Determination of the underlying cause of death in three multicenter international HIV clinical trials

Purpose: Describe processes and challenges for an Endpoint Review Committee (ERC) in determining and adjudicating underlying causes of death in HIV clinical trials. Method: Three randomized HIV trials (two evaluating interleukin-2 and one treatment interruption) enrolled 11,593 persons from 36 countries during 1999-2008. Three ERC members independently reviewed each death report and supporting source documentation to assign underlying cause of death; differences of opinion were adjudicated. Results: Of 453 deaths reported through January 14, 2008, underlying causes were as follows: 10% AIDS-defining diseases, 21% non-AIDS malignancies, 9% cardiac diseases, 9% liver disease, 8% non-AIDS-defining infections, 5% suicides, 5% other traumatic events/accidents, 4% drug overdoses/acute intoxications, 11% other causes, and 18% unknown. Major reasons for unknown classification were inadequate clinical information or supporting documentation to determine cause of death. Half (51 %) of deaths reviewed by the ERC required follow-up adjudication; consensus was eventually always reached. Conclusion: ERCs can successfully provide blinded, independent, and systematic determinations of underlying cause of death in HIV clinical trials. Committees should include those familiar with AIDS and non-AIDS-defining diseases and have processes for adjudicating differences of opinion. Training for local investigators and procedure manuals should emphasize obtaining maximum possible documentation and follow-up information on all trial deaths.