A Novel Risk Score for Seizures After Stroke

HOUSTON – Researchers have developed a novel scoring system that may help identify patients at risk of post-stroke seizures. Findings were presented at the 2016 American Epilepsy Society Annual Meeting.

The latent period between stroke and late seizures (>7 days) presents an opportunity for diagnostic and preventive care in patients who are at risk for post-stroke epileptogenesis.

Using backward elimination of a multivariate logistic regression model, Marian Galovic, MD, of University College London, United Kingdom, and colleagues developed a prognostic model for late seizures following ischemic stroke. The SeLECT score was derived from data on 1200 patients with ischemic stroke from the Stroke and Epilepsy Registry of Eastern Switzerland. The score was then validated in a cohort of 1427 patients through concordance statistic (c), calibration plots, and the Hosmer-Lemeshow calibration test.

Of the total cohort (n= 2627), 5% of patients experienced late seizures; 3% of those patients experienced recurrent seizures. A SeLECT value of 0 was associated with a 5-fold decrease in risk of late seizure, while a SeLECT score of 10, the highest, was associated with a 12-fold increase in relative risk for late seizures. Validation analyses showed a c-statistic of 0.76, and a good fit for predicted and observed outcomes based on calibration plots and nonsignificant Hosmer-Lemeshow tests (P> .3).

“The SeLECT score has the potential to individualize clinical care by identifying those who would benefit from optimal management and appropriate follow-up,” the authors wrote. Additionally, the score may help identify patient populations better suited for inclusion in antiepileptic drug treatment trials and biomarker research.

Click here for more coverage of AES 2016.

Reference

Galovic M, Döhler N, Conrad J, et al. The SeLECT score: A novel tool to predict seizures after ischemic stroke. Presented at: 2016 American Epilepsy Society Annual Meeting; December 2-6, 2016; Houston, TX. Abstract 1.088.