Outcomes of Early Pubertal Timing in Young Women: A Prospective Population-Based Study

The Great Smoky Mountains Study is a longitudinal study of the development of psychiatric disorders in rural and urban youths (14–19). A representative sample of three cohorts of children, ages 9, 11, and 13 at intake, was recruited from 11 counties in western North Carolina using a household equal probability accelerated cohort design (20). The externalizing problems subscale of the Child Behavior Checklist (25) was administered to a parent of the first-stage sample (N=3,896). All children whose scores were in the top 25%, plus a 1-in-10 random sample of the rest, were recruited for detailed interviews. Ninety-five percent of families contacted completed the telephone screen. About 8% of the area residents and the sample are African American, and fewer than 1% are Hispanic; American Indians make up only about 3% of the study area but were oversampled to constitute 25% of the sample. All participants were given a weight inversely proportional to their probability of selection, so that the results are representative of the population from which the sample was drawn. Of all youths recruited, 80% (N=1,420) agreed to participate. The sample was 49% female (N=630). Table 1 presents the study design and participation rates at each wave. In this article, we present data on 4,021 parent-daughter pairs of interviews carried out across an age range of 9 through 21.

The parent and participant were interviewed by trained interviewers separately (as close as possible to the participant's birthday) until participants were 16; at ages 19 and 21, participants were interviewed, but not their parents. Before the interviews began, the parent and child signed informed consent forms approved by the Duke University Medical Center Institutional Review Board. Each parent and child was paid for their participation.

At each assessment up to age 16, self-ratings of current pubertal morphologic status based on the standard Tanner staging system (26) (5-point scale from I [prepubertal] to V [adult]) and age at menarche were assessed. Self-ratings of Tanner stage correlate well with direct observation of pubertal status (27, 28). For the two younger cohorts, early pubertal timing was defined as achieving Tanner stage IV (increased breast size and elevation) by age 12 (18.2% of girls). These physical changes are highly salient to the girls themselves, their family, and their peers. The oldest cohort, however, was first interviewed at age 13. Thus, girls in this group who had already achieved Tanner stage IV or V could not be unambiguously assigned to a pubertal timing group using the Tanner staging method. Instead, these girls were classified as early maturers if they reported an onset of menarche before age 11. The proportion of early maturers identified this way was similar to the proportion identified using Tanner staging. Together, 19.0% of girls (unweighted N=115) were categorized as early maturers.

Outcome status was derived by aggregating observations across two periods: adolescence (ages 13 to 16) and young adulthood (ages 19 and 21). All adolescent data except juvenile justice status were collected through parent and child interviews with the Child and Adolescent Psychiatric Assessment (26, 27). Parent and adolescent reports were combined using an either/ or rule for adolescent outcomes unless otherwise specified. In young adulthood, all outcomes except officially recorded criminal offenses were assessed through participant interviews with the Young Adult Psychiatric Assessment (28). The time frame for both interviews was the 3 months preceding the interview unless otherwise stated. Our goal was to assess a range of domains of functioning within each developmental period (e.g., crime, psychiatric functioning).

All associations were tested using weighted logistic regression models in a generalized estimating equations framework implemented by SAS PROC GENMOD. Robust variance (sandwich type) estimates were used to adjust the standard errors of the parameter estimates for the stratified design effects. Therefore, the parameters reported here are representative of the population from which the sample was drawn. Associations between early puberty status and each outcome were tested in adolescence (ages 13 to 16) and in young adulthood (ages 19 and 21).

Across all waves, an average of 82% of possible interviews was completed, ranging from 75% to 94% at individual waves (Table 1). Three or four assessments (depending on the cohort) were possible in adolescence, and two assessments were possible in adulthood. If missing assessments was associated with pubertal timing, then our analyses might be biased. Pubertal timing, however, was not associated with missing assessments overall or in young adulthood only. The overall statistical trend suggested that early maturers missed fewer assessments than their peers.

Early maturation was associated with at least one indicator in each of the six adolescent outcome domains (10 of 26 specific indicators; see Table 2) and predicted having had any problems in substance use, school/peers, and psychiatric functioning domains (one or more elevated indicators). The strongest associations (odds ratios >3.0) were seen with substance use and conduct problems.

No significant associations were observed between early maturation and any of the five young adult domains (Table 3), suggesting that attenuation of the negative effects of early maturation was the rule. Only three of the 21 specific outcome measures were significantly predicted by early puberty; these included higher rates of lifetime sexual partners, higher rates of young adult depression, and lower levels of illicit drug abuse. Depression was the only current young adult outcome for which early maturers were at increased risk. Depression in young adulthood might reasonably be related to psychiatric status in adolescence (32). Testing associations with both disorders that were predicted by early puberty in adolescence—conduct disorder and depression—suggested that it was adolescent conduct disorder, not depression, that accounted for the higher rates of young adult depression. An interaction term between pubertal timing and adolescent conduct disorder significantly predicted young adult depression (odds ratio=221.8, 95% CI=9.7–5094, p=0.0007), such that 9% of early maturers with no history of adolescent conduct disorder displayed depression in young adulthood, compared with 81% of early maturers reporting a history of adolescent conduct disorder. After accounting for this interaction, the main effect of pubertal timing on young adulthood depression was no longer significant (odds ratio=1.9, 95% CI=0.5–7.4, p=0.38). Overall, there was little evidence of persistent, long-term effects of early pubertal timing, with the exception of a small group of girls at greater risk for depression in young adulthood.

Attenuated effects might suggest either recovery by early maturers or “catch-up” by on-time and late maturers. Longitudinal analyses were conducted for indicators that were measured similarly across developmental periods (e.g., alcohol abuse or dependence, marijuana abuse or dependence, illicit drug abuse or dependence) to test whether they decreased from adolescence to young adulthood for early maturers (recovery) or increased across developmental periods for on-time and late maturers (catch-up). Figure 1 shows that early maturers recovered (i.e., indicators decreased over time) in terms of their illicit drug abuse or dependence, whereas their on-time and late maturing peers caught up (i.e., indicators increased over time) in their alcohol abuse or dependence. Neither the recovery nor the catch-up hypothesis was fully supported for marijuana abuse and dependence.

Despite the overall appearance of attenuation of the adolescent effects of early maturity in young adulthood, it is possible that adverse outcomes persisted for a subset of early maturers, such as those with the highest overall levels of adolescent problems. To test this hypothesis, we compared early maturers who had elevated levels of negative outcomes in adolescence (the upper 25% on a cumulative negative outcome scale, which corresponded to exhibiting five or more negative adolescent outcomes) with on-time maturers who had similar levels of adolescent negative outcomes to determine whether early maturation predicts a more persistent course of disturbance. These groups were compared on the total number of negative adult outcomes as well as on scales for total outcomes related to crime, substance use, socioeconomic status, sexual behavior, and mental health. The most troubled adolescent early maturers did not differ from similarly troubled on-time maturers in any adult outcomes (Table 4).

Finally, it is plausible that early maturers with particular types of adolescent problems are more likely to have persistent problems. In fact, this pattern was already observed for adolescent conduct disorder predicting young adult depression. To further test for this pattern, interaction terms between significant adolescent outcomes and pubertal timing were entered into models predicting each domain of young adult functioning as well as specific young adult psychiatric disorders. There was little evidence that early maturation moderated the effect of adolescent indicators on young adult outcomes. Significant interactions occurred at the rate expected by chance and were as likely to indicate decreased as increased risk for early maturers compared with on-time and late maturers. (Full results are available on request from the first author.)

The Great Smoky Mountains Study has several strengths besides its longitudinal prospective design. The key variable, early puberty, was assessed yearly and prospectively for most girls from middle childhood through adolescence; a population-based design minimized selection biases; and a wide range of functional domains in adolescence and young adulthood were assessed. The sample is not representative of the U.S. population, however; Native Americans are overrepresented and African Americans and Latinos are underrepresented. Also, the gold standard measure of pubertal development involves recurrent physical examinations by a medical professional. Such repeated intrusive assessment, however, would have been financially prohibitive and might have led to unacceptable sample attrition. Fortunately, there is adequate to good reliability between self- or parent-reported physical development and official medical records (38). As evidenced by the findings in adolescence, our pubertal measure was sufficiently sensitive to detect significant effects.