Platform Session – Electroencephalography/Epilepsy: Temporal development of cEEG patterns as predictors of prognosis after cardiac arrest

Abstract

Continuous EEG-monitoring (cEEG) is used to detect electrographic seizures and is a promising prognostic tool in comatose patients after cardiac arrest (CA). The objective of this study was to investigate the temporal development of epileptiform activity and EEG background activity as predictors of prognosis. In this prospective substudy of the Target Temperature Management trial six sites recorded simplified cEEG on comatose patients resuscitated after CA. The cEEG-data was blinded to the treating team and assessed, blinded to outcome and clinical data, at 12, 24, 36, 48 and 60 h after CA. The standardized EEG terminology of the American Clinical Neurophysiology Society was used. The time-points of recovery of a normal-voltage continuous background and the appearance of an epileptiform EEG were determined, the latter defined as abundant epileptiform discharges (⩾0.1 Hz), abundant periodic or rhythmic discharges or electrographic seizure activity for 30 min. The cEEG-data was not part of the multimodal prognostication protocol for decisions on withdrawal of care. Poor outcome was defined as best Cerebral Performance Category 3–5 until 180 days. 134 patients were included and 65 had a good outcome. 103 patients recovered a continuous normal-voltage background. Background recovery within 24 h occurred in 72 patients, 55 (76%) had good outcome. Among the 31 patients with background recovery beyond 24 h, 10 (32%) had good outcome. Thus a late recovery of background increased the odds for a poor outcome six times (sens 55%, spec 85%). All 31 patients that failed to recover their background within the monitoring period (60 h after CA) had a poor outcome (sens 45%, spec 100%). Appearance of an epileptiform EEG occurred in 66 patients. Among 38 patients with an epileptiform EEG within 24 h, 34 (89%) had poor outcome (sens 49%, spec 94%). 28 patients had start of an epileptiform EEG beyond 24 h, 16 had a poor outcome (sens 46%, spec 80%). Thus an early epileptiform EEG increased the odds for a poor outcome seven times. The logistic regression model showed no significant interactions between background and epileptiform activity regarding predictive ability. Thus assessing both the time to recovery of background and the time to appearance of epileptiform activity improved prognostication. 16 of the 65 patients with good outcome had an epileptiform EEG and 14 (88%) of these patients recovered their background at the same time or before the onset of epileptiform activity. Abundant periodic or rhythmic discharges were seen in 3 survivors with good outcome and all had a recovery of background within 24 h and before the onset of discharges. Time to epileptiform activity and EEG background recovery are independent prognostic indicators. Patients with early background recovery combined with late appearance of epileptiform activity may have a good prognosis.