Study design of randomized controlled clinical trials of breast cancer screening.

Abstract

Evaluation of population screening must be based on a randomized clinical trial (RCT) with the study population randomized into two arms: an intervention group invited to screening and a control group not invited to screening. Reduced mortality in the intervention group is evidence of a benefit from screening. Individual randomization is the ideal, but cluster randomization is often used for logistical and ethical reasons. The use of volunteer subjects is methodologically acceptable, but results cannot be generalized. Seven RCTs of breast cancer screening by mammography have been carried out in the United States, Canada, Sweden, and Scotland. All the studies, except the Canadian, were designed to assess the effect of screening across a wide range of ages at entry. The question of the efficacy of breast cancer screening at younger ages (< 50 years) arose early, after the first results were reported. To address this question, basic elements of the screening protocol must be considered when interpreting the results; these are screening modality (e.g., mammography with or without physical examinations), interscreening interval, and number of screening rounds. This article examines the possible influence of these factors and reviews the design choices and the characteristics of the seven RCTs.