Post-stroke seizures: classification, risk prediction, and management

Stroke is the leading cause of adult-onset epilepsy. The International League Against Epilepsy (ILAE) classifies seizures occurring within 7 days of stroke as acute symptomatic seizures (ASyS); later seizures are classified as unprovoked/remote symptomatic. After a single remote symptomatic seizure, a diagnosis of post-stroke epilepsy (PSE) requires an estimated 10-year recurrence risk of at least 60%. We review three frequently conflated questions: how post-stroke seizures should be classified, how the risk of subsequent epilepsy should be estimated, and how patients should be managed. The 2010 ILAE Task Force introduced this definition for epidemiological use, acknowledged that its criteria were somewhat arbitrary, and identified lesion-based refinement in cerebrovascular disease as an unresolved objective; a dedicated ILAE working group is now revisiting the definition. Risk prediction has already progressed beyond a binary approach, largely by incorporating rather than discarding early seizure timing. Temporal classification and multivariable risk estimation are complementary. The principal gap concerns management and diagnostic consequences. Estimated post-stroke seizure risk spans approximately an order of magnitude, whereas the epilepsy label and its driving, occupational, and insurance consequences remain dichotomous. Antiseizure medication prescribing represents a related but distinct problem: treatment should already be individualised, yet available evidence suggests that prescribing is inconsistently calibrated to recurrence risk. Existing prediction models address neither mismatch as they were not designed to do so. We define the evidence required to support a graded framework, the properties such a framework must preserve, and the settings in which its applicability remains unproven.