Covariates and confounding in epidemiologic studies using metabolic gene polymorphisms.

The relationship between exposure and disease when biomarkers are introduced in an epidemiologic study is explored and summarized. In molecular epidemiologic studies, biologic measurements play a major role as markers of exposure, disease or susceptibility to disease and/or exposure. In this scenario, the definition and management of confounding factors may change. Sometimes the presence or activation of the biomarker is partially caused by the relevant environmental exposure, and therefore the 2 variables (exposure and biomarker) should not be always treated as confounders of each other. Models of exposure-disease association in the presence of biologic markers are presented. The concept of confounders is reviewed in light of the role of biomarkers in the pathway between exposure and disease. Copyright 2002 Wiley-Liss, Inc.