Ethnic Differences in the Presence of Secondary Sex Characteristics and Menarche Among US Girls: The Third National Health and Nutrition Examination Survey, 1988–1994

METHODS

The NHANES III data set was used for study purposes. Briefly, this survey used a stratified multistage probability sampling design to obtain nationally representative information on the health and nutritional status of the US population using interviews and direct examinations. The NHANES III includes information from 39 695 individuals and was designed to represent the noninstitutionalized civilian US population who were aged 2 months or older and resided in the 50 states and the District of Columbia between 1988 and 1994. The NHANES III designates 4 racial/ethnic groups based on self-reported information given by the primary respondent in the screening and family interview portion of the survey: non-Hispanic white, non-Hispanic black, Mexican American, and other (ie, Hispanics not of Mexican origin or non-Hispanics from racial groups other than white or black). For our purposes, we used only the first 3 racial/ethnic groups because there were only 86 other race/ethnic girls with information on Tanner staging and other variables of interest. Detailed information on the sample design and operation of NHANES III can be found elsewhere.^9,10

There were 7050 girls aged 1 to 16 years in the youth sample of NHANES III, of whom girls aged 8 to 16 years were eligible for pubertal assessment. Physicians used the Tanner staging¹¹ to determine the status of pubic hair and breast development. Staging for breast development was done during the auscultation of the heart with the examinee in the seated position, and pubic hair was evaluated in the supine position during the clinical examination. Details on performing the assessments are described in the Physician Examiner’s Training Manual.¹⁰ Refusal rates for the Tanner staging varied by girl’s age with the lowest rates encountered for girls aged 8 years and the highest rates for girls aged 15 years. Specifically, refusal rates ranged from 8% to 15% for pubic hair staging and from 6% to 13% for breast staging. Girls aged 10 to 16 years were asked about whether they had attained period/menstrual cycle and their age (years) at first period/menstrual cycle. Accordingly, this information was used in this study to define menarche status (both the prevalence of having attained menarche at the time of NHANES III examination for girls of various ages and self-reported age at menarche). In sum, 1623 girls aged 8 to 16 years (466 non-Hispanic white, 589 non-Hispanic black, and 568 Mexican American) were available for the analysis of pubic hair and breast development, and 1168 girls aged 10 to 16 years (330 non-Hispanic white, 419 non-Hispanic black, and 419 Mexican American) for the analysis of menarche.

Tanner stages for pubic hair and breast development range from I to V.¹¹ In accordance with previous reports, presence of pubic hair or breast development was defined in this study if Tanner stage II or higher was met at the time of the NHANES III examination. Briefly, pubic hair was said to be present when there was either sparse growth of long, slightly pigmented downy hair or straight or only slightly curled hair seen along the labia. Breast development referred to the elevation of breast and papilla at least as small mounds.

Race/ethnicity, presence/absence of the pubertal milestone, age at NHANES III examination (recorded in months), and self-reported age at menarche (recalled in years) are the primary variables of interest. Other variables such as family size, rural/urban residency, poverty income ratio, and current BMI were included in the study to describe the characteristics of the sample and to serve as covariates. These variables were defined on the basis of the operational definitions set forth in the NHANES III. Specifically, family size referred to the number of family members who were living in the household; and rural/urban residency referred to the rural/urban code based on US Department of Agriculture codes that described metropolitan and nonmetropolitan counties by degree of urbanization and nearness to metropolitan areas.¹² These codes were later recoded into 2 categories in NHANES III data set to ensure confidentiality of data: 1) metropolitan, ie, central counties of metropolitan areas of 1 million population or more or fringe counties of metropolitan areas of 1 million population or more, or 2) nonmetropolitan, all other areas. The poverty income ratio was the ratio of the reported family incomes divided by the poverty threshold, which was produced annually by the Census Bureau and adjusted for changes caused by inflation. BMI (kg/m²) was computed using weight (in kilograms) divided by height (in meters) squared.

The percentages of girls who had pubic hair and breast development and had reached menarche were compared across racial/ethnic groups. Mean age at the onset of pubic hair, breast development, and menarche was estimated on the basis of the status quo data with the application of the probit model, assuming that the onset of each pubertal milestone followed a normal distribution.¹³ Multivariate logistic regression analysis was used for analyzing race/ethnicity with the presence/absence of each pubertal characteristic adjusting for family size, rural/urban residency, family income ratio, and current BMI. Mean age at menarche was also estimated on the basis of the self-reported age at menarche using failure time model under the assumption of a normal distribution of the event.¹⁴ For each girl who had reached menarche, 0.5 years was added to the self-reported age at menarche to reduce the potential reporting bias. For girls who did not obtain menarche, age at NHANES III examination was treated as the end of follow-up time for the analysis. A Cox proportional hazard model was used to examine the association of self-reported age at menarche with race/ethnicity after adjustment for the covariates.¹⁴ The NHANES III used disproportional sampling with stratification and clustering, which generally increases the variance for the point estimates as compared with assuming simple random sampling in the analysis.¹⁰ To address the complex sampling strategy used in NHANES III, we used the SUDAAN statistical software¹⁵ and weights for the NHANES III examination sample to estimate the percentage of girls with a pubertal characteristic and for multivariate analysis. Statistical software that takes into account the complex sampling feature of NHANES III is not readily available for the estimation of the mean age at onset of a event by the methods described above; therefore, the estimation of mean age at onset of each pubertal measure was performed both with and without the application of weights for the NHANES III examination sample using SAS statistical software.¹⁶

RESULTS

Select characteristics for the study sample are shown in Table 1. In general, the study sample was similar to the NHANES III sample with regard to the distributions of family size, rural/urban residency, family income level, and current BMI. Racial/ethnic differences were apparent in both the NHANES III sample and the study sample. Compared with white girls, black and Mexican American girls were more likely to have a larger family, lower family income, and higher BMI and to live in metropolitan areas.

Black girls were more likely to have pubic hair, breast development, and attainment of menarche at younger ages than white girls; Mexican American girls were generally between black and white girls with respect to the prevalence of each pubertal milestone at younger ages (Table 2). For example, 52.7% of black girls aged 9 years had pubic hair compared with 26.6% of Mexican American and 23.2% of white girls. Although 35.1% of black and 22.0% of Mexican American girls reported obtaining menarche at 11 years of age, only 11.8% of white girls reported obtaining menarche at this age. The observed racial/ethnic difference became smaller as girls aged.

Mean ages at the onset of pubic hair or breast development and menarche are estimated by race/ethnicity (Table 3). On average, black girls reached each pubertal milestone at a younger age than white girls and, to a lesser extent, Mexican American girls. Specifically, the weighted mean ages at onsets of pubic hair, breast development, and menarche, estimated on the basis of the status quo data, were 9.5, 9.5, and 12.1 years, respectively, for black girls; 10.3, 9.8, and 12.2 years, respectively, for Mexican American girls; and 10.5, 10.3, and 12.7 years, respectively, for white girls. The weighted mean ages at menarche, estimated on the basis of the reported age at menarche, were 12.3 years for black girls, 12.5 years for Mexican American girls, and 12.7 years for white girls.

Results from multivariate analyses are presented in Tables 4 and 5. In all models, white girls are used as the reference group. Black girls were significantly more likely to have each of the 3 puberty characteristics than white girls even after adjusting for socioeconomic status, as indicated by the odds ratios and hazard ratios. When BMI was added as a covariate, the odds ratio of having obtained menarche for black girls in comparison with white girls lost statistical significance (Table 4), whereas the hazard ratio of menarche onset remained significant (Table 5). The odds ratios of having obtained menarche and breast development were significantly higher for Mexican American girls than for white girls after adjustment for the social and economic variables (Table 4). However, after adjusting for BMI (Table 4) or in the failure time model of the reported age at menarche (Table 5), Mexican American girls were not significantly more likely to have achieved menarche or pubic hair in comparison with white girls. The social and economic variables studied were not independently associated with the measures of pubertal development (results not shown), and the point and interval estimates changed little after their inclusion in the models. Inclusion of current BMI had some influence for the attainment of menarche as noted on Table 4.

DISCUSSION

This study of a national sample confirms that the onset of pubic hair, breast development, and menarche begins earlier on average among black girls than among white girls. Social and economic factors studied here do not seem to account for these observed differences, although current BMI was associated with having achieved menarche. Values of the point estimates for Mexican American girls lie between black and white girls for mean age at onset of each milestone. These findings suggest that race/ethnicity should be considered when discussing the average ages at which girls begin and attain pubertal maturation. This underscores the need for normative data for subgroups of the US female population.

Studies of pubertal maturation of girls in the US are scattered.^1–4,17,18 As early as 1948, Reynolds and Wines¹⁷ reported that mean ages at onset of pubic hair, breast development, and menarche were 11.0, 10.8, and 12.9 years, respectively, based on a follow-up study of 49 white girls. Another follow-up study of 18 white girls found that mean ages were 11.9, 11.2, and 13.3 years for the onset of pubic hair, breast development, and menarche.¹⁸ In 1980, Harlan et al¹ analyzed a national sample of girls who were aged 12 to 17 years and participated in the US Health Examination Survey, and reported a mean age at menarche of 12.8 years for white girls and 12.5 years for black girls. In 1989, Villarreal et al² conducted a cross-sectional study of 699 Mexican American girls aged 10 to 17 years and reported that mean ages at onset of pubic hair and breast development were 11.3 years and 11.0 years, respectively. In a more recent report, Herman-Giddens et al³ studied 17 077 girls aged 3 to 12 years who were seen in a sample of pediatric practices from across the United States. They reported that the mean ages at onset of pubic hair for black and white girls were 8.78 years and 10.51 years, respectively, and 8.87 years and 9.96 years, respectively, for breast development. Menarche occurred on average at 12.16 years in black girls and 12.88 years of age in white girls. The generalizability of results from previous studies to the US population is limited because of small sample size, late ages of girls at entry, or use of participants from clinical settings.

Our study used data from a nationally representative sample; thus, the results are more likely to be generalizable to the US population. In addition, we examined the role of social and economic factors such as family size, family income level, and urban residence, because these variables have been previously reported to be associated with pubertal development.^7,8,19,20 However, we did not observe a significant association between measures of pubertal development and social and economic factors as measured by family size, rural/urban residency, and the poverty income ratio, and these social and economic factors could not explain the observed racial/ethnic differences in pubertal development. This suggests that general social and economic status plays a more minimal role in influencing sex maturation of US girls or that these covariates did not fully adjust for social and economic status as intended.

Improvement in energy intake and nutrition has been linked to an earlier age at menarche.^5,6,11 Body fat mass is hypothesized to play an important role in the initiation of and progression toward puberty.^16,21 Therefore, nutritional status or body size might account for the ethnic difference observed in the study. Analysis of data from NHANES III demonstrated that black girls tended to be heavier than white girls of the same age before they enter puberty.²² To attempt to address this concern, we adjusted for current BMI in the final stage of our multivariate analysis. The NHANES III data are limited in this respect, because BMI before the onset of the pubertal milestone is not available given the cross-sectional nature of this survey. Adding current BMI to the models might represent a statistical overadjustment because having attained puberty is associated with a greater current BMI. Nevertheless, the inclusion of current BMI in our analyses did not modify the relation between race/ethnicity and pubertal status except for the attainment of menarche. The role of prepubertal nutrition cannot be adequately addressed with the available data in the NHANES III.

Human evidence exists that exposure to endocrine-disrupting agents such as polybrominated biphenyls can induce earlier onset of puberty as measured by pubic hair, breast development, or menarche.^23–26 Changes in menstruation have been reported for women who have been exposed to lead or polychlorinated biphenyls via sport fish consumption.^27,28 Neither exposure was available for analysis in our multivariate modeling, despite recognition that minority populations may have higher environmental contaminant exposure. It is possible that nonwhite girls might be more likely to be exposed to these agents and, as a result, reach puberty earlier than white girls. However, we do not have data in this study to examine this important issue.

Several limitations should be kept in mind when interpreting our study findings. First, the study was based on the analyses of cross-sectional and retrospective data. This approach has been widely used but is different from a cohort study in which the actual onsets of an event are followed. The validity of the study results thus is dependent on the validity of the models used and the accuracy of recall. This limitation may explain why the mean ages at menarche for black and Mexican American girls estimated from probit model were slightly different from the ones estimated from the failure time model. Second, we were unable to adjust fully for the unique sampling features of NHANES III in the estimation of the confidence interval for the mean age at onset of a puberty measure. However, our analysis was stratified by race/ethnicity, a major factor determining the sampling frame of NHANES III. The weighted and unweighted point estimates for the mean age at onset of a puberty measure were very similar if not identical, indicating that consideration of the sampling feature of NHANES III does not make an important difference in these analyses

CONCLUSION

Data from NHANES III suggest that girls enter puberty at slightly different chronological ages based on their reported race and ethnicity. In general, black girls enter puberty first, followed by Mexican American and non-Hispanic white girls. The racial/ethnic differences seem to be independent of select social and economic factors. This finding underscores the need for normative data for subgroups of the US female population. Additional studies are warranted to elucidate factors, particularly prepubertal nutrition and exposure to endocrine-disrupting agents, that may account for the racial/ethnic difference in the pubertal development.