Abstracts

Rationale: Extracorporeal Membrane Oxygenation (ECMO) is an advanced cardio-respiratory support for critically ill patients with cardiorespiratory failure and a part of cardiopulmonary resuscitation (CPR) when conventional CPR fails. Acute neurologic sequelae (i.e., intracranial hemorrhage, stroke, seizures) as well as chronic (developmental disability) are known complications. Studies have shown that neuro-imaging severity significantly correlates with a worse neurologic outcome. However, neuroimaging during ECMO can be challenging with risks associated. Early identification of neurologic injury can help prognosticate the degree of developmental disability and can aid decision making. We identified pediatric patients at our tertiary care center who underwent EEG during ECMO to determine the value of EEG as an early identifier of neurologic sequelae. Methods: Neonatal and pediatric patients who had underwent ECMO and EEG between 2009-2018 were identified at Children’s Medical Center in Dallas, Texas. Inclusion criteria included: an EEG during ECMO or a maximum of 48 hours post ECMO discontinuation; and acute neuroimaging either with head ultrasound (HUS), computed tomography (CT) or magnetic resonance imaging (MRI) during or within 3 weeks of ECMO stop. The EEG was ranked according to severity using defined categories for mild, moderate, or severe. Two distinct severity scales were used for neonatal EEG (<1 month) and childhood EEG (>1month – 18 years). Neuroimaging scores were done according to a severity scale with weighted categories. ECMO related information were obtained including laboratory data, ECMO type, cannulation site, vasoactive-inotropic score (VIS), and pediatric mortality/morbidity scores. Statistical analysis was performed using Fisher’s Exact Test. Results: A total of 50 patients met inclusion criteria. Six patients were off ECMO for 24-48 hours prior to the EEG. One patient had only a MRI, 7 patients had only CT scans, 8 patients had only HUS, and 34 patients had a combination of studies. There was 1 normal EEG, 12 with mild abnormalities, 17 had moderate abnormalities, and 20 had severe abnormalities. Of the severe, two had isolated electrographic and/or electroclinical seizures, eight had isolated severe background abnormalities, and ten had a combination of both electrographic and/or electroclinical seizures and severe background disturbance. EEG severity correlated significantly with neuroimaging severity (p<0.0001). Both the normal to mild EEG group and moderate to severe EEG group were statistically significant when compared to the same scale neuroimaging findings (p<0.0001). When analyzing the severe EEG findings, those with isolated seizures, severe background abnormalities, and combination of a severe background with seizures did not show a statistically significant difference in severity of neuroimaging studies. The EEG severity was not statistically significant when compared with the VIS and the serum lactate. Conclusions: In our single center study EEG severity of EEG, including the presence of electrographic and electroclinical seizures was associated to severity of neuro-imaging findings. Hence EEG may be a useful tool in the setting of ECMO in predicting severity of neurological imaging abnormalities. It can be used as an early marker of neurologic injury and help guide management. Further subgroup analysis differences between seizures or severe background EEG was not helpful in predicting imaging abnormalities. Future large multicenter studies may help distinguish these subgroups further. These findings help contribute to the growing body of literature concerning the utility of EEG in the acute care of pediatric ECMO patients, and its use to help predict long term neurologic sequelae. Funding: No funding