Daily Mortality and Air Pollution in Atlanta: August 1998–December 2006

Abstract

ISEE-0879 Background and Objective: Authors have previously published using two years of mortality data for the two largest counties in Atlanta, GA (Inhalation Toxicology 16, 2004). In response to questions regarding limiting the geographic area to only two counties, the authors investigate whether the estimated effects remain consistent in terms of magnitude and/or statistical significance with the addition of an extended mortality data series and/or larger geographic area of decedents. Methods: The authors estimate the effects of pollution on human mortality for various AQI among decedents at least 65 years of age during 8/1/1998–12/31/2006. The authors also compare the effects for a four-county geographic area (Fulton, DeKalb, Gwinnett and Cobb counties) to the two-county area (Fulton and DeKalb counties) investigated previously. Results: The estimated effects differ depending on the time series and geographic area. The estimated effect of PM2.5 is lower in both magnitude and significance when estimated over the extended time series. Specifically, the published estimated effect for PM2.5 was .0054 (2.96 t-value) for the period 8/1/1998–7/31/2000 compared to .0016 (1.83 t-value) for 8/1/1998–12/31/2006. The estimated effect declined further when estimated using mortality within the four-county geographic area. Results for other AQI are similar in that the estimated effects using the larger geographic area are generally lower, although t-values are higher in some instances. Conclusion: Since it is difficult to define the “best” or most appropriate geographic area related to a specific measurement site, a comparison of various geographic definitions can illustrate the affect of different opinions on the estimated effects. The authors have shown the impact for a particular study location which should be replicated in other locations. It is likely, however, that no general pattern exists.