In randomised controlled trials (RCTs), patients can experience low tolerance to the treatment or adverse events and drop out of the study after discontinuing the treatment and therefore result in failure to measure the study outcomes. Consequently, patients with measured data may no longer be representative of the initial study population and this post-randomisation selection bias can partially invalidate randomisation. Ignoring or inadequately handling missing data can lead to estimation biases and loss of statistical power. Although a substantial literature on missing data has been developed, there remains variability in practice on strategies for handling missing data in clinical research and the lack of consensus on which approach was desired under different primary statistical analyses and missing data mechanisms.

The objectives of this study are to compare common approaches to the analysis of missing data in RCT via a series of statistical simulation studies and to provide recommendations on how to handle missing data in RCT. In our simulation study, we will consider continuous, binary, and repeatedly measured study outcomes, and compare missing data methods including complete-case analysis, imputation, and methods that do not require a complete data set. The DSI-SUDS trainee working on this project will gain knowledge and expertise in missing data, the design and analysis of standard RCT, and conducting statistical simulation study in R statistical software.