Introduction: Large vessel occlusion (LVO) triage scales aim to identify LVO on clinical grounds and direct ambulances to endovascular centers, but current tools are reported to have poor performance in independent datasets. We investigated situations where these tools were performing poorly and examined the possible impacts of misclassification.
Methodology: Patient data were collected prospectively from consecutive ambulance initiated stroke codes at two major stroke centers. Patients with symptom onset >6 hours and pre-existing deficits were excluded. LVO was defined as ICA, M1 or proximal M2 occlusion on CT angiography. National Institutes of Health Stroke Scale (NIHSS) scores assessed by stroke clinicians were used to derive five published triage tools (RACE, LAMS, CPSSS, PASS, FAST-ED). Patients were classified as 1) Typical presentations where LVO presented with severe symptoms (weakness and cortical sign) and non-LVO did not, or 2) Atypical presentations where non-LVO had severe symptoms and LVO did not.
Results: A total of 391 patients were included, of whom 59 were atypical presentations which included 25 ICH and 5 mimics. Most triage tools were able to correctly identify >95% of typical presentations but only 10-25% of atypical presentations. Within the atypical cases, bypassing the nearest stroke center would have delayed tPA for 6 patients (3.5% of all infarcts) with non-LVO infarcts. Conversely, of the missed LVO cases 10 patients (19.6% of all LVO) would have been immediately eligible for endovascular therapy. A further 6 M2 occlusions with NIHSS <6 would have been missed but were unlikely to have received intervention due to mild severity.
Conclusion: Misclassification by LVO triage tools is largely due to ICH and the uncommon situations where presence of LVO does not correspond to a severe clinical syndrome (and vice versa). Inaccuracies however, would have resulted in relatively few potential tPA patients bypassing the nearest center. Whilst a small proportion of LVO will be missed, overall delivery of LVO to endovascular centers will still be dramatically increased. LVO triage tools therefore have excellent potential to expedite endovascular treatment, but efforts are needed to improve prehospital paramedic accuracy to achieve this goal.