Dependent Exposure in a Matched Case-Control Study

• Activity Number: 638
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 8:30 AM to 10:20 AM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Mark Delorey*+
• Companies: CDC
• Keywords: Case-control study ; Matched design ; Profile likelihood ; Bayesian methods

Abstract

In a matched, case-control study, the goal is to determine whether or not a specific risk factor is observed more (or less) frequently in the cases than in the controls. An assumption under this design is that the presence or absence of the risk factor is an independent event across cases and across controls within the same matched group. This assumption is violated when only a fixed number of subjects within a matched group are allowed to exhibit the risk factor. As an illustration, we describe a study in which the risk factor for plague is whether or not a pet sleeps in the same bed as one of its family members. Both frequentist and Bayesian estimates of the odds ratio are derived for the case in which the number of individuals allowed to exhibit the risk factor is the same in each group.