Friday Night Covid Plotting #62

New Topic
This topic has been archived, and won't accept reply postings.
 wintertree 22 Jan 2022

Post 1 - Four Nations

Welcome to thread #62 - Rebound and Isolation Edition!

Cases are turning to rise in all 4 home nations.  I’ve only looked at demographics for England, but this looks like a rise in children with the return of schools, followed by a rise in downstream, household infections.  Assuming double or triple vaccination as the norm, hopefully this is mostly people getting their first dose of more-or-less endemic covid and not going to hospital. 

One of these data points in the last week is yours truly, currently self-isolating and champing at the bit to get out in to the wider world.  I'd hoped to hold out until self isolation rules were dropped, but was't trying very hard to avoid it any longer...

Link to previous thread:

https://www.ukhillwalking.com/forums/off_belay/friday_night_covid_plotting_61-...


OP wintertree 22 Jan 2022

Post 2 - England 1

Cases are rebounding and the next post looks at this a bit more.

Both hospitals measures are falling; and these plots don’t capture the context that around half of recent hospital admissions have been “incidental” - people who have Covid but who are admitted for another reason - this I think reflecting just how high prevalence of Covid has been recently.

Deaths look to have peaked and be preparing to fall; what we’d expect given hospital admissions.


OP wintertree 22 Jan 2022

Post 3 - England 2

Plot 18 shows that the rate constant for cases is turning towards a rise in all regions; a few lead the way.  I’ve chosen the South East, and done a demographic plot D1.c for it.  The growth of cases in under 15s is visible as an emergence of gold colour on the bottom right of the cases plot, and as a turn to orange on the rate constant plot.  There’s an expanding wedge of orange indicating cases turning to decay, starting with ages 35 to 45 and aged either side turning later.  This looks very much like household infections; ages where cases were in decay through other societal interactions but where the links to school cases are “forcing” growth

So, when omicron runs out of steam in school ages we’d expect that downstream growth in adults to stop, as it has before.   The big question now is “when does Omicron turn to decay in school ages”.  Given the significant under-detection of omicron - in part due to the reduction in severity of symptoms due to the sheer quantity of immunity now out there - my guess is sooner rather than later.

Once school cases go in to decay, we might expect those adult ages to as well, but we’re seeing a big tranche of control measures dropped along with a lot of messaging over returning to offices.  So, who knows.  We’ll find out.

I put in a plot P1.c.  This shows for a given data the relative distribution of cases across the ages; not the absolute number.  To my eye, I can see the centre of cases shift, coalescing in younger adults during school holidays and splitting to children and parental ages during term times.   I’ve added an annotated version with what I see scribbled on;  the lines jump about I think because the prevalence in other ages not related to schools shifts about.

Still no sign of university outbreaks in the top level data.

One wild card is the effect of the weather on R and the rate constant for cases.  The usual swing between nice and grotty periods has been suspended by this blocking high which looks set to last there month out.  Once it does break, we're going to see R rise across the board I expect; with rate constants as they are right now - except for schools and downstream infections - we'd expect to see decay continue.  Where we'll be by the time bad weather arrives is another matter however.

I don't recall a January like this, although Netweather says January 2006 was similar [1].  Now I've got to figure out why I don't remember January 2006.  I suspect the answer can be found in a wide range of taverns where I was hard at work on my studies, about half of which haven't made it through the pandemic.  

[1] https://www.netweather.tv/weather-forecasts/news/11260-bored-of-high-pressu...

Post edited at 18:53

OP wintertree 22 Jan 2022

Post 4 - Wrap Up

A very reduced set of posts this week; still winding things down for #64 to be the last weekly post, and I’m shattered - been tackling a lot of garden projects to relieve the tedium of self isolation.  

As someone else raised on another thread in The Pub (I hope they don’t mind me re-posting it)

“Exactly, nice that my experience has been one of administrative hassle, and a couple of nice days spent with the kids playing Castlevania on my N64 and painting LOTR minis rather than monitoring everyone's O2 levels and on the phone to 111. “ 

I’m not sure what a Castlevania is, but the change in the risk factors and consequences over the last 22 months has been staggering.  I was reflecting earlier to myself on the seismic shift to working from home, the vaccine developments and the repurposing trials.  It all shows to me what society can achieve when it has a clear purpose beyond self-perpetuation.

The got me lamenting how little purpose society seems to have these days beyond continuing on; and it’s not like we’re short of what should be motivation for a Purpose over decarbonisation and achieving energy independence.  

1
 Bobling 23 Jan 2022
In reply to wintertree:

Don't mind at all, rather flattered in fact.  You therefore must know what I meant about painting LOTR minis ;  )

Now COVID is in my house too I've rather taken my eye off the ball but I'm not suprised that cases are rising in your plots, about 1/3 of both my kids classes have been off this last week. To use an over used word during the pandemic cases in my community at the moment are at a completely *unprecedented* level.

Hope your self-isolation passes quickly and productively, but that you get some quiet time too, it's a rarity with young kids as I am sure you know.

 mountainbagger 23 Jan 2022
In reply to Bobling:

> To use an over used word during the pandemic cases in my community at the moment are at a completely *unprecedented* level.

Yes, feels like that around here too. I've never known so many people personally with Covid pretty much all at the same time, and it's mainly families with kids at school.

Thankfully they're all vaccinated and there are no hospitalisations amongst the people I know. There's still a few that have a harder time of it, and a few who take quite a few weeks to shift some of the symptoms.

I dread to think what it would be like if we were all immunonaive. Well, I don't need to think as I can remember from the first few months of the pandemic in 2020. I also knew quite a few people personally who caught it back then, both March to May 2020 and then around Jan 2021. The outcomes were much worse, including two people with an outcome of the worst kind. Nothing like that now.

Vaccines have accelerated our progress, there is no doubt in my mind, and I'm grateful to everyone who contributed. Natural infection is now possible (or less dangerous) for many more people than it was.

 Misha 23 Jan 2022
In reply to wintertree:

Thanks. Hope you’re finding things to do while in isolation.

The question in my mind is how long it will be until cases reduce to whatever will be the endemic baseline (which will no doubt be seasonal). As you say, the general reduction is being offset by growth in school ages and their parents. The return of office workers and, to a lesser extent, the end of mask mandates will also contribute to growth. However what we’ve previously seen is a gradual change in behaviours after various rule relaxations. I suspect we’ll see that again, which would help to moderate growth but would stretch out the peak. Things like half term and worse weather will also confuse things. My guess is cases will pogo around for at least another couple of months like they did in autumn but around a much higher baseline - perhaps in the 80-100k a day range. I’d expect a reduction from April though.

Still not clear if they’re going to scrap mass testing. I think that would be unwise for the time being. Hopefully the ONS survey and hospitalisation / ICU / death reporting will keep going to give people an idea of what’s going on. 

In terms of outcomes, the reduction in ICU occupancy is great to see. Deaths are higher than before but what we really need is the data on deaths ‘from’, which of course takes longer to feed through. Same for admissions. Apologies if this was discussed last week, I didn’t follow the thread due to work being a bit crazy.

I’m interested to see how the return to the office goes at our place. I know some people are keen to get back, others less so. I’m not in any rush while cases are so high. 

1
 Michael Hood 24 Jan 2022
In reply to Misha:

Unless the various statistics indicate the proportions of "with Covid" and "from Covid", then as Covid prevalence increases, they become less useful for our understanding of what's going on.

This applies to hospital admissions, occupancy and deaths. I'd realised this about the hospital stats but only recently twigged that it applies to the deaths as well. It's the "Covid but died in a RTA within 28 days" thing all over again but with some actual justification this time.

As for return to the office, once again my company are behaving like numpties and insisting that everyone comes back in.

Post edited at 06:42
1
 Michael Hood 24 Jan 2022
In reply to wintertree:

Why are you aiming for #64 being the last?

We're not being geeky about powers of 2 are we 😁

 mountainbagger 24 Jan 2022
In reply to Michael Hood:

> Why are you aiming for #64 being the last?

He's getting older, losing his hair, (not so) many years from now

OP wintertree 24 Jan 2022
In reply to Michael Hood:

Fits with the general trend of doubling times….  

 mountainbagger 24 Jan 2022
In reply to mountainbagger:

When I get older losing my flair
(Not so) Many years from now
Will you still be reading me at half past nine
Covid stats, denialists whine

If I'd been up till quarter to three
Would you open the thread
Will you still need me, will you still read me
When I'm sixty-four

Etc.

Edit: I wanted to do the whole song but got to go to work in a minute 🙂

Post edited at 08:56
 Ramblin dave 24 Jan 2022
In reply to Michael Hood:

The Government dashboard actually includes both "died with 28 days of a positive test" and "had COVID on the death certificate", although the latter presumably takes some time to collate because it seems to run at about a month's delay:
https://coronavirus.data.gov.uk/details/deaths

That means that we probably won't get much clarity on the "of vs with" question for another couple of weeks.

 Offwidth 24 Jan 2022
In reply to Ramblin dave:

I'd bring it all back to where is the main pressure on the NHS. Long Covid is way more important that the 'with' question. Let's hope long covid rates for omicron are much lower.

https://www.theguardian.com/society/2022/jan/24/long-covid-nearly-2m-days-l...

3
 Offwidth 24 Jan 2022
In reply to mountainbagger:

Can't resist.....

I could be handy, digging up clues
When your fight has gone
Read it and feel better by the fireside
Sunday mornings stemming the tide
Doing the data, digging for leads
Who could ask for more

Will you still need me, will you still read me
When I'm sixty-four

.....

OP wintertree 24 Jan 2022
In reply to Misha:

> The question in my mind is how long it will be until cases reduce to whatever will be the endemic baseline 

Indeed; I think the wild-type-vaccine-3rd-dose granted  immunity-against-infection-by-omicron is going fade pretty quickly so there's probably a lot of water yet to go under this bridge.   Likewise, the gradual behaviour shifts you describe. 

> My guess is cases will pogo around for at least another couple of months like they did in autumn but around a much higher baseline - perhaps in the 80-100k a day range. I’d expect a reduction from April though.

I wouldn't bet against your guess or how you got there; perhaps it'll almost disappear over the summer then raises its head around September.

> I’m interested to see how the return to the office goes at our place. I know some people are keen to get back, others less so. I’m not in any rush while cases are so high. 

The only reason I'd been holding out against infection was the sheer hassle of self isolation and the risk of both adults in the household having to isolate at once which would have been hassle-cubed.

OP wintertree 24 Jan 2022
In reply to Ramblin dave:

> That means that we probably won't get much clarity on the "of vs with" question for another couple of weeks.

There's a plot I run off every few months.  I take the demographic cases data for England and ask how many of those people would be expected to die within 28 days from all-cause-mortality.  (form pre-Covid ONS data).  If Covid had not association with lethality, then deaths under the 28-day measure would match the all-cause-mortality model.

If people close to death are more likely to catch Covid, and/or if Covid is more likely to kill those people, deaths under the 28-days measure are higher than the all cause mortality model.

The left plot shows the deaths (black) and all cause mortality model applied to cases (red), and the right plot shows the ratio between them.

Back early on in October 2020, a positive test meant someone was about 50x more likely to die than their all cause mortality; vaccination crashed this down to ~10x by summer last year; now under breakthrough Omicron infections it looks to be down to ~5x.

The mortality model is over-simplistic and in particular it doesn't include seasonality.  

If the plot keeps moving in the way it looks like, we'll soon have about 4x the all cause mortality rate associated with a Covid positive; at that point about 20% of people would be dying "with Covid" and not from "Covid" - a fraction that could be higher if people close to death are more likely to catch Covid.  At that point I bow out of the analysis because it needs proper expertise in the relevant fields to do this properly.  

But things are still moving in the right direction, and there's not far left for them to move.


OP wintertree 24 Jan 2022
In reply to Offwidth:

> I'd bring it all back to where is the main pressure on the NHS. Long Covid is way more important that the 'with' question. Let's hope long covid rates for omicron are much lower.

2 million days sounds like a lot, doesn't it.  Actually 1.8 million over 18 months, so call it 1.2 million days lost a year.

The NHS employs about 0.6 million FTEs across the areas that number is applied too if I've got the right measures.  

So, another way of looking at that number is around 2 days of absence per staff member per year.

Not such a big, scary number in context is it?  Yes, it's another added pressure to healthcare.  There's a lot of them, and it's worrying how many are getting worse even as Covid gets better.

> Let's hope long covid rates for omicron are much lower.

Long Covid largely appears to be due to immune dysregulation in infections; this is a hallmark of the infection process in people with no prior immunity; so as immunity continues to increase it seems likely that the rates will continue to decrease.  

Chronic fatigue style illness following viral infection is not unique or novel to Covid, and it's unlikely to go away any more than it does for other infections.  The immediacy and scale of the problem is giving research in the area a big push, and I hope that as things are solved for long Covid sufferers, they are solved for many others too.

1
 Offwidth 24 Jan 2022
In reply to wintertree:

An average of 2 days extra staff absence a year is a loss of almost 1% of the workforce. That's at a critical time on top of all the other staff pressures (vacancies, departures of experienced staff, covid positive absences and covid isolation and general exhaustion and stress related illness). You know full well how such numbers in a stressed system can do disproportionate damage. By definition long covid statistically won't have impacted much of the first wave (but staff with serious lung damage caused by catching covid - especially early on when working with inadequate PPE - won't have been at work then)

On the prior immunity point, there doesn't seem to be any significant decline in the latest ONS newer delta-related data, despite reported antibody levels in the ninety percentile range. I'd love to see better data overall on this, and it must exist in some form as all staff absence gets categorised. We just don't have any data for omicron long covid yet.

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

2
OP wintertree 24 Jan 2022
In reply to Offwidth:

> . You know full well how such numbers in a stressed system can do disproportionate damag

I do.  Did you read my post?  I said:

Yes, it's another added pressure to healthcare.  There's a lot of them, and it's worrying how many are getting worse even as Covid gets better.

I'm not clear what you think should happen at this point.  Do you think we should stop Omicron from spreading indefinitely through NPIs and start recharge the pandemic potential?  If not, at this point, what will be will be, and there are going to be staff absences in all sectors related to people catching omicron.

To me the news is not a Big Scary Number of days of staff absence as the guardian give, bereft of all context.  Presented like that, it's easy for readers to assume this can all be blamed on Covid.

Flip the context - 2 additional days of staff absence a year is an unsustainable level of stress. Presented like that, it's clear the issues are systemic with politics not Covid behind them, and that the ways out of solving this start with the politics of healthcare provision and not of Covid control.  

 Offwidth 24 Jan 2022
In reply to wintertree:

I said I know you know!!?? I was making a supplementary post, not disagreeing with you. I also agree the long covid absence rates need to be treated in proportion to their effect.. not ignored nor sensationalised.

I've been clear for a while now I see the key data factors are now down to politics and in that especially our government failure to fund the NHS, Care and PHE systems properly, to workforce plan properly, and to stop shooting those sectors in the foot making systems under terrible pressure struggle even more. Covid is just another pressure from now on (one that with cheap and easy restrictions like masks in indoor public space and better ventilation would be reduced a bit...but hey when Boris needs red meat they don't give a damn).

4
 Si dH 25 Jan 2022
In reply to wintertree:

Provisional data suggests cases are falling again already and only ever flattened rather than rose (at the top level.) 

Because case data is now mostly LFTs, the delays between reported and specimen dates are much shorter. The data for Monday just gone (penultimate day in the graph below) is already very mature. (Edit, I've added the second graph, of daily change in reported cases by specimen date, to show why I'm confident in this.)

The behaviour is basically that last week, LFTs rose and offset the drop in PCRs, causing a flattening (and concerns about a turn to rise) but this week it looks like LFTs are matching last week while PCRs continue to fall off.

All good!

Post edited at 17:21

OP wintertree 25 Jan 2022
In reply to Si dH:

Good stuff.

The regional rate constants plot show that the cases rate constant is becoming more negative in all regions - turning to decay where it was in growth, and accelerating decay elsewhere. So stagnation at the top level should rapidly turn to decay as the balance between growing and decaying regions passes...

Updated demographics for one growth region below -  the South East.  Data is pretty noisy broken down by age and region but it looks to me like growth in all the school ages has peaked (except maybe 15-19), and we'd expect growth to turn to decay in parental ages after that.  What's really interesting is that there's a third stream of growth appearing in the demographics - child > parent > grandparent?  The spacing and timing looks right.  I don't recall seeing causality so clearly beyond child > parent in the data before.  Perhaps that says something about the transmission characteristics of omicron, or perhaps it's just that the decay elsewhere is so strong it gives more clarity to the signal.

Regionally, hospital admissions started to move to a plateau (rate constants rising towards 0) but have resumed decay; perhaps related to the demographic shifts throughout the main peak; perhaps related to incidental admissions as much as anything. 

On the subject of incidental counting, as Ramblin dave said up-thread, the difference between the 28-days and death certificates measures should bring us come clarity on incidental deaths; it looks to me like the certificates measure has moved from being slightly over the 28-days measure to now being under it; a move in the right direction.  As Ramlbin dave also said, it'll take another couple of weeks to get proper clarity; especially if there's much reporting lag in this measure.  


 Si dH 25 Jan 2022
In reply to wintertree:

Multiple infection episodes to be reported from 31 Jan. That means if you test positive twice within 90 days that's still one case, but if more than 90 days apart, it's two cases.

https://coronavirus.data.gov.uk/details/whats-new/record/af008739-ffa3-47b8...

In reply to Si dH:

Better late than never, but aren't we planning to draw a willy on it from march 24th? So we can expect just under 8 weeks of decent reinfection data :-/

OP wintertree 25 Jan 2022
In reply to Si dH:

Whammy.  That’s going to shake things up a bit.  

I probably shouldn’t read too much in to their choice of a 90-day cutoff….

 Michael Hood 25 Jan 2022
In reply to Si dH:

So 90 days for re-infection but anything beyond 28 days and it's nothing to do with Covid - hmm.

 Si dH 25 Jan 2022
In reply to wintertree:

It has always asked, when you register for a PCR test, whether you have tested positive in the previous 90 days. At least it has for a long time. I think they obtain this data because there is supposedly a risk of false positives due to bits of RNA(?) hanging around after you have cleared the infection.

[We've had a bit of discussion about this at work recently due to the high prevalence - we previously had a requirement to undertake a PCR before visiting a site (as well as daily LFTs while there) but we've quite a few people with covid in recent weeks who are due to do a visit soon and don't want a false positive, so they are changing the policy.]

Post edited at 20:27
OP wintertree 25 Jan 2022
In reply to Michael Hood:

> So 90 days for re-infection but anything beyond 28 days and it's nothing to do with Covid - hmm.

There is a 60-day measure in the dataset as well - it's even advertised in the API documentation since the recent overhaul, and there's the death certificate measures.  I had a look at these a couple of threads back [1].

Short answer - the 28 day measure tracks the death certificates measure closely, the 60 day measure doesn't.  It's also 28 days from detection not infection which might make a surprisingly crucial difference. 

[1] https://www.ukhillwalking.com/forums/off_belay/covid_ctd_since_the_plotting_th...

In reply to Si dH:

> [We've had a bit of discussion about this at work recently due to the high prevalence - we previously had a requirement to undertake a PCR before visiting a site (as well as daily LFTs while there) but we've quite a few people with covid in recent weeks who are due to do a visit soon and don't want a false positive, so they are changing the policy.]

It's going to be interesting to see how the wind down of workplace control measures goes; with isolation requirements set to end soon enough but whilst it remains a notifiable disease it seems things are a bit confused...

Post edited at 20:42
 Michael Hood 25 Jan 2022
In reply to wintertree:

Knew about the 60 day but didn't know that it was "worse" at tracking the death certificates than the 28 day.

Regardless, the point was really that if you die 29 days after a positive test then you're not considered to have died from Covid (even if you have) by the most published statistic, yet if you test positive after 89 days you're still considered to be suffering from the "initial" Covid infection.

It reinforces the lies, damn lies and statistics feeling; comparing apples to pears, etc.

Using different time ranges may be perfectly valid, it just doesn't feel that way.

Post edited at 22:17
1
OP wintertree 25 Jan 2022
In reply to Michael Hood:

> Using different time ranges may be perfectly valid, it just doesn't feel that way.

But they're measuring different things.  The 28 days measure is strongly validated by the death certificates data.   Too long or too short a measure has major issues.  The 28 days measure may soon fail due to incidental infection; at that point only medical judgement is left.  The problem with the embodiment of that judgement - death certificates - is that there's too much latency for them to have been useful in informing policy in the past; that urgency for rapid data is fading away as the situation improves however.

The cutoff for a positive test is quite different; it's harder to understand how to validate that without genotyping every single detection to identify different viral infections; something utterly unthinkable at present - and if it was possible it would delay publication of a critical measure used for real time policy formation.    Again this urgency is fading away as we leave the pandemic phase.

The dashboard measures are designed to be low latency during a pandemic for policy purposes; I have no doubt they are entirely fit for purpose there.  But arbitrary seeming compromises have to be made to bound the level of detail to the point the measures can be made real time.  As we exit the pandemic phase, that urgency fades, and the complexity behind the measures rises (incidental hospitalisations and deaths, re-infections vs re-detections).  

The answer I think is...

  1. Shut down the dashboard systems and P1/P2 sampling at some point not so far from now - it is a system perfectly fit for a pandemic, and one that is difficult to shoe horn in to an endemic world.
  2. To use some of the testing resource freed up to raise the cadence of the ONS random sampling survey to twice weekly, with more demographic and regional statistical power
  3. To determine a criteria with which to prioritise ONS positives and hospital admissions positives for sequencing as part of a sentinel system for potentially worrying variants
  4. To work on reducing the latency of the death certificates reporting.

> It reinforces the lies, damn lies and statistics feeling; comparing apples to pears, etc.

To an extent that's because the inputs are apples and pears.

Post edited at 22:31
In reply to Michael Hood:

> Unless the various statistics indicate the proportions of "with Covid" and "from Covid", then as Covid prevalence increases, they become less useful for our understanding of what's going on.

I don't think it is as clear cut as 'of' vs 'with'.   No matter what underlying condition you've got getting Covid as well is very likely to make it worse.  Maybe the high numbers of people going to hospital 'with Covid' is a sign that other conditions which might not have been serious enough to put them in hospital that particular week become serious enough to put them in hospital when they get Covid as well.  

The idea that 'with Covid' means the Covid is just coincidental and should be ignored is probably simplistic.  

2
In reply to wintertree:

> 2 million days sounds like a lot, doesn't it.  Actually 1.8 million over 18 months, so call it 1.2 million days lost a year.

You can't just scale it like that because the Covid infections are not evenly distributed over the 18 month period and the long covid condition can last for a year.  So you will be accumulating more people with long covid over time and very likely the number of people affected by long covid at the end of the 18 month period will be much higher than the number at the beginning. 

2
OP wintertree 26 Jan 2022
In reply to Si dH:

React data suggesting reinfections are quite concentrated (or, more correctly as the future hasn’t happened yet, I suspect front loaded) in certain sub populations.

https://www.bbc.co.uk/news/health-60132096

Be really interesting to look at the rate constant before and after the changes and see how much it is or isn’t affected by this reporting change.  Obviously the cumulative cases curve is going to diverge a lot…

 Si dH 26 Jan 2022
In reply to wintertree:

> Be really interesting to look at the rate constant before and after the changes and see how much it is or isn’t affected by this reporting change.  Obviously the cumulative cases curve is going to diverge a lot…

They are going to retrospectively change all the specimen-date data to include reinfections so if you want to compare before/after the change, make sure you have the old data saved somewhere in the format you want it. I guess it wouldn't do to interpolate off your old jpeg graphs on these threads

OP wintertree 26 Jan 2022
In reply to Si dH:

>  I guess it wouldn't do to interpolate off your old jpeg graphs on these threads 

I could (I think) live with myself if I had to graphically blend old and new graphs together with identical axes etc as a step short of backing the data out…. But yes; good point - a lot of the plots use the API not downloads, so I need to get suitable downloads in place first!

 Toerag 26 Jan 2022
In reply to wintertree:

 

> Updated demographics for one growth region below -  the South East.  Data is pretty noisy broken down by age and region but it looks to me like growth in all the school ages has peaked (except maybe 15-19), and we'd expect growth to turn to decay in parental ages after that.

That's interesting as our school age cases are ramping back up here - 66 <9s on the 21st, 110 yesterday. This would signify that UK SE is running out of kids to infect, I know we have plenty left to infect here.

OP wintertree 26 Jan 2022
In reply to tom_in_edinburgh:

> I don't think it is as clear cut as 'of' vs 'with'.   

I don't think anyone has said it's that "clear cut", except for this post I'm quoting?

> No matter what underlying condition you've got getting Covid as well is very likely to make it worse.

That's simply not true.  For many younger people, they are now so highly immune that it's a mild cold for them - one many people aren't ever aware that they have.  The main challenge these people present as incidental admissions are the additional resource consumed through infection control around Covid.  There are plenty of other patients for whom Covid will complicate pre-existing conditions, but it's by no means a universal.

> Maybe the high numbers of people going to hospital 'with Covid' is a sign that other conditions which might not have been serious enough to put them in hospital that particular week become serious enough to put them in hospital when they get Covid as well.  

NHS England has quarterly hospital admissions - pre-Covid - of about 1.5 million people. [1]

That works out at about 16,000 admissions per day.  With the ONS having Covid prevalence top out at about 6%, we'd expect 960 daily admissions with purely incidental Covid if there was no link between Covid and tendency to be admitted.  At that time English admissions peaked at around 2,000 a day.

So, I don't think your "maybe" is well justified by any means.  For sure, it will be tipping over some people from sick by non-admission to admission, and it will be making life worse for some of those admitted, if Covid had no health effects at all and was randomly distributed across the population, we'd expect a similar number of incidental admissions to what is being reported in the breakdowns.

A second take is to use demographic all-cause mortality measures to estimate the number of deaths under the "within 28 days" measure.  This is the plot below; it's heavily caveated [2], but it's at the point where deaths under this measure are about 3.5x those we'd expect if Covid had no association with lethality (did't cause it, didn't concentrate in those anyway close to it).  This suggests that about 30% of the most recent deaths under this measure are incidental - we wait to see how that does or does not develop in the death certificates.

Yes, some "incidental" Covid makes things worse, but not all of it.  It would be a nonsense to expect its severity to remain the same in terms of outcomes as it stops being serious enough to be the primary reasons for admissions.

Things are moving on.

> The idea that 'with Covid' means the Covid is just coincidental and should be ignored is probably simplistic.  

Again, I think you are the only person to say they should be "ignored" - but it's also clear that the measures are becoming less and less appropriate, and that the situation is much better than one would infer by simply comparing them to previous waves.

[1] https://www.statista.com/statistics/504012/number-of-nhs-hospital-admission...

[2] https://www.ukhillwalking.com/forums/off_belay/friday_night_covid_plotting_62-...


1
 Offwidth 26 Jan 2022
In reply to wintertree:

I agree with your broad analysis but you're not taking into account the different demographics of those hospitalised (mainly old), compared to those with covid (mainly young). Anecdotally a good proportion of the 'with' group catch it in hospital and either way they do tend to get worse outcomes (more ill, more likely to die and longer stays) than those with similar conditions 'without' covid. When consultants say don't understate the potential seriousness of the 'with' covid group that's for good reasons. Infection control is also a real pain but I can't see that disappearing soon because of the immunity degraded vulnerable and the unvaccinated (hospitals can't just let people get exposed to covid if it's likely to be very serious for them). Improved ventilation should be a high priority.

1
 Offwidth 26 Jan 2022
In reply to wintertree:

React saying 2/3rds of omicron infections are re-infections.

https://spiral.imperial.ac.uk/handle/10044/1/93887

In reply to wintertree:

> That's simply not true.  For many younger people, they are now so highly immune that it's a mild cold for them - one many people aren't ever aware that they have.  

It's not my opinion.   I heard it in an interview of a hospital doctor who was being asked about admissions 'with' rather than 'for' Covid and he said something along the lines of 'I've not seen any condition Covid doesn't make worse'.

It seems eminently possible to me that if you've got a heart problem or cancer and you also catch a dose of Covid that might be the week your underlying problem gets bad enough to put you in hospital.  But I am not a medical doctor.

1
OP wintertree 26 Jan 2022
In reply to Offwidth:

> I agree with your broad analysis but you're not taking into account the different demographics of those hospitalised (mainly old), compared to those with covid (mainly young). 

I did recognise this with the caveats: "if Covid had no health effects at all and was randomly distributed across the population, we'd expect a similar number of incidental admissions to what is being reported in the breakdowns.".  In previous times I've discussed this I've detailed the anti-correlation between the age distributions of Covid and those going to hospital, but if I ended up repeating every caveat and qualifier every time I mention anything, I'd be hitting post size limits...

As it happens, the population normalised distribution of Covid cases is a lot less demographically sensitive than it used to be - if you look at the heat map [1] and read off the rates from individual age buckets (hover the mouse and get a tool tip), there's only a 2:1 bias between young adults and old adults now.

> When consultants say don't understate the potential seriousness of the 'with' covid group that's for good reasons.

Nor am I claiming otherwise, but it's also clear that 1000 hospital admissions reported through the dashboard now is highly incomparable to 1000 hospital admissions admissions reported through the dashboard a year ago.  Yes, Covid is making the problem worse for some of the people with it, but if someone has broken a bone playing rugby whilst not even knowing they had Covid, they're in for a broken bone.  

If we tested for all other circulating viruses on all admissions, there'd be a lot of those as well, and no doubt it would have potential seriousness for the more vulnerable people in those "with" groups.   

It also seems likely that we're reaching the point of significant incidental deaths; it seems to me very hard to argue that Covid is affecting the outcome there. 

> Infection control is also a real pain

Yes, I've said that repeatedly.  I totally agree on ventilation.  

> React saying 2/3rds of omicron infections are re-infections.

Yup, there's a whole discussion with Si dH on that up-thread and a link to another article discussing the React outputs.

In reply to tom_in_edinburgh:

> It seems eminently possible to me that if you've got a heart problem or cancer and you also catch a dose of Covid that might be the week your underlying problem gets bad enough to put you in hospital.  But I am not a medical doctor.

Sure, and the same could happen if they caught some other endemic disease.  The same does happen.  A lot.  But we don't measure for it or report for it in the way we do with Covid.

There does remain a cohort who are never going to reach endemic immunity levels to Covid because it's come too late in their life for their immune system to be able to produce a broad and enduring response, be-it through immune suppression, disease or sheer age.  For that group, incidental Covid is going to remain a worse problem going forwards than other endemic virus. I expect these people are over-represented in hospital admissions.  For many other people, the times are changing. 

[1] https://coronavirus.data.gov.uk/details/cases?areaType=nation&areaName=...

 neilh 26 Jan 2022
In reply to wintertree:

There was an article in the economist this week highlighting the excellent work that the ONS had done on Covid.

https://www.economist.com/britain/2022/01/22/britains-office-for-national-s...

 Offwidth 27 Jan 2022
In reply to neilh:

Just imagine if this one famous musician's decision to leave Spotify,  over Rogan's platform for covid misinformation, became hundreds of famous musicians. In any sensible view on freedom of speech, Public Heath impacts of a pandemic should always take priority over freedoms to distribute covid misinformation to millions.

https://www.theguardian.com/technology/2022/jan/26/spotify-neil-young-joe-r...

4
 neilh 27 Jan 2022
In reply to Offwidth:

I am sure the likes of Van Morrison and the Gallagher brothers will have a different view................

 Offwidth 27 Jan 2022
In reply to neilh:

Who cares?...if even a small fraction of other big names threaten to follow Young the company will have to force Rogan to cut out interviews with covid deniers and anti-vaxers on their platform.

2
 neilh 27 Jan 2022
In reply to Offwidth:

I see no rush to follow...do you?

OP wintertree 27 Jan 2022
In reply to thread:

Annotated version of plot 22.  We're seeing the Lissajous figure of the two kinds of hospital occupancy turn to decay.

Normally the lag between admission and ITU admission (on the fraction of people ending up in ITU) means that the curve turns counter-clockwise as it heads for decay, but this time we're turning clockwise - meaning that decay in ITU occupancy came before the decay in general occupancy.

What's changed?  This is the first time we've had a wave ended by immunity not control measures, and not unrelated to that, it's the first time we've had rising incidental admissions.  This wave also happened pretty rapidly with a lot of less-serious hospitalisations in the highly immune landing just as what I think may have been the tail end of zero-immunity delta era hospitalisations where making up a significant chunk of ITU occupancy.   I suspect the explanation lies somewhere in those details.

Post edited at 11:53

 TomD89 27 Jan 2022
In reply to Offwidth:

There is so little actual substance to that article (many assertions, zero evidence) that to me it's clearly a smear piece designed to influence those that are only peripherally aware of Rogan and poorly informed on the subject matter.

It does say "The letter highlighted a highly controversial episode from last December that features Robert Malone, a virologist who was involved in the mRNA vaccine technology that led to some of the leading Covid-19 vaccines but has since been criticised for spreading vaccine misinformation."

Huh so someone who is deeply knowledgeable about mRNA vaccine technology shouldn't be allowed a public opinion because Neil Young thinks it's misinformation. Guess I'll listen to Neil Young>virologist, no question.

Good journalism would include who was criticising, what exactly they were critcising, and what the vaccine misinformation in question was. This article is devoid of such critical info. 

> the company will have to force Rogan to cut out interviews with covid deniers and anti-vaxers on their platform.

Could you name anyone who has outright denied covid and/or is universally anti-vaccine that's been on the Rogan podcast please? I've bothered watching, you'll struggle.

4
OP wintertree 27 Jan 2022
In reply to TomD89:

I would just point out that being a highly capable scientist is no iron clad guarantee against becoming a raving lunatic or blithering idiot some decades later.  Latterly the noble prize winner James Watson of DNA fame has been going round pushing the view that science says darker skinned people are hornier and less intelligent than pale skinned people.  William Shockley moved on from co-inventing the transistor to advocating a new welfare system that would pay darker skinned Americans to be sterilised as part of his eugenic vision. AIDS denialism and shocking levels of sexism also abound in a minority of other prize winners.

There won’t be many experienced scientists who haven’t worked with a colleague with some serious reality disconnects or perhaps a tendency to start a monologue about butt plugs over the conference dinner.  It turns out scientists are people too, just like everyone else, and just like with everyone else a fringe and minority opinion needs robust testing against the evidence, and in that arena Robert Malone falls way short.  

Post edited at 12:45
 Offwidth 27 Jan 2022
In reply to TomD89:

I'd assume something taking pot shots at journalism would do some research first and place trust in the large number of doctors who outed Malone's misinformation. Malone has been saying increasingly problematic things since the start of the pandemic:

https://en.wikipedia.org/wiki/Robert_W._Malone

https://www.politifact.com/article/2022/jan/06/who-robert-malone-joe-rogans...

1
 Toerag 27 Jan 2022
In reply to wintertree:

>  I suspect the explanation lies somewhere in those details.

Interesting comparisons between waves - Original variant at peak had ~15% of admissions in ITU, Alpha 9%, Delta 10% and Omicron had 3%.  Shows how Delta was deadlier than Alpha as even though the majority of the vulnerable had had some sort of vaccination the percentage going to ITU was higher. Omicron's definitely having a less severe effect on the community, it'll be interesting to know how much of that is due to it's absolute 'weakness' and how much is due to vaccination.

 elsewhere 27 Jan 2022
In reply to Toerag:

>Omicron's definitely having a less severe effect on the community, it'll be interesting to know how much of that is due to it's absolute 'weakness' and how much is due to vaccination.

Mostly due to vaccination, otherwise the unvaccinated x% would only make up x% of hospital admissions (& worse) rather than the unvaccinated being disproportionally represented in admissions etc.

Although to really know you'd have match unvaccinated demographic (known?) & behaviour (unknown?) with the same amongst those vaccinated.

Post edited at 15:13
 TomD89 27 Jan 2022
In reply to wintertree:

Yes I get all that and largely agree, but this guy is a virologist/immunologist and his opinion holds more weight than Neil Young's, especially as he had a focus/involvement on mRNA technology. Certainly that doesn't mean he can't be wrong, but lends at least some credibility. It would at least be worth hearing first hand what he is saying and assess wouldn't it? You have to admit that Guardian article is a bit on the light side with any evidence or direct, in context, quotes?

I'm fine with dissecting and critiquing his views, assertions, opinions and welcome conversation on that, but throwing unsubstantiated labels at people is wrong. I wouldn't pipe up if he'd been accurately called an "mRNA vaccine skeptic", but trying to throw 'anti-vax, covid denier' at everyone is disingenuous, and trying to get them banned from platforms is insidious, especially if leveraging the mass opinion of popular musicians or celebrities to achieve this aim, or off the back of news articles with sparse information or clear political bias. 

In reply to Offwidth:

If we're throwing toilet paper at each other:

https://www.dailymail.co.uk/news/article-10418589/Most-270-signed-anti-Joe-...

Using wikipedia and politifact as evidence isn't a good sign. The first thing it says on politifact is 'he was banned from twitter for misinformation' as if that actually means anything in the real world. You must be able to see the tactics being employed by these websites?

I'll just repeat the challenge to your claim above:

> the company will have to force Rogan to cut out interviews with covid deniers and anti-vaxers on their platform.

Could you name anyone who has outright denied covid and/or is universally anti-vaccine that's been on the Rogan podcast please?

6
 Offwidth 27 Jan 2022
In reply to TomD89:

The 'doctors' on the recent letter are just a subset of those medics and academics who have debunked various claims made by Malone during the pandemic in the normal peer review format that is applied to all scientific output. He then chose to use right wing journalistic routes to spread what was then proven misinformation.  Wikipedia has evidence of this in the links and politicfact is just one example of reported fact checking of his multiple false claims (again linked) that have been peer reviewed during the pandemic.

As for Rogan he has expressed vaccine scepticism and told under 21s not to get vaccinated.  For outright anti-vax views on his show he has hosted Alex Jones.

https://www.bbc.co.uk/news/world-us-canada-56948665

Here is another example of a Rogan guest who spouted nonsense on covid and vaccines:

https://healthfeedback.org/claimreview/joe-rogan-interview-with-peter-mccul...

Post edited at 18:04
2
 elsewhere 27 Jan 2022
In reply to wintertree:

Lissajous - an interesting point.

> What's changed? 

In news today, COVID positive occupancy with COVID now exceeds 50% of total COVID positive patients.

​​​​​​That may be part of it. Or maybe not!

Post edited at 18:30
 oureed 27 Jan 2022
In reply to Offwidth:

> As for Rogan he has[...] told under 21s not to get vaccinated.  

While we're on the subject of misinformation, Rogan's 'controversial' comment was actually:  

"If you're a healthy person, and you're exercising all the time, and you're young, and you're eating well...like, I don't think you need to worry about this."

When challenged about his comment he said: "I believe [vaccines are] safe and encourage many people to take them."

2
OP wintertree 27 Jan 2022
In reply to thread:

The regions that took to have chains of infection driven by growth in school cases all look to be hading for decay in cases going off the rate constant plots.  We can also see how that growth in cases translates to a bit of growth in hospitalisation in those regions. 

So, hopefully that's it and we're back to decay for now; bit of an open question about what'll inevitably change that - the ongoing waning of immunity-against-infection-by-omicron induced by the 3rd doses, or the next spell of bad weather temporality raising R?  Although I'm starting to think the idea of water falling from the sky is a mass delusion, this is quite the exceptional January.  


 neilh 27 Jan 2022
In reply to oureed:

I know very few Americans who eat really well……. There are the odd exception… and let’s be honest they will not be listening to Rogan . Not his target audience. 
 

Only an observation.  

 Misha 27 Jan 2022
In reply to wintertree:

The lissajous is a great illustration of what’s going on. I wonder which of your two reasons is the greater factor. Either way, all I can say is we’ve been pretty ‘lucky’ with Omicron. Let’s hope Pi won’t be a lot worse.

 Misha 27 Jan 2022
In reply to oureed:

> "If you're a healthy person, and you're exercising all the time, and you're young, and you're eating well and you’ve had all the recommended vaccines... like, I don't think you need to worry too much about this."

FTFY.

2
 Misha 27 Jan 2022
In reply to wintertree:

The idea of a return to decay is nice and you can see it in the headline numbers but I suspect we’re into a period of pogoing within a gradually reducing range. Lots of factors at play. Here’s another one - socialising in pubs etc is usually much reduced in January.

In reply to TomD89:

> Using wikipedia and politifact as evidence isn't a good sign. 

You did just link to the Daily Heil...

 Toerag 28 Jan 2022
In reply to Misha:

>  Lots of factors at play. Here’s another one - socialising in pubs etc is usually much reduced in January.

Our local stats show the demographic effects really well. In early December cases were heavily driven by schoolkids and their infected parents, then in late December it was the 20-30yr olds from xmas partying and the school age rates dropped off a cliff due to school hols, now it's schoolkids again. Today's graph attached, the school age cases have nearly doubled since the 21st, and in Jersey they've more than tripled in 3 weeks. I'm expecting the parent infections to be lower than early december due to the effects of boosters. The CI have had less infections than the UK so our rates will be higher due to lots more infectable people, but the behaviours and demographics will be the same. Virtually no-one goes out on the razzle in January until payday weekend at the earliest.   You can also see the effects of secondary school mask use and possibly vaccination in the graph too - <9s are ~4x more likely to become infected than secondary school kids.

Post edited at 00:38

 Si dH 28 Jan 2022
In reply to thread:

Mainwood put some interesting demographic case/infection ascertainment data on Twitter.

(Leading edge a data blip I think.)

Post edited at 13:09

OP wintertree 28 Jan 2022
In reply to Si dH:

Well, that's very interesting once I finished thinking about it.

So, we can likely multiply the school aged cases by a factor of several when comparing to older ages' perhaps more as symptoms and so detections drop with breakthrough omicron infections.

That re-writes the demographic cases/100k plots on the dashboard quite dramatically; no wonder growth is running out of steady in the young...

In reply to Si dH:

> Mainwood put some interesting demographic case/infection ascertainment data on Twitter.

I think that graph is trying to tell us that there is a lot of under-reporting, especially in the young.

The multiple brown and black lines confused me until I figured they must each represent a 5-year range within the stated age range...

 Si dH 28 Jan 2022
In reply to wintertree:

I would have guessed ascertainment would be very low in young kids although the difference is dramatic.

I was surprised how high it is in some working age adults and how much lower in over 80s. I'm not sure why that is.

In reply to Captain Paranoia: yes, me too.

OP wintertree 28 Jan 2022
In reply to Si dH:

> I was surprised how high it is in some working age adults 

I'm not sure it is "high" - his use of percentage as the y-axis unit is IMO a bit unintentionally misleading; he's going (10 day rolling sum of dashboard cases) / ONS prevalence.  

The ratio of dashboard cases to ONS prevalence is one thing, but one is not a percentage of the other, as they measure different things.  A day day rolling sum is - I very much doubt - a calibrated way of bridging that gap. 

Given that we know re-infections are missing from the dashboard, I suspect the % values are way too high.  Really the y-axis scale should be purely relative and dimensionless - the plots would look the same, but interpretation might differ slightly. 

> and how much lower in over 80s. I'm not sure why that is.

That is curious, isn't it.  There are noticeable differences in how over 60s and adults under 60 track the weather wobbles in the rate constants as well.  All sorts of really involved stuff going on I don't think the models have captured well.

> I would have guessed ascertainment would be very low in young kids although the difference is dramatic

Perhaps related to what fraction of toddlers have nostrils larger than the LFT bog brush...

Post edited at 14:54
 Toerag 28 Jan 2022
In reply to Si dH:

> and how much lower in over 80s. I'm not sure why that is.

1) they're triple or quadruple jabbed.

2) The rest of the population will hang out with each other but not go see granny if they have a sniffle, so the 80+ are being shielded to an extent.

2) I think that given the recent revelations how 2/3rds of current infections are reinfections, I think some people are simply more susceptible than others to being infected.  Those 80+ who have been infected have been very likely to die of covid, or die naturally, thus leaving a population of 80yr olds who are naturally unlikely to be infected.

Post edited at 16:13
 Si dH 28 Jan 2022
In reply to wintertree:

> Perhaps related to what fraction of toddlers have nostrils larger than the LFT bog brush...

The first time my son had a PCR test in summer 2020, it was a total nightmare and we only managed to put it about half a cm up one nostril for a couple of seconds. He was also full of snot at the time and unable to blow his nose to get rid of it. Came back positive.

We've done 4-5 tests on him since then, all but the most recent (last autumn) were a horrific ordeal and I only did them when symptoms have absolutely required it beyond any doubt. On occasions in winter 2020 during lockdown I seriously considered us all taking the 14 day isolation hit in preference to him doing the test. I suspect many others in our situation would just not do it at all. Anecdotally the test staff said most young kids behaved similarly.

Post edited at 16:18
1
 Si dH 28 Jan 2022
In reply to Toerag:

> 1) they're triple or quadruple jabbed.

> 2) The rest of the population will hang out with each other but not go see granny if they have a sniffle, so the 80+ are being shielded to an extent.

> 2) I think that given the recent revelations how 2/3rds of current infections are reinfections, I think some people are simply more susceptible than others to being infected.  Those 80+ who have been infected have been very likely to die of covid, or die naturally, thus leaving a population of 80yr olds who are naturally unlikely to be infected.

I think perhaps you've misunderstood. What's it's showing is that proportionally fewer infections in that age group are detected as cases - not that case or infection rates are lower, necessarily although I agree they generally are for some of the reasons you give.

 Toerag 28 Jan 2022
In reply to Si dH:

> I think perhaps you've misunderstood. What's it's showing is that proportionally fewer infections in that age group are detected as cases

Ah yes you're right, sorry. The current drop for most demographics resembles that seen in the summer and I'd call it a 'covid is over, I must have a cold / I don't want to be off work / I want to go on holiday' mentality.  I'm not sure the leading edge is a data blip, cases are now rising here and in Jersey too in the past few days. Can't remember if I said upthread, but Jersey's school cases have tripled since the start of term and ours have virtually doubled in a week.

Post edited at 16:27
 Offwidth 29 Jan 2022
1

New Topic
This topic has been archived, and won't accept reply postings.
Loading Notifications...