|
Abstract:
|
Most previous studies examining health effects of source-specific air pollution have used monitor-specific source contributions estimated under the implicit assumption that pollution levels are homogeneous within a defined range of a monitor or within a city or county, which is often violated. Such methods are subject to non-ignorable exposure measurement error resulting from spatial misalignment between estimated source-specific exposures and health outcomes, leading to biased health effects estimates. This bias can be substantial particularly when assessing differential health effects across racial or ethnic groups whose residence locations are spatially clustered (e.g., within certain census tract or zip code areas). In this study, we assess source-specific exposures using multipollutant spatial data in Harris County, TX, and apply a spatially enhanced source apportionment method that accounts for spatial correlation and enables prediction of source-specific exposures at unmonitored locations. This allows us to more accurately estimate health effects of source-specific exposures among different racial/ethnic groups by reducing spatial misalignment errors.
|
