Longitudinal Development of Secondary Sexual Characteristics in Girls and Boys Between Ages 9½ and 15½ Years FREE

Most research on pubertal development is cross-sectional, involving different adolescents assessed at different ages. Such research is not positioned to characterize change in secondary sexual characteristics—genitals, breasts, and pubic hair—that represent different physiological processes. The available longitudinal evidence is outdated or covers a limited age span.¹ Current information on timing of puberty (ie, age at which adolescents are in each sexual maturity stage) and change in secondary sexual characteristics (ie, how quickly individuals progress from one stage to the next) is important for understanding the etiology of pubertal disorders as well as normal development. It is also important to understand individual-level synchrony (ie, characteristics developing concurrently) or asynchrony (ie, one characteristic develops before the other) in pubertal development, given that asynchrony has implications for health problems.²

This investigation was motivated by new findings² and previous work.^3- 5 It is not known whether patterns of thelarche and pubarche established at pubertal onset are sustained longitudinally, or whether patterns of asynchrony are observed among boys. The present study aimed to (1) identify ages when adolescents were in sexual maturity stages 2 through 5 (Tanner criteria)^6- 8 for each secondary sexual characteristic; (2) explain the relation of development across pubertal characteristics; and (3) evaluate synchrony of pubertal development across characteristics. A secondary goal was to investigate whether racial differences in pubertal timing and change are accounted for by family socioeconomic status.

The multisite Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) Study of Early Child Care and Youth Development was designed to evaluate the effects of nonmaternal care on children's functioning. In 1990, 10 university-based sites recruited participants from designated hospitals in Little Rock, Arkansas; Irvine, California; Lawrence, Kansas; Boston, Massachusetts; Philadelphia and Pittsburgh, Pennsylvania; Charlottesville, Virginia; Seattle, Washington; Hickory and Morgantown, North Carolina; and Madison, Wisconsin. Children were followed up from birth to age 15½ years (1991-2006) with a common study protocol. Annually from ages 9½ to 15½ years (2000-2006), staging of secondary sexual characteristics was performed by a nurse practitioner or physician. Institutional review boards at each university and the data coordinating center approved the study. Families gave written consent and children gave assent. See a previously published NICHD Early Child Care Research Network study⁹ for detailed recruitment and selection procedures and http://secc.rti.org for study procedures.

Healthy newborns of English-speaking mothers were recruited. Attempts were made to contact 3015 families meeting enrollment eligibility requirements 2 weeks after birth. Of 1526 families contacted, 1364 (89.4%) completed the 1-month home visit before becoming study participants. There were no significant differences between these 1364 families and the 1990 US population,¹⁰ except that the study sample had slightly more married couples than the US population (76.7% vs 74.2%; P = .04).

From enrollment to age 9½ years, approximately 20% of the sample was lost to follow-up; moving and time constraints were the major reasons for attrition. Dropouts were more likely to have lower maternal education and to be nonwhite. Between ages 9½ and 15½ years, the cohort lost an additional 8.1%, but this attrition was random in terms of sex, annual family income, years of maternal education, or race/ethnicity.

Child sex, race/ethnicity, and years of maternal education were reported at 1 month. Only white and black participants are included in this report because the numbers of other minorities were too small. Family income when the child was 9½ years old was converted to an income-to-needs ratio based on federal poverty guidelines (ratio <2.0 represents low income).¹¹

Boys' and girls' sexual maturity stage was assessed using Tanner criteria.^7- 8,12 Girls' sexual maturity staging was based on instructions from the American Academy of Pediatrics Assessment of Sexual Maturity Stages in Girls,⁵ augmented with breast bud palpation (except for the assessment at age 9½ years). The assessment of stage of sexual maturity is based on photos of the 5 sexual maturity stages (1 indicates prepuberty and 5, full sexual maturity) that were used in the Pediatric Research in Office Settings Network study of pubertal development.⁵ Boys' sexual maturity stage was assessed using Tanner's original criteria.⁶ For boys and girls, sexual maturity stage is determined by comparing the adolescent's sexual development with the series of photographs showing sexual maturity as stages of breast, genital, and pubic hair development. Annual examinations identified sexual maturity stage (1-5) for breast development in girls, genital development in boys, and pubic hair development in both. If a child was between 2 stages, the examiners gave the child the lower stage rating. Examinations continued for girls until menarche and stage 5 breast and pubic hair development were achieved and for boys until stage 5 genital and pubic hair development were achieved. Beginning at age 10½, girls reported annually whether menstruation had begun and age at first menses. Annual maternal report was used if daughter’s self-report was missing (n = 117).

Centralized training was provided to all examiners before the first pubertal assessments. Physicians and nurse practitioners were annually recertified (requiring 87.5% accuracy) by rating stages of pubertal development using photos from the Pediatric Research in Office Settings Network study of pubertal development⁵ for girls and Tanner photos for boys.¹³

Participation at each annual examination for 859 children was as follows: age 9½ years, 781 (90.9%); age 10½ years, 700 (81.5%); age 11½ years, 668 (77.8%); age 12½ years, 662 (77.1%); age 13½ years, 645 (75.1%); age 14½ years, 616 (71.7%); and age 15½ years, 598 (69.6%). To address missing-data problems, we imputed 20 data sets (10 each for boys and girls)^14- 16 and conducted analyses separately on each data set. Test statistics and coefficients were averaged across analyses, separately for boys and girls, and coefficient standard errors were calculated by combining within- and between-model variability.¹⁵

All statistical analyses were conducted using SAS statistical software, version 9.1.3 (SAS Institute Inc, Cary, North Carolina); all tests were 2-tailed. Significant differences between children with and without sexual maturity data for all categorical variables were evaluated using χ² tests; continuous variables were evaluated using t tests, assuming unequal variance.

The ages that girls and boys were in each sexual maturity stage (2-5) were estimated for each secondary sexual characteristic in separate logistic regressions: breast (girls), genital (boys), and pubic hair (girls and boys). For this analysis, whether an adolescent was in a particular sexual maturity stage at each assessment was dichotomized (0 vs 1). The probability of being in a stage was estimated at a given age and was modeled using a random coefficient logistic model with the dichotomized indicator of being in a particular stage as the dependent variable and age at each assessment (in years) as the independent variable. The intercept was treated as random (ie, individual-level variation was allowed), whereas the effect of age was fixed. The ages at which the individual had a 50% or greater likelihood of being in the particular sexual maturity stage for breast and pubic hair development (girls) or genital and pubic hair development (boys) were estimated.

Once the ages of being in the various sexual maturity stages for the different secondary sexual characteristics were estimated, differences across time and across characteristics were modeled using a doubly repeated measures analysis of variance. For these analyses, the estimated ages were the dependent variables, and sexual maturity stage (2-5) and secondary sexual characteristics (breast or pubic hair for girls; genital or pubic hair for boys) were repeated (independent) factors. Additional models added between-subjects factors for menarche (girls only), race (black vs white), years of maternal education, and family income-to-needs ratio. Finally, the length of time (in years) that girls and boys remained in puberty was calculated as the difference between the ages that individuals were estimated to be in sexual maturity stage 5 and sexual maturity stage 2 (eAppendix).

The estimated age that each individual was predicted to be in each sexual maturity stage for breast (girls) or genital (boys) and pubic hair (girls and boys) was used to determine synchrony of pubertal maturation. Given that there is no consensus about what qualifies as a synchronous pattern, we defined synchronous development as secondary sexual characteristics being in the same sexual maturity stage within approximately 4 months of each other. This standard was based on prior longitudinal work.¹⁷ In an individual, if the difference in the estimated age of the 2 secondary sexual characteristics was more than 4 months, development was classified as breast or genital first or pubic hair first. Using age estimates obtained from the random coefficients logistic regression, the differences between the estimated ages for breast vs pubic hair development (girls) and for genital vs pubic hair development (boys) were calculated. Children were then classified as to whether they exhibited synchrony (difference ≤4 months) or asynchrony (breast or genital development >4 months earlier than pubic hair development or pubic hair development >4 months earlier than breast or genital development) for each sexual maturity stage.

As shown in Table 1, 373 white girls (43.4%), 59 black girls (6.9%), 364 white boys (42.4%), and 63 black boys (7.3%) had valid pubertal measurements for at least 1 of the 7 assessments (taken annually from ages 9½ to 15½ years), thereby qualifying them for inclusion in the study. Girls had more valid data at all ages than did boys. Overall, 373 girls (86.3%) and 364 boys (85.2%) were white, 93 girls (23.4%) and 98 boys (24.5%) were classified as low income at age 9½ years, and mother's mean (SD) years of education was 14.5 (2.4) for girls and 14.3 (2.4) for boys. (The low income percentages are based on numbers of participants who had income data at age 9½ years [ie, 397 of 432 girls and 400 of 427 boys].) Maternal age at the time the child was born was greater for the analysis group (mean, 28.9 years; 95% confidence interval, 28.4-29.4 years) than for girls not included in the analysis (27.7 years; 26.7-28.6 years). No significant differences existed for race, early family income-to-needs ratio, or years of maternal education. For boys, no significant differences emerged on any of these background factors for those included and not included in the analyses.

The eFigure shows the distribution of each sexual maturity stage, by age and race, for breast and pubic hair development for girls and for genital and pubic hair development for boys. At age 9½ years, most white girls showed no evidence of breast (56.2%) or pubic hair (71.5%) development; in contrast, most black girls showed some evidence of breast (69.0%) or pubic hair (59.2%) development. By age 15½ years, most white girls were in sexual maturity stage 5 for breast (79.5%) or pubic hair (81.9%) development; similarly, most black girls were in sexual maturity stage 5 for breast (84.1%) or pubic hair (92.5%) development. At age 9½ years, more than two-thirds of white boys showed no evidence of puberty for genital (72.8%) or pubic hair (91.9%) development; less than half of the black boys showed some evidence of puberty for genital (47.9%) or pubic hair development (33.8%). At age 15½ years, most white boys were in sexual maturity stage 5 for genital (67.3%) or pubic hair (58.7%) development; similarly, most black boys were in sexual maturity stage 5 for genital (81.0%) or pubic hair (71.3%) development.

Overall, girls' breast development occurred earlier than pubic hair development (F_1,431 = 52.25; P <.001). This effect, however, was moderated by the stage being examined (F_3,1293 = 53.29; P <.001). Girls were in stage 2 breast development a mean of 5.0 months earlier than stage 2 pubic hair development and were in stage 3 breast development a mean of 2.2 months earlier than stage 3 pubic hair development. There were no statistically significant differences between breast and pubic hair development in terms of the ages at which girls were in stages 4 and 5 (Table 2).

On average, girls took 1.5 years to develop from one sexual maturity stage of breast development to the next; girls took slightly less time (mean, 1.3 years) to develop from one sexual maturity stage of pubic hair to the next. Overall, girls took between 4 and 4½ years to go from sexual maturity stage 2 (beginning of puberty) to sexual maturity stage 5 (full maturity) (Table 3).

When age of menarche was added to the repeated-measures analysis of variance as a between-subjects factor, we found a significant main effect for age of menarche (F_1,407 = 435.72; P <.001), and significant interactions between age of menarche and sexual maturity stage (F_3,1221 = 7.62; P <.001) and between age of menarche, characteristic (breast or pubic hair), and sexual maturity stage (F_1,1221 = 4.54; P = .004). As expected, if girls were older when they experienced menarche they were also older when in all sexual maturity stages for breast and pubic hair development (B's range, 0.41-0.48 for age at breast development stages 2-5 and 0.36-0.53 for age at pubic hair development stages 2-5; P < .001 for all). (B indicates unstandardized regression coefficient.) Age of menarche had a stronger positive relationship with age at developmental stages 3 and 4 (B's range, 0.47-0.53) than it did with age of being in stage 2 (B = 0.41 for breast development and 0.36 for pubic hair development). The relationship between age of menarche and age at breast development stage 5 declined to a B of 0.42, whereas the relationship between age of menarche and age at pubic hair development stage 5 remained relatively higher (B = 0.48).

Overall, boys were in each sexual maturity stage for genital development before they were in the same stage for pubic hair development (F_1,426 = 821.26; P <.001); this effect was moderated by the stage being examined (F_3,1278 = 411.40; P <.001). Stage 2 genital development was earlier than stage 2 pubic hair development by a mean of 1.1 years; stage 3, 3.9 months earlier; stage 4, 1.5 months earlier; and stage 5, 2.3 months earlier (Table 4).

On average, boys took 2 years to go from stage 2 to stage 3 genital development and 1.3 years to go from stage 2 to stage 3 pubic hair development. The change from stage 3 to stage 4 was shorter, taking 1.1 years for genital development and 10.8 months for pubic hair development. The change from stage 4 to stage 5 took approximately 1.4 years for both genital and pubic hair development. Overall, boys took approximately 4½ years and 3½ years to go from sexual maturity stage 2 (beginning of puberty) to sexual maturity stage 5 (full development) for genital and pubic hair, respectively (Table 5).

At the onset of puberty, most girls (66.2%) were estimated to be at sexual maturity stage 2 for breast development at least 4 months before stage 2 for pubic hair development (ie, breast-first pattern) (Table 6). At later maturity stages, girls' development tended to become more synchronous, with 51.4% estimated to be in sexual maturity stage 5 for breast and pubic hair development within approximately 4 months of each other.

At the onset of puberty, most boys (91.1%) were estimated to be in sexual maturity stage 2 for genital development at least 4 months before stage 2 for pubic hair development (ie, genital-first pattern) (Table 6). At later maturity stages, boys' development remained asynchronous, with 74.7% still showing the genital-first pattern for being in sexual maturity stage 5.

On average, black girls were in each sexual maturity stage for breast and for pubic hair development approximately 9 months earlier than white girls (Table 2) (F_1,430 = 72.05; P <.001). The other relationships described regarding differences in ages of being in each stage for breast and pubic hair development and the length of time between stages showed no significant relations with race. These findings remained the same when controlling for family income-to-needs ratio (measured at age 9½ years) and years of maternal education at the time of participants' birth. Only one interaction between puberty characteristic and family income-to-needs ratio was significant (F_1,393 = 8.33; P = .004) and indicated that girls from higher-income homes experienced later breast development than did girls from lower-income homes.

Similar to black and white girls, black boys were in each stage of genital and pubic hair development earlier than white boys (Table 4) (F_1,425 = 89.07; P <.001). The discrepancy between ages that black and white boys were in each stage was greater for earlier stages (range, 9.6 months-1.0 years earlier for stage 2) than it was for later stages (7.2 months earlier for stage 5) (F_3,1275 = 7.36; P < .001). With demographic factors controlled, these results remained unchanged.

This report, to our knowledge, presents one of the most detailed longitudinal studies of pubertal development in boys and girls conducted to date. Yearly assessments of breast, genital, and pubic hair growth allowed us to analyze the ages at which adolescents were in each sexual maturity stage and the time it took to go from one stage to the next. We also examined synchrony and asynchrony in secondary sexual characteristics in white and black youth.

Breast development in girls and genital development in boys tended to begin before pubic hair development. But pubic hair development occurred faster across the stages than did genitals or breasts. In addition, synchrony between breast or genital and pubic hair stages tended to increase with advancing stages.

Our findings are consistent with cross-sectional studies showing that girls' pubertal development occurs earlier than boys’, a putative consequence of earlier estrogen synthesis.¹⁸ Our observation that breast development in girls and genital development in boys precedes pubic hair development is also consistent with previous studies,^1- 2,17 with some exceptions.³

Although Biro et al² estimated asynchrony in early puberty, we extended this analysis by assessing synchrony or asynchrony at 4 sexual maturity stages (2-5). At the early stages of puberty, we typically observed asynchrony between breast and pubic hair stage in girls and genital and pubic hair stage in boys. At later stages of sexual maturity, although development in boys remained asynchronous, breast development and pubic hair development in girls tended to become more synchronous. The pattern in girls is believed to result from the earlier timing of hormonal changes, starting with the initiation of cyclic gonadotropin-releasing hormone of a magnitude sufficient to stimulate gonadotropin and estrogen production,¹⁹ which stimulates growth of breast tissue. Pubic hair growth is stimulated by testosterone and adrenal hormones but appears to respond more slowly to neuroendocrine changes in early puberty. Girls who follow the breast-first pattern may have higher estrogen levels and be more at risk as adults for estrogen-related diseases, such as breast cancer, than those following a pubic hair–first pattern.

Findings indicate that, at any age, black adolescents are more advanced than white adolescents in pubertal development, consistent with prior research.^5,20- 21 The earlier onset of puberty in black youth has been attributed to genetic factors, higher body mass index, and nutritional factors.²⁰ No ethnic differences were observed in the length of time it took to progress from one sexual maturity stage to the next for girls. In contrast, sexual development in black boys took longer than in white boys.

These detailed findings suggest that pubertal development cannot be described exclusively in terms of a simple onset and termination because different secondary sexual characteristics develop on different schedules that show varying degrees of timing, depending on sex and ethnicity. For future research, the present findings raise the question of whether previously reported risks for morbidity related to the timing of puberty onset (eg, metabolic syndrome and reproductive malignancies²²) are associated solely with entry into puberty or whether continuing asynchrony in secondary sexual characteristics might also be associated with morbidity.

Although drawn from sites distributed throughout the United States, and reflecting a range of economic, social, and educational backgrounds, the sample is not representative of the US population because it is not a probability sample. The findings may also be biased because of dropouts and missing data. Otherwise-eligible children whose mothers were known drug users or who did not intend to continue living near a study site were excluded from the study. An additional limitation of the study is that at the assessment at age 9½ years, we did not use breast bud palpation, which may have led to an underestimation of the percentage of girls at developmental stage 2. Nonetheless, this research used a highly reliable protocol for assessing pubertal development.

The descriptive findings from this study are useful in understanding normative variation in the timing and change in secondary sexual characteristics at puberty. They will help identify children with atypical changes in sexual maturation and unusual progression of sexual maturation and growth disorders. Replication of this study in a representative sample is now needed.

Correspondence: Elizabeth J. Susman, PhD, Department of Biobehavioral Health, The Pennsylvania State University, Health and Human Development Building, Room E-315, University Park, PA 16802 (ejs5@psu.edu).

Accepted for Publication: July 29, 2009.

Author Contributions: Drs Susman, Houts, Steinberg, Belsky, and Roisman had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Susman, Houts, Steinberg, Belsky, Cauffman, DeHart, Friedman, and Roisman. Acquisition of data: Susman, Steinberg, Belsky, Cauffman, and DeHart. Analysis and interpretation of data: Susman, Houts, Steinberg, Belsky, Roisman, and Halpern-Felsher. Drafting of the manuscript: Susman, Houts, Steinberg, Belsky, DeHart, and Friedman. Critical revision of the manuscript for important intellectual content: Susman, Houts, Steinberg, Belsky, Cauffman, DeHart, Roisman, and Halpern-Felsher. Statistical analysis: Houts, Steinberg, and Halpern-Felsher. Obtained funding: Susman, Steinberg, Belsky, Cauffman, Friedman, and Roisman. Administrative, technical, or material support: Cauffman, DeHart, Friedman, and Halpern-Felsher. Study supervision: Steinberg and DeHart.

Eunice Kennedy Shriver NICHD Early Child Care Research Network Members: This study was directed by a steering committee and supported by NICHD through a cooperative agreement (U10), which calls for scientific collaboration between the grantees and the NICHD staff. Current members of the Steering Committee of the NICHD Early Child Care Research Network, listed in alphabetical order, are: Dr Belsky; Cathryn Booth-LaForce, PhD, University of Washington, Seattle; Robert H. Bradley, PhD, University of Arkansas at Little Rock; Celia A. Brownell, PhD, University of Pittsburgh; Margaret Burchinal, PhD, The University of North Carolina at Chapel Hill; Susan B. Campbell, PhD, University of Pittsburgh; Dr Cauffman; Alison Clarke-Stewart, PhD, University of California, Irvine; Martha Cox, PhD, The University of North Carolina at Chapel Hill; Robert Crosnoe, PhD, The University of Texas, Austin; James A. Griffin, PhD, NICHD Scientific Coordinator, Bethesda, Maryland; Dr Halpern-Felsher; Willard Hartup, PhD, University of Minnesota, Minneapolis; Kathryn Hirsh-Pasek, PhD, Temple University, Philadelphia; Daniel Keating, PhD, University of Michigan, Ann Arbor; Bonnie Knoke, MS, RTI International, Research Triangle Park; Tama Leventhal, PhD, Tufts University, Boston; Kathleen McCartney, PhD, Harvard University, Cambridge, Massachusetts; Vonnie C. McLoyd, PhD, The University of North Carolina at Chapel Hill; Fred Morrison, PhD, University of Michigan, Ann Arbor; Philip Nader, MD, University of California, San Diego; Marion O’Brien, PhD, The University of North Carolina, Greensboro; Ross Parke, PhD, University of California, Riverside; Robert Pianta, PhD, University of Virginia, Charlottesville; Kim M. Pierce, PhD, University of Wisconsin-Madison; A. Vijaya Rao, PhD, RTI International; Dr Roisman; Susan Spieker, PhD, University of Washington; Dr Steinberg; Dr Susman; Margaret Tresch Owen, PhD, The University of Texas at Dallas; Deborah Lowe Vandell, PhD, University of California, Irvine; and Marsha Weinraub, PhD, Temple University.

Financial Disclosure: None reported.

Funding/Support: This study was supported by the NICHD.

Role of the Sponsor: The NICHD provided scientific oversight of the study through scientific directors Sarah L. Friedman, PhD (1991-2006), and James Griffin, PhD (2006-2009).

Disclaimer: The content is solely the responsibility of the named authors and does not necessarily represent the official views of the NICHD, the National Institutes of Health, or individual members of the network.