Patients with large vessel occlusion strokes (LVOS) may be better served by direct transfer to endovascular capable centers avoiding hazardous delays between primary and comprehensive stroke centers. However, accurate stroke field triage remains challenging. We aimed to develop a simple field scale to identify LVOS.
The FAST-ED scale was based on items of the NIHSS with higher predictive value for LVOS and tested in the STOPStroke cohort, in which patients underwent CT angiography within the first 24 hours of stroke onset. LVOS were defined by total occlusions involving the intracranial-ICA, MCA-M1, MCA-2, or basilar arteries. Patients with partial, bi-hemispheric, and/or anterior + posterior circulation occlusions were excluded. Receiver operating characteristic (ROC) curve, sensitivity, specificity, positive (PPV) and negative predictive values (NPV) of FAST-ED were compared with the NIHSS, Rapid Arterial oCclusion Evaluation (RACE) scale and Cincinnati Prehospital Stroke Severity Scale (CPSSS).
LVO was detected in 240 of the 727 qualifying patients (33%). FAST-ED had comparable accuracy to predict LVO to the NIHSS and higher accuracy than RACE and CPSS (area under the ROC curve: FAST-ED=0.81 as reference; NIHSS=0.80, p=0.28; RACE=0.77, p=0.02; and CPSS=0.75, p=0.002). A FAST-ED ≥4 had sensitivity of 0.60, specificity 0.89, PPV 0.72, and NPV 0.82 versus RACE ≥5 of 0.55, 0.87, 0.68, 0.79 and CPSS ≥2 of 0.56, 0.85, 0.65, 0.78, respectively.