Studies with sufficiently large samples for analyzing interactions with common exposures may still have sample size issues for a relatively rare exposure or genotype of interest, or when a large sample size is created through a meta-analysis of many small studies. In these situations there may be small cell counts where asymptotics break down, leading to unstable results.
