An examination of exposure measurement error from air pollutant spatial variability in time-series studies

• Type of Material: Article
• Authors: Stefanie Ebelt Sarnat, Emory University; Mitchel Klein, Emory University; Jeremy Sarnat, Emory University; W Dana Flanders, Emory University; Lance Waller, Emory University; James A. Mulholland, Georgia Institute of Technology; Armistead G. Russell, Georgia Institute of Technology; Paige Tolbert, Emory University
• Language: English
• Date: 2010-03-11
• Publisher: Nature Publishing Group: Open Access Hybrid Model Option B
• Publication Version: Author Accepted Manuscript (After Peer Review)
• Copyright Statement: © 2009 Rights Managed by Nature Publishing Group
• Final Published Version (URL): http://www.nature.com/jes/journal/v20/n2/full/jes200910a.html
• Title of Journal or Parent Work: Journal of Exposure Science and Environmental Epidemiology
• ISSN: 1559-0631
• Volume: 20
• Issue: 2
• Start Page: 135
• End Page: 146
• Grant/Funding Information: This work was supported by grants to Emory University from the U.S. EPA (R82921301-0), the National Institute of Environmental Health Sciences (R01ES11294), and the Electric Power Research Institute (EP-P27723/C13172 and EP-P4353/C2124), as well as to the Georgia Institute of Technology from the U.S. EPA (RD832159, RD831076 and RD830960).
• Keywords: cardiovascular disease; air pollution; exposure assessment; time-series; emergency department visits; measurement error
• Research Categories: Environmental Sciences; Health Sciences, Public Health; Health Sciences, Epidemiology

Abstract

Introduction Relatively few studies have evaluated the impacts of heterogeneous spatiotemporal pollutant distributions on health risk estimates in time-series analyses that use data from a central monitor to assign exposures. We present a method for examining the impacts of exposure measurement error relating to spatiotemporal variability in ambient air pollutant concentrations on air pollution health risk estimates in a daily time-series analysis of emergency department visits in Atlanta, Georgia. Methods We used Poisson generalized linear models to estimate associations between current day pollutant concentrations and circulatory emergency department visits for the 1998–2004 period. Data from monitoring sites located in different geographical regions of the study area and at different distances from several urban geographic subpopulations served as alternative measures of exposure. Results We observed associations for spatially heterogeneous pollutants (CO and NO2) using data from several different urban monitoring sites. These associations were not observed when using data from the most rural site, located 38 miles from the city center. In contrast, associations for spatially homogeneous pollutants (O3 and PM2.5) were similar regardless of monitoring site location. Conclusions We found that monitoring site location and the distance of a monitoring site to a population of interest did not meaningfully impact estimated associations for any pollutant when using data from urban sites located within 20 miles from the population center under study. However, for CO and NO2, these factors were important when using data from rural sites, located greater than 30 miles from the population center, likely due to exposure measurement error. Overall, our findings lend support to the use of pollutant data from urban central sites to assess population exposures within geographically dispersed study populations in Atlanta and similar cities.