Children are really, REALLY unlikely to die from Covid-19
New data shows the infection fatality rate keeps getting lower
It has been well recognised since the start of the pandemic that children are much lower risk of severe disease or dying from Covid-19 than adults. The vigorous testing performed in many countries and the rarity of bad outcomes in children has posed challenges in interpreting the data around deaths in children from Covid-19, due to the possibility of incidental positive tests (they happened to test positive with Covid-19 by chance when they actually died of something else).
This has led to a number of spurious claims over the risks of death to children, with professional Fact Checking organisations needing to get involved.
A new study out as a pre-print uses detailed analysis of these cases in the UK to shed some new light on the risk of death to children from Covid-19.
Spoiler alert: it is really, really, really low.
What did they do?
Thanks to the nationalised provision of healthcare in the UK, the author group from the UK Health Security Agency were able to link up data from children with a positive Covid-19 test between March 2020 and December 2021 with healthcare data from the NHS. They looked for anyone aged <20 years who died within 100 days of a positive test for Covid-19. The family doctors of the patients were surveyed for clinical details, and hospital discharge paperwork, death certificates and post-mortem reports reviewed. The medical teams who cared directly for the children were contacted where more information was needed.
Decisions on whether Covid-19 contributed to the death was made by 2 independent study team members and conflicts were resolved by author group discussion. Where it was unclear, Covid-19 was assumed to have contributed.
What did they find?
The headline result was that during this period there were 185 deaths within 100 days of a positive Covid-19 test, of which 81 were due to Covid-19. Using estimated Covid-19 infections in each age group during this time period, the highest infection fatality rate (IFR) was in children <1 year (1.7 deaths per 100,000 infections), followed by 16 - 19 years (1.5/100,000), then 12 - 15 years (0.9/100,000) and lowest in 1 - 11 years (0.3/100,000). Every death of a child is tragic, but these numbers are reassuringly small.
Looking at each variant, the risk of death declined quite significantly for each subsequent wave. The IFR for the original virus was 1/100,000, for Alpha was 0.8/100,000 and for Delta was 0.6/100,000 (no data yet for Omicron, but given we know it is significantly less virulent than Delta we can expect this to fall further still).
The overall risk of death for people aged <20 years from Covid-19 during this time period was 0.7 deaths per 100,000 infections (7 per million, or 0.0007%). This is the same as the average risk of death to someone runnjng a marathon, going skiing for 10 days, or going on a return flight from London to New York City.
More than half of all deaths due to Covid-19 occurred within a week of the positive test, and >90% within 30 days. Covid-19 was responsible for 1.2% of all deaths in children during this time period.
What about comorbidities?
One of the most important part of the analysis is regarding the comorbidity status of the children who sadly passed away due to Covid-19. Of the 81 deaths, 61 (75%) occurred in children with significant comorbidities, including severe neuro-disability, immunocompromise, congenital syndromes or chronic heart disease.
Current estimates are that in the UK around 8.3% of school-age children have a medical comorbidity. This means that 75% of the deaths occurred in ~8.3% of the population, which emphasises the importance of vaccination for children with the types of medical problems described above (the reality is that children with these conditions make up an even smaller proportion of the 8.3%). It’s important to recognise that many other respiratory viruses pose a similar threat to children with these conditions, and that vaccination can provide substantial protection from severe disease - even if protection from symptomatic infection wanes over time.
The flip side to this is that the other 25% of deaths occurred in the ~91.7% of children without these comorbidities. We can adjust our overall numbers accordingly using their data, which gives us 20 deaths from 10,664,166 infections (0.2 deaths per 100,000 infections). This is the equivalent to the average lifetime risk of death to the average person from being alive for 2 days.
Risk per infection, or risk from infection?
The reason we focus on the risk conditional on being infected (i.e. what is your risk of dying once you are already infected) is that this is relatively constant and easy to understand. Trying to evaluate an overall risk, including the risk of getting infected, gets really complicated, really quickly. This has changed dramatically over the past 2 years already, and will continue to change with time. It is also heavily dependant on where you live, your behaviours, the behaviours of those around you, and community disease prevalence.
What we can say is that for your child, if and when they do get infected, the risk is extraordinarily low - especially if they have no serious underlying health conditions.
It is possible for a disease with a low severity per infection to cause more harm overall due to lots of infections (for example, Flu causes more harm in the UK than Ebola despite the latter being far more lethal). Comparisons for Covid-19 taking this into account become less and less helpful as we transition out of the acute pandemic phase, until we have settled in to something more like endemic disease (where there is a more consistent and predictable rate of infections year on year), as what was true for that time period is no longer relevant.
What we do know is that almost every child in the UK (and most in the US and Europe) have already been infected at least once, and that this almost certainly represents the infection with the highest risk profile, because prior to vaccination or infection you encounter the infection without protection from immunity.
The first infection carries the highest risk (especially prior to vaccination).
What’s more, we know the risk of the serious multi-system inflammatory syndrome (MIS-C) decreases substantially following vaccination (and almost certainly following infection). Even better news is that the risk of MIS-C has decreased substantially over time, being at its lowest for the Omicron variant (we are unsure why, but it may be to do with the changes of the spike protein of the virus).
This means that the risk per infection at a population level is almost certainly well past it’s peak, and is likely to continue to decline.
The risk of any child dying from a Covid-19 infection is extremely low, and keeps getting lower. For children without underlying health conditions, the risk of death from a Covid-19 infection is similar to the risk of dying from an average 2 days of life (2 per million). Vaccination can significantly reduce risk for children with comorbidities. The risk of MIS-C after infection has declined significantly.
Thanks for reading The Munro Report! Subscribe to make sure you don’t miss out on new posts.
Would young children who have Down Syndrome but no other comorbidities be classified as a child with a medical comorbidity? Or are they considered to have a severe neurological impairment? (My thought would be no, but I’m biased as all get out on the subject.)
I know it’s a terribly specific question, but here in the U.S., the risk of Covid to Down Syndrome children has basically been answered with, “We don’t know, but probably, so never leave your house and definitely mask your two-year-old.”
The US had approved a third shot for the 5-11 demographic a couple months back. Since that cohort has the lowest inherent risk, is there a rationale based upon outcome that justifies a third dose for children without comorbidity? I acknowledge the short term increases to antibodies and the increased diversity of T cells with a third dose, but do those elements changes the outcome related to severity, duration, MIS-C, PASC, etc?