On the uses of routine patient-reported health outcome data.

It has long been standard practice to ask patients in clinical trials about their health status, but the practice is now being extended to patients receiving treatment on a routine basis. In this paper, we examine four types of analyses that these health data might inform: comparisons of alternative treatments for the same condition, of health care providers, of changes in performance over time and of treatments of different types of condition. Analytical challenges arise because counterfactuals cannot be observed and because health status cannot be measured continuously. The implications of these challenges and the ability to meet them vary according to the comparative exercise. We argue that, provided with a sufficient number of health status measures for each patient and proper risk adjustment, health status measurement has great potential to inform the first three types of comparison. However, we believe that it is not yet possible to use such data to make secure comparative judgements about the outcomes from treatment for different types of condition.