Controlling Family Wise Error for Multiple Endpoints within a Group Sequential Trial with Sample Size Re-estimation
View Presentation View Presentation
*Cyrus R. Mehta, President, Cytel Inc. 

Keywords: Secondary endpoints, adaptive design

The fixed sequence procedure of testing a secondary hypothesis at level alpha only if the primary hypothesis is first rejected at level alpha has been shown by Hung, Wang and O’Neill (J. Biopharm Statist., 2007) to inflate the family wise error rate if a group sequential or adaptive design is adopted. We present a general and flexible procedure for guaranteeing control of type-1 error in this setting. The method is applied to two-stage trials with early stopping boundaries and sample size re-estimation based on interim results.