Health Outcomes Should be Viewed as an Area under the Curve

Abstract

Time is an essential component of any outcome worth measuring in the clinical evaluation of drugs. Unfortunately, temporal information is often suppressed by the way clinical endpoints are measured and reported. This is particularly a problem when point-in-time comparisons are made using physiological test results or with most of the available quality-of-life instruments. One consequence of ignoring the area under the outcome curve (AUC) is unnecessary measurement error. Another is that point-in-time comparisons may not provide useful information on the magnitude of health effects. This is important because clinical policy makers are increasingly concerned with questions of value for money. They are asking not whether a new therapy is better, but by how much is it better. The goal of therapy is not to make patients better at a point in time, but over time. Clinical trial design should therefore emphasize measures that can be integrated over the temporal duration of interest. Application of AUC to describe health outcomes will, however, require attention to the following issues: (1) the cyclicity of biological (and sociological) parameters, (2) the relationship between test results and outcomes of more direct concern to patients, (3) health effects that extend beyond the study period, and last but not least, (4) the relationship between quality of life and survival.