Aerobic fitness relates to the ability of the circulatory and pulmonary systems to take in, transport, and use oxygen during sustained physical activity (9). In young people, the best single measure of aerobic fitness is peak oxygen uptake (V˙O2peak) (6,22). V˙O2peak is the highest rate at which oxygen can be consumed during exercise.
Body size strongly correlates with V˙O2peak. Appropriate scaling will therefore underpin the interpretation of V˙O2peak data in children. Perhaps the most robust method to partition out the influence of body size is to scale V˙O2peak allometrically to a direct quantification of the involved muscle mass (25).
Tolfrey et al. (25) reported that allometric scaling by body mass failed to account correctly for body size in 15 boys (12.3 ± 0.3 yr) and was an invalid allometric scaling denominator for V˙O2peak. Conversely, estimated lower leg muscle volume, derived from magnetic resonance imaging, was an appropriate scaling variable to partition out the influence of body size (25). These results confirmed findings from previous studies that right thigh muscle volume (28) and calf and thigh muscle girths (20) are more appropriate body size dimensions compared with body mass for scaling V˙O2peak in children and V˙O2max in adults, respectively.
Although calf and thigh muscle girths and upper or lower leg muscle volumes are more appropriate than body mass to scale V˙O2peak, they do not provide a direct measure of the total involved leg musculature during physiological assessment. Dual-energy x-ray absorptiometry (DXA) can quantify lean mass of the legs. To our knowledge, no study has examined the appropriateness of scaling V˙O2peak in children to the lean mass of both legs, as quantified using DXA. Further, no study has examined the appropriateness of scaling V˙O2peak to the lean mass of both legs compared with total lean mass and whole body mass in a large, heterogeneous sample.
Independent of body size and body composition, maturation is associated with increased V˙O2peak in young people (4,5). Controlling for maturation and body size might therefore improve the interpretation of sex differences and longitudinal changes in aerobic fitness in pediatric populations. However, it is unknown whether maturation exerts an independent influence on V˙O2peak after controlling for lean mass.
The aim of this study therefore was to compare three candidate body size index models for the scaling of V˙O2peak in 9- to 11-yr-old children: whole body mass, total lean body mass, and lean body mass of both legs. The independent influence of maturation and sex on V˙O2peak after controlling for body size was also investigated.
Participants and settings
Participants were randomly selected from the SportsLinx database (7). The SportsLinx project (24) has annually collected anthropometric and physical fitness data in 9- to 10-yr-old Liverpool children since 1998. One hundred and thirty-six girls and 103 boys age 9–11 yr were recruited. Participants and their parents/guardians gave written informed consent and assent, respectively. The study was approved by the university ethics committee. Parents/guardians completed medical questionnaires on behalf of their child. Children were eligible if they were free from the presence of chronic disease and metabolic disorders.
Experimental design and procedures
Participants visited the laboratories on one occasion to complete several physiological examinations including anthropometrics, V˙O2peak testing and body composition scan. All measurements occurred in thermoneutral conditions.
Anthropometry and maturation assessment
Stature and sitting stature were measured to the nearest 0.1 cm using a Leicester Height Measure (Seca Ltd., Birmingham, UK) and body mass to the nearest 0.1 kg using a calibrated mechanical flat Seca scale (Seca Ltd.) with participants wearing light clothing and without shoes (17). Body mass index (BMI) was calculated as mass divided by the square of stature (kg·m−2). Assessments were taken by a level 1 ISAK anthropometrist. To estimate maturity status, stature, sitting stature, leg length, body mass, chronological age and their interactions were used in sex-specific equations to predict each participant’s maturation offset (years from age of peak height velocity [PHV]) (19).
Measurement of aerobic fitness (V˙O2peak)
V˙O2peak was assessed during a discontinuous incremental treadmill test (HP Cosmos, Traunstein, Germany) (15). After practicing walking (4 km·h−1) and running (8 km·h−1) on the treadmill until comfortable with the activity, participants walked at 4 and 6 km·h−1 and then ran at 8, 10, 12, and 14 km·h−1 to the point of volitional exhaustion. The test consisted of 3-min stages followed by a 30-s rest interval. Treadmill gradient remained at 1.0% throughout. A pediatric facemask (Hans Rudolph, Kansas City, KS) covering the nose and mouth was secured via an adjustable nylon harness. During the test, V˙O2 and carbon dioxide production (V˙CO2) were assessed breath by breath by an online gas analysis system (Jaeger Oxycon Pro; Viasys Health Care, Warwick, UK). HR was monitored continuously (Polar, Kempele, Finland). Pulmonary variables were expressed as mean values for 15-s epochs. The V˙O2peak was accepted as the highest 15-s V˙O2 value when participants exhibited subjective indicators of peak effort (unsteady gait, hyperpnea, facial flushing, and sweating) that were confirmed by an RER ≥ 1.00 and/or HR ≥ 195 beats per minute (3). Before and after testing on all test days, vanes were calibrated using known volumes of flow rate (0.2 and 2.0 L·s−1) and the gas analyzers against known gas concentrations (0.5% CO2 and 20.5% O2).
Measurement of body composition
A whole body DXA (Hologic QDR series Discovery A, Bedford, MA) scan assessed absolute fat mass and lean mass (excluding bone mass) and percentage body fat. Total lean mass of both legs was determined by segmental analysis (left + right leg lean mass). DXA assessment of body composition has been validated against hydrodensitometry (26) and recommended as a measure of fat and lean mass in children (16). Participants were scanned in the supine position while wearing a T-shirt and a pair of shorts. All scans were performed and analyzed by the same qualified researcher. The DXA scanner was calibrated daily using the lumbar spine and step phantom. Scans were analyzed after each assessment using Hologic QDR software for Windows version 11.2 (©1986–2001 Hologic Inc.) and stored in secure data files.
Statistical analyses and modeling
All analyses were performed using SPSS (version 17.0; IBM SPSS Inc., Chicago, IL). We used allometric regression models of the form V˙O2peak = a×body size indexb adjusted for biological sex and maturity offset (years from PHV). Maturity offset and sex were included as covariates, as they contributed substantially to the model even after controlling for body size (however measured). All analyses were conducted using nonlinear models in the arithmetic space (1,21). A common slope was fitted for boys and girls, as preliminary analyses revealed no substantial sex by body size interaction. Three candidate models were compared, each using a different index of body size: whole body mass, total lean body mass, and lean mass of both legs. We assessed goodness of fit using the Akaike information criterion (AIC) with AIC differences, Akaike weights, and evidence ratios were computed from the derived AIC values for each model, according to methods detailed by Burnham and Anderson (8). The AIC is not a null hypothesis test but rather a measure of the relative goodness of fit of a candidate model. The AIC difference for a particular body size index model was derived as the observed AIC minus the AIC for the best model (the model with the lowest AIC value). Hence, the AIC difference for the model estimated to be the best of the candidate set is zero. We adopted the following rules of thumb for AIC difference from the estimated best model: 0–2, substantial support for the model; 4–7, considerably less support; > 10, essentially no support. Akaike weights permit inference about the relative likelihood of competing models and provide the probability that a particular model is the best from the set of candidate models (these probabilities sum to one for the set of three models). Evidence ratios are simply the ratio of Akaike weights and indicate the relative likelihood of one model being better than another.
One hundred and twenty-six girls and 87 boys who provided acceptable V˙O2peak scores were included in analyses. Anthropometric, body composition, and aerobic fitness data are presented in Table 1.
Body mass and body fat classifications for the group and by sex are reported in Table 2. The prevalence of overweight and obese was 36.4% based on the BMI classifications of Chinn and Rona (10). The prevalence of overfat and obese was 52.2% based on total body fat classifications using the cutoff points of McCarthy et al. (18). Both classification methods indicated that the number of overweight or overfat and obese participants were greater in girls than boys. The sample was heterogeneous for body mass (21.9–73.0 kg), total lean body mass (16.6–44.2 kg), and total body fat percentage (13.3%–45.3%).
Scaling methods: comparison of the three candidate models
Table 3 reveals the results of the AIC analyses for each model. The model estimated to be the best is that involving the lean mass of both legs. With an AIC difference substantially >10, the whole body mass model has essentially no support. The total lean body mass model has substantial support, with an AIC difference from the best model of <2. The Akaike weights reveal that if, theoretically, we could draw multiple independent samples and repeat the analyses, then the lean leg mass model would emerge as the best 69 of 100 times, versus 31 of 100 times for the lean body mass model. The evidence ratio for the lean leg mass model versus the total lean body mass model is 2.2 (the lean leg mass model is twice as likely to be the best), which is relatively weak support for the best model.
The size exponent (b) for the lean mass of both legs was 0.55 (90% confidence interval = 0.46–0.64). Size independence of the best model was confirmed by an essentially zero correlation (r = 0.001) between the allometric model residuals and the lean mass of both legs (with respect to size-correcting techniques, the model residuals represent the V˙O2peak after having controlled for the effect of the body size variable, sex, and maturity offset).
Sex and maturity offset contributed substantially to the model. After controlling for the lean mass of both legs and maturity offset, V˙O2peak in girls was 17% lower than that in boys (13%–21%). After controlling for body size and sex, a 1-yr increase in maturity offset (closer to PHV) was associated with a 6% higher V˙O2peak (4%–9%).
This study investigated whether lean mass of both legs, total body mass, or total lean body mass was the most appropriate scaling variable for V˙O2peak in 9- to 11-yr-old children. The main finding in this study of 213 children heterogeneous for body size and composition was that the regression model using the lean mass of both legs was the best of the three candidate models. The large AIC difference for the model using total body mass as the indicator of body size revealed that this model has essentially no support. The model using total lean body mass had substantial support, with a small AIC difference compared with the best model. These results imply that the models involving either the lean mass of both legs or the total lean body mass are substantially superior to the model involving total body mass.
The second key finding was that maturation exerted an independent influence on V˙O2peak after controlling for lean mass of both legs. Previous studies that controlled for body size and composition using body mass and fat-free mass revealed similar findings (4,5), although the present study extends the knowledge base by using the superior method of DXA-derived lean mass to quantify anthropometric scaling variables. This implies that maturation should be controlled for when sex-based differences in physiological and performance measures or longitudinal changes in aerobic fitness of children are investigated.
In humans, pulmonary oxygen uptake during moderate and vigorous exercise is purported to reflect the rise in muscle O2 consumption (2), with coherence within ±10% for muscle V˙O2 and phase II pulmonary V˙O2 kinetics (14). The primary muscles used during walking and running in humans are in the lower (gastrocnemius, soleus, tibialis anterior, plantar flexors, and dorsiflexors) and upper leg (quadriceps, hamstrings, gluteals, and iliopsoas) (13,23). The biceps brachii and abdominals are supporting muscles. Using lean mass of both legs as a scaling variable for V˙O2peak therefore provides direct quantification of a large proportion of the metabolically active muscle mass during physiological assessment.
In contrast to lean mass of both legs, body mass is confounded as a scaling variable for V˙O2peak in humans because of the heterogeneity of body composition (12). Previous research has found that body mass cannot remove the influence of body size properly from V˙O2peak and provide a size-free index (25,27). Inappropriate body size scaling can confound comparisons of aerobic fitness in terms of metabolic or cardiac performance and the physiological ability to maximally consume oxygen (11). In 128 boys and 97 girls age 8–11 yr, DXA-assessed body fat correlated with V˙O2peak relative to body mass (11). However, there was no relationship between body fat and V˙O2peak relative to total lean body mass. Further, differences in V˙O2peak relative to body mass between children in the lowest and highest quartile for body fat disappeared when V˙O2peak was scaled to total lean body mass. It was subsequently concluded that total lean body mass was a more appropriate, body fat independent scaling variable for V˙O2peak than total body mass, which our findings for total lean mass support. Our findings extend knowledge in this area by showing that the best model uses the lean mass of both legs, as a direct index of involved musculature.
Similar to the problems with scaling by body mass, total fat-free mass estimated from skinfold assessment was precluded as an appropriate scaling variable for V˙O2peak in young people because lower leg muscle volume did not represent a constant proportion of fat-free mass (25). In the current study, using the lean mass of both legs, this problem did not arise, as the lean mass of both legs was directly proportional to total lean body mass. Consequently, both the total lean body mass and lean mass of both legs variables were appropriate models for removing the influence of size, with the latter being a marginally superior fit. To improve our interpretation of aerobic fitness in children, more appropriate techniques must be used to quantify metabolically active muscle mass in the legs during physiological assessment. Techniques might include DXA, MRI (25,28), calf and thigh muscle girths (19), and assessments that represent a constant proportion of fat-free or lean mass.
To our knowledge, this is the first study to examine the appropriateness of scaling V˙O2peak in children using DXA-assessed lean mass of both legs and total lean mass, compared with total body mass. Tolfrey et al. (25) used MRI to measure lower leg muscle volume in 15 boys age 12.3 ± 0.3 yr. This scaling variable for V˙O2peak had an exponent of b = 0.62, which lies within our 95% confidence interval when V˙O2peak was scaled for lean mass of both legs. Welsman et al. (28) used MRI to measure thigh muscle volume in 16 boys age 9.9 ± 0.3 yr. The exponent of b = 0.55 for thigh muscle volume is also consistent with our point estimate. Direct comparisons cannot be made across these studies because of differences in samples, scaling variables, and additional covariates (maturity offset). However, although DXA-derived lean mass should be seen as the superior method of quantifying anthropometric scaling variables, the similarity of these scaling exponents lends support to the view of Tolfrey et al. (25) that MRI-assessed leg volume (upper or lower) is adequately representative of the involved musculature in exercise at V˙O2peak.
The statistical analyses used here included AIC. It is important to note that the AIC is grounded in information theory as opposed to a null hypothesis testing framework. The method does not assume a “true” model, and individual AIC values are only interpretable in relation to other AIC values in the candidate model set (8). That is, it would not be possible to determine from AIC values if all three candidate models were poorly specified. However, as body mass is a widely used scaling variable for V˙O2peak and body mass was the worst of the three candidate models, our findings inform scaling of aerobic fitness data in children. Moreover, body size and absolute peak oxygen uptake are strongly related, so all three candidate models examined are biologically plausible, that is, they indicate involved musculature as a sink for oxygen consumption. Although all participants in the study met at least one criterion for providing an acceptable V˙O2peak score from the fitness test, it is important to note that some children did not meet end points both for RER and HR for the test.
CONCLUSIONS AND IMPLICATIONS
To improve the interpretation of aerobic fitness data in children, scaling V˙O2peak by the lean mass of both legs or by DXA-estimated total lean body mass is superior to scaling by total body mass. The best of the three candidate models was that involving the lean mass of both legs, although it was only marginally superior to the total lean body mass model. The influence exerted by maturation on V˙O2peak independent of lean mass suggests a measure of maturation should be assessed and controlled for when interpreting sex differences in aerobic fitness in pediatric populations.
This study was supported by funding from Liverpool City Council through the SportsLinx project and a Neighbourhood Renewal Fund.
The authors have no conflicts of interest.
The results of the present study do not constitute endorsement by the American College of Sports Medicine.
1. Albrecht GH, Gelvin BR, Hartman SE. Ratios as a size adjustment in morphometrics. Am J Phys Anthropol
. 1993; 91: 441–68.
2. Armstrong N, Barker AR. Oxygen uptake kinetics in children and adolescents: a review. Pediatr Exerc Sci
. 2009; 21: 130–47.
3. Armstrong N, Welsman JR. Aerobic fitness
. In: Armstrong N, Van Mechelen W, editors. Paediatric Exercise Science and Medicine
. Oxford (UK): Oxford University Press; 2000. p. 65–76.
4. Armstrong N, Welsman JR. Peak oxygen uptake in relation to growth and maturation
in 11- to 17-year old humans. Eur J Appl Physiol
. 2001; 85: 546–51.
5. Armstrong N, Welsman JR, Kirby BJ. Peak oxygen uptake and maturation
in 12-yr olds. Med Sci Sports Exerc
. 1998; 30 (1): 165–9.
6. Armstrong N, Welsman J, Winsley R. Is peak V˙O2
a maximal index of children’s aerobic fitness
? Int J Sports Med
. 1996; 17: 356–9.
7. Boddy LM, Hackett AF, Stratton G. Changes in fitness, body mass index and obesity in 9–10 year olds. J Hum Nutr Diet
. 2010; 23: 254–9.
8. Burnham KP, Anderson DR. Model Selection and Multimodel Inference: A Practical Information-Theoretic Approach
. 2nd ed. Springer-Verlag; 2002. p. 98–149.
9. Caspersen CJ, Powell KE, Christenson GM. Physical activity, exercise, and physical fitness: definitions and distinctions for health-related research. Public Health Rep
. 1985; 100 (2): 126–31.
10. Chinn S, Rona RJ. Prevalence and trends in overweight and obesity in three cross-sectional studies of British children, 1974–94. BMJ
. 2001; 322: 24–6.
11. Dencker M, Wollmer P, Karlsson MK, Lindén C, Andersen LB, Thorsson O. Body fat, abdominal fat and body fat distribution related to V˙O(2PEAK) in young children. Int J Pediatr Obes
. 2011; 6 (2–2): e597–602.
12. Döbeln WV. Maximal oxygen uptake, body size, and total hemoglobin in normal man. Acta Physiol Scand
. 1956; 38: 193–9.
13. Gottschall JS, Kram R. Energy cost and muscular activity required for leg swing during walking. J Appl Physiol
. 2005; 99: 23–30.
14. Grassi B, Poole DC, Richardson RS, Knight DR, Erickson BK, Wagner PD. Muscle O2
uptake kinetics in humans: implications for metabolic control. J Appl Physiol
. 1996; 80: 988–98.
15. Hopkins ND, Stratton G, Tinken TM, et al. Relationships between measures of fitness, physical activity, body composition and vascular function in children. Atherosclerosis
. 2009; 204 (1): 244–9.
16. Lohman TG, Going SB. Body composition assessment for development of an international growth standard for preadolescent and adolescent children. Food Nutr Bull
. 2006; 27 (4): S314–25.
17. Lohman TG, Roche AF, Martorell R. Anthropometric Standardisation Reference Manual
. Champaign (IL): Human Kinetics; 1991.p. 55–70.
18. McCarthy HD, Cole TJ, Fry T, Jebb SA, Prentice AM. Body fat reference curves for children. Int J Obes
. 2006; 30: 598–602.
19. Mirwald RL, Baxter-Jones ADG, Bailey DA, Beunen GP. An assessment of maturity from anthropometric measurements. Med Sci Sports Exerc
. 2002; 34 (4): 689–94.
20. Nevill AM, Markovic G, Vucetic V, Holder RL. Can greater muscularity in larger individuals resolve the 3/4 power-law controversy when modelling maximum oxygen uptake? Ann Hum Biol
. 2004; 31: 436–45.
21. Packard GC, Birchard GF, Boardman TJ. Fitting statistical models in bivariate allometry. Biol Rev
. 2011; 86: 549–63.
22. Rowland TW. Does peak V˙O2
in children? Evidence from supramaximal testing. Med Sci Sports Exerc
. 1993; 25 (6): 689–93.
23. Sloniger MA, Cureton KJ, Prior BM, Evans EM. Lower extremity muscle activation during horizontal and uphill running. J Appl Physiol
. 1997; 83: 2073–9.
24. Taylor S, Hackett A, Stratton G, Lamb L. SportsLinx: improving the health and fitness of Liverpool’s youth. Educ Health
. 2004; 22: 3–7.
25. Tolfrey K, Barker A, Thom JM, Morse CI, Narici MV, Batterham AM. Scaling of maximal oxygen uptake by lower leg muscle volume in boys and men. J Appl Physiol
. 2006; 100: 1851–6.
26. Wallace JA, George K, Reilly T. Validity of dual-energy X-ray absorptiometry for body composition analysis. In: Bust PD, editor. Contemporary Ergonomics
. London (UK): Taylor and Francis; 2006. p. 513–5.
27. Welsman JR, Armstrong N. Scaling for size: relevance to understanding the effects of growth on performance. In: Hebestreit H, Bar-Or O, editors. Encyclopedia of Sports Medicine: The Young Athlete
. Oxford (UK): Blackwell; 2007. p. 50–62.
28. Welsman JR, Armstrong N, Kirby BJ, Winsley RJ, Parsons G, Sharpe P. Exercise performance and magnetic resonance imaging-determined thigh muscle volume in children. Eur J Appl Physiol Occup Physiol
. 1997; 76: 92–7.