Prakash 2019.

Study characteristics
Methods	Randomised controlled trial, not blinded. A block randomisation model (blocks of 4) was used with sealed envelopes. The study was conducted at a tertiary care neonatal intensive care unit at Bihar, India, between April 2018 and September 2019.
Participants	80 newborns with clinically apparent seizures, after acute metabolic disorders were ruled out. Inclusion was based on appearance of motor or autonomic phenomena suggestive of seizures. Exclusion criteria were prematurity, major congenital malformation, intubation at time of admission, newborns presenting with myoclonic jerks. Demographic data at baseline comparable
Interventions	Group A (n = 42) LEV loading 10 mg/kg IV, if seizures persisted additionally 5 mg/kg. Maintenance with dosage that had proved to control seizures. If seizures persisted, 'cross‐over' to PB. Group B (n = 38): PB loading 20 mg/kg IV, if seizures persisted, additional 10 mg in aliquots up to a maximum dosage of 40 mg/kg. Maintenance after 24 h. If seizures persisted, 'cross‐over' to LEV. Third‐line drug midazolam 0.2 mg/kg/dose followed by continuous infusion
Outcomes	Primary outcome variable: cessation of clinical seizure activity for 5 d Secondary outcome variables: time to control seizures, survival at discharge, short‐term adverse effects, neurodevelopmental outcome at 12 m, EEG control of seizures
Notes	Cessation of clinical seizures not different between groups. Adverse events (cardiorespiratory depression, sedation) in 3/42 in the LEV group vs 20/38 in the PB group. EEG only after clinical seizures were controlled, not different between groups. Neuromotor developmental delay, 'mental retardation' and comorbidities more frequent in PB group.