Invited commentary: the action in the interaction and exposure modification

Abstract

The study of disease variability in populations is a goal of modern epidemiology. Because most common diseases arise out of a combination of factors and events (exposures, heritability, comorbidities, and chance), developing simple models of characterizing joint events is a daunting task. Dr. Weinberg argues successfully in this issue of the Journal (Am J Epidemiol. 2012;175(7):602-605) that additive null models can capture pure forms of independent causal effects in studies of rare conditions. Moreover, the concept of exposure modification, which characterizes most gene-environment interactions reported to date, is introduced. More cross-talk between biologists and epidemiologists is needed to tackle key issues in chronic disease etiology, and the argument for the use of parsimonious joint models in epidemiology is convincing.