Oral Presentation Asia Pacific Stroke Conference 2024

The HERMES-24 Score: External Validation in a Contemporary Clinical Cohort (112661)

Vincent Trinh 1 2 , Minyan Zeng 1 2 , Rebecca Scroop 3 4 , Jim Jannes 4 5 , Timothy Kleinig 4 5 , Lyle J Palmer 1 2
  1. Australian Institute for Machine Learning, University of Adelaide, Adelaide, South Australia, Australia
  2. School of Public Health, University of Adelaide, Adelaide , South Australia, Australia
  3. Department of Radiology, Royal Adelaide Hospital, Adelaide, South Australia, Australia
  4. School of Medicine, Faculty Health and Medical Science, University of Adelaide, Adelaide, South Australia, Australia
  5. Department of Neurology, Royal Adelaide Hospital, Adelaide , South Australia, Australia

Background: The HERMES-24 score derived from age and 24-hour National Institute of Health Stroke Scale (NIHSS) has been recently proposed for prognosticating 90-day outcome following large vessel occlusion (LVO) stroke. Excellent predictive performance was achieved in the HERMES, ESCAPE-NA1 and INTERRSect trials cohorts. External validation in routine clinical cohorts is lacking.

Aim: Our study aimed to externally validate and assess the predictive performance of the HERMES-24 score in a large contemporary clinical cohort.

Methods: Patients with confirmed anterior circulation LVO stroke aged >18 years and receiving endovascular thrombectomy (EVT) within 24 hours were included. The association between HERMES-24 score and 90-day functional outcome (mRS ≤2 and ≤3), ordinal mRS and mortality was investigated using generalised linear models. Predictive performance was assessed via the c-statistic.

Results: A total of 1052 patients were included for analysis. Median age and NIHSS at presentation was 75.0 years (IQR 64.0 – 82.3) and 14.0 (IQR 8.0 – 20.0) respectively. TICI 2b or greater reperfusion was achieved in 862 (81.9%). At 90 days, 202 (21.5%) patients had died and 486 (51.8%) regained functional independence (mRS ≤2). The HERMES-24 score demonstrated (similarly to previously) excellent performance for predicting 90-day mRS ≤2 (c-statistic, 0.862, 95% CI: 0.839 – 0.885), mRS ≤3 (c-statistic, 0.870, 95% CI: 0.846 – 0.893), mortality (c-statistic, 0.859, 95% CI: 0.830 – 0.888) and ordinal mRS (c-statistic, 0.808, 95% CI: 0.792 – 0.826).

Conclusion: The HERMES-24 score has utility as a prognostic tool for LVO stroke in the routine clinical setting. Further validation in basilar occlusion LVO stroke would be beneficial.