Background and Aims:
Potential concerns of severity-based clinical tools for endovascular thrombectomy triage include comprehensive stroke center (CSC) overburdening and delayed thrombolysis in non-thrombectomy patients. We analyzed the implications of using the 2-step ACT-FAST paramedic triage algorithm in a real-world prehospital stroke population.
Data were included from the ongoing in-field paramedic validation study since November 2017. Ambulance Victoria paramedics assessed pre-hospital ACT-FAST (arm drift followed by speech if right arm weak, or shoulder tap for inattention if left arm weak) in suspected stroke patients in the state of Victoria, Australia after watching a 8min training video. Algorithm results were validated against prespecified comparators of ICA/M1-occlusion and requirement for CSC or neurosurgical care.
Of n=196 pre-hospital assessments (n=72 ICA/M1-occlusion), ACT-FAST had 93.8% sensitivity and 74.4% specificity. Positive predictive value for ICA/M1-occlusion was 41.7%, with remaining positive assessments comprising: basilar occlusion(2.7%), proximal M2 occlusion(5.5%), dissection/intracranial atherosclerosis(4.2%), intracranial hemorrhage(22.2%), intracranial tumor(1.4%), distal infarcts(15.3%) and normal imaging(6.9%). Positive predictive value for requiring CSC/neurosurgical care (large vessel occlusion, hemorrhage and tumor) was 77.8%. Median additional travel time in metropolitan Victoria from closest thrombolysis hospital to CSC was 7min(IQR 4-9) calculated using prior transfer cases.
Paramedic pre-hospital assessment of the severity-based ACT-FAST triage algorithm identified the majority of patients requiring thrombectomy and comprehensive stroke unit care. If used to bypass patients to CSC, the false-positive workload appears manageable. Relatively few non-thrombectomy patients would bypass the nearest thrombolysis centre, and potential delay in thrombolysis is likely minimal in metropolitan areas given the transport time differential.