ABSTRACT
Missing data is common in clinical trials but is often inadequately handled which could have an impact on estimation and testing of treatment effects. Based on the National Research Council (NRC) recommendations, regulatory agencies are requesting more thorough analyses on missing data from sponsors in recent years. With advancement of flexible general purpose Bayesian software packages such as WinBUGS, SAS Proc MCMC, and Stan, it is relatively simple to develop Bayesian methods or multiple imputation approaches to address complex missing data problems while incorporating the uncertainty. In this chapter, we present several case studies to demonstrate how to utilize Bayesian, multiple imputation, and some unconventional approaches for missing data analysis in clinical trials. The case studies are selected to represent different types of endpoints in various therapeutic settings including analysis of daily diary data in an insomnia study, analysis of a schizophrenia study with a continuous endpoint, and analysis of functional scale data in a pediatric spinal muscular atrophy study with missing outcomes in presence of death.
