Abstracts

Rationale: Extracorporeal membrane oxygenation (ECMO) is a form of cardiopulmonary support that can be life saving for children with cardiac or respiratory failure. There is a high risk of neurological injury during ECMO such as stroke. Recently, a high-risk of non-convulsive seizures has also been recognized. Our purpose was to analyze EEG in order to predict imaging abnormalities in children requiring ECMO. We hypothesize that specific EEG patterns will predict abnormal imaging and mortality.Methods: We retrospectively examined the utility of continuous EEG (CEEG) for ECMO in the intensive care units (ICU) at a quaternary care center. ECMO protocol was recently revised to include CEEG in most patients. We studied children ages 0-21 years undergoing ECMO between January 2010 and April 2015. Primary outcome measures were electrographic seizures (EGSz), abnormal brain imaging (ABI) and death.Results: 172 cannulations were performed, 50 following the protocol change (Fig 1). Clinical seizures occurred in 23% (n=8); 3 without electrographic correlate. EGSz were recorded in 26% (n=10), 5 without clinical correlate. Status epilepticus occurred in 15% (n=5). Most seizures occurred at least 24 hrs after ECMO initiation; only 1 was within 6 hrs. 43% (n=21) had epileptiform discharges and 91% (n=31) had slowing. Imaging was abnormal in 41% (n=19) (Fig 1). CEEG findings, such as focal attenuation, focal slowing, or epileptiform discharges correlated with ABI in 47% (n=8). 80% (n=40) survived during ECMO, but then 38 % (n=19) died. There was no correlation between arrest to cannulation time or cannulation site and EGSz, ABI or death. A trend for a higher rate of EGSz occurred with a respiratory versus a cardiac diagnosis (p=0.074, OR 4.4), but no difference for ABI or death. A longer duration of ECMO correlated with a higher death risk (p=.03), however there was no trend with regards to EGSz or ABI. There was a significant correlation between EGSz and ABI (p=0.054, OR 7.3). There was no significant correlation between EGZ or ABI and death.Conclusions: Our CEEG data demonstrated 23% and 26% incidences of clinical and EGSz during ECMO in children, respectively, many of which were subclinical. The risk of seizures in this cohort study are comparable to prior studies of seizures in ECMO patients, however in our cohort there is a large number of patients with difficult-to-control seizures, often fulfilling criteria for status epilepticus. There was a high number of strokes in this cohort, many of which have corresponding EEG abnormalities. The death rate was slightly higher than what is reported in more recent studies, however this is likely due to selection bias; ECMO is more often imposed on sicker, less stable patients in this specialized care center. We found the longer patients were on ECMO, the higher risk of death. With further analysis our hope is that we will be able to find certain EEG predictors for prognosis in these children and to teach future physicians about the neurological risks of ECMO to improve the lives of these children.