Abstract TMP63: Simplified SAVE Large Vessel Occlusion Prediction Scale Performs Similarly to More Complex Scales

Abstract

Introduction: Prehospital Large Vessel Occlusion (LVO) prediction scales report accuracies near 80% but vary in complexity and exam items tested. NIHSS subscores of Best Gaze, Motor Arm, Visual fields, and Best Language and/or Extinction are most predictive of ICA, M1, or M2 LVOs. Complex scoring and exam distinctions could limit the adoption of prehospital LVO scales. Hypothesis: We hypothesized that SAVE, a 4 item clinical scale with absent/present scoring and avoidance of nuanced exam distinctions, would predict LVOs as well as the 5 item and 9 point Field Assessment Stroke Triage for Emergency Destination (FAST-ED) scale. Methods: Retrospective analysis of the prospective Screening Technology and Outcomes Project in Stroke (STOPStroke) Study included 735 patients with suspected acute ischemic stroke or TIA evaluated in emergency departments with NIHSS scores and computed tomography angiography. Receiver operating characteristic analyses assessed the ability of the 4 point SAVE scale to predict the presence of ICA, M1, M2, or Basilar artery LVOs. The SAVE scale assigns 1 point for any NIHSS abnormalities in Speech (Best Language and/or Dysarthria), Motor Arm, Visual fields, or Eye deviation (Best Gaze). The results were compared to the published FAST-ED scale, which was derived from the same STOPStroke cohort and performs similarly to other published scales. Results: The prevalence of ICA, M1, M2, or Basilar artery LVOs was 31%. The SAVE scale area under the curve (AUC) was 0.79 and the FAST-ED scale AUC was reported as 0.81. SAVE ≥1 and ≥2 are equivalent to FAST-ED ≥1 and ≥2. SAVE ≥3 is equivalent to FAST-ED ≥4. SAVE ≥4 is equivalent to FAST-ED ≥6. Conclusions: The SAVE scale performed similarly to the FAST-ED scale at predicting ICA, M1, M2, or Basilar artery LVOs. The SAVE scale focuses on 4 items with present/absent scoring and avoids the nuances of aphasia and neglect to reduce the complexity of prehospital LVO prediction. Prospective prehospital validation is needed.