Residence location as a measure of environmental exposure: a review of air pollution epidemiology studies

Abstract

Residence location has long been used to indicate environmental exposure in many epidemiological studies. This indicator is easy to establish, requires little exposure or monitoring data, and is potentially applicable to many types of investigations. The validity, accuracy and utility of residence location as an exposure indicator, however, is challenged by current concerns regarding multiple exposure pathways, persistent and toxic contaminants, and cumulative exposures from non-point, mobile and point sources. This paper reviews 45 epidemiological studies that use residence location to identify study populations and estimate air pollution exposures. Thirteen (29%) of the studies determined environmental exposures based on “proximity” measures, usually the distance from a subject's residence to a pollutant source. Other studies used “zones” presumed to have equal pollutant levels. Several studies combined zone and proximity approaches. Exposures were quantified using monitoring data in 27 (60%) studies and dispersion modeling in two (4%) studies. Sixteen (36%) studies did not use any environmental data to quantify exposure. A total of 31 (69%) of the studies reported significant associations between health endpoints and the pollutant exposures represented by residence location. In general, comprehensive and systematic approaches to identify and estimate population exposures were not used, and the exposure estimates were therefore deemed likely to have great uncertainty. Unless exposure levels among groups are verified, it cannot be determined whether nonsignificant associations between exposures and health endpoints indicate a lack of measurable health effects, or are merely a result of exposure misclassification. Site-specific and quantitative exposure assessments are needed to better quantify and confirm exposures within such studies, as well as to permit interpretations and comparisons across studies.