OBJECTIVE: Identify risk factors and treatment strategies for neonatal seizures requiring midazolam infusion. 

DESIGN: A single center retrospective study was undertaken for patients admitted to the neonatal ICU with EEG-confirmed seizures over a four-year period. 

RESULTS: Of 67 patients with EEG-confirmed neonatal seizures, 23 required midazolam infusion to control seizures. Of these patients, five received one additional antiseizure medication, 8 received two, 9 received three, and 1 received four.  Patients requiring midazolam after two or more medications failed to control seizures had >20% rates of mortality, epilepsy and cerebral palsy. None of the patients receiving only one medication prior to midazolam developed epilepsy or cerebral palsy.

Of the 44 patients who did not receive midazolam infusion, 11 responded to monotherapy with phenobarbital or fosphenytoin, 19 required two medications, 9 required three, and 2 required four. Of those requiring ≥ 3 medications, over 63% developed epilepsy or cerebral palsy.

Groups (+/- midazolam) did not vary by gestation age, birth weight or APGAR scores.

The typical sequence of medications was phenobarbital, then levetiracetam, fosphenytoin, and topiramate. The sequence of medications used for patients who did not receive midazolam for seizure control was more variable which may reflect application of our institution’s neonatal seizure protocol preferentially to patients with status epilepticus.

CONCLUSIONS: Refractory neonatal seizures increase the risk of unfavorable developmental outcomes.  Birth characteristics are not predictive of seizure response to medications.  Use of two seizure medications prior to midazolam infusion, and three seizure medications in the group not requiring midazolam were associated with worse outcomes.