Statistical Inference Following an Adaptive Design for Case-Driven Efficacy Study

• Activity Number: 513
• Type: Contributed
• Date/Time: Wednesday, August 4, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Xiaoming Li*+ and Ivan S.F. Chan and Keaven M. Anderson
• Companies: Merck Research Laboratories and Merck Research Laboratories and Merck Research Laboratories
• Keywords: Adaptive ; Inference ; Vaccine ; Efficacy

Abstract

In vaccine studies the efficacy endpoint usually is a (rare) single infection/disease event, the exact method conditional on the total number of events is commonly used for testing the hypothesis that study vaccine lowers the risk of the event, and the treatment effect is usually quantified by vaccine efficacy (VE = 1-vaccine risk/placebo risk). A two-stage adaptive design strategy built on the traditional fixed-number-of-events design was proposed for the scenarios in which the incidence rate may not be known. While the 'naïve' point estimate (MLE) of the vaccine efficacy is usually used, it is well known that it may be biased due to interim analysis/adaptation. The performance of the MLE following the proposed adaptive design will be evaluated. It will also be discussed as how to obtain valid confidence interval for VE following the potential adaptation in this design.