Thread by @AtomsksSanakan

1/G

On other threads I criticized estimates of COVID-19's fatality. Here I'll highlight the best estimate I've seen:
0.9% from Neil Ferguson's team at Imperial College.

It's being falsely criticized again.
(h/t @thereal_truther)



tabletmag.com/sections/news/…

2/G

Ideologues often criticize the 0.9% estimate in order to downplay the severity of COVID-19 + evade policies they dislike. John Ioannidis resorted to that

judithcurry.com/2020/04/01/imp…
cato.org/blog/how-one-m…
freopp.org/jay-bhattachar…
reason.com/2021/06/22/the…

3/G

In March 2020, Ferguson's team applied work from Verity et al. on China, to Great Britain (GB).

That led to an estimate of 0.9% of SARS-CoV-2-infected people dying of COVID-19; i.e. 0.9% infection fatality rate (IFR).



spiral.imperial.ac.uk:8443/bitstream/1004…

4/G

That result can be checked using antibody (seroprevalence) studies that estimate the number of infected people.

Great Britain IFR inferred from UK BioBank study is ~0.9%.

ukbiobank.ac.uk/media/x0nd5sul…
static-content.springer.com/esm/art%3A10.1…
coronavirus.data.gov.uk/details/deaths…

5/G

ONS uses the same antibody test as UK BioBank.
ONS' IFR is ~0.9% for England, + about the same or higher for the rest of GB.

(England has the largest impact on GB's IFR, since it makes up ~84% of GB's population)

ons.gov.uk/peoplepopulati…

thelancet.com/journals/lanpu…

6/G

The 1st round of REACT-2 gets about the same result as ONS.

(Later rounds of REACT-2 are less reliable because of the antibody test used in REACT-2; that problem doesn't apply to ONS + BioBank)




nature.com/articles/s4146…

7/G

The WHO + the USA's CDC relied on Levin et al.'s IFR estimate:

web.archive.org/web/2021032419…


Levin et al. estimated IFR by examining antibody studies mostly from Europe + the USA. Their results matched those of Ferguson's team:

link.springer.com/article/10.100…

8/G

So that's at least 4 different sources supporting Imperial College's IFR estimate.

I can't think of another estimate with that much support.

(ONS, BioBank, + REACT-2 are independent of each other. Levin et al. uses all 3, but in combination with dozens of other sources.)

9/G

On to Verity et al., which the ~0.9% was inferred from:
medrxiv.org/content/10.110…
thelancet.com/journals/lanin…

Verity et al.'s reportedly assumed 12 - 13 deaths for their ~0.7% IFR estimate, when only 7 deaths happened so far.

That made contrarians mad:
judithcurry.com/2020/03/25/cov…

10/G

Turns out 14 people died:
science.sciencemag.org/content/368/64…
en.wikipedia.org/wiki/COVID-19_…

And Verity et al.'s IFR of ~0.7% for China held up well:



So Imperial College's ~0.9% estimate for Great Britain was based on work that held up.

11/G

IFR estimates from Ferguson's team also did well in other European nations with decent death reporting:


So it'd be nice if ideologically-motivated deniers stopped fabricating / lying about the accuracy of the team's work. 🙄

onlinelibrary.wiley.com/doi/full/10.11…

12/G

And 0.15% IFR from the top of the thread fails because:

- it's a global IFR, while the Imperial College team's IFR was for Great Britain (IFR varies across populations)
- 0.15% is from a nonsensical paper by John Ioannidis


washingtonpost.com/opinions/witho…

13/G

Some tweets addressing more of the disingenuous denialism from the Tablet article mentioned at the top of the thread.





tabletmag.com/sections/news/…