Assessment of Clinical Scales for Detection of Large Vessel Occlusion in Ischemic Stroke Patients from the Dijon Stroke Registry

(1) Background: The limited availability of thrombectomy-capable stroke centres raises questions about pre-hospital triage of patients with suspected stroke (IS) due to large vessel occlusion (LVO). Aims: This study aimed to evaluate the diagnostic accuracy of clinical stroke severity scales available for LVO detection. (2) Methods: Patients with IS were prospectively identified among residents of Dijon, France, using a population-based registry (2013-2017). Clinical signs and arterial imaging data were collected. LVO was defined as an occlusion site affecting the terminal intracranial internal carotid artery, the M1 segment of the middle cerebral artery (MCA), or the basilar artery (restricted definition). A wide definition of LVO also included the M2 segment of the MCA. For each of the 16 evaluated scales, a receiver operator characteristic (ROC) analysis was performed, and the c-statistic representing the area under the ROC curve was evaluated to assess discrimination for predicting LVO. (3) Results: 971 patients were registered, including 123 patients (12.7%) with an LVO according to the restricted definition. The c-statistic for LVO detection ranged between 0.66 and 0.80 according to the different scales, with a sensibility varying from 70% to 98% and a specificity from 33% to 86%. According to the wide definition of LVO (174 patients, 17.9%), the c-statistic was slightly lower, ranging between 0.64 and 0.79. The sensitivity was 59% to 93%, and the specificity was 34% to 89%. (4) Conclusion: The clinical scales failed to combine a high sensitivity and a high specificity to detect LVO. Further studies are needed to determine the best strategy for pre-hospital triage of IS patients.