Daily numbers, graphs, analysis thread 8

[Barracker]

Welcome to thread 8 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.

Their deaths, even.

In fact cancer deaths fell by around 7% overall. Presumably in fact covid-19 finished them off. But many different cancers, so unless grouped it doesn't show up. Note that female cancers (ovarian, uterus, breast) were the only ones not to fall. This likely reflects both the much lower risk that women have from covid-19, but also the lower age of profile female cancer deaths.

That is March to April. We had 4,379 deaths recorded in March from covid-19 and 27,764 in April. Unfortunately the ONS does not seem to have thought to compare February (no covid-19 deaths).

Incidentally there were no deaths by suicide or homicide registered at all in April. However we should be careful drawing conclusions from this as this by date of registration, and it is to be presumed that such deaths would not be registered for many months, following an inquest, which presumably none took place at all in April.

Sorry if this is common knowledge but is there easily available graphs/csv data showing numbers of hospitalisations over time by UK region? Can't see any from a quick search of govt websites.

A rise in hospitalisations in the past week would presumably mean they got infected 1-2 weeks ago, is this right?

There are other death falls we should be very careful of for similar reasons. 'Accidental falls' fell from 130 to 27, land transport accidents from 6 to 0, homicides from to 0, but is this because coroners stopped doing their jobs, or because there was actually such a fall in these death causes.

Here's a chart showing death cause changes between March and April.

I've excluded things that seem like they should require an inquest, plus this is the top 20 causes, and I've grouped cancers.

We see a massive rise in Alzheimers death, 20% rises in

Symptoms, signs and ill-defined conditions
Heart failure and complications and ill-defined heart disease
Cardiac arrhythmias
Parkinson disease

There are ~20% falls or more in:

Acute respiratory infections other than influenza and pneumonia
Influenza and pneumonia

(but we should see February to be sure, as there were only 3,000 such deaths in March)

Septicaemia
Cardiomyopathy
Congenital malformations, deformations and chromosomal abnormalities
Aortic aneurysm and dissection
Diseases of the musculoskeletal system and connective tissue

I found the death causes here:

www.ons.gov.uk/releases/leadingcausesofdeathuk

Unfortunately it only seems to list the top 5 causes of death by year and sex and age group, so it's not that helpful, however 'Symptoms, signs and ill-defined conditions' is something that is only a leading cause in the oldest group of women

So it does look like there were thousands of old people dying of covid-19 who were NOT given a covid-19 death certificate but rather dementia, etc., since the 'old people' causes rose. While on the other hand perhaps 1000 people who would have died of other causes anyway in April have been given covid-19 death certificates.

Most of these cancer, etc. deaths would have been old men, so it's not a big deal so long as we focus on excess mortality by age and sex (none at all for young people and children, huge for older people). We can see plenty of excess mortality in 50+ year olds etc.

Sorry graph

The daily briefings have the graph that shows hospitalisations and you can access the data they use via the government website

www.gov.uk/government/publications/slides-and-datasets-to-accompany-coronavirus-press-conference-15-may-2020

article about prolonged COVID symptoms in patients. It's a self-referred questionnaire study of 640 respondents, but has some pretty graphs and raises the issue of ongoing medical needs of COVID survivors, and the limits of reliance on testing data.

Superb analyses, SHoots
I'm mentally digesting them

Many of the excess deaths not listed as COVID,
e.g. the astonishing increase in dementia / Alzheimer deaths, also increases in Parkinsons, serious CVD and lung problems, liver & kidney disease, cancer in the elderly ....

are likely to be because COVID weakened the very vulnerable and accelerated the progress of their disease(s)

I'd expect a drop in RTAs and work accidents,
also not surprised if there is a drop in deaths from non-domestic murders, which are the majority, especially alcohol or gang-related deaths of young males.

Reportedly about 5% of people who reported symptoms have suffered prolonged & mutliple ailments,
which are continuing human and economic cost.

We can't evaluate the importance of these atm, because we don't know if they'll nearly all be fit again within 6 months,
or if a significant number will have longterm continual or recurring health issues

imo, it will take at least a year true impact of COVID and also the merits of the vaious strategies of tackling it
I also think we'll find later that the best strategy varies for groups of countries that have many characteristics in common, but are different to other groups

e.g.
developing countries
Scandi / Nordic countries
Eastern Europe
rest of Europe - incl UK, France , Spain
USA
South America

I'm sure there is a drop in RTAs, but I'm just wondering aloud in a bit now in that we see a drop in mortality and deaths occurrences for the younger groups, but if perhaps these groups are happening in normal numbers but they are being underreported.

In other words we have both deaths by date of death and registration since covid-19 started, and both show big falls in the younger age group, but if the normal process of investigating these broke down because of covid-19 then we'd appear to have falls in both that don't actually exist.

This is the generalised 'death delay' report from 2018

www.ons.gov.uk/peoplepopulationandcommunity/birthsdeathsandmarriages/deaths/articles/impactofregistrationdelaysonmortalitystatisticsinenglandandwales/2018

There is official advice to indefinitely postpone major inquests

www.kingsleynapley.co.uk/insights/blogs/public-law-blog/the-inquest-process-during-covid-19-restrictions

Even with these measures, disruption and delay of inquest proceedings is inevitable. The Chief Coroner suggests that coroners should consider adjourning inquests, especially those which are likely to occupy significant time and judicial / staff effort.

The nature of inquests means that they are more reliant than other proceedings on medical expertise, and more likely to involve an investigation of conduct in a medical setting. The very substantial strain that COVID-19 is placing on the healthcare system is going to significantly reduce the availability of such expertise to coroners, and is likely to prevent medical organisations and their staff from participating in inquests.

Clearly the 0 counts for suicide and homicide in April are ridiculous nonsense and cannot possibly be true reflections of what happened.

So we'd need to look at the underlying causes of death for younger age groups

If we look at the 2008-2018 report, the largest death cause among males 20-34 was suicide.

There are 6000 suicides per year, which is 500 monthly.

The ONS stats for March show just 5 suicides, which is clearly wrong for occurrences and 0 for April, which is impossible.

The ONS data for 2008-2018 show for men 20-34

about 10 monthly homicides
about 25 monthly land transport accidents
about 75 monthly suicides
about 50 accidental poisonings (alcohol? overdose?)

In total this is 2000 deaths per year, in categories currently not being registered at all. That's 40 deaths per week, or 56% of the deaths for the male 20-34 group.

So we really can make no conclusions about young people's mortality for months and months, except to say that:

covid-19 death counts are tiny and not a significant risk for young people
said counts are unlikely an overstatement of risk, in that only 24 out of 385 aged 0-44 had cancer as the main other cause, and cancer as death cause is down by under 10% generally, so if the 385 (a small number, to be sure) is overstated it's by no more than a handful of people
we don't know whether total mortality for young people is down or up

The ONS stats for March show just 5 suicides, which is clearly wrong for occurrences and 0 for April, which is impossible.

Sorry that's just me being thick.

The data are by occurrence but they obviously don't include deaths which have happened but haven't been reported.

Therefore, perhaps 500 suicides occurred in March but only 5 have been registered to 5 May (the cut-off date), and none from April.

That seems entirely correct to me. Presumably as we go back by month we will see more and more deaths 'occurring' until it tails off (as eventually 100% of suicide deaths are registered). However of course the number of suicides is already fixed at the end of the month, just we don't report it till it's official!

Returning to R0:
We've discussed that it varies around the country

After lockdown is eased, most relevant for the country as a whole may be to exclude local hotspots which have been locked down again

As always, we must remember that R0 is estimating infection events that happened 1-2 weeks earlier - hence each estimate sits within a wide prediction band.

In Germany, the RKI (public health Inst) is using 2 different methods to calculate R0 - which give different numbers

• but both based on the number of days considered in the calculation, still a national R0, not regional:

https://www.rki.de/DE/Content/InfAZ/N/NeuartigesCoronavirus/Situationsberichte/2020-05-15-de.pdf??__blob=publicationFile

"The R-value reported to date reflects the trend in the number of new cases and can indicate possible changes in trend.

However, this value is sensitive to short-term changes in the number of cases

• such as those caused by individual outbreaks -

which can lead to relatively large fluctuations, especially in the case of a small number of new cases.

In addition to this sensitive R-value, the RKI therefore now provides a second more stable 7-day R-value,
which refers to a longer period of time and is therefore subject to less short-term fluctuations.

Both R-values are estimated on the basis of nowcasting.
The nowcasting ended on the 10th of May 2020, so no reliable statement can be made about the number of new cases in the last 3 days.

The previously reported sensitive R-value can be estimated by using a moving 4-day average of the number of new cases estimated by nowcasting.

It then compares the 4-day average value of the new cases on one day with the corresponding average value four days before.

Because the new cases were infected 4 to 6 days before onset of illness, this means that they ocurred 8 to 13 days ago.

The previous R-value, which is reported today, thus, maps the infection events occurring approximately one to two weeks ago.

The current estimate is R= 0.80 (95% prediction interval: 0.67 – 0.97)
and is based on electronically notified cases as of 15/05/2020, 12:00 AM.

Similarly, the 7-day R-value is estimated by using a moving 7-day average of the nowcasting curve.

This compensates for fluctuations more effectively.
The 7-day R-value then compares the 7-day average of the new cases on one day with the 7-day average four days earlier.

Because the cases were infected 4 to 6 days before onset of illness, this means they ocurred 8 to 16 days ago.

The 7-day R thus maps the infection events of about one to a little more than two weeks ago.

The 7-day R-value is estimated at 0.90 (95% predictation interval: 0.83 - 0.98)
and is based on electronically notified cases as of 15/05/2020, 12:00 AM."

The RKI "Nowcasting" chart illustrates how R0 is very much a rough approximation:

InMySpareTime

Really not good news about the length of illness. And interesting that many were more ill after 2nd week. When many are expected back at work.

Not good news it is longer in asthmatic and vit d deficiency when it looked like they were taking vit d many of them. Although i would expect an asthmatic to have a long cough.
If people know they are vit d deficient surely they are takig supplements already?? So shouldnt still be deficienct, unless they maybe have some issue using it.

anyone know what country the survivors survey is from? Vitamin D tests are not often available here so I'm assuming it's a mix of American and British and the true level of vitamin D deficiency would be higher.

Illness tends to deplete vitamin D levels anyway.

Survivors survey was from several countries, mostly US and UK, there's a table in the article showing locations of respondents.

There's definitely been murder during the period of April. I've seen reports in the news.

Very good that Jenny Harries highlighted today that with increasing testing you'd perhaps expect to find more cases but it's still dropping (although levelled out at approx 3.5k). That must show our cases have dropped.
Interesting R has risen. I had thought it would be after VE Day and cases coming out but I understand this is from further back data? Did something happen 3/4 weeks ago that would account for this? Nothing springs to my mind!

Easter Bank Holiday weekend/heatwave?

They have said that the rise in R value can principally be attributed to cases in care homes and hospitals rather than in the community.

Globally, deaths exceed 300,000 but falling now.
Interesting history charts of how this pandemic has shifted betwen world regions

The USA accounts for 30% of all daily COVID deaths, with NY having its own strip in the regional chart.
Unfortunately so does the UK, along with Spain & Italy

Brazil 2nd to US in daily deaths now

This is probably a very stupid question but my brain is fried at the moment so please forgive me!

I'm a bit puzzled by the PHE data and the data given for the number of cases per day.

On PHE, a weekly report is produced of data. There is a graph showing number of cases according to specimen date. As an example, of specimens taken on 9th May, 777 cases were identified. In the few days after, figures are lower (though may not be complete yet). But, on the daily briefing and other data dashboards, we see new cases in the thousands e.g. 3000+

Are these data positive tests from a range of specimen dates, say 3 or 4 days? Which is why they are much higher than the specimen date figures on the PHE graph?

As I say, probably a daft question but my brain has given up.

Thanks for these threads I've been following them for a while now and the information and data analysis here is so helpful.

I have gathered from shootsfruitandleaves posts that what covid-19 does is increase the risk of death from disease for everyone of all ages and that this then means that the older you are the greater your likelihood of dying is as your "normal" underlying risk of death from disease is higher as you get older.

My first question is, could you point me to a chart showing risk of death if you catch covid for each age group?

My second question is, is the increase of risk the same at each age? Ie is it always you are for example two times more likely to die of Covid19?

Thank you

My local newspaper has reported at least 3 suicides in the last month or so. I thought it unusual but at the same time I've been looking out for stories about younger people dying more.

If that's just my area I think it's safe to say there is a gap in the data.

Also interesting news that they've found mini blood clots in a third of hospitalised patients in the UK and a study from Wuhan has picked up three markers that can predict death within 10 days with 90% accuracy.
news.yahoo.com/3-clues-coronavirus-patients-blood-115100205.html

This feels like significant progress in understanding (and therefore being able to treat) the disease better.

All seems to be pointing to the blood.

I'm not v active so bloodcots are one of my anxieties. Does 10 day prediction of death pretty much mean they could tell you "btw you will get worse and worse and then doe by day 10!?" That's frightening.