Thread by @apsmunro

Because England has excellent population data on #COVID19, we can make pretty accurate estimates of risk to children 🧵

Fortunately, these risks are extremely low

1/

Note when possible that we want to know rates *per infection*, not per positive test

This is because tests are contingent on how much you test, which is different across time and places

Risks per infection are much less changeable, and what we really care about

2/

Let's take the first 12 months of the pandemic up until Feb 28th 2021

Estimates based on seroprevalence are around 25% children <18 were infected by this point (16+ was >30%, younger will be a bit lower)

There are 12mil children in England, so this is 3mil infections

3/

By that point there were 5000 admissions <18

We know from ISARIC data at least 20% of these were incidental positive tests, so max 4000 admissions due to COVID-19

That's a hospitalisation : infection ratio of 1 :750

4/

By that time there were 61 deaths with a positive test

Of these, only 25 were caused by COVID-19

That's a death : infection ratio of 1 : 120,000

5/

As there is no definition for "long covid" in children, the best we can do is estimate relative risk from the CLoCK study of having 3 or more symptoms 12w after a COVID-19 test

30% if tested positive, 16% if tested negative

RR=1.8 if positive for SARS-CoV-2

6/

Approximate risks to children from SARS-CoV-2 infection summary:

Hospitalisation: 1 in 750
Death: 1 in 120,000

Test +ve for SARS-CoV-2, symptoms at 12w RR=1.8

7/

References:
researchsquare.com/article/rs-689…

coronavirus.data.gov.uk/details/downlo…

ons.gov.uk/peoplepopulati…

researchsquare.com/article/rs-798…

8/