Epilepsy Diagnosis and Antiseizure Therapy: Choosing the Right Medication

Clinical Practice Update — Classification, First-Line ASM Selection, Special Populations, and Treatment-Resistant Epilepsy in Adults

This is an original clinical education article informed by current guidelines and evidence. See References below for source documents.

MDA-EPI-2026 · 15 min read

Clinical Focus

Epilepsy classification, when to start treatment, first-line antiseizure medication selection for focal and generalised epilepsy, management in women of childbearing age, the elderly, and treatment-resistant epilepsy

Target Audience

Neurologists, epileptologists, general physicians, emergency physicians, pharmacists, residents

Setting

Outpatient neurology, epilepsy clinics, emergency departments, primary care

Source Evidence

•AAN/AES 2018 (reaffirmed 2024) — Efficacy of New ASMs for New-Onset and Treatment-Resistant Epilepsy
•NICE NG217 (2022, updated 2025) — Epilepsies in Children, Young People, and Adults
•ILAE 2013 Updated Evidence Review — ASM Efficacy as Initial Monotherapy
•SANAD II Trial (Lancet, 2021) — Lamotrigine vs LEV/ZNS for Focal; Valproate vs LEV for Generalised Epilepsy
•AAN/AES/SMFM 2024 — Teratogenesis and ASM Exposure in Pregnancy
•AAN/AES 2015 (reaffirmed 2024) — Management of an Unprovoked First Seizure in Adults

Key Clinical Takeaways

The most important actionable points from this Practice Update on epilepsy diagnosis and antiseizure therapy. Each links to the full discussion below.

Epilepsy diagnosis and antiseizure therapy decision pathway showing medication selection for focal and generalised epilepsy in adults — Overview of the clinical approach to epilepsy classification and antiseizure medication selection in adults.

1Classify the epilepsy as focal or generalised before choosing a medication — incorrect classification leads to wrong drug selection and treatment failure → Getting the Diagnosis Right
2Lamotrigine is first-line for new-onset focal epilepsy — SANAD II showed it is superior to levetiracetam and zonisamide in the per-protocol analysis, with fewer adverse reactions → Focal Epilepsy
3Valproate remains the most effective first-line option for generalised epilepsy — but is contraindicated in women of childbearing potential due to teratogenicity → Generalised Epilepsy
4Avoid valproate and topiramate in all women of childbearing age unless no alternative exists and a pregnancy prevention programme is in place → Women of Childbearing Age
5In the elderly, prefer lamotrigine or gabapentin for focal epilepsy — avoid enzyme-inducing agents due to drug interactions and bone health concerns → Older Adults
6If two appropriately chosen ASMs fail to achieve seizure freedom, the epilepsy is treatment-resistant — refer for epilepsy surgery evaluation without delay → Treatment-Resistant Epilepsy
7Cenobamate has emerged as one of the most effective adjunctive therapies for treatment-resistant focal epilepsy (21% seizure-freedom at 400 mg in phase 3 trials) — but requires slow titration due to DRESS risk → Treatment-Resistant Epilepsy
8Sodium channel blockers (carbamazepine, oxcarbazepine, phenytoin) can worsen absence and myoclonic seizures — never use them in idiopathic generalised epilepsy → Drugs to Avoid
9Strongly consider starting treatment after a first unprovoked seizure if EEG shows epileptiform activity, MRI shows a structural lesion, or the seizure occurred during sleep → When to Start Treatment
10Levetiracetam is not first-line for newly diagnosed generalised epilepsy — SANAD II showed it is inferior to valproate for seizure remission → Generalised Epilepsy

How Should You Classify the Epilepsy?

Correct classification is the single most important step before prescribing, because the wrong drug for the wrong epilepsy type causes treatment failure and can worsen seizures. The ILAE classification framework distinguishes seizure type, epilepsy type, and epilepsy syndrome — all three levels should be considered when choosing an antiseizure medication.

Perform an EEG in all patients with suspected epilepsy. A routine interictal EEG showing epileptiform discharges supports the diagnosis and helps classify the epilepsy as focal or generalised.

Strong Rec High Evidence NICE NG217 AAN/AES 2015

Perform brain MRI (epilepsy protocol) in all adults with newly diagnosed epilepsy to identify structural causes such as hippocampal sclerosis, tumours, vascular malformations, or cortical dysplasia.

Strong Rec High Evidence NICE NG217

Consider starting antiseizure therapy after a single unprovoked seizure if there are features indicating a high recurrence risk: epileptiform abnormalities on EEG, a structural lesion on MRI, or a seizure that occurred from sleep.

Moderate Rec Moderate Evidence AAN/AES 2015

Clinical Pearl: A normal EEG does not exclude epilepsy — the sensitivity of a single routine EEG is only around 50%. If the first EEG is non-diagnostic but clinical suspicion remains high, repeat the EEG (ideally with sleep deprivation) or arrange prolonged or ambulatory EEG monitoring. The yield increases substantially with repeated studies.

Which Medication Should You Choose for Focal Epilepsy?

Focal epilepsy accounts for approximately 60% of all epilepsy in adults. The SANAD II trial (2021) provides the best comparative evidence for first-line therapy. In the per-protocol analysis, lamotrigine achieved faster 12-month seizure remission than both levetiracetam and zonisamide. In the intention-to-treat analysis, levetiracetam failed to meet non-inferiority to lamotrigine, while zonisamide did meet non-inferiority. Adverse reactions were also less frequent with lamotrigine (33%) than with levetiracetam (44%) or zonisamide (45%). The overall conclusion: SANAD II does not support levetiracetam or zonisamide as first-line. This aligns with NICE NG217, which recommends lamotrigine or carbamazepine as first-line for focal-onset seizures.

Prescribe lamotrigine as first-line monotherapy for adults with newly diagnosed focal epilepsy. In SANAD II, it was superior to both levetiracetam and zonisamide in the per-protocol analysis for time to 12-month remission, and had significantly fewer adverse reactions (33% vs 44% for levetiracetam and 45% for zonisamide).

Strong Rec High Evidence SANAD II 2021 NICE NG217 AAN/AES 2018

Consider levetiracetam as an alternative first-line for focal epilepsy when rapid titration is needed (lamotrigine requires slow uptitration over weeks). Levetiracetam can reach therapeutic doses within days, which is clinically useful when seizures are frequent.

Moderate Rec High Evidence AAN/AES 2018 ILAE 2013

Consider oxcarbazepine or carbamazepine as alternatives for focal epilepsy when lamotrigine and levetiracetam are not suitable. The 2013 ILAE update gives Level A evidence to carbamazepine, phenytoin, levetiracetam, and zonisamide for adults with focal-onset seizures — however, SANAD II comparative data favour lamotrigine over levetiracetam and zonisamide for long-term effectiveness. Carbamazepine has more drug interactions and tolerability issues than lamotrigine.

Moderate Rec High Evidence ILAE 2013 NICE NG217

Consider lacosamide as an alternative monotherapy option for focal epilepsy in adults. It has undergone successful monotherapy conversion trials and has a favourable interaction profile.

Moderate Rec Moderate Evidence AAN/AES 2018

Antiseizure Medications for Focal Epilepsy: A Patient-Centred Comparison

Drug	Typical Maintenance Dose	Time to Reach Target	Best Suited For	Key Practical Issues
Lamotrigine	100–200 mg BID	6–8 weeks (slow titration required)	First-line for focal and generalised epilepsy; safe in pregnancy	Risk of serious rash (SJS) with fast titration; dose halved with valproate co-prescription; HLA-B*1502 testing in East/Southeast Asian populations
Levetiracetam	500–1500 mg BID	Days (rapid titration possible)	Alternative first-line when fast onset needed; broad spectrum; renal clearance	Psychiatric side effects (irritability, depression, aggression) in up to 15%; less effective than lamotrigine in SANAD II
Oxcarbazepine	600–1200 mg BID	2–3 weeks	Focal epilepsy; fewer interactions than carbamazepine; safe in pregnancy (AAN/AES/SMFM 2024)	Hyponatraemia risk (especially in the elderly); can worsen generalised epilepsy
Lacosamide	100–200 mg BID	1–2 weeks	Focal epilepsy; monotherapy conversion; IV formulation available for acute settings	PR prolongation — obtain baseline ECG; avoid with other sodium channel blockers; limited pregnancy data
Carbamazepine	400–600 mg BID (CR formulation)	2–4 weeks	Focal epilepsy when newer ASMs unavailable; ILAE Level A for adults	Enzyme inducer (multiple drug interactions); hyponatraemia; rash; not first choice in elderly; HLA-B*1502 testing required

Verify all doses against local formulary. Lamotrigine dose is significantly affected by valproate and enzyme-inducing ASM co-prescription.
SANAD II did not include lacosamide, cenobamate, or eslicarbazepine — comparative data for these agents are more limited. Note: the 2013 ILAE update granted Level A evidence to levetiracetam and zonisamide alongside carbamazepine and phenytoin for adults with focal seizures, but SANAD II comparative data show lamotrigine is preferred for long-term effectiveness.

What Works Best for Generalised Epilepsy?

Generalised epilepsy encompasses conditions including juvenile myoclonic epilepsy, absence epilepsies, and epilepsy with generalised tonic-clonic seizures alone. SANAD II demonstrated that valproate remains the most effective first-line agent, achieving 12-month seizure freedom in 33% of patients compared with 24% for levetiracetam. However, valproate carries serious teratogenic risks that fundamentally limit its use in women of childbearing age.

Prescribe valproate as first-line therapy for generalised epilepsy in men and in women who are not of childbearing potential. It is the most effective single agent across all IGE syndromes including JME, absence epilepsy, and GTCS alone.

Strong Rec High Evidence SANAD II 2021 NICE NG217

Do not prescribe valproate to any woman of childbearing potential unless a pregnancy prevention programme is in place and no suitable alternative exists. In utero valproate exposure is associated with major congenital malformations in approximately 10% of pregnancies and neurodevelopmental disorders in 30–40% of exposed children.

Strong Rec High Evidence AAN/AES/SMFM 2024 NICE NG217

Prescribe lamotrigine or levetiracetam as first-line therapy for generalised epilepsy in women of childbearing potential. Both are associated with lower teratogenic risk, although both are less effective than valproate for seizure control. Counsel patients about this trade-off.

Strong Rec Moderate Evidence AAN/AES/SMFM 2024 NICE NG217

Prescribe ethosuximide as first-line for childhood absence epilepsy when GTC seizures are not present. It has ILAE Level A evidence for this indication and was superior to lamotrigine in a double-blind head-to-head trial.

Strong Rec High Evidence ILAE 2013

Do not prescribe sodium channel blockers (carbamazepine, oxcarbazepine, phenytoin, eslicarbazepine) for patients with idiopathic generalised epilepsy. They can exacerbate absence and myoclonic seizures and worsen overall seizure control.

Against High Evidence NICE NG217

Warning

Lamotrigine can worsen myoclonic seizures in some patients with JME, even while controlling GTC seizures. Monitor closely when using lamotrigine as a valproate alternative in this syndrome. If myoclonus worsens, consider switching to levetiracetam or adding a low-dose valproate (in men or when pregnancy is not a concern).

Clinical Pearl: The SANAD II result for generalised epilepsy is perhaps the most clinically difficult to act on. Valproate is clearly the best drug — but you cannot use it in half your patient population. For women of childbearing age with JME or generalised epilepsy, the conversation must be explicit: lamotrigine or levetiracetam may not control seizures as well, but the teratogenic risk of valproate is unacceptably high. Document this discussion carefully.

What Should You Do When the First Two Drugs Fail?

Approximately one-third of people with epilepsy do not achieve seizure freedom with the first appropriately chosen and tolerated antiseizure medication. After failure of two adequately trialled ASMs, the probability of achieving seizure freedom with subsequent medication changes drops substantially — landmark data suggest only around 10–15% will become seizure-free. This is the ILAE definition of drug-resistant (treatment-resistant) epilepsy, and it should prompt referral for epilepsy surgery evaluation.

Refer all patients with drug-resistant epilepsy (failure of two appropriately chosen and tolerated ASMs) to a comprehensive epilepsy centre for evaluation, including consideration of epilepsy surgery. Do not delay this referral for sequential drug trials.

Strong Rec High Evidence NICE NG217

Consider cenobamate as adjunctive therapy for treatment-resistant focal epilepsy in adults. In the Krauss et al. phase 3 dose-response trial, seizure-freedom rates reached 21% at the 400 mg dose (vs 1% placebo) — substantially higher than historical rates for other adjunctive ASMs. Titrate very slowly (starting at 12.5 mg daily, increasing over 12+ weeks) due to the risk of DRESS syndrome with rapid dose escalation.

Moderate Rec Moderate Evidence FDA-approved 2019

Consider lacosamide, eslicarbazepine, brivaracetam, or perampanel as adjunctive therapy options for treatment-resistant focal epilepsy. Selection should be guided by the patient’s existing regimen, comorbidities, and side-effect profile.

Moderate Rec High Evidence AAN/AES 2018

Reassess the diagnosis before labelling epilepsy as treatment-resistant. Up to 20% of patients referred to epilepsy surgery centres are found to have non-epileptic events (psychogenic non-epileptic seizures). Video-EEG monitoring is essential for confirmation.

Strong Rec Moderate Evidence NICE NG217

Clinical Decision Pathway

Choosing an Antiseizure Medication: 4 Questions

Question 1: Is this focal or generalised epilepsy?

Use EEG, MRI, and seizure semiology to classify. Focal features: aura, unilateral motor activity, impaired awareness with automatisms. Generalised features: bilateral tonic-clonic from onset, absences, myoclonus.

Question 2: Is this patient a woman of childbearing potential?

If yes → Avoid valproate and topiramate. Prefer lamotrigine, levetiracetam, or oxcarbazepine (all with lower teratogenic risk per AAN/AES/SMFM 2024). Ensure folic acid supplementation and contraception counselling.

If no → Drug selection based on epilepsy type, comorbidities, and interactions.

Question 3: Which first-line ASM should I start?

Focal epilepsy → Lamotrigine (first choice), levetiracetam (if rapid onset needed), oxcarbazepine or carbamazepine (alternatives).

Generalised epilepsy (men / non-childbearing) → Valproate.

Generalised epilepsy (women of childbearing age) → Lamotrigine or levetiracetam.

Absence epilepsy without GTC → Ethosuximide (first choice) or valproate.

Elderly → Lamotrigine or gabapentin (fewer interactions, better tolerability).

Question 4: What if two ASMs have failed?

Confirm diagnosis (consider video-EEG to exclude PNES). Refer to comprehensive epilepsy centre for surgery evaluation. Consider adjunctive therapy: cenobamate, lacosamide, brivaracetam, perampanel, or clobazam depending on epilepsy type.

Evidence in Context

SANAD II: What It Showed and What It Means

SANAD II was a pragmatic, open-label UK trial that recruited 1,510 people with newly diagnosed epilepsy. For focal epilepsy (990 patients randomised to lamotrigine, levetiracetam, or zonisamide), levetiracetam failed to meet non-inferiority to lamotrigine in the ITT analysis (HR 1.18, 97.5% CI 0.95–1.47), while zonisamide did meet non-inferiority (HR 1.03, 0.83–1.28). However, in the per-protocol analysis, lamotrigine was superior to both drugs. Adverse reactions were significantly less common with lamotrigine (33%) than levetiracetam (44%) or zonisamide (45%). For generalised or unclassified epilepsy (520 patients), valproate was superior to levetiracetam: 33% achieved immediate 12-month freedom vs 24% for levetiracetam, and levetiracetam was more likely to fail due to inadequate seizure control. The trial was unblinded, which is a limitation, but its pragmatic design reflects real-world clinical practice.

The Valproate Dilemma in Women

The 2024 AAN/AES/SMFM guideline on teratogenesis makes clear that valproate and topiramate carry the highest risks among ASMs. Valproate is associated with major congenital malformations in roughly 10% of exposed pregnancies and neurodevelopmental disorders (including autism spectrum disorder and reduced IQ) in 30–40%. Topiramate also carries increased risk of oral clefts and is now subject to pregnancy prevention requirements in some jurisdictions. Lamotrigine, levetiracetam, and oxcarbazepine are associated with the lowest risks of major congenital malformations and are the preferred agents in this population.

Where NICE, AAN/AES, and ILAE Agree and Differ

All three major guideline bodies agree on the importance of epilepsy classification before drug selection, the primacy of monotherapy, the strong efficacy of valproate for generalised epilepsy, and the need to avoid valproate in women of childbearing age. They differ mainly in the granularity of their first-line recommendations: NICE explicitly recommends lamotrigine or carbamazepine for focal epilepsy; ILAE provides a broader evidence-graded list; and AAN/AES focuses on the comparative efficacy of newer agents against traditional drugs. SANAD II results are most directly reflected in the 2022 NICE update.

Cenobamate: A Game-Changer for Treatment-Resistant Focal Epilepsy?

Cenobamate (FDA-approved 2019) has a dual mechanism of action — enhancing fast and slow inactivation of sodium channels and positive allosteric modulation of GABA-A receptors. In the Krauss et al. phase 3 dose-response trial (Lancet Neurol 2020), seizure-freedom rates during the maintenance phase were 1% (100 mg), 11% (200 mg), and 21% (400 mg) versus 1% for placebo — substantially higher than historical rates for other adjunctive ASMs in treatment-resistant populations (typically 2–5%). Early clinical development saw cases of DRESS syndrome with an initial rapid titration protocol; the amended slow-titration schedule (starting at 12.5 mg daily) has largely mitigated this risk. Long-term open-label extension data show seizure-freedom rates of 10–16% at 12-month intervals. Cenobamate is not yet included in the AAN/AES or NICE guidelines due to timing, but it is covered in the 2025 Continuum ASM update and is increasingly used in specialist practice.

What We Still Don’t Know

Optimal ASM for generalised epilepsy in women: Valproate is the best drug but cannot be used. Lamotrigine and levetiracetam are inferior. There is an urgent need for head-to-head trials of newer broad-spectrum agents (e.g., perampanel, cenobamate) in IGE.

When to safely withdraw ASMs: Evidence on ASM discontinuation after prolonged seizure freedom remains sparse, and relapse rates are poorly characterised by epilepsy syndrome.

Biomarkers for treatment resistance: Why do some patients respond to the first ASM while others fail multiple drugs? Genetic predictors and neuroinflammatory markers are under investigation but not yet clinically actionable.

Comparative effectiveness of newer ASMs: SANAD II did not include lacosamide, cenobamate, brivaracetam, or perampanel. Large pragmatic trials comparing these agents against established first-line drugs are needed.

References

1.Marson A, Burnside G, Appleton R, et al. The SANAD II study of the effectiveness and cost-effectiveness of valproate versus levetiracetam for newly diagnosed generalised and unclassifiable epilepsy. Lancet. 2021;397(10282):1375–1386. doi:10.1016/S0140-6736(21)00246-4
2.Marson A, Burnside G, Appleton R, et al. The SANAD II study of the effectiveness and cost-effectiveness of levetiracetam, zonisamide, or lamotrigine for newly diagnosed focal epilepsy. Lancet. 2021;397(10282):1363–1374. doi:10.1016/S0140-6736(21)00247-6
3.Kanner AM, Ashman E, Gloss D, et al. Practice guideline update summary: Efficacy and tolerability of the new antiepileptic drugs I: Treatment of new-onset epilepsy. Neurology. 2018;91(2):74–81. doi:10.1212/WNL.0000000000005755
4.Pack AM, Oskoui M, Williams Roberson S, et al. Teratogenesis, Perinatal, and Neurodevelopmental Outcomes After In Utero Exposure to Antiseizure Medication: Practice Guideline From the AAN, AES, and SMFM. Neurology. 2024;102(11):e209279. doi:10.1212/WNL.0000000000209279
5.Glauser T, Ben-Menachem E, Bourgeois B, et al. Updated ILAE evidence review of antiepileptic drug efficacy and effectiveness as initial monotherapy for epileptic seizures and syndromes. Epilepsia. 2013;54(3):551–563. doi:10.1111/epi.12074
6.National Institute for Health and Care Excellence. Epilepsies in children, young people and adults. NICE Guideline [NG217]. 2022 (updated 2025). nice.org.uk/guidance/ng217
7.Krauss GL, Klein P, Brandt C, et al. Safety and efficacy of adjunctive cenobamate (YKP3089) in patients with uncontrolled focal seizures: a multicentre, double-blind, randomised, placebo-controlled, dose-response trial. Lancet Neurol. 2020;19(1):38–48. doi:10.1016/S1474-4422(19)30399-3

How to Read the Evidence Tags

Every recommendation carries two tags. These are Medaptly’s own simplified interpretations for educational clarity.

Recommendation Strength

Tag	What It Means	In Practice
Strong Rec	Benefits clearly outweigh risks for most patients.	Standard practice.
Moderate Rec	Evidence favours benefit; some uncertainty remains.	Most patients should receive this.
Conditional Rec	Benefit less certain; individualise.	Use shared decision-making.
Against	No benefit or risks outweigh benefits.	Avoid this intervention.

Evidence Quality

Tag	What It Means	Confidence
High Evidence	Multiple RCTs or meta-analyses.	Very confident.
Moderate Evidence	Single RCT or large observational studies.	Reasonably confident.
Low Evidence	Expert consensus or small studies.	Less certain.

These are Medaptly’s simplified interpretations. For the full classification systems used by each source, consult the original documents in References.

Article Information

For Educational Purposes Only. This is original clinical education content informed by current published guidelines and clinical evidence. It does not constitute medical advice, is not endorsed by any guideline body (AAN, AES, ILAE, NICE, or any other organisation), and does not replace individualised clinical judgement, institutional protocols, or local formulary guidance. Drug dosages should always be verified against current prescribing information before prescribing. Readers are encouraged to consult the original source guidelines listed in the References section for the full evidence review and complete recommendation sets.