Read on Twitter
I'm going to take apart a terrible article in the British Medical Journal which uses a cynical reporting of so called asymptomatic Covid claims to argue against lockdowns. The last paragraph makes the intent clear. I'm not a Dr so its alarming how easy this is to dismantle /1
It's based around recent testing in China where 80% of domestic cases were asymptomatic. BUT China is now testing fast and symptoms take 5 days (see chart) to develop so the reasonable description is asymptomatic SO FAR - /2
It's noted that in contrast passengers arriving in the country who tested positive were symptomatic. If that 80% were universal this would not be the case, you'd expect the same ratio. So we can be sure speed of testing in China is the key - catching before symptoms appear /3
The article draws heavily on a so called 'evidence based medicine'Â piece, however it fails to reproduce the 'not peer reviewed, personal opinion' disclaimer that piece carries. I'll note in passing my comment pointing out the obvious is still unpublished by the authors. /4
Piece draws on a Guardian article on Vo, the town that locked down & did mass testing as further 'proof' but again the AS YET test remains. There is no evidence the authors checked to see if symptoms later developed - a simple test of their theory /5 https://www.theguardian.com/commentisfree/2020/mar/20/eradicated-coronavirus-mass-testing-covid-19-italy-vo
Back to the BMJ which assembles some very strange further proof including "Users of Chinese social media have expressed fears that carriers with no symptoms could be spreading the virus unknowingly" Elsewhere users think its really 5g despite that fact there is no 5g in Iran /6
If you want to really test this theory S. Korea has a huge amount of testing done & an entire month past peak. If there really was undetected significant asymptomatic transmission it would appear as unexpected cases & deaths in South Korean data /7 https://www.worldometers.info/coronavirus/country/south-korea/
Instead deaths in S. Korea decline exactly where we would expect them to, 23 days after the case peak (average time to death is 23 days from infection). With 15,000 tests a day across the period you'd expect to randomly hit more & more asymptomatic cases if there were there
OK here is a link to the terrible article itself, withheld it till I'd presented the flaws and counter argument but I've some more to come /9 https://www.bmj.com/content/369/bmj.m1375
You'll see similar 50% asymptomatic claims made for Iceland but also not on an understanding of the data. The so called random data is people who volunteer and the CEO of DeCode has admitted thats more likely to be people who thing they might have been exposed /10
But the full Iceland data set disproves the idea of widespread asymptomatic cases. The blue bar here is conventional testing, the orange DeCode 'random' testing. Despite similar testing numbers DeCode has found only a handful of the positives /11
50% of a very small not really random sample that may actually be NOT YET symptomatic is not much 'evidence based medicine' to scrap the lockdowns which are clearly working in many paces on. So what's actually up? /12
Most straightforward explanation is the one given at the end of the article - it costs too much. If you've a big stock portfolio then perhaps the downwards curve in the value of that is more alarming than the upwards curve in deaths. /13
The sunk costs fallacy is another factor, if you've been one of those experts saying 'its only the flu' for a couple of months and are now facing the enormous not the flu death percentages then you may well grasp as any straws that would make you 'right' for a day or two longer
But this argument also has a more general appeal to those who are afraid because it tells them they need not be. Or to those who are fed up because it tells them this could end with small consequences. You can see why people want to believe with the best of intentions
Meanwhile another 684 died in the UK in the last 24 hours and every day that death toll increase increases by more - except where there have been effective stay at homes for period approaching 23 day. Places like Italy, China, S. Korea. We have the evidence stay at homes work.
This piece by @zeynep on modeling is very worth a read in the context of everything above - my problem is not with 'what if' modelling but it's use here to argue for an end to lockdowns as if that what if is proven https://www.theatlantic.com/technology/archive/2020/04/coronavirus-models-arent-supposed-be-right/609271/
