Association of Phthalates, Parabens and Phenols Found in Personal Care Products With Pubertal Timing in Girls and Boys

Kim G. Harley; Kimberly P. Berger; Katherine Kogut; Kimberly Parra; Robert H. Lustig; Louise C. Greenspan; Antonia M. Calafat; Xiaoyun Ye; Brenda Eskenazi


Hum Reprod. 2019;34(1):109-117. 

In This Article


The study population was almost exclusively Latino, with 73% of mothers having lived in the USA for fewer than 11 years at the time of their pregnancies (Table I). Most mothers (79%) had not completed high school, and 69% of children at age 9 were living below the federal poverty threshold. At age 9, 55% of children were overweight or obese and 65% of mothers had been overweight or obese prior to the pregnancy.

Timing of pubertal onset in this cohort was similar to Latino children participating in NHANES (Sun et al., 2002). In girls, the median age at thelarche (B2) was 9.2 years, at pubarche (PH2), 10.3 years and at menarche, 11.7 years. In boys, the median age at gonadarche (G2) was 10.8 years and at pubarche (PH2), 12.2 years. At age 9, the start of pubertal observation period, 39% and 20% of girls had already reached B2 and PH2, respectively, and 11% and 1% of boys had reached G2 and PH2.

All biomarkers were detected in >90% of samples except triclosan, which was detected in 73% of prenatal and 69% of peripubertal samples (Table II), and butyl paraben, which was detected in <40% of samples and therefore was not included in the analyses. Creatinine-corrected biomarker concentrations are shown in Table II. Most levels were similar to those of Mexican-American women in NHANES, although our cohort had slightly higher concentrations of MEP, methyl paraben and 2,4-dichlorophenol (CDC, 2018). The intraclass correlation between the two pregnancy samples was lowest for the phthalates (MEP = 0.39, MBP = 0.19, MiBP = 0.26) but showed moderate correlation for the parabens (methyl paraben = 0.46, propyl paraben = 0.41) and phenols (triclosan = 0.46, benzophenone-3 = 0.56, 2,4-dichlorophenol = 0.46, 2,5-dichlorophenol = 0.55).

Table III shows the associations of prenatal and peripubertal biomarker concentrations with timing of pubertal milestones in girls. With prenatal urinary biomarkers, we observed an association of earlier pubertal milestones with higher concentrations of MEP, triclosan and 2,4-dichlorophenol. Specifically, each doubling of prenatal MEP concentration was associated with a shift in timing of pubarche by −1.3 months (95% CI: −2.5, −0.1) and each doubling of concentrations of prenatal triclosan and 2,4-dichlorophenol (a degradation product of triclosan) was associated with a shift in menarche of −0.7 months (95% CI: −1.2, −0.2) and −0.8 months (95% CI: −1.6, −0.0), respectively.

With peripubertal exposure in girls, we observed associations of earlier thelarche (mean shift = −1.1 months, 95% CI: −2.1, −0.0), pubarche (mean shift = −1.5 months, 95% CI: −2.5, −0.4), and menarche (mean shift = −0.9, 95% CI: −1.6, −0.1) with each doubling of urinary concentrations of methyl paraben and earlier pubarche (mean shift = −0.8, 95% CI: −1.6, −0.1) with each doubling of propyl paraben concentrations at age 9. We also observed later onset of pubarche with peripubertal concentrations of 2,5-dichlorophenol (mean shift = 1.0 month, 95% CI: 0.1, 1.9).

The associations of urinary biomarkers with pubertal timing in boys are presented in Table IV. No prenatal biomarkers were associated with pubertal timing in boys. With peripubertal concentrations, we observed an association of earlier gonadarche with each doubling of propyl paraben (mean shift = −1.0 months, 95% CI: −1.8, −0.1).

When we examined urinary biomarker concentrations as categorical variables by quartiles (Supplementary Table S1), results were similar to the continuous results shown in Table III. Figure 1 shows that the associations in girls between prenatal MEP and triclosan with earlier pubarche and menarche, respectively, were also present when MEP and triclosan concentrations were treated as categorical variables. Associations with peripubertal methyl paraben, propyl paraben and 2,5-dichlorophenol in girls also persisted in categorical analyses (Figure 1), although these findings had reduced power.

Figure 1.

Association of quartiles of exposure with timing of puberty in girls.

In sensitivity analyses, the results were very similar when urinary dilution was corrected using specific gravity rather than creatinine (not shown), and when prenatal and peripubertal biomarker concentrations were included in the same model (not shown). Mediation analysis found no evidence of natural indirect effects for any of the exposures and outcomes (Supplementary Table S2 and S3), suggesting that childhood obesity did not mediate the observed associations.