Abstracts

Rationale: Seizures are a significant complication for pediatric patients requiring extracorporeal membrane oxygenation (ECMO) care and have been associated with increased mortality and neurodevelopmental sequelae (Campbell et al., 1991, Okochi et al., 2018). No official consensus guidelines exist on the use of continuous EEG (cEEG) monitoring in pediatric patients undergoing ECMO. However, there are efforts to place cEEGs earlier, as the majority of seizures occur within the first 24-48 hours after cannulation (Okochi et al., 2018). Our institution implemented a protocol for pediatric ECMO to improve time to cEEG initiation and seizure detection. This study was conducted to determine whether these goals were met.

Methods: This was a single center retrospective study at Johns Hopkins Children Center. We reviewed the cEEGs of patients that underwent monitoring while on ECMO between April 6, 2014 and June 27, 2020. The protocol started on April 2019 and asked that cEEG be placed on any ECMO patient as soon as possible from time of cannulation for a minimum of 72 hours. Prior to implementation, cEEGs were placed when there was a clinical suspicion for seizures. This study focused on addressing two questions: 1) does protocol implementation decrease time from ECMO cannulation to cEEG initiation? and 2) do the majority of seizures occur early after ECMO initiation?

Results: There were 50 patients in this cohort. The majority were infants (76%). Analysis of 33 pre- and 18 post- protocol cEEGs (one patient had 2 ECMO runs) demonstrated that the number of cases where EEG was started within 24 hours significantly increased after protocol implementation, indicating reduced time from ECMO cannulation to cEEG initiation (Χ²_(L) = 11.93, p < 0.01). There was no significant difference between groups for ECMO cannulation to seizure detection (pre-protocol 42 hours, IQR 16-299 and post-protocol 48 IQR 23-101, p = 0.9). However, the pre-protocol group had a 15 times greater odds of having ongoing seizures at the start of the cEEG compared to the post-protocol group in which seizure onset was captured after cEEG initiation (OR 15, 95% CI 1.37-191, p < 0.05). Comparing all seizures pre-protocol to those post-protocol, there was a significant difference in the median interval between cEEG recording start time and seizure detection, with ongoing seizures at cEEG initiation in the pre-protocol group, compared to seizures detected at 42 hours post (0 minutes, IQR 0-16.5, 42 hours, IQR 18-70, p < 0.05 respectively).

Conclusions: This study demonstrates that our protocol resulted in faster clinical care, with earlier time to cEEG initiation. In addition, protocol implementation resulted in more cases where seizures were recorded immediately at onset.

This has important implications for improving clinical outcomes, as with lead-time to seizure onset, we can intervene earlier to prevent seizures, treat them immediately and gather data that may help to predict impending seizures or changes in patient condition.

Funding: Please list any funding that was received in support of this abstract.: No direct funding for this study. CW Habela received salary support from the Doris Duke Charitable Foundation and KO8NS102526 NIH/NINDS.