1/P
Interesting paper below co-authored by John Ioannidis, published in a journal Ioannidis was editor-in-chief of for a decade.
It makes some interesting points, but also illustrates the dangers of under-estimating COVID-19.
ncbi.nlm.nih.gov/pmc/articles/P…
2/P
Let's start with 2 concepts:
- infection fatality rate (IFR)
- herd immunity
The 'herd immunity threshold' is the proportion of people who need to be immune to infection in order for the pandemic to not accelerate, even if we lived life like usual.
3/P
Proteins known as antibodies (+ the immune cells that make them, such as B cells / plasma cells) are crucial for preventing re-infection, + thus to getting herd immunity.
Seroprevalence studies measure how many people have increased antibodies levels
4/P
Seroprevalence studies can also be used for estimating the number of infected people, since antibody levels increase after infection.
That can then be used to estimate IFR, which is the proportion of SARS-CoV-2-infected people who die of COVID-19.
institutefordiseasemodeling.github.io/nCoV-public/an…
5/P
So over-estimating the proportion of people with increased antibody levels (i.e. seroprevalence) is dangerous for at least 2 reasons:
1) under-estimating IFR by over-estimating the number of infected people
2) over-estimating how close society is to herd immunity
6/P
Early in the pandemic, Ioannidis under-estimated the risk of IFR by giving unrealistic under-estimates of both IFR and the number people SARS-CoV-2 would infect.
See @JHowardBrainMD for more emphasis on this.
statnews.com/2020/03/17/a-f…
7/P
He also gave a scenario he did not espouse, with an IFR of ~1% and a higher herd immunity threshold that allowed for ~60% of people to be infected.
He later moved on to over-estimating infection rates to under-estimate IFR.
statnews.com/2020/03/17/a-f…
8/P
And Ioannidis peddled the debunked idea that the herd immunity threshold (HIT) was low.
So his trifecta is:
1) under-estimate IFR,
2) over-estimate how close herd immunity is by over-estimating infection rates and, 3) under-estimating HIT
9/P
That brings us to his newer paper. He + his co-authors show previously infected people are at lower risk of infection than people who were not previously infected.
Makes sense, since antibodies increasing after infection + help prevent re-infection.
ncbi.nlm.nih.gov/pmc/articles/P…
10/P
I'm skeptical of the claim that immune protection following infection is as good as immune protection following vaccination.
But that's not the main focus of this thread. So folks can go elsewhere for discussion of that:
11/P
Their paper over-estimates seroprevalence in India, similar to Ioannidis' fellow Stanford COVID-19 contrarian Jay Bhattacharya.
60% seroprevalence would also be incompatible with Ioannidis' previous speculation on low HIT.
ncbi.nlm.nih.gov/pmc/articles/P…
12/P
Since at least May 2020, I + others warned of the risk of saying HIT is low:
archive.is/Xjyec#selectio…
But contrarians like claiming HIT is low to say policies they dislike are not needed.
Here's how that turned out for India:
ourworldindata.org/explorers/coro…
13/P
But that's India. How about Austria, the country's Ioannidis paper was about?
Here's a perspective to contrast with Ioannidis:
IFR is high, HIT is low, + infection rates in 2020 were low. So letting SARS-CoV-2 infect more people is dangerous.
14/P
Ioannidis instead gives IFR estimates of ~0.3% - ~0.4%, with his early speculation of HIT being low and a substantial number of people infected.
onlinelibrary.wiley.com/doi/10.1111/ec…
15/P
Results from the ski resort to town of Ischgl, Austria seems to support his claim, with ~42% of people infected with a low IFR.
But that IFR estimate is non-robust, as the article notes, especially in combination with the low population.
medrxiv.org/content/10.110…
16/P
Results in Vienna are more robust.
Calculating IFR as Ioannidis does (deaths 1 week after the study's mid-point), gives an IFR of ~0.8%, with ~1% infected.
Not what Ioannidis predicted.
ncbi.nlm.nih.gov/pmc/articles/P…
covid19-dashboard.ages.at/dashboard_Tod.…
nature.com/articles/s4159…
17/P
Similar pattern in Austria overall months later, with an IFR of ~0.5% and ~5% seroprevalence.
(with Ioannidis' method under-estimating IFR by not including a long enough lag)
covid19.who.int/region/euro/co…
statistik.at/web_de/presse/…
18/P
And as with India, it was ridiculous to suggest Austria achieved herd immunity by February 2021.
Austria's cases/day and COVID-19 deaths/day increased, which would not happen with herd immunity.
covid19.who.int/region/euro/co…
19/P
So Austria + India illustrate the danger of under-estimating COVID-19 in the way John Ioannidis, Jay Bhattacharya, + their ilk did.
I don't expect most of their fans to learn from this, though. Ideology trumps facts for them
20/P
So Ioannidis just went to Austria to talk about IFR, without admitting he was wrong.
Oh well. Maybe he'll find more people who don't know his long history of being wrong on COVID-19. 🤷♂️
20:16 - 25:20:
Share this Scrolly Tale with your friends.
A Scrolly Tale is a new way to read Twitter threads with a more visually immersive experience.
Discover more beautiful Scrolly Tales like this.