Coronavirus - worldwide spread and control

The independent SAGE meetings will be starting today at 12pm (UK time). You can watch on YouTube:

https://twitter.com/user/status/1257218607932637185

 

Brilliant, thanks for posting!

Here is the link to the livestream (edited):

https://www.youtube.com/watch?v=cFLAwZeNfxE

 

Alternative Sage meeting streaming now:

https://www.youtube.com/watch?v=cFLAwZeNfxE

 

These results seem reassuring i.e. a test which found 99.5% of positives seems very good and consistent with what I'd heard anecdotally regarding PCR tests i.e.for other pathogens. However, reports on coronavirus testing seemed to show the PCR test was only 70% accurate [Professor Anthony Costello - COVID Report]; possibly this lower value takes into account errors due to the original sampling i.e. rather than the PCR test.

We don't know the rate of false positives; however, the 99.5% figure seems to suggest that the test will find 199 out of 200 true positives. So I'm a bit surprised by the high rate of false positives "6-7 of the 98 positives to be false positives". My level of education is much lower than yours - might explain it!

 

Independent SAGE will be doing a press conference today at 4pm

https://twitter.com/user/status/1257307051622936577


https://twitter.com/user/status/1257313419117518851

 

I found the Independent Sage interesting but was disappointed about how much of the airtime, especially in journalists questions in the press conference, was given over to the mathematical modeller (Prof Karl Friston) of the group.

He also spent a large chunk of the beginning of the actual Independent sage meeting (I think up to the 25 minute mark!) talking about how it was important to know which proportion of society had been infected, that it could be 15%, could be 30%, could be this, could not be that... the difference that makes to strategy and that antibody testing is important. He also talked about second waves.

He talked about herd immunity and if a large proportion were infected, the virus could burn itself out/ would have an impact on the spread. I can see that it’s interesting to him, but now (and this is what I’ve been hearing from the other members of independent sage both outside the group- like on Twitter- and in the meeting), it seems it more important to put focus on antigen testing, tracing and the isolating strategy. Social distancing, what policies will be followed, how community testing will be carried out. I felt that especially in the journalists questions, that message was drowned out, which I thought was a shame. He did raise a few good points in the meeting though. About locally based responses and decision making.

In the independent sage meeting itself I thought Prof Alyson Pollock and dr Zubaida Haque were particularly excellent. It was also good that there were other public health experts, virologists etc. I didn’t listen to everything and skipped in and out though. Points were raised about the decentralisation of testing, tracing etc and it how it needs to be community led - which is not at all what is happening now. Also the shortcomings of the data the govt is producing. And how it’s “5 point plan” of getting out of lockdown don’t align with what the WHO is saying. It was also a message throughout that we shouldn’t be relying on a vaccine coming quickly.

( Edited to add some points) Independent Sage will be collaborating and producing some documents for the govt to look at. They talked about who is doing what, in the meeting.

For those interested, here’s an earlier thread about prof Karl Friston’s work:
https://twitter.com/user/status/1248614842912497668

 

The Heinsberg study by Streeck has now been published (preprint, not peer reviewed)

His conclusions are being discussed controversially; it's going to be interesting what other scientists will make of it.

"While the number of infections in this high prevalence community is not representative for other
parts of the world, the IFR calculated on the basis of the infection rate in this community can
be utilized to estimate the percentage of infected based on the number of reported fatalities in
other places with similar population characteristics.

Whether the specific circumstances of a
super-spreading event not only have an impact on the infection rate and number of symptoms
but also on the IFR requires further investigation.

The unexpectedly low secondary infection
risk among persons living in the same household has important implications for measures
installed to contain the SARS-CoV-2 virus pandemic."

 

Thank you @Snow Leopard for your comments. I wonder whether you find the evidence for asymptomatic transmission presented in the following study anymore compelling? (note it’s a preprint not peer reviewed)

The extract below is from the results section of the paper:

Suppression of COVID-19 outbreak in the municipality of Vo, Italy

https://www.medrxiv.org/content/10.1101/2020.04.17.20053157v1.full.pdf

 

I only looked at seem of the 4pm press release. I agree that there is a sense of disappointment that there is quite a lot of banging own drums here just as much as elsewhere. On the other hand there was some very good material and the idea of being up front and transparent is good.

 

I think you might be getting specificity and sensitivity mixed up (it’s a bit confusing so easily done!). 99.5% specificity means that 99.5% of healthy people are correctly identified and 0.5% are false positives. Specificity doesn’t tell you anything about how many true positives there are - for that you need to look at the sensitivity.

The reports you have seen on the tests being 70% accurate will be referring to the sensitivity, which you’re right in thinking means that 70% of people with SARS-CoV-2 are correctly identified (and 30% are false negatives). @Snow Leopard mentioned earlier in this thread that the sensitivity for RT-PCR tests are 80%, but also pointed out that reliability is worse if testing is carried out in a random community sample (as opposed to testing a highly selected population showing key symptoms).

https://en.m.wikipedia.org/wiki/Sensitivity_and_specificity

ETA: the reason for false positives is because sometimes SARS-CoV-2 might be mistaken for other pathogens such as the common cold coronaviruses

 

I thought I'd put up the list of people who were in the Independent SAGE group. I haven't had time to check them all out yet, so mostly they just had their name up and not their title. So I don't know which were doctors, statisticians, professors or what!

Oh, around the 1.45 mark there was some talk about recovery, rehab, and how to manage ongoing disease after recovery. I didn't notice any talk of post viral fatigue or anything similar, but I was fading by then!

Sir David King - Chair
Karl Friston - Modelling
Anthony Costello
Martin McKee
Gabriel Scully
Kamlesh Khunti
Deenan Pillay
Susan Michie
Dr Allison Pittard - Critical Care
Zubaida Haque
Allyson Pollock
Christina Pagel
Elias ? (I think he might have been on film, but there were some tech probs)

I'm so sorry I haven't found details about any of the participants out - watching for about two and a half hours really left me flattened!

I agree with @lunarainbows that Allyson Pollock was amazing!

Oh, and the last question was about masks. But I'd really lost it by then. I think they suggested to use (cloth ones) to keep off others, not for self. Costello said he's currently in Yorkshire, so when walking goes without. But he would wear a mask in London.

One of the other people (forgotten name) said that the Royal Society (Medicine? not stated) just issued advice. I haven't had time or capability to check that out yet.

I do hope the video is put up for people to see. It's a long watch, it would have been much easier in 15 minute chunks with tea breaks in between for me!

 

List of the participants and what their specialism is, have been put on this thread starting here :

https://twitter.com/user/status/1257257487327211520

 

Rival Sage group says Covid-19 policy must be clarified

And more...

https://www.theguardian.com/world/2020/may/04/rival-sage-group-covid-19-policy-clarified-david-king

 

The FDA has approved an intravenous antibody test made by Swiss company Roche.

"Roche said it's antibody test has a specificity rate exceeding 99.8% and sensitivity of 100%, meaning tests would show very few false positives and no false negatives.

A false-positive result could lead to the mistaken conclusion that someone has immunity. Roche said its test relies on intravenous blood draws, with higher accuracy than finger-prick tests."

https://www.reuters.com/article/us-health-coronavirus-testing-roche-hldg-idUSKBN22F02Q

 

"Accuracy" is a misleading figure, because it doesn't say whether the inaccuracy was due to false positives or negatives. The assumption of 99.5% specificity was from a previous meta analysis of RT-PCR studies. The difference in the figures is due to the fact that many of the subjects are not true positives.

The actual rate of false positives or negatives depends on the underlying true positive rate according to Bayes' Theorem:
https://en.wikipedia.org/wiki/Bayes'_theorem#Statement_of_theorem

If you had 1000 test subjects, 5% (50) are true positives, the test has 90% sensitivity, but only 99.5% specificity, then you would have 50 positive results. Which is what you'd initially expect based on your knowledge of which are true positives, but 5 of those are false positives!

 

Those figures are far higher than real-world test results of similar tests and I simply won't believe them unless they publish a large scale community study comparing the test to other diagnostic tests such as RT-PCR, CT scanning and a symptom based diagnostic protocol.

The problem with their internal testing is the sample sizes are too small and their samples are not representative of the community as a whole.

 

All the members of (the actual) SAGE group have now been published, except for two members who did not give permission for their names to be released.

You can read the full list of names here:
https://news.sky.com/story/coronavi...ed-after-criticism-over-transparency-11982941

Simon Wessely isn’t on the list.

 

The IHME projections have been updated again. They have changed considerably. In fact, the model has been altered to a new "hybrid model" to try to compensate for erratic reporting (by smoothing data) and a slower than expected down slope in daily deaths.

[05/04/20 current estimates projected through August 4th. 04/21/20 prior US estimates were projected through June 1st. Prior European estimates were probably also from 04/21, but might be from 04/01.]


Example changes:

New (05/04) US estimates using new "hybrid" model.

US total:..134,475 deaths (up. 81.54% from.74,073): range (95,092 to 242,890)
Texas:.......3,632 deaths (up 211.49% from. 1,166): range ( 1,470 to. 10,721)
Georgia:.....4,913 deaths (up 148.01% from. 1,981): range ( 2,013 to. 12,125)
New York:...32,132 deaths (up. 35.34% from 23,741): range (29,248 to. 37,136)
California:..4,666 deaths (up 167.70% from. 1,743): range (29,248 to. 37,136)

New(05/04) estimates for Europe still using the prior model.

UK:.........40,555 deaths (up. 49.65% from 27,100): range (29,657 to. 74,539)
Italy:......31,458 deaths (up. 13.25% from 27,777): range (29,605 to. 34,969)
France:.....28,859 deaths (up. 14.99% from 25,096): range (25,280 to. 38,798)
Spain:......27,727 deaths (up.. 9.89% from 25,231): range (25,720 to. 32,130)
Germany:.....8,543 deaths (up. 23.42% from. 6,922): range ( 7,006 to. 12,150)


[ My feeling is that, given the magnitude of the changes, it's hard to know if these predictions will be any more accurate than the previous ones.]

 

I remember it by thinking of net fishing for tuna (apologies to non-meat eaters!). You want to your net to get as many tuna as possible from a particular area (that's sensitivity), but you want to minimise the number of other types of fish that get caught up (that's specificity).

 

New op-ed by Harvard epidemiologist Mark Lipsitch in the Washington Post (no paywall, article free to view)

‘Serology’ is the new coronavirus buzzword. Here’s why it matters.'

"...in the rush to generate and interpret serologic data, it can be hard for anyone, including experts, to understand what serologic studies can and cannot teach us. Why are they so important, and what decisions will they inform?"

https://www.washingtonpost.com/opin...ew-coronavirus-buzzword-heres-why-it-matters/