Examining outcome measures in a clinical study of stroke.

Abstract

We investigated the relation between outcome and sample size for six selected stroke outcome measures to assist investigators in selecting end points for stroke studies. Data from a clinical trial of 167 stroke patients assessed shortly after admission to the hospital and 5 weeks later provided information on clinical, motor, and functional outcomes measured using a neurologic status scale, a stroke severity scale, the Fugl-Meyer Scale, the Barthel Index, and the activities of daily living and cognition subscales of the Level of Rehabilitation Scale. Data were examined using Pearson correlation coefficients and power analyses. All measures were significantly correlated. There was also substantial congruency between the subscales of a measure and its total score. The measures had variable efficiencies; the Barthel Index was the most efficient and therefore required the fewest subjects to identify a significant effect. These data suggest that careful consideration must be given to the choice of stroke outcome measures in terms of their numbers, interrelationships, and statistical properties, as these factors have important implications for the design, analysis, and conduct of clinical stroke studies.