactivator 1 alpha (PGC1α), MCT2, and MCT1. Figure 3. Exercise with lactate administration did not augment exercise-induced adult hippocampal neurogenesis, but lactate administration promoted adult hippocampal neurogenesis. (A) Quantification of Ki67-positive cell and (B) of DCX-positive cell in subgranular zone of dentate gyrus of mice. (C) The represent image of immunohistochemistry. Arrow indicates a represent positive cell. Scale bar: 100 μm. Two-way ANOVA was performed, and independent Student’s t-test was used for post hoc test (n = 4–5 per group). VEH sedentary without lactate, LAC sedentary with lactate, EXE + VEH exercise without lactate, EXE + LAC exercise with lactate; *p < 0.05, **p < 0.01, and ***p ≤ 0.005. Data are presented as the mean ± standard deviation. 5 Vol.:(0123456789) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ Exercise significantly affected the expression of all relevant proteins [Fig. 6A–D; two-way ANOVA; FNDC5 (p = 0.001), BDNF (p = 0.003), PGC1α (p = 0.001); Fig. 7A; MCT2 (p = 0.028)], except MCT1 (Fig. 7B,C; two- way ANOVA; p = 0.079). Lactate intake did not affect the hippocampal FNDC5, BDNF (Fig. 6A,B), and MCT2 (Fig. 7A) protein expression. However, the hippocampal PGC1α protein expression in LAC tended to be higher than that in VEH (Fig. 6C; p = 0.08). Notably, the hippocampal FNDC5 protein expression in EXE + LAC was significantly elevated compared to EXE + VEH (Fig. 6A; p = 0.007), and the hippocampal BDNF, PGC1α, and MCT2 protein expression in EXE + LAC was higher than that in EXE + VEH by 20%, 9%, and 19%, respectively, although the difference was not statistically significant (Figs. 6B,C and 7A). Exercise, lactate, and co‑treatment did not affect hippocampal VEGFA or HCAR1 protein expression. To further elucidate the proteins relevant in the context of exercise- and lactate-mediated effects on AHN and cognitive behavior in EXE + LAC, we investigated the angiogenesis-related proteins vascu- lar endothelial growth factor A (VEGFA) and hydroxycarboxylic acid receptor 1 (HCAR1). We did not find any difference in hippocampal VEGFA and HCAR1 protein expression among the groups (Fig. 8A–C). Figure 4. Exercise with lactate administration augmented exercise-induced improvement in reference and retention memory. The test was conducted using eight-arm radial arm maze (RAM). Day 1 to 5 is learning phase of RAM task and Day 11 is retention memory trial of RAM task. (A) The entire results of RAM task. Two-way repeated ANOVA was performed. In order to improve readability of indications of post hoc results, (B) the results of comparison of changes over time within the same group (paired t-test was used for post hoc test) and (C) the results of comparison of difference among groups within the same day (independent Student’s t-test was used for post hoc test) are separately presented. VEH sedentary without lactate, LAC sedentary with lactate, EXE + VEH exercise without lactate, EXE + LAC exercise with lactate; n = 8 per group, *p < 0.05, **p < 0.01, and ***p ≤ 0.005; #p < 0.05, vs day 1 within group. Data are presented as the mean ± standard deviation. 6 Vol:.(1234567890) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ Discussion Lactate has been implicated as a major molecule in mediating exercise-induced AHN. Accordingly, we reasoned that exogenous lactate intake could partially mimic the effect of exercise on AHN. Therefore, we hypothesized that exercise with oral intake of lactate augments exercise-induced AHN. To validate this assumption, we exam- ined the effect of exercise with lactate intake on the proliferation and neuronal differentiation of the neurogenic Figure 5. Exercise with lactate administration promoted spatial working memory. The test was conducted using Y-maze. (A) The total number of arm entries and (B) spontaneous alternation behavior (spatial working memory). Two-way ANOVA was performed, and independent Student’s t-test was used for post hoc test (n = 14 per group). **p < 0.01 and ***p ≤ 0.005. Data are presented as the mean ± standard deviation. Figure 6. Exercise with lactate administration effectively enhanced the hippocampal molecules relevant to exercise-induced adult hippocampal neurogenesis. Level of protein expression of hippocampal (A) FNDC5, (B) BDNF, and (C) PGC1α in mice. (D) The represent image of western blot. Two-way ANOVA was performed, and independent Student’s t-test was used for post hoc test (n = 4 per group). The original blots are presented in Supplementary Figs. 1 and 2. VEH sedentary without lactate, LAC sedentary with lactate, EXE + VEH exercise without lactate, EXE + LAC exercise with lactate, FNDC5 fibronectin type III domain-containing protein 5, BDNF brain derived neurotrophic factor, PGC1α peroxisome proliferator-activated receptor gamma coactivator 1-alpha; *p < 0.05, **p < 0.01, and ***p ≤ 0.005. Data are presented as box plot. 7 Vol.:(0123456789) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ pool in the DG, reference and retention memory, spatial working memory, and the expression of relevant hip- pocampal proteins in mice. MCTs presents in endothelial cells of blood–brain barrier28, which indicates that circulating lactate can transport into the central nervous system. Indeed, circulating lactate transports into the central nervous system and is utilized in the brain both in human26 and rodents27. Furthermore, MCTs are particularly abundant in the hippocampus compared to other regions of brain28, which indicates that the hippocampus is susceptible to the circulating lactate. Indeed, in the previous study measuring the hippocampus extracellular lactate concentration by microdialysis after intraperitoneal injection of lactate, time course changes in the hippocampal extracellular lactate concentration occurred simultaneously with the time course changes in blood lactate concentration, and the elevation in the hippocampal extracellular lactate level is blocked by MCT inhibitor25. Elevation in blood lactate concentration to more than 6 mM was sufficient to result in a significant increase in the hippocampal lactate levels22,25. Therefore, we assert that oral administration of 3 g/kg lactate in our study is enough to increase the hippocampal lactate concentration; thus, our applied treatment can affect the hippocampus. A previous study showed that intraperitoneal injection of lactate promotes AHN in an MCT2-dependent manner23. Similarly, in the present study, we observed that lactate enhances proliferation and neuronal differentia- tion of the neurogenic pool in the DG (Fig. 3). However, lactate did not augment exercise-induced AHN (Fig. 3). The lack of an augmented effect of exercise with lactate on AHN may be related to the physiologically limited capacity of neurogenesis, i.e., an upper limit to the increase in neurogenesis. In several typical rodent strains, an approximate 1.5- to 2.0-fold increase in neurogenesis by exercise compared to baseline seems to be the upper limit8,32,36,37. In previous studies, the exercise protocol was usually set at 10–15 m/min (velocity) for 40–60 min (duration), 5 days/week (frequency) over a period of 4 weeks. Considering these observations, the exercise pro- tocol we used (15–25 m/min for 40–50 min, 5 days/week for 5 weeks) should induce a sufficient increase in AHN because the amount of exercise in our study is higher compared to the previous studies. Indeed, in the current study exercise increased the number of Ki67- and DCX-positive cells in the DG by 2- and 1.5-fold, respectively, compared to the control condition (Fig. 3). Thus, there may be little (or no) opportunity for neurogenesis induced Figure 7. Exercise with lactate administration effectively enhanced the hippocampal MCT2 protein expression but not MCT1. Level of protein expression of hippocampal (A) MCT2 and (B) MCT1 in mice. (C) The represent image of western blot. Two-way ANOVA was performed, and independent Student’s t-test was used for post hoc test (n = 4 per group). The original blots are presented in Supplementary Fig. 3. VEH sedentary without lactate, LAC sedentary with lactate, EXE + VEH exercise without lactate, EXE + LAC exercise with lactate, MCT1/2 monocarboxylate transporter 1/2; *p < 0.05. Data are presented as box plot. Figure 8. Exercise, lactate, and co-treatment did not affect hippocampal VEGFA and HCAR1 protein expression. Level of protein expression of hippocampal (A) VEGFA and (B) HCAR1. (C) The represent image of western blot. Two-way ANOVA was performed, and independent Student’s t-test was used for post hoc test (n = 4 per group). The original blots are presented in Supplementary Figs. 4 and 5. VEH sedentary without lactate, LAC sedentary with lactate, EXE + VEH exercise without lactate, EXE + LAC exercise with lactate, VEFGA vascular endothelial growth factor A, HCAR1 hydroxycarboxylic acid receptor 1. Data are presented as box plot. 8 Vol:.(1234567890) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ by lactate, as exercise had a substantial effect on neurogenesis. This interpretation is partially supported by previ- ous studies using mice with a lower than normal level of neurogenesis, where mice with abnormal conditions of neurogenesis showed more potential for increased neurogenesis compared to mice in normal conditions. Indeed, in previous studies, AHN impeded by a chronic stressful environment6 and aging38 was significantly increased to a greater extent by co-treatment of exercise and a supplement than by exercise alone; additionally, it is noteworthy that the exercise intensity used was lower than that used in the current study. Therefore, we suggest that further investigation focusing on level of exercise intensity and/or pathological models is required to validate the augmented effect of exercise with lactate on AHN. While we were not able to observe an augmented effect of exercise with lactate on AHN, we found that exercise-induced improvement in reference and retention memory were augmented by lactate (Fig. 4). And spatial working memory was promoted by co-treatment (Fig. 5), also hippocampal FNDC5, BDNF, PGC1α (Fig. 6A), and MCT2 (Fig. 7A) protein expressions were effectively enhanced by co-treatment. We speculate that this additive effect of co-treatment on memory and the relevant factors may be explained by neuronal plasticity rather than neurogenesis alone. Neuronal plasticity occurs at the cellular level during learning and memory18. A previous study identifying the action of lactate on neuronal plasticity showed that lactate promoted the expression of synaptic plasticity-related genes (Arc, c-Fos, and Zif268) both in primary neuronal cultures of the mouse neocortex and in vivo39. This result was corroborated by transcriptome analysis identified that expression of 15 neuronal plasticity-related genes was upregulated by lactate in primary cultures of cortical neurons40. Furthermore, the beneficial effect of lactate on neuronal plasticity39 and memory15 was negated by MCT inhibition. This finding suggests the promoting effect of lactate on neuronal plasticity and the importance of MCTs when lactate acts as a promoter of neuronal plasticity. Brain FNDC5 plays a significant role in neuronal plasticity, especially in the context of exercise41,42. Hip- pocampal neuronal function was impaired by the knockdown of brain Fndc5 in wild-type mice, and impaired hippocampal neuronal plasticity in an AD model was rescued by boosting brain FNDC5. Finally, the study showed that exercise-induced improvement in hippocampal neuronal function was blunted by the downregula- tion of brain FNDC5 expression; these results indicate the role of brain FNDC5 as a key mediator of exercise on neuronal plasticity and memory41. Hippocampal BDNF is an important molecule of effects of exercise on many aspects of both AHN19 and neuronal plasticity18. Hippocampal BDNF is regulated via several pathways; however, in the context of exercise, it has been reported that a PGC1α/FNDC5-dependent mechanism is an important way to regulate hippocampal BDNF42. This previous study showed that the knockdown of PGC1α significantly downregulated hippocampal Fndc5 gene expression both in vitro and in vivo. Hippocampal Bdnf gene expression was significantly upregu- lated by forced expression of FNDC5 both in vitro and in vivo. Notably, the expression of important neuronal plasticity-related genes (Arc, c-Fos, Npas4, Zif268) was also sharply increased by forced expression of FNDC5 both in primary cultures of hippocampal neurons and in the hippocampus of wild-type mice. These results demon- strate that hippocampal BDNF affects hippocampal neuronal plasticity in a PGC1α/FNDC5-dependent manner. Consequently, the improvement in memory by exercise with lactate intake (Figs. 4 and 5) may have resulted from enhanced neuronal plasticity due to augmented hippocampal FNDC5 protein expression and small increase in hippocampal BDNF, PGC1α, and MCT2 protein expression (Figs. 6 and 7A). Additionally, HCAR1 is a lactate receptor abundant around cerebral blood vessels (but sparsely expressed in skeletal muscles) and is involved in stimulating the exercise-induced cerebral VEGFA expression and angiogenesis43. It has been sug- gested that the beneficial effect of exercise on brain functions partially results from improved cerebral perfusion via angiogenesis44. Thus, our speculation that the improvement in memory by exercise with lactate intake may results from the enhanced neuronal plasticity is partially supported by the finding that there was no change in hippocampal VEGFA and HCAR1 protein expression either by exercise or lactate (Fig. 8). In summary, our study shows that exercise with lactate did not augment exercise-induced AHN. However, the physiologically limited capacity for neurogenesis in normal mice coupled with the robust neurogenesis response to exercise may have occluded the potential contribution of lactate. Nevertheless, exercise with lactate augmented exercise-induced improvement in reference and retention memory. Also, spatial working memory was promoted by co-treatment. Consistent with this result, hippocampal FNDC5 protein expression was significantly augmented, and hippocampal BDNF, PGC1α, and MCT2 protein expression were slightly more upregulated by exercise with lactate compared to exercise (the changes were not statistically significant). These positive changes are likely to result in enhancing hippocampal neuronal plasticity and, subsequently, may induce the improve- ment in memory. Herein, we partially evaluated AHN (proliferation and neuronal differentiation) and did not explore neuronal maturation, i.e., the final phase of neurogenesis. Investigating neuronal maturation requires tracing methods such as injection of bromodeoxyuridine. Considering that AHN is vulnerable to stress, we decided not to use this method to avoid confounds from excessive stress that can occur when both oral and intraperitoneal injections are used. Therefore, an experiment that specifically evaluates neuronal maturation phase is needed to establish a better understanding of the complete effect of lactate on AHN. Lactate is a highly relevant signaling molecule that regulates brain functions14. However, it is unknown whether lactate is a primary mediator of exercise-induced AHN. Furthermore, it is unknown whether lactate acts equally even under exercise conditions that differ physiologically from resting conditions in the brain. So far there is no critical evidence that exercise-induced AHN is mediated by lactate. This aspect may be a major limitation of the present study. Therefore, we plan to conduct further studies to find direct evidence to reveal the relationship between lactate and exercise-induced AHN. In conclusion, to our knowledge, this is the first study to investigate the effect of co-treatment of exercise and lactate on AHN. We demonstrated the effect of lactate on AHN via oral administration, i.e., through an applied 9 Vol.:(0123456789) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ approach rather than an invasive approach. Our results suggest that lactate has a potential to be developed as a novel supplement that improves the positive effects of exercise on the hippocampus and its cognitive function. Methods Ethical approval. The animal study was reviewed and approved by the Konkuk University Institutional Animal Care and Use Committee (No. KU19149). All methods were performed in accordance with the relevant guidelines and regulations. The study was carried out in compliance with the ARRIVE guidelines: https:// arriv eguid elines. org/. Furthermore, efforts were made to minimize discomfort and stressful situations. Animals. Before starting the experiment, 6-week-old male ICR mice (33.2 ± 1.4 g, Orient Bio Inc., Seong- nam, Republic of Korea) were habituated to the laboratory environment for at least a week. All mice were housed in standard transparent plastic cages under a controlled temperature at 23–25 °C with 40%–50% humidity and a 12-h light/dark cycle (lights on: 07:00–19:00). Standard chow diet (Orient Bio Inc., Seongnam, Republic of Korea) and water were provided ad libitum. Pilot experiment. In the pilot experiment with 6-week-old male ICR mice distinct from a set of mice used in the main experiment (cohort 1 and cohort 2), blood lactate concentration was measured using a lactate ana- lyzer (LT-1730, Lactate Pro 2, ARKRAY, Kyoto, Japan) after a single oral administration of lactate over time or immediately after exercise from tail vein blood at D1, D15, and D29 (Fig. 1). Experimental design. This study comprised two independent experiments except the pilot test. Mice of experiment 1 were mainly for biochemical analysis (Fig. 2A, n = 9 per group) and mice of experiment 2 were for behavioral analysis (Fig. 2B, n = 8 per group). Mice were randomly divided into four groups: VEH, LAC, EXE + VEH and EXE + LAC. LAC mice were orally administered 3 g/kg of sodium lactate, which was a mixture of a stock solution of sodium lactate (195–05,965, Wako Chemical, Osaka, Japan) and distilled water at a ratio of 1:1, and VEH mice were administered an equal solution excluding sodium lactate. EXE mice were administered the solution immediately after every exercise training. Exercise training was conducted five times per week for five weeks. The treadmill exercise was performed at 15 m/min for 40 min in the first week, 20 m/min for 40 min in the second week, 22 m/min for 50 min in the third week, and 25 m/min for 50 min in the fourth and fifth weeks. The treadmill slope was fixed at 8° (Fig. 2). For motivating mice to run, mild electrical stimulation on a grid at the rear of the treadmill was given. Electri- cal stimulation was set at a constant current of 0.4 mA, which is appropriate electrical level not to cause major distress45–47 and not to increase the circulating lactate level (Fig. 1B). Tissue processing. Mice were dissected under deep anesthesia with 10 μL/g of 1.25% avertin, 48 h after the last treadmill exercise and lactate administration. The reason that mice are sacrificed 48 h after the treat- ments is to avoid acute effects of exercise and/or lactate on hippocampal protein and/or mRNA expression. Considering the previous studies, single exercise and/or lactate injection can increase hippocampal protein and/or mRNA expression in 12 h including BDNF, PGC1α, and so on22,25. Mice were transcardially perfused with cold 0.9% saline. For immunohistochemistry, we randomly selected 5 out of 9 mice of experiment 1 and brains were removed, postfixed in 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS) for 48 h, and stored in cold 30% sucrose in 0.1 M PBS until completely sunken. Brains were then cryosectioned into coronal 40-µm-thick slices (at this stage, one brain sample of LAC was damaged, thus we excluded it from the results); the slices were stored in a cryoprotectant solution (30% ethylene glycol + 30% glycerol in 0.1 M phosphate buffer) at − 20 °C until further analysis. For immunoblotting, bilateral hippocampi of the other 4 mice of experiment 1 were dissected on an ice-cooled plate, immediately frozen in liquid nitrogen, and stored at − 80 °C until further analysis. Immunohistochemistry. Every fourth section was taken from the region between brain bregma − 1.46 mm and − 2.18 mm. Six randomly selected sections per brain were used for analysis. Free-floating sections were incubated in 0.3% H2O2 to inhibit endogenous peroxidase and in 10% normal goat serum (NGS, S-1000, Vector Laboratories, Burlingame, CA, USA) prepared in PBS containing 0.1% Tween 20 (PBS-T) to block nonspecific protein binding. Sections were incubated overnight at 4 °C with rabbit anti-Ki-67 (1:1,000, ab15580) and rab- bit anti-DCX (1:2000, ab18723) primary antibodies (Abcam, Cambridge, MA, USA) in 3% NGS prepared in 0.1% PBS-T and subsequently for 1 h at 23–25 °C in biotinylated goat anti-rabbit secondary antibodies (1:300, BA-1000, Vector Laboratories, Burlingame, CA, USA) in 3% NGS prepared in 0.1% PBS-T. The sections were further incubated with ABC reagent (1:200, VECTASTAIN Elite ABC kit, PK-6101; Vector Laboratories, Burl- ingame, CA, USA) for 90 min at 23–25 °C. Finally, the sections were visualized using a DAB Substrate Kit (SK-4100, Vector Laboratories, Burlingame, CA, USA) and mounted. To determine the subgranular zone of the DG, the granule cell layer was divided into three layers48. The granule cell layer width was determined using the gridlines. Then, the granule cell layer was divided into three layers of approximately equal thickness. The number of positive cells in the most inner layer was manually counted using EVOS M5000 microscopy (Thermo Fisher Scientific, Waltham, MA, USA) under 20 × and 40 × objective lenses and normalized by length of the border line between the subgranular zone and hilus. The length was measured using Image J software (NIH Image Engineer- ing, Bethesda, MD, USA). 10 Vol:.(1234567890) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ Immunoblotting analysis. Hippocampi were homogenized using a TissueRuptor (QIAGEN, Hilden, Germany) in 400 μL of protein extraction buffer (EzRIPA Lysis kit, WSE07420, ATTO, Tokyo, Japan). Lysates were centrifuged at 20,000×g at 4 °C for 15 min. Thereafter, the lipid layer (top layer) was removed, and clear supernatants were transferred to a new tube. The supernatants were centrifuged again at 20,000×g at 4 °C for 15 min. Finally, the supernatants were transferred to new tubes. Protein concentration was determined using a Pierce™ BCA Protein Assay Kit (23225, Thermo Fisher Scientific, Waltham, MA, USA). Proteins were denatured by heating at 100 °C for 5 min. Total protein (40 μg per lane) was separated using 10% or 12% SDS-PAGE at 60 V for 30 min, followed by 100 V for 120 min, and then transferred to polyvinylidene difluoride membranes (ISEQ00010, Millipore, Billerica, MA, USA) at 100 V for 2 h. The membranes were blocked for 1 h at 23–25 °C in 5% non-fat dried milk (NFDM, F141511, Cellconic, Hanam, Republic of Korea) in 0.1% PBS-T, then incubated overnight at 4 °C in primary antibodies in 3% NFDM in 0.1% PBS-T, and subsequently incubated for 90 min at 23–25 °C in horseradish peroxidase-conjugated secondary antibodies in 3% NFDM in 0.1% PBS-T (the infor- mation on antibodies is provided as Supplementary Table 1). Immunodetection was performed using ECL™ Prime western blotting Detection Reagents (GERPN2232; Cytiva, Marlborough, MA, USA). All images showing the results of the quantitative analysis were assessed using the ImageJ software. Radial arm maze. To measure reference and retention memory, eight-arm radial arm maze (RAM)49,50 was performed with mice of cohort 2 (Fig. 2B). To acclimatize to the maze and a reward (sunflower seeds)7,50, mice were allowed to explore and feed freely in the RAM 30 min once a day for 3 days. The rewards were scat- tered in all arms. The learning phase was started following the acclimatizing phase. During the learning phase, each mouse was performed one trial daily for 5 days. The same three arms were rewarded each day and across trials. The arms placed to be rewarded were never changed for a given mouse but varied among mice. A trial ended when 5 min had elapsed or all the rewards had been received, whichever occurred first. The light was kept dim during all trial to reduce the anxiety of mice. Entry into a never-rewarded arm was considered a reference memory error. Therefore, errors ratio refers to the entry number of reference memory errors divided by the total entry number. Retention memory test was conducted 6 days after the last learning trial. Spontaneous alternation behavior test using Y‑maze. To measure spatial working memory, the spontaneous alternation behavior test was conducted with all mice of experiment 1 and 2 (Fig. 2). Each mouse was randomly placed in one arm of the symmetrical Y-maze and allowed to explore freely for 6 min. The light was kept dim during test to reduce the anxiety of mice. The sequence and total number of arms entered were recorded except for the first 1 min, which was considered as the habituation period. The number of arm entries was counted when the hind paws of the mouse were completely placed in the arm. An alternation was defined only as entries into all three arms on consecutive occasions. Therefore, the number of maximum alternations was the total number of arm entries minus two, and the percentage of alternations was calculated as (actual alterna- tions/maximum alternations) × 100. Additionally, in case of mice that recorded 3 or less the total number of arm entries, we were not able to obtain data. Finally, 14 out of 17 mice were included in the results. Statistical analysis. All data were analyzed using IBM SPSS Statistics 25 software. Graph construction was performed using the GraphPad Prism software (version 9.0). All data were checked for normality of distribu- tion using the Shapiro–Wilk test, and all data were verified for normality, except for blood lactate concentration data of LAC at 120 min (Fig. 1A). Therefore, a comparison of LAC and VEH at 120 min was performed using a two-tailed Mann–Whitney test. For other data with normal distribution, comparison of two or more groups over time was performed using two-way repeated analysis of variance (ANOVA), and post hoc tests were per- formed using one-way repeated ANOVA, paired t-test, or independent Student’s t-test. Comparisons between two groups were performed using an independent Student’s t-test. Comparisons of four groups were performed using two-way ANOVA, and post hoc tests were performed using an independent Student’s t-test. Blood lactate concentration (Fig. 1), immunohistochemistry (Fig. 3), RAM (Fig. 4) and Y-maze (Fig. 5) data are presented as the mean ± standard deviation (SD), and immunoblotting data are presented as box plots (Fig. 6-8). A value of p < 0.05 was considered statistically significant. Data availability The datasets generated and/or analyzed during the current study are available from the corresponding author upon reasonable request. Received: 15 July 2022; Accepted: 5 April 2023 References 1. Babcock, K. R., Page, J. S., Fallon, J. R. & Webb, A. E. Adult hippocampal neurogenesis in aging and Alzheimer’s disease. Stem Cell Rep. 16, 681–693 (2021). 2. Landry, T. & Huang, H. Mini review: The relationship between energy status and adult hippocampal neurogenesis. Neurosci. Lett. 765, 136261 (2021). 3. Toda, T., Parylak, S. L., Linker, S. B. & Gage, F. H. The role of adult hippocampal neurogenesis in brain health and disease. Mol. Psychiatry 24, 67–87 (2019). 4. Lee, T. H. Y. & Yau, S. Y. From obesity to hippocampal neurodegeneration: Pathogenesis and non-pharmacological interventions. Int. J. Mol. Sci. 22, 1–33. https:// doi. org/ 10. 3390/ ijms2 20102 01 (2020). 5. Speisman, R. B., Kumar, A., Rani, A., Foster, T. C. & Ormerod, B. K. Daily exercise improves memory, stimulates hippocampal neurogenesis and modulates immune and neuroimmune cytokines in aging rats. Brain. Behav. Immun. 28, 25–43 (2013). 11 Vol.:(0123456789) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ 6. Leem, Y.-H., Kato, M. & Chang, H. Regular exercise and creatine supplementation prevent chronic mild stress-induced decrease in hippocampal neurogenesis via Wnt/GSK3β/β-catenin pathway. J. Exerc. Nutr. Biochem. 22, 1–6 (2018). 7. Choi, S. H. et al. Combined adult neurogenesis and BDNF mimic exercise effects on cognition in an Alzheimer’s mouse model. Science 361, eaan8821 (2018). 8. So, J. H. et al. Intense exercise promotes adult hippocampal neurogenesis but not spatial discrimination. Front. Cell. Neurosci. 11, 1–12. https:// doi. org/ 10. 3389/ fncel. 2017. 00013 (2017). 9. Cahill, S. P., Cole, J. D., Yu, R. Q., Clemans-Gibbon, J. & Snyder, J. S. Differential effects of extended exercise and memantine treat- ment on adult neurogenesis in male and female rats. Neuroscience 390, 241–255 (2018). 10. Pereira, A. C. et al. An in vivo correlate of exercise-induced neurogenesis in the adult dentate gyrus. Proc. Natl. Acad. Sci. U. S. A. 104, 5638–5643 (2007). 11. Pedersen, B. K. Physical activity and muscle–brain crosstalk. Nat. Rev. Endocrinol. 15, 383–392 (2019). 12. Jodeiri Farshbaf, M. et al. Does PGC1α/FNDC5/BDNF elicit the beneficial effects of exercise on neurodegenerative disorders?. NeuroMol. Med. 18, 1–15 (2016). 13. Delezie, J. & Handschin, C. Endocrine crosstalk between Skeletal muscle and the brain. Front. Neurol. 9, 698 (2018). 14. Magistretti, P. J. & Allaman, I. Lactate in the brain: From metabolic end-product to signalling molecule. Nat. Rev. Neurosci. 19, 235–249 (2018). 15. Suzuki, A. et al. Astrocyte-neuron lactate transport is required for long-term memory formation. Cell 144, 810–823 (2011). 16. Müller, P., Duderstadt, Y., Lessmann, V. & Müller, N. G. Lactate and BDNF: Key mediators of exercise induced neuroplasticity?. J. Clin. Med. 9, 1136. https:// doi. org/ 10. 3390/ jcm90 41136 (2020). 17. Nalbandian, M. & Takeda, M. Lactate as a signaling molecule that regulates exercise-induced adaptations. Biology 5, 1–12. https:// doi. org/ 10. 3390/ biolo gy504 0038 (2016). 18. Bettio, L., Thacker, J. S., Hutton, C. & Christie, B. R. Modulation of Synaptic Plasticity by Exercise. International Review of Neuro- biology Vol. 147 (Elsevier Inc, 2019). 19. Liu, P. Z. & Nusslock, R. Exercise-mediated neurogenesis in the hippocampus via BDNF. Front. Neurosci. 12, 1–6 (2018). 20. Hu, J. et al. Elevated lactate by high-intensity interval training regulates the hippocampal BDNF expression and the mitochondrial quality control system. Front. Physiol. 12, 629914. https:// doi. org/ 10. 3389/ fphys. 2021. 629914 (2021). 21. Pötzsch, A. et al. L-lactate exerts a pro-proliferative effect on adult hippocampal precursor cells in vitro. iScience 24, 102126. https:// doi. org/ 10. 1016/j. isci. 2021. 102126 (2021). 22. El Hayek, L. et al. Lactate mediates the effects of exercise on learning and memory through sirt1-dependent activation of hip- pocampal brain-derived neurotrophic factor (BDNF). J. Neurosci. 39, 2369–2382 (2019). 23. Lev-Vachnish, Y. et al. L-lactate promotes adult hippocampal neurogenesis. Front. Neurosci. 13, 1–13. https:// doi. org/ 10. 3389/ fnins. 2019. 00403 (2019). 24. Bergersen, L., Rafiki, A. & Ottersen, O. P. Immunogold cytochemistry identifies specialized membrane domains for monocar- boxylate transport in the central nervous system. Neurochem. Res. 27, 89–96 (2002). 25. Park, J., Kim, J. & Mikami, T. Exercise-induced lactate release mediates mitochondrial biogenesis in the hippocampus of mice via monocarboxylate transporters. Front. Physiol. 12, 736905. https:// doi. org/ 10. 3389/ fphys. 2021. 736905 (2021). 26. Van Hall, G. et al. Blood lactate is an important energy source for the human brain. J. Cereb. Blood Flow Metab. 29, 1121–1129 (2009). 27. Hui, S. et al. Glucose feeds the TCA cycle via circulating lactate. Nature 551, 115–118 (2017). 28. Pierre, K. & Pellerin, L. Monocarboxylate transporters in the central nervous system: Distribution, regulation and function. J. Neurochem. 94, 1–14 (2005). 29. Tsuchida, R. et al. Exercise type influences the effect of an acute bout of exercise on hippocampal neuronal activation in mice. Neurosci. Lett. 783, 136707 (2022). 30. Tsumiyama, W. et al. Evaluation of the lactate threshold of rats using external jugular vein catheterization. J. Phys. Ther. Sci. 24, 1107–1109 (2012). 31. Hattori, S., Omi, N., Yang, Z., Nakamura, M. & Ikemoto, M. Effect of ginger extract ingestion on skeletal muscle glycogen contents and endurance exercise in male rats. Phys. Act. Nutr. 25, 15–19 (2021). 32. Inoue, K. et al. Long-term mild, rather than intense, exercise enhances adult hippocampal neurogenesis and greatly changes the transcriptomic profile of the hippocampus. PLoS ONE 10, 1–25 (2015). 33. Kim, J. et al. Inhibition of androgen receptor can decrease fat metabolism by decreasing carnitine palmitoyltransferase I levels in skeletal muscles of trained mice. Nutr. Metab. 16, 82. https:// doi. org/ 10. 1096/ fasebj. 2020. 34. s1. 04779 (2019). 34. Jang, I., Kim, J., Kyun, S., Hwang, D. & Lim, K. Acute administration of exogenous lactate increases carbohydrate metabolism during exercise in mice. Metabolites 11, 553. https:// doi. org/ 10. 3390/ metab o1108 0553 (2021). 35. Hwang, D., Seo, J., Kim, J. & Lim, K. Effect of mild-intensity exercise training with capsiate intake on fat deposition and substrate utilization during exercise in diet-induced obese mice. Phys. Act. Nutr. 24, 1–6 (2020). 36. Kim, J. W. et al. Comparison of adult hippocampal neurogenesis and susceptibility to treadmill exercise in nine mouse strains. Neural Plast. 2017, 1–13 (2017). 37. Lorens-Martín, M., Torres-Alemán, I. & Trejo, J. L. Exercise modulates insulin-like growth factor 1-dependent and -independent effects on adult hippocampal neurogenesis and behaviour. Mol. Cell. Neurosci. 44, 109–117 (2010). 38. Koltai, E. et al. Combined exercise and insulin-like growth factor-1 supplementation induces neurogenesis in old rats, but do not attenuate age-associated DNA damage. Rejuvenation Res. 14, 585–596 (2011). 39. Yang, J. et al. Lactate promotes plasticity gene expression by potentiating NMDA signaling in neurons. Proc. Natl. Acad. Sci. U. S. A. 111, 12228–12233 (2014). 40. Margineanu, M. B., Mahmood, H., Fiumelli, H. & Magistretti, P. J. L-lactate regulates the expression of synaptic plasticity and neuroprotection genes in cortical neurons: A transcriptome analysis. Front. Mol. Neurosci. 11, 1–17 (2018). 41. Lourenco, M. V. et al. Exercise-linked FNDC5/irisin rescues synaptic plasticity and memory defects in Alzheimer’s models. Nat. Med. 25, 165–175 (2019). 42. Wrann, C. D. et al. Exercise induces hippocampal BDNF through a PGC-1α/FNDC5 pathway. Cell Metab. 18, 649–659 (2013). 43. Morland, C. et al. Exercise induces cerebral VEGF and angiogenesis via the lactate receptor HCAR1. Nat. Commun. 8, 15557. https:// doi. org/ 10. 1038/ ncomm s15557 (2017). 44. Paillard, T., Rolland, Y. & de Barreto, P. S. Protective effects of physical exercise in Alzheimer’s disease and Parkinson’s disease: A narrative review. J. Clin. Neurol. 11, 212–219 (2015). 45. Marcaletti, S., Thomas, C. & Feige, J. N. Exercise performance tests in mice. Curr. Protoc. Mouse Biol. 1, 141–154 (2011). 46. Sharp, M. et al. Marine phytoplankton improves recovery and sustains immune function in humans and lowers proinflammatory immunoregulatory cytokines in a rat model. Phys. Act. Nutr. 25, 42–55 (2021). 47. Kim, H. J. & Kwon, O. Aerobic exercise prevents apoptosis in skeletal muscles of high-fat-fed ovariectomized rats. Phys. Act. Nutr. 26, 1–7 (2022). 48. Mathews, E. A. et al. A distinctive layering pattern of mouse dentate granule cells is generated by developmental and adult neu- rogenesis. J. Comp. Neurol. 518, 4479–4490 (2010). 49. Wenk, G. L. Assessment of spatial memory using the radial arm maze and morris water maze. Curr. Protoc. Neurosci. 26, 1–12 (2004). 12 Vol:.(1234567890) Scientific Reports | (2023) 13:5838 | https://doi.org/10.1038/s41598-023-33017-1 www.nature.com/scientificreports/ 50. Whishaw, I. & Tomie, J. Of mice and mazes: Similarities between mice and rats on dry land but not water mazes. Physiol. Behav. 60, 1191–1197 (1996). Acknowledgements This study was supported by the KU Research Professor Program of the Konkuk University. This study was sup- ported by the Ministry of Education of the Republic of Korea and the National Research Foundation of Korea (NRF-2021R1G1A1011987). We would like to thank Editage (www. edita ge. co. kr) for English language editing. Author contributions D.H., J.K., H.P., and K.L. contributed to the study hypotheses and design. D.H., S.K., I.J., and T.K. conducted the experiments. D.H., S.K., I.J., and T.K. contributed to data acquisition. All authors contributed to the data curation. D.H. and J.K. wrote the first draft of the manuscript. All authors contributed to manuscript revision, read, and approved the submitted version. Competing interests The authors declare no competing interests. Additional information Supplementary Information The online version contains supplementary material available at https:// doi. org/ 10. 1038/ s41598- 023- 33017-1. Correspondence and requests for materials should be addressed to K.L. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2023

Exogenous lactate augments exercise-induced improvement in memory but not in hippocampal neurogenesis.

04-10-2023

Hwang, Deunsol,Kim, Jisu,Kyun, Sunghwan,Jang, Inkwon,Kim, Taeho,Park, Hun-Young,Lim, Kiwon

eng

PMC6342873

Vol.:(0123456789) 1 3 Archives of Orthopaedic and Trauma Surgery (2019) 139:25–33 https://doi.org/10.1007/s00402-018-3035-5 ORTHOPAEDIC SURGERY A complete posterior tibial stress fracture that occurred during a middle-distance running race: a case report Jun Komatsu1  · Atsuhiko Mogami2 · Hideaki Iwase3 · Osamu Obayashi2 · Kazuo Kaneko1 Received: 5 August 2018 / Published online: 7 September 2018 © The Author(s) 2018 Abstract Posterior tibial stress fractures are more frequent than anterior tibial stress fractures, and they are considered to have a good prognosis for returning to sports; cases leading to a complete fracture are rare. A 17-year-old male involved in high school athletics middle-distance running had a 3-week history of pain with training. He was running up to 300 km/week on streets and cross-country in an even distribution. He had posterior tibial stress fractures, but despite the lower leg pain, he contin- ued running. One year later, he was brought to the emergency department after having sustained an injury to the right lower leg while running in a middle-distance race; bilateral tibial stress fractures, with one side complete and the opposite side incomplete, had developed simultaneously. This relatively rare case of bilateral posterior stress fractures, with one side a complete fracture and the opposite side an incomplete fracture, that was treated surgically via exchange intramedullary nail- ing is reported. The patient could begin light jogging from 3 months after surgery and was without symptoms at 5 months after surgery. He could resume middle-distance racing after 1 year. Posterior tibial cortical fractures are more common and respond better to conservative treatment than anterior tibial stress fractures, and they are a common fracture type in runners. We believe that close, careful follow-up is necessary if patients continue excessive training. Keywords Stress fractures · Running · Overuse injuries · Posterior tibial stress fracture · Runner-type stress fracture · Intramedullary nailing Introduction Highly committed athletes commonly develop stress frac- tures. In the general population, long-distance running is a particularly common form of exercise, physical activity, and leisure activity. It has become increasingly popular due to its easy accessibility and a growing interest in disease prevention. Although many positive health effects have been attributed to distance running, it can cause injuries. Repetitive tissue stress frequently causes overuse injuries affecting the lower extremities [1]. Adolescent athletes place high physical demands on their bodies that vary depending on the given sports activity, which may cause stress or avul- sion fractures due to repetitive microtrauma that overloads the bone. A runner who suddenly increases the intensity and duration of training is at risk for developing a stress fracture [2]. A tibial shaft stress fracture is the most frequent such fracture in athletes [3, 4]. The tibia is reported to be the most common site of stress fractures, accounting for 35–56% of all stress fracture injuries [5]. Tibial stress fractures can be classified into two groups depending on the location, anterior and posterior, causing anterior and posterior/posteromedial stress fractures, respec- tively. Anterior stress fractures occur in sports with frequent jumping, and they are characterized by prolonged healing due to excessive fibrous growth [6]. Anterior cortical frac- tures are less common than posterior stress fractures [7], and they often heal poorly due to constant tension exerted by relatively poor vascular and posterior muscular forces; they are located on the anterior, tension side of the tibial * Jun Komatsu jkomatsu@juntendo.ac.jp 1 Departments of Medicine for Motor Organs, Juntendo University Graduate School of Medicine, 2-1-1 Hongo, Bunkyo-ku, Tokyo 113-8421, Japan 2 Department of Orthopaedic Surgery, Juntendo University Shizuoka Hospital, 1129 Nagaoka, Izunokuni 410-2295, Shizuoka, Japan 3 Department of Bio-Engineering, Juntendo University Institute of Casualty Center, 1129 Nagaoka, Izunokuni 410-2295, Shizuoka, Japan 26 Archives of Orthopaedic and Trauma Surgery (2019) 139:25–33 1 3 shaft, and are prone to delayed union and nonunion [8]. In some instances, anterior tibial stress fractures can progress to complete fractures [9–11]. On the other hand, posterior tibial cortical fractures are more common and respond adequately to conservative treat- ment, but they are a significant clinical problem for runners. In most cases, conservative treatment is sufficient, and sur- gical treatment is very rarely needed. The most predomi- nant type is the low-risk posteromedial cortex (compression side) stress fracture [3, 12, 13]. However, to the best of our knowledge, no previous reports have specifically examined complete posterior tibial stress fractures. A relatively rare case of simultaneous bilateral posterior tibial stress fractures, in which one side was a complete frac- ture and the opposite side was an incomplete fracture, which was treated surgically via exchange intramedullary nailing, is reported. Case report A 17-year-old male involved in high school athletics middle- distance running presented with a 3-week history of pain with more training. He was running up to 300 km/week on streets and cross-country in an even distribution. Although he had taken analgesics, the pain during exercise did not improve, and he presented to our emergency department with lower leg pain (Fig. 1). There was no clear abnormality on the radiographs of the tibia, but STIR magnetic resonance imaging (MRI) confirmed a high-intensity area of the distal one-third of the tibia, and the diagnosis of stress fracture and shin splint was made. The patient was instructed to sus- pend training, and the injury was treated conservatively with follow-up on an outpatient basis (Fig. 2). Follow-up radio- graphs were checked at 2 and 3 months. With this treatment, the fracture healed with no complications, and he decided to return to running after 3 months. At 6 months, radiography showed thickening of the bone cortex in the back one-third of the right tibia and in the back of the distal part of the left tibia, so that he was again instructed to stop training (Fig. 3). However, he discontinued coming to the outpatient clinic on his own after 6 months. He was then seen in the emergency department, having sustained an injury to the right lower leg while running a middle-distance race, 1 year after the initial examination. He described how, when he had just started and passed through the first corner, he had felt a ‘‘snap’’ in his right calf, sud- denly could not run, and fell and had to abandon the race. He said that his leg was deformed in an impossible direc- tion. It became impossible to run because of the lower leg deformities, and he was brought to our emergency depart- ment. He was admitted to hospital, and X-ray examination showed a greatly displaced oblique fracture in the proximal Fig. 1 Initial radiographs show suspected tibial stress fracture or shin splints. There is no clear abnormality in the radiographs of the tibia in the antero-posterior and lateral views. A Right side; B left side 27 Archives of Orthopaedic and Trauma Surgery (2019) 139:25–33 1 3 1/3 of the left lower leg, but, fortunately, there was no open wound (Fig. 4). Before the race that day, the patient had not experienced any pain or discomfort in his left lower leg while running. In the emergency department, the primary assessment showed moderate direct and indirect tenderness of the opposite proximal tibia. There was no malalignment, swelling, or discoloration of the left lower leg. The radiographs showed a full-thickness fracture of the proximal one-third posterior tibial shaft (Fig. 4). On MRI, T1-weighted imaging showed a high-signal area at the mid- dle one-third and a low-signal on T2-weighted imaging, and STIR showed an abnormal high signal at the same site (Fig. 5). These findings suggested abnormalities such as edema and bleeding in the bone marrow. On bone scintig- raphy, there was moderate accumulation in the vicinity of the left lower third of the thigh, so a left tibial stress fracture was diagnosed (Fig. 6). Fig. 2 Initial coronal MRI scans diagnosed with a stress fracture show a strikingly wide low-signal intensity on the T1-weighted scan (A), and a high-signal intensity on the T2-weighted scan (B) and STIR fat-suppressed scan (C) in the localized bone marrow. The abnormal finding is more detectable on the STIR fat-suppressed MRI scan 28 Archives of Orthopaedic and Trauma Surgery (2019) 139:25–33 1 3 The following day, surgical treatment for the injury was performed under general anesthesia after written, informed consent was obtained from the patient and parents. In addi- tion, the patient gave written, informed consent for pub- lication of the case, including the accompanying images. At our institution, ethical approval is not required for reporting individual cases. Closed reduction and internal fixation of the radial shaft fracture were performed, with intramedullary nailing (10-mm-diameter, T2 Nailing sys- tem, Stryker, Kalamazoo, MI) for both the left complete tibial fracture and the right tibial stress fracture (Fig. 7). The injury was then treated conservatively, with 9 weeks in an ROM knee cast and no weight-bearing on the affected leg. Healing was monitored through a series of follow-up radiographs. With this treatment, the fracture healed with no complications. Although the patient was asymptomatic and clinical healing of the fracture was apparent 10 months after the nailing, a fracture line was still visible on radio- graphs (Fig. 8). The patient could begin light jogging from 3 months after the operation and was without symptoms at 5 months. He returned to middle-distance racing after 1 year. Discussion If operative treatment is not performed, a stress fracture that is persistently painful may develop delayed union, nonun- ion, or even a complete fracture. An established technique for treating delayed or nonunion tibial stress fractures is intramedullary nailing [14]. There are two types of tibial stress fractures, anterior and posterior, based on their loca- tion. Furthermore, there are many frequent running-type stress fractures (posterior fractures), which are thought to resolve better than jumping-type stress fractures (anterior fractures), which often develop in field athletes. In particu- lar, posterior tibial cortical fractures, which are more fre- quent and show an adequate response to conservative ther- apy, are often seen in runners. Ohnishi said that actually they were widely distributed proximally to distally including the middle third, so runner-type stress fractures are more likely to be generated from the posterior tibia [15]. The distance run per week can also be a factor in stress injuries. It has been shown that running more than 64 km/week (approxi- mately 40 miles/week) is a significant risk factor for lower extremity injuries [16]. In the present case, the patient had posterior tibial fractures, but, despite the lower leg pain, he continued running, so that bilateral tibial stress fractures, with one side complete and the opposite side incomplete, occurred simultaneously. To the best of our knowledge, there have been no previous reports of a posterior tibial stress fracture with a complete fracture as in the pre- sent case. Because the symptoms improve with rest for a short period of time, many patients may not seek treat- ment. Moreover, there may be intrinsic elements, that are internal factors, which result in additional stresses to the bone. Such intrinsic elements include anatomical variations, footwear, running mechanics, training regi- mens, and running surfaces, as well as individual health factors, such as poor bone health (osteoporosis and low Fig. 3 Radiographs of both legs 6 months after the initial visit to our hospital. Obtained 6 months after the first examination, callus forma- tion is seen at the lateral and posterior side of the tibia in the antero- posterior and lateral views. The arrows indicate callus formation at the lateral and posterior sides of the tibia 29 Archives of Orthopaedic and Trauma Surgery (2019) 139:25–33 1 3 bone density) [17]. According to Reeder et al. [18], it is important to focus on the runner’s training regimen and history to identify potential injury-causing factors. A pes cavus foot is linked to stress fracture incidence; because this foot type is more rigid, it does not absorb shock and Fig. 4 Radiographs at the emergency department with deformity of the right lower leg. Full-length tibial radiographs were requested in keeping with the clinical picture, and they confirm complete tibial and fibular fractures of the right side. A Right-side tibial radiography and 3D computed tomography; B Left-side tibial posterior stress frac- ture. Arrows indicate callus formation sites 30 Archives of Orthopaedic and Trauma Surgery (2019) 139:25–33 1 3 passes impact forces to the tibia, therefore, increasing the risk for a tibial stress injury [19]. Most patients with the symptoms of stress fractures reduce the distance, frequency, and intensity of their activ- ities. Distance runners have an increased risk for stress fractures because of the high impact and repetitive loads. With repetitive mechanical loading of bone, cumulative bone strain can cause bone damage and stress fractures if net bone damage is chronically greater than bone repair [20]. Especially in runners, posterior tibial stress fractures occur commonly on the compressive posterior surface of the proximal and distal thirds of the tibia; these fractures can be considered low-risk fractures compared to anterior fractures and managed by relative rest. Nonsurgical treat- ment of posterior tibial stress fractures begins with rest and stopping the aggravating activity. A stress fracture is a mechanical failure of the bone in which the activity of the osteoblasts cannot keep pace with the activity of the osteoclasts. Repetitive, cyclical loading of the bone with inadequate recovery occurs, and the bone is unable to repair itself between exercise sessions [18]. With heavy loading of a bone, microcracks may appear within the bone tissue. Such microcracks are thought to contribute to acti- vating the remodeling process, which is required for adap- tation of the bone to the functional demands on the tissues caused by loading. However, with excessive loading, either in magnitude or frequency, a stress fracture can develop due to the insufficient time for remodeling to repair the microcracks [6]. Unfortunately, the repetitive and high loading nature of running creates an ideal environment for the development of stress fractures. Furthermore, a complete fracture may occur due to large cracks. In the present case, due to repeated periods of rest and resumption of competition, at 1 year, a complete fracture finally resulted. Restricting excessive exercise and ensur- ing sufficient rest when lower extremity pain appears may have made it possible to return to racing at an early stage. The present case had an insidious onset, present- ing with moderate clinical signs and symptoms; therefore, diagnosis might have been delayed by analgesic medica- tion, resulting in a complete fracture. At the same time, intramedullary nailing was performed for the opposite tibial stress fracture to facilitate an early return to training. In conclusion, our observations demonstrate that the posterior tibia stress fracture is a more frequent tibial stress fracture than the anterior stress fracture, and it is considered to have a good prognosis for returning to sports; although rare, there are cases that result in a com- plete fracture, and when excessive training is continued, careful follow-up is needed. Fig. 5 MRI of the left lower leg after right complete fractures. On MRI, STIR shows an abnormal high signal 31 Archives of Orthopaedic and Trauma Surgery (2019) 139:25–33 1 3 Fig. 6 Bone scintigraphs show abnormal local uptake in the antero- posterior (A) and postero-anterior (B) and lateral views of the right side (C) and left side (D) of the patient with stress fractures. Arrows indicate longitudinal linear uptake in the bone scintigraph views of the patient with stress fractures 32 Archives of Orthopaedic and Trauma Surgery (2019) 139:25–33 1 3 Open Access This article is distributed under the terms of the Crea- tive Commons Attribution 4.0 International License (http://creat iveco mmons .org/licen ses/by/4.0/), which permits unrestricted use, distribu- tion, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. References 1. Van Gent RN, Siem D, van Middelkoop M, van Os AG, Bierma- Zeinstra SM, Koes BW (2007) Incidence and determinants of lower extremity running injuries in long distance runners: a systematic review. Br J Sports Med 41(8):469–480 2. Ballas MT, Tytko J, Cookson D (1997) Common overuse run- ning injuries: diagnosis and management. Am Fam Physician 55(7):2473–2484 Fig. 7 Radiographs show fixation of both tibial fractures, complete and incomplete, with intramedullary nailing. Closed reduction and internal fixation of the radial shaft fracture were performed using intramedullary nailing for both the right com- plete tibial fracture (A) and the left tibial stress fracture (B) 33 Archives of Orthopaedic and Trauma Surgery (2019) 139:25–33 1 3 3. Brukner P, Bradshaw C, Khan KM, White S, Crossley K (1996) Stress fractures: a review of 180 cases. Clin J Sport Med 6(2):85–89 4. Jones BH, Thacker SB, Gilchrist J, Kimsey CD Jr, Sosin DM (2002) Prevention of lower extremity stress fractures in athletes and soldiers: a systematic review. Epidemiol Rev 24(2):228–247   5. Pohl MB, Mullineaux DR, Milner CE, Hamill J, Davis IS (2008) Biomechanical predictors of retrospective tibial stress fractures in runners. J Biomech 41(6):1160–1165 6. Bargfeldt C, Krogsgaard M, Rasmussen SW (2011) Stress fracture in combination with avulsion from the tibia in a marathon runner: a case report. Scand J Med Sci Sports 21(2):330–332 7. Orava S, Hulkko A (1984) Stress fracture of the mid-tibial shaft. Acta Orthop Scand 55:35–37 8. Orava S, Hulkko A (1988) Delayed unions and nonunions of stress fractures in athletes. Am J Sports Med 16:378–382 9. Kaeding CC, Yu JR, Wright R, Amendola A, Spindler KP (2005) Management and return to play of stress fractures. Clin J Sport Med 15(6):442–447 10. Martinez SF, Murphy GA (2005) Tibial stress fracture in a male ballet dancer: a case report. Am J Sports Med 33(1):124–130 11. Plasschaert VF, Johansson CG, Micheli LJ (1995) Anterior tibial stress fracture treated with intramedullary nailing: a case report. Clin J Sport Med 5(1):58–62 12. Matheson GO, Clement DB, McKenzie DC, Taunton JE, Lloyd- Smith DR, MacIntyre JG (1987) Stress fractures in athletes: a study of 320 cases. Am J Sports Med 15:46–58 13. Bennell K, Brukner P (1997) Epidemiology and site specificity of stress fractures. Clin Sports Med 16:179–196 14. Hattori H, Ito T (2015) Recurrent fracture after anterior tension band plating with bilateral tibial stress fracture in a basketball player. Orthop J Sports Med 3(10):1–5 15. Ohashi J, Goto T (2003) Stress fracture in the tibia in long-dis- tance runners: fracture level in the tibia and radiographic charac- teristics. Jpn J Orthop Sports Med 23(3):254–258 (In Japanese) 16. Macera CA, Pate RR, Powell KE, Jackson KL, Kendrick JS, Cra- ven TE (1989) Predicting lower-extremity injuries among habitual runners. Arch Intern Med 149(11):2565–2568 17. Korpelainen R, Orava S, Karpakka J, Siira P, Hulkko A (2001) Risk factors for recurrent stress fractures in athletes. Am J Sports Med 9:304–310 18. Reeder MT, Dick BH, Atkins JK, Pribis AB, Martinez JM (1996) Stress fractures. Current concepts of diagnosis and treatment. Sports Med 22(3):198–212 19. Bennell K, Brukner P (2005) Preventing and managing stress frac- tures in athletes. Phys Ther Sport 6(4):171–180 20. Warden SJ, Burr DB, Brukner PD (2006) Stress fractures: patho- physiology, epidemiology, and risk factors. Curr Osteoporos Rep 4:103–109 Fig. 8 Postoperative antero-posterior and lateral radiographs show newly healed tibial stress fractures. Postoperative radiographs show antero-pos- terior and lateral low views. The patient was asymptomatic and clinical healing of the fracture is apparent 10 months after the nailing, and a frac- ture line is hardly visible on the radiographs. A Right side; B left side

A complete posterior tibial stress fracture that occurred during a middle-distance running race: a case report.

09-07-2018

Komatsu, Jun,Mogami, Atsuhiko,Iwase, Hideaki,Obayashi, Osamu,Kaneko, Kazuo

eng

PMC7559068

International Journal of Environmental Research and Public Health Article High-Performance Handball Player’s Time-Motion Analysis by Playing Positions Carmen Manchado 1 , Juan Tortosa Martínez 1 , Basilio Pueo 1 , Juan Manuel Cortell Tormo 1 , Helena Vila 2,*, Carmen Ferragut 3, Francisco Sánchez Sánchez 4, Sonia Busquier 5, Sergio Amat 5 and Luis Javier Chirosa Ríos 6 1 Faculty of Education, University of Alicante, 03690 San Vicente del Raspeig, Spain; carmen.manchado@ua.es (C.M.); juan.tortosa@gcloud.ua.es (J.T.M.); basilio@gcloud.ua.es (B.P.); jm.cortell@gcloud.ua.es (J.M.C.T.) 2 Faculty of Education, University of Vigo, 36905 Pontevedra, Spain 3 Faculty of Medicine and Health Sciences, University of Alcalá, 28871 Alcalá de Henares, Spain; cferragutfiol@gmail.com 4 Faculty of Sport Science, University of Castilla La Mancha, 45071 Toledo, Spain; Fco.Sanchez@uclm.es 5 Department of Applied Mathematics and Statistics, University of Cartagena, 30203 Cartagena, Spain; sonia.busquier@upct.es (S.B.); sergio.amat@upct.es (S.A.) 6 Department of Physical Education and Sports, University of Granada, 18011 Granada, Spain; lchirosa@ugr.es * Correspondence: evila@uvigo.es Received: 31 July 2020; Accepted: 13 September 2020; Published: 17 September 2020



 Abstract: The purpose of this study was to analyze the on-court demands of handball players during the European Handball Federation Champions League Final Four (VELUX EHF FINAL4) 2019 to define time–motion characteristics (played time; covered distances) both in offense and defense. Furthermore; we aimed to define position-specific demands and differences among them. Forty players from three teams were analyzed during the tournament using a local positioning system (LPS) for the first time in top handball. Players covered similar distances both in offense (1388.28 ± 2627.08 m), and in defense (1305.47 ± 5059.64 m) and remained on court for a similar average time (15.69 ± 8.02 min and 15.40 ± 8.94 min respectively). When locomotion activities were normalized according to the time they spent on court; significant differences were found for defense compared to offense in walking (+20%; p < 0.000; Cohen’s effect size (ES) = 1.01) and jogging (−29.6%; p = 0.000; ES = 0.90), as well as a tendency for high-intensity running (+ 25.2%; p = 0.077; ES = 0.31). Per playing position; center and left back (CB = 94.86 ± 10.98 m·min−1; LB = 96.55 ± 24.65 m·min−1) showed the highest running pace in offense and mid-left; front center defender and outside right for the defense (ML = 90.38 ± 30.16 m·min−1; FCD = 87.04 ± 14.94 m·min−1; OR = 89.64 ± 34.93 m·min−1). In conclusion; profile differences existed among players’ position activity; both in offense and defense; which should be taken into account when designing specific physical training programs Keywords: running pace; running distance; competition load; LPS 1. Introduction Handball is an Olympic sport, belonging to so-called team sports. It is characterized by fast transitions between offensive and defensive actions during the game with the ultimate objective of scoring a goal [1,2]. To this end, offensive players (six field players and one goalkeeper) attempt to create spaces that allow them to throw the ball towards the goal in advantageous conditions, while the defense tries to avoid it, causing a great amount of physical confrontations between players [2]. These attack phases in handball are dynamic, characterized by fast movements and a high frequency of fast passes, so physical demands are important [1]. Furthermore, these physical demands are not Int. J. Environ. Res. Public Health 2020, 17, 6768; doi:10.3390/ijerph17186768 www.mdpi.com/journal/ijerph Int. J. Environ. Res. Public Health 2020, 17, 6768 2 of 15 only the same if the team is in the offensive or defensive phase and even if the player plays in one position or another [3–5]. The very nature of the game implies that players must be physically trained to maintain the game’s speed and intensity throughout a match [4,6–8], whether they play in offense or defense. Therefore, knowing and understanding the sport’s physical demands (distances, speeds, intensities) [1], as well as technical–tactical actions [4,7] (passes, throws, jumps, marking, change of direction, etc.) is essential to correctly plan players’ training. [1,8]. All these elements are of great importance in handball and are also closely related to each other, which makes handball a particularly complex sport [4,5,9]. Likewise, it is important to note that the playing position, the game phase (offense or defense), as well as the team’s playing style can lead to big differences in each player’s physical demands. Therefore, the physical load cannot only be determined generally, but according to each player’s specific position on the court both in offense and defense [4,6,10]. All this information could help coaches to better individualize training loads and thereby improve performance [4,6]. This necessity to understand handball’s physical characteristics has raised great interest among researchers who have studied these demands using different methodologies [1]. The most widely used method has been time–motion analysis, based on observing players in the competition followed by an analysis of a video, taken with one camera [11,12] or two cameras [13]. The video-recorded matches are analyzed and the actions encoded. However, this method is time-consuming and depends on a subjective analysis of the observer, thus not being an objective or precise method when determining the different locomotion speeds. Notwithstanding, no method exists to date that allow one to accurately measure the physical and physiological demands of handball players during the competition. In order to overcome this gap, the European Handball Federation (EHF), Select® and Kinexon® jointly developed the Kinexon® tracking system for handball players (Kinexon: München, Germany; Select Sport 1947: Glostrup, Denmark) in addition to a monitored ball, the iball, which has been recently validated [14] and used in studies on handball [15] and other team sports [16]. This technology provides us with values regarding movements, accelerations, changes of direction, jumps, as well as data on the speed at which the ball is transferred (game speed) and the speed and position of the throws in real time, opening up new possibilities in the study of handball competition requirements [16]. With this fully automatic tracking system, the inconveniences mentioned for the conventional time–motion analysis are solved. Despite a great interest in understanding the requirements of high-level players, only a few studies have focused on analyzing the real demands of an elite handball competition in male handball [3,13,17,18]. Cardinale, Whiteley, Hosny, and Popovic [3], studied players’ movements during the men’s world championship using three cameras, and provided new data on players’ movements (distances and intensities) during the match. They concluded that there was no significant difference in terms of distance covered in different locomotion categories, but they did not distinguish between offense—and defense—specific playing positions. In the same line, González de Haro [17] reported the analysis of only one match with Global Positioning System devices (WIMU PRO™, Realtrack Systems S.L.: Almeria, Spain). These researchers [3,17] concluded that specific physical conditioning is necessary to maximize performance of handball players and minimize the occurrence of fatigue. To the best of our knowledge, no study has been conducted considering in detail the two phases of the game, offense and defense, and analyzing all the playing positions by using a technology that allows load individualization and automation, a local positioning system (LPS). Better knowledge of on-court demands of handball players at the highest level is necessary to improve the individualization of physical preparation [3,6,7,17,18]. Thus, the aim of this study was to analyze on-court demands of handball players during the VELUX EHF FINAL4 to define time–motion characteristics (played time, covered distances) both in offense and defense, including position-specific demands and differences among them. Int. J. Environ. Res. Public Health 2020, 17, 6768 3 of 15 2. Materials and Methods 2.1. Subjects Data were obtained from players participating in the VELUX European Handball Federation (EHF) Champions League Final Four 2019/20, held in Cologne (Germany). The teams that participated in the Final Four were FC Barcelona (Spain), Telekom Veszprém (Hungary), HC Vardar (The Republic of North Macedonia), and KS Kielce (Poland). Barcelona’s players were not included in the study because their sensors were not placed properly, causing interferences in the signal and thus unreliable data. Dainis Krištop¯ans (HC Vardar) did not wear the sensors during the games so his data were not available for the analysis either. Finally, 40 players were analyzed during both semifinals, the final championship game and the bronze medal game. Goalkeepers were excluded from the analysis as distance and motion characteristics do not reflect their performance needs. Anthropometric characteristics and the age of the players are presented in Table 1. This information was collected from the official statistical data provided by the EHF. Table 1. Physical characteristics of the players (Mean ± Standard Deviation). Teams n Height (cm) Body Mass (kg) BMI (kg/m2) Age (Years) TELEKOM VESZPRÉM 14 193.0 ± 8.8 92.9 ± 13.6 24.8 ± 1.8 31.0 ± 4.2 HC VARDAR 13 190.2 ± 10.4 90.5 ± 14.3 24.9 ± 2.4 29.7 ± 4.2 KS KIELCE 13 190.1 ± 6.4 90.1 ± 9.9 24.9 ± 2.1 28.2 ± 6.1 Total 40 191.1 ± 8.6 91.2 ± 12.5 24.8 ± 2.1 29.7 ± 4.9 Legend: BMI = Body mass Index. 2.2. Instruments The players’ position data were collected through a Local Positioning System (LPS) (Kinexon Precision Technologies, Munich, Germany), which has been recently validated [7] and used in studies on team sports [8,9], showing adequate between-device reliability (coefficient of variation around 5%) when compared to well-known systems such as GPS. Firmware versions and application software versions corresponded to the latest releases on the testing date (August 2019). Figure 1 shows the setting of the 9 antennae around the playing field, connected via ethernet to the main server, and 10 anchor antennae distributed at 3 different levels above the ground in the Lanxess Arena. The LPS system was installed, calibrated, and checked for accuracy by a technician who worked for the manufacturer as follows: The exact position of the anchors in reference to the playing field was measured (blue numbered positions in Figure 1). Then, the anchor positions and the playing field position and size were transmitted to the Kinexon application. The location of one sensor at pre-defined positions (corner, penalty line, center point) was checked. In addition, two paths were followed to test the data quality and calculated distance—walking on the sideline and walking on a meander inside the field (black discontinued line in Figure 1). The devices worn by players comprised a sensor (player tag) positioned between the player’s shoulder blades using a pouch sewn onto the player’s jersey. The functionality of the sensors was tested in the venue by randomly walking and checking if signals were received from all units with adequate signal strength. These sensors transmit time signals via radio-technology to the antennae, which send signals via a wide local area network (WLAN) to local static base stations at known locations. A player’s momentary position is determined via 20 Hz frequency by calculating the time-of-flight (TOF) of ultra-wide-band radio signals traveling from the transmitter to the base stations, which calculate the actual 2D position of the devices within the playing field. Subsequently, instantaneous speed, i.e., scalar magnitude of velocity, as per the rate of change in horizontal x, y positions, and acceleration, as per the rate of change in speed, are derived by calculating the difference between two consecutive positions, i.e., approximating the derivative of the player’s position. The raw position and speed data are then filtered and smoothed by means of a Int. J. Environ. Res. Public Health 2020, 17, 6768 4 of 15 Kalman filter for position data and an exponential moving average with a window length of 1 s for speed and position data. Data were split into offensive and defensive moments of play automatically. To this end, there was automatically a change from offensive to defensive for the team and vice versa at the moment where the ball possession changed. The respective offensive shift started with the ball possession of the team. Moreover, the system also checked if the players and the ball were moving in the direction of the opponent’s goal. In the event the ball was outside the court, the shift was interrupted. All data were analyzed using the system software (Kinexon Web Application, version 3.2.6, Munich, Germany). TELEKOM VESZPRÉM 14 193.0 ± 8.8 92.9 ± 13.6 24.8 ± 1.8 31.0 ± 4.2 HC VARDAR 13 190.2 ± 10.4 90.5 ± 14.3 24.9 ± 2.4 29.7 ± 4.2 KS KIELCE 13 190.1 ± 6.4 90.1 ± 9.9 24.9 ± 2.1 28.2 ± 6.1 Total 40 191.1 ± 8.6 91.2 ± 12.5 24.8 ± 2.1 29.7 ± 4.9 Legend: BMI = Body mass Index. 2.2. Instruments The players’ position data were collected through a Local Positioning System (LPS) (Kinexon Precision Technologies, Munich, Germany), which has been recently validated [7] and used in studies on team sports [8,9], showing adequate between-device reliability (coefficient of variation around 5%) when compared to well-known systems such as GPS. Firmware versions and application software versions corresponded to the latest releases on the testing date (August 2019). Figure 1 shows the setting of the 9 antennae around the playing field, connected via ethernet to the main server, and 10 anchor antennae distributed at 3 different levels above the ground in the Lanxess Arena. Figure 1. Local positioning system (LPS) setting: nine antennae connected to the server in red locations; ten reference antennae (anchors) in blue locations; meander path inside the field followed to check calibration accuracy (black discontinued line). Figure 1. Local positioning system (LPS) setting: nine antennae connected to the server in red locations; ten reference antennae (anchors) in blue locations; meander path inside the field followed to check calibration accuracy (black discontinued line). 2.3. Procedure In this study, a descriptive observational cross-sectional study was used to examine the physical demands according to playing positions during competitive matches. This time–motion analysis is used with team [5,19] and beach handball [19], as well as with other team sport studies [20,21]. The study was approved by the EHF. The clubs signed an informed consent in the initial contract with the EHF to take part in the competition, where they accepted the rules and norms of the EHF, including their participation in different studies. The players’ data were anonymized for the purpose of this study. The players were informed of the purposes, procedures, and risks of the study and provided informed consent before the beginning of the study. All the procedures were conducted in accordance with the Declaration of Helsinki and approved by the Ethics Committee of the University of Vigo (registration number 04-719). The variables described next were measured based on position and speed data. The distances covered during the entire match (total distance/duration of play), distances per minute during play and relative distance in established speed zones were computed. These zones were set as zone 1: standing (≤0.9 m/s), zone 2: walking (1.0–1.9 m/s), jogging (2.0–3.9 m/s), running (4.0–5.4 m/s), high-intensity running (5.5–6.9 m/s) and sprinting (≥7 m/s), in accordance with similar handball studies [3,5,18,19]. We also considered the distinction between offense (when the team was in possession of the ball) and defense (not in possession of the ball), and classified the players by their positions according to handball nomenclature in offense (left wing = LW, left back = LB, center back = CB; line player = LP; right back = RB; and right wing = RW) and defense (center back = CB; mid right = MR; mid left = ML; Int. J. Environ. Res. Public Health 2020, 17, 6768 5 of 15 outside right = OR; outside left = OL; and front center defender = FCD). The descriptive analysis of the data included the mean, the range, the variance and the standard deviation. 2.4. Data Analysis Graphical, analytical and numerical studies were performed using our own developed programs. The Shapiro–Wilk test was performed in order to verify the normality of the data. Group differences were determined by variance analysis (ANOVA) followed by Games–Howell or Tukey post hoc testing, or Student’s t-tests for independent samples, where appropriate. To determine the magnitude of each relationship, Cohen’s effect size (ES) was used with a modified classification (trivial <0.2, small 0.21–0.6, moderate 0.61–1.2, large 1.21–1.99, and very large >2.0) proposed for sports sciences [22] and used in other similar handball studies [3]. The precision of population estimates was reported as 95% confidence intervals, and statistical significance was set at p < 0.05. 3. Results 3.1. Time on Court, Distance Covered in Offense and Defense The average time on court in offense (n = 66) and defense (n = 67) during the VELUX EHF Final 4 was 15.69 min (±8.02 min) and 15.40 min (±8.94 min), respectively. The total average distance covered per player during each game in offense was 1388.28 ± 2627.08 and 1305.47 ± 5059.64 m in defense. When comparing offense and defense with regard to the absolute distances covered (Figure 2), significant differences were found in walking (p = 0.017; ES = 0.61) and jogging (p = 0.03; ES = 0.77), as well as a tendency towards high-intensity running (p = 0.075; ES = 0.45). Int. J. Environ. Res. Public Health 2020, 17, x 5 of 14 2.4. Data Analysis Graphical, analytical and numerical studies were performed using our own developed programs. The Shapiro–Wilk test was performed in order to verify the normality of the data. Group differences were determined by variance analysis (ANOVA) followed by Games–Howell or Tukey post hoc testing, or Student’s t-tests for independent samples, where appropriate. To determine the magnitude of each relationship, Cohen´s effect size (ES) was used with a modified classification (trivial <0.2, small 0.21–0.6, moderate 0.61–1.2, large 1.21–1.99, and very large >2.0) proposed for sports sciences [22] and used in other similar handball studies [3]. The precision of population estimates was reported as 95% confidence intervals, and statistical significance was set at p < 0.05. 3. Results 3.1. Time on Court, Distance Covered in Offense and Defense The average time on court in offense (n = 66) and defense (n = 67) during the VELUX EHF Final4 was 15.69 min (±8.02 min) and 15.40 min (±8.94 min), respectively. The total average distance covered per player during each game in offense was 1388.28 ± 2627.08 and 1305.47 ± 5059.64 m in defense. When comparing offense and defense with regard to the absolute distances covered (Figure 2), significant differences were found in walking (p = 0.017; ES = 0.61) and jogging (p = 0.03; ES = 0.77), as well as a tendency towards high-intensity running (p = 0.075; ES = 0.45). Walking Jogging Running HIRunning Sprinting Distance covered (m) 0 200 400 600 800 1000 1200 1400 1600 1800 2000 2200 2400 Offense Defense 16.27% 23.47% 47.06% 37.65% 24.28% 23.83% 9.27% 13.84% 0.35% 0.85% * * Figure 2. Differences in distance covered in offense and defense during different locomotion characteristics. * Statistical differences; p ≤ 0.05.; HIRunning: High-Intensity running Locomotion activities were then normalized for each player according to the time they spent on court to obtain a true reflection of these demands, both for offense and defense. The running pace per game showed by the complete team in offense was 88.45 ± 20.72 m·min−1, walking: 13.89 ± 2.98 m·min−1, jogging: 40.55 ± 10.12 m·min−1, running: 23.65 ± 12.53 m·min−1, high-intensity running: 9.70 ± 9.39 m·min−1 and sprinting: 0.42 ± 0.94 m·min−1. The running pace per game showed by the complete team in defense was 80.83 ± 27.11 m·min−1, walking: 17.53 ± 4.18 m·min−1, jogging: 28.56 ± 4.18 m·min−1, running: 20.49 ± 11.47 m·min−1, high- intensity running: 12.96 ± 11.54 m·min−1 and sprinting: 0.56 ± 1.29 m·min−1. Figure 2. Differences in distance covered in offense and defense during different locomotion characteristics. * Statistical differences; p ≤ 0.05.; HIRunning: High-Intensity running Locomotion activities were then normalized for each player according to the time they spent on court to obtain a true reflection of these demands, both for offense and defense. The running pace per game showed by the complete team in offense was 88.45 ± 20.72 m·min−1, walking: 13.89 ± 2.98 m·min−1, jogging: 40.55 ± 10.12 m·min−1, running: 23.65 ± 12.53 m·min−1, high-intensity running: 9.70 ± 9.39 m·min−1 and sprinting: 0.42 ± 0.94 m·min−1. Int. J. Environ. Res. Public Health 2020, 17, 6768 6 of 15 The running pace per game showed by the complete team in defense was 80.83 ± 27.11 m·min−1, walking: 17.53 ± 4.18 m·min−1, jogging: 28.56 ± 4.18 m·min−1, running: 20.49 ± 11.47 m·min−1, high-intensity running: 12.96 ± 11.54 m·min−1 and sprinting: 0.56 ± 1.29 m·min−1. When comparing offense and defense, significant differences were found in walking (p < 0.000; ES = 1.01) and jogging (p = 0.000; ES = 0.90), as well as a tendency for total distance (p = 0.71; ES = 0.32) and high-intensity running (p = 0.077; ES = 0.31). 3.2. Positional Differences in Distance Covered and Speeds The distances covered in offense by playing position for each locomotion category are shown in Figure 3. Int. J. Environ. Res. Public Health 2020, 17, x 6 of 14 When comparing offense and defense, significant differences were found in walking (p < 0.000; ES = 1.01) and jogging (p = 0.000; ES = 0.90), as well as a tendency for total distance (p = 0.71; ES = 0.32) and high-intensity running (p = 0.077; ES = 0.31). 3.2. Positional Differences in Distance Covered and Speeds The distances covered in offense by playing position for each locomotion category are shown in Figure 3. Walking LP CB RB LB RW LW Distance covered (m) 0 100 200 300 400 500 Jogging LP CB RB LB RW LW Distance covered (m) 0 200 400 600 800 1000 1200 1400 Running LP CB RB LB RW LW Distance covered (m) 0 100 200 300 400 500 600 HI running LP CB RB LB RW LW Distance covered (m) 0 50 100 150 200 250 300 350 * * * ¥ Sprinting LP CB RB LB RW LW Distance covered (m) 0 10 20 30 40 * * * * Figure 3. Distance covered in different locomotion category by playing position in offense. * Statistical differences with the left wing p ≤ 0.05; ¥ = statistical differences with the right wing p ≤ 0.05. Legend: left wing = LW; left back = LB; center back = CB; line player = LP; right back = RB; right wing = RW. Although not many significant differences were found, high effect size values were obtained (Table 2). Figure 3. Distance covered in different locomotion category by playing position in offense. * Statistical differences with the left wing p ≤ 0.05; ¥ = statistical differences with the right wing p ≤ 0.05. Legend: left wing = LW; left back = LB; center back = CB; line player = LP; right back = RB; right wing = RW. Int. J. Environ. Res. Public Health 2020, 17, 6768 7 of 15 Although not many significant differences were found, high effect size values were obtained (Table 2). Table 2. Effect sizes in different locomotion categories by playing position in offense. Playing Positions Walking Jogging Running High Running Sprinting LP CB 0.80 RB 0.86 LB 0.61 RW 0.76 LW 0.76 CB LP 0.63 LB 0.54 LW 0.57 RB 0.75 RB LP 0.75 LB 0.54 LB RW LP 0.93 LP 1.19 LB 1.23 CB 0.90 RB 1.34 RB 0.79 CB 0.69 LW LP 1.25 LP 1.44 CB 1.07 CB 1.24 LB 1.81 RB 1.11 RB 2.04 LB 0.95 Legend: left wing = LW; left back = LB; center back = CB; line player = LP; right back = RB; right wing = RW. The distances covered in defense for each locomotion category by playing position are shown in Figure 4. Int. J. Environ. Res. Public Health 2020, 17, x 7 of 14 Table 2. Effect sizes in different locomotion categories by playing position in offense. Playing Positions Walking Jogging Running High Running Sprinting LP CB 0.80 RB 0.86 LB 0.61 RW 0.76 LW 0.76 CB LP 0.63 LB 0.54 LW 0.57 RB 0.75 RB LP 0.75 LB 0.54 LB RW LP 0.93 LP 1.19 LB 1.23 CB 0.90 RB 1.34 RB 0.79 CB 0.69 LW LP 1.25 LP 1.44 CB 1.07 CB 1.24 LB 1.81 RB 1.11 RB 2.04 LB 0.95 Legend: left wing = LW; left back = LB; center back = CB; line player = LP; right back = RB; right wing = RW. The distances covered in defense for each locomotion category by playing position are shown in Figure 4. Walking CB MR ML OR OL FD Distance covered (m) 0 100 200 300 400 500 600 700 § § Jogging CB MR ML OR OL FD Distance covered (m) 0 200 400 600 800 1000 1200 1400 Running CB MR ML OR OL FD Distance covered (m) 0 100 200 300 400 500 600 700 HI running CB MR ML OR OL FD Distance covered (m) 0 100 200 300 400 500 Figure 4. Cont. Int. J. Environ. Res. Public Health 2020, 17, 6768 8 of 15 Int. J. Environ. Res. Public Health 2020, 17, x 8 of 14 Sprinting CB MR ML OR OL FD Distance covered (m) 0 10 20 30 40 50 # # # Figure 4. Distance covered in different locomotion categories by playing position in defense. # Statistical differences with front defender p ≤ 0.05; § statistical differences with center back p ≤ 0.05. Legend: center back = CB; mid right = MR; mid left = ML; outside right = OR; outside left = OL; front center defender = FD. Furthermore, moderate, large and very large effect sizes were found in the different locomotion characteristics by playing positions in defense (Table 3). Table 3. Effect sizes in different locomotion categories by playing position in defense. Playing positions Walking Jogging Running High Running Sprinting CB MR 1.31 OR 0.77 ML 0.68 ML 1.38 MR 1.05 OR 1.19 OR 1 ML 0.93 OL 1.26 OL 0.90 FD 1.11 FD 2.65 FD 0.51 MR OL 0.61 OL 0.52. OR 0.65 FD 0.77 OL 0.84 ML OL 0.65 FD 0.84 OR OL CB 1.07 ML 0.85 MR 0.64 FD MR 1.74 MR 1.46 OR 0.50 CB 4.21 ML 1.89 CB 1.17 MR 1.62 OR 1.45 ML 1.04 ML 2.46 OL 0.71 OR 0.71 OR 1.80 OL 0.56 OL 0.57 Legend: center back = CB; mid right = MR; mid left = ML; outside right = OR; outside left = OL; front center defender = FD. 3.3. Running Pace by Playing Positions When the distance covered in each locomotion category normalized according to the time spent on court in different playing positions during offense were analyzed, the ANOVA showed significant differences for jogging (p = 0.029) and sprint distances (p = 0.045) between the different playing Figure 4. Distance covered in different locomotion categories by playing position in defense. # Statistical differences with front defender p ≤ 0.05; § statistical differences with center back p ≤ 0.05. Legend: center back = CB; mid right = MR; mid left = ML; outside right = OR; outside left = OL; front center defender = FD. Furthermore, moderate, large and very large effect sizes were found in the different locomotion characteristics by playing positions in defense (Table 3). Table 3. Effect sizes in different locomotion categories by playing position in defense. Playing Positions Walking Jogging Running High Running Sprinting CB MR 1.31 OR 0.77 ML 0.68 ML 1.38 MR 1.05 OR 1.19 OR 1 ML 0.93 OL 1.26 OL 0.90 FD 1.11 FD 2.65 FD 0.51 MR OL 0.61 OL 0.52. OR 0.65 FD 0.77 OL 0.84 ML OL 0.65 FD 0.84 OR OL CB 1.07 ML 0.85 MR 0.64 FD MR 1.74 MR 1.46 OR 0.50 CB 4.21 ML 1.89 CB 1.17 MR 1.62 OR 1.45 ML 1.04 ML 2.46 OL 0.71 OR 0.71 OR 1.80 OL 0.56 OL 0.57 Legend: center back = CB; mid right = MR; mid left = ML; outside right = OR; outside left = OL; front center defender = FD. Int. J. Environ. Res. Public Health 2020, 17, 6768 9 of 15 3.3. Running Pace by Playing Positions When the distance covered in each locomotion category normalized according to the time spent on court in different playing positions during offense were analyzed, the ANOVA showed significant differences for jogging (p = 0.029) and sprint distances (p = 0.045) between the different playing positions. However, the post hoc analysis did not show any statistically significant differences (Figure 5). Int. J. Environ. Res. Public Health 2020, 17, x 9 of 14 outside right (p = 0.016; ES = 1.19) and the front defender (p = 0.003; ES = 0.37), as well as a tendency between the central back and the mid left (p = 0.074; ES = 1.20). Offense LP CB RB LB RW LW Normalized distance (m.min -1) 0 10 20 30 40 50 60 70 Walking Jogging Running High Running Sprinting Figure 5. Distance covered in each locomotion category normalized according to the time spent in court in different playing positions during offense. Legend: left wing = LW, left back = LB, center back = CB; line player = LP; right back = RB; right wing = RW. Defense Normalized distance (m.min -1) 10 20 30 40 50 60 Walking Jogging Running High Running Sprinting § § Figure 5. Distance covered in each locomotion category normalized according to the time spent in court in different playing positions during offense. Legend: left wing = LW, left back = LB, center back = CB; line player = LP; right back = RB; right wing = RW. When the distance covered in each locomotion category normalized according to the time spent on court in different playing positions during defense was analyzed, the ANOVA showed significant differences only for high-intensity running (p = 0.038) between the different playing positions (Figure 6). Post hoc analysis showed significant differences in this category between the central back and the outside right (p = 0.016; ES = 1.19) and the front defender (p = 0.003; ES = 0.37), as well as a tendency between the central back and the mid left (p = 0.074; ES = 1.20). Int. J. Environ. Res. Public Health 2020, 17, 6768 10 of 15 LP CB RB LB RW LW Figure 5. Distance covered in each locomotion category normalized according to the time spent in court in different playing positions during offense. Legend: left wing = LW, left back = LB, center back = CB; line player = LP; right back = RB; right wing = RW. Defense CB MR ML OR OL FD Normalized distance (m.min -1) 0 10 20 30 40 50 60 Walking Jogging Running High Running Sprinting § § Figure 6. Distance covered in each locomotion category normalized according to the time spent in court in different playing positions during defense. § Statistical differences with center back p ≤ 0.05. Legend: center back = CB; mid right = MR; mid left = ML; outside right = OR; outside left = OL; front center defender = FD. Figure 6. Distance covered in each locomotion category normalized according to the time spent in court in different playing positions during defense. § Statistical differences with center back p ≤ 0.05. Legend: center back = CB; mid right = MR; mid left = ML; outside right = OR; outside left = OL; front center defender = FD. 4. Discussion The aim of this study was to analyze on-court demands of handball players during the VELUX EHF FINAL4 to define time–motion characteristics (played time, covered distances) both in offense and defense, including position-specific demands and differences among them. Significant differences were found between offense and defense in the walking and jogging categories. In offense, significant differences were established in the high-intensity running category between LW and other playing positions. In defense, differences were also identified for CB in walking, and the FCD in sprinting when compared to other playing positions. Several studies have analyzed handball games differentiating intensity categories [3,5,11,17,19,23,24], although they have taken into account neither all playing positions nor the different phases of the game. In this regard, there is a broad consensus among researchers on the need to establish certain categories when analyzing players’ movements, ranging from low-intensity (standing, walking, jogging), medium-intensity (running) and high-intensity (HIrunning, sprinting) situations. However, little consensus exists on the speed ranges for defining the different categories, which makes it difficult to compare between the different studies. For the analysis of the intensity at which the player moves across the field, the categorization proposed by Cardinale et al. [3] used for the Qatar WCh 2015 study was applied. Analyzing the results by locomotion categories, we found that for the offense, the longest distance was performed in jogging (47.06%), while defenders covered a greater distance in the walking category. When comparing offense and defense, results showed a trend for significance in the HIrunning category. These results highlight the needs to differentiate the characteristics of the game phases. As we do not have more studies Int. J. Environ. Res. Public Health 2020, 17, 6768 11 of 15 to compare these results, we present an analysis of the general data with respect to other research carried out. Globally, players covered similar distances both in offense and defense. These results are in line with the study reported by Michalsik et al. [11], which also analyzed distances by phases of the game. Other studies [3,13] have analyzed these variables without differentiating the game phases. They were conducted during the final phase of top level international men´s competition, the Men’s World Handball Championships of Germany (2007) and Qatar (2015), respectively. Our results are consistent with these studies regarding the total average distances covered, being 8% higher in the Germany WCh and 1% smaller in the Qatar WCh 2015. Other studies that analyzed men´s top national leagues showed greater covered distances than ours, ranging from 3157 m on average in the analysis of the German first division [25] to 3627 m in the main Danish league [4], or 4370 m in the main Portuguese league [12]. The problem is that, in most of these studies, standard deviations from the average are very high, thus complicating the use of this criterion as a performance control measure. A better criterion, with a practical application for coaches, is to normalize the data according to a players’ time on court [3,6]. When the data are normalized according to the time spent on court, which in the case of national leagues is larger (above 40 min) than in international competitions, the average running pace does not differ too much between the studies, that is, by about 10%. According to this criterion, at the highest level, a handball player covered between 70 and 90 m·min−1. This data normalization facilitates the comparison between studies and allows the coaches to dispose of a workload reference regarding locomotion activities. Regarding time spent on court, offense and defense presented average times of 15.69 min and 15.40 min, respectively. Summing up both phases, values were similar to those presented by Luig et al. [13] and Cardinale et al. [3], being equal to 32 min and 37 min respectively, but lower that those described in national league studies [3,4,14,15], where values over 40 min were found. Regarding the average running pace values in offense and defense (89 m·min−1 and 81 m·min−1, respectively), our results did not match those presented by Michalsik et al. [11], although the differences were less than 10%. In this line, the results showed in the different studies are very heterogeneous, possibly due to the methodology used, the instrument of control, level of play, the category and the competition analyzed. Regarding specific positions, our main finding was that there were differences among players’ locomotion categories in each playing position, both in offense and defense, which implies a need for greater differentiation and individualization in the training load according to the different playing positions. In the case of the offense, left wing players showed the highest covered distances in the high-intensity running and sprinting categories in relation to the other specific positions, and the right wings covered longer distances than right back players for the HIrunning category. We should bear in mind that it is possible that some statistical comparisons in our study may show no statistical significance even when the means of the two groups are quite different, because the sample is small and the SD are high. In this context, ES might be a good indicator not only of the magnitude of the changes but also of the associations that are likely to present significant differences in larger samples. The ES in the HIrunning and sprinting categories reinforces the observed statistical differences, since they present a large and very large ES. Therefore, we can conclude that the wings have different demands for high-intensity activities than the rest of the players in the offensive phase. These results are reinforced by the idea that these players are those who are responsible for performing most of the counter-attacks or to reach position in the first wave of the fastbreak, which are the fastest actions of the offensive phase. These data are in line with previously reported data by Michalsik et al. [4] and Povoas et al. [5] for the offensive phase. For the defensive phase, the results showed a similar behavior to that in the offense. It is the defense-specific playing positions (CBs and FCDs) that have the highest values during the defensive phase, showing high ES values. CBs have higher values in the covered distance for Int. J. Environ. Res. Public Health 2020, 17, 6768 12 of 15 the walking category than those found for mid defenders. These results are consistent with the work performed by CBs during this phase, as they are the players who move depending on the area where the ball is directed, and these displacements are usually of low intensity. In the same line, when the locomotion categories were normalized according to the time spent on court, we observed that in the HIrunning category, CBs covered less distance than the ORs and FCDs, which indicates that CBs carried out most of their activities in the low-intensity categories. In the sprinting category, the highest covered distances correspond to the FCDs, which showed significant differences with the CBs and Mid defenders, with a large and very large ES. These data are consistent with the specific role of this player, who carries out his activities mainly in the front defense line, covering the central area of the defense, moving from side to side. Again here, as an application to training, coaches should differentiate training by specific positions, for example, by creating very intense tasks for FCDs and wings. To our knowledge, this is the first time that specific defensive playing positions have been analyzed, so these data cannot be compared. It provides a novel and in-depth knowledge of the real needs of this phase of the game. The total distances covered by each playing position are smaller, for both phases of the game, when compared to those reported by Michalsik et al. [4]. Variations in the methodology and the different technology used, as well as the different competitions analyzed in both studies, may account for these differences. On the one hand, LPS technology that allows load individualization and automation through micro sensors [16] was used for this study. However, Michalsik et al. [4] performed a manual estimation of intensities based on distance references on the court, following the player’s individual monitoring with a camera. Differences also existed between these two studies regarding the time spent on the court. Danish players stayed clearly longer on court in all cases. This may be because we are comparing a national league with a European final league tournament. The game-sharing times can be altered by the number and quality of players taking part in the Velux EHF Final 4, which gathers the best teams and players in the world, as well as the nature of the competition (final phase of the biggest club tournament). Therefore, the time sharing, a larger use of rolling substitutions as well as more rotations to maintain the intensity of the game may be greater than in a national league. Other studies have also analyzed the playing positions in men’s championships with senior high level players, but have neither differentiated the phases of the game nor analyzed each individual playing position. Some of these studies have also shown greater high intensity values by the wing players compared to the rest of the playing positions [13,25].What seems to be clear is that, in general, studies that analyze the maximum distance covered per playing position varies widely, making a comparison with our study difficult because of the different procedures used [3,12,13,25]. A trend that can be observed in the available studies is that regardless of the category, procedure used, level of play or gender, all studies analyzing locomotion activities in handball have in common the differentiation of loads according to positions [3,5,6,10,19,25]. For this reason, given the great variability according to playing position, we propose to differentiate the physical work according to the role adopted in the game, in line with the conclusions of most studies. Several limitations were found that made it difficult to discuss this study and compare it to others. The first is the small number of works focusing on the playing load in high-level men’s professional handball in final tournaments, such as the Velux EHF Final 4. Moreover, it is complicated to compare studies because of the lack of unified criteria to determine locomotion categories. It would be necessary and essential to standardize criteria so that they could be taken into consideration in future studies. Other limitations present in the work are related to the lack of development of multivariate analysis techniques. Additionally, only one championship has been analyzed, corresponding to a high performance level in the senior men’s category. Further studies will be needed to deepen our knowledge of handball’s total load through the individualized use of sensor technology (EPTS), which allows us to learn about the other previously mentioned parameters. In addition, it would be advisable to combine the advances in the physical Int. J. Environ. Res. Public Health 2020, 17, 6768 13 of 15 understanding of the game with its impact on the game’s technical–tactical component. Furthermore, there is also a need for extending the analysis to other competitions, categories and gender. 5. Conclusions Offensive players covered longer distances in the jogging category and defensive players in the walking category. Profile differences existed among players’ position activity, both in offense and defense. In fact, more activity in high-intensity categories was found for wing players in offense. In the case of defense, it was the CB that covered the largest distances in low-intensity categories, and the FCD covered most of the distance in high-speed categories. Practical Applications Our findings suggest the need to differentiate the training load specifically for each position, and differentiate between the phases of the game, creating specific exercises, that is, very short work (less than 2 m of displacement) involving high-intensity movements (above 5 m−1) and repeated in a random way over time, with high active rest time between sets. For example, you can do integrated training with simulated game situations, where FCDs in offense and LWs in defense have greater involvement. Another possibility is to design tasks that raise the fatigue threshold at each position and phase to check their impact on the game. These integrated exercises can also include explosive resistance training that improves performance in decisive final actions such as 1v1, blocks, etc. In addition, knowing the specific load of a top-level tournament will allow coaches to determine the maximum levels of physical requirements in elite handball and set them as references based on the category. Furthermore, knowing that the different demands for the playing positions are differentiated will allow coaches to individualize and plan their workouts accordingly, as well as consider it in the match load dosing and in players’ substitutions, for example, if possible, giving more rest to the LWs and the FCDs to maintain the level of intensity. At a high performance level, coaches should work to improve training control. The normalization of locomotion activity data allows disposing of a workload reference, in addition to facilitating the comparison between sessions. Very little information is available about the demands of the game in the different national leagues. Currently, the system is only being used in the German Bundesliga. The use of the system in the VELUX EHF Champions league would undoubtedly provide us with new relevant information about the highest competition among European clubs. In the future, studies could be carried out to analyze players’ rotations in offense and defense. In addition, future research should relate workload on the court to the workload outside the court, such as in the fitness room. Finally, it is also possible that the results obtained in this study are useful for the future design of more specific physical tests related to the demands of the game. Author Contributions: Conceptualization, C.M., H.V. and F.S.S.; data curation, B.P.; formal analysis, J.T.M., S.B. and S.A.; investigation, C.M.; methodology, J.T.M. and B.P.; resources, J.M.C.T.; supervision, H.V., F.S.S. and L.J.C.R.; writing—original draft, C.M., H.V. and L.J.C.R.; writing—review & editing, J.T.M. and C.F. All authors have read and agreed to the published version of the manuscript. Funding: This research was partially funded by Séneca-CARM—grant number 20928/PI/18, by MINECO/FEDER grant number PID2019-108336GB-100 and by Consejo Superior de Deportes, grant number 24/UPB/19 Conflicts of Interest: The authors declare no conflict of interest. References 1. Manchado, C.; Tortosa-Martinez, J.; Vila, H.; Ferragut, C.; Platen, P. Performance factors in women’s team handball: Physical and physiological aspects—A review. J. Strength Cond. Res. 2013, 27, 1708–1719. [CrossRef] [PubMed] 2. Fasold, F.; Redlich, D. Foul or no Foul? Effects of Permitted Fouls on the Defence Performance in Team Handball. J. Hum. Kinet. 2018, 63, 53–59. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2020, 17, 6768 14 of 15 3. Cardinale, M.; Whiteley, R.; Hosny, A.A.; Popovic, N. Activity Profiles and Positional Differences of Handball Players During the World Championships in Qatar 2015. Int. J. Sports Physiol. Perform. 2017, 12, 908–915. [CrossRef] [PubMed] 4. Michalsik, L.B.; Aagaard, P.; Madsen, K. Locomotion characteristics and match-induced impairments in physical performance in male elite team handball players. Int. J. Sports Med. 2013, 34, 590–599. [CrossRef] 5. Povoas, S.C.; Ascensao, A.A.; Magalhaes, J.; Seabra, A.F.; Krustrup, P.; Soares, J.M.; Rebelo, A.N. Physiological demands of elite team handball with special reference to playing position. J. Strength Cond. Res. 2014, 28, 430–442. [CrossRef] 6. Luteberget, L.S.; Spencer, M. High-intensity events in international women’s team handball matches. Int. J. Sports Physiol. Perform. 2017, 12, 56–61. [CrossRef] 7. Pereira, L.A.; Nimphius, S.; Kobal, R.; Kitamura, K.; Turisco, L.A.; Orsi, R.C.; Abad, C.C.; Loturco, I. Relationship between change of direction, speed, and power in male and female national olympic team handball athletes. J. Strength Cond. Res. 2018, 32, 2987–2994. [CrossRef] 8. Ortega-Becerra, M.; Belloso-Vergara, A.; Pareja-Blanco, F. Physical and physiological demands during handball matches in male adolescent players. J. Hum. Kinet. 2020, 72, 253–263. [CrossRef] 9. Kniubaite, A.; Skarbalius, A.; Clemente, F.M.; Conte, D. Quantification of external and internal match loads in elite female team handball. Biol. Sport 2019, 36, 311–316. [CrossRef] 10. Karcher, C.; Buchheit, M. On-court demands of elite handball, with special reference to playing positions. Sports Med. 2014, 44, 797–814. [CrossRef] 11. Michalsik, L.B.; Madsen, K.; Aagaard, P. Match performance and physiological capacity of female elite team handball players. Int. J. Sports Med. 2014, 35, 595–607. [CrossRef] [PubMed] 12. Povoas, S.C.; Seabra, A.F.; Ascensao, A.A.; Magalhaes, J.; Soares, J.M.; Rebelo, A.N. Physical and physiological demands of elite team handball. J. Strength Cond. Res. 2012, 26, 3365–3375. [CrossRef] [PubMed] 13. Luig, P.; Manchado, C.; Pers, J.; Kristan, M.; Schander, I.; Zimmermann, M.; Henke, T. Motion characteristics according to playing positions in international men’s team handball. In Proceedings of the 13th Annual Congress of the European College of Sports Science, Estoril, Potugal, 9–12 July 2008; pp. 241–247. 14. Hoppe, M.W.; Baumgart, C.; Polglaze, T.; Freiwald, J. Validity and reliability of GPS and LPS for measuring distances covered and sprint mechanical properties in team sports. PLoS ONE 2018, 13. [CrossRef] [PubMed] 15. Link, D.; Weber, M.; Linke, D.; Lames, M. Can positioning systems replace timing gates for measuring sprint time in ice hockey? Front. Physiol. 2019, 9, 1882. [CrossRef] 16. Fleureau, A.; Lacome, M.; Buchheit, M.; Couturier, A.; Rabita, G. Validity of an ultra-wideband local positioning system to assess specifc movements in handball. Biol. Sport 2020. [CrossRef] 17. Gonzalez-Haro, P.J.; Gómez-Carmona, C.D.; Bastida-Castillo, A.; Rojas-Valverde, D.; Gómez-López, M.; Pino-Ortega, J. Analysis of playing position and match statusrelated differences in external load demands on amateur handball: A case study. Rev. Bras. Cineantropom. Hum. 2020, 22, e71427. [CrossRef] 18. Hansen, C.; Sanz-Lopez, F.; Whiteley, R.; Popovic, N.; Ahmed, H.A.; Cardinale, M. Performance analysis of male handball goalkeepers at the World Handball championship 2015. Biol. Sport 2017, 34, 393–400. [CrossRef] 19. Belka, J.; Hulka, K.; Safar, M.; Weisser, R.; Samcova, A. Analyses of time-motion and heart rate in elite female players (U19) during competitive handball matches. Kinesiology 2014, 46, 33–43. 20. Higham, D.G.; Pyne, D.B.; Anson, J.M.; Eddy, A. Movement patterns in rugby sevens: Effects of tournament level, fatigue and substitute players. J. Sci. Med. Sport 2012, 15, 277–282. [CrossRef] 21. Akenhead, R.; Hayes, P.R.; Thompson, K.G.; French, D. Diminutions of acceleration and deceleration output during professional football match play. J. Sci. Med. Sport 2013, 16, 556–561. [CrossRef] 22. Hopkins, W.G. A Scale of Magnitudes for Effect Statistics. A New View of Statistics. Available online: www.sportsci.org/resource/stats/effectmag.html (accessed on 8 September 2020). 23. Chelly, M.S.; Hermassi, S.; Aouadi, R.; Khalifa, R.; Van den Tillaar, R.; Chamari, K.; Shephard, R.J. Match Analysis of elite adolescent team handball players. J. Strength Cond. Res. 2011, 25, 2410–2417. [CrossRef] [PubMed] Int. J. Environ. Res. Public Health 2020, 17, 6768 15 of 15 24. Pers, J.; Bon, M.; Kovacic, S.; Siblia, M.; Dezman, B. Observation and analysis of large-scale human motion. Hum. Mov. Sci. 2002, 21, 295–311. [CrossRef] 25. Buchel, D.; Jakobsmeyer, R.; Doring, M.; Adams, M.; Ruckert, U.; Baumeister, J. Effect of playing position and time on-court on activity profiles in german elite team handball. Int. J. Perform. Anal. Sport 2019, 19, 832–844. [CrossRef] © 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

High-Performance Handball Player's Time-Motion Analysis by Playing Positions.

09-17-2020

Manchado, Carmen,Tortosa Martínez, Juan,Pueo, Basilio,Cortell Tormo, Juan Manuel,Vila, Helena,Ferragut, Carmen,Sánchez Sánchez, Francisco,Busquier, Sonia,Amat, Sergio,Chirosa Ríos, Luis Javier

eng

PMC7423319

RESEARCH ARTICLE Effects of preferred music on physiological responses, perceived exertion, and anaerobic threshold determination in an incremental running test on both sexes Felipe Marroni RasteiroID, Leonardo Henrique Dalcheco Messias, Pedro Paulo Menezes Scariot, João Pedro Cruz, Rafael Lucas Cetein, Claudio Alexandre Gobatto, Fu´lvia Barros Manchado-Gobatto* Laboratory of Applied Sport Physiology - LAFAE, School of Applied Sciences, University of Campinas, UNICAMP, Limeira, São Paulo, Brazil * fgobatto@unicamp.br Abstract This study aimed to investigate and compare the effects of preferred music on anaerobic threshold determination in an incremental running test, as well the physiological responses and perceived exertion at this intensity, in physically active men and women. Additionally, by using area under the curve (AUC) analysis of the parameters of interest during the graded test, we studied the effects of music at two physiological moments—before and after anaerobic threshold intensity (iAT)—in men and women. Twenty (men = 10; women = 10) healthy and active participants completed four visits to the laboratory. The first and second sessions were used for sample characterization. In the third and fourth sessions, partici- pants performed an incremental running test (started at 7 km.h-1 with increments of 1 km.h-1 at each 3-minute stage) under preferred music and non-music conditions. Blood lactate ([Lac]), heart rate (HR), and perceived exertion were measured by two scales (RPEBorg and the estimation of time limit – ETL) during all tests, and the total time of effort (TT) was consid- ered as performance. Individual curves of the “intensity vs blood lactate” analyzed by the bissegmentation method provide the iAT and the AUC of [Lac], HR, RPEBorg, and ETL before and after the iAT attainment were calculated. The iAT for men (non-music: 11.5 ±0.9km.h-1 vs music: 11.6±1.1km.h-1) and women (non-music: 9.8±0.7km.h-1 vs music: 9.7 ±0.7km.h-1) was not affected by music, and for both sexes, there was no difference between non-music and music conditions in all variables obtained at iAT. The AUC of all variables were not affected by music before the iAT attainment. However, [Lac], HR, and RPEBorg pre- sented higher values of AUC after iAT for the female group with preferred music. This may be due to the fact that 70% of women have increased TT under music conditions. Overall, preferred music did not affect the iAT determination in an incremental running test. How- ever, some physiological responses and perceived exertion after iAT of female subjects seems to be influenced by preferred music. PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 1 / 16 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Rasteiro FM, Messias LHD, Scariot PPM, Cruz JP, Cetein RL, Gobatto CA, et al. (2020) Effects of preferred music on physiological responses, perceived exertion, and anaerobic threshold determination in an incremental running test on both sexes. PLoS ONE 15(8): e0237310. https://doi.org/10.1371/journal.pone.0237310 Editor: Daniel Boullosa, Universidade Federal de Mato Grosso do Sul, BRAZIL Received: September 10, 2019 Accepted: July 23, 2020 Published: August 12, 2020 Copyright: © 2020 Rasteiro et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the manuscript and its Supporting Information files. Funding: We thank the subjects for participating in the procedures. The study was partially supported by São Paulo Research Foundation – FAPESP (2012/06355-2, 2016/50250-1, 2018/05821-6, and 2019/10666-2), National Council for Scientific and Technological Development – CNPq (307718/ 2018-2, 308117/2018-2), and Coordenac¸ão de Introduction The effect of music on exercise was investigated by Ayres [1], who described the influence of a music band playing during a six-day bicycle race. However, only in recent decades has there been a significant increase in research on the importance of this topic to the performance in different physical activities and exercise situations [2–8]. Many investigations involving the effect of music on exercise have been focused on some contributing factors of music, such as genre [9], rhythm [10], tempo [11, 12], and auditory–motor synchronization [5, 8]. On the other hand, there are still few studies in which the musical preference of the participants is freely guaranteed. It has been speculated that emotional components of the preferred music could be an effective aid to increase personal motivation [13–15]. The literature presents conflicting information regarding the influence of music (preferred or not) on exercise at different intensity domains [16–19]. Based on evidence that music can distract exercisers from the unpleasant and uncomfortable sensations associated with physical effort [8, 20], it should be expected that preferred music would help not only in exercise at moderate or heavy domains but especially at severe domain, in which physiological instability and exhaustion meet. Although researchers have concerned with distinct exercise characteris- tics in experiments with music (e.g. type, intensity and volume) [21–24], few investigations considered this context regarding the preferred music [16, 17]. To the best of our knowledge, despite the significance of the anaerobic threshold intensity (iAT) determination to define the aerobic–anaerobic transition and thus to obtain an accurate performance diagnosis on the exercise domains [25, 26], none investigations followed this way using the preferred music. Among many evaluative protocols, the incremental (graded test) is the most common pro- cedure for iAT determination in laboratory and field conditions [27]. During this application, the intensity is incremented, inducing an exponential behavior of blood lactate ([Lac]) accu- mulation. Therefore, the highest workload that still leads to an equilibrium between lactate production and removal is termed as the iAT [25] and can be determined by reliable mathe- matical analysis, such as the bissegmentation of two linear regressions along with linear inter- polation [28–30]. Before the iAT attainment (moderate and heavy domains), physiological responses are not expected to abruptly increase, reflecting a stability that is favorable to main- tenance of exercise [31, 32]. On the other hand, at intensities higher than iAT (severe domain), the physiological balance is gradually lost, leading to exhaustion [33]. We believe that investi- gations on the physiological responses (e.g., heart rate and [Lac]) and perceived exertion (obtained by perception scales) at two moments during the incremental test (before and after iAT) can improve the understanding of the effects of music at these different intensity domains. For this, the area under the curve analysis (AUC) commonly used in other scientific approaches [34, 35] seems to be an interesting method. Although there are studies documenting the effects of music on exercise in both men and women [5, 11, 12, 36–40], there is still an ongoing debate about sex differences in music pro- cessing. It is reasonable to consider that music’s effects on physical performance could be sex- dependent, as there are reports showing that men and women seem to differ in their percep- tion of music [41–45]. In this way, Macone et al. [39] and Cole and Maeda [36] have demon- strated that women, but not men, had increased physical performance in a music condition compared to a non-music condition. Thus, an important question to be answered is whether music affects differently the physiological responses, perceived exertion, and, consequently, the iAT determination in men and women in the same physical condition (e.g., active individ- uals but non-athletes). Still, the understanding of the effects of music on men or women depending on the intensity domains (before or after iAT) seems to add knowledge for exercise PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 2 / 16 Aperfeic¸oamento de Pessoal de Nı´vel Superior - Brasil (CAPES) - Finance Code 001. So, we would like to express our thanks for this support. Competing interests: The authors have declared that no competing interests exist. prescription in these conditions. Obviously, in this sense of application, the music to be used needs to be the subject’s preference. This study aimed to investigate and compare the effects of preferred music on anaerobic threshold determination in an incremental running test, as well as the physiological responses and perceived exertion at this intensity, in physically active men and women. Additionally, by using AUC analysis of the parameters of interest during the graded test, we studied the effects of music at two physiological moments (before and after iAT) in men and women. Materials and methods Study design Participants were requested to maintain the same individual hydration/food habits and avoid alcohol/caffeine ingestion as well as hard physical activity at least 96 hours prior to testing. Twenty healthy, non-athletes, non-smoking, and active male (n = 10; age = 23±2 years; body mass = 73.3±11.7 kg; height = 175±1 cm; body fat = 8.5±2.3%) and female (n = 10; age = 20±1 years; body mass = 59.7±5.3 kg; height = 165±1 cm; body fat = 18.4±3.6%) were selected for this study. As inclusion criteria, individuals should be active and experienced with at least two years of weekly practice in running exercise. The present study was conducted according to the norms of Helsinki and previously approved by the Research Ethics Committee of The School of Medical Sciences, located at the University of Campinas (protocol number – 64648617600005404). Four visits to the laboratory were completed (Fig 1). The first session was conducted to explain the study’s procedures and obtain agreement to participate in the research, which was approved by the university’s local ethics committee. Moreover, at the end of this session, the subjects were asked to provide 10 songs they consider motivational during their daily physical activity. In line with this, the second session was conducted to identify the motivational level of each selected song via the Brunel Music Rating Inventory-2 [46]. During the same session, participants were evaluated for body composition (i.e., lean mass, fat mass, and body fat), physical activity level, and physical activity readiness (PAR-Q) [47]. Skinfold measurements were performed by the same experienced researcher via a clinical adipometer/plicometer (Ces- corf, Cardiomed, PR, BR). Lohman [48] and Jackson and Pollock’s [49] approaches were used to estimate the body composition of the men (i.e., triceps, subscapular, and abdominal skin- folds) and women (i.e., triceps, suprailiac, and thigh skinfolds), respectively. The International Physical Activity Questionnaire (IPAQ) [50] was adopted for analysis of physical activity level (men = 3535±2425 metabolic equivalent-min/week; women = 3568±1860 metabolic equiva- lent-min/week). The third and fourth sessions were dedicated to the exhaustive incremental protocol on a motorized treadmill (Super ATL, Inbramed, RS, BR). All procedures were conducted in a con- trolled environment (temperature = 22˚C±1˚C; luminosity = ~300lx). Additionally, these ses- sions were conducted in an isolated room (length = 4.83 m; width = 2.11 m). Therefore, participants did not maintain contact with other people except for the evaluators, who com- municated (when strictly necessary) through gestures. Moreover, information regarding the duration of the test or stages was avoided. These sessions were randomized and separated by 48–72 hours (S1 File). In one of the sessions, the protocol was performed under non-music conditions. In the remaining session, subjects were allowed to listen to their preferred music during the exhaustive incremental protocol. In both evaluations, [Lac], HR, RPEBorg, ETL, and total time (TT) were analyzed. PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 3 / 16 Music classification The BMRI-2 evaluates the motivational quality of music during exercise. It is comprised of six items, each concerning a musical component (rhythm, style, melody, tempo, instrumentation, and beat). Each item is comprised of a seven-point Likert scale, where 1 is “strongly disagree” and 7 is “strongly agree" [46]. Taking into account hygiene and preference aspects, each indi- vidual was asked to bring his or her own headphones. For BMRI-2 application, the previously selected songs were inserted into a musical player (iPod Shuffle A1373, Apple, SP, BR), which was also used in the incremental protocol. The music’s volume was standardized at five clicks below the maximum, ranging from 70– 85 dB. These strands were previously tested for the application of the BMRI-2, and it was found that the aforementioned range would be ideal for working within the present study pre- cisely because it assures auditive safety. The instrument application consisted of the individual playing of the song for 90 seconds. Subsequently, the scale was applied and the song score established. In order to eliminate the effect of listening to the previous song, a concentration grid [51] was applied between songs. These processes were repeated until the establishment of the score referring to the 10 songs. During the incremental protocol, the songs were ranked according to the score previously obtained, with the highest score at the top of the playlist and the others placed in descending order (S2 File). Since all songs were considered preferred by the evaluated participants, the music tempo was not controlled. However, all bpm values are presented in the S3 File. Fig 1. Experimental design adopted in the study. a) First session was conducted to explain the study’s procedures and obtain agreement to participate in the research. Moreover, subjects were instructed to bring 10 songs for exercising. Subsequently, during the second visit, subjects were evaluated for body composition, physical activity, and readiness. In the same session, the motivational quotient of the 10 songs provided by the subjects was determined via the Brunel Music Rating Inventory– 2 (BMRI-2). The incremental protocol in non-music and preferred music conditions was randomly conducted in the third and fourth sessions. b) Incremental protocol started at 7 km.h-1 with increments of 1 km.h-1 in 3-minute stages. Blood samples were collected at rest and at the end of each stage to [Lac] analysis. During the same interval, two perceived exertion scales were applied (RPEBorg and ETL). HR was measured throughout the protocol. c) Anaerobic threshold intensity (iAT) was determined by the intersection between two linear fits resulting from the bissegmentation method. https://doi.org/10.1371/journal.pone.0237310.g001 PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 4 / 16 Blood lactate concentration and heart rate analysis Capillarized blood samples (25 μL) were taken from the earlobe and deposited into microtubes (Eppendorf 1.5 ml) containing 50 μl of NaF. The [Lac] was analyzed by the electrochemical method using a lactimeter YSI2300-STAT-Plus (Yellow Springs, OH, USA). The participants’ HR was recorded (beat to beat) using Polar heart monitors (Polar, RS800, RJ, BR). Data were recorded during all protocols. Perceived exertion scales Two psychometric scales were considered for measurement of perceived exertion. The scale originally proposed by Borg [52] with a range of 6–20 (RPEBorg) was adopted. Moreover, the estimation of time limit (ETL) scale proposed by Garcin et al. [53] was also considered. Incremental protocol The incremental protocol started at 7 km.h-1, with increments of 1 km.h-1 in 3-minute stages. The slope of the treadmill was maintained at 1% during all tests. At the end of each stage, the effort was interrupted for 30 seconds for blood collection. During this interval, participants indicated with their fingers the perceived exertion in two psychometric scales. In both tests (non-music or music), the participants used the same auricular headsets adopted to answer the BMRI-2 in the second session. For determination of the iAT, individual curves of intensity (km.h-1) vs blood lactate (mM) were plotted. After visual inspection, performed by two experienced researchers, the bisseg- mentation analysis proceeded and iAT was identified by the intersection between fits [28, 29]. [Lac], HR, RPEBorg, and ETL at iAT ([Lac]iAT, HRiAT, RPEBorg iAT, and ETLiAT, respectively) were determined by linear interpolation. Relativization in percentage (%) was performed by dividing the iAT by the maximum value recorded of intensity (ipeak) and then multiplied by 100 (iAT [% ipeak]). The same procedure was applied to calculate the [Lac]iAT (% [Lac]peak) and HRiAT (%HRmax). TT was considered when the individual achieved maximum HR (i.e., 220-age) [54] or asked to stop (voluntary exhaustion). To calculate the time taken to reach the anaerobic threshold (TBiAT [%TT]) as well as the remaining effort time (TAiAT [%TT]), the intensity (km.h-1) and time (total seconds of each stage) were plotted as x axis and y axis, respectively. Thus, the first-degree equation was replaced by known values, identifying the spe- cific time that the iAT occurred. Area under the curve analysis Measurements obtained multiple times from the incremental protocol were also used to inves- tigate whether music would be able to differently influence the responses before and after iAT attainment. Following the iAT determination, individual curves of intensity (km.h-1) vs the variables studied ([Lac], HR, RPEBorg, and ETL) were plotted. The curve was divided into two moments, before and after iAT. Then the trapezoidal method was applied stage by stage until reaching the stage corresponding to iAT. The AUC values obtained for each stage interval were then summed, and the total was considered as the AUC before iAT. The same was applied in the stages after the iAT. Fig 2 indicates an individual example of the AUC analysis of the heart rate variable. Statistical analysis Data (S4 File) were calculated and analyzed using STATISTICA 7.0. The figures were elabo- rated by the software GraphPad Prism 5. Data are presented as mean and standard deviation PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 5 / 16 of the mean. The normality and homogeneity of the data were confirmed by the Shapiro–Wilk and Levene tests, respectively. Two-way ANOVA was adopted to determine the effects of music (non-music vs music) and sex (male vs female), as well as their interaction (music vs sex) on parameters obtained from the incremental test. AUC data were analyzed by repeated measures ANOVA considering the effects of music (non-music vs music) and moment (before vs after iAT), as well as their interaction (music vs moment). The Newman–Keuls post hoc analysis was adopted in all cases. The relationship between variables was analyzed using Pear- son’s correlation. In all cases, the level of significance was set at 5%. Results Preferred music did not influence iAT determination through the incremental test, regardless of sex (Table 1). No significant effect of music on any of the variables studied was detected by two-way ANOVA. Additionally, we found an effect of sex on iAT and TT, showing that males exhibit higher aerobic fitness and physical performance (iAT and TT) than females. However, the female group presented higher HRiAT values than the male group in the non-music condi- tion, but no significance was observed for the music condition. No interaction effect was observed in any of the variables studied. In addition, most of the variables presented a signifi- cant relationship in the intra-group analysis under non-music and music conditions for the male group (iAT–r = 0.92, p = 0.001; [Lac]iAT−r = 0.79, p = 0.006; [Lac]iAT (% [Lac]peak)– r = 0.80, p = 0.005; TT–r = 0.93, p = 0.001; RPEBorg iAT−r = 0.65, p = 0.042; ETLiAT−r = 0.91, p = 0.001). Likewise, [Lac]iAT (r = 0.83, p = 0.003), HRiAT (r = 0.98, p = 0.001), HRiAT (% HRmax) (r = 0.97, p = 0.001), TT (r = 0.86, p = 0.001), RPEBorg iAT (r = 0.69, p = 0.028), and ETLiAT (r = 0.87, p = 0.001) were significantly correlated for females. Individual responses regarding TT can be seen in Fig 3. Figs 4 and 5 show a comparative analysis of the AUC from before and after the iAT attain- ment, under non-music and music conditions, for both sexes. For the male group, [Lac] pre- sented a significant difference only for the moment effect (Fig 4a). On the contrary, HR did Fig 2. Individual example of the AUC analysis by Heart Rate (HR) responses. (a) Present the HR responses during the incremental running test under the non-music condition; (b) present the HR responses during the incremental running test under the preferred music condition. The dotted line indicates the anaerobic threshold intensity (iAT) in their respective conditions (non- music and music). Values represent the total area under the curve before and after the iAT. au indicates arbitrary unit. https://doi.org/10.1371/journal.pone.0237310.g002 PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 6 / 16 not present a significant difference when the moments were compared (before and after iAT). In the same way, RPEBorg and ETL did not present any significance among the effects (music, moment, or interaction). On the other hand, [Lac] (Fig 5a), HR (Fig 5b), and RPEBorg (Fig 5c) were significantly higher with the preferred music than non-music condition after iAT attain- ment for female subjects. The same was not observed for the ETL (Fig 5d). Discussion Our main results demonstrate that in general terms, the preferred music did not significantly affect the physiological and perceptual responses during an incremental test, or the iAT deter- mination. However, the significant effect for sex in terms of iAT and TT shows that men had higher aerobic fitness and performance in the incremental test when compared to women. Table 1. Parameters obtained from the incremental protocol performed under non-music and music conditions, in both sexes. Male Female Music Effect Sex Effect Interaction Non-music Music Non-music Music p F p F p F iAT (km.h-1) 11.5 ± 0.9 11.6 ± 1.1 9.8 ± 0.7† 9.7 ± 0.7γ 0.972 0.001 < 0.001 40.344 0.795 0.069 iAT (% ipeak) 74.4 ± 3.0 73.3 ± 3.2 78.9 ± 5.4 77.1 ± 4.9 0.297 1.119 0.004 9.691 0.800 0.065 [Lac]iAT (mM) 3.6 ± 1.0 3.6 ± 0.7 4.6 ± 2.4 5.1 ± 1.6 0.668 0.188 0.016 6.460 0.630 0.236 [Lac]iAT (% [Lac]peak) 48.5 ± 5.0 45.5 ± 6.2 55.3 ± 13.5 53.0 ± 12.8 0.398 0.731 0.032 5.000 0.911 0.013 HRiAT (bpm) 153 ± 10 152 ± 10 164 ± 13† 165 ± 13 0.971 0.001 0.002 11.279 0.829 0.047 HRiAT (%HRmax) 77.4 ± 4.8 77.1 ± 5.0 82.1 ± 6.2 82.5 ± 6.1 0.974 0.001 0.007 8.338 0.821 0.052 TT (s) 1644 ± 248 1710 ± 269 1073 ± 248† 1115 ± 293γ 0.525 0.413 < 0.001 48.395 0.887 0.021 TBiAT (%TT) 60.3 ± 4.4 58.5 ± 5.1 61.1 ± 8.2 59.6 ± 5.6 0.382 0.782 0.618 0.254 0.940 0.006 TAiAT (%TT) 39.7 ± 4.4 41.5 ± 5.1 38.9 ± 8.2 40.4 ± 5.6 0.382 0.782 0.618 0.254 0.940 0.006 RPEBorg iAT (score) 13 ± 1 12 ± 1 13 ± 2 13 ± 1 0.488 0.492 0.041 4.498 0.299 1.113 ETLiAT (score) 12 ± 3 11 ± 3 12 ± 3 12 ± 3 0.374 0.809 0.758 0.097 0.508 0.447 iAT − anaerobic threshold intensity; iAT (% ipeak)–relativization of anaerobic threshold intensity in relation to the maximum intensity reached in protocol; [Lac]iAT−blood lactate concentration at iAT; [Lac]iAT (% [Lac]peak) – relativization of the lactacidemia referring to the iAT in relation to the lactate peak value obtained in the protocol; HRiAT−heart rate at iAT; HRiAT (%HRmax) − relativization of the heart rate referring to the iAT in relation to the product of the equation 220-age; TT– total time effort; TBiAT (%TT) − relativization of the time to reach the iAT in relation to the total time of effort; TAiAT (%TT) − relativization of the total time after reached the iAT in relation to the total time of effort; RPEBorg iAT−rating of perceived exertion at iAT; ETLiAT−estimation of time limit at iAT. γ significant difference between male and female in the preferred music condition. † significant difference between male and female in the non-music condition. Significance was pre-fixed at p  0.05. https://doi.org/10.1371/journal.pone.0237310.t001 Fig 3. Individual results of the total time of effort (TT) obtained from the incremental protocol performed in non-music and music conditions. https://doi.org/10.1371/journal.pone.0237310.g003 PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 7 / 16 Fig 4. AUC analysis of male subjects on the [Lac], HR, RPEBorg, and ETL measured during the incremental test performed in music and non-music conditions. The AUC before and after the iAT in terms of (a) Lactate concentration [Lac], (b) Heart rate (HR), (c) Rate of perceived exertion (RPEBorg), and (d) Estimation of time limit (ETL) were compared. # indicates differences for the moment effect. https://doi.org/10.1371/journal.pone.0237310.g004 Fig 5. AUC analysis of female subjects on the [Lac], HR, RPEBorg, and ETL measured during the incremental test performed in music and non-music conditions. The AUC before and after the iAT in terms of (a) Lactate concentration [Lac], (b) Heart rate (HR), (c) Rate of perceived exertion (RPEBorg), and (d) Estimation of time limit (ETL) were compared. # indicates differences for the moment effect. https://doi.org/10.1371/journal.pone.0237310.g005 PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 8 / 16 Although we did not observe the effect of music on performance, independently of sex, intra- subject analysis revealed that most of the males evaluated (70%) had 2–11% improvement in TT in the presence of preferred music. The same was observed for the females evaluated (70%), who had 2–20% improvement in TT when the graded test was performed listening to preferred music. Moreover, AUC analysis revealed that [Lac] and perceived exertion (i.e., RPEBorg and ETL) are elevated after iAT determination for both sexes (i.e., moment effect). On the other hand, women seem to be more susceptible than men to preferred music after iAT in terms of [Lac], HR, and RPEBorg, and this can partially explain the individual performance improvements. As far as we know, this study is the first to investigate the effect of preferred music in a running incremental test applied for male and female. Effects of sex and music on iAT determination and incremental test outcomes Sexual dimorphisms and gender disparity in sports and exercise science have been highlighted [55]. Few evidences consistently demonstrate that men present higher performance than woman in incremental testing [56, 57]. A higher performance in men than women could be explained by differences in body composition components and their distribution. Men and women may differ in the amount and distribution of body fat [58] as well as lean body mass and body size like stature [59–61]. Hoffman et al. [62] showed that men have higher iAT than women in a cycle-ergometer. Moreover, only 11% of women maintained [Lac] in a steady state in exercise performed above iAT for 30 min. Estradiol may impact [Lac] dynamics in luteal and follicular menstrual phases [63], and Hoffman et al. [62] explained these marked oscilla- tions in [Lac] are due to this ovarian hormone. However, a recent study came to the opposite conclusions, demonstrating that the power output and key physiological variables at maximal lactate steady state were not affected by the menstrual cycle [64]. On the other hand, it is important to state that besides [Lac]iAT, the HRiAT and RPEBorg iAT were also higher in women than men regardless of the adoption of music. We cannot affirm that our female data was affected by the menstrual cycle, but the early [Lac] increase may be associated with the circulat- ing ovarian hormones, although this remains to be elucidated. Overall, although we cannot directly discuss the influence of hormonal status on the incremental test outcomes, we can affirm that, at least in our sample, the oxidative system of the female group was lower than in the male group, and this is not affected by preferred music. Despite the comparisons between sexes had advanced on some scientific questions, the effects of music on the incremental test outcomes are a novel finding. To the best of our knowl- edge, a similar experimental design was not found, and our study provides new insights on this context. Music may affect the central nervous system by downregulating theta waves in brain regions during exercise [65]. Probably through these central mechanisms, the music seems to reduce the perceived exertion during exercise [8]. Other studies have also demonstrated that music can influence peripheral variables [66, 67]. In short, music may have an ergogenic effect on physical exercise [16, 68, 69]. We believe that our data offer two major insights on the music–exercise association. To begin with, music did not affect iAT and related parameters, regardless of sex. This important finding demonstrates the robustness of iAT determination. Moreover, although ANOVA did not reveal any significant effect for music, 70% of the female group and 70% of the male group had 2–20% and 2–11% improvements in TT, respec- tively (Fig 3). Overall, our data suggest that women were more susceptible to music’s effects than men. In a mixed sample, Cole and Maeda [36] showed that only women had better per- formance in running while listening to preferred music. These authors suggested that women pay more attention to music while exercising than men, explaining the divergent outcomes. PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 9 / 16 We must recognize that other music characteristics (e.g., synchronous, asynchronous) are also tested during exercise [2–5, 69–74] and may likely influence sex comparisons. The preferred characteristics were chosen for two main reasons. First, studies have demonstrated their ergo- genic effect on exercise [16, 18, 75, 76], and this model matches our aims. Second, athletes and/or merely active subjects routinely use preferred music during exercise [8, 76–80]; there- fore, our results have relevant practical applications. Although the analysis of preferred music on iAT determination and performance can reveal important outcomes, it does not allow fur- ther insights on the behavior of physiological variables and perceived exertion throughout the incremental test. Therefore, the AUC analysis supports this context. Physiological variables and perceived exertion before and after iAT As far as intensity is incremented during the graded test, [Lac] is expected to abruptly increase when pyruvate oxidation exceeds its maximal rate of production. Therefore, higher [Lac] is expected after iAT attainment when compared to its counterpart. This is confirmed by our [Lac] AUC analysis for both groups (Figs 4a and 5a). However, preferred music may affect, only for the female group, AUC of [Lac] throughout the incremental test, after the iAT attain- ment. On the other hand, this result may be due to the increase of the TT by more than half of the female subjects. Studies analyzing the effects of music on [Lac] are scarce. Eliakim et al. [66] demonstrated that motivational music leads to higher lactate clearance after subjects performed a 6-min run exercise at peak aerobic power. This result was explained by the fact that music kept subjects active after exercise, promoting lactate clearance. This context, however, does not apply to our study, since we measure [Lac] during the incremental test. Although authors have showed that music can influence the central nervous system during exercise [7, 65], we cannot observe a direct relationship between preferred music and myocyte response in terms of lactate production. However, we observed a possible relationship between the preferred music and blood lactate response in the female group (Fig 5a), but further studies are required. Music is capable of modifying the cardiovascular profile during exercise [81, 82]. Distinct from kinetic [Lac], HR increases linearly throughout the incremental test. The similarity of HR AUC between moments can be explained by a slight right-shift on the iAT determination for three subjects. This outcome reduced the AUC of these subjects after iAT attainment and explains the non-significant effect for moment (Fig 4b). The same results are not transposed to women. Music and moment were factors that modulated HR throughout the incremental test- ing. Since women tend to focus on some elements of music more than men [36, 43], it is possi- ble that music increased the HR AUC of women mainly after iAT determination. Moreover, this partially explains why 70% of women had better performance (i.e., TT) in the incremental test with preferred music. The effect of music on perceived exertion during exercise is one of the most discussed [8, 16, 18, 19, 67, 75, 78]. Nakamura et al. [16] showed that preferred music increases cycling dis- tance performed at high intensity. Supported by the psychobiological model, Marcora et al. [83] suggest that exercise tolerance increases by the potential motivation of preferred music; others have supported this hypothesis [19, 77, 78, 81]. Thus, the significant interaction for RPEBorg (Fig 5c) can be explained by the fact that preferred music improved exercise tolerance (TT), leading female subjects to present higher values of AUC. These inferences, however, are aligned only regarding our female subjects, and the same explanation in terms of TT and HR differences for both sexes fits in this case. Lastly, the ETL has been considered an important exercise context [84–86]. However, we do not know to what extent the complexity in estimate exercise duration is affected by music. Thus, our data cannot confirm that ETL is not sensitive to music effects, so further studies are required. PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 10 / 16 Finally, some studies have highlighted the importance of the music tempo on the running cadence [21, 23, 24, 87], but this effect was not considered over the preferred song. Interest- ingly, Dyer and McKune [88] investigated the tempo of individual favorite song on the perfor- mance, psychological and physiological responses of well-trained cyclists in time trial cycling. For a better investigation of the preferred music, the authors modified the music tempo according to three experimental conditions (100, 120 and 140 bpm). The authors observed a negative effect of the fast music tempo (i.e., 140 bpm) on the performance. Although they used a creative alternative to investigate the music tempo during the evaluation, the preferred char- acteristics of the song (for example, style, rhythm and harmony) had to be changed [89], possi- bly generating a different condition of that aimed in our study. For this reason, our group chose to evaluate the “pure effect” of the preferred music (without manipulating any property of music) in an incremental running test with controlled exercise cadence. Future perspectives and limitations In this study, we investigated the effects of preferred music in both sexes. However, despite its importance, the menstrual cycle was not controlled in our experimental design. On the other hand, no female subject waited more than 72 hours to return to the laboratory to perform the second incremental test. Thus, although we cannot affirm that all female subjects performed tests restricted to the follicular or luteal phase, it is possible that huge variations of ovarian hor- mones in systemic circulation did not occur between tests. Future studies are encouraged to investigate if our results can be transposed to other music characteristics (e.g., synchronous and asynchronous) or in other exercise types. Moreover, other physiological measurements during an incremental test, such as oxygen uptake and mus- cle oxygenation, can shed light on the effects of music during exercise. Conclusion In summary, preferred music did not affect the iAT determination in an incremental running test, nor the physiological and perceptive responses at this intensity independently of sex. However, more than half of our female subjects had improved performance in the graded test with the preferred music, which may be more related to responses after iAT (severe domain) in this condition. These outcomes were not found for male subjects. Therefore, the effects of preferred music seem to be more pronounced for female subjects when compared to males. Supporting information S1 File. Parameters obtained from the incremental protocol performed in Trial 1 and Trial 2. (DOCX) S2 File. Table with descriptive data of the average and standard deviation, as well as per- cent in relation to the maximum score (i.e. 42 points), of each song score (BMRI-2) in their respective position in the playlist, as well as the mean value of the 10 songs. (DOCX) S3 File. Table with descriptive data of the average and standard deviation of each music tempo (bpm) in their respective position in the playlist, as well as the mean value of the 10 songs. (DOCX) PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 11 / 16 S4 File. (XLSX) Acknowledgments We would like to thank the subjects for the participation on the procedures. Author Contributions Conceptualization: Fu´lvia Barros Manchado-Gobatto. Data curation: Felipe Marroni Rasteiro, Leonardo Henrique Dalcheco Messias. Formal analysis: Felipe Marroni Rasteiro, Leonardo Henrique Dalcheco Messias, Pedro Paulo Menezes Scariot. Funding acquisition: Claudio Alexandre Gobatto, Fu´lvia Barros Manchado-Gobatto. Investigation: Felipe Marroni Rasteiro. Methodology: Felipe Marroni Rasteiro, Leonardo Henrique Dalcheco Messias, Pedro Paulo Menezes Scariot, João Pedro Cruz, Rafael Lucas Cetein, Claudio Alexandre Gobatto, Fu´lvia Barros Manchado-Gobatto. Project administration: Fu´lvia Barros Manchado-Gobatto. Supervision: Fu´lvia Barros Manchado-Gobatto. Visualization: João Pedro Cruz, Rafael Lucas Cetein. Writing – original draft: Felipe Marroni Rasteiro, Leonardo Henrique Dalcheco Messias, Claudio Alexandre Gobatto, Fu´lvia Barros Manchado-Gobatto. Writing – review & editing: Felipe Marroni Rasteiro, Leonardo Henrique Dalcheco Messias, Pedro Paulo Menezes Scariot, João Pedro Cruz, Claudio Alexandre Gobatto, Fu´lvia Barros Manchado-Gobatto. References 1. Ayres L. P. The influence of music on speed in the six day bicycle race. American Physical education Review. 1911; 16:321–324. 2. Bacon C, Myers T, Karageorghis C. Effect of music-movement synchrony on exercise oxygen con- sumption. Journal of Sports Medicine and Physical Fitness. 2012; 52(4): 359. PMID: 22828457 3. Crust L, Clough PJ. The influence of rhythm and personality in the endurance response to motivational asynchronous music. Journal of Sports Sciences. 2006; 24(2): 187–195. https://doi.org/10.1080/ 02640410500131514 PMID: 16368629 4. Karageorghis CI, Mouzourides DA, Priest D-L, Sasso TA, Morrish DJ, Walley CL. Psychophysical and ergogenic effects of synchronous music during treadmill walking. Journal of Sport and Exercise Psy- chology. 2009; 31(1): 18–36. https://doi.org/10.1123/jsep.31.1.18 PMID: 19325186 5. Karageorghis CI, Priest D, Williams L, Hirani R, Lannon K, Bates B. Ergogenic and psychological effects of synchronous music during circuit-type exercise. Psychology of Sport and Exercise. 2010; 11(6): 551–559. 6. Karageorghis CI, Hutchinson JC, Jones L, Farmer HL, Ayhan MS, Wilson RC, et al. Psychological, psy- chophysical, and ergogenic effects of music in swimming. Psychology of Sport and Exercise. 2013; 14 (4): 560–568. 7. Schneider S, Askew CD, Abel T, Stru¨der HK. Exercise, music, and the brain: is there a central pattern generator? Journal of Sports Sciences. 2010; 28(12): 1337–1343 https://doi.org/10.1080/02640414. 2010.507252 PMID: 20845211 PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 12 / 16 8. Terry P. C., Karageorghis C. I., Curran M. L., Martin O. V., & Parsons-Smith R. L. Effects of music in exercise and sport: A meta-analytic review. Psychological Bulletin. 2020; 146(2): 91. https://doi.org/10. 1037/bul0000216 PMID: 31804098 9. Moss S. L., Enright K., & Cushman S. The influence of music genre on explosive power, repetitions to failure and mood responses during resistance exercise. Psychology of Sport and Exercise. 2018; 37:128–138. 10. Szabo A., Small A., & Leigh M. The effects of slow- and fast-rhythm classical music on progressive cycling to voluntary physical volitional exhaustion. Journal of Sports Medicine and Physical Fitness. 1999; 39:220–225. PMID: 10573664 11. Karageorghis C., Jones L., & Stuart D. P. Psychological effects of music tempi during exercise. Interna- tional journal of sports medicine. 2008; 29(07):613–619. 12. Karageorghis C. I., Jones L., Priest D. L., Akers R. I., Clarke A., Perry J. M., el al. Revisiting the relation- ship between exercise heart rate and music tempo preference. Research quarterly for exercise and sport. 2011; 82(2):274–284. https://doi.org/10.1080/02701367.2011.10599755 PMID: 21699107 13. Koelsch S. Towards a neural basis of music-evoked emotions. Trends in cognitive sciences. 2010; 14 (3): 131–137. https://doi.org/10.1016/j.tics.2010.01.002 PMID: 20153242 14. Laukka P., & Quick L. Emotional and motivational uses of music in sports and exercise: A questionnaire study among athletes. Psychology of Music. 2013; 41(2):198–215. 15. Saarikallio S. H., Maksimainen J. P., & Randall W. M. Relaxed and connected: Insights into the emo- tional–motivational constituents of musical pleasure. Psychology of Music. 2019; 47(5): 644–662. 16. Nakamura PM, Pereira G, Papini CB, Nakamura FY, Kokubun E. Effects of preferred and nonpreferred music on continuous cycling exercise performance. Perceptual and Motor Skills. 2010; 110(1): 257– 264. https://doi.org/10.2466/PMS.110.1.257-264 PMID: 20391890 17. Yamashita S., Iwai K., Akimoto T., Sugawara J., & Kono I. Effects of music during exercise on RPE, heart rate and the autonomic nervous system. Journal of Sports Medicine and Physical Fitness. 2006; 46(3):425. PMID: 16998447 18. Ballmann C. G., McCullum M. J., Rogers R. R., Marshall M. M., & Williams T. D. Effects of Preferred vs. Nonpreferred Music on Resistance Exercise Performance. Journal of strength and conditioning research. 2018. 19. Ballmann C. G., Maynard D. J., Lafoon Z. N., Marshall M. R., Williams T. D., & Rogers R. R. Effects of Listening to Preferred versus Non-Preferred Music on Repeated Wingate Anaerobic Test Performance. Sports. 2019; 7(8):185. 20. Hutchinson J. C., Jones L., Vitti S. N., Moore A., Dalton P. C., & O’Neil B. J. The influence of self- selected music on affect-regulated exercise intensity and remembered pleasure during treadmill run- ning. Sport, Exercise, and Performance Psychology. 2018; 7(1):80. 21. Lim H. B., Karageorghis C. I., Romer L. M., & Bishop D. T. (2014). Psychophysiological effects of syn- chronous versus asynchronous music during cycling. 22. Kreutz G., Schorer J., Sojke D., Neugebauer J., & Bullack A. (2018). In dubio pro silentio–Even loud music does not facilitate strenuous ergometer exercise. Frontiers in psychology, 9, 590. https://doi.org/ 10.3389/fpsyg.2018.00590 PMID: 29867622 23. Waterhouse J., Hudson P., & Edwards B. (2010). Effects of music tempo upon submaximal cycling per- formance. Scandinavian journal of medicine & science in sports, 20(4), 662–669. 24. Van Dyck E., Moens B., Buhmann J., Demey M., Coorevits E., Dalla Bella S., et al. Spontaneous entrainment of running cadence to music tempo. Sports medicine-open, 1(1), 15. https://doi.org/10. 1186/s40798-015-0025-9 PMID: 26258007 25. Faude O, Kindermann W, Meyer T. Lactate threshold concepts. Sports Medicine. 2009; 39(6): 469– 490. https://doi.org/10.2165/00007256-200939060-00003 PMID: 19453206 26. Svedahl K., & MacIntosh B. R. Anaerobic threshold: the concept and methods of measurement. Canadian journal of applied physiology. 2003; 28(2):299–323. https://doi.org/10.1139/h03-023 PMID: 12825337 27. Bentley DJ, Newell J, Bishop D. Incremental exercise test design and analysis. Sports Medicine. 2007; 37(7):575–586. https://doi.org/10.2165/00007256-200737070-00002 PMID: 17595153 28. Hinkley DV. Inference about the intersection in two-phase regression. Biometrika. 1969; 56(3): 495–504. 29. Manchado-Gobatto F, Vieira NA, Messias LD, Ferrari H, Borin J, de Carvalho Andrade V, et al. Anaero- bic threshold and critical velocity parameters determined by specific tests of canoe slalom: effects of monitored training. Science & Sports. 2014; 29(4): e55–e8. 30. Messias LHD, Polisel EEC, Manchado-Gobatto FB. Advances of the reverse lactate threshold test: non-invasive proposal based on heart rate and effect of previous cycling experience. PloS One. 2018; 13(3): e0194313. https://doi.org/10.1371/journal.pone.0194313 PMID: 29534108 PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 13 / 16 31. Billat V. L., Sirvent P., Py G., Koralsztein J. P., & Mercier J. The concept of maximal lactate steady state. Sports medicine. 2003; 33(6):407–426. https://doi.org/10.2165/00007256-200333060-00003 PMID: 12744715 32. Molinari C. A., Palacin F., Poinsard L., & Billat V. L. Determination of Submaximal and Maximal Training Zones From a 3-Stage, Variable-Duration, Perceptually Regulated Track Test. International Journal of Sports Physiology and Performance. 2020; 1(aop):1–9. 33. Glaister M. Multiple sprint work. Sports medicine. 2005; 35(9):757–777. https://doi.org/10.2165/ 00007256-200535090-00003 PMID: 16138786 34. Tai M. M. A mathematical model for the determination of total area under glucose tolerance and other metabolic curves. Diabetes care. 1994; 17(2):152–154. https://doi.org/10.2337/diacare.17.2.152 PMID: 8137688 35. Tey S. L., Salleh N. B., Henry C. J., & Forde C. G. Effects of non-nutritive (artificial vs natural) sweeten- ers on 24-h glucose profiles. European journal of clinical nutrition. 2017; 71(9):1129–1132. https://doi. org/10.1038/ejcn.2017.37 PMID: 28378852 36. Cole Z., & Maeda H. Effects of listening to preferential music on sex differences in endurance running performance. Perceptual and motor skills. 2015; 121(2):390–398. https://doi.org/10.2466/06.PMS. 121c20x9 PMID: 26447745 37. Hutchinson J. C., & Sherman T. The relationship between exercise intensity and preferred music inten- sity. Sport, exercise, and performance psychology. 2014; 3(3): 191. 38. Hutchinson J. C., Karageorghis C. I., & Jones L. See hear: Psychological effects of music and music- video during treadmill running. Annals of Behavioral Medicine. 2015; 49(2):199–211. https://doi.org/10. 1007/s12160-014-9647-2 PMID: 25142042 39. Macone D, Baldari C, Zelli A, Guidetti L. Music and physical activity in psychological well-being. Perceptual and Motor Skills. 2006; 103(1): 285–295. https://doi.org/10.2466/pms.103.1.285-295 PMID: 17037673 40. Madison G., Paulin J., & Aasa U. Physical and psychological effects from supervised aerobic music exercise. American journal of health behavior. 2013; 37(6):780–793. https://doi.org/10.5993/AJHB.37. 6.7 PMID: 24001627 41. Corso J. Sex and age difference in pure tone thresholds. Arch Otolaryngol. 1963; 77: 53–73. 42. Kageyama T. Loudness in listening to music with portable headphone stereos. Perceptual and motor skills. 1999; 88(2):423–423. https://doi.org/10.2466/pms.1999.88.2.423 PMID: 10483631 43. Karageorghis C I, Terry PC, Lane AM. Development and validation of an instrument to assess the moti- vational qualities of music in exercise and sport: The Brunel music rating inventory. J Sports Sci. 1999; 17:713–724 https://doi.org/10.1080/026404199365579 PMID: 10521002 44. Kellaris JJ, Rice RC. The influence of tempo, loudness, and gender of listener on responses to music. Psychology & Marketing. 1993; 10(1): 15–29. 45. McCown W., Keiser R., Mulhearn S., & Williamson D. The role of personality and gender in preference for exaggerated bass in music. Personality and individual differences. 1997; 23(4):543–547. 46. Karageorghis CI, Priest D-L, Terry PC, Chatzisarantis NL, Lane AM. Redesign and initial validation of an instrument to assess the motivational qualities of music in exercise: The Brunel Music Rating Inven- tory-2. Journal of sports sciences. 2006; 24(8): 899–909. https://doi.org/10.1080/02640410500298107 PMID: 16815785 47. Thomas S, Reading J, Shephard RJ. Revision of the physical activity readiness questionnaire (PAR-Q). Canadian Journal of Sport Sciences. 1992; 17(4): 338–345. PMID: 1330274 48. Lohman TG. Skinfolds and body density and their relation to body fatness: a review. Human Biology. 1981; 53(2): 181. PMID: 7239496 49. Jackson AS, Pollock ML. Practical assessment of body composition. The Physician and Sportsmedi- cine. 1985; 13(5):76–90. https://doi.org/10.1080/00913847.1985.11708790 PMID: 27463295 50. Craig CL, Marshall AL, Sjo¨stro¨m M, Bauman AE, Booth ML, Ainsworth BE, et al. International physical activity questionnaire: 12-country reliability and validity. Medicine & Science in Sports & Exercise. 2003; 35(8): 1381–1395. 51. Harris DV, Harris BL. The athlete’s guide to sports psychology: mental skills for physical people: Human Kinetics; 1984. 52. Borg GA. Psychophysical bases of perceived exertion. Medicine & science in sports & exercise. 1982. 53. Garcin M, Vandewalle H, Monod H. A new rating scale of perceived exertion based on subjective esti- mation of exhaustion time: a preliminary study. International Journal of Sports Medicine. 1999; 20 (01):40–43. 54. Karvonen MJ, Kentala E, Mustalo O. The effects of training on heat rate; a longitudinal study. Ann Med Exp Biol Fenn. 1957; 35:307–315 PMID: 13470504 PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 14 / 16 55. Costello J. T., Bieuzen F., & Bleakley C. M. Where are all the female participants in Sports and Exercise Medicine research?. European Journal of Sport Science. 2014; 14(8): 847–851. https://doi.org/10. 1080/17461391.2014.911354 PMID: 24766579 56. Neder J. A., Nery L. E., Peres C., & Whipp B. J. Reference values for dynamic responses to incremental cycle ergometry in males and females aged 20 to 80. American journal of respiratory and critical care medicine. 2001; 164(8):1481–1486. 57. Reinhard U., Mu¨ller P. H., & Schmu¨lling R. M. Determination of anaerobic threshold by the ventilation equivalent in normal individuals. Respiration. (1979); 38(1):36–42. https://doi.org/10.1159/000194056 PMID: 493728 58. Geer E. B., & Shen W. (2009). Gender differences in insulin resistance, body composition, and energy balance. Gender medicine, 6, 60–75. https://doi.org/10.1016/j.genm.2009.02.002 PMID: 19318219 59. Dionne I., Despres J. P., Bouchard C., & Tremblay A. (1999). Gender difference in the effect of body composition on energy metabolism. International journal of obesity, 23(3), 312–319. https://doi.org/10. 1038/sj.ijo.0800820 PMID: 10193878 60. Makovey J., Naganathan V., & Sambrook P. (2005). Gender differences in relationships between body composition components, their distribution and bone mineral density: a cross-sectional opposite sex twin study. Osteoporosis International, 16(12), 1495–1505. https://doi.org/10.1007/s00198-005-1841-4 PMID: 15838718 61. Kirchengast S. (2010). Gender differences in body composition from childhood to old age: an evolution- ary point of view. Journal of Life Sciences, 2(1), 1–10. 62. Hoffmann S. M., Skinner T. L., Osborne M. A., Emmerton L. M., & Jenkins D. G. The Efficacy of the Lac- tate Threshold: A Sex-Based Comparison. Journal of strength and conditioning research. 2018. 63. Jurkowski J. E., Jones N. L., Toews C. J., & Sutton J. R. Effects of menstrual cycle on blood lactate, O2 delivery, and performance during exercise. Journal of Applied Physiology. 1981; 51(6):1493–1499. https://doi.org/10.1152/jappl.1981.51.6.1493 PMID: 6798000 64. Mattu A. T., Iannetta D., MacInnis M. J., Doyle-Baker P. K., & Murias J. M. Menstrual and oral con- traceptive cycle phases do not affect submaximal and maximal exercise responses. Scandinavian jour- nal of medicine & science in sports. 2020; 30(3):472–484. 65. Bigliassi M., Karageorghis C. I., Nowicky A. V., Orgs G., & Wright M. J. Cerebral mechanisms underly- ing the effects of music during a fatiguing isometric ankle-dorsiflexion task. Psychophysiology. 2016; 53 (10):1472–1483. https://doi.org/10.1111/psyp.12693 PMID: 27346459 66. Eliakim M, Bodner E, Eliakim A, Nemet D, Meckel Y. Effect of motivational music on lactate levels during recovery from intense exercise. The Journal of Strength & Conditioning Research. 2012; 26(1): 80–86. 67. Szmedra L., & Bacharach D. W. Effect of music on perceived exertion, plasma lactate, norepinephrine and cardiovascular hemodynamics during treadmill running. International journal of sports medicine. 1998; 19(01): 32–37. 68. Lima-Silva A. E., Silva-Cavalcante M. D., Pires F. D. O., Bertuzzi R., Oliveira R. S. F., & Bishop D. Lis- tening to music in the first, but not the last 1.5 km of a 5-km running trial alters pacing strategy and improves performance. International journal of sports medicine. 2012; 33(10):813–818. https://doi.org/ 10.1055/s-0032-1311581 PMID: 22592542 69. Simpson S. D., & Karageorghis C. I. The effects of synchronous music on 400-m sprint performance. Journal of sports sciences. 2006; 24(10):1095–1102. https://doi.org/10.1080/02640410500432789 PMID: 17115524 70. Crust L. Effects of familiar and unfamiliar asynchronous music on treadmill walking endurance. Percep- tual and motor skills. 2004; 99(1):361–368. https://doi.org/10.2466/pms.99.1.361-368 PMID: 15446664 71. Crust L. Perceived importance of components of asynchronous music during circuit training. Journal of Sports Sciences. 2008; 26(14):1547–1555. https://doi.org/10.1080/02640410802315427 PMID: 18949662 72. Hutchinson J. C., Sherman T., Davis L., Cawthon D., Reeder N. B., & Tenenbaum G. The influence of asynchronous motivational music on a supramaximal exercise bout. International Journal of Sport Psy- chology. 2011; 42(2):135–148. 73. Hutchinson J. C., & Karageorghis C. I. Moderating influence of dominant attentional style and exercise intensity on responses to asynchronous music. Journal of Sport and Exercise Psychology. 2013; 35 (6):625–643. https://doi.org/10.1123/jsep.35.6.625 PMID: 24334323 74. Terry P. C., Karageorghis C. I., Saha A. M., & D’Auria S. Effects of synchronous music on treadmill run- ning among elite triathletes. Journal of Science and Medicine in Sport. 2012; 15(1):52–57. https://doi. org/10.1016/j.jsams.2011.06.003 PMID: 21803652 75. Ghaderi M., Rahimi R., & Azarbayjani M. A. The effect of motivational and relaxation music on aerobic performance, rating perceived exertion and salivary cortisol in athlete males. South African Journal for Research in Sport Physical Education and Recreation. 2009; 31:29–38. PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 15 / 16 76. Stork M. J., Kwan M. Y., Gibala M. J., & Martin K. G. Music enhances performance and perceived enjoy- ment of sprint interval exercise. Medicine and science in sports and exercise. 2015; 47(5):1052–1060. https://doi.org/10.1249/MSS.0000000000000494 PMID: 25202850 77. Clark I. N., Baker F. A., & Taylor N. F. The modulating effects of music listening on health-related exer- cise and physical activity in adults: a systematic review and narrative synthesis. Nordic Journal of Music Therapy. 2016; 25(1):76–104. 78. Clark J. C., Baghurst T., & Redus B. S. Self-Selected Motivational Music on the Performance and Per- ceived Exertion of Runners. Journal of strength and conditioning research. 2018. 79. Loizou G., & Karageorghis C. I. Effects of psychological priming, video, and music on anaerobic exer- cise performance. Scandinavian journal of medicine & science in sports. 2015; 25(6):909–920. 80. Smirmaul B. P. Effect of pre-task music on sports or exercise performance. The Journal of sports medi- cine and physical fitness. 2017; 57(7–8):976–984. https://doi.org/10.23736/S0022-4707.16.06411-2 PMID: 27244132 81. Karageorghis C. I., & Priest D. L. Music in the exercise domain: a review and synthesis (Part I). Interna- tional review of sport and exercise psychology. 2012; 5(1):44–66. https://doi.org/10.1080/1750984X. 2011.631026 PMID: 22577472 82. Karageorghis C. I., & Priest D. L. Music in the exercise domain: a review and synthesis (Part II). Interna- tional review of sport and exercise psychology. 2012; 5(1):67–84. https://doi.org/10.1080/1750984X. 2011.631027 PMID: 22577473 83. Marcora SM, Staiano W, Manning V. Mental fatigue impairs physical performance in humans. Journal of Applied Physiology. 2009; 106(3): 857–864. https://doi.org/10.1152/japplphysiol.91324.2008 PMID: 19131473 84. Coquart J. B., Dufour Y., Groslambert A., Matran R., & Garcin M. Relationships between psychological factors, RPE and time limit estimated by teleoanticipation. The Sport Psychologist. 2012; 26(3):359– 374. 85. Garcin M., Fleury A., Mille-Hamard L., & Billat V. Sex-related differences in ratings of perceived exertion and estimated time limit. International journal of sports medicine. 2005; 26(08):675–681. 86. Garcin M., Coquart J., Salleron J., Voy N., & Matran R. Self-regulation of exercise intensity by estimated time limit scale. European journal of applied physiology. 2012; 112(6):2303–2312. https://doi.org/10. 1007/s00421-011-2197-4 PMID: 22009018 87. Van Dyck E. (2019). Musical Intensity Applied in the Sports and Exercise Domain: An Effective Strategy to Boost Performance?. Frontiers in psychology, 10, 1145. https://doi.org/10.3389/fpsyg.2019.01145 PMID: 31156525 88. Dyer B. J., & McKune A. J. (2013). Effects of music tempo on performance, psychological, and physio- logical variables during 20 km cycling in well-trained cyclists. Perceptual and Motor Skills, 117(2), 484– 497. https://doi.org/10.2466/29.22.PMS.117x24z8 PMID: 24611252 89. Nakamura P. M. (2015). Music tempo’s effect on exercise performance: comment on dyer and mckune. Perceptual and motor skills, 120(3), 860–863. https://doi.org/10.2466/29.PMS.120v20x5 PMID: 26057422 PLOS ONE Effects of preferred music on physiological responses, perceived exertion and anaerobic threshold determination PLOS ONE | https://doi.org/10.1371/journal.pone.0237310 August 12, 2020 16 / 16

Effects of preferred music on physiological responses, perceived exertion, and anaerobic threshold determination in an incremental running test on both sexes.

08-12-2020

Rasteiro, Felipe Marroni,Messias, Leonardo Henrique Dalcheco,Scariot, Pedro Paulo Menezes,Cruz, João Pedro,Cetein, Rafael Lucas,Gobatto, Claudio Alexandre,Manchado-Gobatto, Fúlvia Barros

eng

PMC7451842

J Exerc Nutrition Biochem. 2018;22(2):007-011, http://dx.doi.org/10.20463/jenb.2018.0010 30 Physical Activity and Nutrition. 2020;24(2):030-037, http://dx.doi.g/1.20463/pan.2020.0012 30 Muscle oxygenation, endocrine and metabolic regulation during low- intensity endurance exercise with blood flow restriction 1. Graduate school of Sport and Health Science, Ritsumeikan University, Shiga, Japan 2. Department of Physical Education, Hanyang University, Seoul, Korea 3. Physical Activity and Performance Institute (PAPI), Konkuk University, Seoul, Korea 4. Research Center of Health, Physical Fitness and Sports, Nagoya University, Nagoya, Japan 5. Department of Sports Science, Japan Institute of Sports Sciences, Tokyo, Japan 6. Research Center for Urban Health and Sports, Osaka City University, Osaka, Japan [Purpose] The present study investigated the effect of endurance exercise with blood flow restriction (BFR) performed at either 25% maximal oxygen uptake (V 3 O2 max) or 40% V 3 O2 max) on muscle oxygenation, ener- gy metabolism, and endocrine responses. [Methods] Ten males were recruited in the present study. The subjects performed three trials: (1) endur- ance exercise at 40% V 3 O2 max without BFR (NBFR40), (2) endurance exercise at 25% V 3 O2 max with BFR (BFR25), and (3) endurance exercise at 40% V 3 O2 max with BFR (BFR40). The exercises were performed for 15 min during which the pedaling frequency was set at 70 rpm. In BFR25 and BFR40, 2 min of pressure phase (equivalent to 160 mmHg) followed by 1 min of release phase were repeated five times (5 × 3 min) throughout 15 minutes of exercise. During exercise, muscle oxygenation and concentration of respiratory gases were measured. The blood samples were col- lected before exercise, immediately after 15 min of ex- ercise, and at 15, 30, and 60 minutes after completion of exercise. [Results] Deoxygenated hemoglobin (deoxy-Hb) level during exercise was significantly higher with BFR25 and BFR40 than that with NBFR40. BFR40 showed significantly higher total-hemoglobin (total-Hb) than NBFR40 during 2 min of pressure phase. Moreover, exercise-induced lactate elevation and pH reduction were significantly augmented in BFR40, with concom- itant increase in serum cortisol concentration after ex- ercise. Carbohydrate (CHO) oxidation was significantly higher with BFR40 than that with NBFR40 and BFR25, whereas fat oxidation was lower with BFR40. [Conclusion] Deoxy-Hb and total Hb levels were sig- nificantly increased during 15 min of pedaling exercise in BFR25 and BFR40, indicating augmented local hypoxia and blood volume (blood perfusion) in the muscle. Moreover, low-and moderate-intensity exercise with BFR facilitated CHO oxidation. [Key words] low-intensity exercise, blood flow restric- tion, muscle oxygenation, endocrine response, energy metabolism Received: 2020/06/13, Revised: 2020/06/24, Accepted: 2020/06/26, Published: 2020/06/30 ©2020 Hyejung Hwang et al.; Licence Physical Activity and Nutrition. This is an open access article distributed under the terms of the creative commons attribution license (http:// creativecommons.org/licenses/by/2.0), which permits unre- stricted use, distribution, and reproduction in any medium, provided the orginal work is properly cited. *Corresponding author : Kazushige Goto Graduate school of Sport and Health Science, Ritsumeikan University, Shiga, Japan. Tel: +81-77-599-4127 / Fax: +81-77-599-4127 E-mail: kagoto@fc.ritsumei.ac.jp ©2020 The Korean Society for Exercise Nutrition OPEN ACCESS http://dx.doi.org/10.20463/pan.2020.0012 2020;24(2):030-037 INTRODUCTION In traditional training procedures aimed to increase muscular strength and muscle hypertrophy, exercise intensity above at least 70% of one repetition maximum (1RM) is commonly recommend- ed1 However, high-intensity exercise entails the risk of injury due to excessive stress on muscle joints as well as connective tissues in un- trained or older people. In contrast, low intensity exercise (e.g., 20% of 1RM) with blood flow restriction (BFR) has beneficial effects even with short periods of training2-4. In particular, resistance exercise with BFR is effective in improving muscle strength and muscle hypertro- phy5-8. Exercise with BFR augments local hypoxia in muscle. The low- ered muscle oxygenation during exercise is expected to elicit erythro- poiesis with subsequent increases in oxygen transport capacity9, cap- illary density, mitochondrial biosynthesis, and myoglobin level in the tissues10-11. These cascades are stimulated by increased expression of hypoxia-inducible factor-1 (HIF-1) and vascular endothelial growth factor (VEGF), which are two major factors involved in angiogene- sis12. Several studies have shown that low-intensity endurance exercise (30-40% V. O2 max) with BFR increases oxygen uptake, heart rate, and metabolite levels during and after exercise compared with nor- mal exercise without BFR13-15. However, the influence of endurance exercise with BFR and the difference in effects with respect to low and extremely low intensity exercise on muscle oxygenation and metabolic regulation is currently unknown. Therefore, the purpose of the present study was to investigate the effects of endurance exercise with BFR performed at 25% V. O2 max or 40% V. O2 max on muscle oxygenation, energy metabolism, and endocrine responses. Hyejung Hwang1,2,3 / Sahiro Mizuno4 / Nobukazu Kasai5 / Chihiro Kojima5 / Daichi Sumi6 / Nanako Hayashi6 / Kazushige Goto6** Physical Activity and Nutrition. 2020;24(2):030-037, http://dx.doi.org/10.20463/pan.2020.0012 31 Muscle oxygenation, endocrine and metabolic regulations during exercise with blood flow restriction METHODS Subjects Ten males (mean± standard deviation [SD] age: 24.7 ± 2.1 years, height: 171.2 ± 5.7 cm, and body weight: 68.0 ± 7.8 kg) were recruited for the present study. They were healthy and had regular physical activity (few days/week, e.g., resistance exercise, endurance exercise). However, none of the subjects were involved in any training pro- gram at the start of the study. All subjects were explained the purpose of experiment, procedures, and the potential risks of the study. A written informed consent was sub- sequently obtained from each participant. The present study was approved by the Ethics Committee for Human Experiments at Ritsumeikan University. Experimental design All subjects visited our laboratory four times during the experimental period. At the first visit, an incremental pedaling test was conducted to assess maximal oxygen uptake (V. O2 max) using an ergometer (Aerobike 75XLIII; Konami Corporation, Tokyo, Japan). From second through fourth visits, three experimental trials were performed in a random order. The three trials consisted of endurance exercise at 40% V. O2 max without BFR (NBFR40), endur- ance exercise at 25% V. O2 max with BFR (BFR25), and endurance exercise at 40% V. O2 max with BFR (BFR40). At least 7 days were prepared among trials. For BFR25 and BFR40, specially designed tourni- quets (E20 Rapid Cuff Inflator and Rapid Version Cuff, Hokanson, USA) were used to apply pressure during exercise, and the tourniquets were inflated at 160 mmHg pressure. Necessary information to accustom the subjects with the device was shared during the preliminary ses- sion. The tourniquet was placed at the proximal site of the middle thigh, both legs. Blood flow restriction and exercise protocols The tourniquet was designed to be 11 × 85 cm wide. It was used in conjunction with a rapid cuff inflator. The air inflator was controlled to maintain a stable level of required pressure (160 mmHg) during the pressure phase. Based on previous studies, we had set up 15 min pedal- ing exercise with a BFR protocol using an ergometer4,32. During the 15 min exercise in each trial, the pedaling fre- quency was set as 70 rpm. In BFR25 and BFR40, 2 min of pressure phase (equivalent to 160 mmHg) followed by 1 min of release phase were repeated five times (5 × 3 min) throughout the exercise. In NBFR40, the subjects wore a tourniquet, but no pressure was applied through- out the exercise (Fig. 1). Muscle oxygenation During exercise, the muscle oxygenation level in the vastus lateralis muscle was evaluated noninvasively using near infrared spectroscopy (NIRS) (Hb14-2, Astem Co., Ltd. Kanagawa, Japan). The probe emitted two different wavelengths from the LED and photo diode, and detected the light transmitted through the body with the help of the light receiving element. The probe was placed on the right vastus lateralis (VL) muscle (at midpoint between the greater trochanter and lateral condyle of the femur), and the sampling rate was 10 Hz. The data were expressed as relative values to the baseline values obtained during rest. The oxygenated-hemoglobin (oxy-Hb), deoxygenated hemoglobin (deoxy-Hb) and total hemoglobin (total-Hb) levels were determined. Respiratory variables Respiratory samples were collected using breath by breath method and analyzed using an automatic gas ana- lyzer (AE300S, Minato Medical Science Co., Ltd., Tokyo, Japan) to determine V. O2, carbon dioxide output (V. CO2), minute ventilation (VE), and the respiratory exchange ra- tio (RER). The carbohydrate and fat oxidation rates were also calculated from V. O2 and V. CO2 using the following equations16. The collected data were averaged every 30 s. Figure 1. Exercise protocol with or without BFR Physical Activity and Nutrition. 2020;24(2):030-037, http://dx.doi.org/10.20463/pan.2020.0012 32 Muscle oxygenation, endocrine and metabolic regulations during exercise with blood flow restriction Exercise energy metabolism equations: CHO oxidation(g/min) = 4.210 × V. CO2 −2.962 × V. O2 FAT oxidation(g/min) = 1.695× V. O2 −1.701× V. CO2 EE(kcal/15min) = 4.07×CHO oxidation + 9.75×FAT oxidation Blood sampling and analysis Subjects arrived at the laboratory at approximately 8:00 AM following an overnight fast (at least 10 h after the previous meal). They rested about 20 min before the first blood collection. After rest, a 22-gauge polyethylene catheter was inserted into an antecubital vein and base- line blood sample was obtained. After exercise, subjects rested on the chair for an hour for blood collection. Blood samples were collected before exercise, immediately after 15 min of exercise, and at 15, 30, and 60 min after com- pletion of exercise. Blood glucose and lactate concentra- tions were measured using a glucose analyzer (FreeStyle, Nipro Co., Osaka, Japan) and a lactate analyzer (Lactate Pro, Arkray Co., Kyoto, Japan) immediately after blood collection. The serum samples were obtained after 10 min of centrifugation at 4 ºC, and these samples were stored at -80 ºC until analysis. Serum growth hormone (GH), corti- sol and myoglobin (Mb) concentrations were measured at a clinical laboratory (SRL, Inc., Tokyo, Japan). Heparin syringes (2.5 mL) were used to collect blood samples for determination of blood gas and electrolyte levels. From obtained blood samples, blood pH, HCO3−, base excess (BE), partial pressure of oxygen (pO2), partial pressure of carbon dioxide (pCO2), and sodium (Na+) and potassium (K+) concentrations were measured using an automatic blood-gas analyzer (OPTI CCA TS, Sysmex Co., Hyogo, Japan). Blood gas and electrolyte analyses were per- formed immediately after blood collection. Statistical analysis All data are expressed as means ± SD. Time-dependent changes in variables were analyzed using two-way re- peated measure analysis of variance (ANOVA) to confirm significant interaction (trial × time) and main effects for trial and time. When a significant interaction (time × trial) or main effect was detected, a post-hoc Tukey test was performed to identify differences. A P-value < 0.05 was considered to indicate statistical significance. RESULTS Muscle oxygenation Figure 2 shows the relative changes in the variables of muscle oxygenation during 15 min of exercise. Oxy- Hb did not show a significant interaction (trial × time, p=0.83). Moreover, significant main effects of trial (p=0.24) and time (p=0.24) were not noted. The oxy-Hb level rapidly reduced during the pressure phase in BFR40, while NBFR40 revealed a slight increase in the over oxy- Hb level during the 15-min exercise session. Deoxy-Hb showed a significant interaction (trial × time, p<0.001), and the main effects of time (p<0.001) were noted. Al- though a marked increase in the deoxy-Hb levels were noted in during the pressure phase when exercise was performed with BFR (for the BFR25 and BFR40 trials), this decrease rapidly recovered during the subsequent re- lease phase (1 min). In contrast, NBFR40 revealed slight elevation over 15 min of exercise. Total Hb showed a significant interaction (trial × time, p<0.001), and the main effects of time (p<0.001) were noted. The total-Hb levels increased during the pressure phase when exercise was performed with BFR (for the BFR25 and BFR40 tri- als), with a decrease in the levels during the subsequent release phase. In the NBFR40 trial, the total-Hb level gradually increased during the 15-min exercise session. Figure 3 presents the averaged muscle oxygenation variables during the pressure phase (2 min) and release phase (1 min) of the 15-min exercise session. The oxy- Hb level was significantly lower during the pressure phase in the BFR40 trial (p<0.05, Fig. 3A), while no significant difference was noted during the release phase. Particularly, the deoxy-Hb level was significantly increased during the pressure phase in the BFR25 and BFR40 trials. Further- Figure 2. The percent changes of muscle oxygenation vari- ables during exercise in the NBFR40, BFR40 and BFR25 every 5 s. (A) : The changes of oxygenated hemoglobin during exercise in the NBFR40, BFR25 and BFR40. (B) : The changes of deoxygenated hemoglobin during exercise in the NBFR40, BFR25 and BFR40. (C) : The changes of total hemoglobin during exercise in the NBFR40, BFR25 and BFR40 trials. Physical Activity and Nutrition. 2020;24(2):030-037, http://dx.doi.org/10.20463/pan.2020.0012 33 Muscle oxygenation, endocrine and metabolic regulations during exercise with blood flow restriction Figure 3. The percent changes of muscle oxygenation variables during 2 min pressure and 1 min release phase of 15min exercise in NBFR40, BFR25 and BFR 40. * p<0.05 between trials Physical Activity and Nutrition. 2020;24(2):030-037, http://dx.doi.org/10.20463/pan.2020.0012 34 Muscle oxygenation, endocrine and metabolic regulations during exercise with blood flow restriction more, increased deoxy-Hb levels were noted during the release phase in the BFR40 trial (p<0.05, Fig. 3B). The total-Hb level significantly increased during the pressure phase in the BFR40 trial. During the release phase, the total-Hb level was significantly lower in the BFR25 trial than in the NBFR40 and BFR40 trials (p<0.05, Fig. 3C). Blood variables Table 1 presents the changes in the blood variables before exercise and during the 60-min post-exercise period. The blood glucose, HCO3 −, pO2, pCO2, Na + and K + concen- trations did not differ significantly at any time point among the three trials. However, the blood lactate concentrations significantly increased after exercise only in the BFR40 trial (main effect for time, p<0.05), whereas a significant change was not observed over time in the NBFR40 and BFR25 trials. Immediately after exercise, the blood pH was sig- nificantly lower in the BFR40 trial than in the NFR40 and BFR25 trials. A significantly lower blood base excess was noted immediately after the 15-min exercise session in the BFR40 trial than in the BFR25 and NBFR40 trials. Serum GH, cortisol and myoglobin Figure 4 presents the changes in the serum GH, cortisol and myoglobin concentrations. After exercise, the serum GH concentrations tended to be higher in the BFR40 trial than in the BFR25 and NBFR40 trials. However, there was no significant interaction (trial × time, p=0.07) or main effect for trial (p=0.17, Fig. 4A). The serum cortisol con- centration showed a significant interaction (trial × time, p<0.001), and the main effects of trial (p<0.001) and time (p<0.001) were noted. Moreover, serum cortisol concen- trations were significantly higher in the BFR40 trial than in the BFR25 and NBFR40 trial immediately after exercise and at 15 and 30 min after exercise (p<0.05, Fig. 4B). The serum myoglobin concentration showed a significant inter- action (trial × time, p<0.001), and the main effect of time (p<0.001) was noted. In the BFR40 trial, the serum myo- globin concentration 60 min after the exercise session was significantly higher compared to that in the NBFR40 trial (p<0.05). However, no significant difference was observed between the NBFR40 and BFR25 trials (Fig. 4C). Energy metabolism during exercise The averaged CHO and fat oxidation during the 15-min exercise session are presented in Fig. 5. CHO oxidation was significantly higher in the BFR40 trial than in the BFR25 and NBFR40 trials. Moreover, no significant difference was observed between the BFR25 and NBFR40 trials, although exercise intensity was dif- ferent (25% V. O2 max for BFR25 and 40% V. O2 max for NBFR40, Fig. 3A). Fat oxidation was significantly lower in the BFR25 and BFR40 trials than in the NBFR40 trial. Furthermore, the lowest fat oxidation value among Variable Trials Pre Post-exercise (min) 0 15 30 60 Glucose (mmol/L) NBFR40 88.8 ± 3.9 81.8 85.0 ± 5.1 87.4 ± 4.7 83.1 ± 5.3 BFR25 88.1 ± 4.8 86.8 ± 5.1 86.3 ± 3.9 85.6 ± 5.2 87.1 ± 5.5 BFR40 89.2 ± 8.0 89.3 ± 10.0 94.3 ± 10.6 90.6 ± 9.7 88.9 ± 7.2 Lactate (mmol/L) NBFR40 1.1 ± 0.3 1.5 ± 0.4 1.1 ± 0.2 1.1 ± 0.3 1.3 ± 0.2 BFR25 1.3 ± 0.31 2.0 ± 0.5 1.5 ± 0.3 1.3 ± 0.3 1.4 ± 0.3 BFR40 1.3 ± 0.3 5.2 ± 1.5 *#† 3.1 ± 1.0 *#† 2.3 ± 0.5 *#† 1.7 ± 0.4 pH NBFR40 7.41 ± 0.01 7.41 ± 0.02 7.42 ± 0.01 7.42 ± 0.01 7.41 ± 0.02 BFR25 7.42 ± 0.02 7.40 ± 0.02 7.42 ± 0.02 7.41 ± 0.01 7.41 ± 0.03 BFR40 7.41 ± 0.02 7.36 ± 0.05*#† 7.39 ± 0.03 7.42 ± 0.04 7.41 ± 0.02 HCO3- (mmol/L) NBFR40 27.3 ± 2.0 27.3 ± 1.8 27.5 ± 1.6 27.5 ± 1.3 26.4 ± 6.6 BFR25 27.2 ± 1.3 26.6 ± 1.7 26.1 ± 1.9 27.3 ± 1.3 27.5 ± 1.3 BFR40 26.8 ± 1.7 23.0 ± 2.2 23.8 ± 2.7 25.2 ± 2.5 27.1 ± 1.8 Base Excess (mmol/L) NBFR40 2.3 ± 1.8 2.1 ± 1.6 2.6 ± 1.3 2.7 ± 0.9 2.9 ± 1.3 BFR25 2.3 ± 1.4 1.4 ± 1.7 1.4 ± 1.7 2.3 ± 1.1 2.4 ± 1.0 BFR40 1.9 ± 1.4 -2.5 ± 2.3*# † -1.0 ± 2.4*# 0.8 ± 1.6 2.1 ± 1.7 PO2 (kPa) NBFR40 8.45 ± 1.83 9.40 ± 1.66 8.87 ± 2.71 9.20 ± 1.85 6.53 ± 2.60 BFR25 8.21 ± 2.31 8.84 ± 2.24 9.37 ± 2.65 5.77 ± 3.15 6.34 ± 1.55 BFR40 9.14 ± 2.87 7.06 ± 1.84 9.06 ± 1.60 8.47 ± 2.14 7.88 ± 2.64 PO2 (kPa) NBFR40 5.80 ± 0.32 5.93 ± 0.33 5.76 ± 0.36 5.75 ± 0.37 6.11 ± 0.43 BFR25 5.73 ± 0.20 5.85 ± 0.28 5.52 ± 0.38 5.80 ± 0.26 5.91 ± 0.48 BFR40 5.70 ± 0.45 5.63 ± 0.73 4.85 ± 1.81 5.37 ± 0.83 5.78 ± 0.37 Na+ (mmol/L) NBFR40 138.7 ± 1.1 140.0 ± 1.4 138.8 ± 0.5 138.3 ± 1.1 138.7 ± 1.0 BFR25 138.8 ± 1.6 139.1 ± 1.3 138.2 ± 1.1 138.7 ± 2.0 138.3 ± 1.1 BFR40 137.9 ± 1.8 140.0 ± 1.5 138.9 ± 2.5 138.6 ± 2.0 138.6 ± 1.1 K+ (mmol/L) NBFR40 3.65 ± 0.26 4.17 ± 0.19 3.83 ± 0.20 3.76 ± 0.27 3.50 ± 1.06 BFR25 3.52 ± 0.15 3.96 ± 0.16 3.71 ± 0.15 3.69 ± 0.15 3.70 ± 0.11 BFR40 3.49 ± 0.11 4.26 ± 0.38 3.72 ± 0.30 3.63 ± 0.12 3.76 ± 0.15 Table 1. Changes in blood variables before exercise and during post-exercise period. Values are presented as means ± SD. * Significant different from Pre ; p<.000, # Significant different from NBFR40 : p<0.05, † Significant dif- ferent from BFR25 : p<0.05 Physical Activity and Nutrition. 2020;24(2):030-037, http://dx.doi.org/10.20463/pan.2020.0012 35 Muscle oxygenation, endocrine and metabolic regulations during exercise with blood flow restriction all the trials was noted in the BFR40 trial, and the value was significantly lower than those noted in the NBFR40 and BFR25 trials. The total energy expenditure during the 15-min exercise session was significantly lower in the BFR25 trial than in the NBFR40 and BFR40 tri- als (NBFR40: 99.7 ± 21 kcal; BFR25: 64.9 ± 16 kcal; BFR40: 112.1 ± 23.4 kcal, p<0.05). However, significant differences were not noted between the NBFR40 and BFR40 trials with respect to the total energy expenditure during the 15-min exercise session. DISCUSSION The primary findings of the present study were that the deoxy-Hb level was significantly higher during the exercise session in the BFR25 and BFR40 trials than in the NBFR40 trial. A significantly higher total-Hb level was noted during the 2-min pressure phase in the BFR40 trial than in the NBFR trial. Moreover, exercise-induced increase in the lactate level and decrease in the pH were significantly higher in the BFR40 trial, with a concomi- tant increase in the serum cortisol concentration after ex- ercise. Notably, the substrate oxidation pattern was altered with BFR during low-intensity endurance exercise. CHO oxidation was significantly higher in the BFR40 trial than in the NBFR40 and BFR25 trials, while fat oxidation was lower in the BFR40 trial. These findings suggest that BFR during low-intensity endurance exercise promotes muscle deoxygenation and CHO metabolism compared to that when the same exercise is performed without BFR. During endurance exercise, the muscle blood flow is increased in response to the metabolic demands of the muscle17,18. NIRS is commonly used for evaluating the oxygenation levels and hemodynamics in a work- ing muscle. As an individual starts the exercise, oxygen consumption and delivery to the skeletal muscle rapidly increases, up to 50-fold or more19. In the present study, significantly lower oxy-Hb levels were noted during the 2-min pressure phase in the BFR40 trial compared to those in the NBFR40 trial. Moreover, the deoxy-Hb and total-Hb levels were higher during the 2-min pressure phase and 1-min release phase in the BFR40 trial than in the NBFR40 trial. Several studies have reported that the oxy-Hb dissociation curve promoted the rate of deoxy-Hb at or close to the lactate and ventilatory thresholds [33, 34]. Moreover, the deoxy-Hb level during the pressure phase was significantly higher in the BFR25 trial than in the NBFR40 trial, despite lower exercise intensity in the BFR25 trial. The total-Hb level measured using NIRS reflects the blood volume in the muscle. As shown in Figure 3, the total-Hb level was significantly higher in Figure 4. Exercise-induced changes in serum growth hor- mone (A), cortisol (B) and myoglobin concentrations (C) in NBFR40, BFR25 and BFR40. * Significant different from BFR40 at Pre ; p<.000, # Signifi- cant different from NBFR40 at Post 60 ; p<0.002, Values are presented as means ± SD. Figure 5. Carbohydrate (A) and fat oxidation (B) rate during 15 min of exercise. Values are presented as means ± SD. **p<0.05 between trials. Physical Activity and Nutrition. 2020;24(2):030-037, http://dx.doi.org/10.20463/pan.2020.0012 36 Muscle oxygenation, endocrine and metabolic regulations during exercise with blood flow restriction the BFR40 trial than in the NBFR40 trial, suggesting that the cuff pressure during low-intensity endurance exercise augmented the muscle blood volume (blood perfusion). Endurance exercise with BFR augmented hyperemic blood flow in the local region, leading to increased shear- ing stress on the vascular endothelial cells20. Therefore, the augmented blood volume in the BFR40 trial may be because of the increased nitric oxide production induced by augmented shear stress21-23. Notably, no difference was noted in the blood lactate lev- els or pH between the BFR25 and NBFR40 trials, despite the difference in the exercise intensity. In a previous study24, unilateral plantar flexion (30 repetitions/min) using 20% 1RM with BFR promoted a decrease in the in intramuscular phosphocreatine (PCr) and intramuscular pH, as measured by 31P-magnetic resonance spectroscopy (MRS), than ex- ercise using 20% 1RM without BFR. Exercise with BFR induces metabolite accumulation and may affect endocrine response. In the present study, the serum GH level increased till 60 minutes after exercise in the BFR40 trial; however, but there was no significant difference in the serum GH levels among the trials. This result may be attributed to the short exercise duration (only 15 min). However, the serum cortisol concentration was significantly elevated in the BFR40 trial till 60 min after exercise; the serum cortisol and myoglobin concentrations were significantly elevated in the BFR40 trial at 60 min after exercise. Significant differenc- es in the serum GH levels were not observed between the NBFR40 and BFR25 trials. Significant differences were not noted in the BFR40 trial; however, the serum GH tended to increase significantly in the BFR40 trial. Although the exercise-induced increase in the GH levels is dependent on the exercise intensity, the present findings suggest that BFR training effectively increases the serum GH levels. This may be an important finding with respect to prescribing exercise for untrained individuals, including elderly people. In the present study, HR during the exercise was signifi- cantly increased in the BFR40 trial than in the NBFR40 and BFR25 trials (NBFR40: 106 ± 11; BFR25: 102 ± 13; BFR40: 137 ± 18, p <0.05). BFR stimulated autonomic cardiovascular (CV) response through a chemical stim- ulus of accumulation of metabolites and a mechanical stimulus, such as muscle exercise pressor reflex (EPR)25. The rating of perceived exertion (RPE) during the 2-min pressure phase was significantly higher in the BFR25 and BFR40 trials than in the NBFR40 trial (NBFR40: 1.6 ± 0.1; BFR25: 4.0 ± 0.4; BFR40: 5.8 ± 1.0, p<0.05). Stimulation of EPR through BFR may increase exercise-induced fa- tigue. Therefore, BFR during low-intensity endurance ex- ercise augmented the score of subjective fatigue, probably owing to augmented central command26,27. Several studies have reported that endurance exercise with BFR enhanced the recruitment of the fast twitch fibers (FT fibers) during muscle activity. Enhanced FT fiber recruitment activates anaerobic glycolysis and alters the substrate oxidation pattern4,28-30. A 30-min low-intensity endurance exercise session with BFR increased CHO metabolism13. The mus- cle glycogen content significantly decreased after low-in- tensity resistance exercise with BFR compared to that when the same exercise is performed without BFR31. Our results indicate that CHO oxidation did not differ between the BFR25 and NBFR40 trials, despite the difference in the exercise intensity. The exercise intensity was same between the BFR40 and NBFR40 trials; however CHO oxidation was higher in the BFR40 trial. Therefore, we found that low-intensity exercise with BFR altered energy substrate utilization during exercise. The present study has several limitations. Firstly, we did not evaluate the long-term training effects (e.g., change in the muscle strength, endurance, and muscle volume). Secondly, the present study recruited only healthy young male subjects. Although we applied the same pressure of 160 mmHg , the pressure intensity may vary depending on the muscle mass in the legs of the sub- jects. To clarify the benefit of low-and moderate-intensity exercise with BFR, further investigations in elderly peo- ple or clinical populations are required. During the 15-min low-intensity (either 25% or 40% of V. O2 max) endurance exercise session, the levels of de- oxy-Hb and total Hb were significantly increased, when BFR was repeatedly applied. Moreover, lower levels of oxy-HB were noted during endurance exercise with BFR at 40% V. O2 max achieved compared to those when the same exercise was performed without BFR. These findings sug- gest that BFR during low-intensity endurance exercise aug- mented local hypoxia and blood volume in the muscle. Fur- thermore, endurance exercise with BFR at 40% V. O2 max promoted exercise-induced acidification in the blood (i.e., lower pH and higher blood lactate levels) compared to that when the same exercise was performed without BFR. Final- ly, BFR during low-intensity endurance exercise augmented CHO oxidation and impaired fat oxidation. Although the present study was performed as an acute experiment, the findings may suggest the spotential benefits of BFR during low-intensity endurance exercise for health promotion. ACKNOWLEDGMENTS This work was supported by the Ministry of Educa- tion of the Republic of Korea and the National Research Foundation of Korea(NRF-2016S1A5B5A01021612). REFERENCES 1. Kraemer WJ, Ratamess NA, French DN. Resistance training for health and performance. Curr Sports Med Rep. 2002;1:165-71. 2. Yasuda T, Brechue WF, Fujita T, Sato Y, Abe T. Muscle activa- tion during low-intensity muscle contractions with varying levels of external limb compression. J sports sci med. 2008;7:467. 3. Takarada Y, Sato Y, Ishii N. Effects of resistance exercise combined with vascular occlusion on muscle function in ath- letes. Eur J Appl Physiol. 2002;86:308-14. 4. Abe T, Fujita S, Nakajima T, Sakamaki M, Ozaki H, Ogas- awara R. Effects of low-intensity cycle training with restrict- Physical Activity and Nutrition. 2020;24(2):030-037, http://dx.doi.org/10.20463/pan.2020.0012 37 Muscle oxygenation, endocrine and metabolic regulations during exercise with blood flow restriction ed leg blood flow on thigh muscle volume and VO2 max in young men. J sports sci med. 2010;9:452. 5. Sudo M, Ando S, Kano Y. Repeated blood flow restriction induc- es muscle fiber hypertrophy. Muscle Nerve. 2017;55:274-6. 6. Ladlow P, Coppack RJ, Dharm-Datta S, Conway D, Sellon E, Patterson SD. Low-load resistance training with blood flow restriction improves clinical outcomes in musculoskeletal re- habilitation: A single-blind randomized controlled trial. Front Physiol. 2018;9:1269. 7. Yasuda T. Fukumura K, Tomaru T, Nakajima T. Thigh muscle size and vascular function after blood flow-restricted elastic band training in older women. Oncotarget. 2016;7:33595–607. 8. Yasuda T, Loenneke JP, Thiebaud RS, Abe T. Effects of blood flow restricted low-intensity concentric or eccentric training on muscle size and strength. Plos one. 2012;7:e52843. 9. Hamlin MJ, Marshall HC, Hellemans J, Ainslie PN, Anglem N. Effect of intermittent hypoxic training on 20km time trial and 30s anaerobic performance. Scand J Med Sci Sports. 2010;20:651-61. 10. Sundberg C, Eiken O, Nygren A, Kaijser L. Effects of isch- aemic training on local aerobic muscle performance in man. Acta Physiol Scand. 1993;148:13-9. 11. Ohno H, Shirato K, Sakurai T, Ogasawara J, Sumitani Y, Sato S. Effect of exercise on HIF-1 and VEGF signaling. J Phys Fit Sports Med. 2012;1:5-16. 12. Lundby C, Jacobs RA. Adaptations of skeletal muscle mi- tochondria to exercise training. Experimental physiology. 2016;101:17-22. 13. Conceicao MS, Gaspari AF, Ramkrapes APB, Junior EMM, Bertuzzi R, Cavaglieri CR. Anaerobic metabolism induces greater total energy expenditure during exercise with blood flow restriction. PLoS One. 2018;13:e0194776. 14. Corvino RB, Rossiter HB, Loch T, Martins JC, Caputo F. Physiological responses to interval endurance exercise at different levels of blood flow restriction. Eur J Appl Physiol. 2017;117:39-52. 15. de Oliveira MF, Caputo F, Corvino RB, Denadai BS. Short- term low-intensity blood flow restricted interval training im- proves both aerobic fitness and muscle strength. Scand J Med Sci Sports. 2016;26:1017-25. 16. Kelly LP, Basset FA. Acute Normobaric Hypoxia Increases Post-exercise Lipid Oxidation in Healthy Males. Front Physi- ol. 2017;8:293. 17. Joyner MJ, Casey DP. Regulation of increased blood flow (hy- peremia) to muscles during exercise: a hierarchy of competing physiological needs. Physiol Rev. 2015;95:549-601. 18. Takano H, Morita T, Iida H, Asada K, Kato M, Uno K. Hemo- dynamic and hormonal responses to a short-term low-inten- sity resistance exercise with the reduction of muscle blood flow. Eur J Appl Physiol. 2005;95:65-73. 19. Hamaoka T, McCully KK, Quaresima V, Yamamoto K, Chance B. Near-infrared spectroscopy/imaging for monitoring muscle oxygenation and oxidative metabolism in healthy and dis- eased humans. J Biomed Opt. 2007;12:062105. 20. Horiuchi M, Okita K. Blood flow restricted exercise and vas- cular function. Int J Vasc Med. 2012;2012:543218. 21. Tran TK, Sailasuta N, Kreutzer U, Hurd R, Chung Y, Mole P, Kuno S, Jue T. Comparative analysis of NMR and NIRS mea- surements of intracellular PO2 in human skeletal muscle. Am J Physiol. 1999 Jun;276:R1682-90. 22. Nioka S, Kime R, Sunar U, Im J, Izzetoglu M, Zhang J. A nov- el method to measure regional muscle blood flow continuous- ly using NIRS kinetics information. Dyn Med. 2006;5:5. 23. Ganesan G, Cotter JA, Reuland W, Cerussi AE, Tromberg BJ, Galassetti P. Effect of blood flow restriction on tissue oxygenation during knee extension. Med Sci Sports Exerc. 2015;47:185-93. 24. Suga T, Okita K, Morita N, Yokota T, Hirabayashi K, Horiuchi M. Intramuscular metabolism during low-intensity resis- tance exercise with blood flow restriction. J Appl Physiol. 2009;106:1119-24. 25. Spranger MD, Krishnan AC, Levy PD, O'Leary DS, Smith SA. Blood flow restriction training and the exercise pressor reflex: a call for concern. Am J Physiol Heart Circ Physiol. 2015;309:H1440-52. 26. Mendonca GV, Vaz JR, Teixeira MS, Gracio T, Pezarat-Cor- reia P. Metabolic cost of locomotion during treadmill walk- ing with blood flow restriction. Clin Physiol Funct Imaging. 2014;34:308-16. 27. Neto GR, Santos HH, Sousa JB, Junior AT, Araujo JP, Aniceto RR. Effects of high-intensity blood flow restriction exercise on muscle fatigue. J Hum Kinet. 2014;41:163-72. 28. Suga T, Okita K, Takada S, Omokawa M, Kadoguchi T, Yoko- ta T. Effect of multiple set on intramuscular metabolic stress during low-intensity resistance exercise with blood flow re- striction. Eur J Appl Physiol. 2012;112:3915-20. 29. Suga T, Okita K, Morita N, Yokota T, Hirabayashi K, Horiu- chi M. Dose effect on intramuscular metabolic stress during low-intensity resistance exercise with blood flow restriction. J Appl Physiol. 2010;108:1563-7. 30. Pearson SJ, Hussain SR. A review on the mechanisms of blood-flow restriction resistance training-induced muscle hy- pertrophy. Sports Med. 2015;45:187-200. 31. Cumming KT, Paulsen G, Wernbom M, Ugelstad I, Raastad T. Acute response and subcellular movement of HSP27, alphaB-crystallin and HSP70 in human skeletal muscle af- ter blood-flow-restricted low-load resistance exercise. Acta Physiol (Oxf). 2014;211:634-46. 32. Smiles WJ, Conceição MS, Telles GD, Chacon-Mikahil MP, Cavaglieri CR, Vechin FC, Libardi CA, Hawley JA, Camera DM. Acute low-intensity cycling with blood-flow restriction has no ef- fect on metabolic signaling in human skeletal muscle compared to traditional exercise. Eur J Appl Physiol. 2017;117:345-58. 33. Amann M, Eldridge MW, Lovering AT, Stickland MK, Pegelow DF, Dempsey JA. Arterial oxygenation influences central motor output and exercise performance via effects on peripheral loco- motor muscle fatigue in humans. J Physiol. 2006;15:937-52. 34. Miura H, McCully K, Nioka S, Chance B. Relationship be- tween muscle architectural features and oxygenation status determined by near infrared device. Eur J Appl Physiol. 2004; 91:273-8.

Muscle oxygenation, endocrine and metabolic regulation during low-intensity endurance exercise with blood flow restriction.

[]

Hwang, Hyejung,Mizuno, Sahiro,Kasai, Nobukazu,Kojima, Chihiro,Sumi, Daichi,Hayashi, Nanako,Goto, Kazushige

eng

PMC9730280

_VO2 kinetics and tethered strength influence the 200-m front crawl stroke kinematics and speed in young male swimmers Kamil Sokołowski1*, Raul Filipe Bartolomeu2,3,4, Tiago Manuel Barbosa2,3 and Marek Strzała1 1Department of Water Sports, Faculty of Physical Education and Sport, University of Physical Education, Kraków, Poland, 2Department of Sport Sciences and Physical Education, Instituto Politécnico de Bragança, Bragança, Portugal, 3Department of Sports Sciences, Polytechnic of Guarda, Guarda, Portugal, 4Research Center in Sports Sciences, Health and Human Development (CIDESD), Vila Real, Portugal Background: The aim of this research was to examine the relationship between the fast component of oxygen consumption developed in 1-min _VO2 and force indices both measured in tethered swimming test and to assess the influence of the gathered indices on speed and swimming kinematics in 200-m front crawl race. Methods: Forty-eight male swimmers (aged 13.5 ± 0.9 years old) participated in this study. Testing included 1) 1-min all-out front crawl tethered swimming while oxygen consumption (breath by breath) and tethered forces were measured, 2) 200-m front crawl race-like swimming featuring kinematic analysis, and 3) biological age (BA) examination. Results: During the 1-min all-out tethered swimming test, a linear increase in oxygen consumption was observed. There were moderate to high partial correlations between particular periods of seconds in the 1-min _VO2: 31–60, 41–60, and 51–60 and Fmax, Fave, and Iave of tethered swimming, while 41–60 and 51–60 _VO2 were moderately to highly interrelated with all the swimming speed indices and SI. The swimming speed indices significantly interplayed with SL, SI, Fmax, Fave, and Iave. Partial correlations were computed with BA control. Conclusion: The ability of reaching a high level of _VO2 fast is essential for a swimmer’s energy production at short- and middle-distance events. Reaching a high level of _VO2 significantly determines tethered strength and swimming kinematics. The level of _VO2 influences the maintenance of a proper pulling force and the stroke technique of front crawl swimming in young male swimmers. KEYWORDS adolescent swimming, oxygen uptake, tethered swimming, front crawl, biological age, kinematic indices OPEN ACCESS EDITED BY Philippe Hellard, Ministry of Education and Sport, Albania REVIEWED BY Sebastian Weber, INSCYD, Switzerland Santiago Veiga, Universidad Politécnica de Madrid, Spain *CORRESPONDENCE Kamil Sokołowski, kamil.sokolowski@awf.krakow.pl SPECIALTY SECTION This article was submitted to Exercise Physiology, a section of the journal Frontiers in Physiology RECEIVED 15 September 2022 ACCEPTED 07 November 2022 PUBLISHED 24 November 2022 CITATION Sokołowski K, Bartolomeu RF, Barbosa TM and Strzała M (2022), _VO2 kinetics and tethered strength influence the 200-m front crawl stroke kinematics and speed in young male swimmers. Front. Physiol. 13:1045178. doi: 10.3389/fphys.2022.1045178 COPYRIGHT © 2022 Sokołowski, Bartolomeu, Barbosa and Strzała. This is an open- access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms. Frontiers in Physiology frontiersin.org 01 TYPE Original Research PUBLISHED 24 November 2022 DOI 10.3389/fphys.2022.1045178 Introduction The ability to increase energy production is considered crucial in various sports, even in swimming where high velocities cause relatively high energy cost of movement. Thus, it is necessary among athletes of different age groups to develop either aerobic or anaerobic metabolic pathways of energy production. This begins with proper and adequate training from early prepubertal age and continues further with aging, while controlling the maturation level of the swimmer (Balyi and Way, 2009; Lätt et al., 2009). The contribution of energy pathways in swimming events is varied and depends on the duration of the race (Olbrecht, 2000). The 200-m front crawl, for example, is a race which requires a high involvement of aerobic and anaerobic pathways of energy production (Gastin, 2001). The aerobic energy system participates in the overall energy production right from the beginning of the all-out effort, and the oxygen uptake almost reaches its maximum level within 60 s of exercising (Gastin and Lawson, 1994; Serresse et al., 1988; Strzała and Tyka 2009). It has been stated that the maximal oxygen uptake ( _VO2 max) assesses the ability in developing and maintaining high speed of sprint swimmers in efforts lasting about 60 s (Ribeiro et al., 2015; Hellard et al., 2018). According to the data presented by Figueiredo et al. (2011), even in 200-m front crawl race, the aerobic pathway engages fast in providing energy for muscle work within half of the race, while at the third (long course) lap, aerobic metabolism provides for around 80% of all energy production. Among swimmers of different age groups, in the 200-m event, the aerobic contribution has been estimated to be 72% (Zamparo et al., 2000) or even 78.6% (Sousa et al., 2011). However, the contribution of the aerobic pathway of energy production in swimming at short and middle distances seems to have been underestimated over the past years (Peyrebrune et al., 2014). Rodriguez et al. (2003) have reported that swimmers not only reached 92.3% of their _VO2 max in the 100-m events but also exhibited _VO2 kinetics that was significantly faster in the 100-m race than in the 400-m one. Their results highlight the significance of fast oxygen kinetics especially while competing in short races, such as the 100-m ones. Despite the existence of research on the relationship between oxygen consumption and swimming performance, there is a need to refresh (Costill et al., 1985) and further investigate the fast component of _VO2 kinetics, i.e., the abrupt oxygen delivery to the body in short- to medium-term exercising periods. Moreover, there is a knowledge gap on the influence and dependence of this type of cardiorespiratory efficacy, present in most swimming races, on the ability to generate propulsion force and stroke kinematics. In swimming, the examination of specific strength abilities is deemed as a key factor when performing an evaluation. For this purpose, swimming tethered tests are often conducted in adults (Kjendlie and Thorsvald, 2006) and swimmers of other age groups (Amaro et al., 2014). Several studies have confirmed a strong relationship between tethered swimming tests (30–120 s) and short-to-middle distance swimming performances (Morouço et al., 2012; Santos et al., 2016). Biomechanical indices such as stroke length (SL), stroke rate (SR), and stroke index (SI) are significant predictors of young swimmers’ performance (Lätt et al., 2009) and are directly related to swimming efficiency (Geladas et al., 2005). The literature reports that strength preparation and a well-developed oxygen system should cause better stroke kinematics in terms of the ability to maintain proper SR and SL along the race (Costill et al., 1985; Sokołowski et al., 2021). Given these premises, the aim of this research was threefold: 1) to examine the relationship between the fast component of oxygen consumption and tethered swimming force production, 2) to examine the relationship between the fast component of oxygen consumption and 200-m front crawl race kinematics, and 3) to assess the relationship between 200-m front crawl race swimming kinematics and performance. It is hypothesized that there would be a significant relationship between oxygen uptake, tethered swimming force, stroke kinematics, and the performance indices. Materials and methods Participants Forty-eight young male swimmers [13.5 ± 0.9 years old; 14.55 ± 1.66 years of biological age (BA)] participated in this study. They were recruited as swimmers with the highest performance level in their age category from the Polish region of Krakow and were at the fifth threshold in the Ruiz-Navarro et al. (2022) classification of competitive level. Participants presented swimming levels which resulted in a mean value of 350.32 ± 60.22 FINA points for the 200-m front crawl race. All participants were clinically healthy and held a license from the Polish Swimming Federation. All swimmers had been through 4–5 years of systematic swimming at the time of conducting this research, encompassing at least 10 sessions per week and had taken part in national-level competitions and national swimming championships for their age group. 1-min Tethered swimming test A tethered swimming test (Figure 1) in a laboratory- controlled environment (temperature and humidity) was conducted. The test consisted of a single bout of 1-min duration of all-out freestyle tethered swimming and was performed in a flume in still water. With due advance notice, the swimmers were asked to rest the day before the test and maintain their daily diet. Before entering the pool, Frontiers in Physiology frontiersin.org 02 Sokołowski et al. 10.3389/fphys.2022.1045178 they were informed about the testing procedure and then underwent a 1000-m in-water warm-up, as before any competition. After the warm-up and before the test, they swam for 1 min in the flume at a slow pace, fully equipped with the testing apparatus for adjusting to the testing conditions. At this time, they got the possibility to familiarize with the specific environment of the flume and potential inconveniences of using the breathing apparatus and tethered swimming. After the initial 1 min of familiarization, the scientist conducting the test received feedback from the participant. To signal the beginning and ending of the test, a whistle was used. For the last minutes of warm-up and the test itself, the swimmers were asked to breathe only through the mouthpiece and avoid losing their nose clip. This procedure is similar to their training sessions done using a snorkel. The swimmers were equipped with a respiratory valve system that featured an ergospirometer (Start 2000 MES, Poland). The valve system was attached to a rod-like construction just above the swimmer’s head. During the duration of the test, the expired air was analyzed continuously (breath by breath) (Ergo 2000M software MES, Poland) and data were saved for further analysis. This has been proved to be a reliable method of calculating oxygen uptake in swimming (Neiva et al., 2017; Ribeiro et al., 2015; Sousa et al., 2011). From the collected data, the following indices were computed: 1) average oxygen consumption from the first 30 s of the test (1–30 _VO2, l·min−1), 2) average oxygen consumption from the last 30 s of the test (31–60 _VO2, l·min−1), 3) average oxygen consumption from the last 20 s of the test (41–60 _VO2, l·min−1), 4) average oxygen consumption from the last 10 s of the test (51–60 _VO2, l·min−1), and 5) oxygen consumption from the total test duration (1–60 _VO2, l·min−1). Additionally, the participants wore a nylon waist belt, connected by a 3.7 m steel cable to a load cell (ZPS5-BTU- 1kN, Poland) which was fixed on a steel pole (the fixing point is 0.49 m above the water surface). Data were recorded by the load cell at 100 Hz and transferred to a computer software program for further analysis (MAX6v0M software, Poland). Three parameters were calculated over a 60-s recording time: 1) maximum value of force (Fmax, N); 2) average value of force in the entire test (Fave, N) and in the first and second 30-s parts: Fave 0-30, Fave 30-60, N; and 3) average impulse per single cycle (Iave, N·s−1) which is defined as the integral of force over a period of time containing all full cycles divided by the number of completed cycles: Iave 
t1 t0Fdt n (1) where t0 is the beginning of the first full cycle and t1 is the ending of the last full cycle in the 60-s period. Tethered swimming has been described as a reliable method to assess swimming force production (Kjendlie and Thorsvald, 2006; Psycharakis et al., 2011; Amaro et al., 2014). 200-m Front crawl race The 200-m all-out test was carried out in a 25-m swimming pool that meets the International Swimming Federation (FINA) requirements. Before the race, the swimmers completed a 1000- m warm-up just like in competitions. Each trial was performed by three to four swimmers in order to mimic competition conditions. The final and split times of each trial were measured with an automatic timing device (Omega, Switzerland; OCP5, StartTime V). All trials were recorded with a camera at 50 Hz framing (GC-PX100BE, JVC, Japan). The velocity of the part of the race containing the first 10-m start zone as well as start, turn, and finish (which resulted in 115 m) was calculated as VSTF (m·s−1). The surface swimming velocity, i.e., the velocity over the effective clean swimming distance (85 m) was deemed Vsurf (m·s−1). The times for separate sectors were measured when the swimmer’s head crosses the imaginary line linking the markers at both sides of the pool. The 200-m front crawl velocity (Vtotal200, m·s−1) was defined as 200 divided by the final time of the race. The video footage, placement of the cameras and markers, video analysis, and computation of the basic kinematic parameters were performed analogically to the ones described in the literature (Sokołowski et al., 2021), but in this study, a swimming distance twice as long was considered. Kinematic parameters For the kinematic analysis, the stroke rate (SR), stroke length (SL), and stroke index were calculated. The SR was defined as the number of full stroke cycles performed within a unit of time (in FIGURE 1 1-min tethered swimming test. Frontiers in Physiology frontiersin.org 03 Sokołowski et al. 10.3389/fphys.2022.1045178 cycles per minute) and was calculated by video analysis of three consecutive stroke cycles (intraclass correlation of 0.99, 95% CI = 0.960–0.997). The SL was defined as the horizontal distance that the body travels during a full stroke cycle and was calculated as SL  v SR (2) where SL (in m) is the stroke length, v is the swimming velocity, and SR is the stroke rate. Finally, the SI was deemed as an overall swimming efficiency estimator and computed as SI  SL · v (3) where SI (in m2·s−1) is the stroke index, SL is the stroke length, and v is the swimming velocity. Biological age Examination of the participants in terms of BA was conducted by an experienced anthropologist and calculated as BA  BHage + BMage 2 (4) where BHage is the age obtained from the percentile charts based on the participant’s body height and BMage is the age obtained from the percentile charts based on the participant’s body mass. The growth charts by the Children’s Memorial Health Institute, which are standardized and validated for the Polish population, were used (the 50th percentile was used to align the height and mass with age). Additionally, pubertal development was assessed. The Tanner stages based on pubic hair scale were estimated (Bornstein, 2018). The great variety of biological maturation levels in the adolescent groups at the same calendar age causes great differences in muscle mass and aerobic and anaerobic capacities of swimmers. Because of differences in maturation specific water abilities of swimmers and specific testing could be less correlated with swimming performance than simple general tests as isometric force or counter movement jump (Garrido et al., 2012; Strzała et al., 2019). BA may cause bias in the statistical analysis and conclusions. The use of partial correlation statistics with age control helps limit the strong influence of BA in the effects of statistical calculations. The data used in biological age calculation are presented in Figure 2. Statistical analysis The values are presented as mean ± standard deviation. The normality of the data was checked with the Kolmogorov–Smirnov test. In oxygen consumption averaged per 10-s periods, the trend that was most suitable for the gathered data (Figure 3) was identified. The paired-sample t-test was used to compare the values of the average tethered swimming force of the first and second parts of the 1-min tethered swimming test. To identify the relationship between all the variables and swimming velocities in the 200-m front crawl, partial correlations controlled for BA were computed for 1) oxygen consumption and force indices; 2) oxygen consumption, swimming speed variables, and kinematic indices; and FIGURE 2 Average data of BHage, BMage, and BA. FIGURE 3 Average oxygen consumption of all participants, in 10-s periods, during the 1-min tethered swimming test. Frontiers in Physiology frontiersin.org 04 Sokołowski et al. 10.3389/fphys.2022.1045178 3) swimming speed variables and kinematic and force indices. The magnitude of the correlations was determined using the modified scale by Hopkins (2000)—trivial: r ≤ 0.1; low: 0.1 < r ≤ 0.3; moderate: 0.3 < r ≤ 0.5; high: 0.5 < r ≤ 0.7; very high: 0.7 < r ≤ 0.9; nearly perfect: r > 0.9; and perfect: r = 1. Results The data shown in Figure 3 represent the increase in oxygen consumption in the 1-min all-out tethered swimming test, in 10-s periods. The analysis of variance revealed significant differences between values measured every 10 s (F = 164,9, p < 0.01). Further trend analysis indicates the linear trend as the best adjusted to the collected data (F = 289,44, p < 0.01). There were moderate to high correlations between 31–60 _VO2, 41–60 _VO2, and 51–60 _VO2 and all the swimming force indices (Fmax, Fave, Iave). Low correlations were observed between Fmax, Iave, and 1–60 _VO2 (Table 1). A significantly higher average of tethered force was noted in the first 30-s duration of the test: Fave 0-30 85.41 ± 21.41 N vs Fave 30-60 67.12 ± 15.22 (t = 14.77; df = 47; p ≤ 0.0000). The 41–60 _VO2 and 51–60 _VO2 were moderately to highly correlated with all the swimming speed indices and SI. Vsurf was also significantly correlated with 1–30 _VO2 (Table 2). There was a positive correlation between SL and 51–60 _VO2. TABLE 1 Partial correlations controlled for BA between oxygen consumption and force indices from the tethered swimming test. Fmax (N) Fave (N) Iave (N·s−1) 250.24 ± 58.39 74.90 ± 20.63 101.93 ± 23.48 1–30 _VO2 (l·min−1) 0.167 0.053 0.134 1.68 ± 0.59 31–60 _VO2 (l·min−1) 0.296* 0.363** 0.372** 3.30 ± 0.76 41–60 _VO2 (l min−1) 0.395** 0.494** 0.502** 3.65 ± 0.81 51–60 _VO2 (l min−1) 0.482** 0.516** 0.559** 3.92 ± 0.97 1–60 _VO2 (l min−1) 0.285* 0.245 p = 0.054 0.290* 2.55 ± 0.59 *p ≤ 0.05; **p ≤ 0.01. TABLE 2 Partial correlations controlled for BA between oxygen consumption indices from the tethered swimming test, and swimming speed variables and kinematic indices from the 200-m front crawl race. Vtotal200 Vsurf VSTF S SL SI (m·s−1) (m·s−1) (m·s−1) (cycles·min−1) (m) (m2·min−1) 1.40 ± 0.09 1.34 ± 0.09 1.46 ± 0.10 41.68 ± 4.52 1.93 ± 0.24 2.53 ± 0.42 1-30 _VO2 (l·min−1) 0.187 0.299* 0.106 0.080 0.076 0.206 1.68 ± 0.59 31-60 _VO2 (l·min−1) 0.294 0.311 0.288 -0.083 0.206 0.283 3.30 ± 0.76 41-60 _VO2 (l·min−1) 0.463* 0.428* 0.487* -0.136 0.310 0.412* 3.65 ± 0.81 51-60 _VO2 (l·min−1) 0.640** 0.584** 0.666** -0.119 0.393* 0.539** 3.92 ± 0.97 1-60 _VO2 (l·min−1) 0.242 0.311 0.201 -0.007 0.155 0.255 2.55 ± 0.59 p = 0.075 *p ≤ 0.05; **p ≤ 0.01. Frontiers in Physiology frontiersin.org 05 Sokołowski et al. 10.3389/fphys.2022.1045178 Regarding the swimming speed and kinematic variables, the strongest relationships were observed between SI and Vtotal200 and Vsurf and VSTF. The swimming speed was also moderately correlated with SL, Fmax, Fave, and Iave (Table 3). As a supplement to the results, it was decided to present the level of selected oxygen uptake and strength indicators, measured in the 1-min test, followed by the kinematics of 200-m front crawl in relation to BA (Table 4). It could be observed that oxygen uptake and strength abilities continuously improve with higher BA. There was also a general increase in values of stroke kinematics through the years of BA. Table 5 shows 200-m front crawl kinematics by each 50- m lap. Discussion Regarding the analysis of _VO2 kinetics, an instantaneous and sudden increase was observed along the 1-min all-out tethered swimming. Despite the increase in _VO2 which could be characterized as a linear increase, the slopes in both initial and final segments of the 1-min consumption were noticeably lower than the one observed at the middle (Figure 3). Slower oxygen uptake at the beginning of the test may be associated with the use of high-energy phosphocreatine resources and yet low ventilation ( _V E); the final slowdown in _VO2 growth is from reaching a peak and increasing fatigue. This study revealed a significant influence of _VO2 (mainly 41–60 _VO2 and 51–60 _VO2) on 200-m front crawl race swimming speed, swimming kinematic indices, and tethered force indices. A highly developed fast _O2 supply to working muscles (represented by 51–60 _VO2) is significantly related to strength (0.482 ≤ r ≤ 0.559, p ≤ 0.01). This strength in swimming is expressed as the ability to TABLE 3 Partial correlations controlled for BA between swimming speed variables and kinematic indices from the 200-m front crawl race and the force indices from the tethered swimming test. SR SL SI Fmax Fave Iave (cycles·min−1) (m) (m2·s−1) (cycles·min−1) (cycles·min−1) (cycles·min−1) Vtotal200 0.168 0.325* 0.680** 0.341** 0.321** 0.406** Vsurf 0.229 0.301* 0.692** 0.321** 0.408** 0.387** VSTF 0.103 0.337* 0.644** 0.355** 0.411** 0.407** *p ≤ 0.05; **p ≤ 0.01. TABLE 4 Average values of oxygen uptake, tethered swimming, and kinematic indices of 200-m front crawl calculated for biological age. BA (years)/number of participants (n) 51–60 1–60 Fave Iave SR SL SI Vtotal200 _VO2 _VO2 (N) (N·s−1) (c·min−1) (m) (m2·min−1) (m·s−1) (l·min−1) (l·min−1) 11 (n = 1) 2.03 1.33 61.9 89.41 38.88 2.12 2.89 1.43 12 (n = 5) 3.10 1.95 47.31 74.05 45.81 1.72 2.21 1.36 13 (n = 15) 3.42 2.39 65.18 88.31 42.38 1.87 2.45 1.38 14 (n = 5) 3.47 2.38 70.82 98.14 38.62 1.96 2.34 1.31 15 (n = 9) 4.54 2.77 88.6 117.78 41.70 1.99 2.69 1.45 16 (n = 7) 4.41 3.00 82.73 112.44 40.96 1.93 2.49 1.40 17 (n = 4) 4.75 2.91 96.65 123.83 40.09 2.13 2.94 1.48 18 (n = 2) 5.56 3.13 100.95 137.69 40.82 2.08 2.90 1.51 TABLE 5 Average values of kinematic indices for each 50-m lap of 200- m front crawl. I 50 II 50 III 50 IV 50 SR (cycles·min−1) 42.91 ± 5.49 40.44 ± 4.63 39.99 ± 4.89 43.93 ± 4.87 SL (m) 1.97 ± 0.29 1.92 ± 0.24 1.90 ± 0.24 1.85 ± 0.23 SI (m2·min−1) 2.75 ± 0.54 2.46 ± 0.41 2.40 ± 0.41 2.49 ± 0.42 Vsurf (m·s−1) 1.45 ± 0.11 1.29 ± 0.09 1.26 ± 0.10 1.32 ± 0.09 Frontiers in Physiology frontiersin.org 06 Sokołowski et al. 10.3389/fphys.2022.1045178 produce propulsive force, which is later translated into higher stroke efficiency and thus better swimming economy (51–60 _VO2 vs SI, r = 0.539, p ≤ 0.01). Similarly, the higher energy demand connected with 51–60 _VO2 translated into significantly higher Vsurf (r = 0.584, p ≤ 0.05), which depended on proper swimming economy, due to the relationship between Vsurf and SI (r = 0.692, p ≤ 0.01) and Iave (0.387, p ≤ 0.01). This study noted a relationship between 51–60 _VO2 and the overall performance in 200-m front crawl (r = 0.640, p ≤ 0.01) which is in tandem with the results of Rodriguez et al. (2003), where a correlation between _VO2 peak values and the performance at 100 m (r = 0.787, p ≤ 0.05) and 400 m (r = 0.752, p ≤ 0.05) was observed. The reason for a weaker correlation in our study could be the longer period considered for the mean _VO2 calculation. We used 10-s periods, while Rodriguez et al. (2003) used 5-s periods. The breath-by-breath acquisition technique can induce a significant variability on acquired _VO2 values, and different sampling periods might produce different outcomes. Moreover, our quite restrictive statistical calculations (including BA control) could also play a role in that difference. In comparison to the results of Sousa et al. (2011), which showed a positive correlation between 200-m front crawl swimming speed and _VO2 peak (r = 0.69, p = 0.03), our partial correlation was somewhat slightly lower (r = 0.640, p ≤ 0.01). Nevertheless, these researchers found high _VO2 values right after the first 50 m that swimmers could almost maintain for the 200-m effort. Researchers have put forward that the need for oxygen in the muscles triggers an instantaneous and sudden increase in O2 uptake from the very beginning of the exercise (Ribeiro et al., 2015; Hellard et al., 2018). Maybe the highest peak of O2 uptake could be reached even faster in our study and show faster kinetics in young athletes, but because it is in swimming, the aim of racing (also through the test) is to withstand the pace as much as possible until the end of the race. Nevertheless, in our research, we recorded a positive distribution of average tethered swimming force (Fave 0-30 85.41 ± 21.41 N vs Fave 30-60 67.12 ± 15.22 N). The question here is how speedily and individually for a competitor, should a race be open to young 13-year-old swimmers in order to allow for the proper engagement of the fast component of oxygen consumption. It is known that positive pacing, or rather starting a race too speedily, can cause excessive fatigue, low oxygen distribution, and lactic acidosis in the skeletal muscles, which slow down energy production in the aerobic pathway. It may also be due to fatigue of the chest breathing muscles during the second part of the 200-m distance (Gastin and Lawson, 1994). It can be stated that for high aerobic capacity, the fast development of high level of O2 supply is crucial while performing middle distance events such as the 200-m front crawl. For this purpose, the 1-min tethered swimming test seems to be appropriate in examining the ability to supply O2 to the swimmer’s muscles to produce propulsion. Serresse et al. (1988) who examined the maximum 90-s ergocycle test observed that the highest _VO2 values occurred at about 60 s into the test. Similar to our study, their results have shown a linear increase in oxygen uptake up to 60 s into the test. Gastin and Lawson (1994) stated that 30–60 s of maximum effort could be enough to reach up to 90% of athletes’ _VO2 max. Ribeiro et al. (2015) claimed that if the majority of the swimming races are 50, 100, and 200 m, performed at high speeds, examining the _VO2 max at low intensities has limited application in the evaluation of the swimmer’s conditioning. Alves et al. (2011) suggested that faster kinetics during the initial phase of V _O2 max testing is directly related to a better performance at middle-distance events in swimming. Based on this reasoning, one could suggest that middle-distance swimmers should undergo long, high-intensity aerobic repeated sprints in training sessions. Regarding tethered force production, in the present study, a significant positive correlation was found between all indices and 200-m front crawl speed (0.321 ≤ r ≤ 0.411, p ≤ 0.01). Other authors have reported similar findings: Santos et al. (2016) have noted a positive correlation (0.61, p < 0.001) between the peak force of the 2-min tethered swimming test and clean velocity of 200-m front crawl race, while Morouço et al. (2012) showed a very strong relationship between average pulling force, peak force, and 200-m front crawl velocity (r = 0.94 and r = 0.93, respectively, p < 0.01). Again, controlling for BA and longer test duration could be the reasons for weaker correlations in our study. Our study showed great diversity in BA (Figure 2; Table 4). It is therefore a practical example of emphasizing the need for each trainer to adapt their training in relation to the BA of their swimmers. If this is the case, even the most gifted swimmers with delays in relation to BA are often frustrated by worse athletic performance when compared to their calendar peers, and in consequence, they overtrain trying to catch up to the others, get disappointed, then quit their swimming training. On the other hand, swimmers more advanced in relation to BA have the potential to develop through more individualized, intense training. Based on the high correlation between 51–60 _VO2 and SI found in the present study (r = 0.539, p ≤ 0.01), we can state that peak oxygen consumption determined the rate of transfer from chemical energy to mechanical energy, thus leveling up the stroke kinematics of the swimmers. This finding backs up the results by Sánchez and Arellano (2002), where the SI was found to be higher in international-level swimmers than their national-level counterparts in all swim strokes. Barbosa et al. (2013) proposed a multidisciplinary model of swimming performance predictors where the SI plays a significant role. In a study by Costill et al. (1985), the predictability of _VO2 max at freestyle was reported to increase significantly when the SI was included in the multiple regression analysis of an approximate 400-m swim. The multiple regression models prepared by Mezzaroba and Frontiers in Physiology frontiersin.org 07 Sokołowski et al. 10.3389/fphys.2022.1045178 Machado (2013) revealed that in young male swimmers, the SI at the 200-m front crawl race explained 76% of the performance. In the study by Nasirzade et al. (2015), 200-m front crawl performance of young swimmers was strongly related to the SL and SI (r = −0.79 and r = −0.72, p < 0.01, respectively). The mentioned studies are in tandem with our results where SI presented the highest positive correlation with all 200-m front crawl variables (0.644 < r ≤ 0.692, p ≤ 0.01). This very high percentage of share of the SI in performance in the abovementioned studies is also because of its link to performance itself, because the stroke index contains the speed (according to the formula: SI  SL · v). The present study, analyzing the relationship between the aerobic conditioning level, force production, and stroke kinematics is in accordance with the one study found in the literature on this matter, where Costill et al. (1985) identified interrelationships between oxygen uptake, energy cost of swimming, and stroking economy (SI). In our study, we found moderate to high correlations between 31–60 _VO2, 41–60 _VO2, and 51–60 _VO2 and Fmax, Fave, and Iave. Low correlations were observed between Fmax, Iave, and 1–60 _VO2. It could be stated that the ability to generate the pulling force is directly and positively related to the fast O2 supply which is linked with the endurance of the swimmer in terms of aerobic energy production and also lactate utilization or turnover to ATP (Greenwood et al., 2008). Conclusion In the 1-min all-out effort, a sudden increase in oxygen uptake was observed, with swimmers reaching high levels of _VO2 by the end of the tethered test. This fast ability of reaching high _VO2 and trainability of this physiological variable is essential for fitting an appropriate pacing in middle-distance racing and must be an important aspect of 13-year-old swimmers’ conditioning and of the older age groups too, in relation to their BA. Furthermore, it is suitable for the physiological preparation for 200-m front crawl performance and can be useful as a predictor of the swimmer’s endurance. The high intensity _VO2 testing used in the present study is appropriate for predicting sprint (100-m) and middle-distance swimming events performed at high speeds. There is a relationship between the fast-developed 1-min high-level oxygen uptake and the tethered strength abilities and high-speed swimming. The fast O2 supply is crucial for maintaining a proper pulling force and stroke technique. Data availability statement The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation. Ethics statement The studies involving human participants were reviewed and approved by the Regional Medical Chamber in Cracow; decision number: 94/KBL/OIL/2020. Written informed consent to participate in this study was provided by the participants’ legal guardian/next of kin. Author contributions KS collected data, performed statistical analysis, and wrote the manuscript. RB cowrote the manuscript. TB reread and corrected the manuscript. MS cowrote the manuscript and collected data. Funding Article processing charge (open access) was funded within the framework of the programme of the Ministry of Science and Higher Education (Poland) under the name “Regional Initiative for Perfection” within the years 2019–2022, project No. 022/RID/ 2018/19 in the total of 11,919,908 PLN. Conflict of interest The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Publisher’s note All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, editors, and reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher. References Alves, F., Reis, J. F., Alves, F. B., Bruno, P. M., Vleck, V., and Millet, G. P. (2011). Oxygen uptake kinetics and middle distance swimming performance. J. Sci. Med. Sport 15, 58–63. doi:10.1016/j.jsams.2011.05.012 Amaro, N., Marinho, D. A., Batalha, N., Marques, M. C., and Morouço, P. (2014). Reliability of tethered swimming evaluation in age group swimmers. J. Hum. Kinet. 41, 155–162. doi:10.2478/hukin-2014-0043 Frontiers in Physiology frontiersin.org 08 Sokołowski et al. 10.3389/fphys.2022.1045178 Balyi, I., and Way, R. (2009). The role of monitoring growth in long-term athlete development Canadian sport for life. Available at: http://lthd.fieldhockey.ca/files/ CS4LDocs/Monitoring Growth.pdf (Accessed February 9, 2022). Barbosa, T., Costa, M., and Marinho, D. (2013). Deterministic model-swimming performance. Available at: http://www.swimkinetics.isosc.org/ (Accessed June 17, 2022). Bornstein, M. H. (2018). The SAGE encyclopedia of lifespan human development. National Institute of Child Health & Human Development, Bethesda, Maryland. doi:10.4135/9781506307633 Costill, D. L., Kovaleski, J., Porter, D., Kirwan, J., Fielding, R., and King, D. (1985). Energy expenditure during front crawl swimming: Predicting success in middle- distance events. Int. J. Sports Med. 6, 266–270. doi:10.1055/s-2008-1025849 Figueiredo, P., Zamparo, P., Sousa, A., Vilas-Boas, J. P., and Fernandes, R. J. (2011). An energy balance of the 200 m front crawl race. Eur. J. Appl. Physiol. 111, 767–777. doi:10.1007/s00421-010-1696-z Garrido, N. D., Silva, A. J., Fernandes, R. J., Barbosa, T. M., Costa, A. M., Marinho, D., et al. (2012). HigH level swimming performance and its relation to non-specific parameters: A cross-sectional study on maximum handgrip isometric strengtH. Percept. Mot. Ski. 114, 936–948. doi:10.2466/05.10.25.PMS.114.3.936-948 Gastin, P. B., and Lawson, D. L. (1994). Variable resistance all-out test to generate accumulated oxygen deficit and predict anaerobic capacity. Eur. J. Appl. Physiol. Occup. Physiol. 69, 331–336. doi:10.1007/BF00392039 Gastin, P. (2001). Energy system interaction and relative contribution during maximal exercise. Sports Med. 31, 725–741. doi:10.2165/00007256-200131100- 00003 Geladas, N. D., Nassis, G. P., and Pavlicevic, S. (2005). Somatic and physical traits affecting sprint swimming performance in young swimmers. Int. J. Sports Med. 26, 139–144. doi:10.1055/s-2004-817862 Greenwood, J. D., Moses, G. E., Bernardino, F. M., Gaesser, G. A., and Weltman, A. (2008). Intensity of exercise recovery, blood lactate disappearance, and subsequent swimming performance. J. Sports Sci. 26, 29–34. doi:10.1080/ 02640410701287263 Hellard, P., Pla, R., Rodríguez, F. A., Simbana, D., and Pyne, D. B. (2018). Dynamics of the metabolic response during a competitive 100-m freestyle in elite male swimmers. journals.humankinetics.Com. 13 (8):1011-20. Available at: https:// journals.humankinetics.com/view/journals/ijspp/13/8/article-p1011.xml (Accessed November 4, 2022). doi:10.1123/ijspp.2017-0597 Hopkins, W. G. (2000). Measures of reliability in sports medicine and science. Sports Med. 30, 1–15. doi:10.2165/00007256-200030010-00001 Kjendlie, P. L., and Thorsvald, K. (2006). A tethered swimming power test is highly reliable. Portuguese J. Sport Sci. 6, 231–233. Available at: https:// www.researchgate.net/publication/267548775 (Accessed April 25, 2022). Lätt, E., Jürimäe, J., Haljaste, K., Cicchella, A., Purge, P., and Jürimäe, T. (2009). Physical development and swimming performance during biological maturation in young female swimmers. Coll. Antropol. 33 (1), 117–122. Mezzaroba, P. V., and Machado, F. A. (2013). Effect of age, anthropometry, and distance in stroke parameters of young swimmers. Int. J. Sports Physiol. Perform. 9, 702–706. doi:10.1123/IJSPP.2013-0278 Morouço, P. G., Vilas-Boas, J. P., and Fernandes, R. J. (2012). Evaluation of adolescent swimmers through a 30-s tethered test. Pediatr. Exerc. Sci. 24, 312–321. doi:10.1123/pes.24.2.312 Nasirzade, A., Sadeghi, H., Sobhkhiz, A., Mohammadian, K., Nikouei, A., Baghaiyan, M., et al. (2015). Multivariate analysis of 200-m front crawl swimming performance in young male swimmers. Acta Bioeng. Biomech. 17, 137–143. paper 17. doi:10.5277/ABB-00160-2014-03 Neiva, H. P., Marques, M. C., Barbosa, T. M., Izquierdo, M., Viana, J. L., Teixeira, A. M., et al. (2017). Warm-up for sprint swimming: Race-pace or aerobic stimulation? A randomized study. The Journal of Strength and Conditioning Research, 31 (9), 2423–2431. Olbrecht, J. (2000). The science of winning: Planning, periodizing and optimizing swim training. F&G Partners, luton, England. Peyrebrune, M. C., Toubekis, A. G., Lakomy, H. K. A., and Nevill, M. E. (2014). Estimating the energy contribution during single and repeated sprint swimming. Scand. J. Med. Sci. Sports 24, 369–376. doi:10.1111/j.1600-0838.2012.01517.x Psycharakis, S. G., Paradisis, G. P., and Zacharogiannis, E. (2011). Assessment of accuracy, reliability and force measurement errors for a tethered swimming apparatus. Int. J. Perform. Anal. Sport 11, 410–416. doi:10.1080/24748668.2011. 11868560 Ribeiro, J., Figueiredo, P., Sousa, A., Monteiro, J., Pelarigo, J., Vilas-Boas, J. P., et al. (2015). VO2 kinetics and metabolic contributions during full and upper body extreme swimming intensity. Eur. J. Appl. Physiol. 115, 1117–1124. doi:10.1007/ s00421-014-3093-5 Rodríguez, F. A., Keskinen, K. L., Malvela, M. T., and Keskinen, O. P. (2003). Oxygen uptake kinetics during free swimming: A pilot study. Biomechanics Med. Swim. IX, University of Saint-Etienne, Saint-Etienne, France, 379–384. Available at: https://www.researchgate.net/publication/234015321 (Accessed May 16, 2022). Ruiz-Navarro, J. J., López-Belmonte, Ó., Gay, A., Cuenca-Fernández, F., and Arellano, R. (2022). A new model of performance classification to standardize the research results in swimming. Eur. J. Sport Sci., 1–11. doi:10.1080/17461391.2022. 2046174 Sánchez, J., and Arellano, R. (2002). Stroke index values according to level, gender, swimming style and event race distance. Proc. XXth Int. symposium biomechanics sports 20, 56–59. Available at: https://ojs.ub.uni-konstanz.de/cpa/ article/view/620 (Accessed June 17, 2022). Santos, K. B., Bento, P. C. B., Pereira, G., and Rodacki, A. L. F. (2016). The relationship between propulsive force in tethered swimming and 200-m front crawl performance. J. Strength Cond. Res. 30, 2500–2507. doi:10.1519/JSC. 0000000000000410 Serresse, O., Lortie, G., Bouchard, C., and Boulay, M. R. (1988). Estimation of the contribution of the various energy systems during maximal work of short duration. Int. J. Sports Med. 9, 456–460. doi:10.1055/s-2007-1025051 Sokołowski, K., Strzała, M., Stanula, A., Kryst, Ł., Radecki-Pawlik, A., Krężałek, P., et al. (2021). Biological age in relation to somatic, physiological, and swimming kinematic indices as predictors of 100 m front crawl performance in young female swimmers. Int. J. Environ. Res. Public Health 18, 6062. doi:10.3390/ijerph18116062 Sousa, A. C., Figueiredo, P., Oliveira, N. L., Oliveira, J., Silva, A. J., Keskinen, K. L., et al. (2011). VO2 kinetics in 200-m race-pace front crawl swimming. Int. J. Sports Med. 32, 765–770. doi:10.1055/s-0031-1279772 Strzała, M., Stanula, A., Krężałek, P., Ostrowski, A., Kaca, M., and Głąb, G. (2019). Influence of morphology and strength on front crawl swimming speed in junior and youth age-group swimmers. J. Strength Cond. Res. 33, 2836–2845. doi:10.1519/JSC. 0000000000002084 Strzała, M., and Tyka, A. (2009). Physical endurance, somatic indices and swimming technique parameters as determinants of front crawl swimming speed at short distances in young swimmers. Med. Sport. 13, 99–107. doi:10. 2478/v10036-009-0016-3 Zamparo, P., Capelli, C., Zamparo, P., Capelli, C., Cautero, M., Nino, A., et al. (2000). Energy cost of front-crawl swimming at supra-maximal speeds and underwater torque in young swimmers. Eur. J. Appl. Physiol. 83, 487–491. Article in European. doi:10.1007/s004210000318 Frontiers in Physiology frontiersin.org 09 Sokołowski et al. 10.3389/fphys.2022.1045178

<math xmlns="http://www.w3.org/1998/Math/MathML"> <mrow> <msub> <mrow><mover><mi>V</mi> <mo>˙</mo></mover> <mi>O</mi></mrow> <mn>2</mn></msub> </mrow> </math> kinetics and tethered strength influence the 200-m front crawl stroke kinematics and speed in young male swimmers.

11-24-2022

Sokołowski, Kamil,Bartolomeu, Raul Filipe,Barbosa, Tiago Manuel,Strzała, Marek

eng

PMC7117710

RESEARCH ARTICLE “Running with cancer”: A qualitative study to evaluate barriers and motivations in running for female oncological patients Alice AvanciniID1*, Kristina Skroce2, Daniela Tregnago3, Paolo Frada2, Ilaria Trestini3, Maria Cecilia Cercato4, Clelia Bonaiuto3, Cantor Tarperi2,5, Federico Schena2, Michele Milella3, Sara Pilotto3, Massimo Lanza2 1 Department of Medicine, Biomedical, Clinical and Experimental Sciences, University of Verona Hospital Trust, Verona, Italy, 2 Department of Neurosciences, Biomedicine and Movement Sciences, University of Verona, Verona, Italy, 3 Department of Oncology, University of Verona Hospital Trust, Verona, Italy, 4 Epidemiology and Cancer Registry Unit, Regina Elena National Cancer Institute, IRCCS, Rome, Italy, 5 Department of Clinical and Biological Sciences, University of Turin, Turin, Italy * alice.avancini@univr.it Abstract Nowadays, it is widely acknowledged that low physical activity levels are associated with an increase in terms of both disease recurrence and mortality in cancer survivors. In this light, deciphering those factors able to hamper or facilitate an active lifestyle is crucial in order to increase patients’ adherence to physical activity. The purpose of this study was to explore barriers and motivations in a sample of female oncological patients, practising running using the ecological model and compare them with healthy controls. Focus group interviews were conducted at Verona University. Participants were 12 female cancer survivors and 7 matched healthy controls who had participated at “Run for Science” project. The interviews were transcribed verbatim and analyzed using content analysis. Transcripts were catego- rized according to the ecological model, identifying barriers and motivations as themes. About motivations, three sub-themes were included: personal, interpersonal and environ- mental/organizational factors. Regarding barriers, another sub-theme was recognized: community/policy factors. Compared to healthy controls, survivors expressed motivations and barriers specifically related to their oncological disease. Running was a challenge with their cancer and a hope to give to other patients. Main barriers were represented by treat- ment-related side effects, inexperienced trainers and external factors, e.g. delivery of incor- rect information. Running programs dedicated to oncological patients should consider intrinsic obstacles, related to cancer and its treatment. The interventions should offer a per- sonalized program performed by qualified trainers, together with a motivational approach able to improve participants’ adherence to an active lifestyle. Introduction In Italy, one out of three women will experience an oncological disease during lifetime [1]. Cancer is the second most common chronic disease in female population and in 2018 more PLOS ONE PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 1 / 13 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Avancini A, Skroce K, Tregnago D, Frada P, Trestini I, Cercato MC, et al. (2020) “Running with cancer”: A qualitative study to evaluate barriers and motivations in running for female oncological patients. PLoS ONE 15(4): e0227846. https://doi.org/10.1371/journal.pone.0227846 Editor: Denis Martin, Teesside University, UNITED KINGDOM Received: December 24, 2019 Accepted: March 12, 2020 Published: April 2, 2020 Peer Review History: PLOS recognizes the benefits of transparency in the peer review process; therefore, we enable the publication of all of the content of peer review and author responses alongside final, published articles. The editorial history of this article is available here: https://doi.org/10.1371/journal.pone.0227846 Copyright: © 2020 Avancini et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: The data contained in the paper constitute our minimal underlying data set. than 1,870,000 women in Italy were living with a cancer diagnosis [1]. The introduction in clinical practice of innovative treatments have allowed cancer survivors to achieve an improved prognosis and quality of life. Nevertheless, cancer patients often experience impor- tant treatment-related side effects, involving both the physical and psychological spheres, hav- ing a potential prolonged impact on patients’ condition even after therapy conclusion [2]. An increasing amount of studies has demonstrated that physical activity (PA) and exercise (EX) are safe and feasible in the oncological setting. PA can support standard therapies, help- ing cancer survivors in reducing their risk of recurrence and mortality [3]. PA and EX can facilitate the management of some disease- and treatment-related effects, as fatigue, nausea and vomiting, increasing patients’ quality of life [4, 5]. Moreover, the EX and PA benefits include improvement in cardiorespiratory fitness, strength, flexibility and body composition [6, 7]. The American College of Sport Medicine recommends patients with cancer to avoid inactivity and engage in at least 90 min/week of moderate-intensity aerobic PA, with strength EX two times per week [2]. One of the most common type of aerobic PA is running, not only for its physical and physi- ological benefits, but also for its accessibility and simplicity. A recent report indicated that there were 17.1 million running participants during the 2015 running season [8]. Running is the most widespread PA also in the cancer setting with an acknowledged beneficial impact [8]. Running confers numerous cardiovascular, metabolic, musculoskeletal and neuropsychiatric benefits and is strongly associated with lower body weight and smaller waist circumference [8]. This PA is shown to increase life-longevity and is often recommended as prevention and control for various chronic diseases, including cancer. Previous studies have identified differ- ent factors related to running motivation, as the desire to affiliate with other runners, an increase in self-esteem, physical motives for general health benefits, improving quality of life, coping with negative emotions and many more [9]. Despite many positive aspects connected with a more active lifestyle, there are many barriers that can interfere with EX adherence, par- ticularly speaking about running, which may be more physically and psychologically difficult than some other activities [10]. These motivations and barriers are connected not only with the momentary health status, but also with the previous health-related experiences [11]. Furthermore, individual behaviour may be influenced by many elements that interact with the person [12] [13]. This approach, also called ecological model assumes that individual competencies, intrapersonal relations, organisational or community structures and political choices can influence or determine the individual’s behaviour [12] in many fields, including physical activity and lifestyle. To date, no study investigated barriers and motivations in female cancer survivors performing running and compared them with their healthy controls. Therefore, the aim of this study was to qualita- tively investigate barriers and motivations, according to the ecological model, in a sample of female cancer survivors practising running and compare them with healthy controls. Materials and methods Design We conducted a series of focus group sessions among female adults affected or not by cancer to qualitatively assess barriers and motivations towards running. The study was approved by the local Ethical Committee (Department of Neurological, Neuropsychological, Morphological and Movement Science, University of Verona, Prot. No. 165038) and followed to Standards for Reporting Qualitative Research (SRQR) guidelines for qualitative research [14, 15]. PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 2 / 13 Funding: The authors received no specific funding for this work. Competing interests: The authors have declared that no competing interests exist. Participants and recruitment A purposive sample was employed to recruit women who had participated at “Run for Science” (R4S) project [16]. Inclusion criteria for the oncological group (OG) were: female participant, had been diagnosed with cancer, being  18 years of age and participating in R4S event. Regarding healthy controls (HC), women participating at R4S, with absence of chronic disease and 18 years of age or older were considered eligible. The inclusion criteria were assessed by AA through the database of R4S. Eligible women were contacted individually via email by the research team to introduce them the study. If they agreed to participate, AA contacted them by telephone to organize the interview. Written informed consent was obtained from included participants the day of the interviews, before starting the focus group. To protect participants’ identity pseudonyms were used to report the data. The “Run for Science” project The R4S, previously described [17], is a research project endorsed by the University of Verona, which involves Italian, European and American scientific institutions. The purpose of this event, coordinated by FS, CT, and KS, is to investigate several aspects regarding the effects of endurance running, and usually involves more than 200 volunteer runners every year. Data collection Focus groups were held, from April 2019-July 2019, in a meeting room at Department of Neu- roscience, Biomedicine and Movement of Verona University and lasted approximately 60 minutes. Overall, five focus groups were organized, three for oncological subjects (n = 4, 5 and 3) and two for healthy participants (n = 4 and 3). Interviews were conducted separately for the groups of women with a cancer diagnosis and the groups of healthy subjects. The reason for this choice was to make a more possible comfortable environment to bring out detailed infor- mation regarding own personal history. The interviews were carried out by ML and observed by AA and PF. ML is Associate Profes- sor in Sport Science and Methodology at Verona University with expertise in PA and health promotion. AA is a PhD student involved in EX in oncological patients, with previous inter- view experience and PF is a master’s degree student in preventive and adapted PA. Participants were asked about barriers and motivators to running, applying the ecological model. AA and the ML developed some semi-structured questions, based on previous studies [18, 19] to guide the interviews (Table 1). The interview guide was reviewed by DT, the dedicated psycho- oncologist working at Oncology Department of Verona University Hospital. All interviews were audio-recorded and transcribed verbatim. Data collection continued until saturation principle was reached, i.e. no new information seemed to emerge from the interviews. After each focus group session, a questionnaire to investigate the socio-demographic data (e.g. birth date, education level, marital status and occupational status) and clinical informa- tion (medical history) was provided to participants to complete. Perceived economic insecurity was assessed with the closed-ended question “How do you get to the end of the month, with your available financial income?” with four possible response (i.e. many difficulties/ some diffi- culties/ easily/ very easily). Analysis ML, AA and PF independently analysed the data, using the content analysis. This approach was performed with Atlas.tiTM software and involved a process of reading, reflection, decoding PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 3 / 13 and re-reading on the meaning of the data collected, in order to analytically interpret the text. First, the text was read several times to identify recurring ideas and to get a sense of the whole discussion. The second point included the formulation of codes summarizing the salient fea- tures of collected data. The third, was grouping the code into themes and eventually sub- themes. The final step involved all three authors with a process called triangulation. This con- sisted in presenting the emerged findings to the research team members, comparing the results and defining the final themes [20]. Moreover, the researchers compared the emerged themes from the HC and OG to find similarities and differences. Results All the invited cancer survivors (n = 12) participated to the study, while only 7 out of 13 healthy females completed the focus group. Table 2 illustrates the socio-demographic and medical characteristics of both groups. The transcripts were analyzed according to the ecologi- cal model and the following common themes were categorized to reflect the levels: 1) motiva- tions and 2) barriers in running. Theme 1: Motivations Features that have stimulated participant’s will to be or become active in everyday life, even after the conclusion of oncological treatments, include three main sub-themes: individual, interpersonal and organizational factors (Table 3). Individual factors. Different aspects connected with running were common in both groups, such as enjoyment, previous experience, as well as mental and physical benefits of exercising. Some women experienced a true well-being during their running workout, as reported by this woman: “I like running, I like the emotion of moving with my own legs in the environment, and the fatigue I feel is pleasant because it means that by this kind of practice I am moving towards my goal.” (Giovanna, OG). Other women perceived their workouts as a time of their everyday life where they enjoy themselves, as reported by this woman: “For me, it is enjoyment and passion. I started practicing sport while I was not young anymore and I literarily fell in love with running.” (Lara, HC). All women reported that their previous EX experience Table 1. Semi-structured interview questions. Motivations • From the personal point of view (thinking of physical and psychological state and previous experience) is there any factor that in your opinion may motivate the adherence to running program? • From the social point of view (thinking of relationships with other people, friends, colleagues, family) is there any factor that in your opinion may motivate the adherence to running program? • From the environmental point of view (thinking of place, organizations and institutions) is there any factor that in your opinion may motivate the adherence to running program? • From the cultural point of view (thinking of politics and national/regional rules) is there any factor that in your opinion may motivate the adherence to running program? Barriers • From the personal point of view (thinking of physical and psychological state and previous experience) is there any factor that in your opinion may limit the adherence to running program? • From the social point of view (thinking of relationships with other people, friends, colleagues, family) is there any factor that in your opinion may limit the adherence to running program? • From the environmental point of view (thinking of place, organizations and institutions) is there any factor that in your opinion may limit the adherence to running program? • From the cultural point of view (thinking of politics and national/regional rules) is there any factor that in your opinion may limit the adherence to running program? https://doi.org/10.1371/journal.pone.0227846.t001 PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 4 / 13 represented a positive motivator in building and maintaining their active lifestyle. Although the mental health benefits from exercise represented a common factor detected in both groups, Table 2. Participant’ characteristics. Oncological group (n = 12) Healthy group (n = 7) Agea, mean (SD) 50.5 (5.9) 47.5 (8.0) Body mass indexb, mean (SD) 21.9 (2.8) 22.1 (0.8) Education, N Secondary 1 0 High school degree 7 4 Undergraduate degree 3 2 Postgraduate degree 1 1 Marital status, N Unmarried 4 3 Married 7 4 Divorced 1 0 Employment, N Part time employed 8 3 Full time employed 4 4 Family incomec, N Many difficulties 1 0 Some difficulties 4 1 Easily 4 5 Very easily 3 1 METs—Physical activity, mean (SD) 3069.9 (1536.5) 2441.3 (1119.1) Tumor site, N Colorectal 2 - Hematologic 1 - Breast 9 - Stage, N Unknown 5 - Early 4 - Advanced 3 - Metastatic 0 - Months from diagnosis, mean (SD) 57.6 (34.5) - Undergone surgery, N 11 - Undergone chemotherapy, N 9 - Undergone radiation therapy, N 8 - Undergone hormone therapy, N 8 - Undergone others treatment, N 0 - Current treatment status, N Incoming 0 - Ongoing 0 - Ended 12 - SD, standard deviation, N, number; Mets, metabolic equivalent of the task expressed in minutes per week a Expressed in years b Expressed in units of kg/m2 c Perceived economic insecurity assessed by the question: How do you get to the end of the month, with your available financial income? https://doi.org/10.1371/journal.pone.0227846.t002 PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 5 / 13 origins and consequences were different. In particular, healthy subjects applied these benefits to deal with work, family or personal stress, as reported by Laura (HC): “If I’m tired and exhausted at the end of my working day, I usually go for a run and reach some kind of mental regeneration.” In contrast, oncological patients benefitted from running experience in terms of better facing the prescribed treatments, as reported by Elisa (OG): “I suffered a lot from the psy- chological point of view after radiotherapy and chemotherapy, but now I am feeling much better and as far as I understand this is due to my running workouts.” Other factors, such as the per- formance results connected to running, the fact that it is a cheap and easy to perform activity, were identified as personal motivation by the healthy group. In the oncological group, a crucial motivation was specifically related to the disease. In this regard, all the participants confirm that running means for them a personal challenge after cancer: “My main motivation is to show to myself that I can do it, I can do something incredible, like a half marathon, even after my cancer.” (Nicoletta, OG). Another important aspect recognized as a potent stimulus to running is to give hope to other patients: “I run to give hope to who is beginning the tumor winding path. Maybe they will see me and say: okay if she won it, I can do it too.” (Stefania, OG). Interpersonal factors. The relationship with others was an important motivator highlighted during the focus group interviews, in both the oncological and healthy groups. Training with other people was recognized as a vehicle of sociality able to increase motivation in running. Moreover, for OG, exercising with someone who shares similar disease-related experiences, helped them to remain motivated and active: “With these women I immediately found myself very well. We speak the same language because we share the same cancer history.” (Stefania, OG) and “Even if I cannot go, I say to myself: no, someone is waiting for me, I cannot skip, I need to go and workout with them.” (Elisa, OG). Family support is common in both groups. In the HC perspectives, partner stimulate the participants to train, as Lara (HC) told: “My husband encouraged me to run. He is a crucial support for me.”. In cancer survivors’ group, the family support resulted overall positive, but sometimes controversial. Some of them were encouraged, as Margherita (OG) remembered: “My dad is 85 years-old and he rides a bike. He Table 3. Motivation and barriers related to running EX identified by cancer survivors compared to healthy controls. Ecological model (level) Motivations Barriers Cancer survivors Healthy controls Cancer survivors Healthy controls Personal factors • Prior EX experiences • Prior EX experiences • Lack of time (in progress) • Lack of time • Enjoyment • Enjoyment • Injury • EX failure • Physical and mental benefits • Physical and mental benefits • Cancer-related treatment side effects • Cancer-related challenge • Positive EX results • Hope for other patients • Ex easy budget Interpersonal factors • EX group support • EX group support • Trainer not qualified • Lack of social support • Family support • Family support • Friends support • Physician support Environmental and organizational factors • Natural environment • Natural environment • Poor personal security • Poor personal security • Organized training • Untended environment • Untended environment • Air pollution Community and policy factors • Traditionalist culture • Running is underestimated compared to • EX only for athletes and body image • other sports • Incorrect information delivery https://doi.org/10.1371/journal.pone.0227846.t003 PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 6 / 13 always encourages me to stay physically active”. By contrary, others had some concerns, as Gio- vanna (OG) reported: “My parents did not want me to run, they told me you will be too much tired, you have to recover” or Nicoletta (OG) explained: “My husband recommended me not to exaggerate, because I could get injured like my colleagues did.” Nevertheless, oncological patients described that friends, as well as the medical staff, support their choice to begin a run- ning program. Daniela (OG) remembered: “When I decided to start a running program, a lot of my friends texted me an encouraging message to continue exercising” and Tony (OG) recounted: “My oncologist told me that I had to do this, that after my cancer I had to rebuild my life”. Environmental and organizational factors. For both groups, running in the natural environment is an important supportive factor to continue the activity. “Sometimes I go run- ning by the Garda lake, with a wonderful landscape, so it is a very pleasant environment for exercising. I feel less fatigue because I am concentrated on what my eyes see around me” said Antonella (OG), or “We live in a beautiful place that gives us the possibility to stay in touch with the nature and I like a lot running in this area” Federica (HC) remembered. Moreover, OG rec- ognized the great impact of training with an organized team, which provided them with a run- ning campus, a trainer to indicate and explain them the workouts they needed to do: “Have someone who follows you, like an organization, this is very motivating for me” (Giulia, OG). Theme 2: Barriers The interviews revealed various aspects that could interfere with the running EX. The identi- fied barriers were categorized into four sub-themes, including: personal, interpersonal, organi- zational and community-policy factors (Table 3). Individual factors. The personal barriers recognized as obstacles to running were differ- ent between the two groups. The only common aspect was lack of time dedicated to running, although the perspective regarding this potential barrier was different between OG and HC. For healthy subjects, lack of time emerged as the principal obstacle that interferes with run- ning: “Unfortunately I must give priority to the work and when I was preparing for my half mar- athon and needed to run for two hours, I could run only one hour and a half” (Erika, HC). Also from cancer survivors’ point of view, lack of time in EX could be a potential barrier, but most of them explained how cancer disease changed this opinion: “In a typical day it is difficult to cut out some time for EX because you have to work, prepare the dinner for your family, stay with your son because these are the priorities. After my cancer, I said to myself that now I exist! Now I can find my space and my time for EX, I demand it!” (Antonella, OG). In OG, a general consensus confirmed that injuries and treatment-related side effects repre- sent potential obstacles for running. In particular, injuries of other training partners were indi- cated as reasons to discontinue running, how Elisa (OG) and Nicoletta (OG) reported: “When I had a knee injury, I was strongly tempted to stop running, to give up the group” and “When four out of eight colleagues were injured, I thought of interrupting my training session because I did not want to hurt myself”. Concerns about cancer- and treatment-related side effects were indicated as strong factors that may obstacle running: “Hormonal therapy causes fatigue and joint pain, therefore sometimes it is very difficult for me to begin any exercise” (Nadia, OG). Mir- ella (OG) also reported: “My chemotherapy cycles were very long and hard. The main side effect that I experienced was peripheral neuropathy. Sometimes I had to interrupt running, because I had serious sensibility problem in my foots and I was afraid of hurting myself”. Finally, HC reported that failing in pre-established running performance was a serious obstacle to main- tain own training: “When you expect to run for example 10 kilometres with a faster pace and you cannot do it, you lose confidence in yourself and sometimes the temptation to give up is really strong” (Erika, HC). PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 7 / 13 Interpersonal factors. The OG referred that their trainers were not well prepared nor spe- cifically qualified for advising a patient with oncological disease and this was a major obstacle. “When I began to run my coach proposed me an overestimated program for my situation. After a month and a half my knees were blocked, I was in pain, I had difficulty to walk, I had to stop for one month and the temptation to interrupt was very strong” (Antonella, OG). Another partici- pant in the OG expressed concerns regarding the knowledge of some instructors: “I did not have a good trainer, I never performed a warm-up phase, or exercised to reinforce my muscle, and also from a human point of view the support was completely missing” (Ilaria, OG). Environmental and organizational factors. Poor personal security and uncontrolled environment were interrelated and represented a barrier for running in both the HC and OG. “I love running in the nature, but sometimes I meet weird people and I think: this way is not secure for running because I should run without listening to music in order to see if the person that stopped is following me” recounted Lara during an interview in the healthy group. Also, Margherita (OG) told: “I used to run on the bicycle lane and I always carried pepper spray with me because the environment was not controlled and I always had this feeling that someone was behind me, I did not feel comfortable”. However, this feeling of insecurity is magnified by poor maintenance of natural environment; in the OG: “Some areas are poorly managed, there is tall grass that nobody cuts, the plants are not pruned and grow everywhere and consequently I'm afraid to run in those places” (Rossella, OG). In addition, another problem for OG was air pol- lution: “Sometimes I decide to postpone my training due to poor air quality; I do not want to breathe toxic air.” (Ilaria OG). Another woman reported the difficulty to run in some areas because of air pollution: “In some places, smog is very high and I have to admit that it is really difficult to go out for a run.” (Margherita, OG). Community-policy factors. Even if both groups recognized that the sport bodies organise several running manifestations, they agreed on the fact that the actual Italian policy situation was not favourable on promoting running. As Paola (HC) said: “We live in a country where the main sport is football, the others are considered second class sports and, for this reason, are penal- ized”. Furthermore, the OG highlighted how the current traditionalist culture hindered the practice of PA in general: “We live in a traditionalist culture, in which we teach our sons to go to school, to work, to have a family. These are the priorities.” (Antonella, OG). Moreover, market- ing was reported as a negative factor that blocks the correct and healthy promotion of running in OG. In fact, it usually appears that running EX is only adequate for athletes or for physically active subjects, and it is always related to body image. In this regard, Rossella (OG) and Nadia (OG) remembered: “The current advertising and culture teach you to follow a woman model: lean, made up, that does not sweat; this is very disheartening for me.” or “Many information is incorrect and confounding; according to certain advertising you should train yourself to be cool and to have a beautiful body, not for health or for preventing or controlling chronic conditions.” Discussion To the best of our knowledge, this research represents the first qualitative investigation explor- ing motivations and barriers about running, as exercise training, in a group of female cancer survivors and compared them with matched healthy controls. We found several factors that stimulate the approach and adherence to running and others that limit them. Regarding running motivations, several points were common in both groups, such as enjoyment, possibility to perform this type of EX in a natural environment, social support given by teammates and attitude towards EX. These results are in line with previous data [21]. McIntosh et al. for example identified physical and psychological benefits together with social support as factors that stimulated patients who have had cancer to maintain their walking PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 8 / 13 activity [18]. Nevertheless, from cancer survivors’ perspective, other strong running motiva- tions, related to their health history, were identified. Running performance was a challenge connected with their disease and a sort of demonstration they could overcome cancer, giving also hope to other cancer patients. Moreover, the focus group highlighted that patients who have had an oncological disease obtained more support from their family, friends, physician and workout teammates compared to healthy controls. This result is supported by Husebø et al., who identified social support as a crucial component in influencing physical EX in women affected by breast cancer [22]. Regarding the environmental and organizational level, other motivations stimulated patients to maintain their running program, such as taking part in an organized training program and performing this activity in a natural environment. Doing EX outside is a common preference found in several other studies, in different cancer populations, while Blaney et al. reported that participating in an EX program, organized and supervised by an EX specialist was a strong motivator that seemed to offer assurance to survi- vors [23]. These findings support a series of recommendations that should be provided to can- cer survivors in order to propose a successful running program, e.g. increase knowledge regarding EX benefits and promote group training, as summarized in Fig 1. Focusing on barriers toward running, some environmental and organizational factors were similar between the oncological group and healthy subjects, such as poor personal security and untended environment. Another study has emphasized these obstacles mentioning that “safety Fig 1. Strategies to increase adherence and compliance in a running program. https://doi.org/10.1371/journal.pone.0227846.g001 PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 9 / 13 issues” were an impediment to patients affected by cancer walking activity [24]. In addition, they expressed many barriers related to their cancer journey [19, 23]. For example, cancer- related treatment side effects, such as fatigue, joint pain or peripheral neuropathy were identi- fied as serious impediments significantly interfering with the maintenance of running EX. Moreover, physical injuries, inexperienced trainer, air pollution and the public scarce attractiv- ity of running training have emerged as issues that can inhibit the adherence to a running pro- gram. Regarding EX security, a recent systematic review with metanalysis has investigated the safety and feasibility of EX among women affected by stage II-IV breast cancer. A total of 60 randomized controlled trials involving 5200 participants were included. The analysis showed no differences in adverse advents between EX and usual care, independently of EX supervision (EX supervised defined as over half of the Ex session involved face-to-face supervision) [25]. These findings support the EX safety, also in an unsupervised context, and therefore suggest that the fear of injuries observed in our oncological patients does not represent a real risk. Nev- ertheless, the psychological disease-related background might justify this concern. Indeed, a cancer diagnosis and its related treatments carry several physical and psychological impair- ments that alter the subject’s perspectives, e.g. changes in body composition and body image, physical deconditioning. Cancer survivors might not feel confident or capable of performing EX, and specifically running, consequently, they are afraid to undergo injuries and want, for this reason, assurance regarding the trainer’ professionality [26]. Therefore, the trainer should be able to reassure the participants about EX safety, personalizing the information and the instructions to provide. Moreover, after diagnosis, they usually search for additional informa- tion about their lifestyle (e.g. nutrition, smoking, alcohol consumption, PA) from several sources [27, 28]. Without adequate competence to correctly evaluate the quality of the col- lected information, there is the concrete risk of finding misleading news leading to unsafe and risky habits or that can induce excessive attention to those environmental factors potentially harmful as air pollution. One last element seems significant, even if ambivalent. The possibility of reliving the posi- tive emotions experienced in previous training experiences are indicated as significant motiva- tions by the OG. This element further supports the promotion of exercise and training experiences also in the general population because its lack, may decrease the possibility of reac- tion in case of illness. Even in this case some suggestions, based on the identified barriers, should be considered while planning a running program for cancer survivors (Fig 1). Nowa- days, some studies were conducted to improve EX adherence in cancer setting. Among them, Rogers and colleagues have proposed the BEAT trial (Better Exercise Adherence after Treat- ment) which aims to implement behaviour changes in breast cancer survivors by using the social cognitive theory. This dynamic model combines behavior, personal and environmental influences and, at the same time, includes barriers and facilitators in order to create a frame- work for the design of a durable physical activity intervention. In this study the participants were significantly more likely to meet physical activity recommendations both immediately post-intervention and after 3 months compared to control group, besides to show better improvements in fitness and quality of life [29]. These results confirm the importance of including EX barriers and motivators in planning an effective EX program. Focusing on run- ning, some projects (i.e. “Cancer to 5K”) proposed an EX training for cancer survivors, but not specific information regarding how the program was planned are available. To the best of our knowledge, some experiences have investigated the physical benefit of running in cancer[30], but no specific studies have organized the running program considering barriers and motivations. Our study has some limitations as the low response rate especially in the healthy group. Although we cannot guarantee that the saturation principle was achieved in HC, our study PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 10 / 13 mainly focused on oncological patients’ experiences and further investigations will be per- formed in order to confirm our findings. Moreover, it has to be acknowledged that the partici- pants with cancer were already motivated to run as demonstrated by their participation in the R4S event. The oncological group was affected by different cancer types and considering the peculiarity of the physical EX evaluated (endurance running), the results are not widely gener- alizable to other activities. Nonetheless, precisely because these conditions represent a real- world situation, we believe that it is interesting to understand factors that induced these sub- jects to approach and adhere to running EX. In conclusion, the current literature shows the strong importance of a constant PA, such as endurance running, after a cancer diagnosis in order to reduce recurrence risk and mortality. Exploring the factors that limit and favour the promotion of an active lifestyle is extremely important to design specific interventions. Our study investigated, using an ecological approach, barriers and motivations towards endurance running in women affected by cancer and compared them with matched healthy subjects. We found that OG had many motivations originating by personal and interpersonal levels. Furthermore, they interfaced with several obstacles, present into all four levels of the ecological model. Among them, the cancer experi- ence appeared significantly important and influenced both motivators and barriers. Develop- ing a running program that considers all these aspects, may increase its success in terms of both adherence and compliance in this kind of patients (Fig 1). Acknowledgments We thank all the participants that took place in this study. Author Contributions Conceptualization: Alice Avancini, Massimo Lanza. Data curation: Alice Avancini, Daniela Tregnago, Paolo Frada. Formal analysis: Alice Avancini, Kristina Skroce, Paolo Frada. Funding acquisition: Massimo Lanza. Investigation: Alice Avancini, Kristina Skroce. Methodology: Alice Avancini, Daniela Tregnago, Sara Pilotto, Massimo Lanza. Project administration: Cantor Tarperi, Federico Schena, Michele Milella, Massimo Lanza. Resources: Alice Avancini, Kristina Skroce, Daniela Tregnago, Ilaria Trestini, Clelia Bonaiuto, Cantor Tarperi, Federico Schena, Sara Pilotto. Software: Alice Avancini, Paolo Frada. Supervision: Federico Schena, Michele Milella, Massimo Lanza. Validation: Massimo Lanza. Visualization: Alice Avancini, Kristina Skroce, Daniela Tregnago, Ilaria Trestini, Maria Ceci- lia Cercato, Clelia Bonaiuto, Federico Schena, Michele Milella, Sara Pilotto, Massimo Lanza. Writing – review & editing: Daniela Tregnago. Writing – original draft: Alice Avancini, Kristina Skroce, Sara Pilotto. PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 11 / 13 Writing – review & editing: Alice Avancini, Kristina Skroce, Paolo Frada, Ilaria Trestini, Maria Cecilia Cercato, Clelia Bonaiuto, Cantor Tarperi, Federico Schena, Michele Milella, Sara Pilotto, Massimo Lanza. References 1. Aiom A. I numeri nel cancro in Italia2018. 2. Campbell KL, Winters-Stone KM, Wiskemann J, May AM, Schwartz AL, Courneya KS, et al. Exercise Guidelines for Cancer Survivors: Consensus Statement from International Multidisciplinary Roundtable. Med Sci Sports Exerc. 2019; 51(11):2375–90. https://doi.org/10.1249/MSS.0000000000002116 PMID: 31626055 3. Cormie P, Zopf EM, Zhang X, Schmitz KH. The Impact of Exercise on Cancer Mortality, Recurrence, and Treatment-Related Adverse Effects. Epidemiol Rev. 2017; 39(1):71–92. https://doi.org/10.1093/ epirev/mxx007 PMID: 28453622 4. Rogers LQ, Courneya KS, Anton PM, Verhulst S, Vicari SK, Robbs RS, et al. Effects of a multicompo- nent physical activity behavior change intervention on fatigue, anxiety, and depressive symptomatology in breast cancer survivors: randomized trial. Psychooncology. 2017; 26(11):1901–6. https://doi.org/10. 1002/pon.4254 PMID: 27530961 5. Chen HM, Tsai CM, Wu YC, Lin KC, Lin CC. Effect of walking on circadian rhythms and sleep quality of patients with lung cancer: a randomised controlled trial. Br J Cancer. 2016; 115(11):1304–12. https:// doi.org/10.1038/bjc.2016.356 PMID: 27811855 6. Stene GB, Helbostad JL, Balstad TR, Riphagen II, Kaasa S, Oldervoll LM. Effect of physical exercise on muscle mass and strength in cancer patients during treatment—a systematic review. Crit Rev Oncol Hematol. 2013; 88(3):573–93. https://doi.org/10.1016/j.critrevonc.2013.07.001 PMID: 23932804 7. Scott JM, Zabor EC, Schwitzer E, Koelwyn GJ, Adams SC, Nilsen TS, et al. Efficacy of Exercise Ther- apy on Cardiorespiratory Fitness in Patients With Cancer: A Systematic Review and Meta-Analysis. J Clin Oncol. 2018; 36(22):2297–305. https://doi.org/10.1200/JCO.2017.77.5809 PMID: 29894274 8. Lee DC, Brellenthin AG, Thompson PD, Sui X, Lee IM, Lavie CJ. Running as a Key Lifestyle Medicine for Longevity. Prog Cardiovasc Dis. 2017; 60(1):45–55. https://doi.org/10.1016/j.pcad.2017.03.005 PMID: 28365296 9. Zach S, Xia Y, Zeev A, Arnon M, Choresh N, Tenenbaum G. Motivation dimensions for running a mara- thon: A new model emerging from the Motivation of Marathon Scale (MOMS). J Sport Health Sci. 2017; 6(3):302–10. https://doi.org/10.1016/j.jshs.2015.10.003 PMID: 30356611 10. van Dyck D, Cardon G, de Bourdeaudhuij I, de Ridder L, Willem A. Who Participates in Running Events? Socio-Demographic Characteristics, Psychosocial Factors and Barriers as Correlates of Non- Participation-A Pilot Study in Belgium. Int J Environ Res Public Health. 2017; 14(11). 11. Henriksson A, Arving C, Johansson B, Igelstro¨m H, Nordin K. Perceived barriers to and facilitators of being physically active during adjuvant cancer treatment. Patient Educ Couns. 2016; 99(7):1220–6. https://doi.org/10.1016/j.pec.2016.01.019 PMID: 26860549 12. McLeroy KR, Bibeau D, Steckler A, Glanz K. An ecological perspective on health promotion programs. Health Educ Q. 1988; 15(4):351–77. https://doi.org/10.1177/109019818801500401 PMID: 3068205 13. King KM, Gonzalez GB. Increasing Physical Activity Using An Ecological Model. ACSM’s Health & Fit- ness Journal. 2018; 22(4):29–32. 14. O’Brien BC, Harris IB, Beckman TJ, Reed DA, Cook DA. Standards for reporting qualitative research: a synthesis of recommendations. Acad Med. 2014; 89(9):1245–51. https://doi.org/10.1097/ACM. 0000000000000388 PMID: 24979285 15. Tong A, Sainsbury P, Craig J. Consolidated criteria for reporting qualitative research (COREQ): a 32- item checklist for interviews and focus groups. Int J Qual Health Care. 2007; 19(6):349–57. https://doi. org/10.1093/intqhc/mzm042 PMID: 17872937 16. Danese E, Salvagno GL, Tarperi C, Negrini D, Montagnana M, Festa L, et al. Middle-distance running acutely influences the concentration and composition of serum bile acids: Potential implications for can- cer risk? Oncotarget. 2017; 8(32):52775–82. https://doi.org/10.18632/oncotarget.17188 PMID: 28881769 17. Lippi G, Schena F. Run for Science (R4S): the history of a successful project of precision and laboratory medicine in sport and exercise. Journal of Laboratory and Precision Medicine. 2017; 2(4). 18. McIntosh M, Opozda M, Galvao DA, Chambers SK, Short CE. Identifying the exercise-based support needs and exercise programme preferences among men with prostate cancer during active surveil- lance: A qualitative study. Eur J Oncol Nurs. 2019; 41:135–42. https://doi.org/10.1016/j.ejon.2019.06. 006 PMID: 31358246 PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 12 / 13 19. Henriksson A, Arving C, Johansson B, Igelstrom H, Nordin K. Perceived barriers to and facilitators of being physically active during adjuvant cancer treatment. Patient Educ Couns. 2016; 99(7):1220–6. https://doi.org/10.1016/j.pec.2016.01.019 PMID: 26860549 20. Erlingsson C, Brysiewicz P. A hands-on guide to doing content analysis. Afr J Emerg Med. 2017; 7 (3):93–9. https://doi.org/10.1016/j.afjem.2017.08.001 PMID: 30456117 21. Hardcastle SJ, Maxwell-Smith C, Zeps N, Platell C, O’Connor M, Hagger MS. A qualitative study explor- ing health perceptions and factors influencing participation in health behaviors in colorectal cancer sur- vivors. Psychooncology. 2017; 26(2):199–205. https://doi.org/10.1002/pon.4111 PMID: 26935994 22. Husebo AM, Karlsen B, Allan H, Soreide JA, Bru E. Factors perceived to influence exercise adherence in women with breast cancer participating in an exercise programme during adjuvant chemotherapy: a focus group study. J Clin Nurs. 2015; 24(3–4):500–10. https://doi.org/10.1111/jocn.12633 PMID: 24890796 23. Blaney J, Lowe-Strong A, Rankin J, Campbell A, Allen J, Gracey J. The cancer rehabilitation journey: barriers to and facilitators of exercise among patients with cancer-related fatigue. Phys Ther. 2010; 90 (8):1135–47. https://doi.org/10.2522/ptj.20090278 PMID: 20558566 24. Frensham LJ, Parfitt G, Stanley R, Dollman J. Perceived Facilitators and Barriers in Response to a Walking Intervention in Rural Cancer Survivors: A Qualitative Exploration. Int J Environ Res Public Health. 2018; 15(12). 25. Singh B, Spence RR, Steele ML, Sandler CX, Peake JM, Hayes SC. A Systematic Review and Meta- Analysis of the Safety, Feasibility, and Effect of Exercise in Women With Stage II+ Breast Cancer. Arch Phys Med Rehabil. 2018; 99(12):2621–36. https://doi.org/10.1016/j.apmr.2018.03.026 PMID: 29730319 26. Hardcastle SJ, Maxwell-Smith C, Kamarova S, Lamb S, Millar L, Cohen PA. Factors influencing non- participation in an exercise program and attitudes towards physical activity amongst cancer survivors. Support Care Cancer. 2018; 26(4):1289–95. https://doi.org/10.1007/s00520-017-3952-9 PMID: 29090387 27. Kostopoulou V, Katsouyanni K. The truth-telling issue and changes in lifestyle in patients with cancer. J Med Ethics. 2006; 32(12):693–7. https://doi.org/10.1136/jme.2005.015487 PMID: 17145907 28. Gavazzi C, Sieri S, Traclò F, Sproviero A, Vandoni G, Ricci R, et al. Changes in food habits in cancer patients in Italy: a survey. AIOM—SINPE—FAVO. Nutrition. 2018; 55–56:140– 5. https://doi.org/10. 1016/j.nut.2018.04.002 PMID: 30005330 29. Rogers LQ, Courneya KS, Anton PM, Hopkins-Price P, Verhulst S, Vicari SK, et al. Effects of the BEAT Cancer physical activity behavior change intervention on physical activity, aerobic fitness, and quality of life in breast cancer survivors: a multicenter randomized controlled trial. Breast Cancer Res Treat. 2015; 149(1):109–19. https://doi.org/10.1007/s10549-014-3216-z PMID: 25417174 30. Casla S, Lo´pez-Tarruella S, Jerez Y, Marquez-Rodas I, Galvão DA, Newton RU, et al. Supervised phys- ical exercise improves VO2max, quality of life, and health in early stage breast cancer patients: a ran- domized controlled trial. Breast Cancer Res Treat. 2015; 153(2):371–82. https://doi.org/10.1007/ s10549-015-3541-x PMID: 26293147 PLOS ONE Barriers and motivations in running for female oncological patients PLOS ONE | https://doi.org/10.1371/journal.pone.0227846 April 2, 2020 13 / 13

"Running with cancer": A qualitative study to evaluate barriers and motivations in running for female oncological patients.

04-02-2020

Avancini, Alice,Skroce, Kristina,Tregnago, Daniela,Frada, Paolo,Trestini, Ilaria,Cercato, Maria Cecilia,Bonaiuto, Clelia,Tarperi, Cantor,Schena, Federico,Milella, Michele,Pilotto, Sara,Lanza, Massimo

eng

PMC10250310

1 Vol.:(0123456789) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports Participation and performance trends in short‑, medium, and long‑distance duathlon Jonas Turnwald 1, Caio Victor Sousa 2, Marilia Santos Andrade 3, Mabliny Thuany 4, Ivan Cuk 5, Pantelis Theodoros Nikolaidis 6, Katja Weiss 1 & Beat Knechtle 1,7* Participation and performance trends of male and female athletes have been thoroughly analyzed in various endurance sports. Knowing these trends can help coaches and athletes prepare for competitions and may influence their training strategy and career planning. However, duathlon events—consisted of two splits of running (Run1 and Run2) interspersed by a split of cycling (Bike)— have not been thoroughly studied, unlike other endurance sports. The present study aimed to compare participation and performance trends in duathletes who competed in duathlon races hosted by World Triathlon or affiliated National Federations between 1990 and 2021. A total of 25,130 results of age group finishers who competed in run‑bike‑run duathlon races of varying distances were analyzed with different general linear models. Races were divided into three distances: short‑ distance (up to 5.5 km run, 21 km bike, 5 km run), medium‑distance (5–10 km run, 30–42 km bike, 7–11 km run) and long‑distance (at least 14 km run, 60 km bike, 25 km run). On average, women represented 45.6% of all finishers in short‑distance, 39.6% in medium‑distance and 24.9% in long‑ distance duathlon races. Throughout the years, men were consistently faster than women in all three race legs (Run 1, Bike, and Run 2) in all three distances across all age groups, and women could not reduce the performance gap. Concerning the age of peak performance, duathletes of the age group 30–34 finished most often in the top three in short‑ and medium‑distance duathlons, whereas male duathletes of the age group 25–29 and female duathletes of the age group 30–34 finished most often in the top three in long‑distance duathlons. Women participated less, especially in longer distances, and were constantly slower than men. Duathletes of the age group 30–34 finished most often in the top three. Future studies should analyze participation and performance trends in further subgroups (e.g., elite athletes) and pacing behaviours. Abbreviations APP Age of peak performance GLM General linear models ITU International Triathlon Union WT World Triathlon Non-professional endurance sports have been consistently growing in popularity during the last decades. Accord- ingly, the scientific community has studied participation and performance trends in various sports such as triathlon1,2, distance running3–5, cycling6,7 and duathlon8–10. Duathlon is a unique multi-discipline sport in which athletes compete in a run-bike-run format. It is internationally governed by Word Triathlon (WT), formerly the International Triathlon Union (ITU). WT distances include a sprint-distance (5 km run, 20 km bike, 2.5 km run), standard-distance (5–10 km run, 30–40 km bike, 5 km run), middle-distance (10–20 km run, 60–90 km bike, 10 km run) and long-distance (10–20 km run, 120–150 km bike, 20–30 km run), but individual race distances can vary11. Participation and performance trends in duathlons have been investigated before. Nonetheless, to the best of our knowledge, the current literature is either based on a specific race8,10,12 or a specific distance9. OPEN 1Institute of Primary Care, University of Zurich, Zurich, Switzerland. 2Health and Human Sciences, Loyola Marymount University, Los Angeles, USA. 3Department of Physiology, University of Sao Paulo, Sao Paulo, Brazil. 4Faculty of Sports, University of Porto, Porto, Portugal. 5Faculty of Sport and Physical Education, University of Belgrade, Belgrade, Serbia. 6School of Health and Caring Sciences, University of West Attica, Athens, Greece. 7Medbase St. Gallen Am Vadianplatz, Vadianstrasse 26, 9001 St. Gallen, Switzerland. *email: beat.knechtle@hispeed.ch 2 Vol:.(1234567890) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ The performance gap between sexes is one of the main points of interest in endurance sports research. Con- sistent with studies on other endurance sports1,3–6,13, slower race times of women were observed in previously studied duathlon events8–10. Recently, Romero-Ramos et al.9 analyzed performance differences, with regard to age and sex, of the top ten age group athletes competing in the ITU Duathlon World Championships from 2005 to 2016 on the standard-distance length. Men outperformed women in all age groups in all race legs and with advancing age, the differences between both sexes increased9. Previous investigations focused on the Powerman Zofingen with its two distances (short-distance: ~ 10 km run, 50 km bike, 5 km run; long-distance: ~ 10 km run, 150 km bike, 30 km run)8,10. A consistent sex difference of ~ 18–19% in all race legs and total times was observed in the annual top ten elite athletes who participated in the long-distance races between 2002 and 201110. When all finishers of the short- and long-distance races from 2003 to 2017 were analyzed, the sex difference was similar in both versions (~ 8%)8. These differences in race times between females and males might be explained by physi- ological, anthropometric, genetic, hormonal and psychological factors13–15. However, the sex gap seems to be dependent on the race distance. In other endurance sports, such as distance running, the sex gap has been shown to decrease in ultra-endurance-distances15–17. In a study by Waldvogel et al.16, a higher sex gap was observed in 50 mile (9.13%) than in 100 mile (4.41%) ultra-marathon races. Age is another important aspect affecting performance in endurance sports. With advancing age, cellular deterioration and loss of tissue function occur, affecting physical performance in different manners based on the specific requirements of the activity18–20. Compared to sprint races, the age of peak performance (APP) seems to be higher in endurance races2,20. This relationship is also reflected in endurance events of different distances. For instance, Nikolaidis et al.12 investigated the APP in the short- and long-distance races of the Powerman Zofingen and reported that the fastest age group was younger in the short-distance race (age group 20–24) than in the long- distance race (age group 25–29). Conversely, Romero-Ramos et al.9 reported a higher APP (age group 30–34) in both genders when the overall performance of the top ten athletes of each age group at the ITU Duathlon World Championships on the standard-distance was compared. As the race distances (~ 10 km run, 40 km bike, 5 km run) were shorter compared to the short version of the Powerman Zofingen (~ 10 km run, 50 km bike, 5 km run), the difference in the observed APP might be explained by the different study designs and the specific characteristics of the Powerman Zofingen. This highlights the importance of a more extensive dataset for a better understanding of the trends in duathlon9,12. Up to now, no study regarding the APP in the sprint-distance exists. The two disciplines, running and cycling, represent different types of locomotion with their own anthro- pometric and physiological correlates21,22. When the age-related performance decline of each discipline was analyzed separately, the cycling performance could be better maintained than the running performance in older athletes8,9,23. This phenomenon was also observed in triathlon events, where the performance decline with increasing age was more prominent in swimming and running than in cycling1. Little is known so far concerning participation trends in duathlon. When investigating finishers of the Power- man Zofingen from 2003 to 2017, 15.2% of all finishers in the long-distance and 15.9% of all finishers in the short- distance were women8. In the shorter ITU Duathlon World Championships (standard-distance), higher participa- tion of women was observed from 2005 to 2016. Romero-Ramos et al.9 reported that 23.5% of all finishers were women. More studies have been conducted on triathlon races, with an increase in female participants observed since the 1980s1. Also, in triathlon, it seems that women tend to compete in shorter than longer distances1,24–26. Although the above-mentioned literature provides some information about participation and performance trends in duathlon, no study has investigated the worldwide trends across different race distances so far. Knowl- edge of these trends would not only be interesting for scientists but could also help athletes and coaches prepare for races and could influence their training strategy depending on the sex and age of an athlete and the specific distance of a race. Furthermore, duathletes who are aware of different APPs in different race distances would be able to plan their career more precisely. Therefore, the present study aimed to investigate the worldwide participation and performance trends in duathlon with an extensive dataset, including results from finishers who participated in duathlon races worldwide across different distances over several decades. Based upon the previously mentioned findings, we hypothesized firstly that more male than female finishers would be recorded for all distances and especially for longer distances, secondly, that men would be faster than women, thirdly that the sex gap would narrow throughout the years and fourthly, that the APP is higher in longer race distances. Methods Ethical approval and consent to participate. This study was approved by the Institutional Review Board of Kanton St. Gallen, Switzerland, with a waiver of the requirement for informed consent of the partici- pants as the study involved the analysis of publicly available data (EKSG 01/06/2010). The study was conducted in accordance with recognized ethical standards according to the Declaration of Helsinki adopted in 1964 and revised in 2013. Duathlon events. Results of international events hosted by WT or affiliated National Federations were obtained from the results section of WT’s official website27. To ensure comparability, we only included regular international duathlon races that were either World Championship or Continental Championship races, and excluded Cross- or Winter-Duathlons. A total of 187 races were identified, which have taken place from 1990 to 2021. However, distances were not stated on the downloadable result lists. Therefore, information about the race distances had to be retrieved in multiple ways. The race distances were listed in the “Program notes” section for some participant groups. If this was not available, we searched the event page of a specific race with the three tabs “Event Info”, “Local Info” and “Contact” for any information. If no distance was available, we scanned the event page for an external link to the official event website of the race. If available, we thoroughly browsed this 3 Vol.:(0123456789) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ website to find the relevant data. An external link was found for some races, but the website was no longer active or contained non-corresponding content. In this case, we searched for archived versions of the event website on the Wayback Machine of the Internet Archive to get any information regarding the distances28. Nonetheless, no distances could be retrieved for some events/participant groups. We only included data of participant groups, if a clear association of a respective participant group with a specific distance was present. As individual race dis- tances differed and did not always match a specific WT distance, we divided the races into three distances: short- distance (up to 5.5 km run, 21 km bike, 5 km run), medium-distance (5–10 km run, 30–42 km bike, 7–11 km run) and long-distance (at least 14 km run, 60 km bike, 25 km run). For the purpose of this study, we only included successful finishers of adult age group categories who com- peted in a duathlon race in a run-bike-run race mode. Except for the age group 18–19 years, each age group covers a five-year period (20–24 years through to 85–89 years). Required data from the race results included the name of an athlete, the sex, the split times, the total time and the age group. The year and name of an event, obtained from the corresponding event page on WT’s website, and the distance were added. No races from the years 1990, 1992, 1996, 1997, and 2004 could be included. In short-distance duathlon, the first race that could be included was in 2011, in medium-distance duathlon in 1991, and in long-distance duathlon in 2002. Excluded were results from the 1997 Guernica ITU Duathlon World Championships, as the split times did not match the overall times in most cases, and the 2003 Affoltern ITU Duathlon World Championships, as the stated distance appeared to be wrong. Moreover, finishers with empty race times and statistical outliers in any of the race legs (slower/faster by three standard deviations from the mean) were excluded. In total, 66 races met the inclusion criteria. Statistical analysis. Descriptive statistics were presented using mean ± standard deviation and frequencies. All data showed parametric distribution and homogeneity of variance through the Kolmogorov–Smirnov’s and Levene’s tests, respectively. Average speed (kilometers per hour (km/h)) was established as the dependent vari- able for all models. General linear models (GLM) with two factors (two-way ANOVA) were applied for each dis- tance (short, medium, and long) considering the independent factors “sex × age group” and “sex × calendar year”. Further GLM were conducted for men and women separately with “event distance × age group” as independent factors. Fisher’s least significant difference was applied as a post-hoc test to identify specific differences between independent factors. Partial eta square (ηp 2) was applied as a measure of effect size, considering ηp 2 = 0.01 as a small effect, ηp 2 = 0.06 as a moderate effect, and ηp 2 = 0.14 as a large effect. Statistical significance was defined as p < 0.05. All statistical analyses were carried out with Statistical Software for the Social Sciences (IBM® SPSS v.25, Chicago, Ill, USA). Results A total of 25,130 finishers were included. Short-distance duathlon included 4641 men and 2118 women (n = 6759), medium-distance duathlon included 9970 men and 3921 women (n = 13,891), and long-distance duathlon included 3587 men and 893 women (n = 4480). Women’s participation in individual races ranged from 18.5 to 55.9% in relation to men in short-distance duathlon (average: 45.6%), 13.3–51.7% in medium-distance duathlon (average: 39.6%) and 4.2–37.3% in long-distance duathlon (average: 24.9%). See Fig. 1 for detailed participation by sex and year. Performance trends across age groups showed significant effects of both sex and age group for short-, medium-, and long-distance duathlon across all three race legs of the duathlon race (Run 1, Bike, Run 2). See Table 1 for details. Pairwise comparisons showed that men had better performances than women across all age groups in all three race legs (Run 1, Bike, and Run 2) in all three duathlon distances. Finally, age group pairwise comparisons showed that, in short-distance duathlon, the age group was always significantly different from the previous one, but in medium-distance duathlon, the run performance started to drop at the age group 45–49 years in men and age group 50–54 years in women, whereas the bike performance started to drop at the age group 50–54 years in men and age group 55–59 years in women. In long-distance duathlon, the first running leg was stable until the age group 30–34 years in men and age group 50–54 years in women, whereas the bike performance and the second running leg were stable until the age group 50–54 years in both men and women. See Fig. 2. Performance trends across calendar years showed significant effects of both sex and calendar years for short-, medium-, and long-distance duathlon across all three race legs of the duathlon race (Run 1, Bike, Run 2). See Table 2 for details. Pairwise comparisons showed that men were consistently faster than women in all race legs and distances across all calendar years and no trend was observed that women reduced the sex gap throughout the years. Additionally, no apparent performance trend was seen in any distance throughout the years. See Fig. 3. The GLM for men showed significant “event” and “age group” effects for Run 1 (event: F = 39.5, p < 0.001, ηp 2 = 0.52; age group: F = 76.9, p < 0.001, ηp 2 = 0.98; interaction: F = 9.5, p < 0.001, ηp 2 = 0.01), Bike (event: F = 24.5, p < 0.001, ηp 2 = 0.14; age group: F = 82.9, p < 0.001, ηp 2 = 0.96; interaction: F = 2.6, p < 0.001, ηp 2 < 0.01) and Run 2 (event: F = 64.2, p < 0.001, ηp 2 = 0.54; age group: F = 75.2, p < 0.001, ηp 2 = 0.97; interaction: F = 6.0, p < 0.001, ηp 2 = 0.01). Similar results were found in women for Run 1 (event: F = 22.7, p < 0.001, ηp 2 = 0.18; age group: F = 65.7, p < 0.001, ηp 2 = 0.96; interaction: F = 2.0, p = 0.004, ηp 2 = 0.01), Bike (event: F = 13.8, p < 0.001, ηp 2 = 0.11; age group: F = 21.1, p < 0.001, ηp 2 = 0.88; interaction: F = 1.8, p = 0.016, ηp 2 = 0.01) and Run 2 (event: F = 38.9, p < 0.001, ηp 2 = 0.32; age group: F = 48.7, p < 0.001, ηp 2 = 0.95; interaction: F = 2.1, p = 0.002, ηp 2 = 0.01). See Fig. 4 for details. Pairwise comparisons for the men models showed that the average speeds in the three race distances differed from each other up to the age group 65–69 years in the running legs and age group 70–74 years in the cycling leg. 4 Vol:.(1234567890) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ For the women models, a slower average speed was observed in long-distance races in comparison to the other distances up to the age group 55–59 years in the running legs and age group 60–64 years in the cycling leg (Fig. 4). In the analyzed short- and medium-distance races, male and female athletes of the age group 30–34 years finished most often in the top three compared to other age groups. In long-distance races, men of the age group 25–29 years and women of the age group 30–34 years finished most often in the top three. Overall, when all distances were considered, the age group 30–34 years was the most prevalent one in the top three in men and women. See Fig. 5 for details. Discussion This study intended to investigate the worldwide participation and performance trends of short-, medium- and long-distance duathlon over several decades. The participation in the investigated races did not increase over the years. This finding might not be generalized to the sport itself, as we considered only events hosted by WT or affiliated National Federations which were listed on WT’s website and did not compare the same races every year. Furthermore, the present study analyzed world and continental championships, but not local events, which may be preferred by age group athletes. Non-elite athletes often face real-world commitments and financial Figure 1. Participation of men and women in short- (A), medium- (B), and long-distance (C) duathlon across calendar years. 5 Vol.:(0123456789) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ Table 1. General linear model results with average speed as the dependent variable. Duathlon Sex Age group Sex × Age group F p ηp 2 F p ηp 2 F p ηp 2 Short Run 1 180.6 < 0.001 0.66 58.2 < 0.001 0.99 5.0 < 0.001 0.01 Bike 147.2 < 0.001 0.20 68.2 < 0.001 0.99 1.5 0.14 < 0.01 Run 2 135.4 < 0.001 0.47 58.3 < 0.001 0.99 3.5 < 0.001 0.01 Medium Run 1 231.0 < 0.001 0.38 108.7 < 0.001 0.99 4.3 < 0.001 0.01 Bike 104.7 < 0.001 0.04 71.0 < 0.001 0.99 1.4 0.16 < 0.01 Run 2 137.3 < 0.001 0.10 156.4 < 0.001 0.99 2.0 0.02 < 0.01 Long Run 1 119.8 < 0.001 0.30 31.7 < 0.001 0.97 1.4 0.16 < 0.01 Bike 119.9 < 0.001 0.29 19.4 < 0.001 0.95 1.4 0.19 < 0.01 Run 2 41.6 < 0.001 0.10 23.0 < 0.001 0.95 1.15 0.32 < 0.01 Figure 2. Average speed in the three duathlon race legs of men and women in short-, medium-, and long- distance duathlon across age groups. * over line: statistical significance between all age groups; *: statistical significance in comparison to the previous age group; # over line: statistical significance for sex across all age groups. 6 Vol:.(1234567890) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ Table 2. General linear model results with average speed as the dependent variable. Duathlon Sex Calendar year Sex × Calendar year F p ηp 2 F p ηp 2 F p ηp 2 Short Run 1 275.2 < 0.001 0.94 8.6 0.001 0.90 3.7 < 0.001 0.01 Bike 292.7 < 0.001 0.94 43.7 < 0.001 0.98 3.9 < 0.001 0.01 Run 2 230.5 < 0.001 0.93 10.4 < 0.001 0.91 3.3 < 0.001 0.01 Medium Run 1 303.8 < 0.001 0.32 31.6 < 0.001 0.97 2.0 < 0.001 < 0.01 Bike 245.1 < 0.001 0.31 98.0 < 0.001 0.99 2.2 0.001 < 0.01 Run 2 177.7 < 0.001 0.24 25.0 < 0.001 0.96 2.1 0.001 < 0.01 Long Run 1 870.6 < 0.001 0.96 70.9 < 0.001 0.98 0.56 0.91 < 0.01 Bike 714.4 < 0.001 0.96 65.2 < 0.001 0.99 0.78 0.19 0.99 Run 2 226.9 < 0.001 0.88 27.6 < 0.001 0.97 0.75 0.77 < 0.01 Figure 3. Average speed in the three race legs of men and women in short-, medium-, and long-distance duathlon across calendar years. # over line: statistical significance for sex across all calendar years. 7 Vol.:(0123456789) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ constraints that can make it difficult for them to travel to events and compete29. In other endurance sports, such as long-distance running and cycling, an increased participation rate was observed in recent years30–32. In addi- tion, the number of members in national federations has grown in the last decades. In Germany, for example, the number of members of the National Triathlon Federation has more than doubled between 2001 and 202233,34. An important finding was the lower number of female duathletes compared to male duathletes in all dis- tances across all years. Women accounted on average for 45.6% of finishers in short-distance duathlon, 39.4% in medium-distance duathlon and 24.9% in long-distance duathlon. The lower rate of women finishers in longer race distances is in accordance with previous findings in triathlon and might be explained by motivational reasons, differences in training behaviour and sociocultural conditions1,35,36. Regarding performance, men were faster than women in all race legs in all distances across all age groups and calendar years. There is extensive literature on factors that explain the differences in performance between Figure 4. Average speed in the three race legs of men and women separately in short-, medium-, and long- distance duathlon across age groups. ** over line: statistical significance between all three events; *: statistically significant from the other two events. 8 Vol:.(1234567890) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ men and women in endurance sports13–15,36. In addition to physiological differences, such as the lower maximal oxygen uptake (VO2max) in female athletes, morphological differences, social factors, psychological factors and differences in training characteristics have to be considered concerning the sex gap13,37. Female athletes were able to reduce the sex gap in ultra-endurance sports throughout the years, for example, in most ultra-marathon distances38 and ultra-cycling distances7. However, our hypothesis was not confirmed, as we did not find such a trend in any of the investigated distances throughout the years. Based on past findings, the analyzed distances in this study were not long enough to see such a trend, as the physiological and morphological advantages of women (e.g., better fatigue resistance, greater substrate efficiency and lesser energetic demands) rather seem to play a role in ultra- and extreme distances15. Moreover, no apparent performance trend could be observed in any of the investigated distances throughout the years. While Nikolaidis et al.8 also found an unchanged performance of male and female finishers in the “Powerman Zofingen” from 2003 to 2017, Gallman et al.39 found an increased performance of the annual top ten male and female triathletes who competed at the Ironman Hawaii from 1983 to 2012. Methodological dif- ferences, including sample size, time frame and statistical procedures, may be related to the differences in these findings. For example, Gallman et al.39 analyzed the results of the top ten elite athletes, whereas we analyzed all successful finishers of adult age group categories. In many studies on performance trends in marathon races, a phenomenon was observed that “the faster get faster and the slower get slower”40–43. It is important to note that we analyzed data from multiple races, with differences in drafting rules, weather conditions, and track specifica- tions, and not one specific race over a period of time, what may have impacted the results. Another finding was that in short-distance duathlon, a statistically significant decline in performance with increasing age groups could be observed from the first age group (20–24 years), whereas the performance in medium- and long-distance duathlon was relatively stable up to a specific age group. In the analyzed medium- distance races, a statistically significant drop in performance in the first and second run was for the first time observed at the age group 45–49 years in men and 50–54 years in women, while the cycling performance dropped later at the age group 50–54 years in men and 55–59 years in women. Previous studies on multi- discipline sports already showed that the age-related performance decline seems to be higher in running than Figure 5. The number of athletes in each age group who finished in the analyzed races in the top three by sex and distance. 9 Vol.:(0123456789) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ in cycling1,19,44,45. This might be related to the distinct characteristics of the two disciplines. Running is a weight- bearing stretch–shortening activity with a predominantly eccentric type of muscle action compared to cycling, which is a non-weight-bearing activity with concentric contractions46,47. However, it is noteworthy that in a study performed by Swinnen et al.48 running-specific training could improve running economy while the cycling economy could not be improved by cycling-specific training. Regarding long-distance duathlon, a statistically significant drop in performance in the first run was for the first time observed at the age group 30–34 years in men and 50–54 years in women, while in the cycling and second running leg performance dropped at the age group 50–54 years in men and women. A reason that the performance in medium- and long-distance duathlon was relatively stable up to a specific age group in contrast to short-distance duathlon might be that generally, more experienced athletes compete in longer race distances and it was postulated before that the amount of experience is highly important for the performance in multi-discipline sports49. When we analyzed the average speeds of men and women separately in the three race distances across the different age groups, particular differences could be observed. In men, a statistically significant difference between the average speeds of the three race distances could be observed up to the age group 65–69 years in both running legs and 70–74 years in the cycling leg. In women, on the other hand, only the average speed in long-distance duathlon was statistically significantly slower compared to the other two distances up to the age group 55–59 years in both running legs and 60–64 years in the cycling leg. Interestingly, no statistically significant difference could be observed between the average speeds of women in short- and medium-distance races. Although this phenomenon does not make physiological sense, it indicates that women have a lot of room for performance improvement in short-distance duathlon. In many endurance sports, it was previously shown that women adopted a more conservative pacing strategy than men50–53. This might be explained by dif- ferences regarding confidence, decision-making, risk perception and willingness53. For example, compared to men, women showed relatively lower speeds in the beginning and relatively higher speeds at the end of a 100 km ultra-marathon race54. One explanation could be that women did not allocate their energy resources in the best suitable manner. More studies are necessary to confirm or refute these results. The only knowledge we have so far regarding pacing in a duathlon is derived from studies by Nikolaidis et al., who analyzed the effect of aging23, sex and performance level50 as well as the combined effect of aging and performance level55 on pacing. However, these studies are based solely on the “Powerman Zofingen” results from 2003 to 2017 with its two distances (10 km run, 50 km bike, 5 km run; ~ 10 km run, 150 km bike, 30 km run). The authors reported that women adopted a steadier pace and were relatively faster in the second run50. To the best of our knowledge, no information regarding pacing behaviours in short-distance duathlon exists. Regarding the APP, male and female athletes of the age group 30–34 years finished most often in the top three in short- and medium-distance races. This confirms past findings by Romero-Ramos et al.9 who analyzed the performance of the top ten athletes of each age group who competed at the ITU Duathlon World Champion- ships from 2005 to 2016 and found that athletes of the age group 30–34 years performed best in the standard- distance (~ 10 km run, 40 km bike, 5 km run). In long-distance duathlon, men of the age group 25–29 years and women of the age group 30–34 years finished most often in the top three in our study. Therefore, our hypothesis that the APP is higher in longer race distances could not be confirmed. This is in contrast to a study by Kne- chtle et al.2, who analyzed the different APPs of world-class triathletes in different race distances. The authors reported that men achieved the best performance at 27.1 ± 4.9 years in the Olympic distance, 28.0 ± 3.8 years in the Half-Ironman distance and 35.1 ± 3.6 years in the Ironman distance, while women were best at 26.6 ± 4.4, 31.6 ± 3.4 and 34.4 ± 4.4 years respectively. However, besides the differences regarding the modes of locomotion, the methodological approach was different and we were only able to determine the age group of the finishers and not their exact age. Limitations, strengths, and implications for future research. A limitation of this study is the use of secondary data. We were not able to consider important factors related to endurance performance in ath- letes of different competitive levels, such as anthropometric and physiological variables, training status, previous experience, drafting rules, technical equipment, track specifications and weather conditions. Due to the use of secondary data, we cannot exclude that some distances have been rounded. Moreover, data was missing in certain years. Other methodological designs, such as longitudinal studies, could offer more information about the effect of aging on duathlon performance. Nevertheless, this is the first study that investigated worldwide participation and performance trends in duathlon with results from finishers who participated in duathlon races worldwide across three different distances over several decades. Future studies should collect data about the above-mentioned variables and analyze participation and performance trends in further subgroups (e.g., elite athletes) as well as pacing behaviours in short-distance duathlon and the association between place of competi- tion, participation and performance trends. Conclusion More men than women competed in all distances and especially in longer distances. Men were generally faster across all age groups and no trend regarding the sex gap was observed at any distance throughout the years. The APP did not increase with an increase in the race distance. Men and women of the age group 30–34 finished most often in the top three in short- and medium-distance races, whereas in long-distance races, men of the age group 25–29 and women of the age group 30–34 finished most often in the top three. 10 Vol:.(1234567890) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ Data availability For this study, we have included official results and split times from the official website of WT https:// triat hlon. org. The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Received: 10 March 2023; Accepted: 28 May 2023 References 1. Lepers, R., Knechtle, B. & Stapley, P. J. Trends in triathlon performance: Effects of sex and age. Sports Med. 43(9), 851–863. https:// doi. org/ 10. 1007/ s40279- 013- 0067-4 (2013). 2. Knechtle, R., Rüst, C. A., Rosemann, T. & Knechtle, B. The best triathletes are older in longer race distances: A comparison between Olympic, Half-Ironman and Ironman distance triathlon. Springerplus 3, 538 (2014) (PMID:25279329). 3. Hanley, B. Pacing, packing and sex-based differences in Olympic and IAAF World Championship marathons. J. Sports Sci. 34(17), 1675–1681 (2016) (PMID:26736042). 4. Nikolaidis, P. T., Onywera, V. O. & Knechtle, B. Running performance, nationality, sex, and age in the 10-km, Half-Marathon, Marathon, and the 100-km Ultramarathon IAAF 1999–2015. J. Strength Cond. Res. 31(8), 2189–2207 (2017) (PMID:28731980). 5. Senefeld, J., Smith, C. & Hunter, S. K. Sex differences in participation, performance, and age of ultramarathon runners. Int. J. Sports Physiol. Perform. 11(7), 635–642 (2016) (PMID:26561864). 6. Gloor, R. U. et al. Sex-related trends in participation and performance in the “Swiss Bike Masters” from 1994–2012. Percept. Mot. Skills 116(2), 640–654 (2013) (PMID:24032336). 7. Baumgartner, S., Sousa, C. V., Nikolaidis, P. T. & Knechtle, B. Can the performance gap between women and men be reduced in ultra-cycling?. Int. J. Environ. Res. Public Health 17(7), 2521 (2020) (PMID:32272640). 8. Nikolaidis, P. T., Villiger, E. & Knechtle, B. Participation and performance trends in the ITU Duathlon World Championship from 2003 to 2017. J. Strength Cond. Res. 35(4), 1127–1133 (2021) (PMID:30363036). 9. Romero-Ramos, O., Fernández-Rodríguez, E., Mayorga-Vega, D., Merino-Marbán, R. & Podstawski, R. Sex and age-related changes in performance in the Duathlon World Championships. Rev. Bras. Med. Esporte 26(3), 234–238. https:// doi. org/ 10. 1590/ 1517- 86922 02026 03190 540 (2020). 10. Rüst, C. A. et al. Gender difference and age-related changes in performance at the long-distance duathlon. J. Strength Cond. Res. 27(2), 293–301. https:// doi. org/ 10. 1519/ jsc. 0b013 e3182 5420d0 (2013). 11. World Triathlon. World Triathlon competition rules 2020 2019 URL: https:// www. triat hlon. org/ uploa ds/ docs/ World_ Triat hlon_ Sport_ Compe tition_ Rules_ 2020_ 20181 1253. pdf [accessed 2021–11–07]. 12. Nikolaidis, P. T. et al. The age of peak performance in women and men duathletes: The paradigm of short and long versions in “Powerman Zofingen”. Open Access J. Sports Med. 9, 125–130 (2018) (PMID:30140162). 13. Lepers, R. Sex difference in triathlon performance. Front. Physiol. 10, 973 (2019) (PMID:31396109). 14. Smith, F. W. & Smith, P. A. Musculoskeletal differences between males and females. Sports Med. Arthrosc. Rev. 10(1), 98–100. https:// doi. org/ 10. 1097/ 00132 585- 20021 0010- 00014 (2002). 15. Tiller, N. B. et al. Do sex differences in physiology confer a female advantage in ultra-endurance sport?. Sports Med. 51(5), 895–915 (2021) (PMID:33502701). 16. Waldvogel, K. J., Nikolaidis, P. T., Di Gangi, S., Rosemann, T. & Knechtle, B. Women reduce the performance difference to men with increasing age in ultra-marathon running. Int. J. Environ. Res. Public Health 16(13), 2377 (2019) (PMID:31277399). 17. Nikolaidis, P. T., Cuk, I., Clemente-Suárez, V. J., Villiger, E. & Knechtle, B. Number of finishers and performance of age group women and men in long-distance running: Comparison among 10 km, half-marathon and marathon races in Oslo. Res. Sports Med. 29(1), 56–66 (2021) (PMID:32046506). 18. Sousa-Victor, P., García-Prat, L., Serrano, A. L., Perdiguero, E. & Muñoz-Cánoves, P. Muscle stem cell aging: Regulation and rejuvenation. Trends Endocrinol. Metab. 26(6), 287–296 (2015) (PMID:25869211). 19. Lepers, R., Stapley, P. J. & Cattagni, T. Variation of age-related changes in endurance performance between modes of locomotion in men: An analysis of master world records. Int. J. Sports Physiol. Perform. 13(3), 394–397 (2018) (PMID:28714746). 20. Allen, S. V. & Hopkins, W. G. Age of peak competitive performance of elite athletes: A systematic review. Sports Med. 45(10), 1431–1441 (2015) (PMID:26088954). 21. Bentley, D. J., Millet, G. P., Vleck, V. E. & McNaughton, L. R. Specific aspects of contemporary triathlon: Implications for physi- ological analysis and performance. Sports Med. 32(6), 345–359 (2002) (PMID:11980499). 22. Millet, G. P., Vleck, V. E. & Bentley, D. J. Physiological differences between cycling and running: Lessons from triathletes. Sports Med. 39(3), 179–206 (2009) (PMID:19290675). 23. Nikolaidis, P. T., Villiger, E., Victor Sousa, C., Rosemann, T. & Knechtle, B. The effect of aging on pacing strategies in short and long distance duathlon. Exp. Aging Res. 45(3), 223–233 (2019) (PMID:31021693). 24. Rüst, C. A., Knechtle, B., Knechtle, P., Rosemann, T. & Lepers, R. Age of peak performance in elite male and female Ironman triathletes competing in Ironman Switzerland, a qualifier for the Ironman world championship, Ironman Hawaii, from 1995 to 2011. Open Access J. Sports Med. 3, 175–182 (2012) (PMID:24198600). 25. Etter, F. et al. Age and gender interactions in short distance triathlon performance. J. Sports Sci. 31(9), 996–1006. https:// doi. org/ 10. 1080/ 02640 414. 2012. 760747 (2013). 26. Knechtle, B., Rüst, C. A., Rosemann, T. & Lepers, R. Age and gender differences in half-Ironman triathlon performances: The Ironman 70.3 Switzerland from 2007 to 2010. Open Access J. Sports Med. 3, 59–66 (2012) (PMID:24198588). 27. World Triathlon. Results—World Triathlon 2023 URL: https:// triat hlon. org/ resul ts [accessed 2023-02-02]. 28. Internet Archive. Wayback Machine 2023 URL: https:// web. archi ve. org/ [accessed 2023-02-02]. 29. Lamont, M., Kennelly, M. & Wilson, E. Competing priorities as constraints in event travel careers. Tour. Manag. 33(5), 1068–1079. https:// doi. org/ 10. 1016/j. tourm an. 2011. 12. 005 (2012). 30. Scheer, V. Participation trends of ultra endurance events. Sports Med. Arthrosc. Rev. 27(1), 3–7 (2019) (PMID:30601393). 31. Thuany, M. et al. Trends in participation, sex differences and age of peak performance in time-limited ultramarathon events: A secular analysis. Medicina (Kaunas) 58(3), 366 (2022) (PMID:35334541). 32. Shoak, M. A. et al. Participation and performance trends in ultracycling. Open Access J. Sports Med. 4, 41–51 (2013) (PMID:24379708). 33. Deutscher Olympischer Sportbund. Bestandserhebung 2022 URL: https:// cdn. dosb. de/ user_ upload/ www. dosb. de/ uber_ uns/ Besta ndser hebung/ Besta ndser hebung_ 2016. pdf [accessed 2023-14-05]. 34. Deutscher Olympischer Sportbund. Bestandserhebung 2001 URL: https:// cdn. dosb. de/ user_ upload/ www. dosb. de/ uber_ uns/ Besta ndser hebung/ Besta ndser hebung_ 2001. pdf [accessed 2023-14-05] 35. Knechtle, B. et al. Training and racing behavior of recreational runners by race distance-results from the NURMI study (step 1). Front. Physiol. 12, 620404 (2021) (PMID:33613312). 11 Vol.:(0123456789) Scientific Reports | (2023) 13:9303 | https://doi.org/10.1038/s41598-023-36050-2 www.nature.com/scientificreports/ 36. Hallam, L. C. & Amorim, F. T. Expanding the gap: An updated look into sex differences in running performance. Front. Physiol. 12, 804149. https:// doi. org/ 10. 3389/ fphys. 2021. 804149 (2022). 37. Bassett, D. R. & Howley, E. T. Limiting factors for maximum oxygen uptake and determinants of endurance performance. Med. Sci. Sports Exerc. 32(1), 70–84 (2000) (PMID:10647532). 38. Knechtle, B. et al. Do women reduce the gap to men in ultra-marathon running?. Springerplus 5(1), 672 (2016) (PMID:27350909). 39. Gallmann, D., Knechtle, B., Rüst, C. A., Rosemann, T. & Lepers, R. Elite triathletes in ‘Ironman Hawaii’ get older but faster. Age (Dordr) 36(1), 407–416 (2013) (PMID:23591938). 40. Reusser, M. et al. Increased participation and decreased performance in recreational master athletes in “Berlin Marathon” 1974– 2019. Front. Physiol. 12, 631237 (2021) (PMID:34262467). 41. Maffetone, P. B., Malcata, R., Rivera, I. & Laursen, P. B. The Boston marathon versus the world marathon majors. PLoS ONE 12(9), e0184024 (2017) (PMID:28863152). 42. Knechtle, B., Di Gangi, S., Rüst, C. A. & Nikolaidis, P. T. Performance differences between the sexes in the Boston marathon from 1972 to 2017. J. Strength Cond. Res. 34(2), 566–576 (2020) (PMID:30664107). 43. Vitti, A., Nikolaidis, P. T., Villiger, E., Onywera, V. & Knechtle, B. The, “New York City Marathon”: Participation and performance trends of 1.2M runners during half-century. Res. Sports Med. 28(1), 121–137 (2020) (PMID:30889965). 44. Lepers, R. & Stapley, P. J. Age-related changes in conventional road versus off-road triathlon performance. Eur. J. Appl. Physiol. 111(8), 1687–1694. https:// doi. org/ 10. 1007/ s00421- 010- 1805-z (2011). 45. Bernard, T., Sultana, F., Lepers, R., Hausswirth, C. & Brisswalter, J. Age-related decline in Olympic triathlon performance: Effect of locomotion mode. Exp. Aging Res. 36(1), 64–78. https:// doi. org/ 10. 1080/ 03610 73090 34186 20 (2009). 46. Bijker, K. E., de Groot, G. & Hollander, A. P. Differences in leg muscle activity during running and cycling in humans. Eur. J. Appl. Physiol. 87(6), 556–561 (2002) (PMID:12355196). 47. Heiden, T. & Burnett, A. The effect of cycling on muscle activation in the running leg of an Olympic distance triathlon. Sports Biomech. 2(1), 35–49 (2003) (PMID:14658244). 48. Swinnen, W., Kipp, S. & Kram, R. Comparison of running and cycling economy in runners, cyclists, and triathletes. Eur. J. Appl. Physiol. 118(7), 1331–1338 (2018) (PMID:29663075). 49. Knechtle, B., Wirth, A. & Rosemann, T. Predictors of race time in male Ironman triathletes: Physical characteristics, training, or prerace experience?. Percept. Mot. Skills 111(2), 437–446 (2010) (PMID:21162445). 50. Nikolaidis, P. T., Villiger, E., Vancini, R. L., Rosemann, T. & Knechtle, B. The effect of sex and performance level on pacing in duathlon. Sports (Basel) 6(4), 152 (2018) (PMID:30477088). 51. Deaner, R. O., Addona, V., Carter, R. E., Joyner, M. J. & Hunter, S. K. Fast men slow more than fast women in a 10 kilometer road race. PeerJ 4, e2235 (2016) (PMID:27547544). 52. Knechtle, B. & Nikolaidis, P. T. Sex differences in pacing during “Ultraman Hawaii”. PeerJ 4, e2509 (2016) (PMID:27703854). 53. Deaner, R. O. & Lowen, A. Males and females pace differently in high school cross-country races. J. Strength Cond. Res. 30(11), 2991–2997 (2016) (PMID:26950352). 54. Renfree, A., do Carmo, E. C. & Martin, L. The influence of performance level, age and gender on pacing strategy during a 100-km ultramarathon. Eur. J. Sport Sci. 16(4), 409–415 (2016) (PMID:26034882). 55. Nikolaidis, P. T. et al. The combined effect of aging and performance level on pacing in duathlon: The “ITU Powerman Long Distance Duathlon World Championships”. Front. Psychol. 10, 296 (2019) (PMID:30833921). Author contributions J.T. obtained the data and drafted the manuscript, C.V.S. performed the statistical analysis and prepared methods and results. M.S.A., M.T., I.C., P.T.N., K.W. and B.K. helped in drafting the final version. All authors read and approved the final manuscript. Competing interests The authors declare no competing interests. Additional information Correspondence and requests for materials should be addressed to B.K. Reprints and permissions information is available at www.nature.com/reprints. Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/. © The Author(s) 2023

Participation and performance trends in short-, medium, and long-distance duathlon.

06-08-2023

Turnwald, Jonas,Sousa, Caio Victor,Andrade, Marilia Santos,Thuany, Mabliny,Cuk, Ivan,Nikolaidis, Pantelis Theodoros,Weiss, Katja,Knechtle, Beat

eng

PMC5944970

RESEARCH ARTICLE The effects of short term detraining and retraining on physical fitness in elite soccer players Chang Hwa Joo* Department of Football Science, Honam University, Gwangsan-gu, Gwangju, South Korea * footballer@honam.ac.kr Abstract Purpose The aim of this study was to examine the effects of aerobic high-intensity training with reduced volume and training cessation on body composition and physical fitness after the end of season and the time required to recapture physical fitness with intensified retraining following two weeks of detraining in elite soccer players. Method Twenty male semi-professional soccer players participated in this study. The soccer players were assigned to either a group that completed high-intensity aerobic training (HAT, n = 10) or to a detraining and retraining group (DHAT, n = 10) for a 5-week period immediately after the end of the season. The first 2 weeks of the period, members of the HAT group performed high-intensity aerobic exercise (80–90% of HRmax, 12 min × 3, three times per week), whereas members of the DHAT group abstained from any physical activity. During the sub- sequent 3 weeks, members of both the HAT and DHAT groups completed high-intensity aerobic exercise. Exercise performance testing and body composition analysis were per- formed before; after 2 weeks of detraining; and at 1, 2 and 3 weeks of retraining. Results Intensified high-intensity training for 5 weeks maintained the performance in the Yo-Yo Inter- mittent Recovery level 2 test (Yo-Yo IR2) and repeated sprints at any time point (P > 0.05). However 2 weeks of detraining resulted in significant decreases in the performance on the Yo-Yo IR2 (P < 0.01) and repeated sprints test (P < 0.05). Performance on the Yo-Yo IR2 enhanced after 2 weeks of retraining and was maintained up to 3 weeks after retraining, with no significant differences between conditions (P > 0.05). In addition, repeated sprint perfor- mance markedly decreased after the detraining period (P < 0.05) and was continuously lower compared to the baseline at 2 weeks after retraining (P < 0.05). Furthermore, this value reached baseline level at the end of the experimental period (P > 0.05). There were no significant differences between conditions in body composition, performance of agility, or sprint ability throughout the 5-week experimental period (P > 0.05). PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 1 / 15 a1111111111 a1111111111 a1111111111 a1111111111 a1111111111 OPEN ACCESS Citation: Joo CH (2018) The effects of short term detraining and retraining on physical fitness in elite soccer players. PLoS ONE 13(5): e0196212. https://doi.org/10.1371/journal.pone.0196212 Editor: Alessandro Zagatto, Sao Paulo State University - UNESP, BRAZIL Received: August 23, 2017 Accepted: March 7, 2018 Published: May 10, 2018 Copyright: © 2018 Chang Hwa Joo. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. Data Availability Statement: All relevant data are within the paper, its Supporting Information file, and Dryad Digital Repository (doi:10.5061/dryad. mc60n0c). Funding: The author has no support or funding to report. Competing interests: The author has declared that no competing interests exist. Conclusions The present data suggest that short-term detraining after the competitive season can markedly decrease performances in the Yo-Yo IR2 test and repeated sprints. To return to a previous level of ability on the Yo-Yo IR2 and/or sprint test with retraining through high-inten- sity aerobic training after a period of detraining, a similar or longer period of retraining is required. However, the high-intensity training with reduced amount of training after competi- tive season can prevent reductions in physical fitness. Introduction Soccer is a high intensity intermittent exercise that requires a high level of physical fitness for players to successfully perform in the game. Elite soccer players perform 587±133 m of high- speed running (19.8–25.2 km/h) and 184 ± 87 m of sprinting (>25.2 km/h) during a typical game [1]. The total distance of high-intensity running depends on the position of the player and team success in a league [2]. The amount of high-intensity running performed during a game also depends on the competitive standards between leagues: top-class professional soccer player perform more high-intensity running compared with moderate professional soccer players [3]. Thus, high level of physical performance is an important factor in determining team success in soccer. Due to the high intensity performance required in soccer, players should perform system- atic and scientific physical fitness training. Several studies have shown that high-intensity training improves soccer players’ fitness levels and skills, such as sprint, strength, and speed endurance [4, 5]. The organization of fitness training for soccer players varies according to the time frame of the periodization along with changes in training volume and intensity. These changes seek to the optimize player’s physical condition and minimize injury [6]. For example, training is conducted to improve physical fitness during the preseason in preparation for the impending competitive season [7, 8]. Elite soccer players normally cease training or perform training with reduced volume and lower intensity for more than two weeks after the end of the competitive season for physical and mental recovery. A prolonged period of rest after the competitive season causes the partial or complete loss of training-induced physiological and performance adaptations, which is defined as detraining [9]. The magnitude of changes during training-induced adaptations after detraining is different depending on the fitness level and the duration of training cessation or insufficient training [9]. Three to six weeks of detraining did not result in changes in aerobic capacity and muscle strength in recreational players and untrained individuals [10–12]. How- ever, decreases in physical fitness are inevitable after such a period of detraining in well-trained elite players who have a relatively higher level of fitness compared to recreational players [9, 13]. Unlike reduced physical fitness after a prolonged period of detraining in elite players, the effects of short-term detraining (~2 weeks) on fitness are controversial. Buchheit et al. [14] observed that short-term detraining after a competitive season improved levels of strength and cardiorespiratory fitness in Australian football players [14]. In contrast, several studies reported that physical fitness was reduced after a short-term detraining period in elite soccer players [5, 15]. The reasons for these contrasting results are not apparent, but may be due to differences in sports and testing methods. Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 2 / 15 During the preseason, the aim of training is mainly to improve physical fitness, while dur- ing the in-season period, it is performed to develop playing strategies and to enhance perfor- mance, while maintaining physical fitness. High-intensity training is a more efficient method of inducing skeletal muscle adaptation in comparison to moderate-intensity training [16]. High-intensity aerobic training has been widely used by athletes to improve physical fitness during the preseason. Indeed, high-intensity aerobic training consisting of 4 bouts of 4 min at 90–95% of the maximum heart rate during the preseason significantly improved aerobic fit- ness and match performance in soccer players [17]. Those results indicated that high-intensity aerobic training might be effective at improving the physical fitness of soccer players and inducing rapid training adaptation in skeletal muscle during the preseason. In order to start the season without injury, athletes must gradually improve their post-sea- son, resting period-induced reduction in physical fitness with an appropriate exercise intensity and volume. However, there is limited information available regarding the effects of retraining during pre-season training in well-trained elite soccer players. Therefore, the aim of the study was to investigate 1) the effects of aerobic high-intensity training with reduced volume and training cessation on body composition and physical fitness after the end of season and 2) the time required to return to the previous level of physical fitness with intensified retraining fol- lowing two weeks of detraining in semi-professional soccer players. Materials and methods Participants Twenty semi-professional male Korea soccer players (age: 22.1±1.8 years, height: 175.5±4.7 cm). The Korean professional soccer league is divided into K League Classic (first division) and K League Challenge (second division). The semi-professional league consists of the National League and K3 Leagues (K3 League Advanced [12 teams] and K3 League Basic [8 teams]). The soccer players participating in this study were members of K3-league teams. All participants had experience of elite soccer players for at least more than 7 years. All partici- pants were non-smokers, no history of neurological disease or musculoskeletal abnormality and none were under any pharmacological treatment during the course of the study. Ethics statement Before testing, all participants gave written informed consent to participate after details and procedures of the study had been fully explained. All of the fitness testing and exercise were performed in the research institute for sport and exercise science at Honam Unviersity. All of the experimental protocols and related procedures were approved by the ethical committee of Honam University. Intervention period and training All players participating in the study trained for more than 2 hours per day for 4–5 times per week (excluding matches) during the previous season. An independent research assistant selected the 20 participants from among 35 players who were between 20 and 23 years of age by drawing a sealed envelope containing a player’s name followed by drawing another sealed envelope containing the name of the group to which they were assigned (i.e., high-intensity aerobic training (HAT) or detraining and high-intensity aerobic training (DHAT) group). The two-week detraining period started immediately after the last match of the season. The fitness tests were conducted two days and one day before the last match as a pre-test; after two weeks of detraining; and at one, two, and three weeks of retraining. Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 3 / 15 During the detraining experimental period, high-intensity aerobic training was performed three times per week for two weeks in the HAT group. After approximately a 20-min warm-up period, the players performed a soccer drill (Fig 1) on an artificial grass surface and three repe- titions of 12 min of exercise at 80–90% of the maximum heart rate (HRmax) measured during Yo-Yo IR2 test. These repetitions were interspersed by 3 min active recovery. The players con- trolled exercise intensity by watching their HR monitor that recorded at 5 s intervals (Polar Team System, Polar, Electro Oy, Kempele, Finland). These data were downloaded to a per- sonal laptop for further analysis. The mean HR during the 12 min exercise sessions was 87.3±1.5% of HRmax. The DHAT group did not perform any exercise sessions during the two weeks of detraining and conducted normal daily activities. Fig 1. Diagram of high-intensity training. https://doi.org/10.1371/journal.pone.0196212.g001 Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 4 / 15 During the retraining period, the HAT and DHAT groups performed high-intensity aero- bic training (12 min × 3) four times per week for three weeks. The mean HR during the 12 min exercise sessions was 86.5±1.4% of HRmax in the HAT group. The DHAT group com- pleted moderate intensity aerobic training (HRmax 70–80%; 76.5±3.2%) for two days before completing the high-intensity training (HRmax 80–90%; 87.7±1.3%). Experimental protocol A schematic illustration of the experimental design is shown in Fig 2. The subjects completed the 30 m sprint test, Yo-Yo intermittent recovery level 2 (Yo-Yo IR2) test, arrowhead agility test, repeated sprint test, and isokinetic strength test. The tests were conducted for two days. The participants refrained from alcohol and caffeine in the 24 h prior to the test. The partici- pants arrived at the laboratory having completed the appropriate diet regime to monitor the diet level. The participants were instructed to ingest water 5 mL of water for every kilogram of their body mass 2 h before arriving at the laboratory. Upon the arrival at the laboratory, body composition (Inbody 520, Biospace, Seoul, Korea) and height (BSM, Seoul, Korea) were mea- sured. Following the completion of the baseline assessments, the participants commenced the tests on an artificial grass surface. A 30-m sprint test, arrowhead agility test, and repeated sprints test were performed in the morning. The Yo-Yo IR2 test was conducted in the evening with 5 hours of recovery after lunch. Isokinetic strength tests were performed in the laboratory the next day. Body composition and exercise tests were completed immediately before the end of the season; after two weeks of detraining; and at one, two, and three weeks of retraining intervention. 30m sprint test The sprint tests which consisted of 2 maximal sprints of 30 m with 2-minute rest between each sprint were conducted. The sprint times at 5, 10, 20 and 30 m were recorded using the photo- cell gates (Microgate, Bolzano, Itaia). The participants started to run 50 cm before the photo- cell gate recordings. The fastest times at the distances were recorded for data analysis. Fig 2. A schematic illustration of the experimental design. https://doi.org/10.1371/journal.pone.0196212.g002 Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 5 / 15 Repeated sprint test The repeated sprint test consisted of seven maximal 34.2 m sprints, interspersed by 25 s of active recovery (40 m jogging distance) [18]. Recovery was timed so that the subjects returned to the start line between the 23rd and 24th second. Additionally, verbal feedback was given at 5, 10, 15 and 20 s of the recovery. Performance was measured as the total sprint time in seconds. Yo-Yo intermittent recovery test (level 2) The Yo-Yo IR2 test was performed on an artificial turf. The Yo-Yo IR2 test consists of 2 × 20 m shuttle runs at increasing speeds, controlled by audio signals from a compact disk. Between each bout of running, the subjects completed 10 s of active recovery, consisting of 2 × 5 m jog- ging [19]. The test was terminated when the subjects failed twice to reach the start line on time and the distance (meters) covered at the end point was recorded [5]. Arrowhead agility test The arrowhead agility tests consisted of 4 sprints (2 right, 2 left), with 2-minutes rest between each sprint [20]. Each subject started 50 cm behind the start line and sprinted 10 m forward to a cone. From the cone, the subjects turned at a right angle to a cone being apart from 5m before turning to a cone 15 m straight from the start line. They turned again from the cone to accelerate in a straight line for 15 m over the initial start line to complete the run. The fastest times were recorded for data analysis. Timing gates were used to accurately assess the time to completion. Isokinetic strength The subjects performed the Isokinetic dynamometry (Cybex MET-300, New York, USA) to evaluate the unilateral strength of the concentric contraction of the flexors and extensors of the knee [21]. The angular speed parameters of 60˚ × s-1, 180˚ × s-1, and 240˚ × s-1 were used for the measurements. The results of the measurements were expressed in absolute peak torque (Nm) for the purposes of the off-seasonal variation comparisons. Statistical analysis All data are presented as means ± SD. Two-way analysis of variance (ANOVA) with repeated measure was conducted to determine any treatment differences between the HAT and DHAT conditions. The assumption of sphericity (homogeneity of covariance) was assessed and cor- rected for using the Huynh-Feldt epsilon. Because there were only 2 levels in the main effect of condition, follow-up multiple comparisons were not necessary. A significant effect of time was followed up with planned multiple contrasts in line with the a priori hypotheses. Therefore, data at the specific time points were compared with the baseline (first) time point using New- man-Keuls multiple contrasts. Where a significant interaction between condition and time was observed, differences between conditions were examined at each time point using New- man-Keuls multiple contrasts. Baseline values were compared using an independent samples t test. The alpha level for evaluation of statistical significance was set at P < 0.05. Effect sizes were assessed by partial eta squared (Z2 P), which were defined as trivial (<0.1), small (0.1–0.3), moderate (0.3–0.5) and large (>0.5) [22]. Results Body weight and body fat were similar between the HAT and DHAT groups throughout the experimental period (P > 0.05; Tables 1 and 2). There was no significant effect of condition Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 6 / 15 nor was there an interaction of condition and time (P > 0.05) in the performance of players on sprint and agility tests (Tables 3 and 4). However, a significant effect of time was observed for sprint test at 5 m, 10 m, and 30 m as well as in the left direction of arrowhead agility (P < 0.05). Isokinetic strength at all angular speeds remained similar to baseline under both conditions throughout the experimental period, with no significant effects of time, condition, or an interaction between the two (P > 0.05; Tables 5 and 6). There was a significant interac- tion in the Yo-Yo IR2 test (F = 3.273; P < 0.05; Z2 P ¼ 0:267), while the measurement time (F = 1.517; P > 0.05; Z2 P ¼ 0:144) and condition were not significant (F = 1.938; P > 0.05; Z2 P ¼ 0:177). Compared to the pre-detraining performance, the Yo-Yo IR2 test performance decreased significantly after the two-week detraining period (P < 0.01) and the values reach before detraining level after two weeks of retraining in the DHAT group (P > 0.05). No differ- ences were detected at three weeks post-retraining between conditions (P > 0.05), whilst val- ues in the HAT group remained stable throughout the experimental period (P > 0.05; Fig 3). A main effect of time was found (F = 3.539; P < 0.05; Z2 P ¼ 0:282), along with a significant interaction between condition and time for repeated sprint performance (F = 3.127; P < 0.05; Z2 P ¼ 0:258). No changes in repeated sprint performance were observed at any time point under HAT conditions (P > 0.05), whereas repeated sprint performance declined after two weeks of detraining (P < 0.05) and remained lower than at baseline by two weeks post- Table 1. Body composition of the subjects before, after two weeks of detraining and at one, two and three weeks of retraining (mean ± SD). Pre 2W DT 1W RT 2W RT 3W RT Body weight (kg) HAT 68.1±7.1 68.5±7.1 68.6±7.2 68.6±7.3 68.4±7.3 DHAT 67.5±7.3 67.8±7.3 67.9±7.3 68.2±7.2 67.9±7.3 Body mass index (kg/m2) HAT 22.7±0.4 22.7±0.5 22.7±0.5 22.5±0.6 22.9±0.4 DHAT 22.1±0.8 22.4±0.9 22.4±0.6 22.5±0.9 22.3±0.8 Skeletal muscle mass (kg) HAT 32.6±3.5 32.5±3.3 32.3±3.2 32.6±3.6 32.4±3.3 DHAT 33.0±3.5 33.2±2.3 33.1±3.2 33.4±3.3 33.3±3.3 Percent body fat (%) HAT 9.6±0.7 9.7±0.7 9.8±1.3 9.5±0.9 9.9±1.3 DHAT 9.3±1.2 9.8±1.3 9.8±1.1 9.7±1.2 9.5±1.4 Values are means ± standard deviation https://doi.org/10.1371/journal.pone.0196212.t001 Table 2. Differences in the body composition of the subjects between conditions in each test (n = 20). F P Z2 P Body weight Condition 0.048 0.831 0.005 Time 12.372 0.001 0.579 Condition x Time 0.628 0.646 0.065 Body mass index Condition 2.524 0.147 0.219 Time 0.716 0.587 0.074 Condition x Time 1.776 0.155 0.165 Skeletal muscle mass Condition 0.228 0.644 0.025 Time 0.178 0.948 0.019 Condition x Time 0.117 0.976 0.013 Percent body fat Condition 0.046 0.834 0.005 Time 2.201 0.088 0.197 Condition x Time 0.653 0.629 0.068 F; testing criteria level, P; level of statistical significance, Z2 P; partial eta squared https://doi.org/10.1371/journal.pone.0196212.t002 Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 7 / 15 retraining under DHAT conditions (P < 0.05). It reached baseline level at the end of the exper- imental period (P > 0.05; Fig 4). Discussion The major findings in the present study were that two weeks of detraining after competitive season decreased performance in the Yo-Yo IR2 test and repeated sprints. The detraining- induced reductions in the Yo-Yo IR2 test performance improved compared to baseline levels after two weeks of high-intensity aerobic training. Meanwhile, three weeks were required to Table 3. Sprint and agility before, after two weeks of detraining and at one, two and three weeks of retraining (mean ± SD). Pre 2W DT 1W RT 2W RT 3W RT 5 m HAT 1.04±0.04 1.05±0.03 1.06±0.04 1.02±0.05 1.02±0.05 DHAT 1.05±0.04 1.04±0.03 1.05±0.03 1.01±0.04 1.01±0.04 10 m HAT 1.75±0.06 1.74±0.10 1.73±0.05 1.71±0.04 1.72±0.06 DHAT 1.78±0.05 1.73±0.05 1.73±0.05 1.71±0.06 1.72±0.07 20 m HAT 3.00±0.09 3.01±0.13 3.02±0.08 2.99±0.06 2.99±0.09 DHAT 3.05±0.05 3.07±0.08 3.03±0.06 2.99±0.09 2.99±0.09 30 m HAT 4.13±0.11 4.22±0.17 4.25±0.12 4.21±0.11 4.23±0.13 DHAT 4.23±0.07 4.30±0.12 4.29±0.09 4.23±0.12 4.25±0.10 Agility (R) HAT 8.04±0.19 8.09±0.22 8.13±0.17 7.99±0.21 8.03±0.22 DHAT 8.06±0.16 8.09±0.22 8.09±0.19 8.04±0.25 8.05±0.25 Agility (L) HAT 7.99±0.17 8.12±0.20 8.10±0.20 7.98±0.23 8.00±0.18 DHAT 8.04±0.18 8.14±0.24 8.14±0.18 8.08±0.15 8.08±0.20 Values are means ± standard deviation. R; right, L; left https://doi.org/10.1371/journal.pone.0196212.t003 Table 4. Differences in sprint and agility between conditions in each test (n = 20). F P Z2 P 5 m Condition 0.095 0.765 0.010 Time 7.657 0.001 0.460 Condition x Time 1.586 0.199 0.150 10 m Condition 0.305 0.594 0.033 Time 4.672 0.004 0.342 Condition x Time 1.010 0.415 0.101 20 m Condition 0.480 0.506 0.051 Time 2.500 0.060 0.217 Condition x Time 1.167 0.342 0.115 30 m Condition 0.879 0.373 0.089 Time 5.357 0.002 0.373 Condition x Time 1.619 0.191 0.152 Agility (R) Condition 0.013 0.912 0.001 Time 2.516 0.058 0.218 Condition x Time 0.357 0.838 0.038 Agility (L) Condition 0.499 0.498 0.053 Time 3.542 0.015 0.282 Condition x Time 0.382 0.820 0.041 R; right, L; left, F; testing criteria level, P; level of statistical significance, Z2 P; partial eta squared https://doi.org/10.1371/journal.pone.0196212.t004 Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 8 / 15 return to the initial level of repeated sprint performance with retraining using high-intensity training. Ultimately, a reduced amount of high-intensity training after the competitive season facilitated the maintenance of physical fitness. The HAT group that continued to perform high-intensity aerobic exercise after the compet- itive season maintained their performance level in the Yo-Yo IR2 test over the five week treat- ment period. These results are supported by previous studies, which indicate that, after the last match of the season, 10 training sessions, consisting of high-intensity training for two weeks, do not change performance in the Yo-Yo IR2 test in elite soccer players [15]. However, Naka- mura et al. [23] observed that running and plyometric training for two days per week for three weeks after the completion of a competitive season did not prevent the decrease in perfor- mance in The Yo-Yo IR2 test in elite soccer players. The reason of these differences in results is unclear but it probably related to exercise intensity. Indeed, there was no significant decrease in performance in the Yo-Yo IR2 test during off-season in the present study and Christensen et al. [15]’s study applying high-intensity exercise despite reduced exercise time compared to that in-season. Furthermore, the exercise intensity was higher than that in the previous study conducted by Nakamura et al. [23], which modulated endurance training (70–80% of HRmax). In the present study, we found that two weeks of detraining after the competitive season markedly decreased performance in the Yo-Yo IR2 test in elite soccer players. Accordingly, Table 5. Peak torques (Nm) during concentric knee flexion and extension before, after two weeks of detraining and at one, two and three weeks of retraining (mean ± SD). Pre 2W DT 1W RT 2W RT 3W RT DL-PT-E-60 HAT 208.2±12.3 208.2±12.3 209.6±27.3 210.0±25.2 205.2±19.8 DHAT 211.5±13.9 212.7±15.9 214.2±15.9 216.7±17.5 208.3±12.4 DL-PT-F-60 HAT 135.8±32.3 135.8±30.6 135.8±37.7 139.2±29.5 135.2±28.2 DHAT 121.2±21.6 137.2±26.9 140.3±29.1 137.2±27.7 136.5±19.1 NL-PT-E-60 HAT 189.9±33.1 194.6±39.0 197.0±35.2 200.9±38.1 191.0±31.0 DHAT 198.1±28.0 189.1±24.2 183.9±26.8 193.0±24.6 187.0±16.4 NL-PT-F-60 HAT 129.1±30.5 127.9±26.4 125.1±34.8 128.1±26.1 127.1±31.5 DHAT 132.6±23.1 127.1±20.2 125.7±23.1 131.9±24.6 135.3±26.2 DL-PT-E-180 HAT 138.8±18.9 146.6±21.4 146.6±23.9 145.3±23.3 140.4±18.8 DHAT 145.2±23.6 146.1±15.1 152.0±20.1 150.3±15.4 152.6±18.8 DL-PT-F-180 HAT 105.4±17.3 105.2±14.8 108.5±13.8 107.3±10.9 107.7±13.4 DHAT 108.5±9.8 107.2±14.2 110.6±15.7 110.5±17.3 106.1±15.3 NL-PT-E-180 HAT 136.9±18.9 137.0±25.4 139.5±21.1 141.1±18.0 137.7±19.5 DHAT 138.1±16.5 139.6±16.6 137.9±21.9 142.7±21.4 138.5±20.8 NL-PT-F-180 HAT 97.4±18.7 93.7±18.4 99.7±24.4 96.4±18.9 95.7±21.6 DHAT 102.6±16.7 100.5±15.3 100.7±19.2 101.7±17.5 102.6±19.9 DL-PT-E-240 HAT 114.9±19.0 115.9±15.9 116.9±17.6 115.2±15.6 113.4±17.1 DHAT 116.2±16.2 118.6±12.4 117.5±13.6 119.5±12.7 115.8±11.7 DL-PT-F-240 HAT 84.6±15.1 83.8±16.8 84.8±14.1 87.3±12.2 86.2±13.8 DHAT 89.1±9.3 87.8±13.4 92.6±17.4 92.9±16.0 89.3±12.7 NL-PT-E-240 HAT 113.3±16.7 112.7±15.4 112.6±13.7 114.4±11.2 110.3±14.6 DHAT 115.2±10.8 110.3±13.1 116.1±14.6 114.2±16.8 115.4±13.2 NL-PT-F-240 HAT 83.3±15.2 83.4±18.6 81.0±19.2 85.1±16.2 86.3±12.8 DHAT 88.4±14.6 81.7±14.5 86.3±18.4 89.2±18.6 87.2±18.6 Values are means ± standard deviation. DL; dominant leg, NL; non-dominant leg, PT; peak torque, E; extensors, F; flexors, 60, 180, 240; angular velocities (˚
s-1) https://doi.org/10.1371/journal.pone.0196212.t005 Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 9 / 15 Thomassen et al. [5] and Christensen et al. [15] observed that the Yo-Yo IR2 test performance after detraining for two weeks decreased from 845 m to 654 m in elite soccer players. In addi- tion, a study from another laboratory reported that a prolonged detraining period can induce an 8% decline in maximal oxygen consumption [24], which is strongly associated with distance on the Yo-Yo IR2 test [25]. The degree of deterioration of physical fitness over the course of the detraining period after the competitive season is closely related to the fitness level of Table 6. Differences in sprint and agility peak torques (Nm) during concentric knee flexion and extension between conditions in each test (n = 20). F P Z2 P DL-PT-E-60 Condition 1.674 0.228 0.157 Time 1.665 0.180 0.156 Condition x Time 0.100 0.982 0.011 DL-PT-F-60 Condition 0.032 0.862 0.004 Time 1.249 0.308 0.122 Condition x Time 1.156 0.346 0.114 NL-PT-E-60 Condition 0.113 0.744 0.012 Time 0.721 0.583 0.074 Condition x Time 1.050 0.395 0.104 NL-PT-F-60 Condition 0.059 0.814 0.007 Time 1.287 0.293 0.125 Condition x Time 0.469 0.758 0.050 DL-PT-E-180 Condition 0.355 0.566 0.038 Time 1.283 0.295 0.125 Condition x Time 0.695 0.600 0.072 DL-PT-F-180 Condition 0.143 0.714 0.016 Time 0.419 0.794 0.045 Condition x Time 0.268 0.896 0.029 NL-PT-E-180 Condition 0.007 0.935 0.001 Time 0.485 0.747 0.051 Condition x Time 0.177 0.949 0.019 NL-PT-F-180 Condition 0.373 0.556 0.040 Time 0.481 0.749 0.051 Condition x Time 0.520 0.721 0.055 DL-PT-E-240 Condition 0.097 0.762 0.011 Time 0.549 0.701 0.057 Condition x Time 0.153 0.960 0.017 DL-PT-F-240 Condition 0.872 0.375 0.088 Time 0.834 0.512 0.085 Condition x Time 0.297 0.878 0.032 NL-PT-E-240 Condition 0.047 0.833 0.005 Time 0.651 0.630 0.067 Condition x Time 0.971 0.435 0.097 NL-PT-F-240 Condition 0.147 0.711 0.016 Time 1.786 0.153 0.166 Condition x Time 0.696 0.600 0.072 DL; dominant leg, NL; non-dominant leg, PT; peak torque, E; extensors, F; flexors, 60, 180, 240; angular velocities (˚
s-1), F; testing criteria level, P; level of statistical significance, Z2 P; partial eta squared https://doi.org/10.1371/journal.pone.0196212.t006 Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 10 / 15 athletes [23]. Therefore, these results can support the notion that performance in the Yo-Yo IR2 test can be reduced despite only a few days of detraining in elite soccer players with a high level of physical fitness. These decreases in performance in the Yo-Yo IR2 test can be explained at the muscle level. Several weeks of detraining lead to a return in muscle capillarization to Fig 3. Yo-Yo IR2 performan for the high-intensity training (HAT; n = 10) and detraining + retraining (DHAT; n = 10) before, after two weeks detraining and at one, two and three weeks of retraining (n = 11, mean ± SD). P < 0.01; significantly different from pre. P < 0.05; significantly different from pre. ##P < 0.01; significantly between conditions. #P < 0.05; significantly between conditions. https://doi.org/10.1371/journal.pone.0196212.g003 Fig 4. Repeated sprint test for the high-intensity training (HAT; n = 10) and detraining + retraining (DHAT; n = 10) before, after two weeks detraining and at one, two and three weeks of retraining (n = 11, mean ± SD). P < 0.05; significantly different from pre. #P < 0.05; significantly between conditions. https://doi.org/10.1371/journal.pone.0196212.g004 Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 11 / 15 baseline before detraining in athletes and a 25%-45% decline in oxidative enzyme activities, which result in reduced mitochondrial ATP production in skeletal muscle [9]. Several previous studies have reported that high intensity training improves the perfor- mance in the Yo-Yo IR2 test of elite soccer players [25, 26]. In line with these results, high- intensity aerobic training after two weeks of detraining was found to improve performance in the Yo-Yo IR2 test in the present study. Two weeks of retraining with high-intensity exercise is required to return close to the baseline level of performance. This result is inconsistent with a previous study that suggested that athletes with a high fitness level must perform exercise training for a period that is at least twice as long as the resting time period in order to improve their physical fitness to a level of before detraining [24]. The discrepancy in time periods required to return to the physical fitness level at baseline can be due to variations in the length of the detraining period (four weeks versus eight weeks) and the fitness level of the athletes (compared to end of season versus before the Olympic game). Indeed, the performance in the Yo-Yo IR2 test decreased by 11% at the end of the season compared to the start of the season and a 42% increase was observed during the eight weeks of pre-season training [25]. This phe- nomenon is likely due to accumulated fatigue experienced during the competitive season. This assumption is supported by the finding from the present study that the performance in the Yo- Yo IR2 test was higher at three weeks of post retraining compared to baseline. Furthermore, Noon et al. [20] and Oliver et al. [27] observed that perceptual well-being (e.g., motivation, sleep quality, recovery, appetite, fatigue, stress, muscle soreness, stiffness) deteriorated with an increase in training exposure and accumulated fatigue as the season progressed in elite athletes. Repeated sprint performance did not change over five weeks of high-intensity training after competitive season in the present study. In contrast to the present study, previous studies reported that two weeks of high-intensity training immediately after the end of season enhanced repeated sprint performance in elite soccer players [5, 15]. These different results may be associated with the high-intensity training method used during the retraining period. Aguiar et al. [28] observed that intermittent training for 12 weeks consisting of 20 minutes per training session resulted in greater improvements in repeated sprint performance than did continuous training. Indeed, the training sessions in the present study largely comprised of high-intensity endurance exercise, whereas the training sessions used in previous studies con- sisted of five high-intensity aerobic training, including small-sided (4 vs. 4 and 3 vs. 3) soccer drills (8 × 2 min) and five speed endurance training (10–12 × 25–30 s sprints) over the course of two weeks. In other respects, since well-trained athletes are more sensitive to changes in physical fitness with inadequate training intensity and do not easily experience improvements following further training due to the ceiling effect [25], the capacity of repeated sprint perfor- mance of the players in the present study might be optimal by the end of the competitive sea- son. This is supported by the observation that repeated sprint performance in players from the present study was similar to previous study conducted with professional soccer players during the competitive season [18]. It is well known that anaerobic exercise performance decreases in highly trained elite soccer players, despite a short period of detraining after the competitive season [9]. There was also a significant decrease in repeated sprint performance over two weeks of detraining after the end of a match in the present study. The detraining-induced decrease in performance gradually increased during the three weeks retraining period. The aerobic high-intensity training- induced increase in repeated sprint performance in the present study is likely to be the result of training-induced biochemical adaptation in skeletal muscles. Thomassen et al. [5] and Christensesn et al. [15] observed that two weeks of high-intensity exercise immediately after the last match of the season enhanced Na+-K+ pump α2-isoform expression by 15%, increased Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 12 / 15 the FXYD1ser68-to-FXYD1 ratio by 27%, increased the level of pyruvate dehydrogenase by 17%, and improved repeated sprint performance. In comparison, cessation of training for two weeks did not affect the expression of Na+-K+ pump isoform expression and resulted in a reduction of the AB_FXYD1ser68 signal by 18%; decreased pyruvate dehydrogenase level by 17%; a drop in citrate synthase and 3-hydroxyacyl-CoA activity to 12% and 18% of maximal, respectively; and a reduction in performance. However, repeated sprint performance at 3 weeks post-retraining was still lower than the performance recorded at baseline. As men- tioned, aerobic high-intensity training is not optimal for improving repeated sprint perfor- mance, which represents the capacity for anaerobic exercise performance. On the contrary, improvements in repeated sprint performance through aerobic high-intensity training might be associated with the training period during the preseason. Recently, Teixeira et al. [29] reported that high-intensity aerobic training involving shuttle-run intervals (4 × 4 min) for five weeks during the preseason enhanced repeated sprint ability with increased aerobic per- formance in elite athletes. When considered, these findings suggest that more than three weeks of high-intensity aerobic training is required to develop repeated sprint performance during preseason in elite players. The observed lack of changes in body composition and sprint performances (10 m, 20 m, 30 m) for five weeks during the study period in both groups disagrees with previous studies that engaged in more than two weeks of detraining [13, 30]. For example, Koundourakis et al. [13] examined the effect of detraining on exercise performance and body composition in professional soccer players. They observed that prolonged detraining period (six weeks) signif- icantly increased body weight and body fat percentage and reduced maximal oxygen con- sumption and performances in squat-jump, countermovement-jump, and sprints (10 m, 20 m). These results suggest that a short period of detraining (approximately two weeks) may not lead to changes in body composition and explosive exercise performance in well-trained soccer players. This is supported by findings that there were changes in neither isokinetic strength at any angular speeds in the present study nor squat, vertical jump, or isometric and isokinetic knee force following two weeks detraining in high fitness athletes [31]. A possible explanation for the absence of changes in explosive exercise performance after a short period of detraining is the lack of changes in muscle fiber characteristics. Mujika and Padilla. [9] reported that two weeks of detraining did not alter muscle fiber distribution in well-trained athletes. However, three weeks of detraining after the first half of a competitive season in elite soccer players resulted in changes in skeletal muscle morphology, including a reduction in mean fast twitch (FT) fiber cross-sectional area and reduction in mitochondrial enzyme activities and exercise performance [32]. Taken together, these data suggest that more than two weeks of detraining in elite soccer players could have resulted in a decrease in explosive exercise performance by reduced ATP production in skeletal muscle. Conclusions In conclusion, the findings demonstrate that two weeks of detraining after the competitive sea- son resulted in a marked decrease in performance in the Yo-Yo IR2 test and repeated sprints. To return to a previous level of physical fitness with retraining through high-intensity aerobic training after a period of detraining required a similar period of retraining for performance in Yo-Yo IR2 and/or more periods for repeated sprint performance. The off-season rest period did not result in changes in explosive exercise performances and body composition. Aerobic high-intensity training with reduced training volume after a competitive season can prevent reductions in performances in the Yo-Yo IR2 test and repeated sprints. On the contrary, the decrease in aerobic and anaerobic performance induced by two weeks of detraining was Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 13 / 15 recovered within a few weeks of adequate training during the preseason. Therefore, these find- ings suggest that elite soccer players can be allowed to take short periods of rest (~2 weeks) without training during the off-season for the release of mental and physical stress that is accu- mulated throughout the competitive season. Supporting information S1 File. Raw data of Figs 3 and 4 and Tables 1, 2, 3, 4, 5 and 6. (XLSX) Author Contributions Writing – original draft: Chang Hwa Joo. References 1. Carling C, Bradley P, McCall A, Dupont G. Match-to-match variability in high-speed running activity in a professional soccer team. J Sports Sci. 2016; 34(24):2215–23. Epub 2016/05/05. https://doi.org/10. 1080/02640414.2016.1176228 PMID: 27144879. 2. Di Salvo V, Gregson W, Atkinson G, Tordoff P, Drust B. Analysis of high intensity activity in Premier League soccer. Int J Sports Med. 2009; 30(3):205–12. Epub 2009/02/14. https://doi.org/10.1055/s- 0028-1105950 PMID: 19214939. 3. Mohr M, Krustrup P, Bangsbo J. Match performance of high-standard soccer players with special refer- ence to development of fatigue. J Sports Sci. 2003; 21(7):519–28. Epub 2003/07/10. https://doi.org/10. 1080/0264041031000071182 PMID: 12848386. 4. Kotzamanidis C, Chatzopoulos D, Michailidis C, Papaiakovou G, Patikas D. The effect of a combined high-intensity strength and speed training program on the running and jumping ability of soccer players. J Strength Cond Res. 2005; 19(2):369–75. Epub 2005/05/21. 5. Thomassen M, Christensen PM, Gunnarsson TP, Nybo L, Bangsbo J. Effect of 2-wk intensified training and inactivity on muscle Na+-K+ pump expression, phospholemman (FXYD1) phosphorylation, and performance in soccer players. J Appl Physiol (1985). 2010; 108(4):898–905. Epub 2010/02/06. https:// doi.org/10.1152/japplphysiol.01015.2009 PMID: 20133439. 6. Mara JK, Thompson KG, Pumpa KL, Ball NB. Periodization and physical performance in elite female soccer players. Int J Sports Physiol Perform. 2015; 10(5):664–9. Epub 2015/01/23. https://doi.org/10. 1123/ijspp.2014-0345 PMID: 25611789. 7. Jeong TS, Reilly T, Morton J, Bae SW, Drust B. Quantification of the physiological loading of one week of "pre-season" and one week of "in-season" training in professional soccer players. J Sports Sci. 2011; 29(11):1161–6. Epub 2011/07/23. https://doi.org/10.1080/02640414.2011.583671 PMID: 21777053. 8. Buchheit M, Racinais S, Bilsborough JC, Bourdon PC, Voss SC, Hocking J, et al. Monitoring fitness, fatigue and running performance during a pre-season training camp in elite football players. J Sci Med Sport. 2013; 16(6):550–5. Epub 2013/01/22. https://doi.org/10.1016/j.jsams.2012.12.003 PMID: 23332540. 9. Mujika I, Padilla S. Detraining: loss of training-induced physiological and performance adaptations. Part I: short term insufficient training stimulus. Sports Med. 2000; 30(2):79–87. Epub 2000/08/31. PMID: 10966148. 10. Moore RL, Thacker EM, Kelley GA, Musch TI, Sinoway LI, Foster VL, et al. Effect of training/detraining on submaximal exercise responses in humans. J Appl Physiol (1985). 1987; 63(5):1719–24. Epub 1987/11/01. https://doi.org/10.1152/jappl.1987.63.5.1719 PMID: 3693207. 11. Izquierdo M, Ibanez J, Gonzalez-Badillo JJ, Ratamess NA, Kraemer WJ, Hakkinen K, et al. Detraining and tapering effects on hormonal responses and strength performance. J Strength Cond Res. 2007; 21(3):768–75. Epub 2007/08/10. 12. Mujika I, Padilla S. Detraining: loss of training-induced physiological and performance adaptations. Part II: Long term insufficient training stimulus. Sports Med. 2000; 30(3):145–54. Epub 2000/09/22. PMID: 10999420. 13. Koundourakis NE, Androulakis NE, Malliaraki N, Tsatsanis C, Venihaki M, Margioris AN. Discrepancy between exercise performance, body composition, and sex steroid response after a six-week detraining period in professional soccer players. PLoS One. 2014; 9(2):e87803. Epub 2014/03/04. https://doi.org/ 10.1371/journal.pone.0087803 PMID: 24586293. Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 14 / 15 14. Buchheit M, Morgan W, Wallace J, Bode M, Poulos N. Physiological, psychometric, and performance effects of the Christmas break in Australian football. Int J Sports Physiol Perform. 2015; 10(1):120–3. Epub 2014/05/09. https://doi.org/10.1123/ijspp.2014-0082 PMID: 24806508. 15. Christensen PM, Krustrup P, Gunnarsson TP, Kiilerich K, Nybo L, Bangsbo J. VO2 kinetics and perfor- mance in soccer players after intense training and inactivity. Med Sci Sports Exerc. 2011; 43(9):1716– 24. Epub 2011/02/12. https://doi.org/10.1249/MSS.0b013e318211c01a PMID: 21311360. 16. Bartlett JD, Close GL, MacLaren DP, Gregson W, Drust B, Morton JP. High-intensity interval running is perceived to be more enjoyable than moderate-intensity continuous exercise: implications for exercise adherence. J Sports Sci. 2011; 29(6):547–53. Epub 2011/03/02. https://doi.org/10.1080/02640414. 2010.545427 PMID: 21360405. 17. Impellizzeri FM, Marcora SM, Castagna C, Reilly T, Sassi A, Iaia FM, et al. Physiological and perfor- mance effects of generic versus specific aerobic training in soccer players. Int J Sports Med. 2006; 27(6):483–92. Epub 2006/06/13. https://doi.org/10.1055/s-2005-865839 PMID: 16767613. 18. Abrantes C, Macas V, Sampaio J. Variation in football players’ sprint test performance across different ages and levels of competition. J Sports Sci Med. 2004; 3(YISI 1):44–9. Epub 2004/11/01. PMID: 24778553. 19. Bangsbo J, Iaia FM, Krustrup P. The Yo-Yo intermittent recovery test: a useful tool for evaluation of physical performance in intermittent sports. Sports Med. 2008; 38(1):37–51. Epub 2007/12/18. PMID: 18081366. 20. Noon MR, James RS, Clarke ND, Akubat I, Thake CD. Perceptions of well-being and physical perfor- mance in English elite youth footballers across a season. J Sports Sci. 2015; 33(20):2106–15. Epub 2015/09/19. https://doi.org/10.1080/02640414.2015.1081393 PMID: 26383605. 21. Kilinc BE, Kara A, Camur S, Oc Y, Celik H. Isokinetic dynamometer evaluation of the effects of early thigh diameter difference on thigh muscle strength in patients undergoing anterior cruciate ligament reconstruction with hamstring tendon graft. J Exerc Rehabil. 2015; 11(2):95–100. Epub 2015/05/12. https://doi.org/10.12965/jer.150100 PMID: 25960982. 22. Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc. 2009; 41(1):3–13. Epub 2008/12/19. https://doi.org/10. 1249/MSS.0b013e31818cb278 PMID: 19092709. 23. Nakamura D, Suzuki T, Yasumatsu M, Akimoto T. Moderate running and plyometric training during off- season did not show a significant difference on soccer-related high-intensity performances compared with no-training controls. J Strength Cond Res. 2012; 26(12):3392–7. Epub 2011/12/31. https://doi.org/ 10.1519/JSC.0b013e3182474356 PMID: 22207263. 24. Godfrey RJ, Ingham SA, Pedlar CR, Whyte GP. The detraining and retraining of an elite rower: a case study. J Sci Med Sport. 2005; 8(3):314–20. Epub 2005/10/27. PMID: 16248472. 25. Krustrup P, Mohr M, Nybo L, Jensen JM, Nielsen JJ, Bangsbo J. The Yo-Yo IR2 test: physiological response, reliability, and application to elite soccer. Med Sci Sports Exerc. 2006; 38(9):1666–73. Epub 2006/09/09. https://doi.org/10.1249/01.mss.0000227538.20799.08 PMID: 16960529. 26. Mohr M, Krustrup P. Comparison between two types of anaerobic speed endurance training in competi- tive soccer players. J Hum Kinet. 2016; 51:183–92. Epub 2017/02/06. https://doi.org/10.1515/hukin- 2015-0181 PMID: 28149381. 27. Oliver JL, Lloyd RS, Whitney A. Monitoring of in-season neuromuscular and perceptual fatigue in youth rugby players. Eur J Sport Sci. 2015; 15(6):514–22. Epub 2015/09/15. https://doi.org/10.1080/ 17461391.2015.1063700 PMID: 26366619. 28. Aguiar M, Abrantes C, Mac¸ãs V, Leite N, Sampaio J, Iba´ñez S. Effects of intermittent or continuous training on speed, jump and repeated-sprint ability in semi-professional soccer players. The Open Sports Sciences Journal. 2008; 1:15–9. 29. Teixeira AS, Arins FB, De Lucas RD, Carminatti LJ, Dittrich N, Nakamura FY, et al. Comparative effects of two interval shuttle-run training modes on physiological and performance adaptations in female pro- fessional futsal players. J Strength Cond Res. 2017. Epub 2017/09/14. https://doi.org/10.1519/JSC. 0000000000002186 PMID: 28902113. 30. Ostojic SM. Seasonal alterations in body composition and sprint performance of elite soccer players. Journal of Exercise physiology online. 2003; 6(3). 31. Hortobagyi T, Houmard JA, Stevenson JR, Fraser DD, Johns RA, Israel RG. The effects of detraining on power athletes. Med Sci Sports Exerc. 1993; 25(8):929–35. Epub 1993/08/01. PMID: 8371654. 32. Bangsbo J, Mizuno M. Morphological and metabolic alterations in soccer players with detraining and retraining and their relation to performance1988. Detraining and retraining affect physical fitness PLOS ONE | https://doi.org/10.1371/journal.pone.0196212 May 10, 2018 15 / 15

The effects of short term detraining and retraining on physical fitness in elite soccer players.

05-10-2018

Joo, Chang Hwa

eng