We describe the analysis of cohort study data with a standard FORTRAN program which should run on most computers. It provides a summary measure of the mortality (or incidence) rate ratio between the study cohort and some standard population, based either on person-years at risk or on proportional mortality, and adjusted for age, sex and calendar period; a test of the statistical significance of the ratio; and a set of observed death rates in the study cohort. Results may also be produced in a form suitable for use with GLIM. The analysis may be subdivided into a range of time intervals since each subject was first exposed to risk. The program provides for movement of subjects between different 'level-of-exposure' subgroups within the cohort, and for various methods of censoring. It allows considerable flexibility in data management, and is available with complete documentation and a worked example. The program should enable epidemiologists with little computing experience to carry out formal analysis of cohort studies.
Coleman et al. (Sat,) studied this question.