Variance estimation of allele-based odds ratio in the absence of Hardy-Weinberg equilibrium

Abstract

In gene-disease association studies, deviation from Hardy-Weinberg equilibrium in controls may cause bias in estimating the allele-based estimates of genetic effects. An approach to adjust the variance of allele-based odds ratio for Hardy-Weinberg equilibrium deviation is proposed. Such adjustments have been introduced for estimating relative risks of genotype contrasts and differences in allele frequency; however, an adjustment of odds ratios for allele frequencies still does not exist. The approach was based on the delta method in combination with the Woolf's logit interval method and the disequilibrium coefficient. The proposed variance adjustment provided better power than the unadjusted one to detect significant estimates of odds ratio and it improved the variance estimation.