The health effects of air pollution: A reanalysis

Using a newly assembled data base, we explore the relationship between mortality and measures of sulfates and suspended particulates in 96 cities, counties, and metropolitan areas (SMSAs) of the United States. The data fit the previously estimated Lave-Seskin model and we cannot reject the null hypothesis of identical air-pollution coefficients in 1960, 1969, and 1974 cross sections. A strong, consistent, and statistically significant association between sulfates and mortality persists. The association is changed little by adding variables for smoking and alcohol consumption, by using a generalized least-squares estimator instead of OLS, by using city or county instead of SMSA data, or by adding medical care and nutrition variables and using a simultaneous equation framework. These results can be used to support stringent abatement of sulfur-oxides air pollution.