In 2014, the US FDA approved liraglutide for weight management. The statistical review of the application presented various challenges related to the handling of missing data. The ability of the drug to cause weight loss was not in question. The challenge centered on obtaining a reliable estimate of the intention-to-treat effect to support the risk-benefit evaluation. Subjects in the trials that stopped treatment prior to the endpoint were encouraged to attend the primary endpoint visit. Data from the subjects that returned for a primary efficacy assessment played a significant role in the statistical review. They were used to illustrate shortcomings of the applicant's primary efficacy analysis and sensitivity analyses. They were also used in the FDA analyses to address missing data. The goal of this article is to illustrate challenges and considerations associated with the handling of missing data in clinical trials.
Keywords: Estimands; missing data; sensitivity analyses.