Limitations in the use of race in the study of disease causation

Abstract

Tremendous variation exists in the rates of many chronic diseases across racial groups. However, serious technical and conceptual limitations hamper the ability of racial comparisons to illuminate the causative pathways. First, race is confounded by social class, which is complex, and like other confounders of race, may not be measured with equal validity across racial groups. Second, statistical "adjustments" for race effects can be misleading since residual confounding may be misconstrued as a genetic effect. Third, the biologic concept of race tempts us to ignore the context dependency of genetic expression. When trying to detect genetic effects, both the environmental and genetic contributions must be measured and potential gene-environment interactions accounted for. Unfortunately, this process is beyond our current technical capabilities. To move forward on the problem of prostate cancer and other diseases distinguished by marked ethnic differentials, investigators need a more comprehensive understanding of the factors that mediate the apparent effect of race combined with valid measures of those factors, as well as novel strategies that can help overcome the technical and interpretive limitations of statistical adjustment. Finally, the "grand" theories of race-based genetic susceptibility must be replaced with rigorous criteria to determine when a trait can be ascribed to some genetic origin.