Data from 6 British trials of phenobarbital and 4 trials of valproate for the prophylactic treatment of febrile convulsions were polled and analyzed on an intention to treat basis at the Dept of Neurology, Royal Manchester and Booth Hall Children's Hospitals, Manchester. The risk of recurrence in the treatment groups compared to controls expressed as an overall odds ratio was as follows: for phenobarbital, 66 of 296(22%) of treated children had recurrence compared to 58 of 236(25%) of controls (overall odds ratio or relative risk of 0.8, nonsignificant difference); for valproate, 49 of 145(34%) treated children had recurrence compared with 36 of 136(25%) controls (overall odds ratio of 1.42, nonsignificant result). The follow-up period ranged from 6 months to a mean of 30 months. An odds ratio of less than 1 suggests benefit; greater than 1 suggests no benefit from treatment. 
COMMENT. Pooled analysis of the British trials data failed to show any overall value in the prophylactic treatment of febrile convulsions with either phenobarbital of valproate. With side-effects reported in up to 40% of the treated group, continuous anticonvulsant therapy in the prophylaxis of simple febrile convulsions cannot be recommended. The same conclusion was reached in comparing the relative value of phenobarbital administered intermittently, at the time of fever, or continously in a group of 40 patients . Long-term phenobarbital was recommended only in children with complex febrile convulsions (those whose seizures are prolonged 20 min, complicated by EEG seizure discharges, or having neurological abnormalities). Alternative methods of treatment such as rectal diazepam are advised in those at risk of recurrence, and parents must be counselled in the first aid management of seizures.