TL27: Bayesian Missing Data Analysis – Methods and Case Studies
*Frank Liu, Merck & Co. Inc. 


Missing data can be inevitable in longitudinal clinical trials. While there is no “best” method for handling missing data, it is recommended by regulatory agencies to conduct sensitivity analysis to check the robustness of the analysis results. Considering the advantages of Bayesian approach on incorporating the uncertainty of missing data into the sensitivity analysis models, DIA Bayesian Scientific Working Group (BSWG) established a team to look into this area. In this roundtable discussion, we will share some findings from the working group. Specifically, some case studies using real clinical trial datasets from pharmaceutical companies will be used to illustrate the applications, and discuss the opportunities on using Bayesian sensitivity analysis for missing data in clinical trials.