Read on Twitter
Covid-19 IFR is *NOT* 0.13% (THREAD)
There's a lot of confusion and some downright appalling falsehoods in the heated debate about Covid-19. One particular point around which the "let the virus rip" brigade (also known as the morons of Great Barrington) have assembled is the
1/
There's a lot of confusion and some downright appalling falsehoods in the heated debate about Covid-19. One particular point around which the "let the virus rip" brigade (also known as the morons of Great Barrington) have assembled is the
1/
infection fatality rate (IFR) of Covid-19. Unlike the case fatality rate (CFR) which is simply total confirmed deaths over total confirmed cases, the IFR - total *actual* deaths over total *actual* cases - is unobserved.
2/
2/
Below is a table with some number for the UK - you can see the latest number of deaths (different methodologies) and the corresponding CFRs.
3/
3/
IFR has to be inferred b/c some cases of Covid-19 go unrecorded, due to insufficient testing & some cases being asymptomatic. On the whole, IFR is lower than CFR for Covid-19 b/c while we have fairly accurate data for deaths, we've missed *many* cases (esp. in Mar/Apr)
4/
4/
IFR then has to be inferred - this can be done by modelling the spread (number of total cases) or by using seroprevalence studies - large-scale testing for antibodies on a representative sample to gauge how many people in the population have had Covid-19.
5/
5/
The early estimates of IFR in March may have been too high. But some of the "estimates" the anti-lockdown cabal pushes are plainly ridiculous. This applies e.g. to Sunetra Gupta, one of the signatories of GBD who claimed IFR could be "as low as 0.1%"
https://unherd.com/2020/05/oxford-doubles-down-sunetra-gupta-interview/
5/
https://unherd.com/2020/05/oxford-doubles-down-sunetra-gupta-interview/
5/
Let's look at a concrete example. At the Treasury Select Committee hearing two days ago, Gigi Foster, assistant professor of economics at UNSW & "an empirical, data-driven scientist" (these are her words - hold on to them) claimed the IFR is 0.13% based on "WHO data".
6/
6/
Now let's look what would that mean for the UK. Below is a table with some *hypothetical* IFR value with Foster's claim highlighted.
7/
7/
That's right - given the confirmed deaths with a positive Covid-19 test in the prior 28 days - i.e. the *most conservative* number - over 34 million people in the UK would have had Covid-19 already. More than a half.
8/
8/
Why is this nonsense? Two main reasons - firstly, there have been seroprevalence studies done in the UK and they found much lower levels of incidence. The chart below is from the ONS Pilot study - even if we take the upper confidence bounds, still *nowhere near* 50%.
9/
9/
Link to the study https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/conditionsanddiseases/bulletins/coronaviruscovid19infectionsurveypilot/englandandwales4september2020
ICL React study finds similar numbers https://www.imperial.ac.uk/news/201893/largest-study-home-coronavirus-antibody-testing/
10/
ICL React study finds similar numbers https://www.imperial.ac.uk/news/201893/largest-study-home-coronavirus-antibody-testing/
10/
The second reason is that we're still seeing *large* increases in new cases every day (21,242 reported two days ago) - this would not be the case if 50% of the population already have already had covid as we would be nearing the heard immunity threshold.
11/
11/
The herd immunity threshold for Covid-19 is estimated to be around 60% of the population being not susceptible (acquiring immunity either through developing antibodies after going through covid or through vaccination). See a very simple SEIR model illustration below.
12/
12/
The daily increases in cases are inconsistent with being close to herd immunity. Thankfully, @t0nyyates called out the BS from the "empirical, data-driven scientist" Foster (incapable of doing basic arithmetic) at the hearing - but these claims are constantly repeated.
13/
13/
In fact, the countries that have seen the worst epidemics so far are also currently seeing the largest increases - suggesting that herd immunity is far off.
14/
14/
The IFR is not constant. It depends on many factors such as the population profile of countries, available care, comorbidities and many more. It's also fallen down over time as we get better at treating covid with dexamethasone, interferon beta or monoclonal antibodies.
15/
15/
But still, a significant proportion of people who contract Covid-19 sadly die. It's not a virus that we want to let rip while we "shield the vulnerable" (as though that's so easy to do - but that's for a whole another thread).
16/
16/
The brazen falsehoods that the proponents of these "strategies" often use are despicable and should be called out - especially if they don't even hold up against back-of-the-fag-packet calculations.
17/
17/
Below are some proper estimates of IFR & the implied incidence from those estimates in the UK. Note that this illustrates that many cases went undetected & that the March estimates were slightly higher, although as said above, treatment improvement have also lowered IFR since
18/
18/
In conclusion - be vigilant to outlandish claims and fact-check as much as you can. And listen to the experts in field like @AdamJKucharski @chrischirp @mlipsitch @nataliexdean @K_G_Andersen - not upjumped economist (like yours truly)
/END
/END
Links to the IFR estimates studies in the last table (top to bottom)
https://www.who.int/bulletin/online_first/BLT.20.265892.pdf
https://www.cebm.net/covid-19/estimating-the-infection-fatality-ratio-in-england/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7524446/
https://gh.bmj.com/content/bmjgh/5/9/e003094.full.pdf
https://www.medrxiv.org/content/10.1101/2020.03.09.20033357v1 https://www.medrxiv.org/content/10.1101/2020.03.05.20031773v2
https://www.who.int/bulletin/online_first/BLT.20.265892.pdf
https://www.cebm.net/covid-19/estimating-the-infection-fatality-ratio-in-england/
https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7524446/
https://gh.bmj.com/content/bmjgh/5/9/e003094.full.pdf
https://www.medrxiv.org/content/10.1101/2020.03.09.20033357v1 https://www.medrxiv.org/content/10.1101/2020.03.05.20031773v2
