EEG based predictions of good outcome after cardiac arrest improve with sevoflurane sedation, as compared with propofol.

Abstract

We investigated the prognostic value of the early electroencephalogram (EEG) in comatose patients after cardiac arrest, sedated with sevoflurane, as compared to those sedated with propofol. This retrospective cohort study included all resuscitated patients aged ≥18 after cardiac arrest (CA) admitted to the intensive care unit of two large Dutch teaching hospitals. In one hospital all CA patients were sedated with sevoflurane only and cooled to 36 °C, while in the other hospital the patients were sedated with propofol and cooled to 33 °C. EEG patterns at 12 and 24 h after CA were analyzed visually and quantitatively and classified as favorable, unfavorable or other EEG patterns. Quantitative parameters including background continuity index, burst-suppression amplitude ratio, and alpha-delta ratio were compared between groups of sedation. Outcome at 6 months was defined as good (Cerebral Performance Category 1 or 2) or poor (Cerebral Performance Category 3, 4 or 5). We included 412 patients of whom 51 sevoflurane-sedated and 361 propofol-sedated. Predicting good outcome at 12 h after CA, we found a higher sensitivity for those sedated with sevoflurane (0.89, 95 %-CI 0.62-1.00), as compared with those sedated with propofol (0.42, 95 %-CI 0.34-0.50), without significant loss of specificity (0.71, 95 %-CI 0.44-0.91 and 0.88, 95 %-CI 0.81-0.92 respectively). No significant differences in sensitivity and specificity were found between sedation with sevoflurane and propofol for predicting good outcome using the EEG at 24 h after CA. For prediction of poor outcome at 12 and 24 h after CA, no significant differences were found between patients sedated with sevoflurane and propofol regarding sensitivity and specificity. In this non-randomised two-centre cohort study, sevoflurane sedation was associated with less frequent discontinuous EEG 12 h after cardiac arrest, enabling earlier and reliable prediction of good outcome. Prediction of poor outcome was reliable with both sevoflurane and propofol. However, potential bias due to differences in temperature management cannot be excluded.