Abstracts

Rationale: Cardiopulmonary support with extracorporeal membrane oxygenation (ECMO) is used in critically ill neonates and children. The American Clinical Neurophysiology Society Guideline on continuous EEG (cEEG) monitoring recommends that in such patients cEEG be used for the detection of electrographic seizures (ES). We assessed the association between EEG background activity and ES and hospital mortality or presence of imaging-confirmed acute brain lesion (ABL). Methods: A seven-year (2010-2017) retrospective cohort of all pediatric patients (0-21 years) who underwent cEEG during ECMO at Boston Children’s Hospital. We excluded cases with a previous history of seizures or ABL before ECMO cannulation. Data collected included demographic and disease information (cardiac, respiratory, infectious), type and site of cannulation (venous-to-arterial [VA, peripheral or central], or venous-to-venous [VV]). ES included seizures with or without a clinical correlate. The initial EEG background (first 24 hours of recording) was assessed as normal, mild, moderate or severe. In neonates a severe background abnormality was defined as extremely low voltage or burst suppression, and in pediatric patients as attenuated or featureless or showing burst suppression. We subsequently evaluated the association of severe EEG or development of ES, with death and presence of new ABL; the potential diagnostic impact is presented as pre-test (Ppre) to post-test (Ppost) probability values. Results: In 201 (109 pediatric, 92 neonatal) patients, 51% were male, and 67% had a cardiac cause leading to ECMO (post cardiac arrest support in 115/201). VA-ECMO was the most common mode of cannulation (87%).  The most common initial EEG background was slow/disorganized in 74% (81/109) of pediatric patients, and excessively discontinuous in 49% (75/92) of neonates.The mortality rate in the group overall (P_pre) was 47% (95% CI 40-54%), while the P_post of death given a severely abnormal initial EEG background was 88% (95% CI 69-98%). Patients with an EEG that showed a background that was not severely abnormal had similar P_post (41% [95% CI 34-49%]) to the P_pre for the entire group (47% [95%CI 40-54%]). The subsequent development of ES, or not, did not add any further information about prognosis regarding death.ABL was present in 68% (95% CI 61-74%) of the study subjects. The presence of ES was associated with P_post of ABL of 97% (95% CI 84-100%). However, the lack of ES was not informative (i.e., low sensitivity 0.24 with P_post for ABL 62% [95%CI 54-69%]). Conclusions: In these ECMO patients with cEEG a severely abnormal initial EEG background was associated with a high mortality rate and potentially useful in the first step for prognostication of death (P_pre to P_post 47% to 88%). Subsequent identification of ES was associated with an acute brain lesion (ABL), but not mortality (P_pre to P_post 68% to 97%). cEEG is an important tool for diagnostic and prognostic evaluation of pediatric patients undergoing ECMO. Funding: None