Particulate Air Pollution, Sulfur Dioxide, and Daily Mortality: A Reanalysis of the Steubenville Data

Abstract

A recent analysis (Schwartz & Dockery, 7992a) reported that particulate air pollution was significantly associated with daily mortality in the Steubenville standard statistical metropolitan area (SMSA), even after adjustment for the effects of weather. This statistically significant association was reported to persist, with no attenuation of the magnitude of the effect, when sulfur dioxide was simultaneously considered in the analysis. We undertook a reanalysis of the Steubenville data. Our main goals were (1) to investigate the robustness of the results obtained from regression analysis of the mortality data, and (2) to examine the effect of specific seasons on the relationship between particulate air pollution and daily mortality. When full-year mortality was analyzed, the effect of particulates was Substantially attenuated when sulfur dioxide was simultaneously included in the regression, and was no longer statistically significant. When mortality data were analyzed by season, similar results were obtained. In addition, there appeared to be seasonal effects in the association of particulates and sulfur dioxide with daily mortality. The results of the analysis were not robust to small perturbations in the data. In view of these findings, it is premature to draw any conclusions about the relationship between individual components of air pollution and daily mortality. Finally, some methodologic issues in the analysis of serial mortality data are discussed.