A FORTRAN program for stratified analysis of data from case-control studies.

Stratification of study subjects by one or more covariates is a commonly accepted method for dealing with confounding and effect modification in epidemiologic case-control studies. A flexible FORTRAN program is described which facilitates simultaneous stratification by several covariates and which produces summary odds ratio estimates and chi-square statistics by the Mantel-Haenszel method. It also facilitates detection of effect modification by each covariate considered individually. Straightforward means are provided for the user to modify input data before analysis or to exclude certain subjects from analysis, simulating such capabilities in larger statistical packages.