Status epilepticus is associated with a high degree of morbidity and mortality. Approximately ⅓ of patients have long term neurologic sequela, and mortality is 3-5%. (Chin 2006; Raspall-Chaure 2006) First line treatment is with benzodiazepines, but benzos will fail approximately 30-40% of the time. (Appleton 2008) Therefore, an effective second line agent is essential.
Despite widespread excitement about levetiracetam (brand name Keppra), all RCTs to date have indicated that it is no better than existing options such as phenytoin. (Mundlamuri 2015; Chakravarthi 2015; Gujjar 2017) In fact, one trial was stopped early because levetiracetam could not beat placebo. (Navarro 2016) However, the previous trials were too small to be conclusive, and there are many theoretical arguments as to why levetiracetam should be better than phenytoin, so the ConSEPT and EcLiPSE trials were highly anticipated in the pediatrics world.
Paper #1: ConSEPT
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
This is a open-label, multicentre trial conducted in 13 emergency departments in Australia and New Zealand.
This was a superiority study, with the hypothesis that levetiracetam would be better than phenytoin.
Children 3 months to 16 years in convulsive status epilepticus who had already received 2 doses of benzodiazepine.
- Convulsive status epilepticus defined: being unresponsive with continuing abnormality of movement (increased tone or jerking) for longer than 5 min; or two or more recurrent convulsions without recovery of consciousness between convulsions; or three or more convulsions within the preceding hour and a current convulsion.
- Exclusions: Previously enrolled in the trial, already on levetiracetam or phenytoin, received second line antiepileptics in the last 24 hours, know contraindication or allergy to a study drug, a treatment plan stating refractory to phenytoin, head injury, or pregnancy.
Levetiracetam 40 mg/kg IV or IO over 5 minutes. (If still seizing 5 minutes after the infusion, phenytoin was given.)
Phenytoin 20 mg/kg IV or IO over 20 minutes. (If still seizing 5 minutes after the infusion, levetiracetam was given.)
The primary outcome was cessation of seizure activity 5 minutes after the infusion of the trial drug.
They enrolled 234 patients, out of 639 candidates. 127 were missed. 278 were excluded, primarily because they were already on the study drug, or they had a treatment plan excluding phenytoin.
There was no difference in the primary outcome. Clinical cessation of seizure activity was seen in 60% of the phenytoin group and 50% of the levetiracetam group (ARR -9.2%, 95% CI -21.9 to 3.5%; p=0.16).
None of the secondary outcomes were different. (The point estimates were almost all on the side of phenytoin being better).
Median time to cessation of seizure from the commencement of the study drug was not changed: 22 minutes with phenytoin and 17 minutes with levetiracetam (difference 5 min; 95% CI -13.5 to 3.5 minutes; p=0.25).
Adverse events were complicated, because a large proportion of the population received both study drugs. There were no differences between phenytoin and levetiracetam. Adverse appear to be higher when the two drugs are combined (although that is an association only, with clear confounders).
There was one death in the study. The patient was in the phenytoin group, but also received levetiracetam, and died 27 days later.
Comments on this paper
This is an unblinded trial with a relatively subjective outcome, and therefore at high risk of bias. Considering the excitement for levetiracetam (and also specifically against phenytoin) I have seen on Twitter, in papers, and at conferences, I would anticipate a clinician bias in favour of levetiracetam. They did video tape about 60% of the patients, and blinded reviewers agreed with clinical assessments, which is reassuring.
The chosen primary outcome doesn’t make any sense to me. It looks at a different time frame for each drug because the infusion time is longer for phenytoin. It would have made a lot more sense to look at the same time for both drugs (say 30 minutes after arrival), or just try to record the total seizure time like the did in EcLiPSE. This probably biases the primary outcome in favour of phenytoin.
There is a reasonable chance of selection bias, both because of missed patients, and because patients were excluded if they had a treatment plan excluding phenytoin (when no such exclusions existed for levetiracetam).
Paper #2: EcLiPSE
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
This is a open-label, multi-centre randomised clinical trial at 30 UK emergency departments.
This is a superiority trial, with the hypothesis that levetiracetam would be better than phenytoin.
Patients 6 month to 18 years with convulsive status epilepticus (generalised tonic-clonic, generalised clonic, or focal clonic seizure).
- Exclusions: absence, myoclonic, or non-convulsive status, infantile spasms, known or suspected pregnancy, contradiction or allergy to a study drug, known renal failure, or already received a second line anti-convulsant during this seizure.
Levetiracetam 40 mg/kg IV over 5 minutes (max dose 2.5 g).
Phenytoin 20 mg/kg IV over 20 minutes (max dose 2 g).
The primary outcome was time from randomisation to cessation of all visible signs of convulsive activity.
404 patients were randomized out of 1432 screened. Only 286 patients were actually treated and included in the analysis.
There was no difference in the primary outcome of time to seizure cessation: 35 minutes with levetiracetam and 45 minutes with phenytoin (p=0.2).
There was also no difference in the total number of patients treated successfully (70% with levetiracetam and 64% with phenytoin).
There was also no difference in need for further anticonvulsants (37% vs 37%), RSI (30% vs 35%), or admission to critical care (64% vs 54%).
There was no difference in adverse events (12% with levetiracetam and 14% with phenytoin). There were 5 serious adverse events (3 in 2 patients receiving phenytoin, 1 in a patient receiving levetiracetam, and 1 in a patient that received both drugs).
Comments on this paper
Once again, this is an unblinded study with a relatively subjective outcome, so it is high risk for bias.
About ⅕ of the patients in this trial had focal seizures, which probably represent a different class of disease, potentially requiring a different treatment algorithm. About ⅕ of these patients were already on levetiracetam, whereas none of the patients were taking phenytoin, which could bias the results. Less than 20% of patients screened were included, potentially limiting generalizability, and introducing selection bias.
For patients who were intubated, they “censored” the data and assumed that the time to seizure cessation was 12 hours after the RSI. I am not sure that makes a lot of sense, and considering that 30-35% of the patients were intubated, a lot of the times reported in this trial are made up, which could significantly impact the results. Given that intubation occured in 5% more patients in the phenytoin group, that could significantly increase the times reported in the phenytoin group. All times reported probably need to be taken with a grain of salt.
They found no difference in adverse events, but the trial was not powered for adverse events. They try to determine which adverse events were “treatment related”, but in an unblinded trial, that is unlikely to be a reliable process.
These trials resulted in a lot of excited cries for levetiracetam on social media, but I think it is essential to highlight that these are both negative trials. These trials were superiority trials, because people were convinced the newer (more expensive drug) was going to be better, which means they were not designed to support the claims of equivalence that are now being made. There is very little here that suggests we should change our standard treatment algorithms from phenytoin to levetiracetam.
Equivalence, superiority, and non-inferiority
Superiority trials are not the same as non-inferiority trials. Both of these trials were designed with the hypothesis that levetiracetam would be better than phenytoin, and both were negative. Levetiracetam is not better than phenytoin.
Because of that result, a large number of people are interpreting this trial as demonstrating that the two drugs are equivalent, but that is not what these trials show. In fact, I think it is pretty clear from the numbers that if you had designed these trials as non-inferiority trials (trying to prove that levetiracetam is not inferior to phenytoin) they would have failed.
A noninferiority trial has to pick a threshold of acceptable difference between the two drugs being studied. With status epilepticus, that threshold should probably be pretty small, but smaller numbers need larger trials, so my guess is that a number of 5% or 10% would have been picked. In other words, we would have been willing to accept that levetiracetam is non-inferior to phenytoin if the trial showed that it was not going to be 5% worse than phenytoin (including the 95% confidence interval). However, in ConSEPT levetiracetam was 10% worse than phenytoin, with a 95% confidence interval that means it could be as much as 22% worse. And those are absolute numbers! There is no way we can conclude it is non-inferior.
A quote from that trial: “Thus, failure to find a difference does not mean that levetiracetam is statistically equivalent to phenytoin.” (Dalziel 2019)
The EcLiPSE numbers are closer. They don’t present the 95% confidence intervals for the primary outcome, but the sample size is small enough that when I run the numbers, I am pretty sure it would also have failed as a non-inferiority trial.
Of course, I am making theoretical calculations on trials not designed for those calculations. The bottom line is that it is incorrect to interpret these trials as demonstrating equivalence between these two agents. The conclusion of these trials is that levetiracetam is not better.
Neither of these studies looked a a truly important outcomes. Although longer seizures are associated with worse outcomes, time to seizure cessation is a surrogate outcome. One can easily imagine a drug that stops seizures earlier, but actually results in worse neurologic outcomes. Ideally, status epilepticus studies should look at long term survival and neurologic outcomes.
If you have been following the debates on twitter, you will know that I think there is a role for much earlier use of anesthetic agents, like propofol, for patients with active ongoing seizures. If you focus only on time to seizure cessation as a surrogate outcome, I think it is very clear that propofol will beat both of these agents. That doesn’t necessarily meant that it is better, which is why I think we need an RCT that looks at long term neurologic outcomes in these patients.
Not all status is created equal
In my experience, there is a wide range of presentations of status epilepticus. Although they are both defined as status epilepticus, there is a big difference between a patient whose convulsions never stop, and a patient who stops with benzos but never full wakes up, and has another seizure 20 minutes later. I think those patients may require different management strategies, but both are mixed in this study population. Furthermore, EcLiPSE included patients with focal seizures, which probably also changes my treatment approach. There is also probably a difference between a child who regularly has 25 daily seizures, and a child presenting with their first ever seizure. I think these considerations are important when trying to apply the results of these trials.
One of the main concerns about phenytoin are the side effects, and especially reported deaths after dosing errors. These are valid concerns, and grounded in reasonable physiology, but require careful assessment. Phenytoin has been used much more widely than levetiracetam, resulting a in greater opportunity for reported adverse events, and therefore possibly biased literature.
Of note, there was no difference at all in adverse events in these two RCTs. Although RCTs will often overlook harm, these trials were specifically looking for harms, and it is reassuring that no harm was seen among the couple hundred patients studied. If there is harm, it is likely to be occurring in relatively small numbers (requiring more than a few hundred people in an RCT to demonstrate).
I think it is reasonable to look at observational data on the harms of phenytoin when making a decision, but we have to remember that observational data is often biased. I think it is also important to be clear that we are choosing levetiracetam based on observational data, rather than pretending that these RCTs were strong evidence in favour of the drug.
One of the reasons for excitement about levetiracetam is ease of administration. That is an important consideration. In resuscitation, I always want to use the easiest, most foolproof medication.
However, I think we should be somewhat careful about our conclusions. They measured the time to start the infusion in EcLiPSE, and it took the exact same length to get both drugs started.
People often worry about the length of time it takes to administer phenytoin, but administration time doesn’t really matter. Time to seizures cessation is what matters. (As an aside, this is one reason fosphenytoin never made much sense to me. It can be given faster, but it is a pro-drug, and so time to therapeutic levels is not changed at all). These trials did not demonstrate a difference in time to seizure cessation, so who cares how long the drug takes to give?
There are other reasons that levetiracetam is easier to give, but it we are going to change practice simply for convenience, we probably also need to consider costs. According to drugs.com, a dose of levetiracetam IV will cost about $550 USD. A similar dose of phenytoin seems to cost between $30 and $80 USD, depending on the source. Is the convenience worth a 5 to 10 fold increase in price?
Not the question I would ask
The most important thing to consider whenever assessing a trial is probably: is this the question I really want answered? There is no doubt that we needed trials looking at levetiracetam, especially considering the incredible hype this agent has in the medical community. However, this is probably not the question I wanted answered. We knew that second line anticonvulsants have a moderate effect, at best, and they all take a long time (in critical care time) to work. Leaving patients seizing for 20-30 minutes while loading these agents never made a lot of sense to me, and numerous sources have suggested earlier use of anesthetic agents in the management of status epilepticus. (Marik 2004; Millikan 2009; Glauser 2016) I would have loved to see a third arm looking at an agent like propofol.
Using time to seizure cessation as a surrogate, I think it is highly likely that propofol would beat both of the agents. However, that doesn’t mean it is the better agent. The trial I want to see would include a propofol arm (and maybe a ketamine arm), but should look at longer term neurologic outcomes. This is the most controversial of my views, and I will follow it up with its own blog post, and probably a debate with Damien Roland, sometime in the near future.
Despite the excitement, these trials do not provide strong evidence in support of levetiracetam. These trials say that levetiracetam is not better than phenytoin. They also don’t show any difference in adverse events. This is consistent with all prior RCTs.
We should be careful not to claim that these trials demonstrate that the two agents are equivalent, because they do not.
I will probably continue to use phenytoin or valproate, until someone can show me any good science that demonstrates benefit from this new, highly hyped agent. (Although, my treatment is often decided by local protocols, rather than pure science). You may have other reasons to choose levetiracetam, and you should always consider all available evidence as well as your own clinical expertise when making a decision. There is nothing here that says you shouldn’t choose levetiracetam, but it is a more expensive agent, with less overall clinical experience, and no evidence of superiority.
There isn’t enough evidence to be definitive either way, but I remain skeptical of the hype surrounding levetiracetam.
Appleton R, Macleod S, Martland T. Drug management for acute tonic-clonic convulsions including convulsive status epilepticus in children. The Cochrane database of systematic reviews. 2008; [pubmed]
Chakravarthi S, Goyal MK, Modi M, Bhalla A, Singh P. Levetiracetam versus phenytoin in management of status epilepticus Journal of Clinical Neuroscience. 2015; 22(6):959-963.
Chin RF, Neville BG, Peckham C, et al. Incidence, cause, and short-term outcome of convulsive status epilepticus in childhood: prospective population-based study. Lancet (London, England). 2006; 368(9531):222-9. [pubmed]
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;
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]
Gujjar AR, Nandhagopal R, Jacob PC, et al. Intravenous levetiracetam vs phenytoin for status epilepticus and cluster seizures: A prospective, randomized study. Seizure. 2017; 49:8-12. [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;
Marik PE, Varon J. The management of status epilepticus. Chest. 2004; 126(2):582-91. [pubmed]
Millikan D, Rice B, Silbergleit R. Emergency treatment of status epilepticus: current thinking. Emergency medicine clinics of North America. 2009; 27(1):101-13, ix. [pubmed]
Mundlamuri RC, Sinha S, Subbakrishna DK. Management of generalised convulsive status epilepticus (SE): A prospective randomised controlled study of combined treatment with intravenous lorazepam with either phenytoin, sodium valproate or levetiracetam–Pilot study. Epilepsy research. 114:52-8. 2015. PMID: 26088885
Navarro V, Dagron C, Elie C. Prehospital treatment with levetiracetam plus clonazepam or placebo plus clonazepam in status epilepticus (SAMUKeppra): a randomised, double-blind, phase 3 trial. The Lancet. Neurology. 15(1):47-55. 2016. PMID: 26627366
Raspall-Chaure M, Chin RF, Neville BG, Scott RC. Outcome of paediatric convulsive status epilepticus: a systematic review. The Lancet. Neurology. 2006; 5(9):769-79. [pubmed]
Justin Morgenstern. Levetiracetam versus Phenytoin in Status Epilepticus (ConSEPT and EcLiPSE), First10EM, 2019. Available at: