Conversation

Nate Silver
@NateSilver538
This would imply somewhere between 800K and 3.6m cases in the UK, as compared to a reported total of 22,000. That would mean cases had been undercounted by a magnitude of ~35x to ~160x.
Quote Tweet
kadhim (＾ｰ^)ノ
@kadhim
New study from Imperial College London estimates 1.2%-5.4% of UK population has been infected by coronavirus so far imperial.ac.uk/media/imperial
Image
38
622
Nate Silver
@NateSilver538
In certain respects, it's good in the long term if there are more undetected cases. It means the infection fatality rate and the rate of severe cases is lower. It also means later on there could be some benefits from herd immunity in some of the worst-affected areas (e.g. Spain).
22
398
Lauren Knoth
@lknoth
Fatalities are also under reported. Know someone who thought they had the flu and died at home. When asked if he would be tested for COVID (he was 57), authorities said they dont test dead people. This will be particularly bad in rural areas where hospitals may be less accessible
3
Linus Wilson
@linuswilson
Says only 15-20% #COVID19 cases are tested. Under testing which I think this study estimates at a factor of 6. Also the 10ish day testing lag which means today’s test results are 4*6=24 times lower than the number of actual numbers infected with infections 2x in 3.5 days.
Andrey Pavlichenkov
@apavlichenkov1
Nate, I have not found any reference to age adjustment of CFR. For example, assuming 1.38 CFR for China you get 2.7 CFR in Italy and 1.7 In New York City using age pyramids for these locations. Obviously, sample size may not correspond to actual age pyramid.
2
Kjetil Høiby
@kettlehelmet
I haven’t actually read the papers, but if they’re using reported fatalities to reconstruct actual spread - are they accounting for differences in the use of antibiotics and the associated prevalence of resistant bacteria in certain countries (which also have high fatality rates)
🇺🇦 Michael Oberst 🇺🇦
@miob1781
This study has severe limitations that the authors mercifully make explicit. One problem is that the CFR is based on Wuhan, but the official Chinese numbers are not trustworthy.
1
Brendan Kay
@brendankntb
Doesn’t work so well for countries showing a current calculated CFR below 1.38% (such as Germany, Australia and NZ). I would guess that the biggest source of error would be that the eventual CFR is unlikely to be the same number in different places (with different health systems)