ANNALS OF HUMAN BIOLOGY,
1991, VOL. 18, NO. 1, 2 3 - 2 9
Skeletal maturity at onset of the adolescent growth spurt and at peak velocity for growth in height: a threshold effect? R. HAUSPIE $, T. BIELICKI~:and J. KONIAREK~ *National Science Foundation, Vrije Universiteit, Brussels, Belgium ~Polish Academy of Sciences, Institute of Anthropology, Wroclaw, Poland
Ann Hum Biol Downloaded from informahealthcare.com by University of Otago on 12/28/14 For personal use only.
Received 20 July 1989; revised 15 June 1990
Summary. In 191 Polish boys of the Wroclaw Growth Study, the relationship between skeletal age and chronological age was examined at the onset of the adolescent growth spurt (take-off) and at peak velocity of height growth (PHV). It was found that, at PHV, skeletal age is markedly less variable than is chronological age, but at take-off no such reduction in variability is visible. The following interpretation of this finding is proposed. The onset of the spurt depends, ultimately, upon some maturational processes going on in the hypothalamus and shows little relationship with the advancement of the long bones at that time. Therefore, the spurt can begin at any level of skeletal maturity within the range normally observed at the chronological age at which it happens to begin in the individual. Peak height velocity, on the other hand, is reached when skeletal maturity is sufficiently advanced for testosterone to change its influence upon the bones from one which consists in stimulating cartilage growth to one which consists in stimulating epiphyseal fusion, Therefore, PHV is bound to occur within a range of skeletal maturity much more restricted than that within which take-off can occur.
1.
Introduction It is well known that a strong association exists between the timing of the adolescent growth spurt in skeletal diameters, especially body height, and the rate of skeletal maturation during adolescence (e.g. Marshall and Tanner 1986, Roche 1986). For example, in the nearly 200 boys of the Wroclaw Growth Study (the material on which the present report is based) a correlation of r = 0" 89 was found between age at reaching peak velocity of growth in height (obtained by graphic interpolation; cf. Bielicki and Welon 1973) and the age at attaining the skeletal maturity score median for 14-year-olds (Bielicki, Koniarek and Malina 1984). The demonstration of the existence of these relationships, however, does not answer the following question: Is there, during the period of adolescence, any association between (a) certain characteristic 'points' on the curve of growth in height and (b) the level of skeletal maturity at which these 'points' are reached by the individual? For example: does take-off, or the moment at which growth velocity begins to decrease (i.e. the PHV point), 'require' for its occurrence the attainment of a certain, more or less invariable, 'threshold level' of maturity of the long bones? Or is it rather that the level of skeletal maturity on reaching take-off, or PHV, is simply as variable as it is at the chronological age of occurrence of these developmental events? The only report known to us, dealing with this question, is an analysis of relationships between maturation and growth carried out on the Harpenden data by Marshall (1974). This author examined relationships between skeletal age and chronological age for a set of developmental landmarks wider than that with which we are concerned here, since in addition to PHV Marshall also considered secondary sex characters and menarche, but not the take-off point. He found that skeletal ages of boys at the beginning of genital and pubic hair development, or of girls at the beginning of breast development, vary as much as, or even slightly more than, chronological ages. 0301-4460/91 $3.00 © 1991Taylor & Francis Ltd.
Ann Hum Biol Downloaded from informahealthcare.com by University of Otago on 12/28/14 For personal use only.
24
R. Hauspie et al.
He also found that for PHV and later stages of genital development the situation tends to be reversed, i.e. variation of skeletal ages is reduced compared to chronological ages, although not to a statistically significant degree. Such reduction was also observed at menarche and in this case it was marked and highly significant. The apparent absence of a similar effect at PHV was singled out by Marshall with the following comment: 'The wide variation in skeletal age at PHV is a particularly interesting observation as it raises questions about the relationship between skeletal maturation and growth in stature.' The result which led Marshall to make the above comment seemed to us counterintuitive. We therefore decided to examine variation in skeletal and chronological age at PHV, and also at the take-off point, on the much larger material available to us. 2.
Material and methods The subjects are 191 Polish boys of the Wroclaw Growth Study, all born in 1953 and examined annually from 1961 to 1971-72 (see Waliszko and Jedlinska 1976, for a general description of that material). Skeletal maturity was, throughout the duration of the study, determined from hand-and-wrist radiographs by one of us (J.K.) by the TW2 method (Tanner, Whitehouse, Marshall, Healy and Goldstein 1975). Bone age was not available at take-off for some subjects ('20 bones': 10 subjects; RUS: 12 subjects). For each individual, Preece Baines model 1 was fitted to the serial data of height for age (Preece and Baines 1978). All curve fittings were successful: the pooled residual variance was 0"479cm2 (d.f. = 1216) and the runs test showed a significantly too high number of runs in 11 cases out of the 191 (=6%), which means that there is no systematic bias in the estimation of the individual growth curves. Age at take-off and age at PHV were calculated from the function parameters. Subsequently, the skeletal scores ('20 bones' and RUS) at exactly the age at take-off, and at PHV, were read off from individual hand-smoothed curves of skeletal maturity, and were then converted to bone ages by means of the converting tables developed for that sample by Koniarek, Slawinska and Komarowska (1983). 3.
Results Tables 1 and 2 show the frequency distribution of chronological age and bone age, respectively at take-off and at peak velocity. The Gauss-adjusted frequencies (expected frequencies under the assumption of normality) were compared to the observed frequencies by means of the x2-test. The results show that only bone age (both '20 bones' and RUS) at PHV shows significant departures from normality. Figures 1 and 2 depict the polygons of the expected frequency for chronological age and the observed frequency for bone age ('20 bones' and RUS), respectively at take-off and at PHV. Table 3 shows various parametric and nonparametric descriptive statistics for the distribution of chronological age and bone age. The standard deviations and interquartile ranges indicate that, at take-off, the dispersion of bone age is slightly greater than that of chronological age. A test for equality of variances of correlated variables (Dagneli 1984) revealed a significantly greater variation of the '20 bones' -bone age than of the chronological age at take-off (t = 2.052; d.f. = 177;p < 0" 05), but not so for the RUS-bone age (t= 1.414; d.f. = 177; 0"2 < p
1991, VOL. 18, NO. 1, 2 3 - 2 9
Skeletal maturity at onset of the adolescent growth spurt and at peak velocity for growth in height: a threshold effect? R. HAUSPIE $, T. BIELICKI~:and J. KONIAREK~ *National Science Foundation, Vrije Universiteit, Brussels, Belgium ~Polish Academy of Sciences, Institute of Anthropology, Wroclaw, Poland
Ann Hum Biol Downloaded from informahealthcare.com by University of Otago on 12/28/14 For personal use only.
Received 20 July 1989; revised 15 June 1990
Summary. In 191 Polish boys of the Wroclaw Growth Study, the relationship between skeletal age and chronological age was examined at the onset of the adolescent growth spurt (take-off) and at peak velocity of height growth (PHV). It was found that, at PHV, skeletal age is markedly less variable than is chronological age, but at take-off no such reduction in variability is visible. The following interpretation of this finding is proposed. The onset of the spurt depends, ultimately, upon some maturational processes going on in the hypothalamus and shows little relationship with the advancement of the long bones at that time. Therefore, the spurt can begin at any level of skeletal maturity within the range normally observed at the chronological age at which it happens to begin in the individual. Peak height velocity, on the other hand, is reached when skeletal maturity is sufficiently advanced for testosterone to change its influence upon the bones from one which consists in stimulating cartilage growth to one which consists in stimulating epiphyseal fusion, Therefore, PHV is bound to occur within a range of skeletal maturity much more restricted than that within which take-off can occur.
1.
Introduction It is well known that a strong association exists between the timing of the adolescent growth spurt in skeletal diameters, especially body height, and the rate of skeletal maturation during adolescence (e.g. Marshall and Tanner 1986, Roche 1986). For example, in the nearly 200 boys of the Wroclaw Growth Study (the material on which the present report is based) a correlation of r = 0" 89 was found between age at reaching peak velocity of growth in height (obtained by graphic interpolation; cf. Bielicki and Welon 1973) and the age at attaining the skeletal maturity score median for 14-year-olds (Bielicki, Koniarek and Malina 1984). The demonstration of the existence of these relationships, however, does not answer the following question: Is there, during the period of adolescence, any association between (a) certain characteristic 'points' on the curve of growth in height and (b) the level of skeletal maturity at which these 'points' are reached by the individual? For example: does take-off, or the moment at which growth velocity begins to decrease (i.e. the PHV point), 'require' for its occurrence the attainment of a certain, more or less invariable, 'threshold level' of maturity of the long bones? Or is it rather that the level of skeletal maturity on reaching take-off, or PHV, is simply as variable as it is at the chronological age of occurrence of these developmental events? The only report known to us, dealing with this question, is an analysis of relationships between maturation and growth carried out on the Harpenden data by Marshall (1974). This author examined relationships between skeletal age and chronological age for a set of developmental landmarks wider than that with which we are concerned here, since in addition to PHV Marshall also considered secondary sex characters and menarche, but not the take-off point. He found that skeletal ages of boys at the beginning of genital and pubic hair development, or of girls at the beginning of breast development, vary as much as, or even slightly more than, chronological ages. 0301-4460/91 $3.00 © 1991Taylor & Francis Ltd.
Ann Hum Biol Downloaded from informahealthcare.com by University of Otago on 12/28/14 For personal use only.
24
R. Hauspie et al.
He also found that for PHV and later stages of genital development the situation tends to be reversed, i.e. variation of skeletal ages is reduced compared to chronological ages, although not to a statistically significant degree. Such reduction was also observed at menarche and in this case it was marked and highly significant. The apparent absence of a similar effect at PHV was singled out by Marshall with the following comment: 'The wide variation in skeletal age at PHV is a particularly interesting observation as it raises questions about the relationship between skeletal maturation and growth in stature.' The result which led Marshall to make the above comment seemed to us counterintuitive. We therefore decided to examine variation in skeletal and chronological age at PHV, and also at the take-off point, on the much larger material available to us. 2.
Material and methods The subjects are 191 Polish boys of the Wroclaw Growth Study, all born in 1953 and examined annually from 1961 to 1971-72 (see Waliszko and Jedlinska 1976, for a general description of that material). Skeletal maturity was, throughout the duration of the study, determined from hand-and-wrist radiographs by one of us (J.K.) by the TW2 method (Tanner, Whitehouse, Marshall, Healy and Goldstein 1975). Bone age was not available at take-off for some subjects ('20 bones': 10 subjects; RUS: 12 subjects). For each individual, Preece Baines model 1 was fitted to the serial data of height for age (Preece and Baines 1978). All curve fittings were successful: the pooled residual variance was 0"479cm2 (d.f. = 1216) and the runs test showed a significantly too high number of runs in 11 cases out of the 191 (=6%), which means that there is no systematic bias in the estimation of the individual growth curves. Age at take-off and age at PHV were calculated from the function parameters. Subsequently, the skeletal scores ('20 bones' and RUS) at exactly the age at take-off, and at PHV, were read off from individual hand-smoothed curves of skeletal maturity, and were then converted to bone ages by means of the converting tables developed for that sample by Koniarek, Slawinska and Komarowska (1983). 3.
Results Tables 1 and 2 show the frequency distribution of chronological age and bone age, respectively at take-off and at peak velocity. The Gauss-adjusted frequencies (expected frequencies under the assumption of normality) were compared to the observed frequencies by means of the x2-test. The results show that only bone age (both '20 bones' and RUS) at PHV shows significant departures from normality. Figures 1 and 2 depict the polygons of the expected frequency for chronological age and the observed frequency for bone age ('20 bones' and RUS), respectively at take-off and at PHV. Table 3 shows various parametric and nonparametric descriptive statistics for the distribution of chronological age and bone age. The standard deviations and interquartile ranges indicate that, at take-off, the dispersion of bone age is slightly greater than that of chronological age. A test for equality of variances of correlated variables (Dagneli 1984) revealed a significantly greater variation of the '20 bones' -bone age than of the chronological age at take-off (t = 2.052; d.f. = 177;p < 0" 05), but not so for the RUS-bone age (t= 1.414; d.f. = 177; 0"2 < p
