Handling Dropouts in a Multiple-Attacks Migraine Study

Abstract

The clinical development of a novel migraine compound often requires the conduct of a multiple-attack clinical trial. The purpose of such a trial is to assess the consistency of response, where a positive consistency response may be defined as the substantial reduction or complete elimination of headache pain in at least three out of four treated migraine attacks. This study design is prone to dropout. A variety of well-known approaches to handling missing (dropout) data exists, including data imputation methods as well as a “completers-only” approach. This manuscript introduces a novel approach to handling dropout data arising from this study design, an approach in which only those patients whose data is known to yield either a positive consistency response (at least three successes) or a negative consistency response (at least two failures) are included in the analyses. This novel approach is defined as the modified full-analysis-set (MFAS) approach and will be compared to other commonly used methods for handling dropout data through the assessment of bias and mean square error.

Keywords: :

• MFAS
• Missing data
• Modified full-analysis set