Learning to Live With Missing Quality-of-Life Data in Advanced-Stage Disease Trials

Abstract

In advanced-stage disease, palliation can be the most important objective. Quality-of-life (QoL) outcomes provide an intuitively reasonable way to monitor palliation from the patient’s perspective. However, patients with advanced-stage disease are often unable to complete a series of QoL assessments because of deteriorating health. This introduces a possible bias in a clinical trial because patients who are healthy enough to complete the assessments may not represent the target population. In this issue, Brown et al report QoL outcomes from a non–small-cell lung cancer trial comparing supportive care regimens with and without chemotherapy. The analysis of completers only was based on half the baseline sample, and the proportion of completers differed by treatment arms. Despite comprehensive descriptions of the missing data problems and sensitivity analyses designed to cope with them, considerable uncertainty remains. Uncertainty is inevitable with this degree of missing data, but a few key questions can both sharpen focus and suggest practical steps to limit the impact of missing data.