Daily numbers, graphs, analysis thread 12

[BigChocFrenzy]

Welcome to thread 12 of the daily updates

Resource links:

Slides & data UK govt pressers
UK dashboard sub-national data, local authorities
Beta Uk dashboard deaths, cases, hospitals, tests, partially sub-national
UK stats updated daily by PHE & DHSC
ONS UK statistics for CV related deaths, released weekly each Tuesday
PHE surveillance report infections & deaths released every Thursday with sep. infographic
NHS England stats including breakdown by Hospital Trust
FT Daily updates
HSJ Healthcare updates
Worldometer UK page
Plot FT graphs compare countries deaths, cases / million pop. / log / linear
Covidly.com filter graphs compare countries
Plot COVID Graphs Our World in Data

We welcome factual, data driven, and civil discussions from all contributors 📈📶👍

This is new I think -

phe.maps.arcgis.com/apps/webappviewer/index.html?id=47574f7a6e454dc6a42c5f6912ed7076

Cases by middle super output area, England; put in your postcode and see how many cases there have been in the past week in your local area.

Yes I think that is new @Patricia. And useful, though only really much use in the areas with higher rates because the numbers under 3 are suppressed. I think they're being over cautious doing that tbh - knowing a number was 1 or 2 or zero in your area isn't going to tell you who the people with it are. Not on its own at that spatial scale.

@bigchoc I think your rough estimates of people dying after having had Covid could be under estimates. Because the age profile of those with a positive test is very skewed towards the older age groups. See figure 2 on the weekly surveillance report www.gov.uk/government/news/weekly-covid-19-surveillance-report-published

(And note the different scales on 2a and 2b)

So you'd need to do something a little more sophisticated to work out the number who would have died of other causes but it's going to be higher than the crude rate you calculated, and as time goes on will become increasingly significant.

@Firefliess

Surely when doctors fill in death certificates they use sensible judgement about what's relevant to the cause of death - so I don't see why they couldn't do so with Covid too. Eg if someone had it months ago and was obviously better then it's not a factor. If they spend the last 10 weeks of their life on a ventilator then it is. Or else just put a sensible time limit on it of 3 months or something.

Is the issue though (looking at the graph Patriciaholm posted) that PHE aren't looking at the death certificate before saying it's a death with covid. So John Smith could die in a car crash and his death cert say that but whoever is correlating deaths in phe will just look at the fact he had a positive test 2 months before and then report him as a death with covid. Surely the sensible thing would be for PHE to look at the death certificate as well and record deaths where there is mention of covid on the death cert and a positive test? Or if they really want to capture all if a respiratory illness was listed (although this might cause problems when we hit flu season not to mention pneumonia is a common cause of death anyway). A time limit is tricky as we know some do die weeks later (the Broadway actor for example) but perhaps is the only sensible cause of action.

Derby, in the same way Ebola can. She explained it better than me and is a Cambridge science graduate! It's been around for a while but didn't have the "right" conditions till now.

I would expect that at some point, the ONS will come up with a sensible rule for what constitutes a death of/with Covid, and when the positive test was far enough removed to not be relevant. This is because such data is relied on and used for many purposes.
I agree with BigChoc that it is ONS analysis, and what we actuaries call analysis of actual versus expected deaths (which is, what BigChoc described - looking at each age group in the population, and comparing actual numbers to expected numbers) that gives the full picture of the impact of Covid, both directly and indirectly. PHE reporting can only be an estimate of this. The ONS Excess death measure is a first order estimate of this. For a more complex calculation see www.actuaries.org.uk/system/files/field/document/Mortality-monitor-Week-26-2020-v01-2020-07-07.pdf which is comparing actual numbers in 2020 to “expected” numbers based on what happened in 2019.

For anyone wanting to dabble in further calculations, the latest mortality table for the population is English Life Table No. 17.

@Aprillady The excess deaths are the best way of understanding the real numbers who died early on when numbers were high. But they're no use at all currently when numbers dying are very low. And given that death rates are one of the useful things to look at to understand whatever cases are rising it falling, it matters to get them right, not just right to within 1,000 a day, or whatever level you can really estimate from excess death figures

@NeurotrashWarrior

Derby, in the same way Ebola can. She explained it better than me and is a Cambridge science graduate! It's been around for a while but didn't have the "right" conditions till now.

Could be like Herpes (chicken pox/shingles); although unlikely.

Recent report in Guardian with 87% mink infected on a farm in Spain, so more likely that just lurking in another species and jumps.

I am not sure about the word "dormant". What the study described in the NY Times article suggests is that COVID was introduced to Europe and US numerous times. Quite a few of the very first cases did not lead to any further transmissions. Some triggered small outbreaks but (for whatever reason) the infections eventually died down.

Around the world, the new study suggests, the coronavirus arrived more than once without starting runaway outbreaks. In these cases, there was little or no transmission, and the virus simply died out.

The lineage that "ran wild" in Europe and the US was supposedly introduced to Europe in early or mid February.

At the same time the study did not seem to find any significant mutations between the "older" and "newer" C19 straits. In other words, it is not obvious that the new strait is more contagious and it is possible that the transmission might have been accelerated by environmental factors (e.g. outside temperature? other unknown variables?).

Of course, there is also a possibility that the study may be wrong.

Most of COVID publications these days are non peer reviewed / pre-prints so they should be taken with a healthy degree of skepticism.

Looks like the issue of over reporting deaths is being tackled BBC News - Hancock calls for urgent review into coronavirus death data in England
www.bbc.co.uk/news/health-53443724

@wintertravel1980

Of course, there is also a possibility that the study may be wrong.

Most of COVID publications these days are non peer reviewed / pre-prints so they should be taken with a healthy degree of skepticism.

Yes the genome studies all indicate a jump late 2019.

firefliess As I said, it was a crude 1st ballpark figure
I over-estimated by taking 300k, instead of 293k - 45k dead - xxk active cases (20k?)
so I should have taken about 230k, but thought this would offset age to some extent

I looked at Pillar 1 & 2 cases by age - probably best to take the mean of the total for a 2nd estimate,
but I haven't seen this anywhere, so it needs calculating

Mean & media age of cases in Germany is 49 - but the v elderly were better protected - so Uk figure looks higher from ONS chart

Also, the number of tests was v low in March and still low in April,
so a more accurate 2nd estimate needs to take the number of accumulated positives per month and calculate at each stage

We should also count as a COVID death the minority still suffering symptoms sveral weeks later

We also need to consider that the average age of cases is falling in many countries
e.g. in USA and Germany, current average age of infection is < 40

==> would be v useful if anyone has a weekly chart for UK mean case age ?

Yes, excess deaths are the usual scientific basis for an overview of the epidemic impact in the various countries

• and help refute the "it was just flu" ! -

but now we need a more accurate metric as excess deaths returned to normal, weeks ago and (UK) COVID deaths reached 100 - 200 or lower

Counting seems specifically an English / PHE problem (hence incl Wales) as that article confirms:

"Scotland and Northern Ireland, for example, only include deaths in their daily count if someone died within 28 days of a positive test."

I thought that was the global standard

• which still includes the v small % of active cases of people hospitalised / ICU for longer than 28 days

The 'overcounting' thing is going to be fuel for stupid arseholes on Facebook for months.

Some utter arsehole is describing it as a 'staggering fiasco'

twitter.com/JoeMurphyLondon/status/1284039715415613440

People with more than two fucking braincells were already clear months ago that there was massive undercounting at the beginning and then as time went on full or even overcounting.

The bit of overcounting now doesn't exceed the undercounting before.

It's not a 'staggering fiasco'

Utter wanker.

Also there is cross-referencing between the PHE deaths and the ONS deaths in the ONS weekly death publications, so it's not as if any of this should come as a surprise to anyone who cares.

there is cross-referencing between the PHE deaths and the ONS deaths in the ONS weekly death publications, so it's not as if any of this should come as a surprise to anyone who cares

Maybe not, but to anyone who just takes the figures at face value, or can't easily access to decent information, it's going to look as if Covid deaths are never-ending even if they actually drop to almost zero

And I'm just not sure who that's helping

@ShootsFruitsAndLeaves

The 'overcounting' thing is going to be fuel for stupid arseholes on Facebook for months.

Some utter arsehole is describing it as a 'staggering fiasco'

twitter.com/JoeMurphyLondon/status/1284039715415613440

People with more than two fucking braincells were already clear months ago that there was massive undercounting at the beginning and then as time went on full or even overcounting.

The bit of overcounting now doesn't exceed the undercounting before.

It's not a 'staggering fiasco'

Utter wanker.

Yes, exactly. The actual numbers involved aren't going to be big. And they haven't exactly been hiding it - it's on the Govt website and in the documents detailing how the stats are calculated and has been for weeks it seems.

What it does, though, is elongate the tail; now we are at a stage of 20 or so hospital deaths a day, the small but overmeasured non-hospital deaths (and the mad swings they take) are likely to be having an overly negative affect on people's perception of the current situation. If daily deaths were steady state-ing at say 30 a day, rather than swinging from 11-150, perception of the current situation might be a bit different.

From the Guardian just now -
"On the PHE anomaly, one of the authors of the article that prompted a review, Dr Yoon Loke, said the method chose by PHE would be “reasonably accurate” at the beginning of the pandemic when there were few people in the community who had survived the virus, but would become increasingly flawed as time went on, with ever more people surviving coronavirus only to die of something else. He estimates that more than 1,000 deaths may have been wrongly attributed to coronavirus by PHE".

So a very small proportion now, but more of an issue going forward.

I've probably underestimated the "natural" deaths in my ballpark daily 7.5
but it won't have made a significant difference to PHE stats,
certainly not before we reached the "normal" excess deaths level

If anyone has a weekly table of #cases and median age, that would enable a more accurate figure

However, as shoots explained, the ONS estimated that a much more significant chunk of COVID deaths were not being counted earlier

e.g. their excellent Nick Stripe:

twitter.com/NickStripe_ONS/status/1268823005305733125

"The balance of evidence so far points to undiagnosed COVID in the elderly being the most likely explanation for a majority of excess deaths that did not mention CV on certs"

However, the "it's just flu" idiots will seize on this 3rd order inaccuracy to try to follow the USA fuckwits into abandoning all measures
and hence increasing infections again

When & how did the Uk import the cultural war on COVID and SD, masks etc ?
Maybe because the govt left / is a vaccuum for others to fill

Can y'all clear something up for me?

Does the Beta site show just Pillar 1 for LTLA or Pillar 1 & 2?

For example Stockton-on-Tees:
coronavirus-staging.data.gov.uk/cases?areaType=ltla&areaName=Stockton-on-Tees

Not my area but I'm trying to get accurate-ish stats for a friend.

@NightmareLoon

Can y'all clear something up for me?

Does the Beta site show just Pillar 1 for LTLA or Pillar 1 & 2?

For example Stockton-on-Tees:
coronavirus-staging.data.gov.uk/cases?areaType=ltla&areaName=Stockton-on-Tees

Not my area but I'm trying to get accurate-ish stats for a friend.

Both Pillars.