Statistical Consideration in Interim Analysis Timing Optimization for Sample Size Re-Estimation

• Activity Number: 173 - Bayesian Methods Applied to Biometric Problems
• Type: Contributed
• Date/Time: Monday, July 29, 2019 : 10:30 AM to 12:20 PM
• Sponsor: ENAR
• Author(s): Yang Zhang*
• Companies: AtaraBio, Inc.

Abstract

Interim analysis for sample size re-estimation is commonly used in concern of lack of power of the study due to the optimistic use of a higher targeted ORR at the design stage. When the interim analysis indicates that the true ORR may be low, additional enrollment of patients into the study is necessary. One question of our interest in practice is when such an interim analysis should be performed. Early interim analysis with imprecise ORR estimate may result in high decision making error, while late analysis is usually less preferred considering the trial operation. We propose a method to determine the optimal timing for the interim analysis, using the expectation of additional enrollment as the target function, and we can search for the optimal value of $m$ such that the total enrollment is minimized (or converged to a stable phase) and the timing $m$ can be as early as possible, with the restriction of decision making error. We will also investigate the tuning parameters by simulation using the overall type I error and power. An phase two single arm study design will be shown as an example of application.