Established Status Epilepticus: What the ESETT Trial Settled

The short answer#

When a seizure refuses to stop and the standard first drug fails, clinicians have long had three reasonable next options and no good way to rank them. The Established Status Epilepticus Treatment Trial (ESETT), reported in the New England Journal of Medicine in 2019, put those three head to head and found a tie. Levetiracetam, fosphenytoin, and valproate each stopped the seizure and improved responsiveness within an hour in roughly half of patients, with similar rates of serious harm. A tie sounds like a disappointment, but here it is a gift: three interchangeable options mean the decision can turn on what your hospital actually stocks, what a patient can safely receive, and what costs less.

Key points#

• ESETT compared three second-line drugs for convulsive seizures that did not respond to a benzodiazepine.
• Success rates were close: about 47 percent for levetiracetam, 45 percent for fosphenytoin, and 46 percent for valproate.
• Serious side effects, including low blood pressure and the need for a breathing tube, were similar across all three.
• A follow-up analysis found the same balance held from young children to older adults.
• Because no drug won, clinicians can choose based on availability, cost, drug interactions, and individual contraindications.

When the first drug is not enough#

A seizure that keeps going on its own is called status epilepticus, and the convulsive form ESETT studied is a true emergency. The longer active seizing continues, the more it can injure brain tissue and the harder it becomes to shut down. The opening move is settled and unambiguous: give a benzodiazepine such as lorazepam or midazolam, fast, at a full dose.

The problem is that benzodiazepines do not work for everyone. In a meaningful share of patients the convulsions continue, and the situation graduates to what clinicians call established, or benzodiazepine-refractory, status epilepticus. This is the moment ESETT was built for. The team has controlled the airway and given the first drug, the seizure is still running, and someone has to decide what goes in next.

Three drugs, no way to choose between them#

For years that second decision rested more on habit than on data. Fosphenytoin, a prodrug of phenytoin, was the traditional pick. Valproate and levetiracetam were both in wide use, favored by different clinicians and different institutions largely because of familiarity and a scatter of small studies rather than any direct comparison. Professional guidelines listed all three side by side without ranking them, an honest reflection of the fact that nobody had run the trial that would let them.

That gap mattered more than it might seem. Hospitals build order sets, stock pharmacies, and train staff around a default drug. When the default is chosen on tradition rather than evidence, an entire system can commit to a preference that may not hold up. ESETT set out to test whether any of the three deserved that role.

A trial engineered to stop as soon as it knew#

ESETT was funded through the National Institutes of Health neurological and pediatric emergency research networks and ran across dozens of United States emergency departments. Patients qualified if convulsive status epilepticus persisted after an adequate benzodiazepine dose. They were then randomly assigned to levetiracetam, fosphenytoin, or valproate.

Two design choices are worth understanding. The first is blinding: neither the treating clinicians nor the people scoring the outcome knew which drug a patient got. In an emergency where a clinician's expectations can color how they read a flickering, partially sedated patient, that matters. The second is a response-adaptive Bayesian design. Instead of locking the three group sizes in advance, the trial updated its randomization as results came in, nudging more patients toward whichever drug looked stronger, and it scheduled interim looks along the way. If one drug were truly better, this structure was built to surface it fast. If the drugs were equal, the same structure was built to notice that and stop, rather than enroll hundreds more patients to confirm a tie.

The outcome it measured was deliberately practical: no visible seizure activity and improving alertness at 60 minutes, without any rescue anticonvulsant. Each half of that definition earns its place. A drug that flattens the convulsion but leaves the patient unrousable has not delivered a good result, and neither has one that only appears to work because a second agent bailed it out.

The verdict: a three-way tie#

The three arms came out almost on top of one another. Seizures stopped and alertness improved in about 47 percent of the levetiracetam group, 45 percent of the fosphenytoin group, and 46 percent of the valproate group. The trial's Bayesian analysis put the probability that each drug was the single best option at roughly 0.41, 0.24, and 0.35. None crossed the line for superiority.

Because the numbers were converging rather than pulling apart, ESETT hit its pre-planned futility threshold and stopped early, after about 400 patients out of a planned ceiling near 720. Stopping for futility is not the trial giving up. It is the design doing precisely what it promised: recognizing that more enrollment would not crown a winner, and ending a comparison whose answer was already in hand.

Safety told the same story. Endotracheal intubation, low blood pressure, seizures returning over the following hours, and death occurred at broadly similar rates in all three arms, with no drug emerging as clearly safer. A companion analysis in The Lancet in 2020 sorted the results by age, from children through older adults, and found the same balance within every age band. The equivalence held across a lifespan.

Why "no difference" is an answer, not a shrug#

It is easy to file a null result under "found nothing." Read it that way and you have it backward. Before ESETT, the belief that one of these drugs was best rested on tradition, and tradition was already shaping real decisions about what to stock and what to give. ESETT swapped that soft assumption for a firm conclusion: for this emergency, on average, the three drugs are interchangeable.

That conclusion does real work at the bedside. Once efficacy and safety are a wash, the decision can rest on the things that genuinely differ between drugs and between patients. Valproate is generally avoided in pregnancy and in some metabolic disorders. The agents differ in how many drug interactions they carry and how quickly they can be drawn up and infused. Cost and shelf availability vary from one hospital to the next. A tie does not leave you empty-handed. It hands the choice back and tells you it is safe to decide on other grounds.

What ESETT does not settle#

A single trial, read carefully, has edges. ESETT reports averages, and an average of no difference cannot rule out that some particular subgroup responds better to one drug, only that none stood out. It tested specific drugs at specific doses for one specific emergency, so its findings do not automatically transfer to other agents, other doses, or seizures that are not convulsive. And equal control at 60 minutes is not a promise about every later outcome. Within the question it was designed to answer, though, ESETT answered plainly, and that clarity is exactly what the field had been missing.