Young people’s aerobic performance has been extensively researched and documented (2), but, despite the fact that children’s and adolescents’ physical activity patterns are characterized by short-burst, predominantly anaerobic activities, the growth and maturation of anaerobic performance remains poorly understood (3,29). Data describing the anaerobic characteristics of girls are sparse. This may, in part, be attributed to the lack of valid testing protocols with which to assess young people’s anaerobic performance (4,9). Direct measurement of children’s and adolescents’ anaerobic metabolism during exercise is not presently possible and research efforts have focused on the measurement of short-term power output. The Wingate Anaerobic Test (WAnT) (20) which allows the determination of peak power (PP), usually over a 1s or 5s period, and mean power (MP) over the 30-s test period, has emerged as the most popular test of young people’s short-term power.
There are several methodological variants of the WAnT and this confounds direct interstudy comparisons. Intrastudy comparisons of boys’ and girls’ performance on the WAnT are limited by small sample sizes. Analyses in relation to body mass are restricted to the simple division of PP or MP by body mass (the ratio standard) despite the well-documented limitations of this methodology in the interpretation of exercise variables during growth (24,33). The available evidence on gender differences is focused on 10- to 12-yr-olds and is not consistent. Studies have reported higher PP and MP in girls (12), higher MP in boys (14) and no gender differences in PP (14) or MP (30). Data reporting power output in ratio with body mass (i.e., W·kg−1) are conflicting with PP and MP reported to be significantly higher (14) and lower (12) in boys than in girls whereas another study found no significant gender difference in MP (30).
Few data on the influence of maturation on PP and MP are available and an appropriate statistical analysis of longitudinal data has not been published. The only available longitudinal study of young people’s short-term power reported the PP and MP of 36 boys measured four times over an 18-month period (16). The boys were classified into maturational stages according to Tanner’s indices for pubic hair (28), but because of the small sample size, the subjects were grouped as prepubertal (Tanner stage 1), midpubertal (Tanner stages 2, 3, and 4), and late pubertal (Tanner stage 5). No change in mass-related MP (W·kg−1) with maturation was apparent, but mass-related PP appeared to markedly increase from prepuberty to midpuberty and from midpuberty to late puberty. However, as the authors pointed out, the experimental design precluded statistical analysis.
The present study was designed to clarify gender differences and to explore maturational influences on short-term power output. Initially, a large representative sample (N = 200) of 12-yr-olds was recruited and a cross-sectional study performed (5). Body mass was controlled for using log-linear ANCOVA and no significant gender differences in PP or MP were detected whether the scores were adjusted for body mass or not. By using Tanner’s indices (stages 1–4), a main effect for maturation independent of body mass was demonstrated. In contrast to an earlier study that did not appropriately control for body mass (15), salivary testosterone levels in boys were found to make no additional contribution to the explained variance in either PP or MP once the contribution of body mass had been statistically removed. These young people were retested 1 yr later to clarify further our understanding of the factors influencing young people’s PP and MP. The purpose of this study is therefore to use multilevel modelling of these longitudinal data to examine the effects of gender, age, body mass, stature, skin-fold thicknesses, and maturity on short-term power output.
All of the children in year 6 (age 11.1 ± 0.4 yr) of the 15 state schools in the city of Exeter (United Kingdom) were invited to participate in a longitudinal study of physical activity patterns, physiological responses to exercise, and body composition. A total of 747 children (70% of those eligible) volunteered, and written consent was obtained from both the children and their parents/guardians. The study received ethical approval from the Exeter District Health Authority Ethical Committee. In an attempt to detect sample bias, the stature and body mass of the volunteers were compared with the stature and body mass of those who declined to participate. No significant difference (P > 0.05) was detected in either gender. Twenty-five percent of the eligible children in each school were randomly selected from those who volunteered. WAnT data from year 2 of the study have previously been published (5). The data reported here are from the participants who satisfactorily completed the WAnT and associated measures during both the second (boys, N = 97; girls, N = 100) and third (boys, N = 95; girls, N = 80) years of the study.
Age was computed from date of birth and date of examination. Stature, body mass, and skin-fold thickness over the triceps and subscapular regions were measured according to the techniques described by Weiner and Lourie (31). Sexual maturity was visually assessed using Tanner’s indices for pubic hair development (28). The same experienced nurse completed the maturity assessments on both occasions.
The children had visited the laboratory on several occasions and were habituated to both the general environment and the experimental procedures. The WAnT was conducted on a friction-loaded cycle ergometer (Monark 814E Monark-Crescent AB, Varberg, Sweden) interfaced with a microcomputer. The same ergometer was used for all tests. The seat height and handlebars were adjusted appropriately for each subject, and the resistance was set at 0.075 kg (0.74N)·kg−1 body mass (8). After a standardized 3-min warm-up involving pedalling at 60 rpm interspersed with three 2- to 3-s all-out sprints, the subject rested on the ergometer. The WAnT commenced from a rolling start, at 60 rpm against minimal resistance (weight basket supported). When a constant pedal rate of 60 rpm was achieved, a countdown of “3–2-1-go” was given, and the test resistance was applied and the computer activated. Subjects were verbally encouraged throughout the test, and the power output was calculated each second for the duration of the test. The PP over 1 s and the MP over the 30-s period were recorded. Immediately after the test, subjects completed a cool-down consisting of continuous light pedalling against minimal resistance for 3 min (19).
Duplicate fingertip blood samples were taken 3 min after the WAnT (30) and immediately assayed for lactate concentration using a YSI 2300 Stat Plus whole blood analyzer (Clandon Scientific, Farnborough, UK). The analyzer self-calibrated with a known concentration of lactate every five samples, and the calibration was checked regularly against commercially prepared standards of verified concentration.
Descriptive statistics (means and standard deviations) for anthropometric variables, peak and mean power, and postexercise blood lactate were computed by gender for each year.
Factors associated with the longitudinal development of WanT performance (gender and maturity) adjusted for differences in anthropometric measures and age were investigated using the multilevel modelling program MLwiN (17). Multilevel modelling is essentially an extension of multiple regression and is appropriate for analyzing hierarchically structured or nested data. In longitudinal data sets, the hierarchy can be defined as level 1 units—the repeated measurement occasions, grouped within the level 2 unit—the individual subject.
Multilevel modelling is preferable to traditional analytical approaches for longitudinal data as, in addition to describing the population mean response, this method recognizes and describes variation around the mean at both levels; e.g., at level 2, individuals have their own growth rates that vary randomly around the underlying population response, and, at level 1, each individual’s observed measurements may vary around their own growth trajectory. Furthermore, unlike traditional methods based upon repeated measures analysis of variance, which require a complete longitudinal data set, this method is able to handle unbalanced data, for example, where one or more measurement occasion has been missed. Similarly, as individual growth trajectories can be modelled, differing intervals between measurement occasions can be accommodated.
In this study a multiplicative, allometric approach was adopted based upon the model proposed by Nevill and associates (23,25) as follows:MATH 1 where all parameters were fixed with the exception of the constant (intercept term) and age parameters, which were allowed to vary randomly at level 2 (between individuals), and the multiplicative error ratio ε, which varied randomly at level 1 describing the error variance between occasions. The subscripts i and j denote random variation at levels 1 and 2, respectively. The variable “age” was centred around the group mean age of 12.7 yr.
This model can be linearized by logarithmic transformation and multilevel regression analysis on loge y used to solve for the unknown parameters. Once transformed, the equation above becomes:MATH 2 From this baseline model, additional explanatory variables were investigated including sum of triceps and subscapular skin-fold thicknesses, gender, and stage of maturity (stages 2–5 for pubic hair development). The latter two variables were incorporated into the model as indicator variables (e.g., for gender, boys = 0, girls = 1), which sets the boys’ constant as the baseline from which the girls’ parameter is allowed to deviate. An interaction term age·gender was also constructed to investigate differential growth in boys and girls.
As demonstrated for cross-sectional data (26,33), this multiplicative, allometric modelling approach has been shown to be theoretically and statistically superior for longitudinal analyses (25) to the alternative additive, polynomial model (10), as it accommodates the skewness and heteroscedasticity that often characterize size-related exercise performance data (24,33).
Descriptive anthropometric, WanT performance and postexercise blood lactate data are presented in Table 1. Values for peak blood lactate were almost identical in year 1 and 2, with no observable difference between boys and girls.
Table 2 summarizes the results of the multilevel regression analysis for PP. Initially, data from the subgroup of subjects with maturity measures (N = 327) were modelled, but as all maturity parameters were not significant (P > 0.05), the models presented in Table 2 were generated using the entire subject population (N = 372). In model 1, mass and stature proved to be significant explanatory variables with exponents of 0.889 (SE 0.055) and 0.856 (SE 0.189), respectively. There was an additional positive effect of age that was similar for boys and girls as no significant age-by-gender interaction term was identified. Girls’ PP was lower than that of boys (denoted by the parameter for gender, −0.057 (SE 0.015), which was deducted for girls’ predicted scores). These fixed parameter estimates describe the subject population mean response while the random parameters reflect individual departures from this mean response at both levels of the analysis, i.e., between individuals (level 2) and within individuals (level 1). Within model 1 at level 2, significant variation in both constant (intercept) and slope (age) parameters was evident reflecting individual departures from the mean response in the magnitude and rate of growth of PP over the 2-yr period. There was a significant negative covariance (−0.004 SE 0.001) between the level 2 random age/constant parameters indicating a smaller rate of increase in PP for higher initial values of PP.
From the baseline model 1, sum of skin-fold thicknesses was investigated as an additional explanatory variable and the results are summarized in Table 2, model 2. The addition of skin folds yielded a significant negative coefficient and rendered the stature term nonsignificant. Mass remained a significant covariate but the value of the coefficient increased considerably above that in model 1 to 1.203 (SE 0.051). The replacement of stature with skin-fold thickness improved the fit of the model as reflected by the significant change (−466.755 to −476.505) in the deviance statistic (−2 * loglikelihood) for the same number of parameters. The random parameter estimates remained essentially unchanged.
Multilevel models for MP are summarized in Table 3. As for PP, mass, stature, and age were initially entered as explanatory variables (see model 1) and yielded significant exponents of 0.547 (SE 0.056), 1.357 (SE 0.197), and 0.146 (SE 0.027), respectively. Girls’ MP was significantly lower than boys’ and, in contrast to PP, a significant, negative age by gender interaction term was identified, reflecting a smaller increase in MP with age in girls than boys. An additional positive effect of maturity on MP, over and above increases due to stature, mass, and age, was identified for the two later stages (4 and 5) of pubic hair development.
The random parameters for MP showed similar findings to those for PP again with a significant negative covariance term (−0.004 SE 0.001) at level 2, indicating that the higher the initial MP the slower the rate of increase with age. The addition of skin folds to this baseline model (model 2) again rendered the stature coefficient nonsignificant and negated the maturity effect but proved a better fit for the data as indicated by the change in the deviance statistic for fewer fitted parameters. In view of the absence of a maturity effect, this model was recomputed on the entire sample (model 3, N = 372). This yielded similar parameter estimates to model 2, but in this larger sample, the random variation association with the age term was not significant.
The analysis of longitudinal exercise performance data presents a formidable challenge to the researcher, in particular the interpretation of data in relation to changes in body size and composition. There is now extensive evidence within the pediatric exercise science literature to demonstrate that simply dividing by body mass is too simplistic a method which often fails to produce a “size-free” variable (33,34). Allometric analyses have provided insights into interpretation of cross-sectional aerobic and anaerobic performance data (1,5,6) but are less easily applied to longitudinal data sets. One approach is to fit ontogenetic allometric models whereby a mass exponent is calculated for each subject separately (11,18,27). Subsequently, a single mean mass exponent may be calculated to describe different groups, e.g., by gender. However, this two-stage approach is statistically inefficient as intermediate statistics from the individual analyses (the slope and intercept parameters) can only be accommodated partially in a subsequent between group analysis (25). The multiplicative, allometric multilevel regression modelling approach employed in the present analysis overcomes these limitations by allowing between group effects to be evaluated while simultaneously describing individual departures from the mean group response and has provided insights into the growth and maturation of aerobic power and isometric strength in young people (7,25).
The development of anaerobic performance during childhood and adolescence determined using the WAnT is not well defined. Gender differences during this period are particularly difficult to elucidate due to the lack of research data, particularly for girls during the teenage years (3). Direct comparisons of power output results between studies are confounded by methodological variations in the performance of the WAnT, but the present absolute power output data presented in Table 1 are in general agreement with values reported by other investigators for subjects of similar age (12,14,19). In accord with the limited available data no significant gender differences in postexercise blood lactate levels were detected (32).
Our initial cross-sectional study examined gender differences in, and maturational effects upon, WAnT performance in 200 12-yr-old girls and boys (5). In agreement with indications from an earlier mixed longitudinal study of boys (16), this study identified significant main effects for maturity upon WAnT over and above increases due to increases in body mass. Although no overall gender difference was demonstrated for either PP or MP once the influence of body mass had been partitioned out using log linear ANCOVA, at each stage of maturity (stage 1–4 for pubic hair development), boys attained higher power outputs than girls with the gender difference increasing with greater maturity (5). The present study reports age, gender, and maturity effects on WAnT determined PP and MP in these same subjects tested 1 yr later.
The multilevel models presented in Tables 2 and 3 were consistent in demonstrating gender differences in PP and MP, with boys demonstrating higher power outputs regardless of which other covariates had been controlled for. In contrast to previous indications (5), this was independent of any maturity effect. When age, mass, and stature were included as explanatory variables a significant, incremental effect of early maturity was observed, but, again in contrast to previous work (5,16), this was only observed for MP. The finding of stature as a significant independent predictor of anaerobic performance echoes the results of previous studies of young people’s performance measures such as aerobic power and strength (25.34). The precise meaning of a significant stature effect has not been determined, but it has been hypothesized that stature acts as surrogate for a disproportionate increase in leg muscle volume relative to increasing body mass (21)—a characteristic which has been demonstrated in adolescent male subjects (22). However, an interesting feature of the multilevel models explored here is that the addition of sum of skin-fold thicknesses as an explanatory variable rendered the stature term nonsignificant for both PP and MP and also explained the apparent maturity effects observed in the initial modelling of MP. The addition of skin folds also substantially altered the observed exponent for body mass, with values increasing to 1.2 and 1.1 for PP and MP, respectively. This highlights the sensitivity of the mass exponent to the effect of concurrent covariates and indicates that at the mean age of 12.7 yr WAnT determined power output is increasing faster than the increase in body mass in both boys and girls. This reflects trends observed in compiled data around this age range (3).
A significant effect of age over and above that due to the other covariates was also observed for both power measures. However, in contrast to indications from the cross sectional data (5), the significant age by gender interaction term for MP demonstrates a relatively smaller increase in MP for girls over the period of the study. The extent to which this reflects a true physiological difference in anaerobic function remains to be resolved as there are reasons to suspect that the WAnT methodology itself might be expected to accentuate gender differences, particularly for MP. The recommended resistance for use with a Monark cycle ergometer is calculated as a simple per body mass ratio (0.74 N·kg−1) and as such is unable to accommodate differences in body fatness. Even in predominantly prepubertal children where overall body composition differences are not exaggerated, girls have been shown to be exercising against a higher resistance than boys when the resistance is expressed relative to their magnetic resonance imaging (MRI) -determined thigh muscle volume (35). Thus, body compositional changes during puberty are likely to emphasize this effect with girls progressively penalized as body fatness increases and the muscle mass to body mass ratio decreases. This supposition is supported by the significant negative effect of skin-fold thicknesses observed in the multilevel models with a greater negative impact observed for the longer duration MP compared with PP.
A physiological explanation for the remaining age effect is difficult to establish within the present data set. However, initial indications from a subsequent and continuing longitudinal study of WAnT performance in relation to anthropometry, isokinetic strength, and MRI-determined muscle volume suggest that muscle volume may be a significant additional explanatory variable for short term power output between the ages of 10 and 12 yr (13).
Previous analyses of anaerobic performance have controlled for body size effects by removing the influence of a single body size variable—body mass—either by computing per body mass ratios (15) or by using more appropriate allometric techniques (5). The multilevel modelling technique used in the present analysis has enabled a more sensitive investigation of the influence of various covariates upon short-term power.
In summary, this study has demonstrated gender differences in the longitudinal growth of WAnT determined PP and MP between the ages of 12 and 13 yr. Although both power indices increased in both boys and girls, the rate of increase in MP was lower in girls. Maturational differences were initially observed only for MP. Stature effects for both power measures and the maturity effect for MP were negated when the sum of two skin-fold thicknesses was introduced as an explanatory variable. Thus, body mass and skin-fold thicknesses appear to be the best anthropometric predictors of WAnT performance in this age group but gender differences in power indices may be an artifact of a per body mass computed resistance. Further work, perhaps using running rather than cycling models, is required to examine whether gender differences in WAnT performance are a true reflection of gender differences in anaerobic capabilities during growth.
We gratefully acknowledge the technical assistance of Jenny Frost, Alison Husband, and Sue Vooght. The work was supported by the British Heart Foundation and the Healthy Heart Research Trust.
1. Armstrong, N., B. Kirby, A. McManus, and J. Welsman. Aerobic fitness of pre-pubescent children. Ann. Hum. Biol. 22:427–441, 1995.
2. Armstrong, N., and J. R. Welsman. Assessment and interpretation of aerobic fitness in children and adolescents. Exerc. Sport. Sci. Rev. 22:435–476, 1994.
3. Armstrong, N., and J. R. Welsman. Young People and Physical Activity. Oxford: Oxford University Press, 1997, pp. 79–96.
4. Armstrong, N., and J. R. Welsman. Anaerobic performance. In Paediatric Exercise Science and Medicine, N. Armstrong and W. Van Mechelen (Eds.). Oxford: Oxford University Press, 2000, pp. 37–45.
5. Armstrong, N., J. R. Welsman, and B. J. Kirby. Performance on the Wingate Anaerobic Test and maturation. Pediatr. Exerc. Sci. 9:253–261, 1997.
6. Armstrong, N., J. R. Welsman, and B. J. Kirby. Peak oxygen uptake and maturation in 12-year-olds. Med. Sci. Sports. Exerc. 30:165–69, 1998.
7. Armstrong, N., J. R. Welsman, A. M. Nevill, and B. J. Kirby. Modeling growth and maturation changes in peak oxygen uptake in 11–13 yr olds. J. Appl. Physiol. 87:2230–2236, 1999.
8. Bar-Or, O. Pediatric Sports Medicine for the Practitioner. New York: Springer Verlag, 1983, pp. 323–325.
9. Bar-Or, O. Anaerobic performance. In: Measurement in Pediatric Exercise Science, D. Docherty (Ed.). Champaign, IL: Human Kinetics, 1996, pp. 161–182.
10. Baxter-Jones, A., H. Goldstein, and P. Helms. The development of aerobic power in young athletes. J. Appl. Physiol. 75:1160–1167, 1993.
11. Beunen, G. P., D. M. Rogers, B. Woynarowska, and R. M. Malina. Longitudinal study of ontogenetic allometry of oxygen uptake in boys and girls grouped by maturity status. Ann. Hum. Biol. 24:33–43, 1997.
12. Carlson, J. S., and G. A. Naughton. Performance characteristics of children using various braking resistances on the Wingate anaerobic test. J. Sports. Med. Phys. Fitness 34:362–369, 1994.
13. Chia, M. Y. H., N. Armstrong, M. B. A. De Ste Croix, and J. R. Welsman. Longitudinal changes in Wingate Anaerobic Test determined peak and mean power in 10 to 12 year olds. Pediatr. Exerc. Sci. 11:272, 1999.
14. Docherty, D., and C. A. Gaul. Relationship of body size, physique and composition to physical performance in young boys and girls. Int. J. Sports. Med. 12:525–532, 1991.
15. Falgairette, G., M. Bedu, N. Fellmann, E. Van-Praagh, and J. Coudert. Bioenergetic profile in 144 boys age from 6 to 15 years with special reference to sexual maturation. Eur. J. Appl. Physiol. 62:151–156, 1991.
16. Falk, B., and O. Bar-Or. Longitudinal changes in peak aerobic and anaerobic mechanical power of circumpubertal boys. Pediatr. Exerc. Sci. 5:318–331, 1993.
17. Goldstein, H., J. Rasbash, I. Plewis, et al. A User’s Guide to MLwiN
. London: University of London, Institute of Education, 1998, pp. 1–140.
18. Gould, S. J. Allometry and size in ontogeny and phylogeny. Biol. Rev. 41:587–640, 1966.
19. Inbar, O., and O. Bar-Or. Anaerobic characteristics in male children and adolescents. Med. Sci. Sports Exerc. 18:264–269, 1986.
20. Inbar, O., O. Bar-Or, and J. S. Skinner. The Wingate Anaerobic Test. Champaign, IL: Human Kinetics, 1996, pp. 1–110.
21. Nevill, A. The need to scale for differences in body size and mass: an explanation of Kleiber’s 0.75 mass exponent. J. Appl. Physiol. 77:2870–2873, 1994.
22. Nevill, A. M. Evidence of an increasing proportion of leg muscle mass to body mass in male adolescents and its implication on performance. J. Sports Sci. 12:163–164, 1994.
23. Nevill, A. M., and R. L. Holder. Modelling maximum oxygen uptake: a case study in non-linear regression formulation and comparison. J. Rl. Stat. Soc. Series C 43:653–666, 1994.
24. Nevill, A. M., and R. L. Holder. Scaling, normalizing and per ratio standards: an allometric modeling approach. J. Appl. Physiol. 79:1027–1031, 1995.
25. Nevill, A. M., R. L. Holder, A. Baxter-Jones, J. Round, and D. S. Jones. Modeling developmental changes in strength and aerobic power in children. J. Appl. Physiol. 84:963–970, 1998.
26. Nevill, A., R. Ramsbottom, and C. Williams. Scaling physiological measurements for individuals of different body size. Eur. J. Appl. Physiol. 65:110–117, 1992.
27. Rowland, T. W., P. Vanderburgh, and L. Cunningham. Body size and the growth of maximal aerobic power in children: a longitudinal analysis. Pediatr. Exerc. Sci. 9:262–274, 1997.
28. Tanner, J. M. Growth at Adolescence
(2nd Ed.). Oxford: Blackwell Scientific Publications, 1962, pp. 28–39.
29. Van Praagh, E. (Ed.). Pediatric Anaerobic Performance.
Champaign, IL: Human Kinetics, 1998, pp. 45–64.
30. Van Praagh, E., N. Fellmann, M. Bedu, C. Falgairette, G. Coudert, and J. Gender. Difference in the relationship of anaerobic power output to body composition in children. Pediatr. Exerc. Sci. 2:336–348, 1990.
31. Weiner, J. S., and J. A. Lourie. (Eds.) Practical Human Biology
. London. Academic Press, 1981, pp. 27–52.
32. Welsman, J. R., and N. Armstrong. Assessing postexercise lactates in children and adolescents. In. Pediatric Anaerobic Performance
, E. Van Praagh (Ed.). Champaign, IL: Human Kinetics, 1998, pp. 137–153.
33. Welsman, J. R., and N. Armstrong. Interpreting exercise performance data in relation to body size. In: Paediatric Exercise Science and Medicine, N. Armstrong and W. Van Mechelen (Eds.). Oxford: Oxford University Press, 2000, pp. 3–9.
34. Welsman, J., N. Armstrong, B. J. Kirby, A. M. Nevill, and E. Winter. Scaling peak oxygen uptake for differences in body size. Med. Sci. Sports. Exerc. 28:259–265, 1996.
35. Welsman, J. R., N. Armstrong, B. J. Kirby, R. J. Winsley, G. Parsons, and P. Sharpe. Exercise performance and MRI determined muscle volume in children. Eur. J. Appl. Physiol. 76:92–97, 1997.