Abstract:
|
We present statistical methods for jointly estimating acute and chronic health effects associated with short- and long-term exposure to air pollution in the US. Our methods encompass statistical approaches to time-series and prospective cohort designs, which estimate acute and chronic effects separately. Specifically, we develop hierarchical models to estimate both acute and chronic effects within counties, geographical regions, and the nation, and to impute key individual-level covariates (such as smoking) from additional data bases.
Our statistical approaches are motivated by an analysis of two large national cohorts (Medicare and Veterans) merged with the national air pollution monitoring network. The merged data base consists of a 1999-2001 follow-up of approximately 2 million people for whom we have individual-level mortality, morbidity, and socio-demographic covariates (age, gender, ethnicity, nursing home residence), and who are living in 327 US counties providing daily air pollution and weather data.
|