Methods for Analyzing Panel Studies of Acute Health Effects of Air Pollution. Technical Report No. 25

摘要

New methods are presented for analyzing repeated binary health measurements of individuals exposed to varying levels of air pollution. The methods involve a separate logistic regression of response against environmental covariates for each individual under study. Parameters reflecting individual susceptibility to pollutants and weather are estimated using the regression techniques developed by Cox. The parameters are then combined over the individuals in the study to obtain summary estimates of environmental effects. The approach does not require the independence of successive health measurements. It is illustrated with data on asthma and air pollution in the Los Angeles area. A special case of the approach can be adapted to continuous health measurements; the details of this procedure are outlined in an appendix. (ERA citation 05:001957)