Early prediction of refractory epilepsy in childhood

PURPOSE To investigate early predictors (6 months after diagnosis) of refractory epilepsy. METHODS STUDY DESIGN prospective cohort study. INCLUSION CRITERIA all consecutive children <14 years with two or more unprovoked seizures 24h apart, who were seen at our hospital between 1994 and 2004. EXCLUSION CRITERIA patients previously examined in other centres. DEFINITIONS refractory epilepsy: failure of >2 drugs plus >1 seizure/month for >or=18 months. ANALYSIS risk of developing refractory epilepsy was calculated using Kaplan-Meier survival curves. Univariable and multivariable analyses of potential predictors of developing refractory epilepsy were carried out using Cox proportional hazards model. RESULTS 343 patients were included. Mean age at diagnosis was 4.8 years (+/-3.8 SD). Mean follow-up period was 76.2 (+/-35.2 SD) months (range 24-139). Risk of developing refractory epilepsy was 8% at 6 years. Risk for idiopathic syndromes was 2%. For non-idiopathic syndromes the risk was 38% for patients with age at onset <1 year plus >1 seizure during the first 6 months after diagnosis, 9% for age at onset <1 year plus 0-1 seizures during the first 6 months, 22% for age at onset >or=1 year plus >1 seizures during the first 6 months and 3% for age at onset >or=1 year plus 0-1 seizures during the first 6 months. CONCLUSION Risk of developing refractory epilepsy is very low in idiopathic syndromes. For the rest of patients, a simple model comprising three variables allows more accurate prediction of risk of refractoriness.