Outcomes and predictors of seizure recurrence in post-stroke epilepsy, a retrospective hospital-based study

Abstract

Background: The occurrence of seizures following a stroke is a well-recognized complication associated with a significant increase in morbidity and mortality. Despite the numerous studies examining outcomes and risk factors related to post-stroke seizures (PSS), there remains a lack of clarity regarding the clinical characteristics, treatment, and PSS recurrence (PSSR) rates in patients experiencing their initial episode of PSS.

Purpose: This study aimed to determine the risk factors for developing recurrent seizures after first PSS and their effects on functional outcomes and mortality.

Methods: All patients underwent an electroencephalography (EEG) and were monitored for a minimum of 24 months following the first PSS. The primary endpoint was the recurrence of seizures. Predictive factors for PSSR were determined by using the Cox-proportional hazards model, and the cumulative latency of recurrence at 90, 180, 360, and 720 days was estimated using Kaplan-Meier analysis.

Results: Seizure recurred in 36.8% (39/106). Significant association of PSSR was noted with female gender, use of older anti-seizure medications (ASMs) (p<0.001), EEG findings as focal slow wave activity (p<0.001), Ictal epileptiform abnormalities (p=0.015), status epilepticus (p=0.015), and with severe disability (p=0.008). However, multivariate cox-proportional hazards model showed significant association of female gender (HR=3.28; 95% CI: 1.42-7.58; p=0.006). Hazard ratio (HR) was increased with older ASMs use, focal aware seizure types, Ictal EAs, and periodic discharges on EEG; though, statistically significant.

Conclusion: Factors such as the type of ASMs, EEG findings, and seizure type were significantly linked to PSSR. Female gender was the only independent predictor established. Additionally, significant functional decline was reported with recurrence.

Keywords: Predictive factors; Recurrence; Stroke; Symptomatic seizure.