Performance of the ERC/ESICM-recommendations for neuroprognostication after cardiac arrest: insights from a prospective multicenter cohort
Résumé
Brief abstract: In a multicentre network of 28 ICUs in France and Belgium, all comatose patients who fulfilled the 2021 ERC-ESICM criteria for poor outcome after cardiac arrest died or survived with severe neurological disability, even after excluding patients with active WLST to limit self-fulfilling prophecy bias. However, in almost half of the patients, these criteria were not fulfilled, resulting in an indeterminate outcome; in these patients, normal NSE levels and benign EEG predicted neurological recovery, helping reduce prognostic uncertainty.
Aim
To investigate the performance of the 2021 ERC/ESICM-recommended algorithm for predicting poor outcome after cardiac arrest (CA) and potential tools for predicting neurological recovery in patients with indeterminate outcome.
Methods
Prospective, multicenter study on out-of-hospital CA survivors from 28 ICUs of the AfterROSC network. In patients comatose with a Glasgow Coma Scale motor score ≤3 at ≥72 hours after resuscitation, we measured: 1) the accuracy of neurological examination, biomarkers (neuron-specific enolase, NSE), electrophysiology (EEG and SSEP) and neuroimaging (brain CT and MRI) for predicting poor outcome (modified Rankin scale score≥4 at 90 days), and 2) the ability of low or decreasing NSE levels and benign EEG to predict good outcome in patients whose prognosis remained indeterminate.
Results
Among 337 included patients, the ERC-ESICM algorithm predicted poor neurological outcome in 175 patients, of whom 106 (60%) had withdrawal of life-sustaining treatment (WLST). Among the 69 patients without active WLST, the positive predictive value for an unfavourable outcome was 100% [95-100]%. The specificity of individual predictors ranged from 90% for EEG to 100% for clinical examination and SSEP. Among the remaining 162 patients with indeterminate outcome, a combination of 2 favourable signs predicted good outcome with 99[96-100]% specificity and 23[11-38%]% sensitivity.
Conclusion
All comatose resuscitated patients not undergoing WLST who fulfilled the ERC-ESICM criteria for poor outcome after CA had poor outcome at three months, even if a self-fulfilling prophecy cannot be completely excluded. In patients with indeterminate outcome (half of the population), favourable signs predicted neurological recovery, reducing prognostic uncertainty.
Fichier principal
2024 Bougoin et al., Performance of the ERC.pdf (466.16 Ko)
Télécharger le fichier
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|