A simple estimator of minimum detectable relative risk, sample size, or power in cohort studies

Abstract

In planning simple cohort mortality studies, researchers need to know what size of relative risk may be confidently detected with the projected size of the cohort and length of follow-up. Although methods for the calculation of such minimum detectable risks have been devised for case-control studies and for cohort studies where internal comparisons are the focus, this has not been explicitly done for cohort studies using external comparisons. This paper describes a simple procedure designed explicitly for investigating the adequacy of cohort size at the planning stage of a study. An example is presented of a retrospective cohort study of men in a Canadian factory. A method is shown for estimating the minimum detectable underlying relative risk for lung cancer in this cohort.