The Scale for Estimating Prognosis of Epilepsy to predict in patients with epilepsy: A 506-Patient Cohort Study

Objective This study aimed to evaluate the predictive performance of the Scale for Estimating Prognosis of Epilepsy (SEPE) in identifying drug-resistant epilepsy (DRE), using the 2010 International League Against Epilepsy (ILAE) definition as the reference standard. Methods A total of 506 hospitalized patients with epilepsy, including 57 who underwent epilepsy surgery, were enrolled. DRE was defined according to 2010 ILAE criteria, and the reproducibility of SEPE’s predictive risk factors was examined. SEPE scores were then assigned to all patients, and the concordance between SEPE-based predictions and ILAE-defined DRE diagnoses was assessed after one to two years of follow-up in the full cohort (n = 506) and in the subset excluding surgical cases (n = 449). Finally, distinct clinical patterns of DRE, including persistence and fluctuating forms, were identified, and the associations between SEPE scores and these subtypes were analyzed. Results Multivariate logistic regression identified seizures occurring across the sleep–wake cycle, multiple seizure types, and at least one seizure within two months as independent risk factors for DRE. The area under the ROC curve (AUC) for SEPE in identifying ILAE-defined DRE was 0.6827 (95% CI: 0.6322–0.7245) in the full cohort and 0.6812 (95% CI: 0.6302–0.7382) after excluding surgical cases. Among patients with SEPE scores ≥ 6, 91.2% met the ILAE criteria for DRE, with sensitivity of 63.9% and specificity of 96.2%, whereas 89.4% of those with scores ≤ 4 were drug-responsive, with sensitivity of 87.2% and specificity of 87.1%. For scores with 4 < SEPE < 6, 34% had DRE, 53% were drug-responsive, and 24.9% of patients with scores ≤ 3 achieved seizure freedom without antiseizure medications (ASMs). In patients with DRE defined by the ILAE criteria, SEPE scores ≥ 6 were associated with intractable DRE without remission, while scores ≤ 4 were frequently linked to seizure control following appropriate ASM adjustments. Conclusion The SEPE may serve as a useful tool for estimating the risk of DRE and for stratifying patients with epilepsy based on its constituent predictive factors. It may be particularly helpful for physicians—especially general neurologists who are not epilepsy specialists.