This is an extract of the talk I gave at #DFTB19 highlighting an important research ethos.
“The Doctor” is a painting by Luke Fildes and was first exhibited in 1891.
The artist had lost his son Philip at the age of one, and the scenes reflect the admiration that he had for the doctor who had looked out for him.
For some, the painting may represent a stereotypical view of medicine in the past—the doctor rubbing his chin in a wise fashion, the child prostrate on a makeshift bed. And there is a parent figure in the background, watching on anxiously.
This painting has recently had a revival despite being over 100 years old. It highlights the triad of care we all know exists in paediatrics—the child, the parents and carers, and ourselves.
This triad has received increased attention recently. There is a need for child-centred care with respect to their engagement and involvement in their care. There is a need for positive communication with families; we remember the cases where parents haven’t acted as their child’s advocates but forget the vast majority of cases when they have. We often let parents down when we should have been, not just listening to them but honestly hearing what they were saying. And most recently, the doctors themselves. An understanding of the importance of well-being and the shackles of rudeness.
There is a fourth component, as well, one that perhaps will never get the attention it deserves because it isn’t a visceral part of our clinical care. It’s something we know exists but are quite willing to ignore. It’s something that perhaps has more impact on our practice than we would like to admit. It’s the variability in the actual care or treatment we provide or the fact that it might not be necessary at all.
When I became chair of PERUKI, Paediatric Emergency Research United Kingdom and Ireland, the international sibling of PREDICT and daughter of PERN, I had a personal vision that I would drive the organization forward in delivering ground-breaking new research highlighting novel interventions that would really make a difference to patients. What actually occurred is that I have realised that perhaps PERUKI has an even more important role. One that obviously includes the need to develop, innovate, and implement, but one that also highlights where we could and should do better. It’s some examples of variation and the need for no treatment I would like to share.
So, this is an original selection of PERUKI members and those who helped us get PERUKI off the ground. I’d like to pay particular tribute to Mark Lyttle at this point, who has worked tirelessly at the outset to drive forward many early projects and is consistently named checked by our research partners for his ceaseless enthusiasm for collaborating and engaging. PERUKI took part in a prioritisation process published in 2015, with members putting forward their preferred research agendas and PERUKI publishing the top 20 via a Delphi process.
Number 4 on this list was: What is the best IV medication for Acute Asthma? PERUKI started this work with essentially a two-phase examination of wheeze management in March 2013. In the first phase, a written questionnaire was undertaken. PERUKI sites responded as departments, and 183 consultants responded individually on their wheeze management.
In the study, 99 (54.1%) used salbutamol as first-line intravenous therapy, 52(28.4%) magnesium sulfate, and 27 (14.8%) aminophylline; 87 (47.5%) gave these sequentially depending on response, and 30 (16.4%) gave them concurrently. Overall, 146 (79.8%) continued inhaled bronchodilators while on intravenous therapy.
When commencing intravenous bronchodilators, there were 10 different infusion rates, with over a 10-fold variation between the lowest and highest.
Everyone tends to have little foibles about which treatment they prefer. Given the range of phenotypes and genotypes that exist in our wheezy cohort, it can’t be the case that there is only going to be one best-fit treatment for all patients. But a 10-fold difference probably pushes the bounds of flexibility.
What makes this more interesting is the second study. Also completed at the time (March 2013) was a prospective observational study. Data was screened from all patients presenting with wheeze, and a detailed proforma was completed for those who received intravenous therapies. Of 3238 children, 101 received intravenous therapies. Intravenous magnesium sulfate (MgSO4) was used in 67 (60.9%), salbutamol in 61 (55.5%), and aminophylline in 52 (47.3%) of cases.
In 35 cases (31.8%), two drugs were used together, and in 18 cases (16.4%), all three drugs were administered.
More than half used salbutamol as the first-line intravenous agent, while fewer preferred magnesium sulfate or aminophylline, suggesting equipoise regarding which is most efficacious. To investigate this, participants were asked whether they would enrol patients on a randomised controlled trial allocating salbutamol, aminophylline or magnesium sulfate as the first-line intravenous agent, to which 148 (80.9%) responded positively. Asking clinicians who regularly prescribe acute medications is vital for the study design and subsequent implementation of the study findings. With all due respect to respiratory paediatricians, the question they may be interested in or want to explore may well be completely out of keeping with the practice habits of emergency and acute paediatricians. PERUKI have welcomed increased engagement with our speciality colleagues in the last year, and we hope we will reap the benefits of this.
This is a clear example of variation, and I feel uncomfortable. Is there any reason to believe this variation has improved six years on? We have a challenge as the evidence base is not as strong as we would like. We look to Simon Craig and his work on developing asthma outcomes here—a PERN study I am very proud that PERUKI is part of.
So what about where we think there is only a small amount of variation (a nationally agreed algorithm for example). Do we need to improve practice, and CAN we improve practice? The EcLIPSE study was published a mere month ago, and I am proud of the Don’t Forget the Bubbles team for being part of the process of sharing this information widely. The Eclipse study compared levetiracetam and phenytoin in the treatment of status epilepticus. It was published on the same day as the ConSEPT trial, which was a similar study from our PREDICT friends. The EcLIPSE paper is available for open access, and there is a Don’t Forget the Bubbles summary. I also recommend the reviews by Justin Morgenstein and Casey Parker.
The primary outcome was time from randomisation to cessation of all visible signs of convulsive activity, defined as cessation of all continuous rhythmic clonic activity, as judged by the treating clinician.
Much debate has centred on what EcLIPSE and ConSEPT showed, and at the heart of this is the difference between superiority and non-inferiority.
If these studies do nothing else, it will be to spread the word about this construct. It is really important that people don’t glaze over or think because this terminology is used, it’s someone else’s problem to analyse. I think this undue deference to academics probably perpetuates variation in care. I am not saying the theory is easy, but neither is managing a sick neonate with congenital heart disease, and we completely commit ourselves to doing that.
Superiority trials aim to demonstrate that one intervention is better than another. Statistics, by convention, dictate that a difference between the interventions needs to be defined. In the case of EcLIPSE, because phenytoin stopped status 60% of the time and it was felt Levetiracetam may terminate seizures at a 75% rate, the statistics calculated that 140 patients would be needed in each group. If a difference exists, this difference is likely to be real and not by chance alone.
If they had wanted to show that levetiracetam was only 1% better, then thousands of patients would probably have been needed. If there was no difference by chance, it would easily be possible that levetiracetam happened to be 1% better in that cohort of patients.
A few interesting facts come out of EcLIPSE.
The first is that while this wasn’t a perfect observational study—i.e., not all patients presenting were recruited across a wide range of hospitals—over 1,400 patients were screened. This is a good cohort of children with seizures. About 10% of those who needed second-line treatment for status were first presentations of afebrile convulsions, and 5% were as a result of CNS infection.
The median time from randomisation to the start of infusion was 11 min (IQR 8–15) for levetiracetam and 12 min (8–17) for phenytoin.
However, the median time from randomisation to seizure cessation was 35 min (IQR 20 to not assessable) in the levetiracetam group and 45 min (IQR 24 to not assessable) in the phenytoin group.
These interventions take time!
In EcLIPSE, convulsive status epilepticus was terminated by levetiracetam in 106 (70%) participants and by phenytoin in 86 (64%) participants. Therefore, according to the statistics, LEVETIRACETAM is NOT better.
Because the results are broadly the same, it doesn’t mean they are equal – a non-inferiority study looks at two drugs. It aims to calculate what is the minimum number of patients needed to be recruited into each intervention arm to demonstrate that one drug is not more than a certain % worse than another. By convention, that number is normally 10%. The reason why 10% is important is that in EcLIPSE. At the same time, it appears levetiracetam may have passed this test if the study had been designed as a non-inferiority in the ConSEPT study, levetiracetam only terminated seizures (albeit as a different endpoint), 50% of the time; 10% worse than phenytoin. We don’t know yet what the meta-analysis may show us, but this is planned.
A further suggestion is whether we should consider adding levetiracetam with phenytoin; we could, but that might delay some RSI intervention even further without overall benefit in seizure termination. This is a messy area where the complexity of clinical practice hints at the required precision of research head-on.
It might well be that you are happy for others to research novel drugs and techniques. You may well be content in supporting research through signposting or perhaps recruiting patients yourself. I would ask, though, that research itself is not scary. There is a false divide between the ivory tower academic and jobbing clinician. Both these terms are probably tribal and derogatory in their way. We should all care about how effective our treatments are and where variability in practice is not in the patient’s interest. It is no more or less important than the three figures in Luke Fildes’s picture, but perhaps it is less visible.
Through PERUKI, I’d like to champion this cause to make research more accessible. We are not doing research because we like to; we are doing it because we have to.
