Abstract WP242: Predictors of Endovascular Thrombectomy Utilization in Clinical Practice: Analysis of Linked Florida Stroke Registry and RAPID AI Data

Introduction: Endovascular thrombectomy (EVT) is an effective treatment for acute ischemic stroke from large vessel occlusion, supported by strong evidence across imaging profiles and time windows. Understanding how routine care adapts to evolving guidelines, emerging trial data, and expanded imaging access is essential. This study leverages a novel linkage between the Florida Stroke Registry (FSR) and RAPID AI to evaluate clinical, imaging, and system-level factors influencing EVT use in real-world practice. Methods: From January 2023 to December 2024, data from 31 hospitals in the legislatively mandated FSR (using Get With the Guidelines–Stroke data) were linked with RAPID AI imaging (n=14,804). Included were acute ischemic stroke cases with large vessel occlusion by CT-angiography (CTA) or perfusion mismatch on CT-perfusion (CTP). Core volume was defined by ASPECTS on CT (large ≤5, small≥6), or CTP (CBF <30%, large ≥70 cc). Mismatch categories were core–penumbra ratio: no mismatch (≤1.2), low mismatch (>1.2–1.8), and favorable mismatch (>1.8). Multivariable logistic regression assessed associations between demographics, stroke characteristics, imaging, hospital designation, and EVT use. Results: Among 6,667 patients (median age 70 [IQR 61-79], 43.6% female; NIHSS 5 [IQR 2-12]; median onset-to-arriva1 142 minutes [IQR 62-430]; 66.1% with pre-morbid mRS 0-2), EVT was performed in 22%. Greater odds of EVT use were associated with female sex (aOR 1.32; 95% CI 1.15–1.50), higher NIHSS (aOR 1.13 per point; 1.12-1.14), favorable mismatch (aOR 11.13; 4.65–26.66), and low mismatch (aOR 6.76; 2.57–17.80). Lower odds were seen at age ≥85 (aOR 0.77; 0.60-0.99), Medicare (aOR 0.80; 0.67-0.96) or Medicaid insurance (aOR 0.52; 0.37-0.73), at primary stroke centers (aOR 0.31; 0.25–0.38), pre-morbid dependency (aOR 0.37; 0.29–0.46), undocumented mRS (aOR 0.41; 0.34–0.50), IV thrombolysis (aOR 0.84; 0.72–0.98), longer onset-to-arrival (aOR per minute 0.99) and higher systolic BP (aOR per mmHg 0.99). Conclusions: In this statewide registry–imaging linked cohort, imaging-defined perfusion mismatch was the strongest predictor of EVT use. Advanced age, pre-morbid disability, and care at a primary stroke center significantly reduced EVT use. These findings highlight the central role of imaging in treatment decisions and the need to address systemic disparities in EVT access.