Grantee Research Project Results

Partitioning evidence of association between air pollution and mortality.

Citation:

Janes H, Dominici F, Zeger S. Partitioning evidence of association between air pollution and mortality. Epidemiology 2007;18(4):427-428.

Abstract:

[Article excerpt-first paragraph] We thank Drs. Pope and Burnett for their insightful commentary and for emphasizing the key point of our paper in studies of air pollution and health, it is useful to quantify the evidence of the association between air pollution and health at distinct temporal and spatial scales of variation in the data because the estimated association at some of these scales might be more subject to confounding bias than others. In this paper, we estimate the association between month-to-month variation in average PM2.5 in the previous year and month-to-month variation in mortality rates in 113 U.S. counties from 2000 to 2002. Our approach decomposes evidence for this association into three components: comparisons across counties of average PM2.5 levels and mortality rates (spatial); comparisons across time of national levels in PM2.5 and mortality rates (temporal); and comparisons across counties of deviations in county-specific trends in PM2.5 and mortality from their respective national trends (space-by-time interaction). In this paper, we set aside the spatial component of association, and we estimate the association at the remaining two scales (temporal and space-by-time interaction). We report the associations at each of these scales separately.