Estimation of Genetic Risk Parameters from Case Control and Family Studies Using a Marginalized Approach

• Activity Number: 12
• Type: Topic Contributed
• Date/Time: Sunday, August 1, 2010 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Jeanine Houwing-Duistermaat*+ and Roula Tsonaka and Marieke De Visser and Frits Rosendaal
• Companies: Leiden University Medical Center and Leiden University Medical Center and Leiden University Medical Center and Leiden University Medical Center
• Keywords: genetic association ; family studies ; random effect models ; marginalized approach

Abstract

In genetic association analysis often two study designs are used: namely case-control and family studies. Combining studies with these different designs can estimate genetic risks more efficiently. In this work we propose a method to obtain pooled estimates using marginalized mixed-effects models. Our approach corrects for the different sampling schemes, accounts for the correlation within families using random-effects and gives pooled parameters with the desired marginal interpretation. The method will be illustrated using two studies on venous thrombosis: (1) the MEGA case-control study in which for each participant, the genotypes, the total number and the number of affected siblings are available, whereas the siblings' genotypes are unknown, and (2) the GIFT family-based study, in which the genotypes of all subjects from sibships with at least two affected siblings are only known.