Simulating Multiple Chemical and Non-chemical Exposures for a Community-based Cumulative Risk Assessment in New Bedford, Massachusetts

摘要

Cumulative risk assessment requires models of joint exposures to multiple chemical and non-chemical stressors, but it is challenging to apply these models with sufficient geographic and demographic resolution to accurately identify high-risk subpopulations. For an effects-based cumulative risk assessment of a low-income community living near a Superfund site (New Bedford, Massachusetts), we developed a modeling platform to provide high geographic resolution exposure estimates. First we developed and applied a simulation approach to combine coarse-resolution multivariable demographic data from the Public Use Microdata Sample of the American Community Survey with US census tract resolution constraints using probablistic reweighting with simulated annealing. The resulting synthetic microdata included detailed individual demographics for all New Bedford simulated residents as well as census tract of residence. We then developed regression models from publicly available datasets predicting exposures as a function of covariates in the microdata, focusing on exposures relevant to two outcomes of interest (attention deficit hyperactivity disorder-like behavior and blood pressure). This included models of smoking, alcohol consumption, fish consumption, and other risk factors from New Bedford-specific data within the Behavioral Risk Factor Surveillance System; models including prenatal PCB, lead, mercury, and DDE exposures from an established birth cohort in New Bedford; and models of adult exposures to lead, cadmium and PCBs from the National Health and Nutrition Examination Survey. Finally we applied the exposure regression models to the synthetic population. The resulting exposure models illustrated important correlations among predictors of chemical and non-chemical stressors in the New Bedford community, and the calculated individual-level exposures provided insight regarding high-risk neighborhoods and demographic groups.