Article Text

Download PDFPDF
Paramedic validation of an australian large vessel occlusion triage algorithm for stroke
  1. Henry Zhao1,
  2. Lauren Pesavento1,
  3. Skye Coote2,
  4. Leonid Churilov3,
  5. Karen Smith4,
  6. Stephen Bernard4,
  7. Nawaf Yassi1,
  8. Stephen M Davis1,
  9. Bruce CV Campbell1
  1. 1Royal Melbourne Hospital and The University of Melbourne Department of Medicine, Melbourne, VIC, Australia
  2. 2Neurology, Eastern Health, Melbourne, VIC, Australia
  3. 3The Florey Institute of Neuroscience and Mental Health, Melbourne, VIC, Australia
  4. 4Ambulance Victoria, Melbourne, VIC Australia


Objectives Endovascular reperfusion is now standard treatment for large vessel occlusion (LVO) but good outcomes depend on timely therapy. Accurate identification of LVO in the pre-hospital setting has therefore become a key priority to allow ambulance bypass to endovascular centres. Several tools such as the Spanish Rapid Arterial Occlusion Evaluation Scale (RACE)1 have been developed, but paramedic studies to date have shown low specificity, despite training.1,2 We aimed to develop an identification algorithm that would provide high accuracy when used by Australian paramedics.

Methods A new LVO algorithm was developed from retrospective review of discriminating clinical features and requires significant unilateral upper limb weakness (arm falls to stretcher <10 secs), plus either severe language deficit or presence of gaze deviation/severe extinction (assessed by response to shoulder tap). Initial retrospective validation was performed in consecutive code stroke patients over a 15 month period at Royal Melbourne and Box Hill hospitals in Melbourne, followed by prospective paramedic assessment at Royal Melbourne Hospital.

Results Of 565 consecutive patients in the retrospective cohort (82 LVO), the overall accuracy of the LVO algorithm was 87.6%. Misclassification inaccuracies affected 4 LVO patients with clear endovascular eligibility (4.8% of all LVO), and 10 smaller infarcts incorrectly identified as LVO (5.7% of all non-LVO infarcts). Use of the algorithm by paramedics without additional training in the first 70 prospective patients (13 LVO) showed 87.9% accuracy, 83.3% sensitivity and 88.9% specificity. No LVOs with clear endovascular eligibility were missed and just 1 non-LVO infarct was misclassified. All discrimination parameters trended superior to RACE.

Conclusions The new 3-item LVO algorithm has excellent sensitivity for endovascular-eligible LVO and misclassified only a small proportion of non-LVO infarcts. Despite greater simplicity, the algorithm shows better performance than the RACE and has potential to improve patient outcomes from endovascular therapy through faster treatment access.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.