Design considerations for AIDS trials

Pressure from a variety of citizen and scientific groups has become increasingly intense to change the ways drugs for treating AIDS are tested. A review is provided of the reactions of a group of statisticians to a document issued by the AIDS Coalition to Unleash Power (ACT UP) in June 1989. The report was entitled 'A National AIDS Treatment Research Agenda', which presented new principles for carrying out AIDS research. This article discusses some of the proposed changes in experimental design. One possibility is to run all three of the traditional drug testing phases concurrently. These include phase I, the study of drug safety; phase II, the study drug effectiveness; and phase III, which compares different drug treatments. Traditionally, phases II and III are randomized. Randomization could be extended to phase I. A discussion of the role of randomization in drug trials is presented. Randomization should be reserved for those patients for whom no treatment will be obviously preferable. Ultimately, randomized trials offer the best chance of accurately evaluating AIDS drugs. The issues of maintaining an untreated control group; the use of placebos; comparisons of different treatments and different drug doses; and the use of factorial designs, are discussed. More patients should be allowed to participate in drug trials, since the treatments tested will potentially be offered to all patients. This will also expand the applicability of results to the general population. Measures of treatment outcome include survival, rate of disease progression, and changes in biologic markers, such as blood cell levels. Patients participating in a trial of an AIDS drug who develop symptoms such as opportunistic infections should be allowed to enter a trial testing drugs against such infections. Other changes that could make clinical trials more accessible to patients and physicians are proposed. (Consumer Summary produced by Reliance Medical Information, Inc.)