We continue to compare case-control designs that use population controls versus those that use controls selected from their relatives (i.e., siblings, cousins, or ``pseudosibs'' based on parental alleles) for estimating the effect of candidate genes and gene-environment interactions. We have now produced three manuscripts investigating these study designs. The first manuscript compares basic familial designs, evaluating the asymptotic bias in relative risk estimates due to using population controls when there is confounding due to population stratification. Using siblings or pseudosibs as controls completely addresses this issue, whereas cousins provide partial control for population stratification. Next, we show that the conventional conditional likelihood for matched case-control studies can give asymptotically biased effect estimates when applied to the pseudosib approach; the asymptoticbias is towards the null and disappears with disease rarity. This bias reflects the assumption that the pseudosib's (non-transmitted) parental alleles do not lead to disease. We then show that the designs using population or pseudosib controls are generally the most efficient for estimating the main effect of a candidate gene, followed in efficiency by the design using cousins. Finally, we show that the design using sibling controls can be quite efficient, however, when studying gene-environment interactions. In the second manuscript, we investigate the bias and efficiency in using familial designs that are restricted based on family history of disease. These designs help regain some of the efficiency loss due to using family-based controls, while dealing with population stratification. The third manuscript delves into prospective and retrospective likelihoods for family-based case-control designs. We show that using the retrospective likelihoods can improve efficiency for estimating the effects of candidate genes or gene-environment interactions. ?
Showing the most recent 10 out of 922 publications
