Analytic methods in matched pair epidemiological studies

Abstract

Two methods of analysis of matched pair epidemiologic studies are presented. The within pair difference between study and comparison individuals is evaluated with respect to additional confounding variables while simultaneously maintaining the matching. The effects of additional confounding variables are quantiated by their association with the response variable (i.e. exposure in a case-control study or disease in a cohort study). For a binary response variable a multiple logistic function is used and a method to estimate the odds ratio is given. For a continous response variable multiple regression is used and a method to estimate an odds ratio function is derived.