ABSTRACT
BACKGROUND AND PURPOSE: Clinical deficits from stroke are diverse, prompting measurement in trials by a range of outcome scales. Statistical and clinical advantage can be gained by combining scales into a global outcome provided combinations are chosen with limited correlations. We aimed to clarify the interdependence of outcome scales by systematic review of published data and by novel analysis of data from completed acute trials. SUMMARY OF REVIEW: We systematically searched ScienceDirect and PubMed to summarize published data on correlations between stroke outcome scales. We generated new data on correlations among salient scales at 90 days poststroke in patients from the Virtual International Stroke Trials Archive (VISTA). We calculated Pearson and Spearman-Rank correlation coefficients for continuous and ordinal measures, respectively. We also assessed partial correlations, adjusted for baseline National Institute of Health Stroke Scale (NIHSS), and age. Published estimates of interdependence were limited to small single-trial cohorts and gave divergent results. From the more extensive VISTA dataset, we found that the modified Rankin Scale at 90 days poststroke explained 80.8% of the National Institute of Health Stroke Scale at 90 days poststroke and 86.5% of the European Stroke Scale. National Institute of Health Stroke Scale explained 75.9% of the Barthel Index and 81.2% of the Scandinavian Stroke Scale. After adjustment, modified Rankin Scale explained 56.6% of National Institute of Health Stroke Scale, 75.2% of Barthel Index. National Institute of Health Stroke Scale explained 60.2% of Barthel Index. CONCLUSION: Correlations and partial correlations among stroke outcome scales in trial datasets are higher than previously reported. The new estimates are more reliable for trial planning due to the sample size and diversity.
