Applied Logistic Regression, 3rd Edition

Chapter 7: Logistic Regression for Matched Case-Control Studies

7.1 Introduction

An important special case of the stratified case-control study discussed in Chapter 6 is the matched case-control study. A discussion of the rationale for matched studies may be found in epidemiology texts such as Breslow and Day (1980), Kleinbaum et al. (1982), Schlesselman (1985), Kelsey et al. (1986), and Rothman et al. (2008). In this study design, subjects are stratified on the basis of variables believed to be associated with the outcome. Age and gender are examples of commonly used stratification variables. Within each stratum, samples of cases and controls are chosen. The number of cases and controls need not be constant across strata, but the most common matched designs include one case and from 1–5 controls per stratum and are thus referred to as matched studies.

In this chapter we develop the methods for analyzing general matched studies. We illustrate the methods for both the matched study and a matched study (as an example of the more general design).

We begin by providing some motivation ...