Dr. Nigel Paneth - Combined Outcomes in Prevention Trials: Rarely A Good Idea

In recent years, many randomized prevention trials have adopted combined outcomes as the primary trial outcome. Combined outcomes reflect more than one medical condition or diagnostic entity, either alternative outcomes (e.g. stroke and/or heart attack) or competing outcomes (e.g. death and/or disability).

Recent advances in neonatal intensive care have profoundly altered neonatal mortality, but increased the net prevalence of major disabilities such as cerebral palsy and severe cognitive impairment. As a result, death and disability has become the gold standard outcome in neonatal prevention trials, based on the argument that the ideal outcome of any neonatal intervention is disability-free survival.

However, a systematic review of recent newborn trials with death and disability as the primary outcome reveals serious problems. In half of all trials, the ORs for death and disability are in the opposite direction. Not only is power reduced, but a significant finding in one outcome can be obliterated by a minor change in the other, commoner outcome. The vexed question of how to value death as compared to disability is rendered moot in trials that count each as of equal value. Finally, recent problems in the ethics of clinical trials, such as the SUPPORT trial controversy, in which the investigators were censured by the federal government, and then defended in press by NIH leadership, are a direct result of the decision by the investigators to inappropriately combine death and disability in the trial.

The use of combined outcomes in prevention trials need to be reconsidered.