Risk Stratification for Early Seizures after Traumatic Brain Injury: A Clinical Scoring Model for Prediction and...

Early seizures within 1 week after traumatic brain injury (TBI) are classified as acute symptomatic seizures, which can worsen secondary brain injury, leading to poor neurological outcomes and increased mortality. Despite prophylactic antiepileptic drug (AED) use, early seizures remain prevalent, particularly in severe cases. This study aims to identify risk factors for early seizures in patients with TBI receiving prophylactic levetiracetam monotherapy and to develop a clinical scoring model for risk stratification. This study was conducted at an emergency medical center in Japan (2018-2022). In total, 168 patients with TBI were screened, and 147 severe cases met eligibility criteria. Exclusion criteria included age under 18, prehospital cardiac arrest, prior AED use, pregnancy, and alcohol dependency. All patients received prophylactic levetiracetam, with dosage and duration determined by neurocritical care physicians. The primary outcome was early seizures and associated risk factors. Secondary outcomes included hospital stay, functional outcomes, and in-hospital mortality. Multivariate logistic regression identified independent risk factors, and a clinical risk scoring model was developed. Early seizures occurred in 25 of the 147 patients. Multivariate analysis identified heart rate (odds ratio [OR]: 1.048, p < 0.001), creatinine level (OR: 1.535, p = 0.026), and acute subdural hematoma (OR: 5.861, p = 0.027) as independent risk factors. A clinical risk scoring model incorporating these variables demonstrated high predictive accuracy (area under the receiver operating characteristic curve: 0.828, 95% confidence interval: 0.725-0.931). Patients scoring >4.5 underwent electroencephalographic monitoring. This study presents a novel risk scoring model integrating heart rate, creatinine, and acute subdural hematoma to predict early seizures in patients with TBI despite prophylactic levetiracetam use. The model enables rapid risk stratification upon admission and may guide targeted neurocritical management.