Scrutable

shpalman wrote: Sun Apr 26, 2020 12:42 pm "we're listening to the scientists"

"which ones?"

"it's a secret"

https://www.theguardian.com/world/2020/ ... s-outbreak

Don't worry I'm sure it's not totally corrupt
https://twitter.com/carolecadwalla/stat ... 51104?s=09

I don't think Brad Pitt will be playing any of them.

Carole's thread is amazing and yet entirely expected. Cummings being frustrated with civil service entrenchment is at least understandable. Him being in discussion of an independent advisory group is understandable if he wants to get the raw discussion and not just the resultant report and/or minutes.
Him taking part in the discussion and even directing it is lunacy and renders all advice partisan.

If the reporting is true, this is a scandal and Cummings should resign. He won't of course.

Data on excess mortality around the world:
https://twitter.com/jburnmurdoch/status ... 3880496128

Full article here: https://www.ft.com/content/6bd88b7d-338 ... 5c6fac846c

plebian wrote: Sun Apr 26, 2020 7:50 pm

shpalman wrote: Sun Apr 26, 2020 12:42 pm "we're listening to the scientists"

"which ones?"

"it's a secret"

https://www.theguardian.com/world/2020/ ... s-outbreak

Don't worry I'm sure it's not totally corrupt
https://twitter.com/carolecadwalla/stat ... 51104?s=09

I don't think Brad Pitt will be playing any of them.

Carole's thread is amazing and yet entirely expected. Cummings being frustrated with civil service entrenchment is at least understandable. Him being in discussion of an independent advisory group is understandable if he wants to get the raw discussion and not just the resultant report and/or minutes.
Him taking part in the discussion and even directing it is lunacy and renders all advice partisan.

If the reporting is true, this is a scandal and Cummings should resign. He won't of course.

Its about time for Cadwalladr to go underground. If she doesn't I expect within the next year she will suffer a nasty accident.

plebian wrote: Sun Apr 26, 2020 7:50 pmCummings being frustrated with civil service entrenchment is at least understandable. Him being in discussion of an independent advisory group is understandable if he wants to get the raw discussion and not just the resultant report and/or minutes.
Him taking part in the discussion and even directing it is lunacy and renders all advice partisan.

Yes, I agree. It might not be a problem if he was an observer. For one thing it would be useful for Johnson to know which parts of the scientific advice was based on consensus and which was contested.

But him playing an active role is bonkers.

Woodchopper wrote: Sun Apr 26, 2020 8:19 pm

plebian wrote: Sun Apr 26, 2020 7:50 pmCummings being frustrated with civil service entrenchment is at least understandable. Him being in discussion of an independent advisory group is understandable if he wants to get the raw discussion and not just the resultant report and/or minutes.
Him taking part in the discussion and even directing it is lunacy and renders all advice partisan.

Yes, I agree. It might not be a problem if he was an observer. For one thing it would be useful for Johnson to know which parts of the scientific advice was based on consensus and which was contested.

Actually this could be a problem - the advisory group might make recommendations - after considering various views. As a direct line to the PM Cummings could give undue weight to minority opinions that the advisory group had already considered and (the majority) had decided not to pass on. The PM could then latch on to these.

The decline on the other side of the peak is really slow. Slower than expected?

I note the models from James Annan, which were successful on the way up, were a bit too optimistic all last week on the way down. A series of readings all above his line, tilting up his curve. Official UK deaths in the past 7 days is 4,672. Compares to his forecast of 3,300. Unlike a certain organisation, he keeps disclosing all his previous charts to reveal any errors:

https://twitter.com/jamesannan/status/1 ... 1098802177

Countries that locked down early appear to have steeper slopes downwards, as well as the lower absolute peak. The UK could be joining the early-looseners group of countries right now, if it had been 11 days earlier with the lockdown. At a guess, every 7 days earlier to start leads to 14 days earlier to finish.

UK number 1 in the world (official deaths per million population, last 7 day average).

Sweden has risen fast, taking it to a highly competitive number 2, but perhaps peaked before it could take our number one spot. Spain in number 3 thanks to a steeper downwards slope than the UK. Italy in 4th, former number 1, France, now well back in 5th due to having the steepest slope.

Oh, that chart didn't include Belgium, which is number 1. Something odd for Ireland as well.

Chart tool here:

https://ourworldindata.org/grapher/dail ... OR+PRT+IRL

lpm wrote: Mon Apr 27, 2020 10:25 am Countries that locked down early appear to have steeper slopes downwards, as well as the lower absolute peak. The UK could be joining the early-looseners group of countries right now, if it had been 11 days earlier with the lockdown. At a guess, every 7 days earlier to start leads to 14 days earlier to finish.

As mentioned in the other thread, the lockdown in Spain etc was much more restrictive. While Britons were debating how many people could safely fit into a park, Spanish kids weren't allowed outside at all.

As if things couldn't get worse - assuming it's correctly identified

https://www.theguardian.com/world/2020/ ... oronavirus

But deaths are diverging from cases? UK seems alright on cases, doesn't have overloaded hospitals, yet has slow decline in deaths?

lpm wrote: Mon Apr 27, 2020 10:25 am The decline on the other side of the peak is really slow. Slower than expected?

I note the models from James Annan, which were successful on the way up, were a bit too optimistic all last week on the way down. A series of readings all above his line, tilting up his curve. Official UK deaths in the past 7 days is 4,672. Compares to his forecast of 3,300.

This is official deaths in hospital, right? Might be going down more slowerer if hospitals have been near capacity the whole time so as beds (or medical staff) are made available (by deaths, or possibly recoveries but there are no official figures for recoveries) the more severe cases from the community are moved into hospital. So the peak should actually have been higher, and is still higher than the official numbers but coming down, only the top bit of it is chopped off is what we see.

I can say that even if the numbers of cases don't seem to be coming down very quickly in Italy, it's because testing is still ramping up, and the numbers in critical- and non-critical care in hospital are decreasing a lot (as are deaths); so it's clear what's happening is that the increased testing capacity allows mild cases to be tested and also allows health workers to be screened. I can't say this for the UK firstly because I don't have the numbers for hospitalized cases (I only got the testing numbers by scrolling through the Department of Health and Social Care's twitter feed, for f.ck's sake)* and secondly because testing isn't ramping anywhere.

* because I'm an idiot and it only just occurred to me now to google for a github.

lpm wrote: Mon Apr 27, 2020 12:05 pm But deaths are diverging from cases? UK seems alright on cases, doesn't have overloaded hospitals, yet has slow decline in deaths?

Official deaths, or actual (excess) deaths?

lpm wrote: Mon Apr 27, 2020 12:05 pm But deaths are diverging from cases? UK seems alright on cases, doesn't have overloaded hospitals, yet has slow decline in deaths?

See here, :https://www.gov.uk/government/publicati ... april-2020

Slide 5: Occupancy of critical care beds has declined a bit, but not as steeply as the increase before the peak. That seems to the best way to predict deaths in a week or two.

Slide 4: there has been a big decline in occupancy of all hospital beds in London, but the curve is much flatter in the rest of the UK.

So it looks like there has been ongoing hospitalization in particular in the North East and North West of England and Scotland. One issue may be that up there there the population is older and more unhealthy, leading to long hospital stays and eventual death for a high proportion of those infected.

One important thing I haven't seen is data on new hospital admissions. That would be a good guide to how much the virus is spreading.

Thanks, interesting slides.

Two models, one for London, one for UK excl London, would be interesting.

Still seems impossible to predict May & June levels in of the major European countries - a fall to S.Korea levels, or a long, long tail. The James Annan forecast swings up and down between 2,000 UK fatalities in May and 12,000.

lpm wrote: Mon Apr 27, 2020 1:24 pm Thanks, interesting slides.

Two models, one for London, one for UK excl London, would be interesting.

London has a much higher population density.

That may explain why it was affected so badly.

Bird on a Fire wrote: Mon Apr 27, 2020 12:23 pm

lpm wrote: Mon Apr 27, 2020 12:05 pm But deaths are diverging from cases? UK seems alright on cases, doesn't have overloaded hospitals, yet has slow decline in deaths?

Official deaths, or actual (excess) deaths?

Here's estimates of daily excess mortality. https://twitter.com/ChrisGiles_/status/ ... 9339494413

Similarly shows a much smoother downward slope.

Woodchopper wrote: Mon Apr 27, 2020 12:55 pm Slide 5: Occupancy of critical care beds has declined a bit, but not as steeply as the increase before the peak. That seems to the best way to predict deaths in a week or two.

There are a couple of factors that might be occurring here. Firstly, the very sick might take a long time to die, so keep occupying a bed making that occupancy relate to a much earlier period of infections. Secondly, when there were more sick people, some of the more hopeless cases might have been left to die quickly rather than keep them alive in the slight chance that they might recover, so their deaths were earlier and we're now just spreading some deaths over a longer time.

Millennie Al wrote: Tue Apr 28, 2020 1:02 am

Woodchopper wrote: Mon Apr 27, 2020 12:55 pm Slide 5: Occupancy of critical care beds has declined a bit, but not as steeply as the increase before the peak. That seems to the best way to predict deaths in a week or two.

There are a couple of factors that might be occurring here. Firstly, the very sick might take a long time to die, so keep occupying a bed making that occupancy relate to a much earlier period of infections. Secondly, when there were more sick people, some of the more hopeless cases might have been left to die quickly rather than keep them alive in the slight chance that they might recover, so their deaths were earlier and we're now just spreading some deaths over a longer time.

That's probably part of it. The upward slope made up of people who die quickly, and the downward slope contains those who took longer.

[/grumble]

It would be really nice if ICU occupancy was somewhere really accessible, like the UK gov data dashboard wouldn't it, not tucked away on those slides Woodchopper kindly linked to for us.

Compare the sparse data on the UK dashboard to the detail on the Scottish page here. Which I went to to see how bad it is up where son#2 is.

You'd think that as the UK dashboard contains county level data for England but only national level data for Scotland and Wales and NI, they'd at least bung in a link to those elsewhere so people could find it easily, but oh no, that'd be way too helpful

Course, at least if I want to look up all-causes mortality I can do that easily at the ONS and the Scottish equivalent. Buggered if I could find that data for the US anywhere yesterday. It must be on the CDC site somewhere, but from what I could make out only as raw data to download. Not as a nice graph plotted against 5-yr weekly averages or anything handy like that.

[/end grumble]

Informal meta analysis of Covid infection fatality rate: https://medium.com/@gidmk/what-is-the-i ... 58f7c90410

Comes out at between 0.5% and 1%. Which is a lot more lethal than influenza.

Discussion here on some of the studies in the testing thread.

Oxford vaccine trial in South London area:
https://www.covid19vaccinetrial.co.uk/p ... -georges-0

Article in Italian on the very annoying Corriere della Sera website says that there were already about 1200 people who got the covids in Lombardy in the 26 days before "patient 1" was discovered on the 21st of February.

For some reason they call the 26th of January "day zero" , on which day there were 46 cases in Milan and 543 in Lombardy in total. Presumably this is because the Corriere della Sera is better at telling you how to disable your adblock and then popping up links to other stories than either understanding the report from Lombardy's health service or even f.cking linking to it. The report is apparently about backdated self-reported symptoms, from patients who tested positive in the early days of the outbreak i.e. they asked the patients who tested positive when they remember symptoms starting or when they thought they caught it or something.

543 is roughly 2^9 so if the doubling time was 2-3 days back then the first case would have been 20-30 days before, assuming it started with the importation of one case, and that's very early given that the Webasto outbreak in Bavaria dates to the 19th of January (and the virus sequenced from Italy's "patient 1" is quite close to the one from Bavaria).

This is the graph showing the date on which positive cases reckoned they had their first symptoms, both for the whole of Lombardy and for Milan:



It's basically not that different to a graph of the number of positive cases per day, shifted a bit to the left, damped at the end, and with a meaningless spike of 543 (or 46) at the beginning of it.

interesting presss release from the COVID-19 KCL app research on genetics and severity of symptoms.

https://covid.joinzoe.com/post/genetics-covid

With a link to the preprint