Abstract
Objective
To examine the association of breastfeeding or its duration with timing of girls’ pubertal onset, and the role of BMI as a mediator in these associations.
Methods
A population of 1,237 socio-economically and ethnically diverse girls, ages 6–8 years, was recruited across three geographic locations (New York City, Cincinnati, and the San Francisco Bay Area) in a prospective study of predictors of pubertal maturation. Breastfeeding practices were assessed using self-administered questionnaire/interview with the primary caregiver. Girls were seen on at least annual basis to assess breast and pubic hair development. The association of breastfeeding with pubertal timing was estimated using parametric survival analysis while adjusting for body mass index, ethnicity, birth-weight, mother’s education, mother’s menarcheal age, and family income.
Results
Compared to formula fed girls, those who were mixed-fed or predominantly breastfed showed later onset of breast development (Hazard Ratios 0.90 [95% CI, 0.75–1.09] and 0.74 [95% CI, 0.59–0.94], respectively). Duration of breastfeeding was also directly associated with age at onset of breast development (p trend = 0.008). Associations between breastfeeding and pubic hair onset were not significant. In stratified analysis, the association of breastfeeding and later breast onset was seen in Cincinnati girls only.
Conclusion
The association between breast feeding and pubertal onset varied by study site. More research is needed about the environments within which breastfeeding takes place in order to better understand whether infant feeding practices are a potentially modifiable risk factor that may influence age at onset of breast development and subsequent risk for disease in adulthood.
Keywords: Puberty, Puberty-early onset, Breastfeeding, Body Mass Index
Introduction
The documented decline in girls’ age at onset of puberty and its clinical and social consequences is a growing public health concern (1–4). Early pubertal maturation not only affects adolescent behavioral and emotional health, but some pubertal events, such as age at menarche, are established risk factors for cancer later in life (2, 4). In addition, studies have shown that early pubertal timing is associated with endocrine-related disorders, breast cancer (4), metabolic syndrome, psychosocial and other health problems (5). The timing of puberty is known or suspected to be influenced by various factors including genetics, body mass index (BMI), endocrine disrupting chemicals, nutritional factors, and socio-cultural determinants including race/ethnicity (6–8). Moreover, peri-natal factors such as maternal smoking during pregnancy, low birth weight, and infant weight gain have been found to influence the timing of pubertal onset (9, 10).
Early infant nutrition and feeding practices may also influence age at pubertal onset. These are some of the few potentially modifiable targets for early intervention; however, research in this area has been scarce. One recent study conducted in Hong Kong found no association between breastfeeding and onset of breast development in Chinese girls (11). In another study, an association of exclusive breastfeeding with later attainment of menarche was reported in a prospective birth cohort of Filipino girls (12). A retrospective study in the U.K. observed a direct association of delayed menarche among women who were breastfed as infants (13). Others have shown that girls who are formula fed have higher body fat deposition and earlier attainment of menarche compared to those who were breastfed (14). As breastfeeding has been associated with a decrease in overweight and obesity in infancy and early childhood (15, 16), and overweight is linked to younger onset of puberty (17), it is reasonable to suggest that breastfeeding may be associated with later onset of puberty. In addition, alternative pathways may exist that can affect pubertal timing such as protective factors in breast milk composition (18, 19), or associations of breastfeeding with behavior patterns in mothers that promote health.
This study is the first known longitudinal study of girls in the U.S. that estimates associations between early infant feeding and their timing of pubertal onset, as determined by physical examination. The objectives were [1] to assess whether breastfeeding or its duration was associated with timing of girls’ pubertal onset, and [2] to examine the role of BMI as a mediator in these associations. We hypothesized that girls who were predominantly breastfed, and those who were breastfed for longer durations, would enter puberty at later ages, compared to girls who were formula-fed or were breastfed for shorter durations. We also expected that breastfeeding would be associated with girls’ lower BMI, which in turn, would be associated with later pubertal onset.
Methods
Data originated from the Puberty Studies of the Breast Cancer and the Environment Research Program (BCERP), a consortium of three collaborative prospective studies examining predictors of early sexual maturation in girls (20). In these studies, 1,237 socio-economically and ethnically diverse sample of girls, ages 6–8 years, were enrolled between 2004–2007 from three locations led by the following institutions: [1] Mount Sinai School of Medicine (MSSM), with recruitment through clinics, schools, and neighborhood centers from East Harlem, New York; [2] Cincinnati Children’s Hospital Medical Center, with recruitment through public and parochial schools in the Cincinnati metropolitan area and through the Breast Cancer Registry of Greater Cincinnati; and [3] Kaiser Permanente Northern California (KPNC), with recruitment of Kaiser Health Plan members in the San Francisco Bay Area. Informed consent was obtained from the primary caregiver and child assent from the study participant. Study procedures were approved by the Institutional Review Boards at each participating institution. Eligibility criteria included age (6–8 years), female sex, no underlying endocrine medical conditions, and at MSSM, black or Hispanic race/ethnicity. This collaborative project is an ongoing prospective cohort study, and the present analyses use data from the first six years.
Data Collection and Variable Definition
Data were collected annually in California and New York and semi-annually in Cincinnati during in-person clinic visits that included child anthropometry and pubertal assessment. Collection of information on other factors using interviews or self-administered questionnaire with the caregiver was conducted annually at all three sites.
Pubertal onset
Girls’ development was assessed using Marshall and Tanner criteria for breast maturation and pubic hair stages (21) conducted by trained staff at each in-person clinic visit, using a five-stage classification scheme for describing the onset and progression of breast and pubic hair changes by inspection and palpation. Details of training, certification and assessment procedures are reported elsewhere (22). For analytic purposes, breast and pubic hair pubertal stages were characterized as “no onset” (Stage 1) or “onset” (Stage 2 or above) for each visit in this investigation.
Infant feeding practices
At baseline, the primary caregiver answered questions related to infant feeding history, including whether the study participant was breast fed and/or formula fed and duration of breastfeeding. Study participants were categorized into three groups based on feeding practices: [1] predominantly breastfed, defined as girls who were exclusively breastfed or received formula no more than 30 days in their lifetime; [2] mixed fed, defined as girls who were breastfed and also received formula for more than 30 days in their lifetime; and [3] formula-fed, defined as girls who were never breastfed and were exclusively formula-fed. Duration of breastfeeding was also examined and was categorized as: never breastfed; less than 3 months; 3 to 6 months; 6 to 12 months; and 12 months or longer.
Other covariates
Body mass index (BMI) at baseline was calculated as weight (kg)/(height (m))2, and the age-specific BMI percentiles based on Centers for Disease Control and Prevention (CDC) standards (23) were estimated for the age in months. Weight and height were measured in clinic using calibrated scales and fixed stadiometers during the baseline examination. BMI percentiles were categorized into overweight (from the 85th up to the 95th percentile), obese (greater than or equal to the 95th percentile), and normal (less than 85th percentile).
Girls’ race and ethnicity was reported by primary caregivers using census-based categories as well as more detailed information regarding country of origin for parents and grandparents. Caregivers could respond to multiple race or ethnicity categories. These data were categorized into four mutually-exclusive groups: African-American, Hispanic, Asian, and non-Hispanic White. Annual household income at baseline was categorized as < $25,000; $25,000 to $50,000; $50,000 to $100,000; and > $100,000. For 84% of girls, the primary caregiver was their biological mother. Mother’s educational attainment was classified into five categories: less than high school; high school graduate; some college or associate’s degree; bachelor’s degree; and master’s/professional degree. Birth weight was reported by the primary caregiver at baseline and categorized as low (<2,500 grams), normal (2,500 to 4,000 grams), and high (>4,000 grams). Mothers’ age at menarche was coded as a categorical variable: early (<12 years), normal (12–14 years), and late (>14 years).
Statistical Analysis
Initial analyses focused on describing potential differences in covariate distributions among infant feeding practice categories. Those that appeared to show a difference were then treated as covariates in subsequent analyses.
Primary analyses examined associations for infant feeding practices (breastfeeding or formula feeding) and breastfeeding duration with age at onset of breast or pubic hair development. Analyses used proportional hazards survival methods with left-, interval-, and right-censoring, where the outcome of interest was time (i.e., age of girls) until an event (i.e. pubertal onset). Observations were interval censored because the exact point within the time interval in which the transition to pubertal stage 2 occurred was unknown. To accommodate censoring, one upper and one lower boundary of the interval (girls’ age) were considered in every case to estimate the “likelihood contribution” of each observation. The interval in which pubertal onset occurred was defined as between two clinic visit dates: [1] the earlier date with pubertal stage 1 and no prior visit with stage 2 or greater; and [2] the later date as the first clinic visit with pubertal stage 2 or greater and no subsequent stage 1 visit. Right-censored girls had not yet experienced pubertal onset by the final wave included, and so the later date was set as missing. About 14% of girls already had experienced pubertal onset at the baseline exam. For these left-censored girls, the earlier date defining the interval was set at age 60 months (5 years), based on the assumption that girls would not have begun breast or pubic hair development before that age.
Survival analyses were conducted using STATA version 11, using the Weibull parametric survival model. Interval-censored data requires the use of parametric survival models rather than the Cox proportional hazards model, which only allows for right censoring. The Weibull model is a widely-used and flexible parametric survival model (24). This model allows for accelerating failure times, which is appropriate in the context of pubertal development, as the older a girl is, the more likely she is to begin pubertal development. Breast and pubic hair onset were treated as separate outcomes. Length of time interval within which puberty began, as defined above, was used as the outcome variable. A final parametric survival model included variables (race/ethnicity, BMI percentile, mother’s menarche age, mother’s education and family income) that altered coefficient estimates by >10% or that were significant at a probability level <0.05. Secondary analyses included testing for trend in breastfeeding duration, stratifying analyses by recruitment site (New York City, Greater Cincinnati, and San Francisco Bay Area) and testing for mediation of breastfeeding-puberty association by BMI using the Baron and Kenny steps (25). We also examined interactions of site with breastfeeding.
Results
Characteristics of the 1,237 girls in the cohort are presented in Table 1. Breastfeeding rates varied across the three sites. Almost two-thirds of San Francisco Bay Area girls were breastfed for at least 6 months, while less than 7% were never breastfed, compared to 34% never breast fed in Cincinnati, and 30% in New York City. The prevalence of predominant breastfeeding was lowest among African-Americans. Among obese girls, 13% were reported to be predominantly breastfed, while almost 25% were formula-fed. Higher breastfeeding rates were found among mothers with higher education and higher household income. Table 2 shows the characteristics of the cohort by pubertal onset. Pubertal onset was noted to occur earlier in girls from Cincinnati and NYC compared to California, in overweight/obese girls compared to those with normal BMI, in African American girls, and in those with lower family income.
Table 1.
Characteristics of the BCERP Puberty Studies Cohort According to Infant Feeding Practices, Breast Cancer and the Environment Research Program Puberty Studies, 2004–2011
| Characteristic | p | Category | Predominantly Breastfed | Mixed fed | Formula-fed | Overall | ||||
|---|---|---|---|---|---|---|---|---|---|---|
| N | % | N | % | N | % | N | % | |||
| Site | <0.001 | San Francisco Bay Area | 148 | 33.4 | 266 | 60.0 | 29 | 6.5 | 444 | 35.9 |
| Greater Cincinnati | 44 | 12.9 | 187 | 54.8 | 110 | 32.2 | 377 | 30.5 | ||
| New York City | 56 | 13.8 | 231 | 56.8 | 120 | 29.5 | 416 | 33.6 | ||
| Race/Ethnicity | <0.001 | African-American | 38 | 10.2 | 204 | 55.1 | 128 | 34.6 | 390 | 31.5 |
| Hispanic | 72 | 19.8 | 232 | 63.9 | 59 | 16.2 | 372 | 30.0 | ||
| Asian | 15 | 27.8 | 33 | 61.1 | 6 | 11.1 | 57 | 4.6 | ||
| White | 123 | 30.4 | 215 | 53.2 | 66 | 16.3 | 418 | 33.8 | ||
| BMI (percentile) | 0.098 | Normal (< 85) | 178 | 22.3 | 452 | 56.7 | 167 | 21.0 | 826 | 66.7 |
| Overweight (85–94) | 41 | 22.1 | 105 | 56.8 | 39 | 21.1 | 193 | 15.6 | ||
| Obese (≥ 95) | 29 | 13.9 | 127 | 60.8 | 53 | 25.4 | 218 | 17.6 | ||
| Mother's age at menarche (y) | 0.008 | < 12 | 55 | 19.9 | 170 | 61.6 | 51 | 18.5 | 278 | 26.1 |
| 12 to 13.9 | 124 | 23.3 | 283 | 53.1 | 126 | 23.6 | 537 | 50.4 | ||
| ≥ 14 | 46 | 18.4 | 165 | 66.2 | 38 | 15.2 | 249 | 23.4 | ||
| Birth-weight (g) | 0.004 | < 2,500 | 11 | 9.9 | 66 | 59.5 | 34 | 30.6 | 112 | 9.6 |
| ≥ 2,500 | 229 | 21.9 | 598 | 57.1 | 219 | 21.0 | 1055 | 90.4 | ||
| Mother's education | <0.001 | < High school graduate | 30 | 16.9 | 103 | 58.2 | 44 | 24.9 | 180 | 15.3 |
| High school graduate | 25 | 13.8 | 94 | 51.9 | 62 | 34.2 | 182 | 15.4 | ||
| Some college | 46 | 12.4 | 218 | 58.7 | 107 | 28.8 | 372 | 31.6 | ||
| Bachelor’s degree | 81 | 28.6 | 174 | 61.5 | 28 | 9.9 | 285 | 24.2 | ||
| Prof./doctoral degree | 61 | 39.6 | 83 | 53.9 | 10 | 6.5 | 158 | 13.4 | ||
| Family Income ($) | <0.001 | < 25,000 | 41 | 15.6 | 137 | 52.1 | 85 | 32.3 | 267 | 21.9 |
| 25,000 to 49,999 | 29 | 12.3 | 141 | 59.7 | 66 | 27.9 | 239 | 19.6 | ||
| 50,000 to 99,999 | 78 | 24.9 | 177 | 56.5 | 58 | 18.5 | 318 | 26.1 | ||
| ≥ 100,000 | 90 | 31.2 | 175 | 60.8 | 23 | 8.0 | 291 | 23.8 | ||
| Unknown | 10 | 11.0 | 54 | 59.3 | 27 | 29.6 | 104 | 8.5 | ||
Table 2.
Characteristics by Pubertal Onset (Crude), Breast Cancer and the Environment Research Program Puberty Studies, 2004–2011
| Characteristic | N | Breast Onset HR (95% CI) | Pubic Hair Onset HR (95% CI) | |
|---|---|---|---|---|
| Site | San Francisco Bay Area | 444 | 1.0 | 1.0 |
| Greater Cincinnati | 377 | 1.7 (1.4, 1.9) | 1.1 (0.9, 1.3) | |
| New York City | 416 | 1.5 (1.3, 1.8) | 1.5 (1.2, 1.7) | |
| Race/Ethnicity | African-American | 390 | 1.7 (1.4, 2.0) | 2.4 (2.0, 2.8) |
| Hispanic | 372 | 1.2 (1.01, 1.4) | 1.3 (1.1, 1.5) | |
| Asian | 57 | 0.7 (0.5, 1.01) | 0.5 (0.4, 0.8) | |
| White | 418 | 1.0 | 1.0 | |
| BMI (percentile) | Normal (< 85) | 826 | 1.0 | 1.0 |
| Overweight (85–94) | 193 | 1.9 (1.5, 2.2) | 1.5 (1.2, 1.8) | |
| Obese (≥ 95) | 218 | 2.2 (2.9, 2.7) | 1.7 (1.5, 2.1) | |
| Mother's age at menarche (y) | < 12 | 278 | 1.0 | 1.0 |
| 12 to 13.9 | 537 | 0.8 (0.7, 1.01) | 0.8 (0.7, 0.9) | |
| ≥ 14 | 249 | 0.7 (0.6, 0.8) | 0.6 (0.5, 0.8) | |
| Birth-weight (g) | < 2,500 | 112 | 0.9 (0.7, 1.1) | 1.0 (0.8, 1.3) |
| ≥ 2,500 | 1055 | 1.0 | 1.0 | |
| Mother's education | < High school graduate | 180 | 1.0 | 1.0 |
| High school graduate | 182 | 0.8 (0.6, 1.0) | 1.1 (0.8, 1.4) | |
| Some college | 372 | 1.1(0.9, 1.3) | 1.3 (1.1, 1.6) | |
| Bachelor’s degree | 285 | 0.8 (0.7, 1.0) | 1.0 (0.8, 1.2) | |
| Prof./doctoral degree | 158 | 0.8 (0.6, 1.0) | 0.9 (0.7, 1.1) | |
| Family Income ($) | < 25,000 | 267 | 1.0 | 1.0 |
| 25,000 to 49,999 | 239 | 1.1 (0.9, 1.3) | 1.3 (1.0, 1.6) | |
| 50,000 to 99,999 | 318 | 0.8 (0.7, 0.9) | 0.9 (0.8, 1.1) | |
| ≥ 100,000 | 291 | 0.6 (0.5, 0.7) | 0.6 (0.5, 0.7) | |
Breastfeeding and onset of breast development
Overall, in crude analyses, using formula-fed girls as the reference group, hazard ratios for breast onset were 0.82 (95% confidence interval (CI) 0.69, 0.97) for mixed fed girls and 0.64 (95% CI 0.52, 0.78) for predominantly breastfed girls. After adjustment for covariates, these associations were somewhat attenuated, but similar associations remained. Hazard ratios from adjusted analyses were 0.90 (95% CI 0.75, 1.09) and 0.74 (95% CI 0.59, 0.94) for mixed fed and predominantly breastfed girls, respectively. Table 3 shows site-specific crude and adjusted associations of breastfeeding and onset of breast development.
Table 3.
Association of Breastfeeding with Age at Breast Onset, Stratified by Site, Breast Cancer and the Environment Research Program Puberty Studies, 2004–2011
| Breastfeeding Category | N | Crude HR (95% CI) | Adjusted* HR (95% CI) |
|---|---|---|---|
| San Francisco Bay Area | |||
| Formula-fed | 29 | 1.0 | 1.0 |
| Mixed fed | 266 | 0.95 (0.59, 1.51) | 1.73 (1.02, 2.96) |
| Predominantly breastfed | 148 | 0.80 (0.49, 1.29) | 1.43 (0.82, 2.47) |
| Greater Cincinnati | |||
| Formula-fed | 110 | 1.0 | 1.0 |
| Mixed fed | 187 | 0.70 (0.53, 0.91) | 0.56 (0.41, 0.75) |
| Predominantly breastfed | 44 | 0.60 (0.40, 0.89) | 0.55 (0.35, 0.86) |
| New York City | |||
| Formula-fed | 120 | 1.0 | 1.0 |
| Mixed fed | 231 | 1.25 (0.95, 1.65) | 1.18 (0.87, 1.59) |
| Predominantly breastfed | 56 | 1.07 (0.73, 1.57) | 1.04 (0.68, 1.61) |
Adjusting for BMI, ethnicity, mother’s age at menarche, girls’ birth weight, mother’s education, and family income
There was an inverse association of breastfeeding duration with younger age at breast development. In unadjusted analyses, in comparison to girls who were not breastfed, the hazard ratios for increasing duration of breast feeding were 0.84 (0.67, 1.05), 0.85 (0.68, 1.07), 0.92 (0.76, 1.13), and 0.62 (0.51, 0.75) for <3 months, 3–6 months, 6–12 months, and 12 months and more, respectively. After adjustment for covariates, the hazard ratios were 0.84 (0.66, 1.06) for < 3 months, 0.98 (0.76, 1.26) for 3–6 months, 1.12 (0.89, 1.40) for 6–12 months, and 0.72 (0.57, 0.90) for 12 months and more. The test for trend (breastfeeding duration) was statistically significant in both crude and adjusted models. Table 4 shows site-specific crude and adjusted associations of breastfeeding and onset of breast development.
Table 4.
Association of Breastfeeding Duration with Age at Breast Onset, Stratified by Site, Breast Cancer and the Environment Research Program Puberty Studies, 2004–2011
| Breastfeeding Duration (months) | N | Crude HR (95% CI) | Adjusted* HR (95% CI) |
|---|---|---|---|
| San Francisco Bay Area | |||
| Never Breastfed | 29 | 1.0 | 1.0 |
| < 3 | 49 | 1.17 (0.67, 2.04) | 1.69 (0.91, 3.10) |
| 3 – 6 | 63 | 0.80 (0.47, 1.37) | 1.48 (0.79, 2.76) |
| 6 – 12 | 118 | 1.06 (0.65, 1.72) | 2.13 (1.21, 3.76) |
| ≥ 12 | 179 | 0.79 (0.49, 1.27) | 1.48 (0.85, 2.60) |
| P trend | 0.093 | 0.93 | |
| Per month (continuous) | 0.97 (0.94, 1.00) | 0.98 (0.95, 1.02) | |
| Greater Cincinnati | |||
| Never Breastfed | 112 | 1.0 | 1.0 |
| < 3 | 51 | 0.58 (0.40, 0.85) | 0.57 (0.38, 0.87) |
| 3 – 6 | 50 | 0.79 (0.54, 1.14) | 0.59 (0.39, 0.89) |
| 6 – 12 | 73 | 0.85 (0.61, 1.19) | 0.72 (0.49, 1.04) |
| ≥ 12 | 45 | 0.50 (0.33, 0.73) | 0.38 (0.24, 0.61) |
| P trend | 0.014 | 0.001 | |
| Per month (continuous) | 0.98 (0.94, 1.01) | 0.97 (0.93, 1.01) | |
| New York City | |||
| Never Breastfed | 122 | 1.0 | 1.0 |
| < 3 | 72 | 1.14 (0.80, 1.64) | 1.04 (0.70, 1.54) |
| 3 – 6 | 57 | 1.54 (1.03, 2.29) | 1.55 (1.01, 2.40) |
| 6 – 12 | 59 | 1.80 (1.23, 2.62) | 1.72 (1.13, 2.61) |
| ≥ 12 | 99 | 0.95 (0.69, 1.32) | 0.88 (0.61, 1.27) |
| P trend | 0.749 | 0.63 | |
| Per month (continuous) | 0.97 (0.94, 1.00) | 0.98 (0.94, 1.01) |
Adjusting for BMI, ethnicity, mother’s age at menarche, girls’ birth weight, mother’s education, and family income
Breastfeeding and onset of pubic hair development
Unadjusted associations of infant feeding practices with pubic hair onset were qualitatively similar to those for breast onset. In unadjusted bivariate models, using formula-fed girls as the reference group, hazard ratios for breast onset were 0.85 (95% CI 0.72, 1.01) for mixed fed girls and 0.67 (95% CI 0.55, 0.83) for predominantly breastfed girls (Table 5). However, adjusted models were more markedly attenuated and did not show statistically-significant associations between predominant breastfeeding and the onset of pubic hair development. There was also an association with duration of breastfeeding in unadjusted models, which was attenuated after adjustment (Table 6).
Table 5.
Association of Breastfeeding with Age at Pubic Hair Onset, Overall and Stratified by Site, Breast Cancer and the Environment Research Program Puberty Studies, 2004–2011
| Breastfeeding Category | N | Crude HR (95% CI) | Adjusted* HR (95% CI) |
|---|---|---|---|
| Total Population | |||
| Formula-fed | 259 | 1.0 | 1.0 |
| Mixed fed | 684 | 0.85 (0.72, 1.01) | 0.95 (0.79, 1.15) |
| Predominantly breastfed | 248 | 0.67 (0.55, 0.83) | 0.84 (0.67, 1.06) |
| San Francisco Bay Area | |||
| Formula-fed | 29 | 1.0 | 1.0 |
| Mixed fed | 266 | 0.75 (0.48, 1.17) | 0.92 (0.57, 1.49) |
| Predominantly breastfed | 148 | 0.64 (0.40, 1.01) | 0.84 (0.51, 1.39) |
| Greater Cincinnati | |||
| Formula-fed | 110 | 1.0 | 1.0 |
| Mixed fed | 187 | 0.85 (0.64, 1.12) | 0.82 (0.60, 1.12) |
| Predominantly breastfed | 44 | 0.86 (0.57, 1.27) | 0.97 (0.63, 1.50) |
| New York City | |||
| Formula-fed | 120 | 1.0 | 1.0 |
| Mixed fed | 231 | 1.04 (0.79, 1.36) | 1.08 (0.80, 1.46) |
| Predominantly breastfed | 56 | 0.65 (0.43, 0.95) | 0.7 (0.45, 1.07) |
Adjusting for BMI, ethnicity, mother’s age at menarche, girls’ birth weight, mother’s education, and family income
Table 6.
Association of Breastfeeding Duration with Age at Pubic Hair Onset, Overall and Stratified by Site, Breast Cancer and the Environment Research Program Puberty Studies, 2004–2011
| Breastfeeding Duration (months) | N | Crude HR (95% CI) | Adjusted* HR (95% CI) |
|---|---|---|---|
| Total Population | |||
| Never Breastfed | 263 | 1.0 | 1.0 |
| < 3 | 172 | 0.91 (0.72, 1.14) | 0.93 (0.73, 1.19) |
| 3 – 6 | 170 | 0.98 (0.78, 1.23) | 1.01 (0.78, 1.29) |
| 6 – 12 | 250 | 0.81 (0.66, 0.99) | 0.97 (0.77, 1.21) |
| ≥ 12 | 323 | 0.67 (0.55, 0.82) | 0.81 (0.65, 1.01) |
| P trend | 0.000 | 0.059 | |
| Per month (continuous) | 0.96 (0.95, 0.98) | 0.98 (0.96, 1.00) | |
| San Francisco Bay Area | |||
| Never Breastfed | 29 | 1.0 | 1.0 |
| < 3 | 49 | 0.81 (0.47, 1.37) | 0.84 (0.48, 1.48) |
| 3 – 6 | 63 | 0.77 (0.46, 1.28) | 0.97 (0.55, 1.71) |
| 6 – 12 | 118 | 0.71 (0.44, 1.12) | 0.93 (0.56, 1.55) |
| ≥ 12 | 179 | 0.67 (0.43, 1.05) | 0.92 (0.55, 1.51) |
| P trend | 0.000 | 0.94 | |
| Per month (continuous) | 0.98 (0.95, 1.01) | 1.00 (0.97, 1.04) | |
| Greater Cincinnati | |||
| Never Breastfed | 112 | 1.0 | 1.0 |
| < 3 | 51 | 0.73 (0.49, 1.08) | 0.77 (0.50, 1.19) |
| 3 – 6 | 50 | 1.16 (0.79, 1.72) | 0.91 (0.58, 1.42) |
| 6 – 12 | 73 | 0.93 (0.66, 1.31) | 0.92 (0.63, 1.34) |
| ≥ 12 | 45 | 0.66 (0.44, 0.99) | 0.72 (0.45, 1.14) |
| P trend | 0.151 | 0.32 | |
| Per month (continuous) | 0.98 (0.95, 1.02) | 0.99 (0.95, 1.03) | |
| New York City | |||
| Never Breastfed | 122 | 1.0 | 1.0 |
| < 3 | 72 | 1.25 (0.87, 1.79) | 1.10 (0.75, 1.64) |
| 3 – 6 | 57 | 1.17 (0.80, 1.71) | 1.11 (0.73, 1.69) |
| 6 – 12 | 59 | 1.02 (0.70, 1.48) | 1.13 (0.75, 1.72) |
| ≥ 12 | 99 | 0.67 (0.48, 0.93) | 0.75 (0.52, 1.08) |
| P trend | 0.002 | 0.07 | |
| Per month (continuous) | 0.94 (0.91, 0.97) | 0.96 (0.92, 0.99) |
Adjusting for BMI, ethnicity, mother’s age at menarche, girls’ birth weight, mother’s education, and family income
In adjusted models, BMI was not found to mediate the effect of breastfeeding on pubertal onset. The association of BMI (being overweight or obese) with younger age at breast development was significant (HR 2.12, p<0.0005). Although somewhat associated in crude models as seen in Table 1 (p=0.098), breastfeeding (or its duration) and BMI were no longer associated after adjusting for other covariates (p>0.5). In addition, there was no evidence of mediation by BMI after stratifying by site in individual site-specific models.
The breastfeeding by site interaction term was significant for onset of breast development (P < 0.005), but not for pubic hair (p>0.5). Site-specific analyses examining the associations between infant-feeding practices and breastfeeding (Tables 3–6) demonstrated substantial variation across sites, and adjustment for covariates did not account for this variation. The overall direct association of breastfeeding with age at onset of breast development was seen in Cincinnati girls only. However, the other sites – in particular, the San Francisco Bay Area girls – had relatively limited variation in breast feeding practices, as described below. We also examined effect modification by race/ethnicity and the coefficients did not vary depending on girls’ race/ethnicity.
Discussion
This study examined the effect of breastfeeding versus formula feeding, as well as duration of breastfeeding, on girls’ breast and pubic hair onset. Predominant breastfeeding and longer breastfeeding duration were associated with a later age at onset of breast development when data were combined across all sites. Compared to formula-fed girls, the ages of breast onset were older for mixed-fed girls and even more so for predominantly breast-fed girls, which indicates a possible dose-dependent effect of breastfeeding on later pubertal onset. This is also supported by the observed inverse association between duration of breastfeeding and early onset of breast development. However, there were notable site differences in results from stratified analyses. These associations were evident only in Cincinnati, indicating that there may be unmeasured modifying variables that varied across sites.
Given the temporal sequence of the events examined (breastfeeding – BMI – pubertal onset); we hypothesized that breastfeeding might influence BMI, which in turn, would influence pubertal onset. Thus, we examined BMI as a mediator of the relationship between breastfeeding and onset of pubertal events. The inverse association of breastfeeding with childhood obesity is widely documented (15, 16). One possible mechanism involves the higher plasma concentration of insulin found in formula-fed infants as compared to breastfed infants (26). Since insulin enhances cell glucose uptake and inhibits lipolysis (fat breakdown), greater deposition of subcutaneous adipose tissue has been seen in bottle-fed infants (26). However, there was no evidence of BMI mediation in the combined data or for each site separately. This may be attributed to the factors other than breastfeeding such as early childhood BMI, physical activity, nutrition, etc (27).
In addition to BMI, alternative pathways to delayed puberty are plausible, such as protective factors in breast milk composition (such as bioactive nutrients) (18, 19), or associations of breastfeeding with behavior patterns in mothers that promote health. For example, mothers who are more likely to breastfeed may also be more likely to have their daughters engage in physical activity, which may in turn delay pubertal onset. Another potential mechanism, based on evolutionary theories, is that breastfeeding might be a proxy for infant attachment security, which has shown to predict later pubertal onset (28).
It is important to note that infant feeding practices across the sites were considerably distinct, as seen by the numbers of girls who fell into the different exposure categories, which may in part account for differences across the sites. A substantial majority of San Francisco Bay Area girls was breastfed and for longer durations, whereas only a minority of girls in New York City was predominantly breastfed, thus limiting variability in the exposure for these two sites. Participants in New York City may be exposed to other risk factors for earlier pubertal onset such as environmental exposures, or lifestyle factors. These associations may have a much stronger impact on pubertal onset that the counter effect of breastfeeding may not have been observable This may also help explain why past studies, originating from different parts of the world, that examine breast feeding and menarche have yielded inconsistent or equivocal results.
This study had several limitations. Participants were recruited from 3 different areas of the U.S. with diverse socio-economic and ethnic backgrounds; however, our sample is not generalizable to the entire U.S. population. For example, sites were predominantly urban therefore participants from rural areas were not represented. Participants had to voluntarily go to the clinic site, adding a possible selection bias. Also, we did not gather detailed data about early infant feeding practices, such as details about exclusivity of breastfeeding, age of formula/solids introduction. Maternal age at menarche, a proxy for genetic contributions to daughter’s age at breast or pubic hair onset, may not be sufficient to control for potential genetic confounders. The differences noted across the three sites may appear to be a limitation. However, by combining data across the sites, we thus have greater and more stable variation in our main exposures of interest and other key covariates. This could thus be considered a marked strength of this multi-site study. Of note, variation in the sampling strategies across the sites may be considered a limitation. In future cross-site studies, consistent sampling approaches should be employed to minimize potential unmeasured confounding.
Despite limitations, our study addresses some distinct gaps in this area of research. For example, studies with cross-sectional designs (2) or those using logistic regression models fail to address the key analytical issue of censoring, potentially creating significant bias in estimates of pubertal onset. Many studies (12, 14) focus mainly on a single pubertal marker, menarche, which occurs relatively late in pubertal development and is thus a weak and indirect measure of pubertal onset. Some studies (11) fail to include BMI, despite its documented association with pubertal timing. Other studies (9) use self-report or parent-report for pubertal stage estimates. This method is not as reliable as pubertal staging by standardized in-clinic examination with palpation, the method used in this study (29).
The unique strengths of this study outweigh its limitations. This is the first prospective study on a multiethnic U.S. cohort to examine the association between breastfeeding and girls’ pubertal onset using clinical Tanner staging to assess pubertal markers. The longitudinal study design, with annual/semiannual follow-up using physical examination and palpation, allowed for better estimates of age at pubertal onset compared to past studies. We assessed the role of BMI as a possible mediator of the association of breastfeeding and pubertal onset, which has not been tested in other studies. In addition, this study is one of the first to use parametric survival analysis models with interval censoring, incorporating more appropriate analytic methods for analyzing longitudinal data with time to onset of puberty as the outcome.
Breastfeeding has been linked to many health benefits across the lifespan for both mothers and their children (30). Such benefits include reduced morbidity due to gastrointestinal and respiratory infections, reduced rates of sudden death syndrome in infants; and lower cardiovascular disease risk and obesity later in adulthood. In this study, an additional potential benefit of breastfeeding was explored, an inverse association with timing of pubertal onset. Our principal findings are that breastfeeding and its duration may be associated with later age at onset of breast development in certain contexts. This study may viewed as a pivot to the research on this topic, with a future representative national sample further exploring the roles of ethnic origins, BMI, maternal factors, environmental factors. Our findings suggest that future research should focus on potential modifiers of these effects and whether these modifying factors may vary systematically across contexts. From a clinical perspective, such research could provide important information for pediatricians, pediatric nurse practitioners and family physicians, who play a role in education and promotion of breastfeeding and work nationally across diverse contexts. If further research confirms that breastfeeding is associated with delayed puberty within certain contexts, then targeted interventions (31) may confer additional health benefits to delay onset of puberty among certain sets of girls and may promote subsequent beneficial psychosocial and health effects throughout the life-span.
Figure 1.
BMI Mediation, Breast Cancer and the Environment Research Program Puberty Studies, 2004–2011
Acknowledgments
This study was funded by U01ES012770, U01ES012771, U01ES012800, U01ES012801, U01ES019435, U01ES019453, U01ES019454, and U01ES019457 from the National Institute of Environmental Health Sciences (NIEHS) and the National Cancer Institute (NCI); P01ES009584 and P30ES006096 from NIEHS; and UL1RR024131, UL1RR029887 and UL1RR026314 from the National Center for Research Resources (NCRR). Its contents are solely the responsibility of the authors and do not necessarily represent the official views of the NIEHS, the NCI, the NCRR, the National Institutes of Health, or the California Department of Public Health. We also gratefully acknowledge support of the Avon Foundation for this research.
References
- 1.Anderson SE, Dallal GE, Must A, et al. Relative weight and race influence average age at menarche: results from two nationally representative surveys of US girls studied 25 years apart. Pediatrics. 2003 Apr;111(4 Pt 1):844–50. doi: 10.1542/peds.111.4.844. [DOI] [PubMed] [Google Scholar]
- 2.Downing J, Bellis MA. Early pubertal onset and its relationship with sexual risk taking, substance use and anti-social behaviour: a preliminary cross-sectional study. BMC Public Health. 2009 Dec 3;9:446. doi: 10.1186/1471-2458-9-446. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Golub MS, Collman GW, Foster PM, et al. Public Health Implications of Altered Puberty Timing. Pediatrics. 2008 Feb;121( Suppl 3):S218–30. doi: 10.1542/peds.2007-1813G. [DOI] [PubMed] [Google Scholar]
- 4.Kadlubar FF, Berkowitz GS, Delongchamp RR, et al. The CYP3A4*1B variant is related to the onset of puberty, a known risk factor for the development of breast cancer. Cancer Epidemiol Biomarkers Prev. 2003 Apr;12(4):327–31. [PubMed] [Google Scholar]
- 5.Bernstein L. Epidemiology of endocrine-related risk factors for breast cancer. J Mammary Gland Biol Neoplasia. 2002;7:3–15. doi: 10.1023/a:1015714305420. [DOI] [PubMed] [Google Scholar]
- 6.Mouritsen A, Aksglaede L, Sørensen K, et al. Hypothesis: exposure to endocrine-disrupting chemicals may interfere with timing of puberty. Int J Androl. 2010 Apr;33(2):346–59. doi: 10.1111/j.1365-2605.2010.01051.x. [DOI] [PubMed] [Google Scholar]
- 7.Toppari J, Juul A. Trends in puberty timing in humans and environmental modifiers. Mol Cell Endocrinol. 2010 Aug 5;324(1–2):39–44. doi: 10.1016/j.mce.2010.03.011. [DOI] [PubMed] [Google Scholar]
- 8.Cheng G, Gerlach S, Libuda L, et al. Diet quality in childhood is prospectively associated with the timing of puberty but not with body composition at puberty onset. J Nutr. 2010 Jan;140( 1):95–102. doi: 10.3945/jn.109.113365. [DOI] [PubMed] [Google Scholar]
- 9.Maisonet M, Christensen KY, Rubin C, Holmes A, et al. Role of prenatal characteristics and early growth on pubertal attainment of British girls. Pediatrics. 2010 Sep;126(3):e591–600. doi: 10.1542/peds.2009-2636. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Karaolis-Danckert N, Buyken AE, Sonntag A, Kroke A. Birth and early life influences on the timing of puberty onset: results from the DONALD (DOrtmund Nutritional and Anthropometric Longitudinally Designed) Study. Am J Clin Nutr. 2009 Dec;90(6):1559–65. doi: 10.3945/ajcn.2009.28259. [DOI] [PubMed] [Google Scholar]
- 11.Kwok MK, Leung GM, Lam TH, Schooling CM. Breastfeeding, Childhood Milk Consumption, and Onset of Puberty. Pediatrics. 2012 Sep;130(3):e631–9. doi: 10.1542/peds.2011-3697. Epub 2012 Aug 20. [DOI] [PubMed] [Google Scholar]
- 12.Al-Sahab B, Adair L, Hamadeh MJ, Ardern CI, Tamim H. Impact of breastfeeding duration on age at menarche. Am J Epidemiol. 2011;173(9):971–977. doi: 10.1093/aje/kwq496. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Morris DH, Jones ME, Schoemaker MJ, Ashworth AA, Swerdlow AJ. Determinants of age at menarche in the UK: analyses from the Breakthrough Generations Study. British Journal of Cancer. 2010;103:1760–1764. doi: 10.1038/sj.bjc.6605978. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Novotny R, Daida YG, Grove JS, Acharya S, Vogt TM. Formula feeding in infancy is associated with adolescent body fat and earlier menarche. Cell Mol Biol (Noisy-le-grand) 2003 Dec;49(8):1289–93. [PubMed] [Google Scholar]
- 15.Kries RV, Koletzko B, Sauerwald T, et al. Breast feeding and obesity: cross sectional study. BMJ. 1999 Jul 17;319(7203):147–50. doi: 10.1136/bmj.319.7203.147. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Metzger MW, McDade TW. Breastfeeding as obesity prevention in the United States: a sibling difference model. Am J Hum Biol. 2010 May-Jun;22(3):291–6. doi: 10.1002/ajhb.20982. [DOI] [PubMed] [Google Scholar]
- 17.Rosenfield RL, Lipton RB, Drum ML. Thelarche, pubarche, and menarche attainment in children with normal and elevated body mass index. Pediatrics. 2009 Jan;123(1):84–8. doi: 10.1542/peds.2008-0146. [DOI] [PubMed] [Google Scholar]
- 18.Hamosh M. Bioactive factors in human milk. Pediatr Clin North Am. 2001 Feb;48(1):69–86. doi: 10.1016/s0031-3955(05)70286-8. [DOI] [PubMed] [Google Scholar]
- 19.Locke R. Preventing obesity: the breast milk-leptin connection. Acta Paediatr. 2002;91(9):891–4. doi: 10.1080/080352502760272524. [DOI] [PubMed] [Google Scholar]
- 20.Hiatt RA, Haslam SZ, Osuch J. The Breast Cancer and the Environment Research Centers: Transdisciplinary Research on the Role of the Environment in Breast Cancer Etiology. Envir Health Perspect. 2009;117:1814–1822. doi: 10.1289/ehp.0800120. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Marshall WA, Tanner JM. Variations in pattern of pubertal changes in girls. Arch Dis Child. 1969;44(235):291–303. doi: 10.1136/adc.44.235.291. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Biro FM, Galvez MP, Greenspan LC, et al. Pubertal assessment method and baseline characteristics in a mixed longitudinal study of girls. Pediatrics. 2010 Sep;126(3):e583–90. doi: 10.1542/peds.2009-3079. Epub 2010 Aug 9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Vital and Health Statistics. [accessed on 12/05/2012];CDC Growth Charts for the United States: Methods and Development. 2000 DHHS Publication No. (PHS) 2002–1696. Available at www.cdc.gov/nchs/data/series/sr_11/sr11_246.pdf. [PubMed]
- 24.Kleinbaum DG, Klein M. Survival Analysis: A Self-Learning Text. 2. New York: Springer publisher; 2005. [Google Scholar]
- 25.Baron RM, Kenny DA. The moderator-mediator variable distinction in social psychological research: Conceptual, strategic and statistical considerations. Journal of Personality and Social Psychology. 1986;51:1173–1182. doi: 10.1037//0022-3514.51.6.1173. [DOI] [PubMed] [Google Scholar]
- 26.Lucas A, Boyes S, Bloom SR, et al. Metabolic and endocrine responses to a milk feed in six-day-old term infants: differences between breast and cow’s milk formula feeding. Acta Paediatr Scand. 1981;70:195. doi: 10.1111/j.1651-2227.1981.tb05541.x. [DOI] [PubMed] [Google Scholar]
- 27.Ebbeling CB, Pawlak DB, Ludwig DS. Childhood obesity: public-health crisis, common sense cure. Lancet. 2002 Aug 10;360(9331):473–82. doi: 10.1016/S0140-6736(02)09678-2. [DOI] [PubMed] [Google Scholar]
- 28.Belsky J, Houts RM, Fearon RM. Infant attachment security and the timing of puberty: testing an evolutionary hypothesis. Psychol Sci. 2010 Sep;21(9):1195–201. doi: 10.1177/0956797610379867. [DOI] [PubMed] [Google Scholar]
- 29.Desmangles JC, Lappe JM, Lipaczewski G, Haynatzki G. Accuracy of pubertal Tanner staging self-reporting. J Pediatr Endocrinol Metab. 2006 Mar;19(3):213–21. doi: 10.1515/jpem.2006.19.3.213. [DOI] [PubMed] [Google Scholar]
- 30.Aguiar H, Silva AI. Breastfeeding: the importance of intervening. Acta Med Port. 2011 Dec;24( Suppl 4):889–96. Epub 2011 Dec 31. [PubMed] [Google Scholar]
- 31.Centers for Disease Control & Prevention. [accessed on 12/05/2012];Breastfeeding Report Card – United States. 2010 Available at http://www.cdc.gov/breastfeeding/data/