Abstract DP089: Using the ROSIER Scale to Decrease Acute Stroke Activations in a Municipal Hospital Emergency Department

Background: Appropriately identifying strokes among patients presenting acutely with neurologic symptoms is a challenge in the emergency department (ED). Acute stroke activations (ASA), while essential for ensuring prompt care, are also resource intensive. As treatment timelines and eligibility expand, the volume of false-positive ASA (stroke mimics) have increased. Because rates of stroke mimics are higher among patients presenting to the ED as walk-ins rather than via Emergency Medical Services (EMS), we sought to determine whether using the ROSIER scale (RS), a validated tool for stroke triage screening, could reduce false-positive ASA for walk-ins. The RS is a 7-item stroke recognition tool, which incorporates clinical history and neurological signs, and ranges from -2 to +5 with scores ≥1 suggestive of stroke. Methods: We retrospectively reviewed consecutive ASA patients presenting to the ED within 6h from last known well (LKW) from 1/1/2024-6/30/2025. Data was abstracted from the extant medical record with an RS score determined based on triage documentation for all walk-ins, and final diagnosis (stroke/TIA/ICH versus stroke mimic) determined by expert consensus for all ASA blinded. An RS≥1 was considered a positive screen in accordance with prior literature. Descriptive statistics were used to describe our findings and standard tests of comparison to compare groups with alpha set at <0.05. Sensitivity and specificity of the RS are reported. Results: There were 500 ED ASA <6h from LKW over the study period; 279 (56%) presented via EMS and 221 (44%) via walk-in. The proportion of patients with a final diagnosis of stroke/ICH/TIA versus mimic (false-positive) was lower among walk-ins (96 [34.4%] vs. 19 [8.6%], p<0.001). Among the 221 walk-ins, 83 (37.6%) had RS<1 and 138 (62.4%) had RS≥1. The final diagnosis of stroke (stroke/TIA/ICH) occurred in 4 (4.8%) patients with RS<1 and 15 (10.9%) with RS > 1 (p=0.12); all walk-ins who got thrombolysis/thrombectomy (n=6) or had ICH (n=4) had RS≥1. The sensitivity of the RS score for stroke detection was 79% [95%CI 56%-92%] and specificity was 39% [95%CI 33%-46%]. Conclusion: Using the RS for triaging ED walk-ins with acute neurologic symptoms could lead to a decrease in the number of false-positive ED stroke codes without jeopardizing recognition of patients eligible for acute stroke intervention. Further hybrid implementation-effectiveness study of this triage tool in a subset of ED patients may be warranted.