August 27, 2020

expert reaction to ISARIC 4C study on clinical characteristics of children admitted to hospital with COVID-19 in the UK

Research, published in the BMJ, looked at the clinical characteristics of children admitted to hospital with COVID-19 in the UK.

This Roundup accompanied an SMC Briefing

Dr Liz Whittaker, Infectious Disease Lead at the Royal College of Paediatrics and Child Health, and Clinical Lecturer and Consultant Paediatric Infectious Diseases and Immunology, Imperial College London, said:

“This study comprises a prospective collection of data on children admitted to hospital with confirmed SARS-CoV-2 infection in UK.  Although not comprehensive, the data are very reflective of clinicians’ experience nationally.

“The study is reassuring.  It echoes other studies which have reported low numbers of COVID infection over a 6 month period (651 children out of a paediatric population of 16 million).  Very low numbers of children have been admitted to critical care and the researchers reported a very low death rate – particularly in comparison to adults, but also in comparison to the death rate due to other infections (influenza, chicken pox, meningitis, group A strep sepsis, etc.) and other causes of childhood death (e.g. road traffic accidents).

“The researchers identified a sub-set of the children with the previously described inflammatory condition – PIMS or MIS-C.  They did not identify any new characteristics which would help with diagnosis of this cohort.  As reported in other studies, despite 2/3 of the children with presumed PIMS in this cohort becoming unwell and requiring critical care, the outcomes were good.

“Also as previously described, the children in this study met the WHO criteria (fever, abdominal pain, diarrhoea, red rashes and laboratory evidence of inflammation) and also had non-specific symptoms as previously described in an earlier JAMA paper, such as headache, fatigue, muscle aches.  The children with PIMS were more likely to have those symptoms than children who tested positive for SARS-CoV-2 who didn’t have PIMS – this reflects how well the majority of children with SARS-CoV-2 infection are, as they are unlikely to experience even these commonly described flu-like symptoms.

“Flu-like symptoms are extremely common in children with infections, and were present in a third or less of children with presumed PIMS – and so were not included in the WHO criteria as they do not help clinicians (or parents) distinguish children with the inflammatory condition from those with other common childhood conditions.  The RCPCH has produced guidance for parents about the condition: https://www.rcpch.ac.uk/resources/pims-covid-19-linked-syndrome-affecting-children-information-families.

“As in other studies, those children of Black ethnicity are more commonly affected with the very rare PIMS condition.  This reflects the higher incidence of SARS-CoV-2 infection in these communities.

“On the whole, parents and paediatricians should be reassured by the findings of this paper.”

Prof Kevin McConway, Emeritus Professor of Applied Statistics, The Open University, said:

“This is a good, thorough piece of research.  There are some inevitable limitations in what it could do, but the researchers explain these limitations, and I think the statistical analyses are thorough and appropriate.  The overall finding that children and young people have less severe covid-19 than adults do, if they are infected, is certainly solid, and of course many previous studies have also found this.  I’ll just comment on a few points where care may be needed with the interpretation of some of the results.

“The results relate only to children and young people, with laboratory-confirmed covid-19, who were admitted to 260 UK hospitals.  This means that the study can’t tell us anything direct about the chances of a child being infected with the virus, or the chance of an infected child being admitted to hospital, because the researchers did not have data on uninfected children, or on infected children who were not admitted.  The report tells us that patients under 19 made up less than 1% of all patients admitted to these hospitals with confirmed covid-19.  That does indicate that children and young people are much less likely to be admitted than older age groups, but it doesn’t directly tell us whether that’s because they are much less likely to get infected at all, or that, if they are infected, they are much less likely to become ill enough to need hospital treatment, or some combination of the two.  On the face of it, it might seem quite concerning that 18% of the children in this study were admitted to critical care – but you have to remember that this is 18% of hospitalised children, that is, of children who were already ill enough to have been admitted to hospital.  The vast majority of children and young people have not been admitted to hospital with covid-19, because either they haven’t been infected at all, or, if they have, they were never ill enough to require hospital admission, and this study does not tell us directly about them.  A related but less important issue is that, as the research report makes clear, the data did not come from all UK hospitals, but only those in the UK ISARIC consortium, and their patients may not be entirely representative of patients from all UK hospitals.  But these hospitals do account for two-thirds of all covid-19 admissions in England, Wales and Scotland so the possible unrepresentativeness is unlikely to cause big problems with the interpretation of the results, I’d say.

“The study is observational.  So there’s an issue that, for instance, there will be many differences between the children who were and were not admitted to critical care, and so it’s not possible to tell for sure which of those differences might be causing the admission to intensive care, and which are effectively coincidences.  The same goes for the question of which children and young people developed the inflammatory syndrome (MIS-C).  That’s why the press release from the BMJ says that the study ‘can’t establish cause’.  The researchers acknowledge this, and also they report that they carried out so-called multivariable analyses, where they adjusted their findings appropriately to take account of this issue as far as they could – but that can never make the issue disappear entirely.  The researchers could not make an adjustment for everything, and in particular they could not adjust for deprivation.  So, for instance, they report that children of black ethnicity were more likely to be admitted to critical care.  That finding remains statistically clear in the multivariable analysis, so it did not arise because, for example, only because the black children were in different age groups than the others, or had a different chance of having other serious illnesses.  (They may or may not have differed in those ways, but there was still a marked association with black ethnicity in addition.)  But because deprivation was not included in the researchers’ statistical models, it may be that some of the extra risk of going into critical care for black children is because they might be more likely to live in deprived areas.  I need to point out that other studies involving covid-19 and ethnicity have generally found that findings of worse consequences of the disease in people with black and minority ethnicities are still there, often to a considerable extent, when deprivation and other similar factors are taken into account.  But we don’t know whether that is the case in this particular study, because the researchers could not take deprivation into account in their statistical modelling.

“This issue about observational studies, arising because different factors are interrelated, also comes up in relation to a difference between what the two press releases mention.  The University of Edinburgh release says that obesity is a risk factor for admission to critical care, but the BMJ one does not mention obesity at all.  The researchers did find an association between obesity and the chance of being admitted to critical care in what’s called a ‘univariable analysis’, where obesity and critical care admission were considered without any other factors being considered in the same analysis.  But that leaves open the possibility that this association is really just a consequence of other differences between obese children and others, and not the obesity itself.  In the multivariable analysis of admission to critical care, obesity does not feature as a risk factor, so there’s not really any clear evidence that obesity is really associated with critical care admission risk at all.  (And we should bear in mind that the vast majority of the children and young people in this study were not obese – under 20 out of the 651 in the study were recorded as being obese.)

“A minor point: the University of Edinburgh press release says that the “typical age” of hospitalised children was five years old.  Well, that depends on what you understand by ‘typical’.  The median age was indeed about 5.  That is, if you lined up all the children in terms of age, the child in the middle of the line would be about 5.  But in fact the patients were spread out right across the age groups from birth to just under 19 years old that were included.  More than a third of them were under one year old, and a fifth were aged 15-19, for example.”

Prof Adilia Warris, Paediatric Infectious Diseases Specialist, University of Exeter, said:

“This study shows the huge value of being well prepared for major outbreaks to enable a coordinated response to new questions and to provide answers as soon as possible.  The established International Severe Acute Respiratory Infection Consortium (ISARIC) immediately took action when the first COVID-19 cases were reported, and in the UK, with support of UKRI (MRC), ISARIC 4C (Coronavirus Clinical Characterisation Consortium) was launched to collect prospectively (real-time) clinical data of patients admitted to hospitals in the UK with proven COVID-19.

“The most important results of this multi-centre study (138 hospitals in England, Wales and Scotland) published today, are, that of all patients admitted to hospitals with proven COVID-19 in the first 6 months of 2020, only 0.9% were under 19 years of age (651 out of 69,516 patients), that severe COVID-19 disease was rare in children and young people, and that death due to COVID-19 was exceptionally rare in this age group.

“The researchers have shared an extensive set of clinical data in their paper being extremely helpful to improve our recognition of the different COVID-19 disease types observed in children and young people, to more precisely define the clinical syndromes observed, and to improve our understanding of this new disease.  The way in which this study was set-up, a unique opportunity was provided to objectively monitor the characteristics of COVID-19 disease in children and young people, and allowed to detect the emergence of a novel multisystem inflammatory syndrome temporarily associated with Sars-CoV-2 infection.”

Prof Athimalaipet Ramanan, Honorary Professor of Paediatric Rheumatology, University of Bristol, said:

“This paper is part of an observational study looking at adults and children admitted with COVID-19 to UK hospitals.  The paper highlights the relatively small number of children (651) admitted across UK hospitals in comparison to the significant adult admissions to hospital.  This demonstrates what has been shown in other parts of the world, that the vast majority of children get minimal or no symptoms with COVID-19.

“Amongst the children needing hospitalisation, two groups stand out.  One is those children with pre-existing diseases which make them more vulnerable.  The second is that a tiny proportion of children appear to develop a multisystem inflammatory syndrome which has been recognised in UK since April.  These children usually present with fever and abdominal pain and appear to respond well to treatment, although many seem to need critical care input initially.  This paper suggests that the syndrome relative to the population appears to be more prevalent in BAME children.

“This is an observational study of hospitalised children. The limitations include that not all cases may have been captured.  This also does not account for the true prevalence of significant COVID-19 infection in children.  However, on balance it’s an important study as it is likely most children with significant illness would have presented to the hospitals.”

‘Clinical characteristics of children and young people admitted to hospital with covid-19 in United Kingdom: prospective multicentre observational cohort study’ by Olivia V Swann et al. was published in the BMJ at 23:30 UK time on Thursday 27 August 2020.

DOI: 10.1136/bmj.m3249

Declared interests

Prof Kevin McConway: “I am a member of the SMC Advisory Committee, but my quote above is in my capacity as a professional statistician.”

None others received.