2021 European Resuscitation Council/European Society of Intensive Care Medicine Algorithm for Prognostication of Poor Neurological Outcome After Cardiac Arrest—Can Entry Criteria Be Broadened?

Abstract

OBJECTIVES: To explore broadened entry criteria of the 2021 European Resuscitation Council/European Society of Intensive Care Medicine (ERC/ESICM) algorithm for neuroprognostication including patients with ongoing sedation and Glasgow Coma Scale-Motor score (GCS-M) scores 4–5. DESIGN: Retrospective multicenter observational study. SETTING: Four ICUs, Skane, Sweden. PATIENTS: Postcardiac arrest patients managed at targeted temperature 36°C, 2014–2018. Neurologic outcome was assessed after 2–6 months according to the Cerebral Performance Category scale. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: In 794 included patients, median age was 69.5 years (interquartile range, 60.6–77.0 yr), 241 (30.4%) were female, 550 (69.3%) had an out-of-hospital cardiac arrest, and 314 (41.3%) had a shockable rhythm. Four hundred ninety-five patients were dead at follow-up, 330 of 495 died after a decision on withdrawal of life-sustaining therapies. At 72 hours after cardiac arrest 218 patients remained unconscious. The entry criteria of the original algorithm (GCS-M 1–3) was fulfilled by 163 patients and 115 patients with poor outcome were identified, with false positive rate (FPR) of 0% (95% CI, 0–79.4%) and sensitivity of 71.0% (95% CI, 63.6–77.4%). Inclusion of patients with ongoing sedation identified another 13 patients with poor outcome, generating FPR of 0% (95% CI, 0–65.8%) and sensitivity of 69.6% (95% CI, 62.6–75.8%). Inclusion of all unconscious patients (GCS-M 1–5), regardless of sedation, identified one additional patient, generating FPR of 0% (95% CI, 0–22.8) and sensitivity of 62.9% (95% CI, 56.1–69.2). The few patients with true negative prediction (patients with good outcome not fulfilling guideline criteria of a poor outcome) generated wide 95% CI for FPR. CONCLUSION: The 2021 ERC/ESICM algorithm for neuroprognostication predicted poor neurologic outcome with a FPR of 0%. Broadening inclusion criteria to include all unconscious patients regardless of ongoing sedation identified an additional small number of patients with poor outcome but did not affect the FPR. Results are limited by high rate of withdrawal of life-sustaining therapies and few patients with true negative prediction.