Abstract:
|
In randomized-controlled trials, interim analyses are often planned for possible early trial termination to claim superiority or futility of a new therapy. While unblinding is necessary to conduct the formal interim analysis in blinded studies, blinded data also have information about the potential treatment difference between the groups. We develop a blinded data monitoring tool for binary outcomes, utilizing the number of subjects and a pooled response rate from blinded data before conducting the formal, unblinded, interim analysis. The tool helps investigators determine if an interim analysis is necessary. We conduct extensive simulation studies to assess the impact of the implementation of our tool. The numerical studies show that when the effect of the treatment difference is small, the monitoring tool dramatically reduces the expected number of interim analyses. The tool can be used to determine whether or not the formal unblinded interim analysis should be conducted, without losing integrity of the study, thereby potentially saving study resources.
|