Session 2: Bayesian Non-Inferiority Trials
Organizer: Guochen Song (Quintiles)
Chair:
Guochen Song (Quintiles) “Controlling Frequentist Type I and Type II Error in Bayesian Non-inferiority Trials: a Case Study”
In phase III biosimilar studies, utilizing Bayesian method and borrowing information from historical data for the control arm can effectively reduce the sample size. From the Frequntist point of view, however, the type I error from such studies can be inflated if not properly controlled. This case study demonstrates how to control the type I and type II error together in a setting where the endpoint variable is binary and the conjugate beta prior is assumed.
Fanni Natanegara (Eli Lilly) “Bayesian considerations for non-inferiority clinical trials with case examples”
The gold standard for evaluating treatment efficacy of a pharmaceutical product is a placebo controlled study. However, when a placebo controlled study is considered to be unethical or impractical to conduct, a viable alternative is a non-inferiority (NI) study in which an experimental treatment is compared to an active control treatment. The objective of such study is to determine whether the experimental treatment is not inferior to the active control by a pre-specified NI margin. The availability of historical studies in designing and analyzing NI study makes these types of studies conducive to the use of the Bayesian approach. In this presentation, we will highlight case examples for utilizing Bayesian methods in NI study and provide recommendations.
Sujit Ghosh (NC State University and SAMSI) “Robust Bayesian Methods for Non-Inferiority Tests Based on Dichotomous Data”
In a non-inferiority trial, the experimental treatment is compared against an active control instead of placebo. The goal of this study is often to show that the experimental treatment is non-inferior to the control by some pre-specified margin. The standard approach for these problems, which relies on asymptotic normality, usually requires large sample size to achieve some desired power level. This talk presents robust Bayesian approaches based on Bayes factor and posterior probability for testing non-inferiority in the context of two-sample dichotomous data. A novel aspect of the proposed Bayesian methods is that the cut-off value for Bayes factors and posterior probabilities are determined from the data that approximately controls the overall errors. Results based on simulated data indicate that both of the proposed Bayesian approaches provide significant improvement in terms of statistical power as well as the total error rate over the popularly used frequentist procedures. This in turn indicates that the required sample size to achieve certain power level could be substantially lowered by using the proposed Bayesian approaches. [This is a joint work with Muhtar Osman and major part of the talk is based on two published papers: 1 and 2]
