Analysis of multiple failure time data from an AIDS clinical trial

The primary endpoint of AIDS prophylaxis trials is the occurrence of opportunistic infections. While the treatments are not expected to have an effect on the underlying HIV disease, an effect of treatments on mortality cannot be ruled out. Therefore, the primary analysis of these trials must be based on a combined endpoint of infection and survival times. There are several methods available for analysis of multiple failure time data. However, there is no standard method for combining mortality and other failures in these analyses. This paper explores the analysis of multiple infections in the context of a study in which treatments may have an effect on mortality. The methods are applied to an AIDS clinical trial of prophylaxis for fungal infections.