You are here:
Application of alternative spatiotemporal metrics of ambient air pollution exposure in a time-series epidemiological study in Atlanta
Citation:
Sarnat, S., J. Sarnat, J. Mulholland, V. Isakov, H. Ozkaynak, H. Chang, M. Klein, AND P. Tolbert. Application of alternative spatiotemporal metrics of ambient air pollution exposure in a time-series epidemiological study in Atlanta. Journal of Exposure Science and Environmental Epidemiology . Nature Publishing Group, London, Uk, 23:593-605, (2013).
Impact/Purpose:
Exposure error in studies of ambient air pollution and health that use city-wide measures of exposure may be substantial for pollutants that exhibit spatiotemporal variability.In this EPA/ORD/NERL sponsored Cooperative Agreement, Emory and EPA researchers evaluated the impact of applying alternative enhanced traffic air pollution exposure indicators in a time-series study of ambient air pollution and cardiorespiratory emergency department visits in Atlanta, GA. Health risk estimates obtained from testing a variety of spatiotemporal metrics of exposure were then examined. We observed stronger associations particularly for traffic-related pollutants when using spatially-refined ambient concentrations compared to a conventional central site exposure assignment approach. For some relationships, estimates of spatially-refined population exposures showed slightly stronger associations than corresponding spatially-refined ambient concentrations. Moreover, by using spatially-refined pollutant metrics, we identified socioeconomic disparities in concentration-response functions that were not observed when using central site data. In some cases, spatially-refined pollutant metrics identified associations with health that were not observed using measurements from the central site. Finally, complexity and challenges in incorporating modeled pollutant estimates in time-series studies were discussed along with recommendations on ways to improve the results from future air pollution epidemiology stduies by using recently available more refined exposure indicators.
Description:
Exposure error in studies of ambient air pollution and health that use city-wide measures of exposure may be substantial for pollutants that exhibit spatiotemporal variability. Alternative spatiotemporal metrics of exposure for traffic-related and regional pollutants were applied in a time-series study of ambient air pollution and cardiorespiratory emergency department visits in Atlanta, GA, USA. Exposure metrics included daily central site monitoring for particles and gases; daily spatially refined ambient concentrations obtained from regional background monitors, local-scale dispersion, and hybrid air quality models; and spatially refined ambient exposures from population exposure models. Health risk estimates from Poisson models using the different exposure metrics were compared. We observed stronger associations, particularly for traffic-related pollutants, when using spatially refined ambient concentrations compared with a conventional central site exposure assignment approach. For some relationships, estimates of spatially refined ambient population exposures showed slightly stronger associations than corresponding spatially refined ambient concentrations. Using spatially refined pollutant metrics, we identified socioeconomic disparities in concentration–response functions that were not observed when using central site data. In some cases, spatially refined pollutant metrics identified associations with health that were not observed using measurements from the central site. Complexity and challenges in incorporating modeled pollutant estimates in time-series studies are discussed.
