A critical review of stroke trial analytical methodology: Outcome measures, study design, and correction for imbalances

Abstract

Despite considerable advances in understanding the pathophysiology of stroke, there has been a lack of success in identifying new therapies to improve outcome. Our work suggests that the execution of stroke trials is not the primary issue. Here, we consider the analysis of clinical trials as a potential source of error. We review several components of stroke trial analysis. We conclude that many of these trials have been plagued by inappropriate use of complex statistical analytical methods that have not considered the underlying assumptions required for their valid application. Unfortunately, many of these methods have been encouraged by publishing, regulatory, granting, and pharmaceutical entities, yet continue to generate flawed results, usually discovered when early results are not confirmed in subsequent large trials. Because these errors may be just as likely to occur when early studies appear negative and so potentially reflect a missed opportunity to identify an effective therapy, we urge a reassessment of these analytical principles and provide some alternative approaches.