Post-Stroke Epilepsy: Early vs Late Seizures, Prophylaxis, and Drug Choice
A day-2 convulsion on the stroke unit and a late seizure eight months after discharge are two different diseases. The 7-day line decides the workup, the treatment, and the counseling.
The 7-day line decides almost everything about a seizure after stroke.
- →A seizure within 7 days of the stroke is an acute symptomatic (early) seizure — a symptom of the injured brain, not epilepsy. A seizure after 7 days is an unprovoked (late) seizure, and its recurrence risk is high enough that a single one meets the ILAE definition of epilepsy.
- →Guidelines recommend against prophylactic antiseizure medication after ischemic stroke and after intracerebral hemorrhage. Treat seizures that happen; do not medicate seizures that might.
- →An early seizure usually earns a short course of medication at most. A late seizure usually earns long-term treatment — and honest counseling that the diagnosis is now epilepsy.
- →Risk of late seizures is predictable at the bedside: severity, cortical involvement, large-artery atherosclerotic etiology, MCA territory, and an early seizure — the variables of the SeLECT score.
- →Drug choice is a stroke-medicine decision, not just an epilepsy one: enzyme-inducing agents (carbamazepine, phenytoin) can undercut DOACs and statins. Levetiracetam and lamotrigine are the usual first choices — renally dosed where needed.
It is day 2 on the stroke unit. A 71-year-old with a large left MCA infarct has a witnessed convulsion — a minute of head turning and right arm jerking, then confusion. By the time you arrive she is waking up. Her daughter asks the question every family asks: "Does this mean she has epilepsy now?"
The honest answer is usually no — not yet, and probably not ever. But eight months later, if the same patient has a seizure at home while making tea, the honest answer becomes yes — after a single event. Those two seizures look identical on the surface. Clinically, they are different diseases. The line between them is drawn at seven days, and it decides the workup, the treatment, the counseling, and the driving conversation.
Early versus late: why the 7-day line exists
Seizures complicate roughly 5–10% of strokes overall, and stroke is the most commonly identified cause of newly diagnosed epilepsy in adults over 60. But the timing of the first seizure carries most of the prognostic information.
An early seizure — within the first week, most often within the first 24–48 hours — is an acute symptomatic seizure: the electrical noise of acutely injured, edematous, biochemically deranged tissue. Once the acute injury settles, most of these patients never seize again, which is why an early seizure does not, by itself, make a diagnosis of epilepsy.
A late seizure — beyond the first week, often months to years out — comes from a different substrate: a matured cortical scar that has rewired into a seizure focus. That substrate does not go away. Recurrence risk after a first late seizure is high enough that the International League Against Epilepsy's practical definition counts one unprovoked seizure with a high probability of recurrence as epilepsy — and a late post-stroke seizure is the textbook example.1
| Early (≤ 7 days) | Late (> 7 days) | |
|---|---|---|
| What it is | Acute symptomatic seizure — a symptom of acute injury | Unprovoked seizure from an epileptogenic scar |
| Does it equal epilepsy? | No | Yes — one seizure meets the ILAE definition |
| Typical treatment | Treat the event; short course of medication at most, then reassess | Long-term antiseizure medication is usually warranted |
| What to tell the family | "A symptom of the injury — most patients never have another" | "This is post-stroke epilepsy, and it is usually very treatable" |
Who develops post-stroke epilepsy? You can estimate it at the bedside.
Late-seizure risk is not evenly distributed, and the risk factors are things you already know on day one. In the SeLECT derivation and validation cohorts, five variables predicted late seizures after ischemic stroke: stroke Severity, Large-artery atherosclerotic etiology, Early seizure, Cortical involvement, and Territory of the MCA.2 The updated SeLECT 2.0 model additionally weights status epilepticus as the presentation of an acute symptomatic seizure — the single strongest predictor in the model.3
You can score it interactively with our SeLECT 2.0 calculator, which pairs each total with the risk band and — just as important — with what the number should and should not change. A high score changes counseling and vigilance: it identifies the patient whose "funny spell" in month six deserves an EEG rather than reassurance. It does not, by itself, justify starting a drug before any seizure has occurred.
Hemorrhagic stroke deserves its own mention: intracerebral hemorrhage carries a higher seizure risk than ischemic stroke overall, with lobar (cortical) location the dominant risk factor. The same early-versus-late logic applies, and the same prophylaxis answer — coming next — applies too.
The prophylaxis question: guidelines are unusually blunt
Should a patient with a fresh cortical infarct, or a fresh lobar hemorrhage, be started on an antiseizure medication before any seizure happens? It remains one of the most common curbside questions on a stroke service, and one of the clearest answers in the guidelines:
- The AHA/ASA guideline for acute ischemic stroke recommends against prophylactic antiseizure medication; recurrent seizures are treated as they are in other neurological conditions.4
- The 2022 AHA/ASA intracerebral hemorrhage guideline reaches the same conclusion for ICH: no prophylaxis in patients without seizures; treat clinical seizures, and use EEG to look for electrographic seizures in patients with a depressed or fluctuating level of consciousness that is out of proportion to the bleed.5
The reasoning is not that seizures are harmless — it is that prophylaxis has not been shown to improve outcomes, older prophylactic agents (especially phenytoin) have been associated with worse recovery, and blanket sedating medication is a real cost to a brain trying to rehabilitate. The practical translation on rounds: the order set should not contain levetiracetam "just in case."
Treat the seizures the patient has. Do not treat the seizures you fear.
Managing the early seizure
When a seizure happens in the first week, the immediate priorities are the same as for any acute symptomatic seizure: protect the airway, check glucose, and ask what else in the acute setting is contributing — hyponatremia (worth re-checking with our sodium correction calculator when the glucose is high), infection, drug withdrawal, or hemorrhagic transformation, which warrants repeat imaging when the seizure comes with new deficit or a change in exam.
Two situations deserve special respect. First, status epilepticus — ongoing or back-to-back seizures without recovery — is an emergency with its own treatment ladder; our status epilepticus note builder walks the weight-based sequence. Second, nonconvulsive seizures: in a stroke patient whose consciousness is worse than the imaging explains, or who is not waking up as expected, a low threshold for EEG is appropriate — this is exactly the scenario the ICH guideline flags.5
After a single early seizure with a clear acute cause, many clinicians use a short course of an antiseizure medication through the acute period and then reassess — often stopping before discharge or at early follow-up. What an early seizure should not automatically produce is a lifetime prescription. It should produce a documented plan: what was started, why, and when it will be reconsidered — the kind of line that belongs in the discharge paperwork alongside the rest of the stroke discharge checklist.
Managing the late seizure — this is epilepsy, so choose the drug like a stroke doctor
A late seizure changes the frame. Recurrence risk is high, a single event meets the definition of epilepsy, and long-term treatment is usually warranted. The good news belongs in the same breath: post-stroke epilepsy is among the more pharmacoresponsive epilepsies, and most patients achieve seizure freedom on a single agent.
Drug selection in this population is where stroke medicine and epilepsy medicine meet, and three considerations dominate:
- Drug interactions. Enzyme-inducing agents — carbamazepine, phenytoin, phenobarbital — induce the metabolism of direct oral anticoagulants, warfarin, and statins. In a population that lives on exactly those drugs, an inducer can quietly strip a patient of stroke prevention. Levetiracetam and lamotrigine avoid the problem and are the usual first-line choices.
- Renal function. Levetiracetam is renally cleared and needs dose adjustment as creatinine clearance falls — check it against the Cockcroft-Gault calculator rather than the eGFR on the chart.
- Tolerability in an older, injured brain. Slow titration (lamotrigine), attention to mood and irritability (levetiracetam), and avoidance of sedating polypharmacy matter more here than in a young epilepsy clinic population. If phenytoin is ever in the picture — usually inherited from elsewhere — remember that a low albumin makes the total level lie; correct it with the Sheiner-Tozer calculator.
And counsel concretely: driving rules (jurisdiction-specific, and the clinician's job to raise), seizure first-aid for the family, and the reassurance that this diagnosis, unlike much of what follows a stroke, usually responds well to treatment. Seizures are only one entry on the list of late complications worth naming out loud — depression is another, and screening for it is the subject of our PHQ-9 page.
The bottom line
Ask two questions of every post-stroke seizure. When did it happen relative to the stroke — inside or outside the 7-day window? And what does this patient's substrate predict — how severe, how cortical, what mechanism, any early seizure already? The first question tells you whether you are treating a symptom or diagnosing epilepsy. The second tells you how vigilant to be. Prophylax nobody, treat the late seizure like the chronic diagnosis it is, and pick drugs that do not sabotage the anticoagulant keeping the next stroke away.
Frequently asked questions.
Should patients get antiseizure medication after a stroke to prevent seizures?
No. AHA/ASA guidelines recommend against prophylactic antiseizure medication after both ischemic stroke and intracerebral hemorrhage. Seizures that actually occur are treated; medication is not started for seizures that might occur.
Does one seizure after a stroke mean epilepsy?
It depends on timing. A seizure within 7 days is an acute symptomatic seizure and does not constitute epilepsy — most of these patients never seize again. A single unprovoked seizure more than 7 days after the stroke carries a high recurrence risk and meets the ILAE practical definition of epilepsy.
What is the SeLECT score?
SeLECT estimates the risk of late seizures (post-stroke epilepsy) after ischemic stroke from five bedside variables: severity, large-artery atherosclerosis, early seizure, cortical involvement, and MCA territory. SeLECT 2.0 adds extra weight when the acute seizure presented as status epilepticus. It is a counseling and vigilance tool, not a reason to start prophylaxis.
Which seizure medication is best after a stroke?
There is no single best agent, but levetiracetam and lamotrigine are the usual first-line choices because they do not induce the metabolism of DOACs, warfarin, or statins and are reasonably tolerated in older patients. Enzyme inducers such as carbamazepine, phenytoin, and phenobarbital can lower anticoagulant and statin levels and are generally avoided. Levetiracetam requires renal dose adjustment.
How long should antiseizure medication continue after an early post-stroke seizure?
There is no fixed rule. Many clinicians treat through the acute period and reassess quickly — often stopping at discharge or early follow-up — because an acute symptomatic seizure does not itself predict long-term epilepsy. The key is a documented plan for stopping rather than an open-ended prescription.
When is an EEG needed after a stroke?
An EEG is warranted when there is a clinical suspicion of seizures — witnessed events, unexplained fluctuations — and particularly when the level of consciousness is worse than the imaging explains, where nonconvulsive seizures are on the differential. It is not a routine screening test after every stroke.
References.
- Fisher RS, Acevedo C, Arzimanoglou A, et al. ILAE official report: a practical clinical definition of epilepsy. Epilepsia. 2014;55(4):475-482. PubMed
- Galovic M, Döhler N, Erdélyi-Canavese B, et al. Prediction of late seizures after ischaemic stroke with a novel prognostic model (the SeLECT score): a multivariable prediction model development and validation study. Lancet Neurol. 2018;17(2):143-152. PubMed
- Sinka L, Abraira L, Imbach LL, et al. Association of mortality and risk of epilepsy with type of acute symptomatic seizure after ischemic stroke and an updated prognostic model. JAMA Neurol. 2023;80(6):605-613. PubMed
- Powers WJ, Rabinstein AA, Ackerson T, et al. Guidelines for the Early Management of Patients With Acute Ischemic Stroke: 2019 Update to the 2018 Guidelines. Stroke. 2019;50(12):e344-e418. PubMed
- Greenberg SM, Ziai WC, Cordonnier C, et al. 2022 Guideline for the Management of Patients With Spontaneous Intracerebral Hemorrhage: A Guideline From the American Heart Association/American Stroke Association. Stroke. 2022;53(7):e282-e361. PubMed
Related guides
Keep building the picture.
- SeLECT 2.0 calculator Score late-seizure risk from the five bedside variables this article explains.
- The first 24 hours after stroke Where families first hear about seizures on the stroke unit.
- Status epilepticus note builder The weight-based treatment ladder for the seizure that will not stop.
- Creatinine clearance (Cockcroft-Gault) Renal dosing for levetiracetam and the DOACs it must not sabotage.
- Anticoagulation timing after stroke The other long-term drug decision running alongside the antiseizure one.
Article written by