UKH

A new guardian article links the various measures of covid deaths, with some useful plots.

https://www.theguardian.com/world/2021/jan/22/uk-official-covid-death-toll-...

Straight off the bat in the very first paragraph there is an appalling statistical abuse.

People who go to hospital with covid are almost all old and/or have serious underlying conditions. This means the chance of them being admitted to hospital within any given 5 month stretch is probably reasonably high. So how much of that "30%" is anything at all to do with covid?? They don't say, but the paragraph is designed to make the reader think that 30% it's the whole 30%. At the end of the paragraph, the "1 in 8 deaths" are not specifically classified as covid related or not but there is a clear inference that they are covid related (I'm sure some are) without supplying any evidence or even an explicit assertion to support this.

When this type of garbage is in the first paragraph, I'm not reading the rest, sorry.

A previous article dealt with that research:

https://www.theguardian.com/society/2021/jan/18/almost-30-of-covid-patients...

"The findings are based on data from the ONS and general practitioners in England. A total of 47,780 individuals who had a hospital episode between 1 January 2020 and 31 August 2020 with a primary diagnosis of Covid-19 were compared with a control group who did not have Covid-19.

Of the 47,780, 29.4% were readmitted within 140 days of discharge and 12.3% died. The rate of readmission was 3.5 times greater, and the death rate seven times higher, than those in the control group, the researchers found.

The risk of post-discharge illness – such as respiratory conditions, diabetes and problems with the heart, liver and kidneys – in Covid-19 patients was higher compared with the control group. That risk was also greater in younger and ethnic minority individuals compared with those aged 70 and above and white people.

“This matters. Long Covid at this level of morbidity and new disease is absolutely as important as the number of people dying,” said Summers."

The paper it is based on is here:

https://www.medrxiv.org/content/10.1101/2021.01.15.21249885v1.full.pdf

"almost one in eight of them die"... so.... brits are now 87.5% immortal? Awesome.

EDIT

So you don't actually disagree with the article.

"According to the paper by the University of Leicester 30% of Covid-19 patients discharged from English hospitals were readmitted within five months and almost one in eight of them die, raising further concerns over the accuracy of the widely quoted official figure."

They should have a bit more data but that does seem high. If they broke it down by age they could look at the chance of say a 70 year old being readmitted in the next 5 months versus a 70 year old with covid.

But even so 30% seems very high. 

Looking at excess death totals in the US it does (or did I've not seen the latest data), that covid deaths are likely underestimated. We have a 1.7% fatality rate in the US from Covid but it's more likely the true death rate is closer to 0.6% (from a Lancet paper), so we're likely missing about 2/3rds of cases (not deaths, we're missing the asymptomatic/low symptom cases).

Anecdotally, I've known more people this year die of heart attacks than previous years, 4 friends/colleagues who were all in decent health, and I wonder if that was a covid issue. It's something we'll hopefully understand more in time but we do not it affects the clotting system.

Why not read the rest of the article though, it does discuss excess deaths. 

Do you have any evidence to back this up? 40% of admissions are under 65 (the medical definition of old), 25% of admissions are in the under 55s (1) They will not all have serious underlying conditions. The median age in ITU is 61, the vast majority do not have any serious underlying conditions, and indeed 7% are pregnant women. (2)

1) https://www.england.nhs.uk/statistics/wp-content/uploads/sites/2/2021/01/Co...

2) https://www.icnarc.org/DataServices/Attachments/Download/4cd9c693-6657-eb11...

(edited as I put 50 instead of 40%)

Can you actually evidence that statement?  If not, the rest of your post is cast in to doubt.

What I find striking is that the various measures including excess deaths look pretty much the same on the graph and all of the various measures of deaths respond in the same way to a lockdown.

There are all sorts of valid* reasons why lockdown might cause some excess deaths. If excess deaths due to lockdown are already happening, lockdown scepticism requires a credible theory why these multiple causes of excess death respond (in a good way) to lockdown and on the same timescale (three week lag behind infections) as the identified infectious disease.

To me this strongly suggests that the deaths (various measures) are overwhelmingly caused by a single infectious disease and not by lockdowns or other causes*.

*people not going to A&E or GP, stress, poverty, postponed hospital treatments etc

For the England data if you define old as over 65 its about 2/3rds....  add in underlying health conditions and maybe over 3/4. Hence, I don't think that's unreasonable but given henwardian was so fussy about stats in the Guardian article and of those readmission deaths (where compared to a control group) 7 out of 8 look like covid related deaths, he is more sloppy than the Guardian.

https://coronavirus.data.gov.uk/details/healthcare?areaType=nation&area...

One point on excess deaths is I think you have to be very careful with them, as flu levels are so low. Current excess deaths compared to an average year will be a significant underestimate.

In reply to elsewhere

I agree its compelling evidence that most of the unknown excess deaths are covid related.

I think the reality is we've become accustomed to folk who may have died in their 70s, bring kept alive longer through better drugs and treatment over the last few decades, covid is basically robbing them of those extra years. It places so much stress on their organs they can't cope, many who make it out of hospital seem to die of 'natural' causes a month or two later and because it's beyond the 28 day covid test rule aren't included in the covid stats.

This is a useful graph:

https://twitter.com/beajayemac/status/1352610843515641859/photo/1

I don't think that is the case.  

People who are young and healthy like to believe this.  

For the whole pandemic it was.  In the last month, that has shifted significantly so that rather than about 33% being under 65, it's not about 40% of admissions.

The ONS report you cite makes it quite clear that Covid is a major causal factor.  I don't really understand the attempt to dismiss the affects on younger people as being due to other underlying conditions - that's a label which fits an awful lot of people many of whom have decades of life ahead of them - including decades of working / tax paying life.  Not all of those conditions will affect their covid outcomes much either.

Odd to dismiss biassed reporting with such a biassed dismissal.  Reminds me of the old Wintertree family motto.  "Two wrongs don't make a right, it normally takes a lot more..."

It would be a lot more useful if the curves were normalised to the proportion of the population in each of the age-group tranches.

I can put it even more simply; if the excess deaths seen this last 10 months are not due to covid 19, then what caused them?

And yes, the underlying expected level of deaths was low in 2020 because of the very benign weather in especially the first 7 months of the year.

I was also struck by one of the patients in the BBC reports from the Royal London; one 28-yr old, otherwise healthy patient had been on ventilation for three weeks. If he was tested and found positive a week before being put on a ventilator (not unreasonable, given that people don't suddenly require ventilation), if he dies, he will not be counted as a covid casualty, since he would fall outside the 28 day window.

How many other cases are there like that...?

I'm amused by the aspect that is similar to the emotiveness that likely causes Muphry's law.  People moaning about statistics, in their criticisms, too often make even bigger errors themselves. Others just hate the Grauniad (I think it's the best quality free online paper in the UK and normally excellent on covid coverage)

There was a graph once...around 10% of those who remain in hospital I think. Plus those readmissions and subsequent deaths will rarely be within the 28 day window.

Except the Leicester paper indicates the risk is greater for the under 70s.

True, excess deaths will just give a very rough idea. It's going to take a lot of unpacking, travel deaths/car accidents presumably declined, but you'd imagine suicides may have increased.

But underlying health conditions are more benign than people think, it can be diabetes, type 1 and type 2, obesity etc. ~40% of the US are obese but many aren't morbidly obese.

I know a 73 year old who just died of covid who was physically in good shape but had had a heart attack a year back. 

There is also a "deaths within 60 days" measure available from the dashboard API - there's a plot of recent data below.  Although this measure does not get in to cause of death, it does bound the number dying in the next 32 days to be quite low, and therefore the number dying due to their infection being equally or more low.  

One notable analysis is that the "60 days" curve is not (60/28)× larger than the "28 days" curve, as one would expect if there was an epidemic of people being run over by busses after their false positive test.  I consider this a gold standard rebuttal of the "casedemic" idiots.  

More clear water appears between the two curves about the time the new variant started to drive things.  That's something I'd be interested to know more about.

Annual road deaths (3 per 100,000 in the UK) are much lower and suicides (11 per 100,00 in the UK)  are  lower than flu deaths (around 30 per 100,000 UK) on a population basis.

This may just be a question of scaling: have you looked at the ratio of the two curves through time?

Definitely not the case and least in my sisters hospital and my BiL's hospital too.

Most people think it's either die (if you're old at least) or get better, there's a massive in-between group who are suffering long after leaving hospital, no one is sure if the damage will be permanent yet either.

Flu cases in Australia this year were almost non-existent, because of people socially distancing and mask use. I'd imagine the number of RTAs has reduced also, less traffic, especially during LD1

Is that actual clear water, or just more visible because everything is scaled up? Does the water stay clear if the figures are scaled by absolute case numbers?

There does seem to be a step change from about 80 days.

We're seeing it at school, no flu at all. But that's fuel for the conspiracy theorists that we are mixing flu and covid together and not that social distancing and masks work..

I think it's "real" clear water.

Here's a plot of the ratio between the two curves. 

1. At the end of summer (left side), the ratio was much higher.  This I think is the tail end of patients from the first wave dominating the data against the falling case level.  The absolute levels of both were very low and so this clear space wasn't visible.
2. The gap - in fractional terms - peaked at around day 100 (early December) on this plot and is decreasing again now.

Thinking about it, I think the second peak in this plot is the same sort of thing - when cases were in decay, the 28-day measure falls more rapidly than the 60-day measure, creating more space between the.  Then, when cases are rising, the 28-day measure rises more rapidly, reducing the space.

So, it's probably nothing more than the effect of latency in rising and falling phases.  Should have figured that one out sooner...

Heard the same myself many times, likewise, "well it's just flu!"

The 28 day rule might miss some covid deaths but I suppose it was chosen to provide a good comparative standard for tracking the effects of the pandemic. (Alternatively some cynics might say it was chosen because alternative measures show more deaths.)

https://www.ons.gov.uk/peoplepopulationandcommunity/healthandsocialcare/con...

See figure 7. Almost all hospital covid admissions are for old people.

Edit:

Ok, looked at that document a bit more thoroughly as it is strange that it looks very different from the one I linked above (they are showing absolute numbers and rates so not directly comparable as the number of people in each age group is a long way from being identical but this shouln't make such a huge difference I don't think). I think one of the principal differences is between "admitted with covid" (my link) and "in hospital but have covid" (your link). I think your link might include a lot of people who either caught covid in hospital while there for other reasons or were admitted for other reasons but also turned out to be covid-positive.

The key factor I was trying to get at was that if the _reason_ a person is in hospital is specifically severe covid, they are almost all old/underlying condition people.

Ah, great to have a bit of context, I'd still point out that as a newspaper columnist, you really shouldn't be writing an article in a way that assumes anyone reading it has already read all your other articles.

That is certainly extremely concerning. I had a bit of a read of the article but at 25 pages I just don't have sufficient motivation to plough through all of it. I couldn't work out exactly how well they matched their covid people with their "everyone else" people in terms of underlying conditions. As far as I understand it, they matched well for age. I guess my primary concern is about how much of that extra readmission and death was due to covid and how much was down to other factors not taken into account by the methodology of the study. I think on balance it seems reasonable to accept their findings and start worrying for any old relatives or friends who have had covid all over again.

If anyone wants to tell me in a post how good the study was at accounting for age and underlying conditions, I would be very happy

I'd like to see more studies done to back it up though and a published article is obviously better than a pre-print that hasn't seen peer review yet.

I don't think I follow you here. You are in the US? Surely if the official rate of 1.7% is low then adding in the excess deaths will increase the rate, not decrease it to 0.6%?

No, you have catastrophically misunderstood figure 7.

That is giving admissions rates per 100,000 people and there aren’t many old people compared to young people.

Whoops.  Embarrassing given the context in your OP.  

We cross-posted. Please re-read my reply.

We didn’t cross post.  I replied to your message.  Do you mean that you edited it whilst I was replying?  Your edited post retains this statement “Almost all hospital covid admissions are for old people”.  

This is simply not true.  About 40% of all Covid hospital admissions are currently for people under 65.  So it’s closer to “about half” than “almost all”.

Unless you can give a different reason why a lot of extra people are suddenly going to hospital for reasons other than their covid status, this has more than a whiff of grasping at straws to it.

I think the point is that if you're older and get covid, there's much more chance of you being admitted with severe covid. However I very much doubt that a lot of the younger people are catching it in hospital or incidentally swabbing positive.  After all, most younger people will rarely go near a hospital if they don't have underlying conditions.

Completely anecdotal - but of the 4 "severe" covid cases that I have seen/spoken to in the past few weeks, either before or after admission, three were late 30s-late 40s with no underlying conditions.

You are making it very hard for me to remain civil. If you can't debate without being insulting then I'll just stop replying and you can consider yourself to have won the debate.

I did also include people "and/or have underlying conditions", I can't seem to find any stats on the percentage of hospital admissions who do and do not have an underlying condition but there is a sun article out there (yeah, I know, great source!) saying that 95% of people who died from covid had an underlying condition so this does at least suggest that a very large majority of those being admitted with covid have underlying conditions.

There’s debate and then there’s being wrong.  As far as I can see I have been quite civil.  You seem to be going off on one however.

Your comment which I pulled you up on was a gross misrepresentation of the data.  You edited your post but did *not* fix your gross misrepresentation of the data.

So you’ve got no actual evidence to back up the argument you are making.

Pointing out that you catastrophically misunderstood that figure and that you haven’t any evidence to back up your wider point isn’t insulting.  

You can't really conflate deaths and admissions. The younger/healthier patients are more likely to make it out of hospital alive after being admitted, that is all. 

(as an aside, the "died with underlying conditions" thing is a load of rubbish as this could be things like asthma,  mental illness, mild high blood pressure or arthritis, hardly significantly life-limiting conditions in most cases )

Don't worry about wintertree.  He is remarkably emotional for someone who claims a scientific background, but apparently he has a bad tooth and I guess that might have a bearing on the matter.

Just press on and try to distinguish the wood from the (winter)trees.

I’m probably not getting enough Vitamin D.

Today I heard (can't remember which podcast) that for covid patients on ventilators it was even "worse" Roughly one third  over 75, one third 55 to 75 and one third under 55. 

This is partly because very few people over 85 are out on mechanical ventilation as it just won’t work for them, but I think there’s more to it than that given the absolute low number of people aged over 85.

Yep, it was "more or less" and actually only 25% over 75. I assume for the reason you describe. Another reason for being cautious about relaxing measures after mid Feb. 

Indeed.  

This thread isn’t the first example I’ve seen of someone getting the wrong end of the stick by conflating deaths/100k and actual deaths.  The graph pretty much reversed in age rank dependant on the choice of normalisation.  The ONS are exceptionally clear on what the graph shows as well.

The ongoing misconception that this is just an old persons disease, combined with ignorance of how widely defined and wide-spread “pre-existing conditions” are in the population are not a good foundation for a release strategy, but they’re very common.