3.1. Description of the Sample
Of the 389 live singletons that were born into the study, 350 completed a five-week NNNS assessment and had at least one prenatal urine sample from their mothers analyzed for BPA or phthalate metabolites. Characteristics of the sample are shown in . Women averaged 30 years old at delivery of the infant. The majority were married, had greater than high school education, and were employed during pregnancy. Racial makeup of the women was 63% white, 30% black, and 7% other. Based on medical chart review, the mean gestational age of infants at delivery was 39 weeks, and mean birth weight was 3389 grams. Forty six percent of infants were male, and they averaged 34.5 days at the time of the five-week NNNS assessment.
Characteristics of the Sample
3.2. BPA and Phthalate Measures
At least 90% of maternal urine samples had detectable concentrations of BPA at the two different measurement time points. All samples had detectable concentrations of phthalate metabolites for DBP and DEHP. () BPA levels were slightly lower than national data on females reported by the CDC(Centers for Disease Control and Prevention, 2009), but were strikingly similar to those reported in other studies (Cantonwine, et al., 2010
; Wolff, et al., 2008
; Ye, et al., 2008
). Most of the phthalate metabolite levels were substantially higher than those of national samples (Centers for Disease Control and Prevention, 2009). The Pearson correlation of log2
urine BPA concentrations across the two collections times was r = 0.29 (p <0.0001). The Pearson correlation of log2
urine phthalate metabolite concentrations across the two collection times were r = 0.41 (p <0.0001) for DBP, and r = 0.24 (p <0.0001) for DEHP. The Pearson correlations of log2
BPA with log2
phthalate metabolites (DBP, DEHP) at 16 weeks were r = 0.50 and r = 0.42, respectively, and at 26 weeks were r = 0.28 and r = 0.21, respectively (all p < 0.0001).
Phthalate Metabolite and BPA Levels in HOME Study Maternal Urinary Samples in Comparison with National Levels
3.3. Early Infant Neurobehavior
The number of infants receiving scores on the habituation and hypertonicity scales of the NNNS was too small for meaningful interpretation, so we excluded these scales from analyses as we have in our previous work with the NNNS (Yolton, et al., 2009
). Mean scores of the NNNS subscales for the sample were within the ranges reported by Tronick (Tronick, et al., 2004
) and Lester (Lester, et al., 2004
). displays bivariate associations between our exposure measures and NNNS subscales. The only significant (p < 0.10) bivariate association between BPA concentration in urine and NNNS outcomes, controlling for urinary creatinine, was with the 16 week sample and increased hypotonia. To further describe associations with BPA exposure, we conducted bivariate analyses of urinary BPA concentrations and descriptive characteristics among the sample. Maternal urinary BPA concentrations were significantly associated with several covariates. Higher BPA concentrations were associated with higher levels of maternal urine creatinine, serum cotinine, BMI, and with black race. Maternal BPA concentrations were inversely associated with income and single marital status.
Unadjusted Bivariate Associations between Maternal Phthalate and BPA Measures at Two Times and Early Infant Neurobehavior
In initial analyses with total urinary DBP metabolite concentrations where we only controlled for urinary creatinine (), associations significant at p < .10 included: increased hypotonia with higher total DBP metabolite concentrations at 16 weeks; decreased arousal, higher regulation, decreased handling, higher movement quality, and increased hypotonia scores with higher total DBP at 26 weeks. In similar analyses with the total DEHP metabolite concentrations controlling for urine creatinine, we found a significant positive association at 26 weeks with more nonoptimal reflexes.
3.4. Multivariable Analyses
We conducted multivariate analyses of the associations between total maternal urinary concentrations of BPA, DBP, and DEHP metabolites at the two independent measurement times and NNNS outcomes for those significant at p < .10 in the preliminary analyses where we had only controlled for urinary creatinine. Final multivariable model associations are presented in .
Mulitvariable Models of Associations between Maternal Phthalate and BPA Measures at Two Times and Early Infant Neurobehavior for NNNS Scales with Significant Bivariate Findings at p<.10
In multivariable analysis, the association between 16 week BPA and hypotonia reflected a trend toward increased hypotonia, or decreased muscle tone, that was not significant (p = .09). This was the only association with BPA exposure that we explored with multivariable analysis.
In multivariable analyses, the associations between total DBP metabolites at both 16 and 26 weeks and hypotonia were not significant [OR=1.20 (95% CI: 0.96 – 1.50) and OR=1.22 (95% CI: 0.96 – 1.55), respectively]. When measured at 26 weeks, higher urinary concentrations of DBP metabolites were significantly associated with lower arousal, higher self-regulation (trend), less handling required, and improved movement quality (trend) during the NNNS. We found a significant interaction of sex by urinary DEHP metabolites measured at 26 weeks with respect to non-optimal reflexes (p = .04). A sex-stratified analysis revealed a significant association between DEHP urinary metabolites and increased nonoptimal reflexes in males only. Log2 creatinine, age at exam, and infant sex were retained in all final models regardless of their impact. Models were relatively parsimonious with other retained covariates including weight change from birth (arousal, regulation, nonoptimal reflexes), maternal alcohol use (arousal, regulation), maternal maximum log2 cotinine (arousal), marital status (regulation, movement quality), high risk infant status (regulation, movement quality), income (nonoptimal reflexes), and black race (hypotonia).