A Dynamic Alpha Spending (DAS) Function with Informative B-Value for a Stratified Study Design

• Activity Number: 371
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Das Purkayastha* and Jagannath Ghosh
• Companies: Novartis and Novartis Pharmaceuticals
• Keywords: B- value ; Dynamic alpha spending function ; Group sequential design ; Brownian motion ; Stratified study design

Abstract

In conventional group sequential designs both conservative and relaxed alpha spending functions are used with sequential monitoring of clinical data. This is an effort to adjust alpha over time with an idea developed through a stepwise approach within each interim look capturing statistical information in an efficient scheme under a stratified study design. This alternative approach considers observed and expected efficacy as realized through a process over time. Thus, informative B-values over time will allow to extrapolate efficacy of the study-end with a proposed stopping rule within each stratum and the whole study, with an ongoing dynamic adjustment of type-I error. In order to evaluate the comparative utility of this proposed alternative approach with standard O'Brien and Fleming and Pocock methods, a simulation will be performed with a hypothetical stratified clinical trial design.