Ensuring Assay Sensitivity in Analysis of Noninferiority Trials with Missing Data

• Activity Number: 582
• Type: Topic Contributed
• Date/Time: Wednesday, August 7, 2013 : 2:00 PM to 3:50 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Brian Wiens*+
• Companies: Alcon Laboratories
• Keywords: per protocol ; intent to treat ; longitudinal data ; clinical trials

Abstract

We consider methods for ensuring assay sensitivity in active control, non-inferiority trials. Bias in assessment of comparative treatment effects can result in an increased or decreased chance of rejecting the null hypothesis compared to the nominal rate. A non-inferiority trial with biased estimation of treatment effects may have an increased rate of concluding no difference, even when the investigational treatment is importantly worse than the active control. Thus, controlling bias and ensuring assay sensitivity are similar goals. One popular, ad hoc method of ensuring assay sensitivity is use a per protocol analysis set. We consider advantages and disadvantages of the per protocol analysis set and conclude that the disadvantages are numerous and all advantages can be found in other, principled methods of controlling bias and assay sensitivity.