Electroencephalographic features in patients undergoing extracorporeal membrane oxygenation

Background Neurologic injury is one of the most frequent causes of death in patients undergoing extracorporeal membrane oxygenation (ECMO). As neurological examination is often unreliable in sedated patients, additional neuromonitoring is needed. However, the value of electroencephalogram (EEG) in adult ECMO patients has not been well assessed. Therefore, the aim of this study was to assess the occurrence of electroencephalographic abnormalities in patients treated with extracorporeal membrane oxygenation (ECMO) and their association with 3-month neurologic outcome. Methods Retrospective analysis of all patients undergoing venous–venous (V–V) or venous–arterial (V–A) ECMO with a concomitant EEG recording (April 2009–December 2018), either recorded intermittently or continuously. EEG background was classified into four categories: mild/moderate encephalopathy (i.e., mostly defined by the presence of reactivity), severe encephalopathy (mostly defined by the absence of reactivity), burst-suppression (BS) and suppressed background. Epileptiform activity (i.e., ictal EEG pattern, sporadic epileptiform discharges or periodic discharges) and asymmetry were also reported. EEG findings were analyzed according to unfavorable neurological outcome (UO, defined as Glasgow Outcome Scale < 4) at 3 months after discharge. Results A total of 139 patients (54 [41–62] years; 60 (43%) male gender) out of 596 met the inclusion criteria and were analyzed. Veno–arterial (V–A) ECMO was used in 98 (71%); UO occurred in 99 (71%) patients. Continuous EEG was performed in 113 (81%) patients. The analysis of EEG background showed that 29 (21%) patients had severe encephalopathy, 4 (3%) had BS and 19 (14%) a suppressed background. In addition, 11 (8%) of patients had seizures or status epilepticus, 10 (7%) had generalized periodic discharges or lateralized periodic discharges, and 27 (19%) had asymmetry on EEG. In the multivariate analysis, the occurrence of ischemic stroke or intracranial hemorrhage (OR 4.57 [1.25–16.74]; p = 0.02) and a suppressed background (OR 10.08 [1.24–82.20]; p = 0.03) were independently associated with UO. After an adjustment for covariates, an increasing probability for UO was observed with more severe EEG background categories. Conclusions In adult patients treated with ECMO, EEG can identify patients with a high likelihood of poor outcome. In particular, suppressed background was independently associated with unfavorable neurological outcome.