The CHDS recruited 15,528 mothers who received obstetric care through the Oakland, California, Kaiser Foundation Hospital for 20,754 pregnancies between 1959 and 1967. More than 90% of those eligible participated in the CHDS. From 2005 through 2008, 80% of the 1,526 daughters who were eligible for the breast density study were successfully traced, and 85% of traced daughters participated. Of the 527 women for whom DDTs, hypertension status, diabetes status, and BMI data were available, 156 were part of the high-birth-weight oversample, 70 were nonsisters, 301 were sisters (140 sets of pairs and 7 sets of “threes”), and 70 reported taking antihypertensive medication due to physician-diagnosed hypertension at the time of interview. shows the distributions of potential confounders according to p,p´-DDT tertiles in the study population. Proportionally more African-American women were in the highest tertile of prenatal p,p´-DDT exposure than other women. Mothers’ age at birth and daughters’ age at interview were highest among women in the highest p,p´-DDT tertile.
Distribution of maternal, early-life, and adult characteristics by prenatal p,p´-DDT exposure tertiles in 527 women of the CHDS [n (%) or LSM (95% CI)].
The unadjusted risk of medicated hypertension was increased substantially among women with prenatal p,p´-DDT in the highest two tertiles of the study population (). There were fewer cases of medicated hypertension among women in the lowest tertile of prenatal exposure to each of the DDTs compared with their respective higher tertiles (). The unadjusted prevalence of medicated hypertension which arose in women < 35 years old was 4.0% among women with prenatal p,p´-DDT in the highest two tertiles of the study population and only 0.6% in the lowest tertile. When we adjusted for confounding by maternal race/ethnicity in model 2, adjusted hazard ratios (aHR) for the middle and highest tertiles of prenatal p,p´-DDT exposure compared with the lowest tertile were 3.6 (95% CI: 1.8, 7.2) and 2.5 (95% CI: 1.2, 5.3), respectively. Consistent with expectations, medicated hypertension was associated with known risk factors including daughters’ race/ethnicity, menopausal status, BMI, diabetes, and age (data not shown), but only race/ethnicity, menopausal status, BMI, and diabetes were retained in the final model 3 based on p-values < 0.05, and associations based on this model were similar to those from model 2. Although both the model of early-life confounding (model 2) and the model of contemporary risk factors that predicted daughter’s hypertension (model 3) were both included in model 4 (mothers’ race/ethnicity, daughters’ race/ethnicity, menopausal status, BMI, and diabetes), only the covariables from model 3 remained significant, and the p,p´-DDT effect size remained comparable (). The other DDTs—o,p´-DDT and p,p´-DDE—were not significantly associated with medicated hypertension in any of the models ().
Hazard ratios (HRs) of the association between prenatal DDTs and daughters’ medicated hypertension.
Birth weight as a continuous measure was not significantly associated with medicated hypertension (model 3 aHR = 1.00; 95% CI: 0.99, 1.00), but low birth weight (< 2,500 g vs. ≥ 2,500 g) was (aHR = 3.3; 95% CI: 1.1, 9.9). However, associations between the second and third tertiles of p,p´-DDT and hypertension increased when adjusted for low birth weight (model 2 aHR = 4.0; 95% CI: 1.9, 8.2 and aHR = 2.9; 95% CI: 1.3, 6.2, respectively, and model 3 aHR = 4.0; 95% CI: 2.0, 8.0 and aHR = 2.8; 95% CI: 1.4, 5.9, respectively). There was also no significant interaction of either maternal (p = 0.6 for both tertiles 2 and 3) or daughter BMI (p = 0.8 and 0.9 for tertiles 2 and 3, respectively) on the association between prenatal p,p´-DDT exposure and medicated hypertension (model 3, data not shown).
We conducted secondary data analyses of alternative exposure classifications. Model 1 HRs for the association between prenatal p,p´-DDT and medicated hypertension were 1.9 (95% CI: 1.2, 3.1) for p,p´-DDT above the median of 9.00 ng p,p´-DDT/mL serum; compared with the lowest quartile of exposure, the HR for the second (6–9 ng p,p´-DDT/mL serum), third (9–13.39 ng p,p´-DDT/mL), and fourth quartiles (≥ 13.59 ng p,p´-DDT/mL) were 3.0 (95% CI: 1.1, 8.0), 4.5 (95% CI: 1.8, 11.6), and 3.2 (95% CI: 1.2, 8.7), respectively. Further, associations between p,p´-DDT and medicated hypertension based on a model mutually adjusted for all three DDTs (without other covariates) were similar to those based on model 1 (tertile 2 HR = 3.5; 95% CI: 1.7, 7.2 and tertile 3 HR = 2.6; 95% CI: 1.1, 6.1).
Associations with prenatal p,p´-DDT and hypertension were weaker when the case definition included all women with a self-reported diagnosis of hypertension (n = 111) (model 3 aHR = 2.1; 95% CI: 1.3, 3.5 and aHR = 1.5; 95% CI: 0.9, 2.7 for tertiles 2 and 3 relative to tertile 1, respectively) compared with associations with cases limited to women who reported using antihypertensive medication only (n = 70, ). HRs increased when the oversampled daughters with birth weight > 3,835 g were excluded (model 3 aHR = 4.2; 95% CI: 1.9, 9.3 and 2.8; 95% CI: 1.2 to 6.9, respectively, n = 345). Last, the unadjusted association of prenatal p,p´-DDT with medicated hypertension was similar when the sample was restricted to nonsisters plus either the oldest (tertile 2 HR = 2.5; 95% CI: 1.1, 5.5 and tertile 3 HR = 2.4; 95% CI: 1.1, 5.3, n = 61 hypertensives) or second-born (tertile 2 HR = 2.6; 95% CI: 1.2, 5.9 and tertile 3 HR = 2.7; 95% CI: 1.2, 6.1, n = 52 hypertensives) sister per sibship (n = 380 total).