Example: Neurobehavioral Effects of Exposure to Trichloroethylene through a Municipal Water Supply (Reif et al. 2003)
The basis for this study was initially a cross-sectional study of exposure to a number of chemicals with documented release in a community adjacent to a Superfund waste site, the Rocky Mountain Arsenal (RMA) near Denver, Colorado, USA. Study participants were randomly selected from an area within 1.61 km (1 mile) that abutted to the north, northwest, and west boundaries of the site, where fugitive chemicals had been detected in ground and surface waters, sediments, and soils (Figure 4). A total of 585 persons who had lived at their current residence for at least 2 years were eligible for the study; 472 participated. Results of the initial study warranted a second study, conducted in 1991, during which the researchers interviewed and conducted neurobehavioral testing of 204 adults originally identified by the first study (ATSDR 1996). Results of the 1991 study showed a trend toward an increased prevalence of neurologic disorders and adverse reproductive outcomes, particularly in the area north/northwest of the RMA, compared with communities at a greater distance from RMA, presumed to be unexposed to the site. However, the researchers again relied on proximity to the RMA as a surrogate for exposure, and there was evidence that this may have resulted in nondifferential misclassification of exposure, which tends to drive the effect estimate or relative risk toward the null value (Copeland et al. 1977). The researchers initiated a revised exposure assessment using a GIS-based analysis of fate and transport of chemicals in the groundwater regimen hydraulically downgradient from the RMA site. The researchers selected trichloroethylene (TCE) as the marker contaminant for the exposure assessment because of its neurotoxi-cologic properties, and because it had been detected in water supply wells in the study area. The researchers constructed an operable MODFLOW (U.S. Geological Survey, Reston, Virginia, USA) simulation model that accurately reflected hydraulic characteristics of groundwater regime in the study area and used a GIS to develop input variables to the model, including source location of TCE on the RMA site. However, the researchers could not validate TCE levels measured in water wells used by the local water district (LWD), where 90% of the study population resided. The researchers expanded the geographic extent of their study area, and determined that the source of TCE in the groundwater was from multiple hazardous waste sites, including some located outside the original study area. Once the primary source was properly identified, the researchers confirmed the measurement results of TCE in the LWD supply wells by the groundwater model. TCE levels in the wells were then used as input to a hydraulic and water quality simulation model, EPANET (Rossman 1994), to predict TCE levels in the distribution system of the LWD. The researchers used GIS to geocode the study population, develop input data for the simulation model, and assign individual exposure to TCE by linking results of the model to the census block group of residence (Figure 4). The study with the refined exposure assessment found a stronger association of risk for neurobehavioral disorders in the study population than was found in the 1991 study, in which exposure was based primarily on proximity to a source of chemical contamination, including TCE. The study demonstrates that GIS-based technology can be used to refine exposure for epidemiologic investigations, improving sensitivity and specificity beyond a simple proximity metric. It also demonstrates the effect that selection of operational scale can have on exposure assessment in an epidemiology study. The operation of the water distribution system could not be discerned when proximate census blocks were used as a surrogate for exposure.