What does it mean when epilepsy is called refractory or drug-resistant?

Published by Unseen Progress, an independent publisher of caregiver research. Last reviewed 2026-05-10. Part of the epilepsy caregiver research overview.

Short answer. Refractory and drug-resistant are the same thing. The International League Against Epilepsy (ILAE) defines drug-resistant epilepsy as failure of two appropriately chosen, tolerated, and adequately trialled anti-seizure medications to achieve sustained seizure freedom (Kwan et al., ILAE, 2010). After the second proper failure, the probability that a third drug will produce freedom falls to around 5%, and the evidence shifts in favour of comprehensive epilepsy center evaluation — including surgery, neurostimulation, and dietary therapy. The diagnosis is not a verdict; it is the threshold at which the conversation changes.

Where the definition comes from

Before 2010 the term refractory was used informally and inconsistently — some clinicians used it after one failure, others after five. The result was that families with surgically remediable epilepsy waited years longer than they should have, and the average gap between epilepsy onset and surgical evaluation in adult cohorts was 20+ years (Berg et al., 2003; Engel et al., 2003). The ILAE Task Force on Therapeutic Strategies convened to fix this — to draw a clean clinical line that would shorten the path from diagnosis to evaluation for the third of patients who would not become seizure-free on medication alone.

The resulting consensus (Kwan et al., 2010) defined drug-resistant epilepsy with the two-drug rule: failure of two adequately trialled ASM regimens. This number was not arbitrary. It tracks the Kwan and Brodie cohort data showing that two-thirds of newly diagnosed patients achieve seizure freedom on the first or second medication (Kwan & Brodie, 2000). The probability of a third drug succeeding in patients who failed two falls steeply — to roughly one in twenty in subsequent confirmatory studies (Brodie et al., 2012; Chen et al., 2018). At that probability, the expected return on trying further drugs falls below the expected return on evaluating for surgery or neurostimulation in candidates.

What "adequately trialled" actually requires

The diagnosis depends on what counts as a trial. The ILAE consensus operationalises this carefully:

1. Appropriately chosen for the seizure type and syndrome. A carbamazepine trial in juvenile myoclonic epilepsy is not a trial — it is a contraindicated choice. A lamotrigine trial in Dravet syndrome is not a trial — it can worsen seizures. Drug-syndrome match is criterion one. 2. Tolerated. A drug stopped at sub-therapeutic dose because of intolerable side effects is not a failed trial of efficacy; it is a tolerability failure. Different clinical implications. 3. Adequately dosed. The drug must have reached the maximum tolerated dose, sustained for long enough to evaluate. Patients who never got past the titration ladder do not have a failed trial. 4. Adequate duration. Sustained at target dose long enough for pharmacokinetic steady state plus enough event-free time to interpret. Practically, 3–6 months minimum for most ASMs at typical seizure frequencies.

A drug that fails on any of these four does not count toward the two-drug rule. Families who arrive at neurology saying "we've tried six drugs and nothing works" often discover, on careful history, that none of the six met all four criteria. That is a different clinical situation than two adequate failures, and the next step is usually to run a proper trial of the most appropriate drug — not to declare resistance.

What the diagnosis opens, not what it closes

The cultural connotation of "drug-resistant" suggests a closed door — nothing more can be done. The clinical reality is the opposite. The diagnosis opens four options that are not available, or not appropriate, before the threshold is met:

  • Epilepsy surgery. For drug-resistant focal epilepsy with a localisable seizure focus, resective surgery is among the most effective treatments in all of neurology. In a landmark randomised trial (Wiebe et al., 2001), 58% of surgical patients with temporal lobe epilepsy were seizure-free at one year, versus 8% of those continued on medical therapy. Outcomes are best when the focus is in a non-eloquent region and when surgery happens earlier rather than later in the course.
  • Vagus nerve stimulation (VNS), responsive neurostimulation (RNS), and deep brain stimulation (DBS). Implantable devices that reduce seizure frequency in patients who are not surgical resection candidates. Each has its own indications and evidence base (covered in the dedicated spoke).
  • Dietary therapy. Ketogenic diet, modified Atkins, and low glycemic index treatment have meaningful efficacy in certain syndromes (Dravet, Lennox-Gastaut, infantile spasms) and in selected drug-resistant patients more broadly.
  • Clinical trials. Drug-resistant patients become eligible for investigational ASMs and devices that earlier-line patients are not eligible for.

The path to all four runs through a comprehensive epilepsy center — typically a National Association of Epilepsy Centers (NAEC) Level 3 or Level 4 facility — and through a presurgical workup that begins with video-EEG monitoring. Referral is the action the diagnosis enables.

Why families delay the referral

The literature is consistent that median delay from drug-resistance criteria being met to surgical evaluation is years, not months (Haneef et al., 2010; Engel, 2008). Several patterns drive the delay:

  • Hope for the next drug. Each new ASM is presented as a possibility. Families and clinicians under-weight the 5% prior and over-weight the recency of the new drug.
  • Fear of brain surgery. Often based on outdated imagery of what epilepsy surgery involves. Modern resective and disconnection procedures have well-characterised risk profiles, and the comparison is not surgery vs. nothing — it is surgery vs. continued seizures.
  • Confusion about what referral commits to. Comprehensive epilepsy center referral is for evaluation, not for surgery. Many patients evaluated turn out not to be surgical candidates — but the evaluation also surfaces VNS/RNS, diet, and trial eligibility.
  • Reluctance to label. The word "refractory" feels like a defeat. The data is the data — two adequate failures is two adequate failures — and the label is the access key to options that hope alone does not provide.

What the diagnosis does not mean

It does not mean the patient will never be seizure-free. Surgical outcomes in well-selected candidates routinely produce seizure freedom or near-freedom. It does not mean further medication is pointless — some drug-resistant patients respond to add-on therapy or to drugs introduced later in the pipeline. It does not mean the patient is a failure of treatment or that the caregiver missed something. The 30% who are drug-resistant are not a treatment failure population — they are the population for whom the next layer of evidence-based options was developed.

What the research suggests doing

1. Ask the neurologist explicitly: "Have we met the ILAE two-drug threshold?" If the answer is unclear, walk through the four criteria for each drug tried. 2. If two adequate trials have failed, ask for referral to a comprehensive epilepsy center — not as a surgical commitment, but for evaluation. 3. Bring structured diary data to the referral conversation. The receiving center needs frequency, type, semiology, medication history, and side-effect profile in organised form. 4. Use the time waiting for the evaluation to compile the medication history: drug, dates, max dose reached, dose-limiting factor if any, response. This is the document the comprehensive center will work from. 5. Resist the pull to add a third or fourth drug while waiting for evaluation, unless the neurologist specifically recommends it. Each additional drug adds side-effect burden and rarely changes the picture.

Related questions

References

  • Kwan, P., Arzimanoglou, A., Berg, A. T., et al. (2010). Definition of drug resistant epilepsy: ILAE consensus proposal. Epilepsia, 51(6), 1069–1077.
  • Kwan, P., & Brodie, M. J. (2000). Early identification of refractory epilepsy. New England Journal of Medicine, 342(5), 314–319.
  • Brodie, M. J., Barry, S. J. E., Bamagous, G. A., Norrie, J. D., & Kwan, P. (2012). Patterns of treatment response in newly diagnosed epilepsy. Neurology, 78(20), 1548–1554.
  • Wiebe, S., Blume, W. T., Girvin, J. P., & Eliasziw, M. (2001). A randomized, controlled trial of surgery for temporal-lobe epilepsy. New England Journal of Medicine, 345(5), 311–318.
  • Berg, A. T., Langfitt, J., Shinnar, S., et al. (2003). How long does it take for partial epilepsy to become intractable? Neurology, 60(2), 186–190.
  • Engel, J. (2008). Surgical treatment for epilepsy: too little, too late? JAMA, 300(21), 2548–2550.
  • Epilepsy Foundation. Drug-resistant epilepsy and comprehensive epilepsy centers. https://www.epilepsy.com/

---

Unseen Progress publishes long-form caregiver research and builds research-backed daily trackers for the families covered. See the full epilepsy caregiver research overview for the complete framework.