Background: Treatment of neonatal
seizures includes etiotropic and
anticonvulsant treatments. However,
anticonvulsant use in neonates is off-label and requires ethical review.Objective: To investigate the efficacy and safety of
levetiracetam for neonatal
seizures and to establish a predictive model.Methods: We retrospectively analyzed 125 neonatal seizure cases (
phenobarbital 66 cases,
levetiracetam 59 cases). The efficacy, safety and tolerability of
levetiracetam were evaluated by cox regression survival analysis and a regression tree prediction model for the 16-week time point.Results: There was no significant difference between
phenobarbital and
levetiracetam treatment group in short-term efficacy (p > 0.05). But the cumulative survival function suggested that
levetiracetam treatment group was better than
phenobarbital (p = 0.026) in long-term efficacy evaluation. Neurodevelopmental assessments at 16 weeks showed that
levetiracetam had better effect on the neurodevelopmental level (Gesell scores in response) than
phenobarbital (p = 0.011). The main adverse events with
levetiracetam were irritability and
anorexia. According to the regression tree prediction model, the top three factors influencing the
therapeutic effect were pre-treatment seizure frequency, age of onset and etiological classification.Conclusion:
Levetiracetam shows good efficacy, safety and tolerability for the long-term neonatal seizure treatment.