Thread by @gianlucac1

PROF IOANNIDIS: LETS TAKE IT ON 1/n
The paper (pic) engages in calculating a so called "Global Covid19 IFR" and reaches the conclusion that Covid19 fatality is 0.15% globally, so not distant from flu (0.04-0.08)

What social scientists are doing when writing a paper?

They try to extract general rules from the observed phenomena.
When data are "dirty" or "late" or "incomplete" they use models to infer if not the "stopping points" at least the dynamic of the observed phenomena

2/n

So the main findings of Ioannidis meta analysis are that Covid19 fatality is slightly higher than flu: here onlinelibrary.wiley.com/doi/10.1111/ec…

for a detailed comment on Ioannidis numbers @GidMK did a fantastic job here and on his TL

https://twitter.com/GidMK/status/1376304539897237508?s=20

3/n

But i am more intrigued by the message that the paper wants to convey.

Which i guess is "Covid19 is not such a big issue, the response is exaggerated, actually the response is probably causing more harm"

4/n

I live in Milan, Lombardy. So far between official deaths (circa 30.500) and undercounted extra deaths we can estimate Covid19 toll at 40K. Residents are 10.06mn.
0.4%of population died of Covid19 in Lombardy

In NY 0.28% of population died after the first wave
5/n

In Uk 0.25% died so far

I could go on showing that there are many countries where more than Ioannidis' .15% of pop died of Covid19 and
SEROPREVALENCE IS LESS THAN 30%

As Cauchemez and others wrote Covid19 IFR in each country is a function of many things
6/n

Demographic pyramid, network structure, response, and so on

So, does it make sense to calculate a GLOBAL IFR?

Is the info coming out of Sudan useful to Italy when facing covid19? Do we care what an average IFR is?

7/n

I will not comment on the dataset, btw @ikashnitsky has written on mortality
And i will not comment on the horrendous gate-keeping and ad hominem attack against @GidMK

These are so low, that are not worth a comment. I would just feel ashamed if i was editor of the paper
8/n

that had the courage to publish them
I
am just discussing the message of the paper. No matter the numbers (which are useless at best)

Is Ioannidis trying to convey that "we can contain the epidemic and go back in business as, in the end, Covid19 fatality is not a big deal"?
9/n

So is he suggesting that what we saw in many countries like Italy, US, UK,
and now Brasil is a results of our "priors"??

Is the aim of the paper to encourage a GBD like approach?

I believe so.

10/n

If that is the aim of the paper, i would degrade the findings of the paper to "political statements".

Not policy statements

We can discuss policies.

We cannot discuss facts.

And this is what i reckon that paper does

11/11

@ThManfredi @CrossWordsCW @Di_SPACE_Lauro @mtizzoni @ciro @AurelianoStingi @Doom3Gloom

cc @CT_Bergstrom

PS: A STINGY ONE FOR A STATISTICIAN LIKE IOANNIDIS

ht @ThManfredi

R0 of Sarscov2 >flu

due to sarscov2 R0 we have hospital overwhelming in an unmitigated epidemic

Hence Covid19 IFR goes up

concentration risk should be clear to a statistician.

it is endogenous

@threadreaderapp please unroll