Send to

Choose Destination
Crit Care Med. 2016 Jun;44(6):e336-43. doi: 10.1097/CCM.0000000000001630.

Prediction of Large Vessel Occlusions in Acute Stroke: National Institute of Health Stroke Scale Is Hard to Beat.

Author information

1Department of Neurology, Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland. 2Department of Neurology, University Hospital Antwerp, Antwerp, Belgium. 3Department of Neurology, University Hospital Bern, Berne, Switzerland. 4Institute of Social and Preventive Medicine, Faculty of Medicine, University of Lausanne, Lausanne, Switzerland. 5Department of Medicine, University of Thessaly, Larissa, Greece. 6Department of Radiology, University Hospital Bern, Berne, Switzerland.



Endovascular treatment for acute ischemic stroke with a large vessel occlusion was recently shown to be effective. We aimed to develop a score capable of predicting large vessel occlusion eligible for endovascular treatment in the early hospital management.


Retrospective, cohort study.


Two tertiary, Swiss stroke centers.


Consecutive acute ischemic stroke patients (1,645 patients; Acute STroke Registry and Analysis of Lausanne registry), who had CT angiography within 6 and 12 hours of symptom onset, were categorized according to the occlusion site. Demographic and clinical information was used in logistic regression analysis to derive predictors of large vessel occlusion (defined as intracranial carotid, basilar, and M1 segment of middle cerebral artery occlusions). Based on logistic regression coefficients, an integer score was created and validated internally and externally (848 patients; Bernese Stroke Registry).




Large vessel occlusions were present in 316 patients (21%) in the derivation and 566 (28%) in the external validation cohort. Five predictors added significantly to the score: National Institute of Health Stroke Scale at admission, hemineglect, female sex, atrial fibrillation, and no history of stroke and prestroke handicap (modified Rankin Scale score, < 2). Diagnostic accuracy in internal and external validation cohorts was excellent (area under the receiver operating characteristic curve, 0.84 both). The score performed slightly better than National Institute of Health Stroke Scale alone regarding prediction error (Wilcoxon signed rank test, p < 0.001) and regarding discriminatory power in derivation and pooled cohorts (area under the receiver operating characteristic curve, 0.81 vs 0.80; DeLong test, p = 0.02).


Our score accurately predicts the presence of emergent large vessel occlusions, which are eligible for endovascular treatment. However, incorporation of additional demographic and historical information available on hospital arrival provides minimal incremental predictive value compared with the National Institute of Health Stroke Scale alone.

[Indexed for MEDLINE]
Free full text

Supplemental Content

Full text links

Icon for Wolters Kluwer Icon for Bern Open Repository and Information System
Loading ...
Support Center