Can the Line Between Primary and Secondary Endpoints Be Blurred? A Method to Adaptively Test Multiple Endpoints with Emerging Data

• Activity Number: 645
• Type: Topic Contributed
• Date/Time: Thursday, August 4, 2011 : 10:30 AM to 12:20 PM
• Sponsor: International Chinese Statistical Association
• Author(s): Yang Song*+
• Companies: Merck & Co., Inc.
• Keywords: Type I error rate ; adaptive design ; adaptive testing ; multiple endpoints

Abstract

Tentative abstract: In a late-stage clinical trial, especially in a Phase III confirmatory study, there is usually a clear differentiation between the primary endpoint and the secondary endpoints. By default, the secondary endpoints will not be tested unless the primary endpoint is shown to be significant. Similarly, when there are multiple secondary endpoints, there is usually an ordering on those endpoints. Endpoints in lower order will not be tested until those in higher order have been shown to be significant. This is to maintain strong control of the Type I error rate by using a special case of closed testing, i.e. a step-down procedure. Pre-specifying an order for the endpoints can be risky to the study power if there is insufficient information on these endpoints. We propose a procedure that will adaptively prioritize the testing for multiple endpoints with emerging data so that power can be optimized while Type I error rate is strongly controlled.