| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
ORIGINAL RESEARCH COMMUNICATION |
1 From the Lifespan Health Research Center, Wright State University School of Medicine, Dayton, OH
2 Supported by the National Institutes of Health (grants HD12252 and HD27063). 3 Address reprint requests to EW Demerath, Lifespan Health Research Center, 3171 Research Boulevard, Kettering, OH 45420. E-mail: ellen.demerath{at}wright.edu.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Objective: The objective was to simultaneously examine the effects of birth cohort (secular trend) and rate of maturation (age at menarche) on the timing and pattern of increases in body mass index (BMI) during adolescence in girls.
Design: We applied mixed-effects polynomial models to serial BMI data, spanning from 6 y before menarche to 6 y after menarche, obtained from 211 girls enrolled in the Fels Longitudinal Study. We examined the effects of birth cohort (defined as girls born 1929–1946, 1947–1964, and 1965–1983) and age at menarche (defined as 
11.9 y, 12.0–13.1 y, and 
13.2 y) on the magnitude and velocity of BMI during adolescence.
Results: BMI and BMI velocity in girls born after 1965 were significantly greater than those of girls of earlier birth cohorts, despite stability in the mean age at menarche. Although girls with early menarche tended to have significantly higher BMIs than did girls with average or later menarche, these differences did not emerge until after menarche.
Conclusion: These data suggest that increases in relative weight are a consequence, rather than a determinant, of the age at menarche and that secular changes in BMI and in the mean age at menarche could be independent phenomena.
Key Words: Body mass index • menarche • longitudinal studies • cohort effect • body composition • growth • sex maturation • female children • adolescence • United States
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
15% of all children and youth aged 6–19 y and upward of 27% of African American girls aged 12–19 y are overweight (2). US national survey data are not available before the late 1960s, and so it is possible that the recent increase in childhood body mass index (BMI) is not unique to this period but is part of a long-term trend, beginning earlier in the 20th century.
It has been suggested that recent increases in the number of US girls beginning puberty at age 6–8 y could be an outcome of the increased prevalence of childhood overweight (3, 4). Significant declines in the age at menarche among African Americans and whites were reported in the Bogalusa Heart Study between 1978–1979 and 1992–1994, which were accompanied by concurrent increases in BMI (5, 6). Two other recent papers documented a small 2- to 3-mo drop in the median age at menarche in all US girls between the 1960s and the early 1990s, from 12.75 y to 12.5 y (7, 8). Such studies raise the possibility that increased childhood overweight is driving earlier maturation in girls.
It has been established that early-maturing girls of a given chronologic age tend to have higher BMI and body fat than later-maturing girls (9-15). There are, however, conflicting interpretations of this association. Increased adiposity in early childhood was suggested by some to precipitate earlier sexual maturation in girls (4-6, 16), whereas other work indicated increased adiposity is solely a consequence of sexual maturation in the early-maturing girl (17-19). Longitudinal data from successive birth cohorts within a population are necessary to examine the potential contribution of population shifts in adiposity to the timing of sexual development, but such data are rare.
The aim of this study was to simultaneously examine the effects of birth cohort (secular trend) and age at menarche (rate of maturation) on BMI during adolescence in girls. We analyzed serial BMI data in girls enrolled in the Fels Longitudinal Study who were born between 1929 and 1983 to examine how the pattern of increase in BMI before, during, and after menarche has changed over this 50-y period. We then compared the BMI growth of early-, average-, and late-maturing girls to determine whether any birth cohort effects on BMI might be due to changes in the timing of menarche. Because the Fels Longitudinal Study began almost 30 y before the initiation of national health and nutrition surveys, these data furthermore offer a glimpse into possible changes in BMI and the tempo of development earlier in the 20th century.
|
SUBJECTS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
9 y respond to questions about whether they started menstruating since their last visit. If the answer is yes, the girls are then asked to provide, if possible, an exact date. In most cases, girls report the onset of their menses at their first study examination after menarche. Eighty percent reported menarche within 6 mo of its onset, and within 5 y of onset for the remainder, which limits the likelihood of recall error. All of the 211 girls had complete serial data, with BMI observations at –6, –4, –2, 0, 2, 4, and 6 y from their respective ages at menarche. Because of inherent individual differences in rates of development, each of these maturational age groups included individuals with a range of chronologic ages (Table 1
). To examine cohort effects, the range of birth years was broken into 3 equal-sized birth cohort groups. Each birth cohort group spanned 18 y and had sufficient numbers of girls for between-group analysis: cohort 1 (born 1929–1946; n = 68), cohort 2 (born 1947–1964, n = 81), and cohort 3 (born 1965–1983, n = 62).
|
11.9 y, average menarche occurred from age 12.0 to 13.1 y, and late menarche occurred at age 13.2 y. The terms early maturers, average maturers, and late maturers refer to girls with early, average, and late age at menarche as defined earlier and does not refer to the timing of secondary sexual characteristic development (ie, breast, genitalia, or pubic hair development). The subjects were informed of the risks and benefits of their participation in the study, and they provided their written consent to participate. Before 1976, these procedures were approved by the Institutional Review Board of the Fels Research Institute (Yellow Springs, OH) and since 1976 by the Institutional Review Board of Wright State University (Dayton, OH).
Statistical methods
Landmarks of somatic growth
With use of each individual’s serial stature data, we generated growth curves for the 211 girls in the analysis (22). These subjects had a total of 5082 measurements of stature, with an average of 24 measures from each participant (range: 6–44). Each girl had data from age 1 to 20 y and had no more than 3 y between successive observations. A triple logistic model implemented in the statistical program AUXAL (version 2) (23) was applied to each individual’s serial stature data. The indicators estimated in this model included the early childhood minimum height velocity (MHV), the prepubertal MHV, and the adolescent peak height velocity (PHV) (all identified in Figure 1) as well as the stature at each of these ages. Because not all individuals demonstrated a mid-childhood growth spurt (a transient increase in height velocity around age 8–10 y), the childhood MHV could be calculated for only 106 of the 211 girls. The root mean squared error and the maximum error of the model were estimated as goodness-of-fit characteristics for each individual’s data.
|
To examine cohort differences in serial BMI during the pubertal period, mixed effects models (random and fixed effects) were used. Third-degree polynomial models of age (in which age was defined as the number of years before or after menarche: –6, –4, –2, etc) were constructed whereby the effects of age (slope), age2 (acceleration), and age3 (change in acceleration) on BMI were tested as fixed effects. Birth cohort was added as a fixed effect to test for secular trends in BMI. To determine the significance of differences between the birth cohorts in the pattern of change in BMI during puberty, interaction effects between birth cohort and maturational age, maturational age2, and maturational age3 were tested in these models as well. The best model was selected on the basis of the Akaike Information Criterion. Because of the correlated nature of serial data, a correlation structure was specified for the serial measures within individuals. Some commonly used structures, namely, compound symmetric, autoregressive order one, and unstructured, were used to model the covariance structure of repeated measures within subjects. The model specifying an unstructured covariance of the repeated measures correlations was chosen as the best, based on the Akaike Information Criterion and Schwartz Bayesian Criterion.
Effects of the timing of menarche (ie, early, average, or late) on BMI were similarly examined, including the modeling of the repeated measures correlation structure, but in this set of analyses the coefficients for the age, age2, and age3 terms were allowed to vary between individuals (ie, they were considered random effects). Interaction terms (eg, age by timing of menarche, age2 by timing of menarche) were examined to assess the significance of differences among early, average, and late maturers in the velocity and the acceleration of BMI during adolescence. A more detailed description of the mixed effects models used to describe BMI differences within and between individuals is provided by Guo et al (24).
Because of the multiple comparisons made across the 3 birth cohorts and the 3 age-at-menarche groups, the Dunnett multiple comparison test was performed at each age point (–6, ..., 6) in the mixed models. The Dunnett test adjusts the critical P value for acceptance or rejection of the null hypothesis of no difference as does the Bonferroni multiple comparison test. All analyses were conducted with use of SAS, version 8.2 (25).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
). The early childhood MHV occurred at age 3.8 y on average, the prepubertal MHV occurred at age 9.2 y on average, and the adolescent PHV occurred at age 11.6 y on average. There were no significant differences in the timing of these growth indicators among the birth cohort groups. The estimated stature at those ages likewise remained relatively constant across birth cohorts, indicating no secular trend in either the timing or the magnitude of linear growth in girls of the Fels Longitudinal Study from 1929 to 1983. Furthermore, the age at menarche was not significantly different among the 3 cohorts; mean age at menarche was 12.7 y in cohort 1, 12.8 y in cohort 2, and 12.6 y in cohort 3 (P = 0.57, NS). The linear correlations between birth year and the ages at which the linear growth landmarks or menarche were reached were likewise nonsignificant (results not shown).
|
It can be seen from the plot of mean BMI in early-, average-, and late-maturing girls (Figure 2) that before menarche, the 3 groups were nearly indistinguishable from each other, but differences became apparent after menarche. With use of analysis of variance at each age (–6 to 6), we found that girls with an earlier age at menarche had a higher mean BMI than average- or late-maturing girls at 4 y (P = 0.03) and 6 y after menarche (P = 0.001) but not at any age before. To avoid the multiple comparisons problem inherent in this approach, we then explored this relation with use of a mixed effects model in which the effects of age, age2, age3 (where age is again –6 to 6 y from the age at menarche), the timing of menarche (a categorical variable with 3 levels), and their interactions with each other were simultaneously estimated. The indicator estimates are provided in Table 3. BMI followed a third-degree polynomial function of age during adolescence. That is, the age, age2, age3 terms were all significant. Although the main effect of the timing of menarche was not significant, there were significant interaction effects between the timing of menarche and age, and between the timing of menarche and age2, which indicate that girls with early menarche had both a greater rate of increase in BMI and a significantly higher acceleration rate in BMI during adolescence than girls with average or late age at menarche. As a result, a significantly higher BMI emerged among early maturers but not until after menarche. There were no significant differences in BMI, the rate of change in BMI during adolescence, or the acceleration in BMI between the average- and late-maturing girls.
|
|
). A significant interaction between maturational age and birth cohort (P = 0.003) was found, which indicates that the higher BMI of girls in the most recent birth cohort was magnified as the number of years after menarche increased and they matured (Figure 3). The addition of the timing of menarche indicator to this model explained no further variation in BMI. Additionally, we conducted a chi-squared test and found no difference among the 3 birth cohorts in the proportion of early-, average-, and late-maturing girls (
2 = 1.7, 4 df, P = 0.79). Thus, the increase in pubertal BMI and BMI velocity in girls of the Fels Longitudinal Study seen between cohort 2 and cohort 3 was independent of the age at menarche effect and was not accompanied by a decrease in the mean age at menarche. The magnitude or pattern of the secular trend in BMI also did not differ between early-, average-, and late-maturing girls.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
At a given chronologic age, girls with earlier age at menarche tend to be heavier and fatter than girls with a later age at menarche (7, 9-14, 26-30). Given this association, it was suggested that the upward trend in BMI and obesity prevalence might result in a higher incidence of precocious puberty (3, 4), as well as in a drop in the mean age at menarche in the population as a whole (5, 7). Increased body fatness can theoretically affect female sexual maturation in several different ways; eg, adipose tissue can contribute to the aromatization of androgens into estrogens (31). However, few longitudinal studies have looked at the temporal relation between increases in body fatness and the initiation of menses.
We found that early-maturing girls had both a greater BMI velocity and BMI acceleration than did late- and average-maturing girls, resulting in a divergence between early-maturing girls and the others by 4–6 y after menarche, or young adulthood. Thus, although girls with early menarche eventually became heavier than other girls, they were no heavier than average-maturing and late-maturing girls during their prepubescent period. Because most studies compare early- and late-maturing girls at particular chronologic ages, these distinctions are blurred. This ambiguity arises because early-maturing girls tend to be more advanced in their progress toward both sexual maturity and adult body composition at every chronologic age; therefore, the temporal relation between the onset of menses and increased adiposity can be obscured. Our findings are in line with the work of Legro et al (19), de Ridder et al (17), and Iuliano-Burns et al (32). For example, Legro et al (19) found only a small increase in percentage body fat in the years leading up to menarche but a larger increase after menarche. They concluded that "menarche results from a rapid and saltatory maturation of the reproductive axis, independent of changes in body fat" (19; page 1024). Others also emphasized that pubertal body composition changes are downstream physiologic consequences of increased gonadal steroid production and are not causes or even antecedents of sexual maturation (18, 33).
Relatedly, Davison et al (16) argued that early childhood fatness is implicated in the timing of pubertal development. Specifically, they reported that girls showing indications of early pubertal development at age 9 (eg, elevated estradiol concentration, breast development) had significantly higher percentage body fat at age 5 and 7 y than girls who did not (16). Their interpretation was that "overweight is causally implicated in the onset of earlier pubertal development" (16; page 820). However, the temporal association between higher percentage body fat and pubertal advancement does not necessarily prove causality, because both phenomena could be consequences of a third causative factor such as accelerated maturation of the neuroendocrinologic system that is already affecting these girls at age 5–7 y.
We also examined birth cohort, or secular trend, effects on BMI during adolescence, the average age at menarche, and the average age at which specific physical growth landmarks were attained. We did not see a change in BMI or BMI velocity in girls born between 1929 and 1964, but between cohort 2 (born 1947–1964) and cohort 3 (born 1965–1983) we found a significant increase in the mean BMI across adolescence of 1.2 kg/m2. This birth cohort effect mirrors the trend toward increasing body weight and overweight prevalence observed in national data between the NHANES II (born 1957–1974) and the NHANES III (born 1969–1988) (1). We found an acceleration of BMI in the 2-y window immediately before and after menarche, so that by early adulthood, girls born after 1965 were 2 BMI units heavier than girls born earlier in the 20th century. Although they do not tell a substantially different story about the timing of the obesity epidemic than do the US national data, the longitudinal data from this study provide a picture of how the increase in BMI during female adolescence has shifted over time.
Declines in the mean age at menarche of 4 mo per decade were seen across Western Europe between 1830 and 1950 (34). In our study, both the tempo of linear growth, as measured by the age at which the girls attained known landmarks in stature, and the mean age at menarche were highly stable across the entire 50-y period under investigation, despite a concurrent increase in BMI during and after puberty and major societal changes in lifestyle and nutrition. A slight (2–3 mo) decline in the age at menarche was recently reported for all US girls over the 25-y period between the National Health Examination Survey and the NHANES III (7), although this difference did not reach statistical significance in a separate analysis (8). More substantial declines from the 1970s to the 1990s were reported for the girls in the Bogalusa Heart Study (5). There is as yet too little information to know whether the small downward shift in the national estimates of the median age at menarche for girls born in the 1980s and 1990s is biologically meaningful. Our long-term serial data suggest that, if a decline in the average age at menarche were to occur, we could expect an increase in female obesity prevalence to arise among older adolescents and young adults. That is, the causal arrow seems to lead from earlier menarche toward increased BMI. In addition, we did not see a decrease in the mean age at menarche despite a secular increase in BMI and in peripubertal BMI velocity. Again, it appears that at the population level, secular increases in childhood BMI are not necessarily linked to a decrease in the age at menarche. The effects of birth cohort and the age at menarche on BMI can be independent of each other. However, because this cohort exhibited no secular trend toward higher BMI during the early childhood period, we cannot address the question of whether a large increase in average BMI among very young girls might affect the timing of menarche at the population level.
Limitations of our study include our use of BMI as a measure of body fatness. The BMI is an inferred measure of adiposity that is currently recommended as the preferred screening measure for childhood obesity (35). However, BMI reflects not only body fatness but also relative leg length, frame size, and the amount of lean tissue in the body (36). Furthermore, the degree to which BMI reflects adiposity is known to change during growth and development (37). These limitations can be particularly important if trends over time are associated with reductions in physical activity that result in lower muscle mass (38). We were limited to the use of the BMI for this analysis because skinfold calipers and other more direct methods of assessing body fatness were not developed or introduced into the Fels Longitudinal Study until 1960, 30 y after the study was initiated. Our conclusions are to some extent limited as well by the study population, which was composed entirely of whites. The timing and nature of change in BMI across adolescence could have taken a somewhat different course in other ethnic groups during the past 50 y.
In conclusion, girls with early menarche were found to have no greater BMI than average- and late-maturing girls until after menarche had occurred. In addition, whereas BMI and BMI velocity increased during perimenarche in girls born after 1965 compared with girls born 1929–1964, this upward shift in pubertal BMI was not accompanied by an acceleration in the pace of maturation, measured by either the timing of menarche or the timing of linear growth landmarks. Our results counter the concern that national increases in childhood adiposity necessarily lead to earlier menarche in girls and furthermore demonstrate the utility of long-term serial data for elucidating the temporal relations between body composition, growth, and development.
|
ACKNOWLEDGMENTS |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONSUBJECTS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. E Buyken, N. Karaolis-Danckert, and T. Remer Association of prepubertal body composition in healthy girls and boys with the timing of early and late pubertal markers Am. J. Clinical Nutrition, January 1, 2009; 89(1): 221 - 230. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Zhang, S. Pollack, A. Ghods, C. Dicken, B. Isaac, G. Adel, G. Zeitlian, and N. Santoro Onset of Ovulation after Menarche in Girls: A Longitudinal Study J. Clin. Endocrinol. Metab., April 1, 2008; 93(4): 1186 - 1194. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. C. Nooyens, L. L. Koppes, T. L. Visscher, J. W. Twisk, H. C. Kemper, A J. Schuit, W. van Mechelen, and J. C Seidell Adolescent skinfold thickness is a better predictor of high body fatness in adults than is body mass index: the Amsterdam Growth and Health Longitudinal Study Am. J. Clinical Nutrition, June 1, 2007; 85(6): 1533 - 1539. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. M. Lee, D. Appugliese, N. Kaciroti, R. F. Corwyn, R. H. Bradley, and J. C. Lumeng Weight Status in Young Girls and the Onset of Puberty Pediatrics, March 1, 2007; 119(3): e624 - e630. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Ibanez, C. Valls, K. Ong, D. B. Dunger, and F. de Zegher Metformin Therapy during Puberty Delays Menarche, Prolongs Pubertal Growth, and Augments Adult Height: A Randomized Study in Low-Birth-Weight Girls with Early-Normal Onset of Puberty J. Clin. Endocrinol. Metab., June 1, 2006; 91(6): 2068 - 2073. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Wardle, N. H. Brodersen, T. J Cole, M. J Jarvis, and D. R Boniface Development of adiposity in adolescence: five year longitudinal study of an ethnically and socioeconomically diverse sample of young people in Britain BMJ, May 13, 2006; 332(7550): 1130 - 1135. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. B Pierce and D. A Leon Age at menarche and adult BMI in the Aberdeen Children of the 1950s Cohort Study Am. J. Clinical Nutrition, October 1, 2005; 82(4): 733 - 739. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
