Data analysis issues for protocols with overlapping enrollment

Many persons with HIV require and take several medications. The efficacy and safety of many of these medications are uncertain. Usually limited data on drug interactions are available. Thus simultaneous and sequential enrolment of patients into multiple studies is desired for reasons of science and efficiency. This paper discusses the analysis of data arising from coenrolment in multiple studies sponsored by the Community Programs for Clinical Research on AIDS (CPCRA). Factorial designs and those in which patients are sequentially instead of simultaneously randomized are compared. Approaches to data analysis, based on intention-to-treat, for individual and pairs of trials are described. An antiretroviral trial and a trial for prophylaxis of Pneumocystis carinii pneumonia (PCP) are used for illustration. We conclude that such analyses may yield useful information on drug interactions and that a more vigorous coenrolment policy should be pursued in AIDS research.