Mid-Course Sample Size Adjustment Using Combination Tests and Sequential Inference: A Comparative Study

• Activity Number: 458
• Type: Contributed
• Date/Time: Wednesday, August 6, 2014 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Shanhong Guan*+
• Keywords: Combination test ; Conditional power ; Sample size ; Sequential probability ratio test ; Type I error

Abstract

Determination of sample size in clinical trials rests upon knowledge of the expected treatment effect size, which is a function of the expected treatment difference and the nuisance parameter of an outcome variable. In this research project, we consider reassessment of the sample size when expected treatment difference is misspecified before the trial. Under the restriction of an upper bound on the allowable sample size in practical situations, we show that considerable power loss can be observed using two-stage sample size re-estimation based on conditional power. This limitation can be alleviated or overcome by choosing alternative approaches that are based on combination tests or multi-stage sequential tests, while maintaining the overall type I error rate. Numerical studies are conducted to compare the operating characteristics of these approaches. An example is used to illustrate the application and for the comparison of these mid-course sample size adjustment methods. General consideration of these procedures is discussed.