Abstracts

Rationale: Continuous video electroencephalography (cEEG) is considered the standard of care for neonates deemed to be at risk of seizures as most neonatal seizures are electrographic (1, 2). However, significant resource and expertise are required to maintain cEEG and to provide a real-time response to seizures detected with cEEG and this is unfeasible in the majority of neonatal intensive care units (3). We hypothesised that the first hour of cEEG could predict whether neonates would subsequently develop seizures, allowing clinicians to distinguish “high risk” neonates in whom monitoring should be reviewed very frequently from those who are unlikely to develop a significant seizure burden and therefore can be reviewed less frequently. Methods: EEG background features of the first hour of 268 untreated term neonates who underwent cEEG monitoring for 24-120 hours for the NEOLEV2 trial were reviewed independently by SLD and EML Records were graded as normal, mildly (A), moderately (B) or severely (C) abnormal using Tharp’s neonatal EEG background classification(4). If seizures occurred in the first hour of monitoring this was also noted. The significance of the association between the background abnormality and/or seizure in the first hour of monitoring and subsequent seizure burden and time to first seizure were determined. Covariates including gender, cord pH, presence of pain relief or sedation, underlying aetiology of seizures and presence of cooling were analysed. Interrater reliability comparing the experienced neurophysiologist and trainee were calculated. Results: Of neonates with an abnormal EEG in the first hour of monitoring 71/145 (49%) subsequently developed seizures within 24 hours while 16/123 (13%) of neonates with a normal first hour of cEEG monitoring subsequently developed seizures within 24 hours (Likelihood ratio: 41.9, p <0.001). Neonates with a normal first hour of monitoring were 6.7 times less likely to have a seizure in the first 24 hours than those with an abnormal background (OR 0.15, 95% confidence interval 0.08-0.29) and 7.1 times less likely to have a seizure during their entire subsequent recording (24-120 hours) (OR 0.14, 95% confidence interval 0.08-0.25). Neonates with normal/A backgrounds were 20 times less likely to develop seizures compared to neonates with B/C backgrounds (OR: 0.05, 95% confidence interval 0.03-0.10). Interrater reliability was good when comparing normal versus abnormal background (Kappa 0.77) and excellent when compared normal/A versus B/C backgrounds (Kappa 0.93). Conclusions: The EEG during the first hour of monitoring in at risk neonates is highly, but not perfectly predictive of whether seizures will occur over the ensuing 24 hours. This finding allows clinicians to identify neonates at high risk of subsequent seizures who require closer observation.References: 1. Shellhaas RA, Chang T, Tsuchida T, et al. The American Clinical Neurophysiology Society's Guideline on Continuous Electroencephalography Monitoring in Neonates. Journal of Clinical Neurophysiology 2011;28:611-6172. Boylan GB, Stevenson NJ, Vanhatalo S. Monitoring neonatal seizures. Seminars in Fetal and Neonatal Medicine 2013;18:202-2083. Sharpe C, Davis SL, Reiner GE et al. Journal of Clinical Neurophysiology 2018;36(1):9-13. 4. Tharp BR. Neonatal and pediatric electroencephalopgraphy. In: Aminoff MJ, ed. Electodiagnosis in Clinial Neurology. New York: Churchill Livingstone: 77-124. Funding: No funding