Validation of the National Institutes of Health stroke scale-8 to detect large vessel occlusion in ischemic stroke

Background Patients with acute ischemic stroke and large vessel occlusion (LVO) may benefit from prehospital identification and transfer to a center offering endovascular therapy. Aims We aimed to assess the accuracy of an existing 8-item stroke scale (National Institutes of Health Stroke Scale-8 [NIHSS-8]) for identification of patients with acute stroke with LVO. Methods We retrospectively calculated NIHSS-8 scores in a population of consecutive patients with presumed acute stroke assessed by emergency medical services (EMS). LVO was identified on admission computed tomography angiography. Accuracy to identify LVO was calculated using receiver operating characteristics analysis. We used weighted Cohen's kappa statistics to assess inter-rater reliability for the NIHSS-8 score between the EMS and the hospital stroke team on a prospectively evaluated subgroup. Results Of the 551 included patients, 381 had a confirmed ischemic stroke and 136 patients had an LVO. NIHSS scores were significantly higher in patients with LVO (median 18; interquartile range 14-22). The NIHSS-8 score reliably predicted the presence of LVO (area under the receiver operating characteristic curve.82). The optimum NIHSS-8 cutoff of 8 or more had a sensitivity of.81, specificity of.75, and Youden index of.56 for prediction of LVO. The EMS and the stroke team reached substantial agreement (κ =.69). Conclusions Accuracy of the NIHSS-8 to identify LVO in a population of patients with suspected acute stroke is comparable to existing prehospital stroke scales. The scale can be performed by EMS with reasonable reliability. Further validation in the field is needed to assess accuracy of the scale to identify patients with LVO eligible for endovascular treatment in a prehospital setting.