Improving outcome assessment for clinical trials in stroke

Abstract Clinical trials are at the centre of advances in our understanding of stroke and its optimal treatment. In this thesis the uses and properties of outcome assessment scales for stroke trials are described, with particular attention given to the modified Rankin Scale (mRS). Through comprehensive literature review I will show that mRS is the most frequently used functional outcome scale in clinical trials but efficacy of the scale is potentially limited by inter-observer variability. Using a “mock” clinical trial design I demonstrate that inter-observer mRS variability in contemporary practice is moderate (k=0.57). Adding these data to systematic review of published data, confirms an overall moderate inter-observer variability across ten trials (k=0.46). Differing strategies to improve mRS reliability will then be described. I will outline development of a bespoke training package, international training scores across 2942 raters again confirms suboptimal reliability (k=0.67). A pilot trial using endpoint committee review of video recorded interviews demonstrates feasibility of this approach. Attempts to improve reliability by deriving mRS from data recorded in patients’ hospital records are not successful (k=0.34). In the final chapters I present a novel methodology for describing stroke outcomes – “home-time”. This measure shows good agreement with mRS, except at extremes of disability. Finally to put mRS in a historical context, the career of John Rankin and the development of his eponymous scale is recounted