A POTENTIAL PITFALL IN CONTROL OF COVARIATES IN EPIDEMIOLOGIC STUDIES

Control of covariates is essential in nonexperimental epidemi ologic s tudies. Important covariates, such as smoking or alcohol consumption, often are crudely categorized in epidemio logic analyses, In this pape r, I illustrate by both hypothetical and empirical examples thar contr ol of crudely categorized covariates can yield strongly misleading res ults. In particular, I show that, under certain conditions, control fo r crudely classified covariates can even be worse than not controlling for such. covariates at ail. I conclude that covariate specification is an issue that requires much more care than it commonly receives in epidemiologic analyses.