Background: Composite end points are common in clinical trials.
Objective: To describe how composite outcomes are used, constructed, and reported in cardiovascular trials and to evaluate the contribution of individual end points to the composite estimate of effect in those trials.
Design: Review of 2-group, parallel-design, randomized cardiovascular trials that used composite end points and were published in 14 clinical journals from 1 January 2000 to 1 January 2007.
Setting: Published randomized trials in cardiovascular medicine and surgery.
Participants: Two-group, parallel-design trials published in 14 leading general medical, cardiology, and cardiothoracic surgery journals from 1 January 2000 to 1 January 2007.
Measurements: The types and numbers of individual events included in the composite outcome and P values and risk estimates for the composite outcome.
Results: Of 304 trials published that used composite outcomes, 221 (73%) reported a composite primary outcome and 83 (27%) reported a composite secondary outcome. Composite outcomes comprised a median of 3 (interquartile range, 3 to 4) individual outcomes; death was the most common individual outcome. The total number of individual events and the total number of events represented by the composite outcome differed in 79% of trials. P values for composite outcomes were less than 0.050 more frequently than they were 0.050 or greater. Death as an individual end point made a relatively minimal contribution to estimates of effect summarized by the trials' composite end points, whereas revascularization made a greater contribution.
Limitation: All-cause and cardiovascular mortality were not distinguished, and the findings might not apply to trials in other fields.
Conclusion: Composite outcomes in cardiovascular trials are frequent and commonly comprise 3 to 4 individual end points that vary in clinical significance. Discrepancies between the total number of individual events in a trial and those reported for composite outcomes are common. Individual outcomes do not contribute equally to composite measures, so the overall estimate of effect for a composite measure cannot be assumed to apply equally to each of its individual outcomes.