Poststroke epilepsy incidence, risk factors and treatment: German claims analysis

Development and validation of an interpretable machine learning model for predicting post-stroke epilepsy

BACKGROUND

Epilepsy is a serious complication after an ischemic stroke. Although two studies have developed prediction model for post-stroke epilepsy (PSE), their accuracy remains insufficient, and their applicability to different populations is uncertain. With the rapid advancement of computer technology, machine learning (ML) offers new opportunities for creating more accurate prediction models. However, the potential of ML in predicting PSE is still not well understood. The purpose of this study was to develop prediction models for PSE among ischemic stroke patients.

METHODS

Patients with ischemic stroke from two stroke centers were included in this retrospective cohort study. At the baseline level, 33 input variables were considered candidate features. The 2-year PSE prediction models in the derivation cohort were built using six ML algorithms. The predictive performance of these machine learning models required further appraisal and comparison with the reference model using the conventional triage classification information. The Shapley additive explanation (SHAP), based on fair profit allocation among many stakeholders according to their contributions, is used to interpret the predicted outcomes of the naive Bayes (NB) model.

RESULTS

A total of 1977 patients were included to build the predictive model for PSE. The Boruta method identified NIHSS score, hospital length of stay, D-dimer level, and cortical involvement as the optimal features, with the receiver operating characteristic curves ranging from 0.709 to 0.849. An additional 870 patients were used to validate the ML and reference models. The NB model achieved the best performance among the PSE prediction models with an area under the receiver operating curve of 0.757. At the 20 % absolute risk threshold, the NB model also provided a sensitivity of 0.739 and a specificity of 0.720. The reference model had poor sensitivities of only 0.15 despite achieving a helpful AUC of 0.732. Furthermore, the SHAP method analysis demonstrated that a higher NIHSS score, longer hospital length of stay, higher D-dimer level, and cortical involvement were positive predictors of epilepsy after ischemic stroke.

CONCLUSIONS

Our study confirmed the feasibility of applying the ML method to use easy-to-obtain variables for accurate prediction of PSE and provided improved strategies and effective resource allocation for high-risk patients. In addition, the SHAP method could improve model transparency and make it easier for clinicians to grasp the prediction model's reliability.

Risk Factors and Temporal Patterns of Poststroke Epilepsy across Stroke Subtypes: Insights from a Nationwide Cohort Study in Korea

INTRODUCTION

We aimed to investigate the risk factors associated with poststroke epilepsy (PSE) among patients with different subtypes of stroke, focusing on age-related risk and time-varying effects of stroke subtypes on PSE development.

METHODS

A retrospective, nationwide, population-based cohort study was conducted using Korean National Health Insurance Service-National Sample Cohort data. Patients hospitalized with newly diagnosed stroke from 2005 to 2015 were included and followed up for up to 10 years. The primary outcome was the development of PSE, defined as having a diagnostic code and a prescription for anti-seizure medication. Multivariable Cox proportional hazard models were used to estimate PSE hazard ratios (HRs), and time-varying effects were also assessed.

RESULTS

A total of 8,305 patients with ischemic stroke, 1,563 with intracerebral hemorrhage (ICH), and 931 with subarachnoid hemorrhage (SAH) were included. During 10 years of follow-up, 4.6% of patients developed PSE. Among patients with ischemic stroke, significant risk factors for PSE were younger age (HR = 1.47), living in rural areas (HR = 1.35), admission through the emergency room (HR = 1.33), and longer duration of hospital stay (HR = 1.45). Time-varying analysis revealed elevated HRs for ICH and SAH, particularly in the first 2 years following the stroke. The age-specific HRs also showed an increased risk for those under the age of 65, with a noticeable decrease in risk beyond that age.

CONCLUSION

The risk of developing PSE varies according to stroke subtype, age, and other demographic factors. These findings underscore the importance of tailored poststroke monitoring and management strategies to mitigate the risk of PSE.

Post-stroke epilepsy: From clinical predictors to possible mechanisms

BACKGROUND

Stroke is the most common cause of newly diagnosed epilepsy in the elderly, ahead of degenerative disorders, brain tumors, and head trauma. Stroke accounts for 30-50% of unprovoked seizures in patients aged ≥ 60 years. This review discusses the current understanding of epidemiology, risk factors, mechanisms, prevention, and treatment opportunities for post-stroke epilepsy (PSE).

METHODS

We performed a literature search in the PubMed and Cochrane Library databases. The keywords "stroke, epilepsy", "stroke, seizure", "post-stroke seizure", "post-stroke epilepsy" were used to identify the clinical and experimental articles on PSE. All resulting titles and abstracts were evaluated, and any relevant article was considered. The reference lists of all selected papers and reference lists of selected review papers were manually analyzed to find other potentially eligible articles.

RESULTS

PSE occurs in about 6% of stroke patients within several years after the event. The main risk factors are cortical lesion, initial stroke severity, young age and seizures in acute stroke period (early seizures, ES). Other risk factors, such as a cardioembolic mechanism or circulation territory involvement, remain debated. The role of ES as a risk factor of PSE could be underestimated especially in young age. Mechanism of epileptogenesis may involve gliosis scarring, alteration in synaptic plasticity, etc.; and ES may enhance these processes. Statins especially in the acute period of stroke are possible agents for PSE prevention presumably due to their anticonvulsant and neuroprotection effects. Antiepileptic drugs (AED) monotherapy is enough for seizure prevention in most cases of PSE; but no evidence was found for its efficiency against epileptic foci formation. The growing interest in PSE has led to a notable increase in the number of published articles each year. To aid in navigating this expanding body of literature, several tables are included in the manuscript.

CONCLUSION

Further studies are needed for better understanding of the pathophysiology of PSE and searching the prevention strategies.

Post-stroke seizures, epilepsy, and mortality in a prospective hospital-based study

Background and objectives

Post-stroke epilepsy (PSE) is a significant concern in the elderly population, with stroke being a leading cause of epilepsy in this demographic. Several factors have shown consistent associations with the risk of developing PSE, including cortical lesions, initial stroke severity, younger age, and the occurrence of early seizures. The primary objectives of this study were two-fold: (1) to determine the incidence of PSE and (2) to identify the risk factors associated with PSE in a prospective cohort of post-stroke patients.

Methods

A prospective single-hospital study was conducted, involving patients diagnosed with acute ischemic and hemorrhagic stroke. The patients were followed up for 2 years (or until death) from the time of admission. Data about seizure occurrence and recurrent stroke were collected. Kaplan-Meyer curves were used for the assessment of PSE incidence and mortality. Possible predictors of PSE and mortality were selected from between-group analysis and tested in multivariable regressions.

Results

Our study enrolled a total of 424 patients diagnosed with acute stroke. Among them, 97 cases (23%) experienced early post-stroke seizures, and 28 patients (6.6%) developed PSE. The cumulative risks of developing PSE were found to be 15.4% after hemorrhagic stroke and 8.7% after ischemic stroke. In multivariable fine and gray regression with competitive risk of death, significant predictors for developing PSE in the ischemic cohort were watershed infarction (HR 6.01, 95% CI 2.29-15.77, p < 0.001) and low Barthel index at discharge (HR 0.98, CI 0.96-0.99, p = 0.04). Furthermore, patients who eventually developed PSE showed slower recovery and presented a worse neurologic status at the time of discharge. The in-hospital dynamics of the National Institutes of Health Stroke Scale (NIHSS) were significantly worse in the PSE group compared to the non-PSE group (p = 0.01).

Discussion

A higher proportion of cases experienced early seizures compared to what has been commonly reported in similar studies. Watershed stroke and low Barthel index at discharge were both identified as independent risk factors of PSE in ischemic strokes, which sheds light on the underlying mechanisms that may predispose individuals to post-stroke epilepsy after experiencing an ischemic stroke.

Post-stroke epilepsy and risk of all-cause mortality: A systematic review and meta-analysis of cohort studies

OBJECTIVE

Post-stroke epilepsy (PSE) has been suggested as a predictor of poor prognosis. We aimed to evaluate the association between PSE and all-cause mortality via a meta-analysis.

METHODS

Relevant cohort studies were identified by search of electronic databases including PubMed, Embase, and Web of Science. Two authors independently performed database search, data collection, and study quality evaluation. A randomized-effect model incorporating the possible between-study heterogeneity was used to pool the results.

RESULTS

Overall, ten cohort studies including 177,929 patients with stroke contributed to the meta-analysis. Of them, 15,836 (8.9%) had PSE during a mean follow-up of 4 years. Compared to stroke survivors without PSE, patients with PSE had a significantly increased risk of all-cause death (odds ratio [OR]: 1.68, 95% confidence interval: 1.51-1.87, p < 0.001; I² = 80%). Subgroup analyses showed consistent association in patients with ischemic and hemorrhagic stroke, in prospective and retrospective studies, in PSE diagnosed as unprovoked seizure occurred more than 7 days and 14 days after stroke, in studies with follow-up durations within or over 5 years, and in studies with different quality scores (p for subgroup effect all < 0.05). The Galbraith plot identified three outlier studies which may contribute to the heterogeneity of the meta-analysis, and all of the three studies were database-linked studies. The between-study heterogeneity was significantly reduced after removing the three database-linked studies (OR: 1.39, 95% CI: 1.18-1.62, p < 0.001; I² = 0%).

CONCLUSION

PSE was associated with a higher risk of all-cause mortality.

Risk Factors for Postischemic Stroke Epilepsy in Young Adults: A Nationwide Population-Based Study in Taiwan

Background

The incidence of ischemic stroke has been increasing in the young population over the past 20 years. Poststroke epilepsy (PSE) is a common complication after stroke. However, few population-based studies with sufficient follow-up have investigated factors associated with PSE, especially factors related to comorbidities and unhealthy lifestyles in the modern young population. Accordingly, this study aimed to determine the long-term incidence and these risk factors for PSE young adults.

Methods

This cohort study was conducted using data from the Taiwan National Health Insurance Research Database (NHIRD) from 2002 to 2018. All patients aged between 19 and 44 years and diagnosed with ischemic stroke from 2002 to 2015 were retrospectively enrolled with a follow-up of at least 3 years. Multivariable Cox regression models were performed to identify predictors of PSE, including patients' demographics, baseline conditions, stroke severity, etiologies, comorbidities, and unhealthy behaviors.

Results

Among 6,512 ischemic stroke patients, 402 cases (6.2%) developed PSE who were with a mean follow-up period of 8.3 years (SD = 4.3 years). During the overall follow-up, stroke severity and manifestations were associated with PSE, including National Institutes of Health Stroke Scale (NIHSS) score ≥10 (aHR, 1.98; 95% CI, 1.50–2.61), seizure at first stroke admission [adjusted hazard ratio (aHR), 57.39; 95% confidence interval (CI), 43.02–76.55], length of hospital stay ≥14 days (aHR, 1.60; 95% CI, 1.26–2.02), recurrent stroke (aHR, 2.32; 95% CI, 1.85–2.90), aphasia (aHR, 1.77; 95% CI, 1.20–2.60), and malignancy (aHR, 2.05; 95% CI, 1.30–3.24). Furthermore, stroke patients with drug abuse were 2.90 times more likely to develop PSE than those without (aHR, 2.90; 95% CI, 1.53–5.50). By contrast, statin use (aHR, 0.62; 95% CI, 0.48–0.80) was associated with a lower risk of PSE. The risk factors at 1-year and 5-year PSE were similar to that of an overall follow-up.

Conclusions

Stroke severity, aphasia, malignancy, and drug abuse were associated increased risk of PSE and statin use may protect against PSE in young adults. Reducing the severity of stroke, statin use and controlling unhealthy behaviors might be able to decrease the development of PSE. Since PSE is associated with poor outcomes, early identification or intervention of PSE based on the risk factors might reduce the harmful effects of PSE.