Individual differences in the timing and tempo of biological maturation are considerable, particularly during the transition into adolescence and during adolescence per se. In the context of competitive sport, biological maturity status is of relevance for several reasons: the wide range of variation among individuals; the relative success of boys advanced in maturation and of girls later in maturation in a number of sports, especially as level of competition increases (e.g., soccer and ice hockey in boys, gymnastics and figure skating in girls); as a potential factor for selection in some sports (e.g., artistic gymnastics) and in providing guidance to young athletes; as a basis for adjusting training loads to individuals, which has the potential to reduce the risk of injury; and in the context of suggestions that intensive training during the adolescent growth spurt and puberty delays the progress of maturation in female athletes (12,13,15).
Biological maturation is often discussed in terms of status and timing. Maturity status refers to the state of maturation of the individual at a given point in time, for example, skeletal age or percentage of mature height attained at a specific chronological age, pre- and postmenarcheal girls of the same chronological age, and prepubertal, pubertal, and postpubertal youth of the same chronological age. Some indicators of maturity status are considered invasive. Skeletal maturity assessment requires radiographs; pubertal status is best assessed at clinical examination implying direct examination of secondary sex characteristics.
Maturity timing, on the other hand, refers to when specific maturational events occur-for example, age at menarche, age at peak height velocity (PHV), age at attaining a specific percentage of mature height, and so on. Estimates of maturity timing in individual youth require longitudinal data (15).
Given the difficulties in obtaining a measure of maturity status and timing, several noninvasive estimates have been proposed. Expressing current height as a percentage of predicted mature height (4,6,15,16,21) requires an estimate of mature height, which can be predicted (1-3,7,9,22,24). During the adolescent growth spurt, however, estimates of skeletal maturity are required to obtain more accurate predictions (1-3,7,22,24). More recently, time before or after PHV, maturity offset in years, has been proposed as a maturity indicator (17). This approach predicts the time before or after PHV from age, height, weight, sitting height, and estimated leg length. Chronological age at the time of measurement plus maturity offset can permit an estimate of age at PHV. This approach has been used with swimmers (23), gymnasts (19), and a combined sample of figure skaters and ballet dancers (18). Since the original publication of the protocol (17), use of maturity offset as a maturity indicator has not been externally validated on an independent longitudinal sample for which age at PHV is known. The purpose of this study is to verify the applicability of the prediction equation for maturity offset in a sample of female gymnasts followed longitudinally through adolescence.
The study is based on the individual records of 15 Belgian female gymnasts from two prominent clubs in the Antwerp region. The study was approved by the medical ethics committee of the Faculty of Kinesiology and Rehabilitation Sciences of the Katholieke Universiteit Leuven. Informed consent was provided by the parents at all time points, and each gymnast provided written consent after she reached 10 yr of age.
The gymnasts were followed longitudinally at yearly intervals for 6-7 yr between 1990 and 1997. Ages at entry into the program ranged from 6.0 to 11.6 yr (8.7 ± 1.5 yr), whereas ages at the conclusion of the study ranged from 12.8 to 17.6 yr (15.5 ± 1.5 yr). The gymnasts entered the program as beginning-level participants and moved to regional and national competitive levels during the study. They were identified as regional/national caliber at 11.6 ± 1.2 yr, with a range from 8.8 to 13.6 yr. The gymnasts trained, on average, 15 h·wk −1 during the course of the study, with a range from 9 to 19 h·wk−1 (25).
Height and sitting height were measured at annual intervals. Estimates of interobserver measurement reliability for height and sitting height were, respectively: mean differences, 0.23 and 0.43 cm; coefficients of variation, 0.3 and 1.0%; and correlation coefficients, 0.99 and 0.95. Sitting height was subtracted from height to estimate leg (subischial) length.
The Preece-Baines Model I (20) was fitted to longitudinal data for individual gymnasts to derive estimates of age at peak height velocity (yr). The curve-fitting protocol was successfully fit to the height records of 13 of 15 gymnasts with standard errors of estimate between 0.02 and 0.28 cm (25). Curve fitting was unsuccessful in two gymnasts. One was followed from 8.1 to 15.1 yr, and peak velocity of growth in height apparently occurred between the last two observations, but the associated error of estimate was large. The other gymnast was followed from 10.8 to 17.8 yr; PHV apparently occurred between the first two observations, and the model did not provide a good fit.
Maturity offset, that is, time before or after PHV, was predicted with the equation of Mirwald et al. (17):
Maturity offset = −9.376 + 0.0001882 × leg length and sitting height interaction + 0.0022 × age and leg length interaction + 0.005841 × age and sitting height interaction − 0.002658 × age and weight interaction + 0.07693 × weight by height ratio
where R = 0.94, R2 = 0.89 and SEE = 0.57. Length measurements are in centimeters and weight measurements are in kilograms; the weight by height ratio is multiplied by 100. Maturity offset was calculated from measurements taken at each observation for the 13 gymnasts for whom a successful Preece-Baines fit was obtained. Estimates were calculated for ages ≥ 9.0 yr and < 16.0 yr because observations for younger and older ages, respectively, were limited. Maturity offset was added to chronological age at each observation for individual subjects to provide estimates of ages at PHV.
Age at PHV derived with the Preece-Baines model was the criterion for each gymnast (25). Differences between the criterion age at PHV and predicted age at PHV were calculated at each observation for individual gymnasts. Differences between medians of predicted age at PHV and the criterion at each age were tested with the nonparametric Wilcoxon signed ranks test. Spearman correlations between age-specific predicted age at PHV and the criterion were also calculated. Finally, Bland-Altman procedures were used to check for systematic under- or overestimation with the maturity offset prediction equation.
Means, standard deviations, and ranges for maturity offset predicted ages at PHV from maturity offset, and differences between the criterion and predicted ages at PHV by age group are summarized in Table 1. Maturity offset for each of the 13 gymnasts is shown in Figure 1, whereas mean maturity offset and associated standard deviations for each age are shown in Figure 2. Mean maturity offset is negative and highest at 9 yr of age and moves towards zero with increasing age. It approaches zero between 12.5 and 13.0 yr, which is close to the mean criterion age at PHV for the sample, 12.9 ± 1.5 yr. Subsequently, maturity offset is positive and increases with age. Interindividual variation in offset is greatest between 11 and 14 yr of age.
Mean ages at PHV predicted from maturity offset and standard deviations are shown by age group in Figure 3. The earliest predicted mean age at PHV occurs at the youngest age, 9 yr. Mean predicted ages at PHV increase systematically with chronological age. Mean predicted ages at PHV vary between 11.9 and 12.9 yr, but are systematically lower than the criterion age at PHV for the sample. With increasing age, mean predicted ages at PHV move gradually towards the criterion; the mean predicted age at a chronological age of 15 yr is identical with the criterion age at PHV. Standard deviations for predicted ages at PHV based on maturity offset range from 0.2 to 0.4 yr and are consistently lower than the standard deviation of the criterion age at PHV, 1.5 yr.
Differences between individual ages at PHV based on the Preece-Baines model and ages at PHV predicted from maturity offset for individual gymnasts are shown in Figure 4, whereas means and standard deviations for the differences between the criterion PHV for each gymnast and PHV predicted from maturity offset are shown by age group in Figure 5. Interindividual variation is considerable. The difference between criterion age at PHV and predicted age at PHV is largest at 9 yr and declines systematically with chronological age. At a mean chronological age of 12.5 yr, the closest to the criterion (12.9 yr), the difference between predicted and criterion ages at PHV is 0.37 ± 1.22 yr (Table 1). At chronological ages of 14 and 15 yr, mean predicted ages at PHV from maturity offset, 12.8 ± 0.40 and 12.9 ± 0.31 yr, respectively, are virtually the same as the criterion age at PHV, 12.9 ± 1.5 yr. However, the differences between predicted and criterion ages at PHV at these chronological ages are 0.46 ± 1.04 and 0.40 ± 1.27 yr, respectively (Table 1).
Results of the Bland-Altman plot of criterion and predicted ages at PHV for 12-yr-old gymnasts are shown in Figure 6. For gymnasts with an early criterion age at PHV, the maturity offset prediction of age at PHV is at a later age (negative values). On the other hand, for gymnasts with a later criterion age at PHV, the maturity offset prediction of age at PHV is at an earlier age (positive values). It thus appears that maturity offset predictions of age at PHV are overestimated in gymnasts with an earlier criterion age at PHV and underestimated in gymnasts with a later age criterion at PHV. Similar trends are apparent for Bland-Altman plots at ages 9-15 yr (not shown).
Spearman correlations between the criterion age at PHV and predicted age at PHV from maturity offset within each age group are shown in Table 2. Most correlations are moderate to moderately high (0.50- 0.76) and are significant only at 11 and 12 yr. Given the small samples, they are not sufficiently high to warrant valid predictions. In the youngest age group, 9 yr, the correlation between the criterion and predicted ages at PHV is low and negative, −0.13, and not significant.
The present study is an external verification of the applicability of the equation to predict maturity offset (17) to a sample of 13 female gymnasts, who were followed longitudinally through adolescence and for whom age at PHV was derived mathematically (25). Maturity offset and predicted ages at PHV based on maturity offset varied with chronological age of the individual gymnasts.
The same pattern was evident when the differences between the criterion age at PHV based on the Preece-Baines model and the age at PHV based on maturity offset were expressed in terms of years before and after the criterion age at PHV. The difference between the two estimates of age at PHV was 1.10 ± 0.97 yr at 3.0 yr before criterion PHV, 0.36 ± 1.12 yr at 1.0 yr before PHV, 0.48 ± 0.97 yr at PHV, and 0.24 ± 1.13 yr at 1.0 yr after PHV. Numbers were too small at 2 and 3 yr after PHV. Nevertheless, the prediction error is similar whether expressed by chronological age or by years before and after the criterion age at PHV.
The deviation between predicted ages at PHV and the criterion is a concern. The results suggest that the prediction equation developed on a nonathletic sample of girls is not applicable to a relatively homogeneous sample such as gymnasts, who are, on average, short with mean heights that approximate the 10th percentiles of reference samples and who also are, on average, later maturing (11). The adolescent pattern of growth of female gymnasts is generally similar to that of short, normal, slow-maturing girls with short parents (13,14).
Studies of height prediction in short girls also give variable results compared with girls of normal stature. Errors in prediction of mature height also tend to be larger in short, normal, slow-maturing girls than for average children (8), and care is suggested in estimating target mature heights for children with short parents (10). In a comprehensive analysis of five height prediction equations in girls with short stature and late puberty and with short parents, mean prediction errors for mature height varied between −1.4 and + 3.6 cm (5). The large errors could be related to differences in biological maturation and the growth process as well as their interaction in short girls compared with the general population. The results suggest that mature height prediction equations based on the general population of girls with normal stature have limitations when applied to short girls.
Predicted ages at PHV based on maturity offset have relatively small standard deviations within each age group of gymnasts, 0.19-0.42 yr (Table 1). This contrasts the standard deviation for the criterion age at PHV based on the Preece-Baines model in this sample of 13 gymnasts, 1.5 yr (25), as well as standard deviations for ages at PHV in longitudinal samples of nonathletes, 0.7-1.2 yr (15). The prediction of maturity offset and age at PHV is, of course, based on anthropometric dimensions taken before the attainment of PHV and thus may not reflect the variation observed at PHV. Moreover, although female gymnasts are relatively homogeneous in size, build, and physique, they are not homogeneous in the timing of biological maturation as reflected in the age at PHV.
In a sample of 45 gymnasts 8-17 yr of age, predicted age at PHV based on maturity offset was 13.0 ± 0.7 yr (19). This age was similar to the age at PHV for the 13 gymnasts based on the Preece-Baines model, 12.9 ± 1.5 yr (25). The estimated mean age at PHV predicted from maturity offset for all observations for the longitudinal sample of 13 gymnasts 8-17 yr (N = 92) was 12.46 ± 0.56 yr.
As emphasized earlier, the assessment of biological maturation is important in dealing with young athletes and also for adolescents in the general population. Although prediction of maturity offset has merit in that it does not require invasive procedures, care is warranted in utilizing maturity offset per se and predicted age at PHV based on maturity offset as an indicator of maturity timing in short girls. Because height and sitting height and the derivation of subischial (leg) length by subtraction are required for the prediction, measurement variability (error) needs to be considered. Further, the equations in the original report (17) lack precision. The weight/height ratio needs to be multiplied by 100, and this is not clearly specified. All predictions have associated errors and outlying predictions, so that application to individuals needs to be made with care. Individual differences in the timing and tempo of the adolescent growth spurt are considerable and also contribute to prediction errors.
Although the present study is limited to 13 gymnasts, a criterion age at PHV based on the Preece-Baines model fitted to longitudinal height records for individual girls was available. The results suggest that predictions of maturity offset have limitations, specifically with short females.
Special thanks are extended to Mrs. I. Wuyts, coach of the gymnasts, and to all of the girls who participated in this longitudinal study.
1. Bayer, L. M., and N. Bayley. Growth Diagnosis: Selected Methods for Interpreting and Predicting Development from One Year to Maturity
. Chicago: University of Chicago Press, 1959, pp. 45-58.
2. Bayley, N. The accurate prediction of growth and adult height. Mod. Probl. Paediatr.
3. Bayley, N., and S. R. Pinneau. Tables for predicting adult height from skeletal age: Revised for use with the Greulich-Pyle hand standards. J. Pediatr.
4. Beunen, G. P., R. M. Malina, J. Lefevre, A. L. Claessens, R. Renson, and J. Simons. Prediction of adult stature and noninvasive assessment of biological maturation. Med. Sci. Sports Exerc.
5. Bramswig, J. H., M. Fasse, M.-L. Holthoff, H. J. von Lengerke, W. von Petykowski, and G. Schellong. Adult height in boys and girls with untreated short stature and constitutional delay of growth and puberty: accuracy of five different methods of height prediction. J. Pediatr.
6. Eaton, W. O., and A. P. Yu. Are sex differences in child motor activity level a function of sex differences in maturational status? Child Dev.
7. Khamis, H. J., and S. Guo. Improvement in the Roche-Wainer-Thissen stature prediction model: A comparative study. Am. J. Hum. Biol.
8. Khamis, H. J., and A. F. Roche. Growth outcome of "normal" short children who are retarded in skeletal maturation. J. Pediatr. Endocrinol. Metab.
9. Khamis, H. J., and A. F. Roche. Predicting adult stature without using skeletal age: The Khamis-Roche method. Pediatrics
94:504-507, 1994. Erratum in: Pediatrics
10. Luo, Z. C., K. Albertsson-Wikland, and J. Karlberg. Target height as predicted by parental heights in a population-based study. Pediatr. Res.
11. Malina, R. M. Physical growth and biological maturation of young athletes. Exerc. Sports Sci. Rev.
12. Malina, R. M. Growth and maturation of young athletes - is training for sport a factor? In: Sports and Children
, K.-M. Chan and L. J. Micheli (Eds.). Hong Kong: Williams and Wilkins Asia-Pacific, 1998, pp. 133-161.
13. Malina, R. M. Growth and maturation of elite female gymnasts: Is training a factor? In: Human Growth in Context
, F. E. Johnston, B. Zemel, and P. B. Eveleth (Eds). London: Smith-Gordon, 1999, pp. 291-301.
14. Malina, R. M. Growth and maturity status of young artistic gymnasts: status, progress, and issues. In: Science in Artistic Gymnastics (Proceedings of the 6th Internationaal Sportwetenschappelijk Symposium)
, M. Lenoir and R. Philippaerts (Eds.). Ghent: Publicatiefonds Voor Lichamelijke Opvoeding vzw, 2001, pp. 21-38.
15. Malina, R. M. C. Bouchard, and O. Bar-Or. Growth, Maturation, and Physical Activity
, 2nd edition. Champaign, IL: Human Kinetics; 2004, pp. 277-365, 382-384, 623-649.
16. Malina, R. M., S. P. Cumming, P. J. Morano, M. Barron, and S. J. Miller. Maturity status of youth football players: a noninvasive estimate. Med. Sci. Sports Exerc.
17. Mirwald, R. L., A. D. G. Baxter-Jones, D. A. Bailey, and G. P. Beunen. An assessment of maturity from anthropometric measurements. Med. Sci. Sports Exerc.
18. Monsma, E. V., K. A. Pfeiffer, R. Harvey, R. Ross, S. Brown, and R. M. Malina. Maturity-offset, age at menarche, and social physique anxiety among female participantsa in aesthetic activities. J. Sport Exerc. Psychol.
27 (suppl):S109, 2005. (abstract).
19. Nurmi-Lawton, J. A., A. D. Baxter-Jones, R. L. Mirwald, et al. Evidence of sustained skeletal benefits from impact-loading exercise in young females: a 3-year longitudinal study. J. Bone Min. Res.
20. Preece, M. A., and M. J. Baines. A new family of mathematical models describing the human growth curve. Ann. Hum. Biol.
21. Roche, A. F., F. Tyleshevski, and E. Rogers. Non-invasive measurement of physical maturity in children. Res. Q. Exerc. Sport
22. Roemmich, J. N., R. M. Blizzard, S. D. Peddada, et al. Longitudinal assessment of hormonal and physical alterations during normal puberty in boys: IV: Predictions of adult height by the Bayley-Pinneau, Roche-Wainer-Thissen, and Tanner-Whitehouse methods compared. Am. J. Hum. Biol.
23. Simmons, S. E., J. C. White, and J. M. Stager. Maturity assessment in competitive swimmers. Med. Sci. Sports Exerc.
36 (Suppl):S103, 2004. (abstract).
24. Tanner, J. M., M. J. R. Healy, H. Goldstein, and N. Cameron. In: Assessment of Skeletal Maturity and Prediction of Adult Height (TW3 Method)
, 3rd edition. Philadelphia: Saunders, 2001.
25. Thomis, M., A. L. Claessens, J. Lefevre, R. Philippaerts, G. P. Beunen, and R. M. Malina. Adolescent growth spurts in female gymnasts. J. Pediatr.