Spatial attributes of blood organochlorine concentrations in Washington County, Maryland, 1974--1989

People living near potential sources of persistent organochlorine compounds may be exposed to higher than background levels of these ubiquitous pollutants. Geographic information is commonly used to evaluate these pollutants in the environment, but overlooked in studies modeling these compounds in humans. The overall goal of this thesis is to evaluate residential location as a potential exposure determinant by investigating the geographic distribution of human biomarkers of exposure to persistent organochlorine compounds in a community with a potential source. The community chosen for this research was Washington County, Maryland because it contains a Superfund site contaminated with organochlorines and a large proportion of the residents had donated their blood for research purposes.; To determine if the Superfund site was a potential source of exposure to County residents, 110 soil samples collected in and around the site were used to predict soil levels of 1,1-dichloro-2,2-bis(chlorophenyl)ethylene (DDE) at unsampled locations with geostatistical methods. Potential risk factors of exposure to organochlorines were then evaluated using generalized least squares regression to model blood samples of 2,184 residents. Participant addresses were geocoded so potential spatial risk factors such as residential distance to the site could be examined. To evaluate the sensitivity of these models to positional bias in the geocoding process, 163 addresses were visited and the distance between the geocoded coordinates and the residence was determined. Conditional simulation was used to estimate this distance in the remaining geocoded addresses, and the robustness of the regression models was tested using a Monte Carlo approach.; The results provide suggestive evidence that the Superfund site is a potential source of organochlorine exposure to surrounding residents. A statistically significant association between blood dieldrin levels and residential distance from the site adjusting for age, gender, smoking status, education and drinking water source was found, although no other associations between the site and other organochlorines were found. The dieldrin association is robust to geocoding positional bias. Overall, spatial information is found to be important for deriving geographic-based predictors of blood organochlorine levels and accounting for spatially dependent model residuals and should be evaluated when dealing with human exposure and biomarkers.