A computer program for incidence density sampling of controls in case-control studies nested within occupational cohort studies

Abstract

A nested case-control study is a case-control study performed with a cohort study. The nested case-control design is useful because the number of study subjects for whom risk factor information is needed is smaller than in the original cohort study. The approach is especially helpful when collection of data is expensive or time consuming, as when it is necessary to contact subjects or next of kin. Selection of controls in nested case-control studies is best performed with a method called "incidence density" sampling. An approach to incidence density sampling that yields a valid estimate of the rate ratio is sampling without replacement from noncases at the time of case occurrence. To implement this approach, a system of computer programs was written that randomly selects a user defined number of controls for each case. The case-control data produced by this program can be analyzed by conditional logistic regression.