It is somewhat sobering to note that despite (1) the strong effect of age at onset on smoking behavior, (2) the strong genetic effect of the rs16969968 genotype on smoking behavior, and (3) the subjectively meaningful interaction between the 2 effects (increasing the OR per genotype from 1.27 to 1.46), the large sample detected this interaction with a somewhat modest P=.01. This result highlights the power needed to detect gene-environment interactions in complex disease.
