Acute symptomatic seizures due to stroke are closely related to the location and severity of brain damage; therefore, a causal relationship should be inferred if the brain lesion causing the seizures is clearly identified and the seizure occurs with close temporal continuity (8). Some previous studies set 7 d after stroke as the threshold for distinguishing between early and late seizures, whereas others determined 14 d after stroke as the threshold (Table 1). The International League against Epilepsy (ILAE) classifies poststroke seizures into early and late seizures using a 7-d threshold; under this classification, seizures occurring within and after 7 d of stroke onset are classified as early and late seizures, respectively (Table 1) (16). Accordingly, the incidences of early and late seizures were 3–6 and 12%, respectively (9, 17, 18). Early seizures are characterized by increased inflammatory responses, changes in neuronal signaling related to protein synthesis, and increased excitatory neurotransmitter (glutamate) release, ionic imbalance, blood–brain barrier (BBB) permeability, breakdown of membrane phospholipids, release of free fatty acids, and oxidative stress (19, 20). Therefore, metabolic disturbances, such as electrolyte imbalances, acid–base disturbances, and glucose instability, may develop; however, most cases are transient and reversible (21). Late seizures are characterized by irreversible changes, such as gliosis, selective neuronal loss, chronic inflammation, angiogenesis, neurodegeneration, collateral synaptic sprouting, and synaptic plasticity (19). Thus, early seizures follow the course of acute symptomatic seizures, whereas late seizures follow the course of unprovoked seizures (5, 22, 23). Moreover, mortality and disability rates were higher in late seizures than in early seizures. In clinical practice, an evidence-based approach showed that the criteria for a situation that could stipulate the seizure recurrence risk as ≥60% (24). Classifying poststroke seizures into the early and late types is useful for distinguishing acute symptomatic and unprovoked seizures. According to the European guidelines, a poststroke seizure occurring after 1 week on stroke onset is considered a late seizure (i.e., an unprovoked seizure) (25).

The prevalence of acute symptomatic seizures with intracranial hemorrhage is higher than that of ischemic stroke (10–16% vs. 2–4%) (27, 28). Ischemic strokes with hemorrhagic transformation have a higher seizure risk compared to ischemic strokes alone (29).

There is an increased incidence of poststroke seizures in cases with cortical involvement, total anterior circulation infarction, severe stroke with larger lesions, and functional deficits (25, 30). Reperfusion injury can manifest as blood brain barrier disruption, cortical irritation, and epileptic seizures (31). In late poststroke seizures, upregulation in endostatin and NCAM, and downregulation in S100B, Hsc70, and TNF-R1 in acute phase blood samples of stroke showed a high correlation (32). However the seizure risk for cerebral venous thrombosis is debatable, as some studies suggest an increase of up to 34%, whereas other studies indicate no association (33–35).

Acute symptomatic or provoked seizures can be induced by direct and immediate causes (27), including metabolic conditions, central nervous system (CNS) infections, sepsis, trauma, drugs, and alcohol consumption (36).

Determining the causes of acute symptomatic seizures other than stroke lesions is important. Acute symptomatic seizures due to metabolic disturbances are associated with the metabolic condition’s rate of deterioration; the faster the deterioration rate, the higher the risk of seizure (37). Metabolic conditions are usually determined by electrolyte tests conducted within 24 h of seizure. However, no absolute cutoff values for seizure prediction have been established, and only few studies have proposed cutoff values as references (Table 2) (27, 36, 37). Seizures in metabolic disturbances with an electrolyte value below the cutoff should be classified into an unknown category and follow-up should be conducted (8). Acute symptomatic seizures due to CNS infection can be considered acute symptomatic seizures even after 7 d, depending on the clinical course or laboratory findings, as the criteria remain unclear (8). Sepsis can induce encephalopathy, which leads to convulsive or nonconvulsive seizures by triggering the electric circuits that promote seizures (42). Alcohol-withdrawal acute symptomatic seizures should be considered in a patient with a history of excessive alcohol use who develops generalized tonic–clonic seizures after alcohol abstinence for 7–48 h (8). As alcohol-induced acute symptomatic seizures may occur, it is very rare and should exclude other etiologies such as metabolic disturbance, trauma, and drug abuse (36). Drug-related acute symptomatic seizures may occur following the administration of meperidine, methaqualone, glutarimide, theophylline, isoniazid, imipenem, cefepime, and chlorpromazine (36, 41). Furthermore, breakthrough seizures can develop when drugs, such as barbiturates and benzodiazepines, are discontinued (41). If any of these factors are determined, correction and treatment are essential.

Poststroke seizures primarily manifest localization-related seizure semiology, depending on the location of the brain lesion. One-third of all seizures are generalized tonic–clonic seizures (GTCS), whereas two-thirds present as focal seizures, with status epilepticus observed in 9% of cases (47, 48). Focal seizures are common in early seizures, whereas generalized seizures are common in late seizures (47). In patients with ischemic stroke due to large vessel occlusion, seizures occurring within 24 h were predominantly focal seizures or GTCS, whereas seizures with impaired awareness were more common after 24 h (49).

Video-EEG monitoring of all patients with stroke is practically impossible because of time and cost limitations and legislations in different countries. Epileptic seizures are commonly missed by the witness, as most poststroke seizures end within 5 min. Consequently, history taking is the simplest and most vital step in diagnosing poststroke seizures. Although clinicians should carefully listen to patients’ subjective complaints, they should also be familiar with seizure semiology and auras. During interviews, clinicians should ask relevant questions to ensure a correct diagnosis because many patients cannot describe their symptoms concretely and objectively. To ensure an accurate diagnosis, history taking should be divided into preictal, ictal (seizure), and postictal phases. In the preictal phase, determining the various auras of the patient in different environments is important. For example, symptoms appearing while eating, talking, walking, or waking up at night to use the bathroom may be helpful in differential diagnoses. Regarding aura history, confirming the presence of a specific and detailed aura, such as an epigastric rising sensation, hallucinatory taste or smell, ambiguous feelings of fear, anxiety, familiarity, or panorama-like scenes, can help diagnose seizures more accurately than nonspecific symptoms such as headache, dizziness, and nausea (50).

However, the symptoms that develop in the ictal phase cannot be described clearly unless the witness is a clinician. It is therefore crucial to determine the patient’s ability to communicate during a seizure, head version, uneven pupils, patient’s posture, and whether the seizure is tonic or flaccid. Furthermore, abnormal movements, tremulous or myoclonic, such as repetitive, regular muscle contractions, should be determined to ensure accurate diagnosis. In the postictal phase, investigating the presence of confusion, one-sided weakness (Todd’s paresis), or dysphagia is crucial to enable localization or lateralization (50).

Although clinicians may witness poststroke seizures directly in some cases, many cases rely on witness statements. Hence, semiological assessment is important for determining epileptic seizures. Moreover, even if a clinician witnesses the seizure, diagnosis may be difficult in cases with nonconvulsive seizures, unusual seizure semiology, or seizure-related sensory symptoms. In such cases, additional tests, such as EEG, magnetic resonance imaging, and laboratory tests, can aid in the diagnosis. In addition, it is necessary to check for an altered mental state, corporeal localization or lateralization, and somatosensory symptoms and autonomic changes and analyze seizure-like activities (51). If the semiology does not indicate a tangible expression of epileptic seizures, another disease should be suspected.

EEG can aid in the effective evaluation of the brain condition of patients, as it provides real-time and dynamic information on brain function in a simple and noninvasive manner. It effectively distinguishes non-convulsive seizures from autonomic syncope, movement disorders, and pseudo-seizures. EEG has a high sensitivity for the immediate detection of brain ischemia and particularly helpful in cases of large acute infarct volumes (52, 53). In cerebral ischemia, the amplitude and frequency of the EEG waves decrease (54). Acute cerebral infarction may show lateralized or focal slowing patterns on EEG (54). Continuous polymorphic delta with suppressed alpha or beta activity observed in the ischemic hemisphere indicates a poor prognosis (54). Good prognosis is indicated with only delta-to-theta slowing, with no or minor slow activity, and a relatively well-maintained background frequency are observed (54). A recent prospective study investigated whether early (within the initial 72 h) EEG abnormalities could predict poststroke epilepsy during the first year after stroke; the study reported that background asymmetry and interictal epileptiform discharge were independent predictors (55). Another study suggested that there was a 3.2-fold increase in the risk of unprovoked seizures when background asymmetry was observed in the first EEG poststroke and a 3.8-fold increase when interictal epileptiform discharge was observed (55).

Antiseizure medications should be administered when clinical seizures are observed in poststroke patients. However, ASM administration in asymptomatic patients with abnormal EEG findings (e.g., sharp waves or sharply contoured lateralized periodic discharge [LPD]) is debatable in critically ill patients of stroke. The ictal-interictal continuum pattern on EEG can be used as a reference for clinical management, as it suggests possible electrographic seizures (Table 4) (56). A brief, potentially ictal, rhythmic discharge pattern on EEG indicates a seizure onset zone; critically ill patients with this pattern may develop refractory seizures (Table 4) (57). In cases where the EEG pattern satisfies the electrographic seizure or ictal-interictal continuum criteria according to the 2021 American Clinical Neurophysiology Society guidelines, ASM should be considered (Table 4) (56). Furthermore, time-locked LPD is clinically correlated with focal motor jerks, strongly suggesting that epileptic seizures require ASM administration. However, if the discharge frequency does not satisfy the ictal-interictal continuum (<1 Hz) criterion and the shape is less sharply contoured (>200 ms) with LPD, ASM prescription may be suspended. This is due to the possibility of a bystander epiphenomenon, resulting from brain injury without an ongoing insult resulting from a simple structural lesion (58, 59). Generally, one unprovoked seizure with EEG showing prominent epileptiform discharge should be considered an epilepsy based on a seizure recurrence rate ≥ 60% (59, 60). However, applying this in poststroke seizures is difficult because a spike or sharp wave may be observed on EEG due to an epiphenomenon induced by a current acute structural lesion during stroke. In this case, EEG changes induced by the structural lesion (epiphenomenon) disappeared over time when the stroke was properly managed. The 2017 European guidelines do not overlook paroxysmal EEG phenomena as a clinical basis for predicting the recurrence of poststroke seizures (25).

Patients with stroke may show impaired autonomic function, such as a high incidence of orthostatic hypotension, arrythmias, and syncope, owing to their decreased ability to maintain cerebral blood flow. Although the precise underlying mechanisms remains unknown, several studies have proposed possible mechanisms. For example, damage in ischemic stroke can affect nuclear tractus solitarius signaling, resulting in a constant sympathetic activity that increases the resistance of adrenergic beta receptors to stimulation (30). Meanwhile, other studies have postulated that impaired autonomic function occurs when cerebral perfusion decreases in response to a significant decrease in central blood pressure, potentially due to the withdrawal of excessive sympathetic tone (25). A heart rate variability test was conducted to check for reduced cardiac baroreceptor reflex sensitivity and vagal inhibitory outflow (50). Because autonomic dysfunction symptoms may resemble epileptic seizures, unnecessary ASM use should be avoided in the differential diagnosis.

Abnormal poststroke movements are not necessarily the manifestations in epileptic seizures. Therefore, understanding semiology is crucial. Poststroke movement disorder (PSMD) is a poststroke abnormal movement that affects 1–4% of all stroke patients (55). A study analyzing 284 published cases revealed that ischemic stroke accounted for 75% of all PSMD cases. Common sites of the stroke lesions included the posterolateral thalamus (23%), putamen (19%), and caudate nucleus (14%) (61). Because approximately 46% of PSMD cases develop within 7 d poststroke, they overlap with early seizures; therefore, differential diagnosis is vital (61). PSMD after an ischemic stroke can appear quickly, usually within 1 month (61, 62). The frequencies of abnormal movements were as follows: dystonia (23%), chorea (16%), and myoclonus (15%) (61, 62). Chorea and myoclonus often occur within 7 d (61, 62). Movement disorders after a hemorrhagic stroke, commonly including dystonia and tremor, appear more frequently after 6 months (61, 62). In PSMD, owing to stroke lesions, the onset time of caudate lesions is relatively long (approximately 6 months). When the lesion is located in the posterolateral thalamus or putamen, PSMD commonly occurs within 7 d (63); therefore, differentiating it from early seizures is important (63). Because 84% of myoclonus cases induced by PSMD improve naturally when appropriately diagnosed, unnecessary use of ASM can be avoided (61).