Development of 11- to 16-year-olds’ short-term power output determined using both treadmill running and cycle ergometry

Young people’s aerobic power (peak oxygen uptake; peak \( \dot{V}{\text{O}}_{2} \)) determined running on a treadmill or pedalling on a cycle ergometer is well documented but short-term power output (STP) principally reliant on anaerobic metabolism is less extensively researched. Ethical and technological constraints restrict direct measurement of the intramuscular rate of energy production and current knowledge of the development of STP is founded on cross-sectional analyses of performance outcomes. Paediatric research has primarily focused on the assessment of STP during cycling and unlike research into the development of aerobic power little attention has been given to its determination during running.

Although several methods of assessing STP have been explored, understanding of STP in childhood and adolescence is principally derived from performance on variants of the Wingate anaerobic test (WAnT) (Williams and Ratel 2017). Power output in the WAnT is calculated from maximal pedalling cadence against a fixed braking force with peak power output (PP) recorded within a few seconds of exercise onset and total power output averaged over the 30 s test period and expressed as mean power output (MP). Since its introduction by Cumming (1973) and subsequent development at the Wingate Institute, the WAnT has been shown to be a robust and reliable test (Inbar et al. 1996). Paediatric physiologists have, however, developed a plethora of different WAnT protocols including the use of pre-test warm-ups (Inbar and Bar-Or 1975), rolling starts (Armstrong et al. 1997), and toe clips (Lavoie et al. 1984). The time (e.g. 1 s, 3 s, or 5 s) over which PP is recorded varies across studies (Williams and Ratel 2017) and some researchers have incorporated estimations of the work done to overcome the inertia of the flywheel and the internal resistance of the cycle ergometer (Chia et al. 1997) into their calculations. With children and adolescents, concerns have been raised about the ideal body mass-related braking force (Bar-Or 1993). Some researchers have identified individual optimal braking forces and, therefore, optimized peak power (OPP) through a series of short sprints performed against a range of randomly introduced braking forces. Across the 11- to 16-year-old age group, PP has been demonstrated to be generally resilient to moderate variations around the conventional braking force of 0.74 N kg⁻¹ (Dotan and Bar-Or 1983; Santos et al. 2002), but if the braking force is set to optimize PP it may be too high for optimal MP determination. As the WAnT progresses, increasing fatigue causes a reduction in pedal cadence, thus affecting the power-to-velocity ratio resulting in a fall in power output and a lower MP. Moreover, MP is a function of interplay between anaerobic and aerobic metabolism with both girls and boys reaching ~ 65–75% of their peak \( \dot{V}{\text{O}}_{2} \) in a WAnT (Chia 2006). The influence of peak \( \dot{V}{\text{O}}_{2} \) on the development of 11- to 16-year-olds’ MP is unknown.

Cross-sectional studies of WAnT-determined PP and MP are plentiful but the use of a variety of methodologies and power output calculations has precluded confident comparisons of data across studies. Prior to the present project, there were no longitudinal studies of the PP and MP of both boys and girls and only two short duration longitudinal studies with small numbers of boys (Duché et al. 1993; Falk and Bar-Or 1993). There are large differences in reports of ‘typical’ age-related absolute values (i.e. in W) of PP and particularly MP which includes significant but, as yet unquantified, contributions from \( \dot{V}{\text{O}}_{2} \). Nevertheless, data trends with age, at least for PP, are consistent and indicate that it increases with age until ~ 13 years with no significant sex difference. From ~ 13 to 14 years of age, the extant data indicate that boys’ PP markedly increases into young adulthood whereas girls experience a smaller increase with age over the same time period (Van Praagh and Doré 2002).

Two large, cross-sectional studies of OPP used an allometric model to investigate the relative contribution of morphological variables to the total variance in OPP. OPP was reported to be correlated with body mass but more strongly related to fat-free mass (FFM) in both sexes (Doré et al. 2000, 2001). There is a paucity of evidence to confirm these relationships as both cross-sectional and longitudinal studies have generally ‘corrected’ for morphological variables by simply dividing PP and MP with either body mass or FFM and expressing them as W kg⁻¹ (i.e. ratio scaling). It has been shown theoretically and empirically (Armstrong and Welsman 2019a; Tanner 1949; Welsman and Armstrong 2019) that ratio scaling has neither a sound scientific rationale nor statistical justification. Widespread and erroneous use of this scaling technique has clouded the interpretation of developmental exercise physiology and these analyses have confounded rather than clarified our understanding of the development of STP.

Running-related activities and sports are far more common in youth than cycling and, having different muscle recruitment and activation patterns, WAnT-determined PP and MP are unlikely to provide a valid assessment of STP during running-related activities. This is reflected by the low-to-moderate correlations (r = ~ 0.5–0.7) between WAnT external power indices and sport-related performances such as sprint running (Almuzaini 2000; Bencke et al. 2002; Hoffman et al. 2000).

The need for a laboratory running test of STP has been recognized since Margaria et al.'s (1966) seminal report of staircase running. Lakomy’s (1984, 1987) studies of elite adult athletes appear to have been the first to use a non-motorized treadmill (NMT) but it was Van Praagh et al. (Fargeas et al. 1993; Van Praagh et al. 1993) who initially explored the performance of elite youth athletes running on an NMT. They used 10 s sprints and published their research in abstract form only before stopping further development due to child safety concerns. Falk et al. (1996) explored the use of an NMT test (NMTT) to assess the PP of youth athletes over the number of ‘complete stride cycles nearest to 2.5 s’ and MP over 30 s. They observed that many participants had to quit earlier than 30 s or their performance became very irregular and deteriorated badly during the final seconds. They reported performance over 20 s rather than 30 s and do not appear to have persisted with development of an NMTT.

Sutton et al. (2000) developed a laboratory-based NMTT of STP suitable for longitudinal analyses of 8- to 18-year olds. The testing station was developed from Lakomy’s (1987) prototype with a strong emphasis on safety and a built-in adjustable safety harness. The safety harness was rarely used even with young children (on ~ 5% of occasions with 8-year olds) but enhanced the confidence of both children and adolescents who following habituation to the test were able to produce maximal performances over a full 30 s period (i.e. to mirror WAnT MP performance). On completion of the testing station, no methodological or safety problems were encountered in reliability studies and the typical error of NMTT-determined values 1 week apart was reported as 6% for PP and 5% for MP which compares very favourably with the test–retest reliability of the WAnT (Williams and Ratel 2017). The NMT testing station has subsequently been regularly and successfully used with 11- to 16-year olds to investigate, for example, repeated sprint ability (Oliver et al. 2006), recovery profiles following maximal cycling and running (Ratel et al. 2004), age-related recovery profiles (Ratel et al. 2006), and simulated sport-related performances (Oliver et al. 2007). The establishment of an NMTT to complement the WAnT enables exercise mode-specific (i.e. running or cycling) comparisons of the development of STP and peak \( \dot{V}{\text{O}}_{2} \).

Collectively studies which have focused on PP and MP in relation to age or a single morphological variable at specific ‘snapshot’ moments in time have provided limited insights into the development of STP. But the emergence (Aitkin et al. 1981) and regular refinement (Rasbash et al. 2018) of multilevel regression modelling has opened up new avenues of research in developmental exercise physiology. Multilevel allometric modelling enables the effects of variables such as age, body mass, FFM, and maturity status to be partitioned concurrently within an allometric framework to provide a flexible interpretation of the development of exercise performance variables.

Nevill et al. (1998) introduced multiplicative, allometric modelling to paediatric sport science with a re-analysis of previously published data from elite youth athletes and simultaneously with the present authors applied it to a prospective study of the development of 11- to 13-year olds’ peak \( \dot{V}{\text{O}}_{2} \) (Armstrong et al. 1999). There have been few subsequent publications of longitudinal data and applications of multiplicative allometric modelling to developmental exercise physiology have been remarkably sparse. There were no published longitudinal studies of the PP and MP of 11- to 16-year-old girls and boys before the present project and no studies had investigated the sex-specific influence of concurrent changes in age, maturity status, body mass, and FFM on the development of either STP or peak \( \dot{V}{\text{O}}_{2} \). The contribution of peak \( \dot{V}{\text{O}}_{2} \) to the development of MP assessed running or cycling had not been explored.

The primary purposes of the present study were, therefore, to enhance understanding of the development of STP in youth by (1) adopting a multiplicative, allometric modelling approach to investigate the sex-specific development of PP and MP from 11 to 16 years of age in relation to concurrent changes in age, maturity status, body mass, and FFM and, in the case of MP, peak \( \dot{V}{\text{O}}_{2} \); and (2) comparing the development of PP and MP from 11 to 16 years of age as assessed by maximal intensity running and cycling exercise. A subsidiary purpose was to compare and contrast the development of STP and peak \( \dot{V}{\text{O}}_{2} \) in the same young people from 11 to 16 years.

Descriptive data focusing on STP in relation to age or a single morphological variable provide limited insights but are briefly outlined herein for cross-study comparative purposes. Ergometer-specific data describing PP and MP in relation to age, body mass, and estimated FFM, respectively, are shown in Figs. 1, 2, 3 and 4. The descriptive data and correlations with age, body mass, and estimated FFM from the WAnT are in accord with those reported in cross-sectional studies (Van Praagh and Doré 2002; Williams and Ratel 2017) but there are no corresponding comparative data from NMTTs. All reported correlations are significant but the strongest correlations in both sexes are between STP and estimated FFM. On both ergometers and in both sexes, the relationships of PP and MP with age and estimated FFM are near-linear although it is notable that there are wider individual variations in WAnT-determined data, particularly in relation to age. In all cases, PP and MP increase with body mass with a tendency to begin levelling-off at ~ 60 kg in girls and ~ 70 kg in boys.

The baseline multiplicative allometric models 1.1–1.4 presented in Table 1 are each founded on 301 determinations of PP or MP and illustrate a similar picture of the development of STP. The positive age and negative sex terms in models 1.1 and 1.3 show that with body mass controlled for both WAnT- and NMTT-determined PP increases with age with the age effect smaller in girls. In models 1.2 and 1.4 describing the development of MP, an additional negative age by sex interaction demonstrates that on both ergometers with body mass controlled for MP increases with age at a greater rate in boys. Girls normally enter puberty before similarly aged boys (Malina 2017) and at the onset of the present study 56% and 23% of 11-year-old boys and girls, respectively, were pre-pubertal (i.e. at PH stage 1). The size of maturity status-driven changes in morphological and physiological variables related to performance is sex specific and it is, therefore, appropriate to analyse the development of boys’ and girls’ STP in sex-specific models rather than combining data as in Table 1.

The patterns of development of PP and MP are remarkably similar in boys and girls regardless of ergometer but the relative size of contribution of the morphological explanatory variables is sex specific. The baseline models (models 2.1, 3.1, 4.1, and 5.1) show that, with body mass controlled for, age exerts a significant additional effect on PP and MP in both sexes. The entry of maturity status to the STP baseline models had no additional effect on either PP or MP. This is in accord with the development of aerobic power in the same girls but in contrast with data from the same boys where PH stages 2–5 each made a significant positive contribution to explaining the development of peak \( \dot{V}{\text{O}}_{2} \) (Armstrong and Welsman 2019c).

The introduction of sum of skinfolds into models 2.2, 2.5, 3.2, 3.5, 4.2, 4.5, 5.2, and 5.5 resulted in significant negative exponents for skinfolds and increased body mass exponents. The marked increase in body mass exponents was also observed in corresponding models of peak \( \dot{V}{\text{O}}_{2} \) (Armstrong and Welsman 2019c) and it has been attributed to the effect that excess fat mass has on increasing body mass without an increase in the exercise performance variable (Vanderburgh et al. 1995). With the exception of the girls’ MP model (model 5.2) where the contribution of age was not significant, reduced but significant age terms were present in WAnT-determined STP models. In all NMTT-determined STP models, the age term was not significant. In all cases, the entry of skinfold thicknesses, which act with body mass as a surrogate for FFM, resulted in models with a significantly better statistical fit than those founded on age and body mass. Models 2.3, 2.6. 3.3, 3.6, 4.3, 4.6, 5.3, and 5.6 founded on FFM estimated from the youth-specific equations reported by Slaughter et al. (1988) reflect those of skinfold thicknesses plus body mass confirming the strong influence of FFM on PP and MP. The models including estimates of FFM are in accord with those describing the development of peak \( \dot{V}{\text{O}}_{2} \) in the same participants where additional effects of maturity status in boys were negated by the introduction to the baseline model of skinfold thicknesses or the replacement of body mass with estimated FFM (Armstrong and Welsman 2019c).

The multiplicative allometric models illustrate the powerful influence of FFM on both PP and MP. The influence of sex-specific maturity status on FFM is evidenced by reported changes in FFM being at their zenith around the time of peak height velocity (PHV). Boys’ FFM increases by ~ 83% over the period 2 years pre-PHV to 2 years post-PHV with girls’ FFM increasing ~ 31% over a 2-year period centred on PHV (Armstrong 2019; Baxter-Jones et al. 2003).

In both cycling and running, STP is primarily developed from the leg muscles and sex differences have generally been attributed to boys’ greater leg or thigh muscle volume (De Ste Croix et al. 2001: Santos et al. 2003; Welsman et al. 1997). On the other hand, it has been argued that as cycling and particularly running involve other muscles (e.g. trunk and arm muscles) FFM may be the most influential morphological variable in the development of STP (Doré et al. 2000, 2001). Certainly with the experimental challenges and costs of determining leg or thigh muscle volumes (Malina et al. 2004; Santos et al. 2003; Winsley et al. 2003), FFM estimated either from body mass and skinfolds directly or calculated from youth-specific equations presents a pragmatic morphological variable with which to scale and compare STP in youth.

FFM is not the only maturity status-driven variable contributing to the development of PP and MP. Other factors intrinsic to muscle but not investigated herein include changes in muscle structure, muscle fibre type, muscle activation (i.e. the decline in activation deficit), and muscle metabolism. These factors have been reviewed elsewhere in relation to the ethical and methodological challenges of exploring their effects on exercise performance but empirical evidence is sparse (Armstrong et al. 2017; Barker and Armstrong 2010; Dotan et al. 2012; Malina et al. 2004).

The morphological variables influencing the development of MP are similar to those describing PP but, unlike PP, MP represents an interplay of anaerobic and aerobic metabolism. MP is an important component of performance in many youth sports (Armstrong 2019) but few studies have investigated the contribution of \( \dot{V}{\text{O}}_{2} \) to MP. It has been demonstrated empirically that the pulmonary \( \dot{V}{\text{O}}_{2} \) kinetics response at the beginning of high intensity exercise slows from 11 to 16 years (Breese et al. 2010; Fawkner and Armstrong 2004) so the anaerobic contribution to MP would be expected to increase with age. The net \( \dot{V}{\text{O}}_{2} \) during a 30 s WAnT has been reported to be 45% higher in 10-year-old boys than 22-year-old men (Hebestreit et al. 1993). It has been suggested that the aerobic contribution to cycling MP lies within the range 16–45% in 9- to 12-year olds (Chia 2006; Chia et al. 1997) but variations in the mechanical efficiency of cycling with age make it difficult to be more precise. Chia (2006) estimated that 10- to 12-year-old boys and girls attain 67% and 73% of peak \( \dot{V}{\text{O}}_{2} \), respectively, during WAnTs. Phase 2 pulmonary \( \dot{V}{\text{O}}_{2} \) kinetics time constants from 11- to 16-year olds reported from a number of sources (tabulated by Barker and Armstrong 2017) lend support to these estimates by indicating that values of ~ 70–75% of peak \( \dot{V}{\text{O}}_{2} \) are likely to be approached within 30 s of high intensity cycling or running.

As illustrated in Table 6, when ergometer-specific peak \( \dot{V}{\text{O}}_{2} \) was entered it made a significant contribution additional to that of either body mass and skinfold thicknesses (as a surrogate of FFM) or estimated FFM in explaining the development of MP in both sexes, on both ergometers. The final models (models 6.1–6.4), including peak \( \dot{V}{\text{O}}_{2} \) and either sum of skinfold thicknesses and body mass (boys) or estimated FFM (girls) and in the case of boys’ WAnT-determined MP age, present the best statistical fit models of the development of MP. The influence of peak \( \dot{V}{\text{O}}_{2} \) on cycling and running MP had not previously been investigated in youth but it is clear that it makes a significant contribution to the development of MP in both sexes, on both ergometers.

The data demonstrate that, (1) in accord with the development of peak \( \dot{V}{\text{O}}_{2} \), maturity status-driven FFM is the most powerful morphological influence on the development of PP and MP in 11- to 16-year olds, on both ergometers in both sexes; (2) with FFM controlled for peak \( \dot{V}{\text{O}}_{2} \) makes an additional contribution to explaining the development of MP on both ergometers in both sexes; (3) an NMTT provides an appropriate model for investigating running STP in youth and offers a complementary test to the established WAnT; and (4) the development of STP is sex specific but within sex multiplicative allometric models of running- and cycling-determined PP and MP were similar, suggesting that either mode of exercise can be used in future studies of STP in youth.