The Impact of Exposure Misclassification and Exposure-Biased Sampling on Power for Detecting Gene-by-Environment Interactions in Case-Control Studies

• Activity Number: 394
• Type: Invited
• Date/Time: Tuesday, August 6, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Bhramar Mukherjee*+
• Companies: University of Michigan
• Keywords: Gene Environment Interaction ; Misclassification ; Design Sensitivity

Abstract

With limited funding and sample availability, determination of an optimal sampling design to achieve the greatest power for detecting gene-by-environment effects is critical. Exposure enriched sampling is often employed to select individuals for genotyping to enhance power of such studies. However, exposure misclassification combined with such biased sampling can affect operating characteristics of different tests for gene-by-environment interaction as well as joint tests for marginal association and gene-environment interaction. The goal of this study is to characterize the impact of exposure misclassification and intentional exposure-biased sampling on the power and bias properties of three common methods for G*E interaction and joint effect detection. We assessed the power, Type I error, bias and mean-squared-error properties of case-only, case-control, and empirical Bayes methods for the test of G*E interaction as well as joint test of a marginal genetic (or environmental) effect and G*E interaction under various scenarios of biased sampling and exposure misclassification. We propose some simple solutions to correct for misclassification.