Numerical Cincinnati Stroke Scale versus Stroke Severity Screening Tools for the Prehospital Determination of LVO

Abstract

Previous research demonstrated that the numerical Cincinnati Prehospital Stroke Scale (CPSS) identifies large vessel occlusion (LVO) at similar rates compared to a limited number of stroke severity screening tools. We aimed to compare numerical CPSS to additional stroke scales using a national EMS database. Methods: Using the ESO Data Collaborative, the largest EMS database with hospital linked data, we retrospectively analyzed prehospital patient records for the year 2022. Stroke and LVO diagnoses were determined by ICD-10 codes from linked hospital discharge and emergency department records. Prehospital CPSS was compared to the Cincinnati Stroke Triage Assessment Tool (C-STAT), the Field Assessment Stroke Triage for Emergency Destination (FAST-ED), and the Balance Eyes Face Arm Speech Time (BE-FAST). The optimal prediction cut-points for LVO screening were determined by intersecting the sensitivity and specificity curves for each scale. To compare the discriminative abilities of each scale among those diagnosed with LVO, we used the area under the receiver operating curve (AUROC). Results: We identified 17,442 prehospital records from 754 EMS agencies with ≥ 1 documented stroke scale of interest: 30.3% (n=5,278) had a hospital diagnosis of stroke, of which 71.6% (n=3,781) were ischemic; of those, 21.6% (n=817) were diagnosed with LVO. CPSS score ≥ 2 was found to be predictive of LVO with 76.9% sensitivity, 68.0% specificity, and AUROC 0.787 (95% CI 0.722-0.801). All other tools had similar predictive abilities, with sensitivity / specificity / AUROC of: C-STAT 62.5% / 76.5% / 0.727 (0.555-0.899); FAST-ED 61.4% / 76.1%/ 0.780 (0.725-0.836); BE-FAST 70.4% / 67.1% / 0.739 (0.697-0.788). Conclusion: The less complex CPSS exhibited comparable performance to three frequently employed LVO detection tools. EMS agency leadership, medical directors, stroke system directors, and other stroke leaders may consider the complexity of stroke severity instruments and challenges with ensuring accurate recall and consistent application when selecting which instrument to implement.