Abstracts

About 50-60% of children with epilepsy eventually outgrow their seizure disorder. A number of predictive factors have been statistically associated with remission but it is unclear how accurate these factors are when applied to an individual child. Two large prospective cohort studies of childhood epilepsy (Nova Scotia and the Netherlands) each developed a statistical model to predict long-term outcome. We evaluated the accuracy of a prognostic model based on the two studies combined. A wealth of clinical and EEG variables were available for patients in both cohort studies. Data analyses with classification tree models and stepwise logistic regression produced predictive models for the combined dataset and the two separate cohorts. The resulting models were then externally validated on the opposite cohort. Remission was defined as no longer receiving daily medication for any length of time at the end of follow-up. The combined cohorts yielded 1055 evaluable patients. At the end of follow up ([ge]5 years in [gt]96%), 622 (59%) were in remission. Using the combined data, the classification tree model and the logistic regression model predicted the outcome (remission or no remission) correctly in approximately 70% (sensitivity [sim]72%, specificity[sim]65%, positive predictive value[sim]75%, negative predictive value [sim] 62%). The classification tree model split the data on epilepsy syndrome and age at first seizure. Independent statistically significant predictors in the logistic regression model were: seizure number before treatment, age at first seizure, absence seizures, epilepsy types of symptomatic generalized and symptomatic partial, pre-existing neurological signs, intelligence and the combination of febrile seizures and cryptogenic partial epilepsy. When the prediction models from each cohort were cross-validated on the opposite cohort, the outcome was predicted slightly less accurately than the model from the combined data. Based on currently available clinical and EEG variables, predicting the outcome of childhood epilepsy is difficult and appears to be incorrect in about one of every three patients. Predictions schemes are statistically robust but clinically relatively inaccurate. We suggest that clinicians should be cautious in applying prediction models when developing management strategies for individual children with epilepsy.