ABSTRACT NUMBER: ESOC2026A1954 PREHOSPITAL VIDEO TRIAGE FOR STROKE IN NORTH CENTRAL LONDON: LARGE-SCALE ANALYSIS OF DIAGNOSTIC ACCURACY AND IMPACT ON HYPERACUTE REPERFUSION TREATMENT TIMES

Abstract Background and aims Accurate prehospital stroke recognition allows rapid delivery of specialist stroke care, including reperfusion treatments. The current standard of care in the United Kingdom, the FAST test, lacks specificity. To address this, a centralized prehospital video triage (PVT) service was established in North Central London in 2020. We evaluated the efficacy of this service using comprehensive, long-term data. Methods We prospectively collected data on consecutive PVT episodes between August 2020 and August 2024. We recorded whether each triaged patient was directed to a hyperacute stroke unit (HASU) and determined whether a diagnosis of stroke was made following triage using a regional stroke registry. We compared the sensitivity, specificity and discrimination of PVT and FAST. We compared door-to-needle and door-to-groin-puncture times for patients receiving reperfusion treatment following PVT to those treated without PVT, using linear regression. Results We included 6378 PVT episodes. 2962 (46.4%) triaged patients were conveyed to a HASU, and 1491 (23.4%) were diagnosed with ischaemic stroke or intracerebral hemorrhage. PVT had similar sensitivity to FAST and considerably higher specificity (Table 1). The c-statistic for PVT was 0.78, for FAST, 0.71; difference 0.07 (95%CI 0.06-0.08, P < 0.0001). Adjusted for patient and organizational factors, PVT use was associated with a 15.3% reduction in door-to-needle time (95%CI 5.3-24.2, P = 0.004) and 15.8% reduction in door-to-groin-puncture time (95%CI 2.3-27.4, P = 0.023). Conclusions Our findings suggest that PVT improves recognition of patients with conditions mimicking stroke without reducing sensitivity for stroke, accelerates treatment with thrombolysis and thrombectomy, and has potential to improve service efficiency and patient outcomes. Conflict of interest Jonathan Best: nothing to disclose; Chak Lam Ip: nothing to disclose; Erla Jónsdóttir: nothing to disclose; Selina Edwards: nothing to disclose; Jonathan Hayton: nothing to disclose; Ken Crossley: nothing to disclose; David Hargroves: nothing to disclose; Salman Haider: nothing to disclose; Rob Simister: nothing to disclose. Table 1 - belongs to Results