Evaluation of cancer screening programs: parallels with clinical trials

The methodologic limitations of available observational evidence for the evaluation of ongoing cancer control programs is reviewed. The specific example of Control Programs for Cervical Cancer is chosen to illustrate these limitations. These limitations pertain to the special features of the natural history of this disease, to the types of data available as well as to problems in the interpretation of these data. Given the unacceptability of conducting a rigorous randomized clinical trial of an ongoing and accepted cancer control procedure, an alternative experimental approach is proposed. This approach consists in the random allocation of census tracts with a high risk population to either of two groups: 1) an experimental group in which active and repeated screening activities will attempt to reach virtually all susceptible women, 2) a comparable control group which will only benefit from regularly available services.