The impact of misclassification in case-control studies of gene-environment interactions.

In this chapter we describe the impact of risk factor misclassification in case-control studies designed to estimate gene-environment interactions. We show that under certain scenarios even small amounts of exposure or genotype misclassification can substantially attenuate the interaction effect and, as a consequence, dramatically increase the sample size required to study these interactions. A consideration of how sample size is affected by exposure and genotype misclassification in the study design phase should help to identify situations where obtaining better risk factor information is crucial for the feasibility of studies.