JC: Enter Sandman – Which Agent as Second Line in Paediatric Status Epilepticus?

I first heard about the EcLiPSE trial into second-line agents for paediatric status epilepticus back in 2012 when the proposed trial protocol was presented at the APEM meeting in Bristol, UK:

It wasn’t long before EcLiPSE’s Australasian cousin, ConSEPT, was registered at ANZCTR (August 2013, to be exact​1​) – and after what has felt like a long wait, both studies’ findings were published together in the Lancet earlier this month.

You can find EcLiPSE here​2​ (it’s open access) and ConSEPT here​3​ (not open access) and as always we would encourage you to read the original articles and form your own appraisal and opinions before reading further.

What are these trials all about?

Both EcLiPSE and ConSEPT are interested in the pharmacological agent used as a second-line medication in paediatric status epilepticus.

Traditionally in the “fitting child” algorithm, benzodiazepines are the first line medication used in an attempt to stop the seizure. The algorithm differs slightly between APLS UK (the 6th edition algorithms aren’t open access online but a reproduction is included in this policy document from Leicester​4​) and APLS Australia​5​, predominantly in the timings used between doses.

Midazolam, diazepam or lorazepam are given via IV, IO, IM, buccal or intranasal routes with a second dose either five (for IV/IO administration) or ten (for buccal, rectal or intranasal routes) minutes later in Australia, and five minutes later irrespective of route on the UK 6th edition of APLS. Following the second dose of benzodiazepine, the algorithm prompts the clinician to start preparing a second line agent – which, in both cases, is phenytoin if intravenous or intraosseous access has been obtained.

Quite aside from these differences, there are other issues with the way that status epilepticus management plays out in paediatric patients in both the Emergency Department and in the prehospital setting.

Some patients do not receive prehospital benzodiazepines at all; others receive more than two doses either prehospital or in ED or between the two – presumably this is deliberately undertaken because phenytoin is not available, but repeated doses come with a corresponding increase in associated respiratory depression. In other cases there are long delays between doses, far larger than those advised in the algorithm. It’s reasonable to think that either circumstance might lead to an increased likelihood of intubation and ventilation, and that certainly corresponds to my experiences. This open access paper from a UK PICU​6​ focused particularly on issues with benzodiazepine administration exactly as described above.

Phenytoin is a bit of a pain to administer, as our ED nursing colleagues can readily tell us. It can only be administered by infusion (maintenance of which has its own challenges in the fitting child) over 20mins; there are concerns about cardiovascular side effects of faster infusion rates although the evidence base for this is not particularly robust​7​.

Harder to evidence is the omission of phenytoin altogether, but it is certainly my experience that this occurs (and I have heard this echoed by colleagues in NETS NSW, for whom I work casually as a paediatric retrieval registrar). Whether it is the trickiness of administration, or a perception that the child has been fitting “too long already”, or a belief that if benzos haven’t worked then only RSI will be successful, I am aware of a number of cases where a second line agent has been omitted and the team has moved straight to induction of anaesthesia and intubation.

Enter levetiracetam, or Keppra (the trade name by which it is often known and which is far easier to spell and pronounce). Since levetiracetam’s introduction onto the market we have seen its use expanding into spaces traditionally filled by phenytoin or other antiepileptic drugs, particularly for adult patients.

So what happened in these studies?

In my own appraisal of these studies, I’ve put together a simple abstract-type summary of each paper (a hangover from FRCEM exams). Hopefully these highlight the methodological differences between the two papers and their results.


Design: open-label, multi-centre (30 UK Emergency Departments though the PERUKI network​8​), randomised controlled trial

Objective: to determine which is more effective and safer as a second line agent

Population: 6months-18yrs, convulsive status (generalised tonic-clonic, generalised clonic, focal clonic) requiring second-line agent. Excluded if: other seizure type, pregnancy, renal failure, had received second line agent already, previously enrolled. Also excluded after randomisation if seizure stopped and treatment was not given.

Computer generated randomisation, blocks of 2-4. No blinding.

Interventions: randomly assigned to 40mg/kg keppra over 5mins vs phenytoin 20mg/kg over 20mins via IV route only

Primary outcome: time from randomisation to cessation of convulsive activity, determined by clinician.

Key Results: 152 received keppra, 134 received phenytoin

Terminated by Keppra in 106 (70%) and phenytoin in 86 (64%)

Median time from randomisation to commencement of second line agent: 11min for Keppra (IQR 8-15), 12min for phenytoin (IQR 8-17)

Median time to cessation of seizure from randomisation: 35 min for Keppra (IQR 20 to ??), 45min for phenytoin (IQR 24-??)

Authors’ Conclusions: did not show Keppra to be superior, however in combination with other data and ease of administration it may be an appropriate alternative


Design: open-label, multi-centre (13 Emergency Departments in Australia and New Zealand), randomised controlled trial

Objective: to determine which is better second-line treatment for emergency management of convulsive status epilepticus

Population: 3months-16years, convulsive status with two doses benzodiazepines given. Excluded if: previously enrolled, on maintenance of either agent, second line agent in last 24h, management plan excluding phenytoin, allergy/contraindication to either agent, head injury, eclampsia.

Computer generated randomisation, blocks of 2-4. No blinding.

Intervention: randomly assigned to 40mg/kg keppra over 5mins vs phenytoin 20mg/kg over 20mins IV or IO

5mins after completion, if still seizing, other drug was given

If still seizing 5mins after other drug, RSI performed.

Primary outcome: clinical cessation of seizure activity five mins after completion of infusion of the first trial drug (10mins for Keppra, 25mins for phenytoin).

Key Results: 119 received Keppra, 114 received phenytoin

Cessation within five mins – 68 (60%) phenytoin group, 60 (50%) Keppra group

Authors’ Conclusions: Keppra is not superior to phenytoin. Consider sequential use to reduce failure rates.

What do the results mean?

The two papers both found insufficient evidence to demonstrate levetiracetam’s superiority over phenytoin as a second-line agent in that small group of patients who continue to fit after benzodiazepines. It’s worth noting that this really is a subgroup; for example, the EcLiPSE study had 1432 patients presenting in status epilepticus and potentially eligible for inclusion, but only 404 made it to randomisation. This also doesn’t capture the patients whose fits stopped before arrival into hospital, so the need for a multi-centre trial over a long recruitment period is easily explained.

There was slightly different methodology between the two papers and their results weren’t directly comparable: EcLiPSE looked at time from randomisation to seizure cessation, while ConSEPT looked at proportions stopping seizing within five minutes of completion of the infusion. With different infusion rates between the two agents, this is an interesting take and would theoretically give phenytoin longer to have an effect (since the infusion takes longer to administer). That does rather leave us with the question: if levetiracetam’s quicker administration is an advantage, how long do we wait before deciding it hasn’t worked? Traditionally in APLS guidelines, the phenytoin administration time was used to prepare for RSI as the next step. I think that in the real world, it’s unlikely that a clinical team will be ready to perform an RSI five minutes after starting levetiracetam. It’s interesting that the time to administration is similar for both agents in the EcLiPSE trial as this somewhat contradicts my perception that phenytoin is a pain to give, but I wonder whether this is Hawthorne effect/selection bias (more below).

Ummm… Superiority?!

There’s a misconception around superiority trials (which both of these studies were). Superiority, non-inferiority and equivalence trials are different beasts and the burden of statistical “proof” is different for each entity. In practical terms, this means that different numbers of subjects are required; the trade-off (and a common misunderstanding) is that when a superiority trial has a non-significant result, we cannot conclude either equivalence or non-inferiority. For these two studies, this means that the insufficient evidence of superiority of levetiracetam does not mean it is “as good” as phenytoin.

This paper​9​ is a bit intense but the abstract summarises the difference nicely:

When the aim of the randomized controlled trial (RCT) is to show that one treatment is superior to another, a statistical test is employed and the trial (test) is called a superiority trial (test). Often a nonsignificant superiority test is wrongly interpreted as proof of no difference between the two treatments. Proving that two treatments are equal in performance is impossible with statistical tools; at most, one can show that they are equivalent. In an equivalence trial, the statistical test aims at showing that two treatments are not too different in characteristics, where “not too different” is defined in a clinical manner. Finally, in a non-inferiority trial, the aim is to show that an experimental treatment is not (much) worse than a standard treatment.

Lesaffre, E. Superiority, equivalence, and non-inferiority trials. Bull NYU Hosp Jt Dis. 2008;66(2):150-4.

If this is still too much for you, there’s a more plain-language example here. And if your head is still spinning after that, take a moment to appreciate the role of statisticians in clinical research!

So Should We Change Practice?

So, we have insufficient evidence of levetiracetam’s superiority – is there anything else that might influence our decision of which agent to use?

Crucially, levetiracetam is easier to administer to phenytoin; it is compatible with either saline or 5% dextrose and while most documentation suggests it should be given over 15min, both EcLiPSE and ConSEPT administered the drug over 5mins. Access in children can be difficult at the best of times, let alone with ongoing seizure activity, so being able to give an infusion quickly while a line is patent is a definite advantage (if you don’t believe me, ask your nursing colleagues – or think of how many times you’ve been told halfway through an aciclovir infusion that the cannula has stopped working).

Theoretically, the difference in administration should translate to a longer time to administration of phenytoin in the studies. While the evidence for this isn’t in the papers, I suspect that’s at least in part because the departments were set up to administer phenytoin. We can almost see this as a selection bias or Hawthorne effect within the paper – for a child to be included in the trial, they had to be in status epilepticus in an ED that was prepared to administer IV phenytoin. And from my (anecdotal) experience, the difficulty around administration of phenytoin is often prohibitive to its administration at all, with an often articulated belief that “it doesn’t work” (both of these studies certainly show that it does work!). We see clinicians skipping the second-line agent and moving straight to RSI – which is not necessarily a smooth process for those same departments who are unprepared to administer phenytoin, unused to its administration or for whom this is a rarely encountered situation. Perhaps these department are more comfortable with paediatric RSI – but it’s a reasonably rare event even for prehospital services like the one I work for​10​, so I would be surprised if this was true.

For clinicians working in prehospital care, as I do, the difference between administering levetiracetam and phenytoin is not trivial and perhaps that is why I hold these opinions; my colleagues at Sydney HEMS are undoubtedly more comfortable with the care of adult patients and in trauma and a case discussion at our joint HEMS/NETS education day around a child in status epilepticus exposed the magnitude of the degree of discomfort.

I am not saying that these papers provide evidence that levetiracetam is safer, but I do think that the practicalities of using these agents in the prehospital environment or in smaller centres should be debated, probably more openly than they have been so far.

So I’m left with these questions:

  • How often do we fail to follow the fitting child protocol, either by giving more benzodiazepines (which leads to respiratory depression and presumably increases intubation rates) or by omitting a second line agent and moving straight to intubation?
  • Is this scenario more likely in non-paediatric centres?
  • Does it cause harm, either because of complications during RSI or because of delays to appropriate therapy?
  • Would having an easier, quicker-to-administer agent negate some of these harms – and if so, is that agent levetiracetam?

I’m not sure the answers to these questions will be forthcoming any time soon – but I suspect we will start to see clinicians reaching for levetiracetam as a second line agent (if they are giving one!) and hopefully more conversation around this particularly sticky problem now that the studies are out.

Finally, hats off to the research teams for trying to answer a relevant but tricky clinical question and giving us plenty of food for thought.

More FOAMed Takes on the Papers

If you want to read more appraisals and opinions on these two papers, try:

Justin Morgenstern’s take at First10EM

Appraisal from the DFTB Team
Casey Parker’s take on the ConSEPT results when presented at DFTB18


Natalie May


  1. 1.
    ANZCTR – Registration. ANZCTR. https://www.anzctr.org.au/Trial/Registration/TrialReview.aspx?id=364600&isReview=true. Published 2013. Accessed May 3, 2019.
  2. 2.
    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. April 2019. doi:10.1016/s0140-6736(19)30724-x
  3. 3.
    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. April 2019. doi:10.1016/s0140-6736(19)30722-6
  4. 4.
    Management of Paediatric Status Epilepticus. Leicester Royal Infirmary. https://secure.library.leicestershospitals.nhs.uk/PAGL/Shared%20Documents/Status%20Epilepticus%20UHL%20Childrens%20Medical%20Guideline.pdf. Published 2019. Accessed May 3, 2019.
  5. 5.
    Status Epilepticus. APLS Australia. https://www.apls.org.au/sites/default/files/uploadedfiles/Algorithms%20-%20Status%20Epilepticus.pdf. Published 2018. Accessed May 3, 2019.
  6. 6.
    Chin RFM. Inappropriate emergency management of status epilepticus in children contributes to need for intensive care. Journal of Neurology, Neurosurgery & Psychiatry. November 2004:1584-1588. doi:10.1136/jnnp.2003.032797
  7. 7.
    Guldiken B, Rémi J, Noachtar S. Cardiovascular adverse effects of phenytoin. J Neurol. December 2015:861-870. doi:10.1007/s00415-015-7967-1
  8. 8.
    PERUKI (@PERUKItweep) on Twitter. PERIUKI. https://twitter.com/PERUKItweep. Published 2019. Accessed May 3, 2019.
  9. 9.
    Lesaffre E. Superiority, equivalence, and non-inferiority trials. Bull NYU Hosp Jt Dis. 2008;66(2):150-154. https://www.ncbi.nlm.nih.gov/pubmed/18537788.
  10. 10.
    Watterson JB, Reid C, Burns BJ, Regan L. Pre-hospital advanced airway management in children: a challenge that training can handle. Scand J Trauma Resusc Emerg Med. December 2017. doi:10.1186/s13049-017-0432-7

4 thoughts on “JC: Enter Sandman – Which Agent as Second Line in Paediatric Status Epilepticus?”

  1. Peter Thomsen

    “And from my (anecdotal) experience, the difficulty around administration of phenytoin is often prohibitive to its administration at all”

    I agree. That was also the take-away from ILAE lectures in status epilepticus in Denmark in 2016. Give the drug, that can be administeret fastest and safest in high enough doses (I have worked in departments, where there was a setup for phenytoin / fosfenytoin, and in my experiance it works as good as the others: valproatic acid and leveteracitam)

    However, I do prefer keppra and valproatic acid as they are faster to give and easier to manage. We’ll have to wait for the ESETT (trial) results https://www.ninds.nih.gov/Disorders/Clinical-Trials/Established-Status-Epilepticus-Treatment-Trial

    All the best
    Peter, Denmark

  2. Pingback: May FOAMed - FRCEM Success

  3. Pingback: SGEM#265: Total EcLiPSE of the Seizure – What a ConSEPT | The Skeptics Guide to Emergency Medicine

  4. Pingback: JC: Top 10 papers 2018-2019 for #RCEM Annual Scientific Conference. St Emlyn's • St Emlyn's

Thanks so much for following. Viva la #FOAMed

Scroll to Top