Meet the Other Phone. Only the apps you allow.

Meet the Other Phone.
Only the apps you allow.

Buy now

Please or to access all these features

Covid

Mumsnet doesn't verify the qualifications of users. If you have medical concerns, please consult a healthcare professional.

Daily numbers, graphs, analysis thread 7

981 replies

Barracker · 28/04/2020 12:53

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.Flowers

OP posts:
Thread gallery
127
GlassOfProsecco · 28/04/2020 14:42

Thanks for the new thread & contributions of those who are experts in statistics. Great to have evidence-based discussion & debate.

FATEdestiny · 28/04/2020 14:58

Checking in

Inniu · 28/04/2020 15:00

ShootsandLeaves
Would the issues you raise suggest that Chris Giles has over or under estimated excess deaths?

Iremembertheelderlykoreanlady · 28/04/2020 15:05

Uk deaths today 633

BigChocFrenzy · 28/04/2020 15:08

Thanks for the new thread, Barracker Brew

Shoots I agree about date of death being the key parameter

Giles keeps giving us all an initial shock,
but then we have to go to the ONS source and take the time to analyse in depth - calmly - what the raw data actually means

ShootsFruitAndLeaves · 28/04/2020 15:18

Also I take issue with his statement twitter.com/ChrisGiles_/status/1255056329275834370

"The ONS chart shows this is no flu and much more prevalent than the hospital death figures suggest. "

"The care home deaths are particlarly bad.

There were 7,316 registrations in week ending 17 April, compared with a normal number of about 2,400."

The point about flu deaths is we expect them to appear mostly in care homes.

That's WHY we don't worry about the flu, because a care home resident stays on average for 26 months before dying.

E.g., over the last 3 weeks of 2014 and first 6 of 2015, there were 14,678 deaths of people aged under 65. Over the next 9 weeks, there were 13,926 deaths of the same age group

Meanwhile those over 75 died in total 80,856 in weeks 50-6, and then 47,593 in weeks 7-12.

We can see that there were 33,000 (+70%) excess dead 75+ year olds for two reasons when comparing December - January to February - March:

  • normal winter = more deaths
  • a known bad flu season (2014-2015) = even more deaths

But the deaths of working age people was only 5% higher.

The point about 'flu' is NOT that it doesn't result in lots of excess deaths - it does. The point about flu is that it kills people who, to be blunt, typically haven't got long left to live.

So saying 'lots of people are dying in care homes' doesn't tell you it's not flu.

Looking at specifically the age group 45-64 is where we see that this is not flu.

Here we see 1689 deaths in the most recent 2 registered weeks, as opposed to around 1050 pre-covid-19. That is an increase of 60%

For January 2015 the average weekly deaths in this age group was 1390, and in the summer it dropped to 1100, i.e. 26% higher in the peak than the trough.

We DO have record deaths of the elderly, but it's the deaths across middle age and up that are the bigger issue in that in the sense that we clearly DON'T shut down the economy to save 30,000+ old people over winter, but we are doing it now to save a quite specific demographic - NOT any number of people in care homes, who if this WAS a flu would plainly (and rightly so) be allowed to die, but rather in view of:

perhaps 3000 extra deaths aged 40-64
another 4000 deaths in the 65-75 group

(the under 40 group are at reduced risk of mortality at the moment due to not behaving as foolishly as normal)

So the numbers dead so far in the demographics that society can be seen to actively sacrifice itself to defer death are relatively much smaller, but clearly if we did nothing at all they would be much larger.

This is the original Dominic Cummings point - if this WAS flu, then the care homes would have emptied by 10% or whatever, and we would have carried on as we do each and every winter. This only proved to be wrong when it was found that it didn't behave as flu (which kills people who are OLDER than average than the average mortality), and is thus fully ignorable in that it's plainly insane to shutdown life to save, say, 10% of care home population where in 2.5 years the entire population of that care home population has died anyway.

It was Boris getting ill, rather than Stanley, that was the key to covid-19, not any number of deaths in care homes....

And we need to consider this in context:

  • for example if we stop people driving like cunts, by putting black boxes etc. in all cars, then we might save 500 lives of people aged 20-30 in a year (say), which is 30,000 years of life saved.
  • meanwhile if we shutdown the country over winter, let's say we save 30,000 care home deaths, each with one year on average left to live.

The former looks a lot more sane than the latter, does it not?

So we mustn't obsess TOO much about raw numbers in that '30,000 deaths' (of which 15,000 were in care homes) are not comparable in terms of lost years of life with 30,000 deaths in China, where the population trends younger.

And we can't exactly use respiratory virus care home deaths as a stick to beat the government with, because realistically we don't attempt extreme measures to stop them any other year.

We CAN attack the government on the basis of the thousands of avoidable deaths of younger (i.e. middle aged) people and observe, incidentally, that we also saved 10,000 people in care homes, but the latter is not our goal!

ShootsFruitAndLeaves · 28/04/2020 15:24

Sorry , misplaced a bracket, should read:

" This only proved to be wrong when it was found that it didn't behave as flu (which kills people who are OLDER than average than the average mortality, and is thus fully ignorable in that it's plainly insane to shutdown life to save, say, 10% of care home population where in 2.5 years the entire population of that care home population has died anyway).

TheCanterburyWhales · 28/04/2020 15:25

Checking in. Brew

Mummypig2020 · 28/04/2020 15:27

following

LockdownLiquidLunch · 28/04/2020 15:27

Thank you for the follow on thread OP.

whatsnext2 · 28/04/2020 15:28

Post from London school of hygiene and TM might be of interest:
Mathematical modelling is supporting governments to make important COVID-19 response and control strategies.
In our next live Q&A, Petra Klepac & John Edmunds from LSHTM's Centre for the Mathematical Modelling of Infectious Diseases will help explain this vital but complex tool.
Submit your questions now to [email protected]. WPP

Bflatmajorsharp · 28/04/2020 15:42

Thank you Barracker and others.

HeadPotato · 28/04/2020 15:48

Is there a graph which shows how the death rate in the UK has increased for say March 2020 compared to March 2019,2018 etc ?

OrangeBlossomsinthesun · 28/04/2020 15:57

following

ShootsFruitAndLeaves · 28/04/2020 16:04

To return to the subject of raw data, here's what we have available:

www.england.nhs.uk/statistics/statistical-work-areas/covid-19-daily-deaths/

Daily release of total and breakdown of individual days, for NHS England only, aggregated by NHS Trust, and region. Also by age group (20 year boundaries), and ethnicity data updated weekly on Thursdays in the total file (no daily breakdown)

Points:

  • Age data to be considered unreliable for calculating age profile of deaths as it only includes NHS hospitals, whereas the oldest will die mostly in hospital
  • NHS Trust data difficult to analyse beyond the simple trend in view of lack of direct mapping between a Trust and population.
  • NHS death figures certainly OVERSTATE (by perhaps 20%) the number of people who died BECAUSE of covid-19, because testing is routine, whereas it is not in other venues
  • Daily releases are for the previous day of REPORTS OF DEATHS and for a date of death usually 1, 2 or 3 days before publication, but anywhere up to a month or more old. Daily headline figures largely noise because of lag over weekend and spike on Tuesday (Monday reports)

www.icnarc.org/Our-Audit/Audits/Cmp/Reports

ICNARC reports on intensive care outcomes, weekly

Caution note:

graphs may show increased incidence* of certain characteristics such as obesity, ethnicity or deprivation, but there is no regression testing, so be wary - obesity correlated with age, ethnicity with major urban areas, and likewise deprivation

ONS:

www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/datasets/weeklyprovisionalfiguresondeathsregisteredinenglandandwales

Weekly death spreadsheets. Released on Tuesday to 11 days earlier. Data by age group for all deaths (compare with normal weeks for age group). By registration date only for all deaths, and by date of death for covid-19 deaths (no fixed interval between the two dates)

We should note that there are separate data for those deaths 'with covid-19', but that this number is OVER stated for hospitals but UNDER stated for all other venues of death. Therefore the covid-19 stats themselves are not nearly as useful as simply comparing 'excess deaths', i.e. deaths by week compared with the average week, EXCEPT that the risk of covid-19 death for the young (under 30) is teeny-tiny, and these are more likely to be tested and are NOT likely to be in care homes, so we need to take some combination of 'deaths with covid-19', and 'excess deaths' depending on the age group in order to find the actual deaths by age, given that fewer young people are dying from accidents now than normal, which is more significant than the extra deaths from covid-19.

To find date of death for all causes refer to

www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/causesofdeath/datasets/deathregistrationsandoccurrencesbylocalauthorityandhealthboard

which is further split by local authority, and location of death, but NOT age. Here we can reasonably estimate excess deaths by LA. The absolute number of young people not dying because of the lockdown is quite small so it doesn't affect lockdown too much.

We should cross-reference the deaths by LA against the age profile of that LA, and also the number of people in care homes.

Test results:

www.gov.uk/guidance/coronavirus-covid-19-information-for-the-public

Government defines FIVE test classes or pillars:

Pillar 1 - people ill in hospital with symptoms covid-19 (first priority), and certain medical staff. Likely to be high % positive (because sick!) without necessarily implying high % positive in the population
Pillar 2 - mass testing of NHS and other 'key workers'. Depending on the type of key worker it's possible that this is not accurate
Pillar 4 - surveillance testing - testing to find out population that had the virus

the other two pillars
Pillar 3 - theoretical immunity testing that is not yet possible
Pillar 5 - mass testing

are not yet in progress.

So far we have around 600,000 tests in P1, 107,000 in P2, and just 6,500 in P4

P1 is probably best ignored although a declining % of positive cases is of course good news, but the headline 'total cases' is just meaningless and in no case comparable between countries.

P2 is more interesting but needs caution as early cases may be more likely to be symptomatic?

P4 is the most interesting but still rather pitifully low. From here with a sufficient sample size (consider 300+ local authorities, and a few dozen a week from each as a good start, hence several thousand weekly tests at a minimum). Using these data we can estimate the level of infection in our community.

Nquartz · 28/04/2020 16:11

@ShootsFruitAndLeaves out of interest, what is your job? I need to be quite analytical/good with numbers for my job but not at your level 

You & @BigChocFrenzy really know what you're talking about!

I'm generally a lurker & find this thread fascinating, thank you

BigChocFrenzy · 28/04/2020 16:19

We can roughly calculate the % COVID deaths saved in each age group
e.g.
age group 40-64 has about 15 million ==> 3,000 is 0.02%
age group 65-75 has just under 5 million ==> 4,000 is 0.08%

That does rather indicate the pressure to restart the economy
BUT
we don't know how those figure would rise if the UK went the Sweden route and we get into exponential infection growth

We could also make a crude national estimate of % COVID deaths in care homes to date

e.g. 100 x (total care home COVID deaths) / (estimated total care home residents in UK)

However, as Shoot has posted, these are probably not a consideration when it comes to the cost of restarting the economy

WhyNotMe40 · 28/04/2020 16:20

Agreed, thanks Shoots and BigChoc. My number crunching days are too long ago for me to make any sort of stab at it, so I really appreciate all the hard work being done for me!

BigChocFrenzy · 28/04/2020 16:29

The ONS is a national UK treasure, which Germany doesn't really have in that form

  • and this crisis shows it is one thing to copy from the UK !

Collating national statistics from the 16 German states, which have 401 administrative districts ... is neither simple nor quick

On Thursday we will - finally - get the German statistics ... probably just for March (!)
and will then be able to examine the "excess" deaths

I'd expect the major politicians to have had a heads up by now on the basic numbers, at least

I suspect the excess deaths are smaller than expected, because several politicians (not the AfD twats) have been calling recently for quicker relaxation
and it is not normal for politicians here to play to the gallery when they can so quickly look reckless

However, if the figures are a high % of the official RKI [hospital+care home]
then there will be much less pressure on Merkel to restart the economy more quickly

Pertella · 28/04/2020 16:31

These threads are moving pretty fast now!

tawnygrisettes · 28/04/2020 16:37

Why isn't the BBC showing the daily hospital death toll in the ticker as usual?

ShootsFruitAndLeaves · 28/04/2020 16:38

@Nquartz former computer programmer but I haven't done it for a while.

BigChocFrenzy · 28/04/2020 16:41

Sky have confirmed the hospital figures we saw earlier:

552 England (total 19,301)
70 Scotland (total 1,332)
17 Wales (total 813)

==> 639 GB + NI to come

BigChocFrenzy · 28/04/2020 16:46

Reminder - official ONS figures, not FT Giles estimates:

https://news.sky.com/story/coronavirus-deaths-of-hospital-patients-with-covid-19-rise-by-552-in-england-11980004

"In its latest estimate, the Office for National Statistics calculated the death toll
as of 17 April was 52% higher than the figure released by the government.

It found 21,284 people died with the virus by that date,
compared to the figure of 13,917 of hospital patient deaths announced on that date."

NewAccountForCorona · 28/04/2020 16:51

This is an interesting article from Sweden (despite my best efforts I'm not getting too far with translating the original stats, but this is a newspaper summary in English) - www.thelocal.se/20200427/sweden-publishes-new-statistics-on-coronavirus-death-toll

It seems that the advice given to Swedish people is very much like the advice given in other countries, just not legally enforced. So I would be interested to see whether, in general, people are actually carrying on as normal.

It also gives some facts about the deaths in Sweden - over 90% in people over 70, less than 1% in people under 50. They appear to be counting deaths where Covid is the direct cause - even if a patient has tested positive, it is only counted if Covid is on the death certificate.

There is a bit about excess deaths (presumably using their equivalent of ONS data) indicating that in Stockholm "Between March 30th and April 19th, the weekly death toll has been more than double the average for the same weeks over the past five years."

Sweden is currently 7th in Europe in the list of deaths per million population and is still climbing.