We describe an approach to examine the association between exposure to chemical mixtures and a health outcome, using as our case study polychlorinated biphenyls (PCBs) and hypertension. The association between serum PCB and hypertension among participants in the 1999-2004 National Health and Nutrition Examination Survey was examined. First, unconditional multivariate logistic regression was used to estimate odds ratios and associated 95% confidence intervals. Next, correlation and multicollinearity among PCB congeners was evaluated, and clustering analyses performed to determine groups of related congeners. Finally, a weighted sum was constructed to represent the relative importance of each congener in relation to hypertension risk. PCB serum concentrations varied by demographic characteristics, and were on average higher among those with hypertension. Logistic regression results showed mixed findings by congener and class. Further analyses identified groupings of correlated PCBs. Using a weighted sum approach to equalize different ranges and potencies, PCBs 66, 101, 118, 128 and 187 were significantly associated with increased risk of hypertension. Epidemiologic data were used to demonstrate an approach to evaluating the association between a complex environmental exposure and health outcome. The complexity of analyzing a large number of related exposures, where each may have different potency and range, are addressed in the context of the association between hypertension risk and exposure to PCBs.
Keywords: PCBs; cumulative risk assessment; hypertension.