Abstract 264: Enhancing Stroke Care: Neurology Residents' Clinical Prediction of Large Vessel Occlusions for Faster Endovascular Activation

Abstract

Introduction Large vessel occlusions (LVO) account for approximately 15 to 30% of acute ischemic strokes, necessitating rapid clinical recognition to facilitate timely endovascular team activation and reduce door‐to‐reperfusion time for mechanical thrombectomy (MT). Existing clinical LVO scales, such as RACE, FASTED, and VAN, designed for utilization by Emergency Medical Services (EMS), lack reliability and have not been widely adopted for pre‐activation of endovascular teams. Typically, endovascular team activation occurs after LVO confirmation through neuroimaging, highlighting the need for improved methods to identify LVO during early assessments by neurologists. Methods A prospective observational study was conducted at University Hospital, involving all stroke codes activated by the Emergency Department (ED) or EMS. Neurology residents responding to the codes performed initial assessments, encompassing patient history, physical examination, and NIH Stroke Scale (NIHSS) evaluations. Based on the onset and severity of symptoms, presence of cortical signs, and NIHSS scores, patients were classified into four categories: A) probable LVO and stroke, B) possible LVO and stroke, C) not LVO but probable stroke, and D) probable stroke mimic. All assessments were completed before CT scanning. Results Over a ten‐month period, 159 stroke codes were evaluated by neurology residents. Among these, 27 patients were diagnosed with LVO, while 132 patients had no LVO. The neurology residents classified 18 patients into the probable stroke group, demonstrating a sensitivity of 48%, specificity of 96%, a positive predictive value (PPV) of 72%, and a negative predictive value (NPV) of 90%. Combining the probable and possible stroke groups, the residents identified 34 patients with a sensitivity of 74%, specificity of 89%, a PPV of 5%, and an NPV of 94%. Regrettably, seven patients with LVO were misclassified as either stroke mimics or non‐LVO strokes. Among these cases, five patients presented with low NIHSS scores, two had LVO in the posterior circulation, and one developed an in‐stent thrombus. Conclusion Our study revealed that neurology residents' ability to predict LVO was comparable to other LVO prediction scales in terms of sensitivity and specificity, but significantly improved in PPV. With a PPV exceeding 70%, the clinical prediction of LVO by neurology residents proves to be a valuable tool for activating the endovascular team promptly. This approach holds promise for enhancing stroke care in both the emergency department and pre‐hospital ambulance‐based settings. Nevertheless, larger studies are warranted to validate and further refine these findings. Implementing the neurology residents' clinical judgment in LVO identification can potentially optimize patient outcomes and treatment timelines in acute ischemic strokes.