The Shanghai-Minhang Birth Cohort Study (S-MBCS) is an ongoing prospective cohort study designed to determine the distributions of a wide spectrum of maternal environmental exposures and examine their effects on pregnant women and their children (19, 20). From April to December 2012, pregnant women were recruited during their first prenatal care visit (at 12–16 weeks of gestation) at Maternal and Child Health Hospital of Minhang district in Shanghai, China. Briefly, women were eligible for inclusion if they were registered residents of Shanghai, had no history of hospital-diagnosed chronic disease, planned to be deliver their baby in the study hospital, and were willing to attend specified interviews during pregnancy and after delivery. In total, 1,292 eligible pregnant women agreed to participate in this study and completed a structured questionnaire at enrollment.

At delivery, 1,225 participants gave singleton live births, including 667 (54.4%) boys and 558 (45.6%) girls. A total of 611 women voluntarily provided cord blood samples during delivery. Due to limited funding, concentrations of THs and TSH were measured in a subset of 348 cord blood samples of infants with complete information at delivery, sufficient cord plasma volume (≥1 ml), and at least one follow-up at 12 or 48 months of age. Among them, two AGD metrics were measured for 186, 153, 154, and 163 boys, and for 144, 115, 105, and 124 girls at birth, 6, 12 and 48 months of age, respectively. Ultimately, 344 children (194 boys and 150 girls) who both have cord plasma concentrations of THs and TSH and at least one AGD measurement at the four time points were included in the present study, as shown in Figure 1 .

The study protocol was approved by the ethical committee of Shanghai Institute of Planned Parenthood Research. All women provided informed consent for themselves and their children before participating in the S-MBCS and postnatal follow-up visits.

Cord blood samples were collected from the umbilical vein at the time of birth. The samples were centrifuged to retrieve plasma, which was immediately frozen at −80°C until shipment to the clinical laboratory of the affiliated hospital of Shanghai Institute of Planned Parenthood Research for the measurement of THs and TSH concentrations. An Electrochemiluminescence immunoassay with kits obtained from Roche Cobas e601 analyzer was performed to measure the concentrations of TSH, TT₄, FT₄, TT₃, and FT₃ (21). The analytical sensitivity was 0.3 nmol/L, 5.4 nmol/L, 0.4 pmol/L, 0.3 pmol/L, and 0.27 μIU/ml for TT₃, TT₄, FT₃, FT₄, and TSH, respectively. We also measured thyroid peroxidase antibody (TPOAb) concentration. TPOAb positivity was defined as a TPOAb concentration of ≥ 34 IU/L.

The methods and procedures for the measurement of AGD have been described in detail elsewhere (22–24). In girls, AGD_AC was measured from the center of the anus to the anterior surface of the clitoral hood, and AGD_AF was measured from the center of the anus to the posterior end of the fourchette. In boys, AGD_AP was measured from the center of the anus to the anterior base of the penis where the penile tissue met the pubic bone, and AGD_AS was measured from the center of the anus to the posterior base of the scrotum where the skin changed from rugate to smooth. All children were placed in a dorsal decubitus position with both legs held back in a frog leg position to obtain accurate AGD measurements. Before each follow-up visit, we conducted standardized training to ensure the accuracy of AGD measurements. AGD measurements were conducted using the same Vernier caliper with increments of 0.1 mm in children during their first 3 days of life and at 6 months ± 2 weeks, 12 months ± 2 weeks, and 48 months ± 2 weeks of age by the same four trained examiners. To evaluate the inter-examiner variability, two examiners took independent measurements on each newborn on the same day using the same method in 15 girls and 15 boys. The intra-class correlation coefficients of AGD_AP, AGD_AS, AGD_AC and AGD_AF were 0.836, 0.731, 0.624, and 0.722, respectively, indicating moderate-to-good inter-rater reliability for the AGD measurements (25). The examiners who participated in the AGD measurements had no knowledge regarding the neonatal THs or TSH concentrations.

The demographic characteristics, health conditions, medical and reproductive histories, and lifestyle factors (e.g., smoking and drinking habits) of pregnant women and their partners were obtained at enrollment through a structured questionnaire. After delivery, information on infant sex, birth weight, date of birth, mode of delivery, and gestational age was extracted from the electronic medical records of the study hospitals. Feeding practices and additional information were obtained from questionnaires filled by the mother or another caregiver at each postnatal follow-up visit.

The distributions of participant characteristics were summarized as mean (standard deviation, SD) for continuous variables or counts and percentages for categorical variables. The distributions of TT₃, TT₄, FT₃, and FT₄ concentrations were normal, whereas that of TSH concentrations were skewed. Thus, TSH concentrations were log₁₀-transformed for further analyses. Since AGD was reported to be dependent on body weight, we used the anogenital index [AGI = AGD/weight (mm/kg)] as a weight-normalized index of AGD, as reported previously (26). We also used mean (SD), and percentiles to describe the distributions of AGI at birth and at 6, 12, and 48 months of age and to describe the distributions of neonatal TT₃, TT₄, FT₃, FT₄, and TSH concentrations.

The characteristics of the 344 mother–child pairs are presented in Table 1 . The mean (SD) age of mothers at conception was 28.09 (3.38) years and the mean (SD) gestational age was 39.67 (1.18) weeks. Most of mothers were Han Chinese (97.66%), nulliparous (86.88%), received a college degree or higher (78.13%), reported a monthly per capita household income >4000 RMB (79.53%), and had normal weight before pregnancy (72.78%). In addition, 58.31% mothers reported that they were not exposed to passive smoking before pregnancy, and 67.44% fathers reported no alcohol consumption within 3 months before conception. All newborns had a normal 5-min Apgar score (≥8). There was no statistically significant difference in demographic characteristics between the included mother–child pairs and those excluded ( Table S2 ).

All AGI measures were normally distributed. In girls, the mean (SD) AGI_AC was 8.75 (1.26), 4.52 (0.92), 4.06 (0.73), and 3.25 (0.67) mm/kg and the mean (SD) AGI_AF were 2.53 (0.70), 1.80 (0.58), 1.71 (0.52), and 1.78 (0.42) mm/kg at birth, 6 months of age, 12 months of age, and 48 months of age, respectively. In boys, the mean (SD) AGI_AP was 11.73 (1.54), 7.50 (1.25), 6.49 (1.09), and 5.29 (0.68) mm/kg and the mean (SD) AGI_AS was 4.32 (1.13), 3.10 (1.19), 2.79 (0.87), and 2.49 (0.53) mm/kg at birth, 6 months of age, 12 months of age, and 48 months of age, respectively ( Table 2 ). The distributions of AGD from birth to 48 months of age are also shown in Figure S1 .

In all subjects, the mean concentrations of TT₃, TT₄, FT₃, FT₄ and TSH in cord plasma were 0.86 nmol/L, and 94.12 nmol/L, 1.82 pmol/L, 14.22 pmol/L, and 6.47 uIU/mL, respectively, as shown in Table 2 .

In univariate analyses ( Table S3 ), children whose parents were younger than 25 years, whose fathers drank alcohol within 3 months before conception, and who were delivered vaginally (vs. cesarean section) had significantly higher neonatal TSH concentrations. Additionally, children born to mothers who were underweight (vs. normal weight) and gave birth via vaginal delivery had lower FT₃ and FT₄ concentrations. Further, children born to mothers who were exposed to passive smoking before pregnancy had higher FT₃ and FT₄ concentrations and children born to fathers who drank alcohol within 3 months before conception had higher FT₃ concentrations.

In general, multiple linear regression models showed that AGI_AC and AGI_AF in girls tended to be lower in the higher tertile groups of TT₄, FT₄, and FT₃ in cord plasma than those of girls in the lowest tertile, although statistical significance was not reached for all time points ( Figure 2 ). Girls who had higher concentrations of TT₄ had lower AGI_AC at birth (β = -0.56, 95% CI: -1.08, -0.04 for highest vs. lowest tertiles) and 6 months of age (β = -0.57, 95% CI: -1.03, -0.11 for middle vs. lowest tertiles), and lower AGI_AF at 6 months of age (β = -0.35, 95% CI: -0.65, -0.06 for middle vs. lowest tertiles). Higher FT₄ concentrations were associated with lower AGI_AC at birth (β = -0.58, 95% CI: -1.11, -0.05 for middle and β = -0.74, 95% CI: -1.26, -0.22 for highest vs. lowest tertiles) and 48 months of age (β = -0.35, 95% CI: -0.67, -0.02 for middle vs. lowest tertiles), and with lower AGI_AF at 6 months of age (β = -0.35, 95% CI: -0.65, -0.05 for middle vs. lowest tertiles) and 48 months of age (β = -0.22, 95% CI: -0.43, -0.02 for middle vs. lowest tertiles). In addition, girls in the highest tertile of FT₃ concentrations had lower AGI_AF at birth (β = -0.39, 95% CI: -0.68, -0.10) than those in the lowest tertile.

Similar inverse associations of TT₄, FT₄, and FT₃ with AGI in girls were also found in the GEE models ( Table 3 ). Higher neonatal TT₄ and FT₄ concentrations were associated with lower AGI_AC (TT₄: β = -0.27, 95% CI: -0.50, -0.03 for middle vs. lowest tertiles; FT₄: β = -0.38, 95% CI: -0.61, -0.16 for middle and β = -0.30, 95% CI: -0.55, -0.04 for highest vs. lowest tertiles). Additionally, girls in the highest tertile of FT₃ concentrations had lower AGI_AF than those in the lowest tertile (β = -0.22, 95% CI: -0.36, -0.08).

For THs, the pattern of the results in multiple linear regression models ( Figure 3 ) was inconsistent from birth to 48 months of age, although there were a positive association between TT₃ concentrations and AGI_AP at 6 months of age (β = 0.61, 95% CI: 0.10, 1.11 for highest vs. lowest tertiles) and an inverse association between TT₄ concentrations and AGI_AP at 6 months of age (β = -0.69, 95% CI: -1.19, -0.19 for middle vs. lowest tertiles).

Boys in the higher tertile of log₁₀-transformed TSH had higher AGI_AS at birth (β = 0.53, 95% CI: 0.13, 0.92 for middle and β = 0.50, 95% CI: 0.09, 0.91 for highest vs. lowest tertiles) and 12 months of age (β = 0.41, 95% CI: 0.04, 0.79 for highest vs. lowest tertiles; Figure 3 ). The positive association between TSH concentrations and AGI_AS remained in GEE models (β = 0.23, 95% CI: 0.0005, 0.46 for highest vs. lowest tertiles; Table 3 ), but the interaction term between TSH and child age at follow-up was statistically significant, indicating that the association between TSH and AGI_AS may change over time (P < 0.1; Table S2 ).

We conducted several sensitivity analyses using GEE models. When we used AGD instead of AGI as the dependent variable, inverse associations between concentrations of TT₄, FT₃, and FT₄ and AGD were also observed in girls, but the association between TSH concentrations and AGD in boys became non-significant in GEE models ( Table S4 ). The results were virtually identical in sensitivity analyses after excluding infants with TPOAb positivity, with gestational age <37 weeks, or those whose mothers reported prior or current diagnoses of thyroid-related disease/related medication use, and after restricting girls whose mothers had a normal weight ( Tables S5–S8 ). Similar patterns were also observed in girls who were delivered vaginally, although some associations were not statistically significant, possibly due to the small sample size ( Table S9 ).