Table 3.
Results from robust linear regression models for PCB and p,p′-DDE maternal serum concentrations (N = 246).
| Dependent variable | Independent variable | Beta | Std. error | Pr(>|t|) |
|---|---|---|---|---|
| DDE | Age | 11.599 | 4.348 | 0.008 |
| SES | 6.230 | 6.412 | 0.332 | |
| BMI | 0.031 | 3.477 | 0.993 | |
| Serum lipids | 23.995 | 8.987 | 0.008 | |
| PCB 74 | Age | 0.071 | 0.047 | 0.129 |
| SES | 0.037 | 0.064 | 0.564 | |
| BMI | −0.024 | 0.034 | 0.481 | |
| Serum lipids | 0.302 | 0.098 | 0.002 | |
| PCB 99 | Age | 0.158 | 0.069 | 0.024 |
| SES | 0.120 | 0.101 | 0.237 | |
| BMI | −0.057 | 0.055 | 0.295 | |
| Serum lipids | 0.607 | 0.151 | <0.0001 | |
| PCB 118 | Age | 0.696 | 0.173 | <0.0001 |
| SES | 0.480 | 0.263 | 0.070 | |
| BMI | −0.073 | 0.144 | 0.613 | |
| Serum lipids | 1.896 | 0.383 | <0.0001 | |
| PCB 138 | Age | 2.649 | 0.546 | <0.0001 |
| SES | 0.702 | 0.799 | 0.381 | |
| BMI | −0.691 | 0.434 | 0.113 | |
| Serum lipids | 7.101 | 1.141 | <0.0001 | |
| PCB 153 | Age | 5.362 | 0.811 | <0.0001 |
| SES | 0.931 | 1.192 | 0.435 | |
| BMI | −1.828 | 0.643 | 0.005 | |
| Serum lipids | 10.632 | 1.689 | <0.0001 | |
| PCB 156 | Age | 0.580 | 0.089 | <0.0001 |
| SES | 0.087 | 0.131 | 0.505 | |
| BMI | −0.234 | 0.071 | 0.001 | |
| Serum lipids | 1.170 | 0.184 | <0.0001 | |
| PCB 170 | Age | 2.307 | 0.290 | <0.0001 |
| SES | 0.122 | 0.417 | 0.771 | |
| BMI | −0.818 | 0.223 | 0.0003 | |
| Serum lipids | 3.811 | 0.591 | <0.0001 | |
| PCB 180 | Age | 4.751 | 0.508 | <0.0001 |
| SES | 0.242 | 0.727 | 0.739 | |
| BMI | −1.745 | 0.391 | <0.0001 | |
| Serum lipids | 6.444 | 1.029 | <0.0001 | |
| PCB 183 | Age | 0.477 | 0.099 | <0.0001 |
| SES | 0.117 | 0.146 | 0.421 | |
| BMI | −0.108 | 0.078 | 0.167 | |
| Serum lipids | 0.912 | 0.213 | <0.0001 | |
| PCB 187 | Age | 0.966 | 0.160 | <0.0001 |
| SES | 0.049 | 0.237 | 0.837 | |
| BMI | −0.298 | 0.126 | 0.019 | |
| Serum lipids | 1.227 | 0.349 | 0.0005 |
The multivariate regression models revealed consistently negative associations between the compounds and maternal BMI, which reached the level of significance for PCB 153, 156, 170, 180 and 187. There were also consistently positive associations between the compounds and maternal age and maternal serum lipid levels. In this model, we included the compounds of interest as a dependent variable and maternal age, BMI, socioeconomic status and serum lipids as covariates.
Serum lipids as the sum of total serum cholesterol and serum triglycerides in mmol/l, SES socioeconomic status of participants determined using the adjusted socioeconomic index as set out by Winkler, Pr(>|t|) p value for the t test, Std. error standard error, PCB polychlorinated biphenyl, DDE dichlorodiphenyldichloroethylene.