Development and validation of a predictive model assessing the risk of seizure recurrence in children with neurocysticercosis

Abstract

IntroductionNeurocysticercosis (NCC) is a significant factor contributing to the incidence of seizures in developing countries. While numerous studies have investigated the recurrence of seizures in NCC, their reliability is often limited. MethodsWe assessed the socio-demographic, clinical, and neuroimaging details of 161 children with seizures caused by NCC. We monitored them for seizure recurrence over a 6-month follow-up period. We divided the children into two groups: those with seizure recurrence and those without. Subsequently, we identified predictive factors associated with seizure recurrence through univariate analysis, followed by multivariate binary logistic regression. We evaluated the prognostic model for discrimination and calibration and then internally validated it using the bootstrap method. ResultsA total of 23 children experienced breakthrough seizures. In multivariate analysis, the presence of epileptiform abnormalities in electroencephalogram (EEG), more than 5 NCC lesions, the presence of perilesional edema greater than 2 cm in maximum dimension, and a cluster of seizures before presentation were significantly associated with seizure recurrence (p < 0.05). These factors were included in the final NEPC (Number of NCC lesions, Epileptiform EEG abnormality, Perilesional edema, and Cluster of seizures) score. The final model exhibited good discrimination (AUC = 89.1 %; 95 % CI=80.5–95.3 %, p < 0.001) and calibration (p = 0.54). A score of 4 appeared to be the optimal threshold for discriminating between individuals with and without seizure recurrence, with sensitivity and specificity values of 85 % and 87 %, respectively. Interrater reliability was very strong between two pediatric neurologists and strong between a pediatric neurologist and a pediatric neurology trainee resident (k = 0.92 and 0.86, respectively). ConclusionThe NEPC score demonstrates good sensitivity and specificity in predicting seizure recurrence in pediatric children with NCC.