Survival versus years of healthy life: which is more powerful as a study outcome?
Studies of interventions that are intended to improve patients' health are often evaluated with survival as the primary outcome, even when a measure adjusted for quality of survival, such as years of healthy life (YHL), would seem more appropriate. The purpose of this article is to determine whether studies based on survival are more or less powerful than studies based on YHL in clinical trials where either measure might be appropriate. We used data from the Cardiovascular Health Study (CHS) to estimate the sample size that would be needed in studies of 156 different health conditions, for the two outcome measures. The median sample size for a 5-year study was 687 if survival was the endpoint and 484 for YHL. YHL usually required lower sample sizes than survival, although survival was more powerful for some health conditions. We also found that lengthy studies, and studies with many follow-up measures per person, did not have appreciably higher power than less intensive studies. We conclude that clinical investigations in which the goal is to improve health may often be performed more efficiently with YHL rather than survival as the primary outcome measure. Such studies can be short in duration, with relatively few measures per person of health status.