Daily numbers, graphs, analysis thread 6

[Barracker]

Welcome to thread 6 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.

The deaths in germany - not normalised wrt pop - are attached

Median age at death is 82
87% of deaths are in the 70+ age group, although only 19 % of cases are

So UK excess deaths were still rising for the data sets Shoots has plotted,
but hopefully we will see this levelling out and then dropping as more recent data becomes available
Maybe with some different timings for London and the regions.

As posted earlier, graphs of the European average of excess deaths has risen sharply,
with the UK and Italy even more than the average

368 UK deaths today.

It's always low Sundays/Mondays

True but that’s over 200 less than last Sunday.

It is always low Sunday and Monday but that is considerably lower than last weeks drop.

www.thetimes.co.uk/article/south-africa-a-nation-ravaged-by-hiv-is-flattening-the-coronavirus-curve-bp2dxhd3q?shareToken=f164976cf262786a6d86c553236ed1f0

Interesting article by Dr Max Price exploring why,South Africa has less cases and lower death rates than other countries, looks at factors including ethnicity, bcg and poverty. No current explanation.

That's true, but the number announced last Sunday's was around 600 iirc, so this would still appear to be a significant decrease.

"The attached graph shows deaths spiking sooner in Salford hence the virus got there first and spread so the lockdown, which the government fucked up on very badly by implementing a week or so too late, saved more Mancunians than it did Salfordians. In particular we see week 12 deaths still normal in Manchester, barely elevated in week 13, but well-elevated at that point in Salford."
As a Salford resident, I find this really interesting. What dates are weeks 12 and 13?
Anecdotally, there were some weird bugs going around here at the end of Feb, early March. I started with a very mild cough around then that I still have now although it is very slowly improving. I'm hoping it means I may have had it. My neighbour almost definitely has as had classic symptoms.

The death rate, whilst obviously always sad, will be last figure to drop. The hospital admissions is more important, regarding the trend (in my opinion) because that shows the actual numbers who have it and are admitted in hospital. That needs to flatten, which will obviously impact on the death rate. Many patients are admitted and some go to ICU and some don't, but it can take at least two or three weeks in hospital with this before they take a turn for the worse or better. Not all, but that is often the case. I know of a lady who was admitted to a Covid ward over 3 weeks ago, she is doing ok, but still not out of the woods. She became ill 5 weeks ago. She has told a friend that there are many on her ward that have been in for some weeks longer than her.

@larrygrylls this article is few weeks out of date but explains some of the variables in the modelling used, it doesn't say how small the population groupings go down to though, which I think was part of your question:

www.wired.com/story/the-mathematics-of-predicting-the-course-of-the-coronavirus/

Calendar weeks 12-13:

Sunday 15 - Saturday 28 March
(In the UK, this might be alternatively be 16-29 March)

Stockport (if anyone in north west) seems to have rocketed compared to its neighbouring borough Tameside.

"But for places in Western Europe and the United States, the scientific evidence is not particularly complicated.
If you act soon, with a firmer response, you limit early deaths and get through the initial phase more quickly,”^
Jenkins says.
That’s based less on models, she says, and more on research like a 2007 articlee_ about the “non-pharmaceutical interventions” different US cities used to respond to the 1918 influenza pandemic.
Places that locked down more quickly had death rates half as high as ones that waited"

That is important:
lockdown wasn't just based on models, but backed by expert opinion from epedemiologists, virologists etc using their own experiences.

Also, some models look OK for predicting the peak, but not what happens later on

e.g. IHME, designed to plan hospital resources, has reasonable peaks,
but then new deaths falling to zero over summer - on what evidence ?
Maybe the model is not really intended for use that far on in an epidemic ?
Or not wholly suitable for COVID ?

So we need - and will hopefully get - new or updated models that predict the long slow decline we are experiencing
and give scenarios for how much each relaxation measure later on may raise deaths

but we will all along need expert opinion based on other evidence and experience, not just models

So it turns out we have both OVER and under-reporting of covid-19 deaths.

Attached charts show deaths up by more than 140% in care homes, 80% in hospitals, and smaller rates elsewhere. No change at hospices.

The result is that care home deaths are a bigger proportion of the pie than usual.

There are a real extra 8,726 deaths in hospitals as a result of covid-19, but 11,118 covid-19 deaths have been diagnosed. This is not wrong, if the definition of covid-19 death is 'someone who is positive with covid-19'. One presumes the dead are all tested.

If we assume that covid-19 hasn't changed the number dying of other causes of death, we can assume that total excess mortality is all due to covid-19, however covid-19 might change the venue of death, e.g., rather than dying in a hospital, more people might be dying (of non-covid-19 causes) in their care home.

Therefore it's not impossible for more than the total number of excess hospital covid-19 deaths to be diagnosed as covid-19 positive without any of them having a different real cause of death than covid-19.

However we shouldn't assume this, as it's reasonable to assume that some people will die incidentally with covid-19 rather than of covid-19, and therefore if the hospitals are doing their jobs properly, this should be reflected in a number of covid-19-related deaths greater than the count of excess deaths, and this is the most reasonable explanation for this.

We can see that the number of covid-19 deaths as a proportion of the total excess in care homes is increasing week by week, but it is still very low at just 30% of excess mortality, meaning 70% of deaths are missed.

It follows therefore that we should adjust the age profile of covid-19 upwards significantly, as whereas we have in week 15 5285 non-care-home covid-19 death diagnoses, there were almost the same number, 5401, of excess deaths (while diagnoses exceed excess deaths for hospital, for deaths at home that is not true) for the same venues; we have 3657 care home excess deaths but only 1100 matching death certificates.

In other words, while 38% of excess mortality is in care homes, only 17% of covid-19 death certificates are from care homes.

The net result, obviously, is a higher average age of death.

I have a feeling the ONS weeks end on a Friday.

@ShootsFruitAndLeaves

Thank you so much. I was losing my patience with the death denier who wouldn’t or couldn’t explain her position but she 100% sure it was being over reported as Covid-19 but you have absolutely blown her (non) argument out of the water.

It is 368 in England, Scotland and Wales. NI haven't announced theirs yet.

And now they have, 5 in NI. Nice to see the numbers a bit lower today, even if it is a back log of reporting or whatever. Still feels like good news.

• • • DAILY UPDATE * * * Sunday APRIL 26th

Total UK cases: 152,840
New UK cases: 4,463
Total UK Deaths: 20,732
New UK Deaths: 413

No. of deaths by Age Group in England till 5PM 25th April.
0-19 Yrs : 9
20 - 39 Yrs : 134
40 - 59 Yrs : 1480
60 - 79 Yrs : 7190,
80+ : 9607
Total : 18420 (ysday 18084) - Increase by 336
Note - Figures provided above only for England nation not for whole UK from hospitals only.

Here's a graph showing that mortalities of young people are very substantially down considered over last 4 weeks of registration, for both sexes.

It shows clearly that young people are at far greater danger from normal daily life than from covid-19, albeit that the risk of death during daily life for the young is, like covid-19, small.

Second graph shows Week 15 only, for both sexes combined. This therefore contains more Covid-19 deaths as a % of the total. We'd expect to see any spikes here in deaths to appear in intermediate age groups as covid-19 starts to kill those infected a couple of weeks earlier.

Here we see 41 mortalities in the 25-29 age group, although 37, 37, 36, 39 are some of the previous weeks this year, so this is not statistically significant - random noise, in effect.

The 30-34 age group is below normal, while the 35-39 age group is the highest this year, while the 40-44 group is barely elevated.

The increase in mortality for older groups is roughly the same for each group, however the 90+ group seems slightly less affected than usual.

The way we expect this to break down is:

1. younger age groups are dying in smaller numbers than usual from random accidents, violence, etc., because of the lockdown. The risk of these causes of death does not necessarily increase with age, and for certain causes, decreases, but in absolute terms in the UK is small (certain other countries have much higher homicide and/or road deaths).
2. in general we expect people of all ages to die of disease/illness (so excluding the random accidents) in a curve that increases exponentially with age, although it is elevated during early infancy. The meaning of this exponential growth is that the risk of dying of illness is extremely small if you are say 20, but is thousands of times higher if you are 80.
3. We know that the risk of covid-19 death grows also exponentially with age and can be considered as such as a multiple of the risk of death from any cause
4. Hence at some point the curve of illness vastly exceeds the random death line. E.g., We expect around 650 per million 75-79 year olds to die in a given week, but only 12 per million 20-24 year olds.

If we consider that many of those 20-24 deaths will be from accidents, etc. then perhaps only 3 per million are from illness, while perhaps 600 per million of the 75-79 deaths are from illness.

If our lockdown reduces the average person's risk of death due to accidents by half, but covid-19 doubles the average person's risk of death from illness, then it's easy to see how we end up with big spikes in death count, because there are far more deaths from illness among the ageing/old as there are accidents in the young.

If we were able to break down deaths by age and cause, what we would expect to find in our graphs is the same % increase in risk of death from illness relative to base risk for all age groups. I.e. covid-19 DOES increase the risk of death from illness among young people, but because the base risk is so small, it's just not visible in the death totals, especially when you start altering behaviour which affects other more significant causes of death.

I.e. if you double the 600 per million deaths in the old to 1200 and double the 3 to 6 deaths in the young, then the former is very noticeable and the latter just disappears in the noise.

Thanks Shoots for that graph. Really interesting.
I really hope we don’t see the age brackets which are currently less than normal increase due to suicide

Bigchocfrenzy Thank you for the information on calendar weeks.