Analysis and review of analytic approaches to dealing with post-randomized ineligible patients resulting from a change in post-randomization disease-state ascertainment

Abstract

In a series of three parts, this thesis explores the available approaches to analyze post-randomized ineligible patients. Part one: a comprehensive systematic review yielding 18 unique articles from the Medline and the Cochrane Methodology Register databases. Part two: the methods found from the review were applied to a dataset where 23% of the patients in the experimental treatment group were found to be post-randomized ineligible patients. The results of this exercise revealed that little can be done when these patients exist and the best approach to dealing with them is to carefully choose an approach and accurately report how these patients were analyzed. Part three: Influenza was explored to further illustrate the importance of accurate reporting and the broader impact of this analytic issue. In sum, no firm conclusion could be reached on one "best" approach to take for randomized ineligible patients, however prevention and accurate reporting were surrogate recommendations.