ISSUES IN THE MORTALITY ANALYSIS OF RANDOMIZED CONTROLLED TRIALS OF CANCER SCREENING

This paper is concerned with the analysis and interpretation of randomized controlled trials (RCTs) of periodic screening programs for the early detection of cancer in which there is a lengthy follow-up period without screening being offered. That is, screening is offered for a limited time with a subsequent follow-up period during which the cancer mortality is observed. The paper focuses on tests for a mortality reduction due to screening. Two approaches are presented, one based on the experience of all those randomized to the test and control groups, and the other based on the experience of selected groups of cancer cases found during the trial. In the latter approach it is necessary that the two groups selected for analysis be comparable groups of cancers. The concept of comparability is discussed with emphasis on factors that determine comparability and on ways to assess it. Examples from completed cancer screening RCTs are used to illustrate the ideas and methods presented.