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.

Bigchoc,

I don’t think there is a better model, really.

I know you replied to my previous post, but all the points you brought up could be taken into account within the parameters of any epidemiological model.

Someone sent me this one:

gabgoh.github.io/COVID/index.html

You can play with the parameters.

There is no point in refining the model better than the confidence of your input data. As many (including you) keep saying, we need far higher quality data as, especially with exponential models, small changes in input lead to huge changes in output.

Whitty briefing:

He said R0 target probably in a range between 0.5 and 1, probably around the mid point.
< so target about 0.75 ? >

and repeated that keeping R0 below 1 is essential, as above that, COVID will spread at an exponential rate.

He says we have to look in the long run, “and this has got a very long way to run”
and that is why he is very cautious about putting numbers on likely deaths.

He presented hospitals & Critical Care useage, indicating peak has passed and curves aew slowly reducing
^

Yep, larry I'm just anxious about the cumulative erors with exponential growth
It is so dangerous that this uncertainty means the govt has to err on the side of caution when relaxing measures
Hence the economy suffers more

Bigchoc,

Thanks for that.

Looking at the chart of number of people in hospital the peaks ALL occurred about April 10th. There was no regional difference, which is as you would expect. Also the London rise was far steeper which, given its population density and international population, is also as expected.

Interestingly, though, London also seems to have come down very steeply. I would have thought that we (I am in London) would struggle to get an R0 close to the other regions post lock down but I think there is a big ‘fear factor’ here, so maybe Londoners have been more observant.

Also, there is always the possibility of a bug in a very old model coded years ago, not documented or released
and then tweaked to handle COVID

Maybe some old data assumptions about NHS beds e.g. like IHME

Alternatively, release how the Imperial forecasts have changed as rl data has been used
e.g. IHME went from 66,000 UK deaths to 37,000 and now 32,000
(
However, IHME have that unconvincing zero daily death rate for the summer
Does the Imperial model have somethung different there ? If not ....

I have also put this link on the studies thread. Chris Witty is the current Professor of Physic at Gresham College
He is doing a video lecture on COVID 19 on Thursday.
www.gresham.ac.uk/lectures-and-events/covid-19

very tentative hypothesis;

Could it be that each severity of lockdown has a "lower limit" to how far it can reduce the curve

So even if peaks were at different levels, London and the regions will still end up at the same plateau level ?

And similarly regions in Italy will eventually end up at a similar (Italy) plateau level for their stricter lockdown ?

That's of course if infections continue to circulate and infection rate becomes more uniform

Won't happen in somewhere like the USA

Thanks, Chaz Noted in my calendar for Thursday

IHME went from 66,000 UK deaths to 37,000 and now 32,000

I thought that once non-hospital deaths were included we were already well over 32,000!

I have read that a 2nd peak could be worse.

Given that so many people have said that they had Coronavirus symptoms going back to last autumn, could that have been the 1st wave and that we are currently in the 2nd wave?

I think IHME only deal with hospital deaths, because their basic purpose is enable hospitals to organise resources

Before we can say that any more than a handful of people had COVID much earlier,
we need data from nationwide sampling and antibody or other tests

There are so many winter bugs around, that such claims are very unreliable

I had a dreadful respiratory illness that lasted 5 weeks, mostly in bed and had many COVID symptoms and took me 4 months to return to my former gym fitness
I would have assumed that was COVID .... except it was 23 years ago and my GP diagnosed flu and a relapse

This is the problem with a novel Coronavirus:
so much is unknown

We don't know what % of people who recover from it will have immunity, what % will have partial immunity

We don't know what % of people, adult or kids, who recover from it will suffer some longterm effects

We don't know if COVID infections and deaths will reduce significantly over summer

• particularly as we will be ending lockdown and relaxing other measures too

We don't know if there will be a possibly much worse 2nd wave afterwards,
since it could build up from September and last until May

And a 3rd ... until a vaccine ... whenever that becomes available

New treatments - not bleach - will presumably be developed in the meantime in parallel,
which should hopefully reduce the death rate,
but we don't know by how much.

Although we cannot know with 100% certainty that people who recover will get immunity, it is a very high probability that they will. Are there any known viruses where, without significant mutation, the majority of non immunocompromised people do not develop immunity? And we know that, due to its structure Corona SARS-2 does not significantly mutate.

Corona viruses have been with us for a long time and we do know a fair bit about them.

Of course, nasty surprises could be ahead but I am very pleased that this came now, rather than 30 years ago, before the huge advances in antivirals and vaccine technologies.

I will stick my neck out and say that I suspect that we will likely get a positive surprise in terms of time frames and treatments.

I am less optimistic about how the world will deal with the economic and political fallout on the other side.

I agree most people should get immunity after recovery, but we don't know if this is 70% or 99% or somewhere between

Those working on vaccines around the world sound optimistic, so personally I won't worry about that unless they change their tunes

The recent Chinese anti-viral study didn't work out,
but I expect there will be either existing meds found useful to treat some patients by EOY
or new meds will be developed, possibly in a similar timeframe to vaccine

As long as exponential growth remains possible and we don't have reliable treatments or a vaccine,
then imo the economy will be running at most at the 90% level
If we relax too much and get further serious waves - with R0 > 1 - we would probably cycle in & out of lockdown again

Norovirus. Winter vomiting bug. Immunity only lasts max 2 years.

2 years would be enough to significantly slow the rate of transmission I would have thought. Plus we’ll be there or there abouts with a vaccine (with any luck).

I’m not suggesting we aim for natural “herd immunity” I just think even short term immunity would help.

Here are some interesting graphs.

Essentially it seems to show clustering of care home deaths rates by region even with wildly different hospital death rates.

We see for the South West, that there are 78 per million care home deaths, 43 per million deaths at home, but just 46 in hospital. Meanwhile the West Midlands has 104 in care homes, 54 at home, and 203 in hospital.

If we look at the age profiles we find an explanation for this - London is much younger than everywhere else, at 36.7, and 3.4% over 80, while the SW is oldest at 42.7, with 6.2% of the population over 80. Other regions have between 5.0% and 5.6% over 80, and an average age of 40.1 to 41.6

So in fact the 78 per million care home deaths in the SW would be the equivalent of about 40 per million in London, which makes the figures less anomalous. And in reality perhaps less, as presumably older Londoners are less likely to be in care homes and instead at home, and maybe be sent to neighbouring regions

So we do indeed need to take logarithms of the age of everyone in the population, and then compare that total with the total excess deaths to work out how bad covid-19 has hit that area. In other words there isn't really a hard limit on covid-19 deaths in that relatively similar death rates in different regions in care homes does not imply that covid-19 just does its thing.

To get a better picture of what covid-19 really could do to care homes we'd need to take the deaths in London and re-base them by the actual care home population of London, not the total population of London.

There don't seem to be up-to-date statistics on this, but the 2011 census had 35,000 care home places in London, which is 0.43% of the then population, and 0.81% in the NE. It's likely that the 0.43% will have shrunk by now, as London has added another million people since then - there might be more than double the number of care homes per capita in the NE than London. Thus we see that with London as the worst case, where the are 1200 extra deaths in 4 weeks to 10 April, and likely to be 2,000+ by this point. This is something around 5% of the total population.

With 400k care home residents in the UK, we can see that covid-19 could kill 20k without much trouble, though so far that figure is below 10k.

With London likely over 1k/million excess deaths by today, it's clear that the 1,000 per million (0.1%) deaths per million is certainly NOT the worst case, as London has such a young population, with no lockdown other cities could have been worse hit and potentially face much higher death rates than that, so in some ways it's a blessing that London is first to be hit in that if you allowed older areas to get to 'totally fucked' stage, as in Northern Italy, then you end up with a much higher death total. When you let everyone get infected AND have an ageing population then you get mass carnage. London was shielded from that by its youthful population.

I wonder about the % of elderly population in some other large UK cities
e.g. Bham, Manchester, Newcastle, Brighton & Hove, Glasgow

To examine the effect of lockdown / no lockdown,

I compared deaths for Sweden to other Scandi countries incl Finland
and also added population:

(These are just official deaths, without care homes etc)
Striking difference in the height of what looks like Sweden's plateau

Now if we think how the UK deaths would look without lockdown ....

Now that most European countries have peaked,

it could at this stage be useful to consider population-normalised deaths, to consider the impact / pressure - political / economic - of those deaths
(and maybe also indicate the kind of infection rate present, but only if the death rate is similar)

However, immediately clear how difficult it is to compare, when some countries include care home deaths and some don't

So, 3 graphs of comparable countries:

1. All countries
2. Just countries without care home deaths - includes UK, which doesn't look too bad when normalised wrt pop.
3. Just countries with care home deaths

and normalised curves of deaths for Scandinavian countries

Again, Sweden very different to the others who locked down