Monitoring the ACTIVE-W trial: Some issues in monitoring a noninferiority trial

Noninferiority comparisons of new to current therapies and the use of composite outcomes represent significant advances in the design of clinical trials. They increasingly characterize trials of new cardiovascular agents, posing new challenges to Data Safety Monitoring Boards (DSMB) and principal investigators. The ACTIVE-W study was a noninferiority comparison of the combination of clopidogrel and acetylsalicylic acid versus oral anticoagulant among patients with atrial fibrillation. When unexpectedly high rates of stroke and then of the composite outcome of stroke, non–central nervous system systemic embolism, myocardial infarction, and vascular death emerged, the DSMB modified its monitoring plan and conducted its first formal interim analysis much earlier than had been planned in the DSMB charter. The study was terminated when only 27% of the anticipated outcomes had occurred. This paper discusses issues of appropriate stopping guidelines for noninferiority trials and the early emergence of significant harm in relation to one component (stroke) of a composite outcome. Conditional power was not determined concurrently with the HRs during the monitoring of ACTIVE-W; however, the members of the DSMB now believe that such calculations should be considered as useful adjuncts to the calculation of HRs and could lead to earlier termination of noninferiority trials whose interim results suggest futility, without the need for convincing proof of harm.