Daily numbers, graphs, analysis thread 7

[Barracker]

Welcome to thread 7 of the daily updates.

Resource links:
Worldometer UK page
Financial Times Daily updates and graphs
HSJ Coronavirus updates
Johns Hopkins Coronavirus Resource Centre
NHS England stats, including breakdown by Hospital Trust
Covidly.com to filter graphs using selected data filters
ONS statistics for CV related deaths outside hospitals, released weekly each Tuesday

Thank you to all contributors for their factual, data driven, and civil discussions.

@BigChocFrenzy

This site www.cebm.net/covid-19/covid-19-death-data-in-england-update-2nd-may/ does a similar graph, but doesn’t show the layers of lava, just the care home deaths from ONS certified by 25th April vs the hospital deaths. Interestingly the 31st March anomaly seems to be smoothed out by an excess of deaths registered as 31st March outside hospitals.

That is an interesting chart, Millicent, and still shows the peak at 8th April. It looks like very few outside hospital deaths on the 29th and 30th of March, and those put in to 31st. Very odd.

The chart shows deaths registered by 25th of April, but only goes up to 14th of April, so I presume is relatively accurate to that date (allowing for a ten day lag).

@ nquarz i believe there was a study in france to which (if i remember correctly) said that the lockdown saved 60,000 people in a month.

The recovery figures for France are
92,421went to hospital
50,562 discharged - went home
25,857 still in hosptial today.

Has anyone else noted that less healthcare workers have died, as a percentage of that workforce, than the population at large, despite lack of PPE and additional viral load.

That's interesting. But we also have to be careful with regards the lack of PPE. Not all areas did suffer as badly and plenty of hospitals had what they needed so there was protection to a great number of them.
The community don't have any.
Does anyone have the stats re bus drivers and supermarket workers who were at risk of viral load and didn't wear any PPE. That percentage would show something.

I had another thought last night. Considering the large numbers of workers in transport and nhs who dies very early on into lockdown our peak of infections must have been end of feb when most people were spreading it. So it came here very early or the R0 was closer to the 4 or 5 it's been predicted it may have been at.
Considering the R of measles is about 18 I think it does show where this herd immunity does have an impact and why it was discussed at first. It's necessary but not in a sacrifice the elderly way it was originally banded about by DC.

On the what countries include what deaths Ireland includes all deaths including those at home and including probable Covid deaths. Occasionally they have to take probable deaths out of the Covid counts when post mortem shows a different cause.

We also need to ascertain what "health workers" means - if it includes all the admin staff, the managers working from home, the people who are working in empty wards or have no patients due to cancellation of normal medical procedures, then we have no idea of the true rate.

I would like to see the figures for front line nurses, doctors and health care workers who are working on wards, vs the population for 18 to 60 year olds in the rest of the population. I bet those healthcare workers (like bus drivers and some others) have a very high rate of serious illness and death. Particularly the front line nurses etc who are working on non-ICU wards who had absolutely no PPE for the first couple of weeks.

@pocketem

Belarus is interesting... The extract from the article in mid-April below appears to show that many Belarussians took matters into their own hands:

“In Minsk, the metro feels empty, hand sanitisers are everywhere and a lot of restaurants are closed. In Vitebsk, too. There Yuri (name changed upon his request) only leaves home for shopping, keeping a 1.5 metre distance from other customers. His wife has been working online for 2 weeks.”

This, combined with:
a) lockdowns in surrounding countries and the fact Belarus isn’t exactly an international hub
b) accusations that stats are being fiddled to artificially reduce official figures
c) the lack of outside media scrutiny in a dictatorial country,

help explain why Belarus has avoided a catastrophe, or even moderately high figures. Whether it can sustain this, especially given rises in Russia, is debatable.

Hi everyone can someone help me understand the CDCs hospitalisation rates

Im not sure i understand the scales used. For example the 18-49 group. Is that 2.5 people out of 100,000 people or 2.5%, so 2500 people?

Also it says its based on confirmed cases so is that out of 100,000 confirmed cases or just the general population?

Sorry if this is really confusing, its really confused me haha!

@NewAccountForCorona

"We also need to ascertain what "health workers" means - if it includes all the admin staff, the managers working from home, the people who are working in empty wards or have no patients due to cancellation of normal medical procedures, then we have no idea of the true rate.

I would like to see the figures for front line nurses, doctors and health care workers who are working on wards, vs the population for 18 to 60 year olds in the rest of the population. I bet those healthcare workers (like bus drivers and some others) have a very high rate of serious illness and death. Particularly the front line nurses etc who are working on non-ICU wards who had absolutely no PPE for the first couple of weeks."

A study by Newcastle University, published in The Lancet, found that the rate of infection amongst clinical staff was no higher than amongst non-clinical staff.

Release from Newcastle University www.ncl.ac.uk/press/articles/latest/2020/04/covid19screening/

Full article in The Lancet www.thelancet.com/journals/lancet/article/PIIS0140-6736(20)30970-3/fulltext

Morning
Hoping one of you could tell me what I am obviously missing: what are the numbers for recovered for UK and for Holland? If neither is publishing/releasing them, why not? Why have they not estimated based on release from hospital?
TIA

Not strictly a "fact' but I saw this interesting sandbox simulation on what comes next with regard to covid19 - what measures get us to and keep us under R1
ncase.me/covid-19/

Five, as I understand it the thing is that the uk has not been testing previous positive cases in the community to confirm recovery (I suppose because our testing capacity was so limited) so it would basically have been guesswork and they weren't willing to put out figures that were just a guess.

Very interesting myrtle

Thanks, Millicent

Jrobhatch re your question:

Your 2nd chart has "hospitalisation rates" with asterisk
When we look at the CDC paper from which these figures come:

https://www.cdc.gov/mmwr/volumes/69/wr/mm6915e3.htm?scid=mm6915e33_w

the asterisk in the footnote states
"Number of patients hospitalized with COVID-19 per 100,000 population"

Sio e.g. for the 18-49 group, 2.5 people out of every 100,000 people
means that for a city of 1 million people, that would be 25 in that age group hospitalised

The numbers are very low because they are only for the period until 30 March,
when the epidemic was in the v early stages in the USA:

only 4,000 hospital deaths total on 30 March compared to 68,000 hospital deaths today
and only 168,000 cases compared to 1.2 million today

Chers, that Lancet article has a few gaping holes in it and I'd be interested to see an updated version. Firstly, they only ascertained the clinical status of a third of the participants. Secondly they were only testing those with symptoms with no clarity as to what proportion of clinical/non-clinical staff had symptons. Thirdly it only seems to go to the end of March (feel free to correct me, but it seems to run for 3 weeks from 12th March).

Most importantly they state that "The small number of non-clinical staff tested meant that it was not possible to meaningfully compare transmission dynamics between these groups" - so it seems very few non-clinical staff were showing symptoms at the time.

I find it hard to understand how they have come to any meaningful conclusion. While I would of course be delighted to hear that front line staff with little or no PPE had no greater incidence of Covid than everyone else, I don't think this paper proves it.

Think I’m posting in the right place. Would someone please explain today’s figures for the number of deaths...
Nhs England reported 327 deaths in last 24hrs
Scotland = 12
Wales = 14
Northern Ireland = 5.
That totals 358. The NHS England’s number doesn’t include the wider community deaths, how come the UK total for today is 315 including the deaths in the wider community?

From what I can gather from the bumpf under the tweet CKBJ it's because some of the hospital figures were counted under a previous day for some reason. The figure counted the "old" way was 396.

mobile.twitter.com/DHSCgovuk/status/1256997151638511620?ref_src=twsrc%5Egoogle%7Ctwcamp%5Eserp%7Ctwgr%5Etweet

I'm thinking someone has changed how something is reported again. Because who cares about consistency when tracking figures....(I know there are probably good reasons for it but annoys me slightly)

Thank you I’m still confused though! I’m just a very cynical person and believe there is an ulterior motive...we are very close to Italy’s death total.

Gove announced 76,496 tests yesterday

Charts:
. Uk tests
. Comparisons of tests / 1,000 population with other countries

As discussed before, the death or case curve for a country is the sum of the curves over its regions and cities,
which may have different timing, gradients, peak level etc

The USA is a notable example:

John Burn-Murdoch@jburnmurdoch (FT)

Look at national & subnational daily deaths side-by-side.^

NY daily deaths descending, but US plateau continues.

Why?

US is compound of multiple peaks like NY.
As one state’s daily toll falls, another rises, keeping daily deaths high at national level

Apparently I was doing Excel wrong. I made it so it automatically updates (if you download the relevant files), also you have to remap all their crappy spreadsheet data into a relational format so it works better.

gofile.io/?c=PLtbHX

Regarding the 31 March trough I can only imagine that this is linked somehow to the release of daily files. The first daily file is dated 2 April, for deaths reported to 1 April and looks like this

24 130 159 44 84 for 28, 29, 30, 31, 1 April respectively

I think that the first batch of files was published 7 April, hence releases 1, 2, 3, 4, 5, 6 April, as well as a totals files dated 7 April but to 6 April. This had 314, 556, 534, 271, 495 deaths respectively for 28 March-1 April In other words

Looking at the individual dailies for deaths on 31 March, we get an implied 102 reported for Tuesday 31 March itself.

I.e.

28 March - 235 (first four days - implied), x, x, x, 24, 19, 17, 7, 2, 1
29 March - 271 (first three days - implied), x, x, 130, 62, 57, 26
30 March - 180 (first two days - implied), x, 159, 102, 46, 29
31 March - 102 (implied), 44, 59, 26, 19, 5, 22
1 April - 84, 179, 134, 51, 14, 33, 17, 19
2 April - 87, 199, 114, 25, 37
3 April - 99, 181, 70, 90, 63

If we look at the 29 March, we see that a day 4 'score' of 130 is improbable, as is 159 for day 3 of 30 March.

Further if we consider deaths released published after the first tranche, there were 84 deaths published 7 April onwards for 31 March, 76 deaths published 6 April onwards for 30 March, and 89 published 8 April onwards for 1 April, despite the death totals being supposedly 400 for 31 March vs. 600 for 30 March and 1 April.

Looking at the specific daily files I would say that the daily files dated 2nd, 3rd, 4th April are all clearly erroneous, with figures of 130, 159, 144; 62, 102, 59; 57, 46, 26. The 5 April (25, 29, 19) might be correct, while subsequent releases (5, 7,5 and 6, 9, 22) are almost certainly correct.

It's worth noting that the Wayback machine reckons the data were first published 4 April <a class="break-all" href="http://web.archive.org/web//www.england.nhs.uk/statistics/statistical-work-areas/covid-19-daily-deaths/" rel="nofollow" target="_blank">web.archive.org/web//www.england.nhs.uk/statistics/statistical-work-areas/covid-19-daily-deaths/ though the capture is blank.

So I would say that many deaths from the 31 March were allocated to the 29 and 30 March, and this occurred right at the beginning of the daily release process. I am not sure how this happened, but if there was a script somewhere that contained hard-coded dates or whatever then you could see how this could occur.

Very plainly there was no 'real' trough, just a data recording error that NHS England should be able to correct.

shoots
Does your graphic include NHS deaths only or carehomes as well? I know they were added in at a late date and wasn't sure if they were then incorporated. I find these charts you and Barraker do so useful