Why do covariates defined by International Classification of Diseases codes fail to remove confounding in pharmacoepidemiologic studies among seniors?

Abstract
No abstract available