T64. Continuous EEG monitoring in extracorporeal membrane oxygenation in the pediatric and neonatal intensive care unit

Abstract

Introduction Electrographic seizures are common in patients on ECMO. The impact of EEG seizures on outcome is unclear. We aim to describe the main clinical and electroencephalographic (EEG) characteristics of neonates and children on extracorporeal membrane oxygenation (ECMO) while focusing on the impact of seizures on outcome. Methods Retrospective study of patients from birth to 21 years with cEEG in the neonatal and pediatric intensive care unit (ICU) while requiring ECMO at Boston Children’s Hospital from October 2010 to September 2016. Inclusion criteria included any patient that achieved adequate flow on ECMO with each ECMO cannulation included as a separate entry. Two patients were excluded due to the presence of seizures on EEG prior to ECMO cannulation. Variables collected included demographics, reason for ECMO, ECMO type (VA vs. VV) and site, time from ECMO to EEG, and EEG background features. Results 401 patients (182 neonates, 219 pediatric) were included. The median age was 38 weeks with 64.3% male and 2.1 years with 50% male for the neonatal and pediatric groups respectively. The most common reason for ECMO was due to a congenital heart defect 41.9% (168/401) and respiratory distress/pulmonary hypertension 25.9% (104/401). Of those, 168 (76 neonates, 92 pediatric) had an EEG while on ECMO. Most common indication for EEG monitoring included detection of subclinical seizures 90.5% (152/168) and to characterize events 9.5%(16/168). The percent of patients on ECMO with cEEG increased from 25.5% (13/51) in 2011 to 64.8% (46/71) in 2016. ECMO cannulation to EEG had a median time of 24 h (23.7 for the neonates and 24.1 for the pediatric). Seizures were detected in 16.7% (28/168) of patients with an EEG while on ECMO (17 neonates, 11 children), the majority were detected in 2015 and 2016 (16/28). Half of the seizures observed (14/28) were electrographic only, the remainder included electro-clinical (12/28) and clinical only (1/28) seizures. The majority of the seizures were on VA ECMO (15/17 in neonatal, 11/11 in pediatric), with the most common cannulation site being the right carotid artery (16/26). Median time from ECMO cannulation to first seizure was 37.5 h and 69 h, and first seizure to first medication was 6.25 and 6.5 h for the neonatal and pediatric groups respectively. One patient was untreated. Mortality was high, as 50% (199/401) of patients on ECMO died before leaving the hospital. Of those on EEG, the difference in mortality between the group with seizures versus no seizures (46% vs. 54%) was not statistically significant (p = 0.54). Conclusion This study supports that patients requiring ECMO are at extremely high risk for seizures. These seizures would go undetected without the use of cEEG given the high percentage of electrographic only seizures. While our study shows that seizures do not have a direct effect on mortality, further study is needed to assess the effects of seizures on developmental outcome.