Analysis of Case-Only Studies Allowing for Genotyping Error

Abstract

To assess possible interactions between genetic and environmental factors a frequently used design is the case-only design. With case-only design, a measure of association between genetic and environmental factors is easily computed as a cross-product ratio and logarithm of this ratio has been suggested as an estimate of interaction. A drawback of such a case-only study is that it is impossible to detect genotyping errors and no method has been developed to allow for the presence of undetected genotyping errors. A simulation study given in this paper shows that ignoring the presence of genotyping errors may seriously increase the bias of the point estimate and reduce the power of the likelihood ratio test. The main focus of this paper is to propose methods for analysis of case-only studies when there are genotyping errors. Two major assumptions will be considered: (1) the genetic and environmental factors are independent in the control population, and (2) the genetic and environmental factors are conditionally independent in the control population. We suggest that the error rates of mistyping are estimated from an internal validation study, such as genotyping some diseased individuals from the case sample twice. Using such dataset, we develop a simple and yet highly efficient conditional likelihood approach to make inference about interaction parameter. The performances of the analysis methods were investigated through simulation studies. The simulation results showed that the new methods have acceptable performances.