Eleven studies (2,159 participants) on phenobarbital for febrile convulsions were included in the review. These were simple and stratified trials and two studies described as non-random and "not trial". Five trials used blinded assessments. Concealment of allocation was reported in four trials. Follow-up ranged between 12 to 53 months. Losses to follow-up ranged from zero to 34%. Intention-to-treat analyses were performed in three trials.
There was a lack of evidence of efficacy with phenobarbital in the prevention of febrile convulsions across the 11 trials. Benefits were observed in only five trials compared to placebo (three trials), intermittently administered phenobarbital (one trial) and carbamazepine (one trial). Trials showed a lack of conclusive evidence of more adverse behavioural effects associated with phenobarbital.
Nine trials (956 participants, range eight to 302) that evaluated phenobarbital in children with epilepsy were included in the review. Follow-up ranged from three months to 44 months. Concealment of allocation and blinded analysis of the results were clearly reported in one trial. Losses to follow-up (where stated) ranged from zero to 34%. Trial quality was judged as being poor.
Phenobarbital was associated with an increase in adverse side effects in three non-blinded studies compared to phenytoin and placebo (one trial) sodium valproate (one trial) and phenytoin and sodium valproate (one trial). There were no differences in side effects observed with phenobarbital and other comparators in six trials, all of which were undertaken in developing countries.