Prediction of Outcomes Following Rehabilitation of Stroke Patients

Abstract

This article gives an overview of major predictors of outcomes following stroke rehabilitation. The emphasis is on functional outcomes, but mortality is also considered. The literature on predictors of outcome following stroke is now large. Optimal predictors of rehabilitation outcome differ from early acute predictors, although there is much overlap. Measures of severity of impairment associated with the initial stroke are robust predictors, but measures of functional limitations upon admission to rehabilitation in conjunction with chronicity improve the accuracy of prediction of functional outcomes. Basic features of recovery curves are emphasized. Accuracy of prediction has improved to the point that the better studies now predict the majority of the variance of functional outcomes. Accurate, clinically useful prediction requires use of multiple predictors, with different predictive sets for different outcome dimensions. Further improvement in the accuracy and robustness of prediction is needed. The emerging bottleneck is an inadequate means of applying an abundance of predictive findings to routine clinical practice.