UKH

Post 1 - Four Nations Plots

To a quick look, everywhere is still in a holding pattern with high daily cases.

Looking at the leading edge of the plots, England is rising quite rapidly, NI and Scotland are just turning to rise and Wales is decaying a bit.  Some of that turn to rise I think is just noise in the data, we'll see next week; the provisional measurements for the leading edge of plot 9 have everywhere in decay except England. 

More news stories are coming through about the cumulative effect of the case load on hospitals and the ambulance service, it seems to me we can’t muddle on forwards like this indefinitely.  I’m going to do a detailed look through the demographic data for England to show why I think the rising cases isn’t as bad news as it seems.  I think another week of data is needed before we get a good idea if cases are going to continue falling in England like they were or not.  If we don't see cases start to fall soon, some difficult questions arise about why that is;  I think it's not quite the time to borrow trouble over that yet...

Link to previous thread: https://www.ukhillwalking.com/forums/off_belay/friday_night_covid_plotting_44-...

Post 2 - England Part I

The cases in plot 6s continue to go down and get up again like Tub Thumping from Chumbawumba.  I’m very much hoping they change the tune to I’m Falling by Comsat Angels soon.

Admissions are falling and have been for a couple of weeks; the plateau at the far right of the curve may well just be the trendline being twitchy to the random variations in the data.  We'll get a better idea next week.

Occupancy has been falling for a couple of weeks - down by about 20% in the last 11 days which is pretty big news, not least as this is the first sizeable decrease in almost 5 months.  A long way to go, mind but it's a much needed start to draining tension out of the system. 

Deaths - again the biggest fall since May, and the lowest level in two months.  Again, a long way to go.

So, all the healthcare/death measures have been falling for a couple of weeks despite cases going up and down like a yoyo.   I think there's a way of understanding all this that just about hangs together...

Post 3 - England Part II

I don’t think there’s any great mystery to why the hospital measures and deaths are falling despite cases being all over the place.   It just needs a lot of squinting at demographic details and rate constant plots to piece it all together.  Read on if you want my detailed take…. Or save your time and start binging Season 3 of Final Space, which has just landed on Netflix.

The first plot is a new take on my week-on-week method rate constant plot; I’ve used the demographic data and split it in to ages 0-14 and 15+.  This demographic data stops 4 days further back in time than the data used for plot 6 and the usual week-on-week method plot.

The second plot is the same one but with some annotations.  If you view this on a computer (not mobile device) you can flip-book between images with the arrow keys which I often find a useful way of comparing plots.

The rate constants these show are a good way of understanding how much the virus wants to spread in different ages regardless of the absolute number of cases - this is where I look for signs of policy, immunity, and other changes to the mechanic of the spread.

To talk through the regions on this annotated plot:

• Region A - This is the English school holidays - changes in adult rate constant tend to happen before changes in child rate constant - visible for about 5 features in the data in this region.  This suggests to me that the main source of child infection is household transmission from adults who brought it in to the house.  By about 08-21 the steam has gone out of adult growth and it’s in to consistent decay with an ever better halving time.  Children just about catch up and enter decay by the end of this period.
• Region B - Whammy.  Everything goes bonkers for one week.  I think this is partly (by the end of the week) a rising baseline from school transmission and mostly the switch-on of more testing associated with schools, leading to a rise in the fraction of infections detected as cases.  Without this “switch on”, I think the rate constant for cases would have better reflected the rate constant for adults, and shown this to be in ongoing decay.  It’s tempting to join the black ‘x’ marks either side of this region by eye with a line showing infections having every 21 days or so. This is why I think adult cases have actually been in decay for close to a month now, which fits with the otherwise mysterious two weeks of decay in the hospital admissions data and the deaths data for England despite growing cases.  The ONS random sampling surveys are also showing a decay, and these monitor live infections with a result that seems to lag new infections by a couple of weeks.  Something that just occurred to me is that the sudden jumps in rate constant at the start and end of region B are synchronous between ages, despite the lagged behaviour either side of the region - another sign that it could be a systemic effect.
• Region C - The additional testing is now embodied in cases on both sides of the calculation for the rate constant (cases on day X and day X+7) so doesn’t affect it any more and won’t until it switches off.   We see rate constants for children rising, and those for adults rising later, following the child ones - a reversal of the inferred causal link from region A; this suggests to me that now children are the primary source of Covid coming in to houses (adults having apparently exhausted their ability to sustain growth, and schools being a clear factor), and some adults are getting it from household transmission.  Plot D1.c shows that this growth is coming to adults aged around 35 to 50, with other adult ages remaining in decay.  That also feels consistent with household spread from school children.   There’s also an annotated version of D1.c with these two age ranges hi-lighted.

The left side of plot D1.c shows just how concentrated cases have rapidly become in ages 10-15 in particular.  It 

What does this all mean?

• The last week of rising cases in England should not translate in to much rising hospitalisation, as it’s almost en entirely been driven by cases in children - so if my interpretation of all this is right, there’s at least another two weeks of falling hospital admissions and occupancy ahead.   I really hope so, but no promises or certainty.
• If school cases continue to rise, it’s bad news for adult cases which have just tipped over in to growth and now appear to be following child cases.
  •  But… in geographically localised outbreaks we’ve never seen the rapid doubling times sustained for long, nor in university outbreaks; all sorts of feedback systems kick in and moderate things.  Hopefully the same will happen with school transmission, and the top level PCR week on week rate constant plot (which has 4 more days of data than the demographic plot) is very encouraging there. 
  • If the rise in adults is short lived it may not even translate in to a measurable rise in hospitalisations, as the variable lag between events blurs out small beaks.  Fingers crossed…

Looking at D1.c, some of the strongest decay is in adults aged 20-25 where we previously saw the highest level around the time of the football; a nice dark blue hole in this plot just waiting for the next round of university outbreaks to fill it?

As I said, this demographic data stops a few days earlier than the leading edge of the top level data.  That’s shown in the week-on-week plot without demographic breakdown.   It looks from that like the growth in top level cases isn’t going to be sustained; that almost certainly means the growth in adult cases remains very slight or goes back to decay.  We’ll have a much better idea by next week.

So, a lot of inference and a bit of speculation, but it feels quite promising to me - neither the last round of reopenings nor the return of schools have sent us in to uncontrolled spread of the virus (tentatively speaking) and when things do grow it doesn’t really seem to have its heart in it.  

Next up in the data, universities....?

Post 4 - Scotland

The leading edge of the cases plot 6s twitches up but I think that’s in the noise; the week-on-week rate constant plot (next post) shows cases continuing to fall.  Hospital admissions have been falling for about a week but so far occupancy creeps up - hopefully that’ll change any day now.  Deaths look to have turned a corner in to decay.  It's odd how occupancy is behaving notably differently to that in England with admissions in both turning to decay; I haven't looked in to possible reasons why.

If cases are returning to decay after both the mega-spike in younger adults after the last round of restrictions were dropped and then after the more sustained but lower round of growth in children from the schools reopening, perhaps the worst really is passed.  A bit soon to be getting hopes up, but I hope to be looking at sustained decay from now on, give or take the wobbles from the weather and a potential undergraduate case bomb.

Post 5 - Scotland Part II

The exponential decay in cases is slacking off as seen in the week-on-week method rate constant plot. 

This lacklustre halving time of around a month reflects the broad demographic; it had been masked by the much more aggressive halving time of the collapsing spike in cases from young adults - but that may have been more about the limited duration events causing the spike (opening night at clubs etc?) than the “true” potential for decay of infections.   

The demographic of the decay is pretty uniform, with only the previously spiked 20-24 age showing a stand out level of decay.

The rate constant is close enough to growth that a spell of bad weather could probably tip it over.

As with England, the booster program for 50 and overs and single dose vaccination for ages 12-15 are coming online.  Hopefully that gives things a bit more of a push in the right direction, as well as the gradual drawdown of the pool of people who've had neither vaccination nor past infection.

Post 6 - Grab Bag Plots

Plots left without space elsewhere.

Plot 18 - Regional Rate Constants (England)

• All regions have hospital admissions in clear decay
• Cases are growing in all regions.  I haven't done a demographic breakdown by region but it's probably not a bad assumption that it's being driven by the <15 years of age demographic in all regions.
• A couple of regions have yet to see deaths turn to decay, but this measure is very noisy at the regional level.

Plot 22 - Hospital Occupancy

• This has been on a long break
• Recent months are red, the rising phase of last autumn's wave is blue
• The lag from general admission to ITU admission "opens up" what would otherwise be a there-and-back-again diagonal line in to an ellipse.
• As I noted a couple of months back, ITU occupancy is much higher in this wave than it was in last year's for the same level of general occupancy; it happens (more or less by chance I think) to land on the return leg of last years wave.  
• ITU occupancy is at a similar level to where it was last year when the Kent variant blew the doors off the lockdown and sent everything rocketing skywards again; that's embodied in the bale blue invagination in the last wave's rising phase where things started looping round to decay but then got thrown in to growth again.
  • Well, let's hope that doesn't happen again...

I've had a pretty optimistic hat on for the last few weeks - there's no real momentum to growth in cases, and things feel like they're all moving in the right direction, but if the turn to decay in hospital measures is not sustained for a good month or so at least, I think that starts to raise difficult questions about the direction the UK is taking.  I hope it doesn't come to that.

Nobody has guessed the waterfall from a few weeks ago yet, so I'm chaining tack - "Name That Oil Rig".  

We had a very windy day last week and it stripped many of the ash trees of their green foliage in a couple of hours; falling like a green snow.   It looks like the dieback has severely weakened them this year.  Out over the magnesium limestone areas along the coast they're stark dead in masses, the view in to Eden Dene was very saddening.  Biosecurity, let's do more of that.

All sounds sensible and logical to me. Be interesting to see if you are right about the direction of the provisional data once it is fully resolved.

I think universities should be starting to show through in top level cases by now and should certainly be showing by next week if there is going to be a big spike? Hopefully by next week we'll be able to see a demographic effect if there is one. Although, the earliest effects will probably be in 15-19 (Freshers week etc) rather than 20-24? So slightly convoluted by school age data.

The other thing to look out for in the data over the next couple of months will be booster rates and, hopefully, hospitalisations of people post-booster when compared to pre-booster.

Intrigued. I have never really thought about how oil rigs are decommissioned - is this where they go to die? Where is it? (Obviously I have no idea of its previous identity!)

There are quite a few ports (Cromarty,  Nigg etc) being developed for decommissioning oil rigs and then commissioning wind turbines.  Decommissioning turbines is also becoming a thing as they have about a 25year life.

https://www.pocf.co.uk/decommissioning/

It started showing in the data around 2020-09-14 last year, as you say that should be in the top level now and visible in the demographics soon.  It’s possible that collapse in the school age growth rate is making rise in university ages in the top level data, so probably sensible to limit conclusions there until the demographics is out.  The timing depends very much on which universities and the pattern of student return; I wouldn’t call “out of the woods” until early October.

Last year’s spike was bigger in 15-19 than 20-24 so that band could be conflated, but the older band gives us a clue.

I’m hopefully that between two rounds of ludicrous speed outbreaks (4 day doubling times pre delta at my local one) and vaccination that there’re very few “zero immunity” students left.  The bigger risk perhaps is the number of people flying in from far away places and going straight in to the kind of sub standard accommodation that drove previous fast doubling times; if any of them has an escape variant it’ll explode.  Quarantine pending PCR clearance and priority sequencing for imported cases would be nice; something I have no idea on how our sequencing capacity is prioritised in times of high cases.

This is Able UK’s Teeside marine decommissioning facility.  I was driving past in the summer of 2019 and was more than a bit confused to see an oil rig near the road - photo below.  Seeing one nearby brought their size home in a way photos don’t.  That one is gone now, as was the next one.  Turns out they’re decommissioning a bunch of them.  Well, except for the toxic sludge filled legs which the owner wants to leave in the sea because it’ll be hundreds of years before the concrete corrodes enough to leak the toxic sludge.

There is a  problem with dead oil rigs without owners because the companies have gone bust.  No one knew quite what to do with them but some decom outfits are beginning to think they can recycle the steel etc. and make money, which should tidying things up a bit.

Thanks again - clear and excellent commentary as ever. I particularly enjoyed post 3 this week. What an interesting contrast between the return to school in England and Scotland - with children apparently driving infection in England, but not Scotland. Could the masks that secondary pupils are still required to wear in Scotland (and perhaps also the marker of the seriousness of Covid that mask-wearing represents) be the difference here?

With regards to the University ticking bomb..... I'm not sure about university terms elsewhere in the UK, but most freshers weeks in Scotland kicked off two weeks ago, and there's no sign of it in the demographic data yet (as you illustrate), which is hopeful at least. Teaching doesn't start till tomorrow though.

A very good point.  More on that from Beth Mole over at Ars - they certainly think so.  Touches on the situation in Florida which is just unbelievable - and the governor there is building a bigger platform than just Florida of his anti public health stance.  

https://arstechnica.com/science/2021/09/as-florida-punishes-schools-study-f...

Rate constant graphs - continuing an established pattern of directionless bouncing about.

The rate constants need to break out of that pattern and have a direction, to turn over into decay for a month (realistically longer for 3 or more halving times) so we stop running hot with fifty or a hundred deaths per day.

Obviously we want deaths to drop as far as possible, but is there any particular logic to your three halving times?

I think an average flu season - from a quick look at a couple of old news articles - is ROM 10000 flu deaths per year. That would be an average of around the 30 mark per day. Is that the sort of average covid deaths that we think would be 'societally acceptable' in the long term?* Or do we think it's more? Or fewer? I could see possibilities both ways - with all the deaths over the last 18 months the current numbers could seem low, but equally the greater awareness of numbers and the having known people who got the disease badly (making it more relatable) might make a bigger difference.

If we are at 50-100 deaths now then three halving times would take us to 6-12 per day. Of course one might expect a rise over mid-winter. In practice the booster programme might well mean the rise doesn't materialise very much because vulnerable peoples' immunity will be increasing as the seasons get colder... but equally I doubt we will ever actually get down to 6-12 per day.

* I hate this term, but what I mean is the level of deaths at which almost everyone might eventually just get on with their lives and no longer argue for enhanced measures to reduce the number of covid deaths or illness relative to the number of other deaths or illness. I don't mean what we as individuals would 'choose' if asked.

Interesting to step back and think about what we might expect of the rate constants over the next 3-6 months.  My take...

• For infections - perhaps we don't expect to see a decay in infection numbers for much longer if, as hoped, they move to largely sub-clinical nuisance level circulation with reinfection every few years, something of a steady state plus seasonality.
  • A really uniformed guess would be around 30,000 sub-clinical infections/day for a hypothetical steady state, if we assume it's re-infection every 6 years once immunity is built up.  We're perhaps at 2x to 3x that much right now.  Might be much less in summer and much more in winter if the disease syncs up to our seasons in the long run.  Colds had been synced like that for a long time, as I think the surveillance data shows over the last year, they have come unstuck 
• For cases - this really depends on the gating conditions for diagnostic tests (e.g. PCR); this is going to need a re-think at some point going forwards.  Probably needs a rethink already given the way symptoms change with vaccination - depends on what it's actually meant to be measuring which has never been that clear to me.
• For hospitalisations - totally agree with you, we need to see solid decay for a long time, indicative of the seriousness of the disease fading as we move to a world where people have all had multiple exposures to the virus through vaccination and infection.  
  • Still, if this virus ends up in the pantheon of winter respiratory viruses it's going to contribute to the winter pressures on hospitalisation once the diseases phase-locks to the seasons.   It's hard to set in advance a level that is "acceptable" as it could add to other winter pressures, or it could subsume some - it was for extreme reasons but we had Covid instead of flu last winter.   
• For deaths - much like hospitalisations; the current daily numbers wouldn't be totally out of the order if they were happening as part of the winter excess mortality in a bad year, and if they had similar demographics.
  • Right now I think we're still a long way from that point... (about 10x is my guess)
  • Why?   This is a hypothetical future where we would make much less effort to control the spread of the virus and accept that it is one of the contributors to winter deaths.  If this hypothetical future comes, there would be a significant demographic shift in the deaths from what we see now, and they'd presumably become highly seasonal.  That shift would happen because risk would reduce for all but the most vulnerable of society as immunity levels build up, and risk would increase for the most vulnerable because care staff and family will no longer be getting tested for Covid any more than they get tested for flu or common cold viruses, and other control measures will be gone.
  • It's a big question really - from all the different angles (moral, financial, pragmatic), can we justify going back to tolerating high winter mortality from respiratory viruses?  We've had a highly successful crash course in how to knock them back and how to better protect the most vulnerable.  I'm not optimistic on a grown up conversation happening over this at the political level.

In reply to thread:

Rate constant updates for England.  The top level rate constants continue to look like they're heading for decay.  The more lagged demographic data is still rising, but the <15s demographic rate constant at least is going to have to come crashing down over the next few days to match the top level data.   Hopefully the 15+ also comes down immediately; if not at least the apparent driving pressure in 15+ is going to fade.

In reply to Si dH:

Jinx; I spent so long trying to find the right wording you'd posted your reply to elsewhere before I closed mine out.

I agree on your asterisk.  I think there's more to it than the headline death numbers however:

1. If we removed all Covid restrictions from society and especially from care homes etc I think we'd see them go up quite a bit to a level that clearly wouldn't be acceptable.
2. Deaths are being held low I think by the amount of control measures; somewhat paradoxically I think that means a lot of the deaths are in the less vulnerable around whom the strongest protections aren't thrown.  So the deaths we are having are taking more life from people perhaps.  
3. The questions over what is acceptable is perhaps about how many years of life lost per death are acceptable as much as about the absolute numbers of deaths; right now my sense is that the situation now is still not comparable with a typical/bad "flu vaccine era" winter.

Enough to indicate something has changed and we have broken out from the pattern on graphs above of rate constant bouncing up and down in a directionless manner.

At some point we might reach an equilibrium when rate of loss of immunity equals rate of immunity gain due to infection and vaccination.

At that point the rate constants might bounce about chaotically* without direction as they do now.

I'm hoping we can break out of the current pattern to bounce around in the same pattern at a lower level of deaths but I'm not seeing it yet.

*infectious diseases (eg flu) appear to be sporadic, unpredictable and chaotic

It's either that or elimination.

I don't think it's chaotic at all with Covid (other than random, unpredictable timing of variant emergence or importation) 

• The strength and statistical certainty of the temperature link in the wobbles are large, and that's at a national scale with a 24-hour average temperature.  It'd be weird if the correlation wasn't stronger on more local scales in space and time, and presumably other weather factors - that are harder to access free online - contribute more of the variance.  
• Then there's so many factors various people have identified as correlating with other wobbles in rate constants - from big one-shot events to schools and universities etc.

Given all that juicy causality reasonably inferable from even the limited public data,  I don't see why we'd expect it to be so different for other infectious diseases.  We've just never instrumented them nationwide with > 50% engagement with testing before.  

I don't think we have to break out of the current pattern of cases stagnating at 10s of thousands per day to see deaths drop.  As I interpret it, we're basing our future on the idea that this virus would have been basically harmless if we'd all been periodically exposed to it from early childhood onwards.  Neither a first adult infection nor and adult vaccination with single-antigen vaccine are going to switch an adult from [lacking any prior immunity] to [like an adult who'd been exposed from childhood onwards].  They're two first steps on a journey to catching up on immunity.  The vaccine seems to have got us perhaps 90% of the way there, the hope I suppose is that the next steps cover the next 9%.   How close I think existing adults and especially older ones can ever get remains an unknown until we get there. 

IMO, we're looking for increased decoupling between infections and hospitalisations/deaths.  As this hopefully manifests in more infections being genuinely asymptomatic and some more falling outside of the currently defined symptoms for PCR testing, what we expect cases to do is  a lot more uncertain...

Anyhow, no sign of the school associated rise running away with itself, hospitalisations in clear decay, deaths turning the corner to decay, and hopefully at least one halving of hospital occupancy already "locked in".  

This has got me wondering what flu death stats would look like if everyone going into hospital was screened for flu with a PCR test every few days and we counted anyone that died within 4 weeks of testing positive. More. I'm gonna bet it'd be more than 10k.

Not sure what if any point I'm trying to make. I guess just that it's not obvious when we draw a line under it.

Not nationally, but locally we have, schools are having to introduce extra measures. uue to rates trebling since the start of September. In primary they are now  1100 and secondary 1400 per 100k. Meanwhile rates in 18 to 24 have dropped to 200. Although I think this is due to the ONS not keeping up with the exodus of this age group from Allerdale in September. 

Realise I could have sounded a bit dismissive there. It's definitely not yet. 

Sorry; I could have been clearer - cases/day are bonkers in that age range, but it doesn't look like growth is going to continue anywhere near like as fast - either exponentially speaking or hopefully in absolute terms.

A good time for teachers everywhere to make sure they're ventilating as well as possible.  It's not something I've seen commented on much but the commitment from our local teachers has been absolutely above and beyond throughout this disaster; I keep mulling over starting a thread on the home schooling experience but it's still a bit to fresh for comfort.

In reply to Longsufferingropeholder:

I agree we aren't there yet.  We're going to rely on evidence based public health to tell us when Covid has become comparable to the other viruses that annoy us as youth and stalk our last months with sudden ferocity.  But the sheer weight of evidence on Covid far outweighs the studies on other regular causes of mortality; for example I don't think the initial infection leading to terminal cases of pneumonia is often identified or categorised for example.  Stepping back to get an unbiassed view of this must surely be difficult.

Random link - BHF statistics on pneumonia; I was surprised how high up the ranked list the UK is.  The deprivation and regional statistics tell a similar story to Covid.  Covid really has ruthlessly exploited the inequality in society; a strong message from this and one we've shown very little interest in heeding (politically) is that both inequality and absolute poverty  are the enemy of public health.

https://statistics.blf.org.uk/pneumonia

Yep, similar thought process. How many "community acquired pneumonia" cases would be chalked up to flu if we tested every hospital admission for flu? 10k is the lower bound. 

But in the same way as has been discussed, what do you do about it and could it do more harm to prevent cases? Does slowing its spread just store up pandemic potential for future years? Sure we've learned a lot about behavioural control of respiratory viruses. How and whether that should change how we deal with the established ones is not clear to me.

Good thread here showing how cases in children have risen sharply in Wales.

https://mobile.twitter.com/LloydCymru/status/1441802857628545025

We are averaging a good bit over 100 deaths a day still (on the lower 28 day measure), with ONS estimating 45% of the population are still being careful with social distancing, significant numbers still working at home, major infection control in hospitals with a significant fraction on OPEL 4, ambulance services struggling even more. There is a chance things could go wrong again this winter and a horrible( if small) possibility of significant vaccine escape. This ain't anything like a normal flu season yet.

I tend to disagree because cases and deaths are not decoupled. The coupling factor is reduced ten or twenty times by vaccination but the chain of infections includes the vulnerable. Simply because they breath the same air as everybody else. Something* has to change for the coupling factor to change.

*variant, therapeutics, vaccination, herd immunity, immunity fade...

What's the likely timescale for that? To me it's an infinite time into the future, beyond the timescale for rational prediction which I see as months (if that). 66M at 40k cases per day is a long time and a lot of deaths at 100 per day for change...

You see halving/doubling times as event driven, I see it as chaotic because events like adoption of social distancing or lockdown are chaotic. 

I agree that something has to change - and it is changing.  Lots continues to change:

• The booster program for over 50s (a much wider net than the most vulnerable)
• A "virgin infection" or vaccination is the first major step in the change of individual immunity status.   Hopefully for most people their first post-vaccination infection is going to be worse than their second, and their second worse than their third etc, with the decoupling increasing along with those steps.  It's not "herd immunity" because it's not likely to set R<1 with newer variants, but its increasing levels of immune protection against severe disease across society.  The vaccine is good enough to make it possible to do this with only minor restrictions and without crashing healthcare.
• Therapeutics continue to come along.

The hope is that as the immune levels ratchet up across society after vaccination that the infection:hospitalisation ratio continues to decrease.

Although those are symptomatic cases and with the pre-vaccination definition of symptomatic being maintained, those cases are going to naturally fall in people more vulnerable to severe infection, it seems reasonable to assume there's lot more infection going under the radar.  I want to spend some time looking at the ONS random sampling data vs the new daily infections to try and get a handle on that.  But I think your denominator of 40k is too small; way to small.

Cases are a slippery customer here as they select those who are symptomatic or who have been exposed to a symptomatic person, and one characteristic of the rising immune levels should be an increased decoupling between infections and the current definition of PCR cases.  

Right now the demographic of cases and the demographic of deaths are quite separate - those cases are falling mostly in younger people, the deaths are coming mostly from older.  About half of cases are now in ages offered either 0 or 1 dose of vaccination which does little against spread of the virus (1 dose offering significant health protection however, and presumably raising the immunity from prior or breakthrough infection).  As that wraps up, case numbers are going to drop and some of the pressure driving R is going to drop.

It's I think close to impossible to put all the pieces together to understand where we currently are in the "immune normalisation" process.  But, I think things are still improving when it comes to the lethality of Covid infection. 

But the causes you list are not chaotic, they're event driven in response to rising pressures on healthcare.  We don't do lockdown because someone rolls snake eyes, we do it because universal healthcare is at risk.

At the individual level it's all very chaotic, but the statistical mechanics of it give a much more understandable view on larger scales.

At very low case levels the super-spreader mechanic means top level data itself is dominated by random events, but by levels much lower than we're seeing now they coalesce in to the rising baseline.

The only really "chaotic" events are the emergence of the pandemic and of subsequent variants, and the forces driving the weather, which drives fluctuations in the rate constant it seems - but the weather is more like an external forcing component than a chaotic component, because there is going to be little feedback from the pandemic to the weather.  The "directionless" wander of the rate constants you mention is largely correlated in a statistically significant way with the whims of the temperature.  Policy is subject to whim and fancy but at it's core it's a negative feedback loop which tends to damp things.

In case it wasn't clear, my reply was in the context of the previous post and I definitely was not suggesting we are in normal flu type territory yet (I don't think the word season is helpful because we don't know yet if that is how covid will play out. It is with flu, but my anecdotal experience is that colds come and go all year and are worse when things like kids starting school happen, not necessarily in winter. Covid could easily be a problem all year in the long term and just be a bit more common in winter, rather than having a clear seasonal pattern like flu.)

I think some of the things that you list, like working from home, some portion of the population being careful to avoid crowded areas and probably enhanced controls in hospitals, will become the new normal. They will just be part of what we do as a society in the long term, so I'm not really worried about what happens to infection growth in a theoretical state when they stop. 

In other ways I agree with you - we still have significant short term impacts on NHS staff and we still need to worry about variants. But, we do need to accept some level of ongoing health impact from covid as a society and I think it's useful to discuss what that acceptable level might be. That can help calibrate decisions, for example, over when extra measures are introduced over winter.

Interesting what Rob says about schools in Cumbria. Rates here are very high (>2000/100k in 10-14s) but no schools have yet shut to my knowledge. I would be very disappointed if they shut my son's school, at any infection rate. I have long ago resigned myself to him getting multiple bouts of covid and, now that everyone who wants to be is double-vaxxed, have decided his first year of education and his development are more important to me now than the downsides of schools remaining open. If necessary, I'd like to see other restrictions reintroduced elsewhere before they started shutting schools or sending bubbles home.

I think by this evening's demographic data we'll have 50% of cases in England in under 15s.

Quote from your Twitter thread link:

With infections this high, it absolutely baffling that the rules that worked to keep schools safe through the spring have been torn up and thrown out the window.

It's almost like it's entirely deliberate or something.  As elephants in the room go, this one is getting quite big.

Christina Pagel has had a lot to say about this on Twitter recently too. One thing I think she and other commentators are missing are the school busses.  Much smaller volumes of air than classrooms with higher densities of people for prolonged periods and no adults to enforce ventilation policies.  Masks aren't out of the window because the windows are all shut.  The school double decker I saw this morning had 0 ventilator windows open downstairs and 1 open upstairs.

In reply to Si dH:

Much the same here.  There's a lot of Covid coming home from schools right now, and a lot of parents who have not engaged with vaccination.  That's going to suck for them, but it's no reason to close the schools.  

Agreed.

Two steps parents/carers can take to minimise impacts are to make sure they are fully vaccinated ASAP and to use good household ventilation for the next few weeks (at least) to make sure any viral load is moderated.  It might be easier for the government to message this if they weren't obscured by the giant elephant in the room.

the issue with other restrictions rather than closing schools is that they will be much less effective than closing schools would be as school children are un-vaccinated.  Hopefully closing schools again would not be required mind. 

I not chaos I'd be looking for rate constants for October 2021.

I have nothing but praise for your intent on these threads and support the vast majority of what you say here but Id expect, for someone normally careful like you , when quoting deaths, not to give a range below the average; and preferably to be clear as you just have (with provisos) when we are speculating on possible futures. There is a very real political pressure to pretend things are 'normal' now, alongside suppression of bad news from the NHS that shows the opposite. I forgot to add in my previous post that we are currently in autumn, after a warm end to September, not in a cold and damp mid winter.

On any new normal I think it's still most likely things will pretty much go back to the old version.

I, like you, think schools should stay open  but if we were serious about ensuring that, vaccinating more older kids in the vacations, improving ventilation, more wearing of masks and other practical cheap actions (like bubbles) would have very much helped.  Where there are really bad outbreaks, temporary closure for all but the children of essential workers and the most deprived will sometimes be inevitable, as the staffing isn't possible. 

It is going to depend if Covid is mainly getting the same people who would have otherwise died of flu or whether it is getting an additional 30 people over and above the 30 who die of flu.

You're right - the first impact is going to be on parents; it's a rather harder to answer question of how much they in turn drive cases in other, more susceptible ages.  

Agreed.

In reply to elsewhere:

The chaos seems to have infected your typing a bit too much....

I don't think so. 30 extra covid deaths a week in an unrestricted society isn't anything like enough to disrupt everything else for. Even a hundred extra  deaths a week in an unrestricted society probably isn't. Given the choices we made this was always about preventing excessive pressure on the NHS. Currently I'd argue the NHS is in too much trouble irrespective of current covid impacts and heading into winter with a recruitment crisis will only get worse. Am I the only one who sees the irony in people being anti-tax rise but now spending on insurance to bypass NHS waiting lists. The US shows us where that leads... the highest per capita state spending in the world before a dollar is spent from insurance.

No, I don't see irony.   

I see new opportunities for profit, and some people getting exactly what they wanted.  There suddenly seems to be a marketing push on private health care in the UK - survey results making the news, lots of reports on rising waiting list times, individual case studies of people raising funding to go for private treatment, at least one large employer advertising private health as part of their package on the radio.  Amidst a fixed and insufficient pool of staff, driving more healthcare private is going to fuel a destructive spiral, not unlike how supermarkets offering higher wages than fuel firms has buggered things up over fuel transport.

I wish I saw irony and not a combination of long running under-investment and a global crises fuelling things in exactly the wrong direction.

I think it no coincidence that a libertarian American "think tank"  been positioning itself as the enemy of the NHS for some time (for example their disgusting and deceitful politicisation over a terminally ill child under NHS care in 2017) and has then taken a certain view over Covid.

The NHS isn't the only large institutional sector that this lot are gunning for either.  Some big US pension fund investments with long term strings attached are gong in to the HE sector as well.

Edit: Bugger, trying to avoid overt politicisation on these threads.  #Fail.

The numbers I used were not mine, they were a direct quote from the thread of posts above.

Bubbles are almost as bad as closures when rates are this high and 4yos won't wear masks without tearing them off after a couple of minutes (mine won't anyway.) If you can find one to fit in the first place. Otherwise I agree. Vaccination for those teenagers who want it, and more investment in ventilation systems, are no brainers. I also think teachers should be prioritised for boosters if they are over 50, unless rates drop fairly soon. I doubt it will happen though.

I just think we need to be as clear and calm as possible as I think science is loosing an information war on covid and hospitals are in real trouble.

Masks are useful in schools in all sorts of scenarios, especially indoor queues. You don't drive secondary school mask policy from 4 year olds. Bubbles could also be useful at times for older kids. We can't have known infectious kids in school so schools should do what they can to avoid making things worse and the govenment focus should be massive in this area.

If not chaos I'd be looking for rate constants for October 2021.

I'm still not sure I follow you, sorry.

If you mean "I can't see the rate constants a month in to the future because of chaos" I still disagree I'm afraid.  

Other than the possibility of a new, bad variant it's not random factors that are liming our view of the next month, it's our lack of understanding of what are quite deterministic systems in a  statistical mechanics sense.  

• We don't have great measurements of our current situation 
• We don't know all the rules to meaningfully run those measurements forwards by more than a week or two.  
• We don't have clear, pre-defined thresholds linking key metrics (e.g. healthcare pressure) to control measures. 

But those barriers to this are not random factors, they're limits to our understanding of the rules and our knowledge of the current situation.  

I think it's all to easy to be drawn in to thinking that seeking better understanding is hopeless or pointless to the random factors, but given how much of the apparently chaotic variation in the past data can be understood with hindsight, it clearly wasn't chaotic or random (other than variant timing) just complicated.  

This might seem like pedantic hair splitting I think it's an important distinction.  

If it were not chaotic, somebody would be able to predict future rate constants for October.

I'm good at this communications lark!

The weather is determinstic with all that thermodynamics, but it is still chaotic. The predictions get more uncertain the further into the future. Covid is similar.

Like I said above,

• I don't think the problem is that it's chaotic
  • (either in a strict mathematical sense, or in a more general sense of being dominated by unpredictable events - excepting variants).
• I think our ability to predict is limited by...
  • our lack of knowledge of sufficient detail of our present situation
  • our lack of understanding of sufficient detail about the systems involved to propagate our knowledge forwards

Other than variants

• I don't think general randomness or specific mathematical chaos are the limits to our ability to predict the future.  
• I think we can predict the future pretty well in sufficiently broad terms -
  • Immunity is going to rise, severity of infections are going to decrease, hospitalisations and deaths are going to decrease vs true infections
  • Short term variation in rate constants in infections are going to continue to reflect the weather, societal events, policy changes in directions that we can anticipate
  • Some degree of seasonality may emerge as the disease syncs up to the seasonality in our environment and behaviour
• What we can't predict so well is ...
  • The rate at which that future is going to arrive, or what the limits to the expected improvements are
  • If/when a problem variant will emerge. I hope the vaccines and the post-vaccination spread are severely curtailing the possibility for future variants to disrupt things in the way the last two notable ones did, building some fences around the one really big source of randomness going forwards.   

Edit to your edit:

It depends on what you mean.

• Short term fluctuations in the weather are chaotic, but longer term trends are not affected by that short term chaos.  
• Covid looks very similar to me - and coincidentally a lot of the medium term fluctuations are driven by the weather.  They're not an intrinsic property of Covid, but an external pattern forced onto the system. 
• The longer term stuff is not very chaotic looking at all, other than variant emergence.  
• Seasonality may emerge as an anchor that limits the scope for chaos to manifest in the timing of disease cycles, as perhaps is had been doing for other viruses before Covid lockdowns and control measures came along. 

I agree that a specific prediction of a day two months in to the future is rendered pointless by many things including randomness and potentially mathematical chaotic effects, but I don't think that is what limits us understanding the future for Covid, any more than our inability to forecast the temperature on a specific day two months in to the future limits our ability to understand where the climate is going.

You can model what a lockdown might achieve but you can't sensibly model when/how a government will apply a lockdown as that is a political decision.

Back in March 2020 you couldn't predict that there'd be two good vaccines announced in November 2020. 

The two biggest long term influences on viral transmission & its consequences were the product of human chaos.

I think we're a long way from settled science, established technology/therapeutics and settled population level immunology where we can be pretty sure deaths will be in a certain range of good & bad years like flu.

I agree (especially of late) but it's not "random" and it's not "chaotic"- it's monotonic - the worse things get, the stricter control measures get.   A pattern we've seen played out all over the world from governments to mass individual action to drugs gangs.  The general shape of a response to this sort of thing has predictable since the Black Death.   Then, as the situation gets less serious, control measures slacken.  Again, the general shape of this is long known and predictable.

I'm not suggesting that we can predict the future of government action, I'm saying that the difficulty in predicting the future is not in chaos but in uncertainty over the timing and extent  of more predictable choices.  This absolutely is not chaos and it's not down that much down to randomness.  Nobody was rolling the magic 8-ball to see if they should lock down yet.

I agree, but it wasn't "random" and it wasn't "chaotic" - again it was monotonic - the more time and money went in to the vaccine programs, the more likely good results would come out.  Again, the uncertainty was over "how good?" and "how long?".  Nobody rolled some dice to see how well vaccination would work, pre-existing platforms were adapted to this virus, trials were carried out and what was "unknown but knowable" became known.  Not being able to pre-empt the results of re-purposing vaccine platforms and trialling them is not randomness and it is not chaos.   As far as I know, none of the leading vaccines and candidates tie back to a single "Eureka moment" of unpredictable breakthrough but to applying known methods and known techniques to known platforms and testing them.  We could't predict their outcomes, but that's became of limits to our knowledge and understanding as I said above, not due to randomness or uncertainty.  

I disagree - lockdown was coming one way or another, there was simply no choice. The unpredictable (not random) factor was in its timing not inevitability.  This is not chaos.  

I think the emergence of the virus in people, and the alpha and delta variants (etc) are the only real, totally unpredictable events in this that are governed by true randomness.   

Nobody has claimed otherwise as far as I can see?   But we're a lot closer to knowing than we were this time last year, and we'll be closer still to knowing by this time next year.  We started with great uncertainty, and the bounds are tightening.

Lockdowns or not, vaccination or not, once elimination was off the cards there was only ever one end game to this pandemic, the questions are over how quickly we get there, how much suffering and loss there is along the way, and how much worse than the pre-Covid world that destination is.  

None of these are settled questions, but the general shape of the destination is clear; it's a journey the species has been through dozens of times.  There's uncertainty over the details - uncertainty that is reducing as time goes on.  But the destination is neither random nor chaotic.  It's uncertain.  Over more than the shortest timescales, or distance from that destination decreases monotonically over time.  

There's a significant difference to what these interpretations of uncertainty vs randomness mean IMO.  We started out with massive uncertainties on everything, and those uncertainties are tightening as we get closer to our destination.  But by and large the uncertainness were down to gaps in knowledge and understanding, not down to people playing dice.

In defense of elsewhere chaotic systems can still lead to broadly predictable outcomes: the weather front which crossed the UK this morning is a good example. If you apply control to a chaotic system you can still modify outputs, liquid flow being a classic example. Chaotic systems are just far more likely to produce unpredicted surprises.

Sure; I thought I agreed with that in my comments on weather vs climate up thread.

But I look at the spread of Covid and I see a lot of causal links, very few of which stem back to random or chaotic factors, rather things that are generally monotonic and known to us in a broad brush sense.  We are limited in our predictive ability by many things, but I think "chaos/random"  isn't high up that list at all other than for variant emergence.

Loads of causal links back to human decisions that are not random and not mathematically provable as anything. But they are pretty chaotic.

Our rates have nearly trebled again last week, so now over 4000/100k. I think all schools are still open, but now siblings of positive cases are expected to isolate as well. 

Given the new local guidance, and if cases continue to rise at the same rate, there will be a couple of schools that will have more kids at home, than in school, next week. Not sure what is the best way to minimise disruption is, but I think the lack of masks, ventilation and distancing that I have observed this term have contributed to the problem. 

Maybe it's your background. To me random is the background of statistical thermodynamics or quantum mechanics and chaos theory is just a clever way of looking at  non linear systems (eg fluid dynamics). The sexy bits get exaggerated for chaos theory, but despite butterfly effects most of it can lead to reasonably predictable outcomes.

The role of randomness in a chaotic system is very different to the role of randomness in statistical mechanics and quantum mechanics.  A pandemic is more like SM than a chaotic system in my opinion.  Run the pandemic in a hundred parallel universes and it’ll end up in basically the same place but with variations on the journey.  That’s not a chaotic system, it’s a statistical mechanics system.

The attractor for a pandemic is barely any more strange than it is for a thermodynamic system when a partition is removed.  The Lissajous figures for this pandemic are basically ellipses with minor deviations, and they end up shrinking to a very dull and small ellipse tied to seasonality.  It just takes several goes around the loop to get there with a pandemic.

A very timely argument with the first ever adaption of Asimov’s retcon’d Robots and Foundation universes landing on TV this week (excluding Star Wars obviously…) 

You could say the same about a well-behaved fluid dynamics system. I think the pandemic resembles that, way more than statistical thermodynamics, where outcomes are so highly predictable on the macro scale they gave rise to classic thermodynamic equations. 

I don't want to hear about Foundation as I don't have Apple TV (and it won't surprise you I feel our best possible futures are being undermined by the tech giants). Psychohistory is very attractive b*llocks...it makes for great fiction but becomes scary when you read about the regard taken of Hari Seldon by powerful 'influencers' convinced of their own greatness. If I had to predict a critic's future view of the show it would be expensive but flawed ambition (then again GoT had that in spades and was a commercial success).

Schools globally: https://www.reuters.com/world/the-great-reboot/defying-delta-back-school-go...

A quick read suggests that keeping measures like masks, ventilation etc. along with giving students/parents the option of being vaccinated before they returned to school minimises the disruption caused by students needing to isolate. 

shocking who'd have thunk it.

Especially given how school cases increased in the last few weeks before the summer holidays?

To be honest, from my, admittedly limited observations (five schools and two DofE groups this term) I'm equally worried about staff and students being under pressure to "catch up" Not sure what I've seen is sustainable or good for their wellbeing. 

I don't normally show the hospital occupancy plots for all 4 nations (plots 7.2 e/s/n/w) but here they are with today's data release. The turn to decay is clearly visible on all four nations now.

Good.

Going to be an interesting update by Saturday I think.  Change is in the air, and perhaps it's not just the onset of autumn.

In reply to joem:

Not so many people given the ongoing obsession with hand sanitising and comparative lack of interesting in ventilation....   I'm gong to stop now before I start ranting about busses and their windows again. 

In reply to Offwidth:

Spoiler free comments on Foundation.

Thankfully I've missed those 'influencers' and wasn't aware of people taking it as gospel.  Shame they've not read the books; it never really worked that well in the books (other than as a self fulfilling prophecy) and took a lot of help in the form of subtle manipulators working behind the scenes to achieve the desired outcomes.  Now I say it out loud you might have a point about the influencers...

I've been a bit confused by the nitpicking over the TV adaption taking liberties with the books; the various books in no way hang together as a single, consistent whole given the efforts to merge the two separate series decades after they started, and Asimov's preference to tell good stories for their own sakes over slavishly adhering to a single consistent universe.  You can't make a faithful and coherent adaption; Chose one.  Chose coherency.

Good news but why did it take so long?

https://www.theguardian.com/technology/2021/sep/29/youtube-to-remove-misinf...

1 in 20 secondary school kids with Covid, ONS estimates. Sounds like a fair few to me.

 
Otherwise some positive sounding news from Merck today re oral drugs preventing serious outcomes. https://www.reuters.com/business/healthcare-pharmaceuticals/mercks-covid-19...

Answers on a postcard to the usual lobbyists’ address.

None too soon given the escalating violence against healthcare workers in the states from the people ensnared by these orchestrated delusions.  https://arstechnica.com/science/2021/09/health-workers-get-panic-buttons-as...

In reply to Kirsten:

Very promising press release.  Got some questions over this one; will be very interesting when they go for MHRA conditional approval to see if they go for approval for people without other risk factors and what the restrictions of a conditional authorisation over here will be.  Covid certainly seems to have accelerated this pre-existing compound through trials.

My cynical side sees anti-vaxers will use this drug (if approved) even if side effects turn out to be as serious or more serious, as its not a vaccine