Flavors in gene-environment interactions.

In this issue, Noel Weiss (Epidemiology. 2007;18:429-430) presents the argument that gene-environment interaction should be approached through analysis of genetic effects within strata of environmental exposure, rather than the analysis of environmental effects within strata of genetic variants. Both approaches depend on a pair of 2-by-2 tables. I suggest that, at least for an initial look, the 2-by-4 table has advantages.