General Multistage Gatekeeping and Graphical Multiple Testing Procedures in Clinical Trial Applications: A Case Study via Simulation

• Activity Number: 325
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Yihan Li* and Xin Wang and Deli Wang and Walt Offen
• Companies: AbbVie and AbbVie and AbbVie and AbbVie
• Keywords: Gatekeeping procedures ; Graphical multiple testing procedures ; Truncated Hochberg ; Multiple endpoints ; Multiplicity ; alpha redistribution

Abstract

Multiplicity adjustment plays a critical role in clinical trials. In the typical case of hierarchically ordered endpoints, strong control of the family-wise error rate (FWER) can be achieved via the general multistage gatekeeping approach as well as the graphical approach. Gatekeeping procedures ensure the testing of higher ordered hypotheses to be independent of the lower ordered ones; while the graphical approach allows more flexible redistribution of ?. Bonferroni based gatekeeping procedures can be represented by the graphical approach; however an equivalent graphical representation of general multistage gatekeeping using truncated Hochberg is not available. In this research, several Bonferroni based graphical procedures were constructed and compared with the multistage gatekeeping procedures using truncated Holm and Hochberg. We compared the power and examined the tradeoff between different approaches via a case study with two doses and three endpoints. For general multistage gatekeeping procedures, the optimal range of the truncation fraction was also explored.