Value of EEG reactivity for prediction of neurologic outcome after cardiac arrest: Insights from the Parisian registry.

Purpose: To evaluate the predictive value of EEG reactivity assessment and confounders for neurological outcome after cardiac arrest.

Methods: All consecutive patients admitted in a tertiary cardiac arrest center between 2007 and 2016 still alive 48 h after admission with at least one EEG recorded during coma. EEG reactivity was defined as a reproducible waveform change in amplitude or frequency following standardized stimulation. Each EEG was classified based on American Clinical Neurophysiology Society nomenclatures and classified in highly malignant (including status epilepticus), malignant, or benign EEG. We assessed the predictive values of EEG reactivity and sedation effect for neurologic outcome at ICU discharge using the Cerebral Performance Category scale (with CPC 1-2 assumed as favorable outcome and CPC 3-4-5 considered as poor outcome).

Results: Among 428 patients, a poor outcome was observed in 80% patients. The median time to EEG recording was 3 (1-4) days and 51% patients had a non-reactive EEG. The positive predictive value (PPV) of a non-reactive EEG to predict an unfavorable outcome was 97.1% (IC95% 93.6-98.9), increasing to 98.3% (IC95 94.1-99.8) when the EEG had been performed without sedation. In multivariate analysis, a non-reactive EEG was associated with poor outcome (OR 12.6 IC95% 4.7-33.6; p < 0.001). In multivariate analysis, concomitant sedation was not statistically associated with EEG non-reactivity. The PPV of a benign EEG to predict favorable outcome was 49.7% (IC95% 41.5-57.9), increasing to 66.2% (IC95% 54.3-76.8) when EEG was recorded earlier, with ongoing sedation.

Conclusions: After cardiac arrest, absence of EEG reactivity was predictive of unfavorable outcome. By contrast, a benign EEG was slightly predictive of a favorable outcome. Reactivity assessment may have important implications in the neuroprognostication process after cardiac arrest and could be influenced by sedation.