An Intermittent Missing Data Analysis Strategy for Clinical Trials with Death-Truncated Data
*Chenguang Wang, Division of Biostatistics and Bioinformatics,John Hopkins University 

Keywords: functional and survival outcome, non-parametric, oncology trial

For clinical studies such as late stage cancer trials, functional endpoints are often unobserved due to skipped or out-of-window visits, loss to follow-up, death, etc. It is well known that inappropriate missing data handling can produce biased treatment comparisons. In this paper, We propose a rank based hypothesis testing procedure that is based on a composite endpoint of the functional and survival outcome. We explore possible limitations of multiple imputation that is commonly used for missing data analysis. We then propose a novel non-parametric intermittent missing data imputation method and the corresponding sensitivity analysis strategy. An illustration is given by analyzing data from a recent non-small cell lung cancer clinical trial.