Predicting outcomes of acute stroke patients initially presenting with mild neurologic deficits is crucial in decision making regarding thrombolytic therapy. We examined the utility of individual National Institutes of Health Stroke Scale (NIHSS) score items or clusters of items as predictors of functional outcomes at 3 months after mild stroke. Using a multicenter stroke registry database, we identified patients with acute ischemic stroke who presented within 4.5 hours of symptom onset and had baseline NIHSS scores less than or equal to 5. Functional outcomes at 3 months were dichotomized as favorable (modified RankinScale [mRS] score 0 or 1) or unfavorable (mRS 2-6). Individual NIHSS items, clusters of items, and the total score were tested for their ability to predict outcomes in multivariable models. Area under the receiver operating characteristic curve (AUC) was used to assess model performance. Of the 2209 patients who met eligibility criteria, 588 (26.6%) exhibited unfavorable functional outcomes at 3 months. Of the 15 items of the NIHSS, all except item 8 (sensory) and item 11 (extinction) were significantly associated with unfavorable functional outcomes in bivariate analysis (P's < .05). Among the multivariable models, the model with the total NIHSS score exhibited an AUC similar to that of the model with all NIHSS items in predicting functional outcomes (.758 [95% confidence interval .739-.775] versus .759 [.740-.776]; P = .75 for pairwise comparison). Simply using the total NIHSS score was as effective as using all individual items in predicting outcomes of mild stroke patients.
Read full abstract