Abstract 5604: Refining Clinical Trial Composite Outcomes With Investigator-Assessed Weights: An Application to the ASSENT-3 Trial
Background: Declining cardiac mortality and rising costs of clinical trials places a new priority on efficient use of all patient outcomes. Whereas traditional time-to-event analysis exclusively assigns equal weight to the first event in the composite end point, this is counterintuitive to stakeholders.
Methods: We surveyed clinical investigators and asked them to assess the relative severity weights they would place on death, recurrent myocardial infarction (re-MI), cardiogenic shock and congestive heart failure (CHF) in ST-elevation myocardial infarction (STEMI) patients. Their assessment was then applied to a modified survival analysis of selected 30 day endpoints in the ASSENT-3 trial which randomized 6095 STEMI patients and compared three reperfusion strategies:
tenecteplase (TNK) + unfractionated heparin (UH);
TNK + enoxaparin (enox);
TNK + abciximab (abx).
Results: Respondents (n=23) allocated relative values from a standard point scale which was converted to a single denominator (2.0) resulting in the following weights: death 1.0, shock 0.5, CHF 0.3 & re-MI 0.2. The Table⇓ provides the event rates calculated according to the first event as well as at any time during 30-day follow-up (Panel A). In Panel B traditional first event versus weighted composites using first event and all events in each patient are shown. Note that when the traditional time-to-event analysis is re-examined with a weighted composite there is an attenuation of the excess events initially observed with UH (p=0.05 vs p=0.18). By contrast, when all events from each patient are used in the weighted composite, a trend for benefit with enox was observed (p=0.15).
Conclusions: This approach adds potential value to traditional analytical techniques by more efficiently incorporating the differential value of all events in each patient thereby providing new options for future trial design and analysis.