Objectives: This study developed and validated a clinical scoring tool for predicting epilepsy development in pediatric patients with first-onset afebrile seizures using easily accessible clinical and laboratory parameters. Methods: We conducted a retrospective, multicenter study involving pediatric patients aged 1 month to 18 years who presented to the EDs of 3 university hospitals in Korea with first-onset afebrile seizures between March 2018 and March 2021. Stepwise multivariable logistic regression analysis was performed to identify predictors of epilepsy. A point-based risk score was derived from the regression coefficients, and the performance of the prediction model was evaluated using a receiver operating characteristic (ROC) curve. Results: In total, 328 children were included, of whom 132 (40.2%) developed epilepsy. Five variables remained significant in the final multivariable model: age group at onset, 2 or more seizures within 24 hours, lactate >2.27 mg/dL, total calcium <9.25 mg/dL, and abnormal brain imaging findings. The area under the ROC curve (AUROC) was 0.813 95% confidence interval (CI): 0.763-0.859, and the mean AUROC from 5-fold cross-validation was 0.810 (95% CI: 0.760-0.857). Conclusions: This clinical scoring tool may help stratify epilepsy risk in children with first-onset afebrile seizures in the ED. If prospectively validated, it could help identify high-risk patients early for appropriate referral and follow-up without requiring specialized equipment such as electroencephalography, thereby supporting clinical decision-making.
Choi et al. (2026) studied this question.