ABSTRACT
Active control non-inferiority trials compare an investigational treatment to an active control with the intention of demonstrating that the investigational treatment is not much worse than the active control. Many issues have been identified with design, analysis and interpretation of non-inferiority trials, and we discuss some of these issues in this chapter. Selection of the non-inferiority margin, the definition of “not much worse,” is of primary importance, and must have agreement from all parties before the trial begins. Decision rules are typically made by comparing the bound of a confidence interval to the margin and concluding non-inferiority if the margin is outside of the appropriate bound. Study conduct must be rigorous to assure assay sensitivity, the ability of a trial to demonstrate a difference if the investigational treatment is not non-inferior to the active control. Other issues may not have universally agreed solutions, including choice of analysis set, handling of missing data, and planned adaptations to the study design. We discuss some potential solutions.
