Neonatal Seizures: Evaluation of Current Classification Systems

Aim: Reliable classification of neonatal seizures remains controversial, but it is a key tool for research and clinical purposes. Currently used classification systems include the one proposed by Volpe (1989; 2001) with an emphasis on clinical criteria, the one proposed by Mizrahi and Kellaway (1987, 1998) including electrographic seizures and the one suggested in the ILAE revised seizure terminology by Berg et al (2010) as part of an overall seizure classification system not considering specific neonatal seizure patterns. In this study, we evaluate these classification systems on the basis of a retrospective analysis.

Methods: Neonates treated at Great Ormond Street Hospital, London, United Kingdom, a tertiary medical center, with seizures captured on video-electroencephalography (EEG) were included if recorded before 28 days of age in term infants or before 44 weeks postconceptional age in premature infants. Recordings were assessed for seizure frequency, duration, and grouped into the following: (1) electroclinical, (2) electrographic, and (3) clinical only. The three current classification systems were used to characterize seizures and their accuracy was assessed (where available demographic data were collected regarding comorbidities, current medications, and seizure etiology).

Results: We identified 85 neonates with seizures recorded on video-EEG from 2005 to 2014 (mean chronological age 15 days; range, 1-53 days) including 21 preterm infants (mean gestational age at recording 28 weeks; range, 30-43 weeks). Average seizure burden was 428 s/h. We identified 268 events consisting of 151 (56%) electrographic seizures, 80 (30%) electroclinical seizures, and 37 (14%) clinical-only seizures. Of the 56 newborns with electrographic seizures, 46 had only electrographic events. Mean duration was 114 seconds for electroclinical seizures and 129 seconds for electrographic seizures. In the Volpe classification, 32% of the events were fully classifiable and 18% were partly classifiable, in the Mizrahi and Kellaway classification 26% of the events were fully classifiable and 76% were partly classifiable, and in the ILAE classification 21% were fully classifiable and 18% were partly classifiable.

Conclusion: Neonatal seizures were often electrographic and showed a considerable duration, which supports the importance of video-EEG monitoring in neonatal intensive care. A large number of neonates presented without associated clinical seizures and would otherwise be missed. All three classification systems could fully classify only up to one-third of the seizures. Overall, the high proportion of electrographic seizures, the complex seizure semiology and the subtle seizure manifestations in neonates contributed to this result. Therefore, we feel the necessity of a distinct classification system in the neonatal period, which should be based on electroclinical phenotypes and ideally, reflect pathophysiologic origin.