Abstract Number: LBA3 Head Pulse For Ischemic Stroke Detection Prehospital Study During COVID‐19 Pandemic (EPISODE‐PS‐COVID): Initial Trial Results

Abstract

Introduction Prehospital identification of large vessel occlusion (LVO) ischemic stroke using stroke scoring tools based upon physical exam findings is highly user‐dependent and associated with poor specificity. Measurement of the “headpulse” by cranial accelerometry has been previously shown to reliably identify LVO in a hospitalized group of suspected stroke patients, but the utility of this technology in the prehospital environment remains unclear. The EPISODE‐PS‐COVID study was designed to evaluate the use of a novel triaxial cranial accelerometry device (Harmony, MindRhythm, Inc.) for predicting LVO using prehospital data. Methods Paramedics from multiple emergency medical services (EMS) agencies were trained in placement of the Harmony device. Patients suspected of LVO stroke by EMS personnel received placement of the non‐invasive headband device in the field, and prehospital data were collected over a 3‐minute period prior to hospital arrival. Headpulse data were then compared to clinical data collected during a subsequent emergency department (ED) and hospital visit, including results of brain computed tomography angiogram (CTA) as the gold standard for LVO diagnosis. Patients were consented for inclusion in the study after hospital arrival, utilizing a delayed consent model, as approved by the local institutional review board (IRB). Los Angeles Motor Scale (LAMS) scores were also calculated by prehospital providers for comparison with headpulse measurements. Results A total of 377 enrollments were made at six participating hospitals over a period of 18 months (April 2021 – September 2022), including 92 subjects with recordings free of technical recording errors who were able to provide informed consent for use of their data. Nine (9.8%) subjects were confirmed to have LVO by subsequent CTA, while 15 (16.3%) were diagnosed with non‐LVO ischemic stroke, seven (7.6%) with intracranial hemorrhage, and 61 (66.3%) with stroke mimics. The Harmony device algorithm correctly identified seven of nine LVO strokes, resulting in 78% sensitivity, as well as 82 of 83 non‐LVO diagnoses (99% specificity, p < 0.0001). In this data set, contemporaneous LAMS score demonstrated 44% sensitivity and 82% specificity (p = 0.087). Device innovation during the trial significantly reduced the number of technical errors. Device performance numerically exceeded the LAMS score, with p = 0.099. Conclusions Data from this pilot study suggest that the Harmony device has a sensitivity and specificity that may be non‐inferior to the LAMS score when tested in an undifferentiated population of suspected stroke patients. Although the sample size is underpowered to test for superiority, enrollment in the trial is ongoing. A blinded, pivotal trial will begin soon. The use of a non‐invasive device to detect LVO has substantial implications for prehospital LVO stroke management, including the potential for improved triage and improved outcomes for patients with LVO stroke, without delaying care for non‐LVO stroke patients.