In the classic algorithm for status epilepticus, we give 2 doses of a benzodiazepine followed by phenytoin. Since the introduction of levetiracetam, we have seen a lot of strong opinions about the best second line agent, but we never had much data. (I have argued that we actually need to be much more aggressive, using propofol or a barbiturate as our second line agent, as I lay out in my approach to status epilepticus.) Earlier this year, we saw levetiracetam compared with phenytoin in pediatric populations in both the ConSEPT and EcLiPSE trials. Both trials failed to demonstrate superiority of levetiracetam. However, the ESETT trial was ongoing, which many people thought would provide us with the definitive answer as to the best second line agent in status epilepticus…
The paper: The ESETT trial
ESETT: Kapur J, Elm J, Chamberlain JM, et al. Randomized Trial of Three Anticonvulsant Medications for Status Epilepticus. The New England journal of medicine. 2019; 381(22):2103-2113. PMID: 31774955 clinicaltrials.gov: NCT01960075
This is a multicenter, randomized, blinded, comparative-effectiveness trial of levetiracetam, fosphenytoin, and valproate for the treatment of patients with established status epilepticus in the emergency department.
Eligible patients were aged 2 years or older in status epilepticus after a “generally accepted cumulative dose of bezodiapeines”. (Although, I actually think the required doses were pretty small – 4mg of lorazepam, 10 mg of diazepam, or 10 mg of midazolam). Patients were enrolled from 57 different hospitals across the United States, over a 2 year period between 2015 and 2017.
- Exclusions: Seizure was caused by major trauma, hypoglycemia, hyperglycemia, cardiac arrest, or anoxia; the patient had already been treated with an anticonvulsant other than a benzodiazepine; the patient had been intubated prior to arrival; there were known allergies or contraindications; the patient was pregnant or incarcerated; or the patient opted out.
There were 3 groups. All groups had the trial drug infused on a pump over 10 minutes.
- Levetiracetam 60 mg/kg to a max of 4500 mg
- Fosphenytoin 20 mg/kg (phenytoin equivalent) to a max of 1500 mg (phenytoin equivalent)
- Valproate 40 mg/kg to a max of 3000 mg
The primary outcomes was the absence of clinically apparent seizures and improving responsiveness at 60 minutes after the trial drug infusion started, without the need for additional medications.
There was also a primary safety outcome: the composite of life-threatening hypotension or cardiac arrhythmia within 60 minutes after the start of trial drug infusion.
There were 400 episodes of status epilepticus among 384 patients enrolled in the study. The trial reached a predetermined futility criteria (less than 1% chance of showing one of the treatments to be more or less effective), so they did not reach their maximum enrollment of 795. 27% of patients actually failed to meet their eligibility criteria, but are included in the intention to treat analysis. This includes 10% that had psychogenic nonepileptic events.
There was no difference in the primary outcome (absence of seizures and clinical improvement), occurring in 47% of the levetiracetam group, 45% of the fosphenytoin group, and 46% of the valproate group. Outcomes were the same in the per protocol analysis.
There also wasn’t a significant difference in the primary safety outcome (life-threatening hypotension or cardiac arrhythmia within 60 minutes): 1.3% with levetiracetam, 3.2% with fosphenytoin, and 1.6% with valproate.
Mortality was 4.7% with levetiracetam, 2.4% with fosphenytoin, and 1.6% with valproate. Although this isn’t statistically significant, ESETT wasn’t powered for this outcome, and a doubling of the most important outcome in the trial warrants attention.
They include intubation as a safety outcome. (I don’t think this is actually a safety outcome. It is just part of good critical care). It wasn’t statistically different: 20% with levetiracetam, 26% with fosphenytoin, and 17% with valproate.
There were no differences in admission to ICU (about 60%), length of ICU stay (1 day), or length of hospital stay (3 days). Among the patients whose seizures stopped, about 80% stopped within 20 minutes of the trial drug being given. About half of each group required an additional anticonvulsant (52% with levetiracetam and 46% with both fosphenytoin and valproate.)
Although ESETT is a really important study, and I have been eager to see the results, I think that most people knew it wasn’t going to be groundbreaking. The insistence that intubation is a bad outcome has resulted in seizure research shying away from the use of anesthetic agents, despite knowing that they are the most likely to stop seizures. Hopefully, now that we know that all of our second line agents are equally bad, we can finally study agents that are more likely to work.
There are a few important critical appraisal points that might affect how you apply the ESETT results to your practice. Only small doses of benzodiazepines were required, and even smaller doses were given, so this population may be healthier than the population I am seeing. The median total dose given to adults was 5 mg of lorazepam (or equivalent). I would have given two 8 mg doses (if I was using lorazepam), but even if you were following the lower dose guidelines, all patients should have received at least 8 mg total. It is likely that a higher dose of benzodiazepine would have stopped seizures in a few extra patients. These less sick patients are also likely to respond to the second line agent. Therefore, I think the second line agents are even less likely to work in my hands. (This is consistent with the data from the Vetran Affairs cooperative study, in which only 7% of patients responded to second line medications). (Treiman 1998)
Whenever looking at a study comparing different drugs, it is also important to consider whether equivalent doses are being used. Although all the ESETT doses are within a range that experts would recommend, both the valproate and levetiracetam doses are higher than what we usually use (and higher than was used in the CONSEPT and ECLIPSE trials). (Glauser 2016; Dalziel 2019; Lyttle 2019) The use of higher doses might have biased the trial in favour of valproate or levetiracetam. Therefore, we might expect to see worse results in the real world if people continue to use the standard doses of those medications.
There are a couple problems with the primary outcome. It is important to remember that seizure cessation is a surrogate outcome. Longer seizures are associated with bad outcomes. Therefore, we would like to see seizures stopped as soon as possible. However, time to seizure cessation is not a patient oriented outcome. What we really care about is long term survival and neurologic function, which wasn’t studied in this trial. (You could imagine a drug that stops seizures very quickly, but does so by causing brain death.) Similarly, hypotension is a surrogate outcome that isn’t necessarily relevant to the patient.
Furthermore, I’m not sure they picked the best surrogate. There is lots of data that suggests that bad outcomes increase significantly if seizures last longer than 20 or 30 minutes. Waiting until 60 minutes seems way too long. If we are going to use a surrogate, I think it makes more sense to look much earlier. (In which case, the results would look even worse, because none of these agents are rapid acting.) To make matters worse, these patients had already been seizing for a full hour by the time they were enrolled in the study!
Finally, the use of “clinically apparent seizures and improving responsiveness” introduces subjectivity into the outcomes. In general, I am relatively comfortable assessing improving responsiveness after giving benzos, but once you add a second medication into the mix, most patients remain lethargic for a long time, and it can be very difficult to determine whether that is the result of ongoing seizures or just a side effect of the medications.
Seeing as we now have multiple studies that demonstrate pretty convincingly that levetiracetam is not more effective than the alternatives, the conversation is bound to turn to side effects. Of course, we now also have 3 trials in a row that show no difference in adverse events, so it is hard to make a convincing argument for levetiracetam on that front (although many will try). However, these trials are relatively small, and not powered to detect rare adverse events. (See this post by Josh Farkas discussing the importance of rare catastrophic complications.)
We should consider all possible sources of scientific data. Anecdote and observational data have suggested that phenytoin may be associated with more adverse events than our other alternatives. However, we need to be very careful extrapolating from those observations. Phenytoin has been used much longer and more commonly than levetiracetam, so there have been far more opportunities to see adverse events occur. Furthermore, the one true clinically important outcome in ESETT was mortality, and it occurred twice as often in the levetiracetam group. With only a handful of outcomes, it is difficult to differentiate signal from noise, but with no indication that levetiracetam provides any advantages, I think that possible signal warrants attention.
On a more philosophical note, I think it is worth considering the processes in place in modern medicine that led so many people to believe (and loudly argue) that levetiracetam was a much better medication (despite no evidence to support the claim). This happens over and over again in medicine. A new agent is released and everyone gets excited. There will often be some mention of the fact that there is no science to support the new drug, but everyone starts prescribing it anyway. Then, a decade later (just as the patent is about to expire and the money has already been made), we finally get our first well done, non-industry funded study, which inevitably reveals that the fancy, new, and much more expensive medication or intervention was no better than what we already had.
Of course, by the time the big study is finally completed, the drug is generic. Therefore, people will argue that, even though it isn’t any better, we might as well continue to use it, because at least it isn’t as expensive as it used to be. Somehow that argument is supposed to excuse the last decade, in which we over-enthusiastically adopted a far more expensive medication that was never any better than the existing treatments. Unfortunately, it is too late in the case of levetiracetam, but we really need to learn our lesson. We need to wait for evidence that new, expensive medications are better than existing treatments before rushing to prescribe them. (I will leave links to posts about idarucizumab and andexanet alfa here.)
Ultimately, none of the studied options were good enough. Long seizures are bad. We don’t know how long is too long, but permanent brain damage seems to start between 20 and 30 minutes. (Zaccara 2017) Longer seizures are harder to stop, and associated with a host of other medical problems, like hyperthermia, rhabdomyolysis, acidosis, and hyperkalemia. Almost half of these patients were still seizing 1 hour after the administration of the study drug, which was 2 hours after the start of their seizure. That is simply way too long.
Instead of telling us which of these three agents should be used, I think ESETT confirms that none of these medications are good enough as a second line agents in status epilepticus. Although there isn’t solid evidence yet, I think we clearly need to be moving towards an early aggressive approach to status epilepticus that uses anesthetic agents like propofol or barbiturates as the second line agents, and only uses these traditional agents as adjuncts or third line agents. I have gone into this at length in my post on using propofol as the second line agent in status epilepticus.
In this multicenter RCT, there was no difference in efficacy or safety between levetiracetam, fosphenytoin, or valproate in the management of status epilepticus. All were equally ineffective, and this trial should point us towards alternative strategies for managing these critically ill patients.
Other FOAMed looking at ESETT
Dalziel SR, Borland ML, Furyk J, et al. Levetiracetam versus phenytoin for second-line treatment of convulsive status epilepticus in children (ConSEPT): an open-label, multicentre, randomised controlled trial. The Lancet. 2019; http://dx.doi.org/10.1016/S0140-6736(19)30722-6
Glauser T, Shinnar S, Gloss D, et al. Evidence-Based Guideline: Treatment of Convulsive Status Epilepticus in Children and Adults: Report of the Guideline Committee of the American Epilepsy Society. Epilepsy currents. 2016; 16(1):48-61. [pubmed]
Kapur J, Elm J, Chamberlain JM, et al. Randomized Trial of Three Anticonvulsant Medications for Status Epilepticus. The New England journal of medicine. 2019; 381(22):2103-2113. [pubmed]
Lyttle MD, Rainford NEA, Gamble C, et al. Levetiracetam versus phenytoin for second-line treatment of paediatric convulsive status epilepticus (EcLiPSE): a multicentre, open-label, randomised trial. The Lancet. 2019; http://dx.doi.org/10.1016/S0140-6736(19)30724-X
Treiman DM, Meyers PD, Walton NY, et al. A Comparison of Four Treatments for Generalized Convulsive Status Epilepticus N Engl J Med. 1998; 339(12):792-798. [article]
Zaccara G, Giannasi G, Oggioni R, et al. Challenges in the treatment of convulsive status epilepticus. Seizure. 2017; 47:17-24. [pubmed]
Morgenstern, J. Our second line agents for status epilepticus all suck (The ESETT trial), First10EM, December 30, 2019. Available at: