A Nomogram for Early Prediction of Post-Traumatic Epilepsy in Children With Traumatic Brain Injury: Development and Internal Validation.

Abstract

Post-traumatic epilepsy (PTE) is a significant complication after pediatric traumatic brain injury (TBI), yet early risk stratification remains challenging. We aimed to develop and temporally validate a prediction model for early risk stratification of PTE in children with TBI. A retrospective cohort of children with TBI (2018-2023) was analyzed. Patients from 2018-2021 formed the training cohort and those from 2022-2023 the validation cohort. A multivariable logistic regression model was constructed and translated into a nomogram. Model performance was evaluated using discrimination, calibration, and decision curve analysis. Exploratory analyses assessed factors associated with drug-resistant epilepsy (DRE). Among 579 patients (training: 402; validation: 177), PTE occurred in 13% and 14%, respectively. Five variables readily available at presentation (intraventricular hemorrhage, Glasgow Coma Scale score, early post-traumatic seizures, abnormal pupillary response, and albumin) independently predicted PTE and were incorporated into the nomogram. The model showed strong discrimination (AUC 0.942 training; 0.916 validation) and good calibration, with favorable clinical net benefit across relevant thresholds. Exploratory analyses identified potential associations between markers of severe intracranial injury and DRE. We present a practical admission-based nomogram for early risk stratification of PTE after pediatric TBI. The model may help prioritize neurological surveillance and follow-up in children at increased risk. External validation is required before routine clinical implementation.