Abstracts

Rationale: Extracorporeal membrane oxygenation (ECMO) is a form of cardiopulmonary life support, which can be associated with significant risk of neurological morbidity. A standardized ECMO neuromonitoring protocol was therefore established at our institution which included continuous video EEG (cEEG) monitoring for all pediatric patients, for the duration of ECMO therapy. ACNS guidelines recommend cEEG to detect seizures or status epilepticus (SE) in the setting of paralytic agents , which indirectly indicate the presence of acute brain injury, as well as an additional tool to predict patient outcome and survival. The aim of the study is to evaluate the risk factors for mortality in pediatric ECMO patients such as seizure activity or other clinical variables. Methods: A consecutive cohort of pediatric patients age 0-18 years, who underwent ECMO and cEEG monitoring at our institution between January 2015-December 2018 were reviewed. Association of recent cardiac arrest, laboratory values, and imaging findings to the clinical outcomes such as time to death and time to seizures, were assessed using Cox proportional hazard regression analysis. Results: Fifty-one children underwent ECMO during this period. Two were excluded due to lack of EEG data, and 1 excluded because ECMO began 1 week prior to hospital transfer. In the remaining 48, average time to cEEG initiation was 0.4 days. Thirty-three patients (68%) had cEEG throughout the entire ECMO course, and the remaining 15 patients had cEEG during an average of 59% of ECMO therapy (range 13%-91%). Eight patients (16%) had seizures, of which three (37%) had SE. Six of eight patients (75%) had exclusively electrographic seizures. Mean time to first seizure was 71 hours after ECMO initiation. Half of the patients had only one focus, and the rest had multifocal EEG onsets. Six patients (75%) with seizure activity died, including all with SE. The two surviving patients with seizures during ECMO did not develop epilepsy and eventually weaned off of seizure medications. Univariable analysis showed mortality was increased for imaging findings of cerebral edema (hazard ratio (HR) 7.28, 95% confidence interval (CI) 2.84-18.66, p=<0.001), with a trend for cardiac arrest immediately prior to ECMO (HR 2.29, 95% CI 1.00-5.26, p=0.051). There was an inverse relationship between mortality and pre-ECMO bicarbonate level (HR 0.88, 95% CI 0.80-0.97, p=0.01). Although all patients with seizures died, this was not a statistically significant risk factor for mortality (HR 1.61, 95% CI 0.58-4.50, p=0.37). In multivariable analysis, increased risk of death was associated with 1) cardiac arrest immediately prior to ECMO (HR 4.94, 95% CI 1.53-15.95, p= 0.008), 2) imaging findings of cerebral edema (HR 13.52, 95% CI 3.95-46.26, p <0.001) and 3) lactate level >100mg/dL (HR 1.21, 95% CI 1.01-1.44, p=0.031). Conclusions: Seizures are common in children undergoing ECMO, with a high rate of subclinical seizures and SE, as in prior studies of pediatric ECMO. cEEG monitoring for at least three days after ECMO initiation should be strongly considered. Risk factors for mortality include cardiac arrest immediately prior to ECMO, presence of cerebral edema and high lactate levels. Funding: No funding