The efficacy and side-effects of sodium valproate (VA) and carbamazepine (CBZ) were compared in 260 children (5 to 16 years of age) with primary generalized epilepsy or partial epilepsy followed for three years at 63 outpatient clinics in the UK and Ireland and results were reported from Addenbrooke’s Hospital, Cambridge, the Hospital for Sick Children, Great Ormond Street, London, and the Trial Office at Sanofi Winthrop Ltd, Guildford, Surrey, UK. Patients with newly diagnosed epilepsy and at least two seizures in the previous six months were randomized to receive either sodium valproate (200mg twice daily initially, mean maximum 17mg/kg daily) or carbamazepine (5mg/kg initially, mean maximum 10mg/kg daily). Doses were increased until seizures were controlled or toxicity ensued. VA or CBZ was stopped in 12% and 13%, respectively, because of poor seizure control, and in 15% and 12% because of adverse side-effects. Approximately 50% of patients were completely free of seizures for the first 6 months, and 75% had been free for 12 months and 50% for at least two years, by the end of the trial. Primary generalized seizures responded better than partial seizures. VA and CBZ were equally effective; a higher remission rate for VA treated patients was not statistically significant. Most frequent VA side-effects were appetite and weight increase (11%), somnolence (10%), and alopecia (4%). CBZ cf VA caused a higher incidence of somnolence (20% v 10%), diplopia (4% v 0%), ataxia (4% v 0%), and rash (6% v 3%). 
COMMENT. The incidence and type of side-effects are the main determinants for the choice of anticonvulsant in a particular patient. As with previous reports of comparative trials of phenobarbital, phenytoin, carbamazepine, and sodium valproate in both adults and children, there were no significant differences in efficacy between drugs regardless of seizure type, generalized tonic-clonic or partial.
However, side-effects with VA and CBZ were significantly different. Weight gain was particularly troublesome with VA, while somnolence, dizziness, and ataxia required modification of dosage of CBZ. Severe rash noted in 10% of patients taking CBZ in previous adult studies necessitated drug withdrawal in only 3% of children in the present report. Carbamazepine-induced skin rash was reported in 10% of 335 children treated at Toyama Medical University, Japan, and additional reports are cited in Progress in Pediatric Neurology II, PNB Publ, 1994, ppl07-109.