Design and Analysis of Efficient (or at Least Useful) Nested Case-Control Studies

• Activity Number: 146
• Type: Invited
• Date/Time: Monday, July 30, 2012 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): Bryan Langholz*+
• Companies: University of Southern California School of Medicine

Abstract

A unified approach to the analysis of nested case-control studies sampled from epidemiologic cohort data will be presented. This approach provides a natural connection between cohort and case-control methods and further provides links between general case-control sampling designs in continuous and grouped time. Case-control studies from simple binary outcome data are a special case. The generality of the framework opens up a wide range of design options so that researchers can choose a study design that exploits the available data resources to address the primary study questions. Even when the sampling design is complex, the analysis is the natural generalization of the cohort analysis methods estimate the same quantities one would estimate from the full cohort so that data analysis requires no special expertise. Power and sample size calculations are straightforward often analytically derived or by simulation. The approach will be illustrated by a two-phase nested study of severity of endometrial hyperplasia and risk of endometrial cancer.