Long-term follow-up in AIDS clinical trials.

There are several different strategies for the follow-up of patients on AIDS clinical trials. In some trials, data are only collected until patients stop protocol therapy for any reason. Alternatively, patients are followed until they have disease progression or the study ends. It has been suggested in some trials that patients have data collected after they have stopped protocol therapy either because they reached a study endpoint or all patients on the trial have been switched to the superior therapy. The effect of these strategies on the conclusions that can be drawn from a trial will be described using simulations. To calculate the operating characteristics of a strategy that involves a primary and a secondary endpoint, it is necessary to formalize the notion that the results of a secondary analysis support or contradict the primary analysis.