The paediatric "long covid liver" study is being interpreted wrong
Confirmation bias is a dangerous thing
A new study titled, “Long COVID-19 Liver Manifestation in Children” has garnered a lot of attention, in view of the recent cluster of children in the UK and elsewhere with acute, severe hepatitis (liver inflammation) of unknown origin, which some have theorised could be linked to Covid-19 (the current primary hypothesis is that it is linked in some way to Adenovirus, which has been identified in a substantial number of cases).
This new study provides almost no useful information in relation to the current cluster under investigation. Allow me to explain why.
What is the study?
This is a case series of 5 children from Israel who at some point tested positive for Covid-19, and subsequently experienced acute, severe hepatitis. Two children aged 3 and 5 months required transplantation, and the remaining 3 (two aged 8 years and one aged 13 years) had resolution of disease following treatment with steroids (it’s unknown if the steroids contributed to the resolution).
Who are the cohort?
The study falls at the first hurdle. There is no definition of who made it into this study and why, other than it is 5 children who had acute severe hepatitis during 2021 and previously had Covid-19.
Is this all the children who had severe, acute hepatitis during this time period? Is this only the children who had some previous history of Covid-19? If so, how many children had this condition who did not have a history of Covid-19?
Acute, severe hepatitis with unknown cause is uncommon but is well recognised, and in fact is responsible for around 40% of cases of liver failure in children. We would expect to have cases like this anyway with or without Covid-19 around, and we would expect a significant number of these children to have tested positive for Covid-19 at some point, given much of the population has been infected. If there were another 10 children who had this condition at this centre with no history of Covid-19, then we would wonder if these positive tests had any relevance at all.
How do these compare to the current UK cluster?
The cohort described here does not match well at all with the current UK cluster under investigation. Of children in the current UK cluster 50% are aged between 2-3 years (and across Europe 78% are < 5 years). None of the children in this paper are in the main age bracket of 2 - 3 years, and most of them are older than the majority of patients in the current cluster (three of them are 8 years or older).
Also, the exceedance in cases of hepatitis with unknown cause (meaning there being more cases than we would expect) has only been since January 2022, and mostly during March to April. This entire cases series is during 2021.
Are these cases actually due to Covid-19?
This is where things get very tricky. Trying to determine a causal connection between these cases and the Covid-19 diagnosis is far from straightforward. Post viral hepatitis is a recognised condition in children, but the mechanism is not well understood. The problem here is that Covid-19 infection has been so common that coincidences are also very common.
Establishing the link between MIS-C and Covid-19 was much easier, because at the time no one had been infected before, so when 100% of the patients with the condition had a positive test and the epidemic wave matched almost perfectly 4-6 weeks after the Covid-19 wave, this made a compelling case.
Take an alternative example of “covid-toes”. The sudden exceedance in cases of chilblains, mainly affecting the toes, was considered secondary to Covid-19 infection. As it turns out, most people who had the condition in this case series had no evidence at all of previous infection. It was never due to Covid at all.
A series of 5 patients, some of whom had hepatitis more than 3 months after their initial infection, does not establish causality. In fact, this case series doesn’t provide any evidence at all that the cases are related to the Covid-19 diagnosis other than in a two of them, there is a close temporal relationship (for the remainder, the duration is unknown, 90 days and 130 days).
As hepatitis following acute viral infections is a recognised cause, it is likely that some people will experience this after Covid-19, as they would after other viruses. We certainly can’t rule out Covid-19 as the cause in these cases, but in the absence of mechanistic evidence or a good matched control series, we cannot be sure either way.
Incidentally, of the three children who were tested for Adenovirus one was positive.
Why are people so interested in this study?
There is a lot of interest in the current cluster of cases of acute severe hepatitis being observed around the world, and there are some people who are convinced that these cases must be related to Covid-19. Any evidence which looks to support that notion is therefore of exceptional interest, no matter the quality of that evidence. This is known as “confirmation bias”, where we pay excessive attention to anything which we feel confirms what we already believe to be true, whilst ignoring factors which might go against our hypothesis.
There are valid hypotheses for how Covid-19 could potentially be linked to the current cluster, but there is no evidence to support them at this point in time. It remains under investigation, and we should definitely remain open to the possibility. That said, we must also try to avoid jumping prematurely to conclusions, or only focussing on evidence which we feel supports that which we already believe.
This is a small study of a series of children who had Covid-19 at some point, and later developed acute severe hepatitis which may or may not have been related. This case series does not look similar to the cluster under investigation in the UK, and does not provide any evidence that the current exceedance of acute, severe hepatitis is due to Covid-19. That does not mean that Covid-19 is not involved, but this study does not bring us any closer to understanding the current international cluster, and we still lack evidence of a mechanism by which this could occur.
For further reading, please see below: