UKC

Friday Night Covid Plotting #29

This topic has been archived, and won't accept reply postings.
 wintertree 05 Jun 2021

If the weather is still having an effect on the rate constant, hopefully there's a bit of respite ahead for case numbers; what a week.  

I'm starting with the same new plots as last week - Plot A - "Stylistic Cases" only shows the full detail for the biggest growth areas, otherwise rising cases are represented by a straight  orange line linking their first and last values on the plot, and falling cases a similar blue line.  I hope this brings some clarity to the split in behaviours - lockdown working in some places, lockdown failing in others (more than last week) with cases gently rising, and a few places having major outbreaks.   Plot B shows the average cases/100k for the non-outbreak rising areas and the falling areas.

It's interesting that the resurgence in cases since May 21st in Blackburn with Darwen and Bolton seems to tally with a rising phase in many of the other outbreak areas. 

Big picture - the very short doubling times seen in Bolton didn't last, and haven't been seen for long anywhere else.  Rapid exponential growth doesn't seem to be getting sustained elsewhere either.  Lots of factors at play around partial gaps in vaccine coverage, importation events and the Indian variant (now named Delta, in a sign of optimism from the WHO that we're not going to see more than 24 notable variants...).  From the weekly Thursday variants reports out of government and the cases data I get the impression that big outbreaks are happening where the Indian variant lands in areas prone to high case rates in earlier times (see the later demographic post) suggesting work/social factors at play as well as the gaps in vaccine coverage.  Whilst the force of the outbreaks are contained, it's still I think driving spread of the variant across the UK - even after it has spread "everywhere" it hasn't spread to all people, and the outbreaks still function as epicentres exporting more infection to other people elsewhere.  

So far, the variant - now known to partially evade immunity from a single jab against catching the virus both from longitudinal healthcare data and from at least serology study -  isn't leading to big outbreak-style behaviours in other places.  But, it's early days for those other places and a week or two to soon to really be certain I think.

Plot C shows the demographic of cases in leading "outbreak" areas vs the rest of England.  A big shift towards younger ages as has been visible for a couple of months in the D1.c demographic rate constant plots.  This is doubly good news because there's much less infection in those most likely to become ill from the virus, and many of those who are infected in the older ages will still be protected by two jabs, meaning they will have better healthcare outcomes vs a few months ago.  So, we don't expect to see hospitalisations happening with the same ratio of cases:hospitalisations as in previous waves, although there are suggestions the Indian variant is worse in this regards.

The process of renormalising risk to the Kent variant and now the Indian variant reminds me of the process of cooking a lobster alive from cold water.  One wonders just how much potential to get worse the virus has; certainly it's a kitten even now compared to its predecessors.

The rate constant plot for English PCR cases is heading for a doubling time of 10 days; this was typically only seen at times when cases were low last year, given that very consistent behaviour and the way the outbreaks areas are not seeing seeing sustained growth, I'd be surprised if it gets much shorter or sustains at this level for long. 

Link to previous thread: https://www.ukclimbing.com/forums/off_belay/friday_night_covid_plotting_28-735318


2
 wintertree 05 Jun 2021
In reply to wintertree:

The four nations plots:

  • Wales looks like it might be about to join Scotland and England in growth in cases.
  • Scotland is only a couple of doubling times away from its peak daily case rate from January.  About 30 days at the current doubling time.  We'll look at Scotland more in another post.
  • Doubling times of ~15 days for England, Wales and Scotland look to be where we are, although it's in the provisional window.  This data is filtered over longer time periods than the new "week on week" method above, which is why the times for England are longer.  

 wintertree 05 Jun 2021
In reply to wintertree:

The English plots.

  • Cases doubling quickly
  • Hospitalisations are rising, having tipped over to growth on plot 9 about 2 weeks after cases.  This was low level growth in cases; the steep rise associated with the outbreaks happened a couple of weeks later, and it's only just now and in the next week that we'd expect to see that big rise in the blue curve on the right of plot 9 start to see a matching trend in the orange curve for admissions.
    • Thinking about it a bit more, doubling times are not the best lens to verify the effects of vaccination on hospitalisation rates; it's more likely a lag-adjusted ratio between the two is more useful - a "case hospitalisation rate".  I'll try that for next week.  The rate constants are a good way to measure that lag I think however.
  • Hospital occupancy is just starting to rise.
  • Deaths - these looked to be rising last week, and indeed a small positive signal has developed on the rate constants plot as a doubling time, but it's now decaying away again.  I wonder if this is related to the initial "fast outbreak" phase exploring gaps in vaccine coverage, and now we're transitioning to wider spread through partially vaccinated older adults and unvaccinated children and younger adults, meaning the brief rise in deaths is not sustained?

 wintertree 05 Jun 2021
In reply to wintertree:

The Scottish plots:

  • Cases - a couple of doublings away from an all time high, perhaps a month at current doubling times
  • Hospitalisations are rising
    • As above, these should be compared to cases with an appropriate lag to understand how the ratio of cases:hospitalisations is changing from the previous wave.
  • Deaths are generally 0, 1 or 2 a day.  Things can't get much better than that.  It does mean the rate constant measurement for deaths is dominated by noise on the right hand side and isn't to be taken seriously.
  • I haven't looked at (can't get?) the demographic data for Scotland as I have for England, but my sense from the news is that the initial outbreaks were again related to various factors including relative gaps in vaccination, and so the initial hospitalisation rates are probably worse than for the more muted rise that would happen as the outbreak segues in to wider spread.   So, lots of reasons to hope it'll be fine, but...
    • At the rate cases are going, if hospitalisations don't reduce sufficiently from cases it'll be too late to stop it going wrong very soon given the latencies in the data from infection and the lag from policy change to effect on transmission.  Feels a little reckless to me, even if there's plenty of reasons to think it's going to be okay.  


1
 wintertree 05 Jun 2021
In reply to wintertree:

Plot 16: The UTLA Watch plot.

  • The main plot annotates UTLAs with over 12 cases/100k/day.
    • This is my "red" list and is what is used for the demographic plot in my first post.
    • Only a couple of additions since last week
  • The "crop" plot show all other UTLAs to have grown over the last 3 weeks - more than last week

Plot D1.c: The demographic rate constant plots

  • England data is split in to "outbreak" and "the rest" 
  • Cases are clearly featuring more at younger ages as per the plot in my first post.  In the outbreak areas, we're just starting to see them in the oldest people as well however.
    • A very visible consequence of the vaccine program - and without the vaccines I think we'd now be in a really bad place.  Pause for thought.
  • The cases plots are absolute numbers and are normalised by the colour map.  It's really interesting that the magnitude of the cases around October 2020 is similar in the outbreak areas compared to in January 2021, where-as in the rest of England it was much lower in October 2020 than January 2021.   This suggests the outbreak areas had it worse than much of England before the Kent variant stormed the scene and broke everywhere.  Now, they're having it bad again and the leading areas had the bad luck to also be where the latest new, worse, lobster boiling variant landed.
    • I should really do plots of cases/age bin/100k people/day, that would be interesting.

Plot 18: Regional rate constants

  • Remember, the far right of cases plots are provisional and can always go up (very unlikely to go down)
  • Interesting that some areas still eschew growth
  • Having different "NHS Regions" for the middle plots to English Regions for the top and bottom plots is not helpful.  If anyone is bored and wants to make a list of individual NHS trusts per English region, trust level data can be repartitioned to a better simulacrum of the English regions...
  • The far right of the rate constants for hospitalisations is very twitchy to the small number statistical noise in the data.  Other than coarse behaviour of growth vs decay, I'm taking it with a pinch of salt until there's more data.
Post edited at 21:36

 abr1966 05 Jun 2021
In reply to wintertree:

Thankyou!! It's all beyond me but if you were in the same pub garden as me I would gladly buy uou a pint....your analysis is very much appreciated and respected! 

 chris_r 05 Jun 2021
In reply to wintertree:

>  (now named Delta, in a sign of optimism from the WHO that we're not going to see more than 24 notable variants...). 

That genuinely made me laugh out loud. 

See, Wintertree isn't a doom-monger. 

 wintertree 05 Jun 2021
In reply to wintertree:

Variants plot:

  • This is data "by reporting date" and lives perhaps a week behind case - on average with more range to it than that.
  • Both the South African and Brazilian variants are really petering out, I'll drop them next week
  • A few new ones have appeared, we'll see what they do over the next week.
  • At this point, it would take whatever the opposite of a miracle is for a new variant to out compete the SA strain, landing as it did in areas perhaps prone to outbreaks and at a time we we're dropping the last restrictions that were keeping R<1, and given its lobster boiling increase in transmissibility and it's increase single-dose partial vaccine evasion.
  • The doubling time of the India variant has again backed off...
    • The fit to the whole period is now 7.3 days (up from 7 days last week and the crazy 5.1 days in earlier weeks).
    • The most recent week represents a doubling time of 8.4 days.
    • Some of this will be down to the stalling of growth in the worst outbreak regions masking behaviour elsewhere.
    • I think everyone expected the initial outbreak growth and 5.1 day doubling times to not continue for long, but it's still a relief when all the data confirms that hope.

Vaccines plot:

  • By my notional matching up for doses, the effort has covered the bumper week of first doses from March.  
  • Doing this has eaten in to slack in the the gap between doses a bit, adding about 1/3rd of a week to the gap.
  • What's known about increased partial vaccine evasion of the Indian variant against just a first dose suggests lowering this gap is a priority; we know people over a certain age are being offered an 8-week gap;
    • I get the impression worries over loosing a convenient time/location by cancelling and then taking pot luck again through the booking system are putting some people off this for various understandable reasons.
  • Only one more week to go before the "second dose debt" starts to drop significantly.
    • In theory, this means first doses speed up ~2x
    • In practice, as most first doses are now down to younger adults, this depends on the relative availability of different vaccines, information on which is a bit opaque and has been discussed at length towards the end of plotting #28.

 Toccata 05 Jun 2021
In reply to wintertree:

You have no idea how important your analysis is to me and and great number of people. The analysis transcends any official source I have. We salute your effort! Thank you.

 mcdougal 06 Jun 2021
In reply to wintertree:

This is where I come to get information on the pandemic. Thanks Wintertree, this stuff is very much appreciated. 

In reply to wintertree:

> At the rate cases are going, if hospitalisations don't reduce sufficiently from cases it'll be too late to stop it going wrong very soon given the latencies in the data from infection and the lag from policy change to effect on transmission.  Feels a little reckless to me, even if there's plenty of reasons to think it's going to be okay.  

I think it's too much of an ask to shut down the summer season for hospitality and tourism when nobody is dying and the unionist parties and media are fighting even the existing restrictions. 

The good thing is the Scottish school term is over on 24th June.  S5 and S6 have had their exams and, at least in my daughter's school aren't in much.   Weather's good, schools will soon be out and the jags are going in fast.  Maybe it will be OK.

 elsewhere 06 Jun 2021
In reply to wintertree:

The last Scottish vaccine usage I looked at has a binary divide - AZ is second dose only (94%), Pfizer/Moderna is first dose only (97%).

Just a few days ago there were significant number of AZ 1st doses but it looks like in the last few days Scotland has run out of over 40s to give the first jab to. Needs a few days of data to confirm but earlier than I had hoped as only 86% of 40-49 and 93% of 50-59 have had first jab. 

Those figures are low only in comparison to astounding levels of uptake so far.

25k 2nd jabs given to ages 18-39 in last 7 days exceeds 5k Pfizer Moderna 2nd jabs 7 day total so this must be AZ. However I would have thought priority group 6 would have had 2nd jab by now so I don't understand that. Did some other healthy under 40s get their 1st AZ before blood clot policy change?

And as ever, thanks. I learn a lot from your data and the thread in general.

Post edited at 09:21
In reply to elsewhere:

>However I would have thought priority group 6 would have had 2nd jab by now so I don't understand that. Did some other healthy under 40s get their 1st AZ before blood clot policy change?

You'd be amazed about the number of >40s who didn't get jabbed when they were supposed to.  3 of my colleagues didn't - one because his his childcare provider was ill and he needed to be fit for looking after his kids, another because he was a bit anti-vax, and another because he was fussy about the date he was offered (our booking system is a 'you get what you're given' kind of affair. They're all over a month later than they were supposed to be jabbed.

1
 wintertree 07 Jun 2021
In reply to Toccata, mcdougal, abr1966 & elsewhere:

Thanks all; it keeps me off the streets...  I know from other posts on here that some of you are working at the sharp end of the pandemic or helping people afterwards, it's nice to think I can help people with that if only by helping them perhaps follow the picture in more detail.  

I have one advantage over official sources - I am unbound and free to add my own interpretation which they can't.  The media is free to interpret the data and I've seen very little I have much time for there excepting John Burns-Murdoch at the FT.  

A quick lunchtime plot - making a stacked plot of cases/day for the areas (UTLAs) colour coded red/orange/blue in plots A and B.

  • The total vertical size of the "river" is the number of cases per day; it's getting bigger.  
  • We can see the 63 "blue" areas have been shrinking, masking some of the growth at the top level.  Over the least 7 days, their halving time has been ~45 days - so barely decreasing at all now, likely about to tip over in to growth.  
  • The 15 red "outbreak" areas have, taken together, a doubling time for cases of ~42 days; their growth has almost stalled, perhaps to turn to decay although some of the newest outbreaks could take over driving growth I think that less likely given that the outbreaks apparently started in the places most amenable to their spread.  
  • Cases from the 71 "orange" areas added together are growing with a doubling time of ~14 days.  

I think the orange ones are the ones to watch... Do the doubling times remain at ~14 days? Do local responses and ongoing vaccination stretch this to a longer doubling time?  Do the live infections start closing the gap to effective herd immunity faster than vaccination?  If doubling times get much shorter that would seem to be a worrying sign that the effects of vaccine induced immunity are failing, so that's one to look out for at the UTLA level; if these orange areas had "outbreak potential", we'd likely already have seen outbreak-grade sub 7-day doubling times; so if any of these areas switch to rapid doubling, that could be a critical early sign of an immune evading variant.

I think there's vaccination status data at the UTLA level from the dashboard now, it would be interesting to plot exponential rate constants against some metric derived from the vaccine data...  See if the blue regions are further ahead there or just "lucky" and are a bit delayed in joining the orange regions in growth.

Post edited at 13:55

 Si dH 07 Jun 2021
In reply to wintertree:

Nice plot. I think any comparison between utla case growth and vaccination %s will be confounded by demographic differences but it might be interesting to see if it's easy to do.

If you are going to track the progress of your three streams, are you going to fix which utlas are in each as of today? Obviously if you redefine the blue/yellow each week by which are growing then it will become a self fulfilling prophecy.

Good to see the reds overall are growing very slowly now. Bolton is slowly dropping but I had been worried lots of other areas might follow it to similar rates before turning over, it seems that isn't happening fast apart from in Blackburn.

 wintertree 07 Jun 2021
In reply to Si dH:

> If you are going to track the progress of your three streams, are you going to fix which utlas are in each as of today? Obviously if you redefine the blue/yellow each week by which are growing then it will become a self fulfilling prophecy.

I’ve been mulling this.  I don’t think I will, because I don’t think this plot is predictive for individual UTLAs beyond the next week or so.   So I’m thinking it’s an interpretation aid for the now, rather than a tool for prediction.  That being said, interesting to look for causal factors over the separation.

Everywhere is probably heading for some growth; the concentration of certain factors and importation events gave leading “red” areas a head start but growth is likely everywhere at some point.

Have you looked at the cases plot for Sefton today?  Another outbreak of similar size to the first is coming along; stochastic stuff at work?  Another reason prediction is hard for now.  

In reply to wintertree:

> Have you looked at the cases plot for Sefton today?  Another outbreak of similar size to the first is coming along; stochastic stuff at work?  Another reason prediction is hard for now.  

Just went and looked at this. The small absolute numbers and age affected are a familiar story. School outbreak or something? 

Am I the only one struggling a bit with the new plot? What does the 'negative' half mean, and what do the callouts on the right point to? Feel like it should be obvious but I'm having a brain fart.

Edit- never mind. Looked at it for a bit longer. I get it if the Y-axis is just a length scale, i.e. cases/inch

Post edited at 18:26
 wintertree 07 Jun 2021
In reply to Longsufferingropeholder:

A plot like this is “floating” so not anchored to y=0.  The vertical extent of the shaded region - height of the envelope - is the total number of cases.  It should really have a vertical scale bar and not an abused y axis numbering. So there is no “negative” half.  The benefit is that where you have several strands with diverging behaviour, they get very hard to interpret on a normal plot as one strand “pushes” another around.  Setting them floating about the centre of mass and carefully ordering then removes that.

The annotations are the doubling or halving times for total cases within each colour.

Sorry, I didn’t explain much about the plot, rushed it together in a lunch break and when I rush I forget that what’s in my head isn’t in other people’s heads...

> Just went and looked at this. The small absolute numbers and age affected are a familiar story. School outbreak or something? 

Once it’s out of the provisional window the demographics may tell all...

Post edited at 18:25
In reply to wintertree:

Yep, crossed edits, thanks all the same.

> Once it’s out of the provisional window the demographics may tell all...

This may or may not represent one or two sevenths of the truth:

https://coronavirus.data.gov.uk/details/cases?areaType=ltla%26areaName=Sefton#card-cases_by_specimen_date_age_demographics_-_above_and_below_60

 Si dH 07 Jun 2021
In reply to wintertree:

> Have you looked at the cases plot for Sefton today?  Another outbreak of similar size to the first is coming along; stochastic stuff at work?  Another reason prediction is hard for now.  

Yer, I've seen it. Interestingly it looks like it might be centred on Formby again, but more broadly spread than before.

Everywhere in Merseyside is rising in a similar way though. All six Liverpool utlas are still 'light green' on the map but increasing quickly, I suspect we are just a couple of weeks behind Manchester.

Cheshire West is the same, there are outbreaks in Northwich and Winsford. 

Edit - age distribution varies across the utlas but I think there is more overall in late teens and early twenties than in school ages.

Post edited at 19:02
 Misha 09 Jun 2021
In reply to wintertree:

Thanks as ever. Great to have the graphs as it’s hard to get a good idea ‘by eye’ when numbers are still relatively low.

Reflecting on the vaccination drive, 77% first doses is fantastic at this stage. I never thought we’d manage that so quickly. Would be approaching herd immunity with the original variant but I’m guessing we now need over 90%. The reality is that there are significant differences by demographic and geographic factors, so we can expect local outbreaks to continue. Not great given the risk of other variants arising.

Here in Brum, we are over 20% behind, so it’s not surprising that our rate is now in the 60s, having been in the teens for a couple of weeks at the low point. Looking at the demographic analysis (now available for local authorities on the dashboard), we are in the low 90%s for over 70s and then steadily dropping by age bin. I don’t know if that reflects a slow local roll out or vaccine hesitancy.

Birmingham and Solihull (total population 1.3-1.4m) have a joint local programme. There are 3 large vaccination centres (one or possibly two of them only just opened) and 27 local ones at larger GP surgeries. I’m not clear on the criteria for getting called to a local centre - I’m guessing older people, but I’m 40 and got called locally, perhaps because the centre is at my local GP surgery. I’m lucky enough to live a 5 min walk from one of the local ones, as well as a 20 min walk from one of the large ones. I guess many people will be further away and it could be a bit of a trek for those relying on the large centres, though the main one has good public transport and car parking. My brother had to go to the large centre in Sandwell as the main Birmingham one was booked up for a week ahead (good and bad I suppose), whereas the one across the road from him wasn’t open yet.

So access could be part of the reason for relatively low uptake but I doubt it’s the main factor. I’m hoping that as the number of second jabs reduces, our vaccination rate will pick up... don’t fancy being in local restrictions by August!

Got my second jab today, bang on 12 weeks (got the invite 2 weeks ago but was up north on holiday).

 Misha 09 Jun 2021
In reply to Si dH:

It’s interesting that whilst Lancs and Manchester are going south and Merseyside is edging up, Sheffield is about as low as it’s been since before the winter, having been dark green for longer than most other areas... Covid moves in mysterious ways. 

In reply to Misha:

> Covid moves in mysterious ways. 

People, I think...

 Toccata 09 Jun 2021
In reply to Misha:

> Reflecting on the vaccination drive, 77% first doses is fantastic at this stage.

I think it’s a bit of spin to report the vaccination rate as a percentage of adults. I make the first vaccination rate 40.57m/66.65m, 60.8% and the second vacc rate 42.2%. Given it’s the latter that counts against delta we have a few months of vaccinations to go before we’re close.

2
In reply to Misha:

> Reflecting on the vaccination drive, 77% first doses is fantastic at this stage. I never thought we’d manage that so quickly. Would be approaching herd immunity with the original variant but I’m guessing we now need over 90%. 

Yep, that's 77% of adults. Not even close to herd immunity. It's not really a practical aim. Endemic nuisance is the realistic endgame.

In reply to Misha:

> ... but I’m guessing we now need over 90%.

I'd query this part too. The classic HIT of 1-1/R0 suggests probably not that high. You'd need R0 of ~10 for it to be that high, which is pessimistic.

 wintertree 09 Jun 2021
In reply to Longsufferingropeholder:

Yes, I think suppression without significant control measures or elimination - through vaccine induced herd immunity - has been off the cards since last November and the Kent variant.

I’ve been trying to understand how much of the harm of this virus is down to immune disregulation resulting from it being entirely novel / unseen to most people’s immune responses and how much is down to actual mechanism(s) of the virus.  Most of it seems to be immune disregulation and downstream consequences of that.  The narrow spectrum immunity against the spike protein conferred by the vaccines feels like a stop gap against this given the amount of host specific optimisation that’s going on with the spike; I’d love to know who if anyone is working on a vaccine against the capsid or membrane fusion proteins to give a broader spectrum response, although there are other ways of achieving that and for those with no vaccine yet that’s looking increasingly unavoidable…. Possibly also for those of us with our first dose only.  Given the vaccine induced demographic shift perhaps that’s not the worst outcome.  Big gaps in my understanding however. Although having this go down before the immune regulating therapeutics are out of the trials pipeline is not what I’d hoped for.

Post edited at 08:20
 wintertree 09 Jun 2021
In reply to Longsufferingropeholder:

> I'd query this part too. The classic HIT of 1-1/R0 suggests probably not that high. You'd need R0 of ~10 for it to be that high, which is pessimistic.

If talking about granting herd immunity through vaccination alone, the variant has a double whammy of increased R0 and decreased efficacy of the current vaccines.  With current vaccines having an efficacy of perhaps 80% against symptomatic infection, if R0 is over 5, herd immunity is off the table without leaning on infection acquired immunity as well.

Edit: That 20% are still protected by much better health outcomes by their vaccination which is the important bit when, as you say, endemic nuisance the least bad accessible way out.

Post edited at 08:22
In reply to Longsufferingropeholder:

> Yep, that's 77% of adults. Not even close to herd immunity. It's not really a practical aim. Endemic nuisance is the realistic endgame.

The endgame is a few years off.

What the world has done is created whole new industries to respond to the pandemic.  I think we're being pessimistic.  If the money keeps flowing these industries will continue to develop and improve like all tech industries and in a few years there's no reason we won't have better vaccine technologies, scaled up vaccine manufacturing, high-throughput automated vaccination, better PPE including masks, further improved testing and more advanced tracking and tracing tech.   

The world needs to be ambitious and try and use all this new tech and industrial capacity to eliminate Covid and then turn it on other viruses.   The fact we've almost never managed to get rid of a virus completely in the past is neither here nor there.   We'd never got to the moon before we got to the moon.  Doing things which used to be impossible is normal.

In reply to tom_in_edinburgh and wintertree:

Yep, we're putting almost about the right amount of effort into dealing with this. Shows what the world can do when there's no option but to deal with something. Would be nice to be more prepared and resilient to things like this, and yet still I see nobody working on Giant Fire-Breathing Lizard defences.

And yes, HIT needs a factor of vaccine effectiveness in there, but I have similar but different guesssumptions on this out-of-my-field stuff. It just leads to that rabbit hole surrounding which effectiveness number you care about. Infection? Probably not really. Onward transmission? Yeah, for sure. What's the gap between those two? We just don't know well enough at all for each vaccine and each dosing schedule.
I guessumed antibodies were only any use against the spike protein because that's what disrupts the binding to cells. Would antibodies stuck to other bits do anything? Would love it if someone could educate me.
And it does seem like the bad consequences (acute phase at least) are entirely down to the cytokine problem, which would imply (guessumption again) that one day, when we're all exposed at a young age and then regularly, this will be another common cold. If therapeutics can be found to moderate it, and produced cheaply and in volume, and supplied worldwide, that could get us moving.
(aside: still no idea why expensively nebulised interferon would be effective but cheaply injected is not seeing any attention. Anyone got any solid knowledge?)

Post edited at 09:24
1
 Offwidth 09 Jun 2021
In reply to wintertree:

More or Less this morning highlighted a covid related statistical gem. UK vaccination rates use ONS population measures (certainly an underestimation) and for England this average is higher than all the local data averages! (as they uses local registrations for the divisor, which is likely an overestimate). This might explain the weird high numbers on national age group vaccination percentages.

 MG 09 Jun 2021
In reply to abr1966:

> Thankyou!! It's all beyond me but if you were in the same pub garden as me I would gladly buy uou a pint....your analysis is very much appreciated and respected! 

Wintertree is now very pissed, judging by the likes.

 wintertree 09 Jun 2021
In reply to Longsufferingropeholder:

> It just leads to that rabbit hole surrounding which effectiveness number you care about. Infection? Probably not really. Onward transmission?

Indeed, but the uncertainty only works in one direction and that's to increase the effective R0 in the calculation you gave.  

A key thing I don't understand about the "80% protection from symptomatic infection" with 2 doses - is the longitudinal?  Does it protect 80% of the people 100% of the time, or does it protect individuals by reducing their chance of becoming infected by 80%?  This distinction has a massive difference to how things peter out at the end of the day, and how broader spectrum virus granted immunity ends up distributed.

> I guessumed antibodies were only any use against the spike protein because that's what disrupts the binding to cells. Would antibodies stuck to other bits do anything?

One of the resident posters with professional immunology experience (DG I think?) posted a paper a few months back looking at epitopes and antigenic coverage for other proteins - there's plenty of epitopes on both the membrane fusion protein and the viral capsid IIRC.  In terms of not ending up in an endless arms race with the virus it's not so clear to me - especially as more therapeutics come online - that leading with neutralising antibodies is the best long term way forwards for the less vulnerable people (*), lowering the importance of the spike compared to other proteins.  As I understand it, the CD4+ activation of CD8+ T-Cells will cause destruction of infected cells and this can be triggered by epitopes presented during synthesis of the spike, membrane fusion and viral fusion proteins.

(*) But it was IMO absolutely the right tool to bring to the problem for where we are now, don't get me wrong.  I do wonder if the membrane fusion protein also provides epitopes for neutralising antibodies, but I don't know.  My career plan is (hopefully) pretty undefined after the next few years right now, but getting in to immunology on some level has a great appeal to it.

 kirsten 09 Jun 2021
In reply to wintertree:

"Around 80% of adults in the UK are likely to have Covid-19 antibodies, either through infection or vaccination, according to new data reported by the PA.

The latest estimates from the Office for National Statistics represent an around 10% increase in each part of the UK, including Scotland where they were the lowest at 73%.

The estimates were based on blood test samples from private households from the week of May 17."  - BBC 

In reply to wintertree:

> Indeed, but the uncertainty only works in one direction and that's to increase the effective R0 in the calculation you gave.  

Yes, absolutely.

> A key thing I don't understand about the "80% protection from symptomatic infection" with 2 doses - is the longitudinal?  Does it protect 80% of the people 100% of the time, or does it protect individuals by reducing their chance of becoming infected by 80%?  This distinction has a massive difference to how things peter out at the end of the day, and how broader spectrum virus granted immunity ends up distributed.

Not sure how we could ever know the answer to this in the wild, but you can bet it's 'a bit of both'. Can't see a way to measure how many times someone was exposed and not infected outside the lab. There are SAR figures in the variant reports but from what I can tell they're not at all rigorous.

> One of the resident posters with professional immunology experience (DG I think?) posted a paper a few months back looking at epitopes and antigenic coverage for other proteins - there's plenty of epitopes on both the membrane fusion protein and the viral capsid IIRC.  In terms of not ending up in an endless arms race with the virus it's not so clear to me - especially as more therapeutics come online - that leading with neutralising antibodies is the best long term way forwards for the less vulnerable people (*), lowering the importance of the spike compared to other proteins.  As I understand it, the CD4+ activation of CD8+ T-Cells will cause destruction of infected cells and this can be triggered by epitopes presented during synthesis of the spike, membrane fusion and viral fusion proteins.

> (*) But it was IMO absolutely the right tool to bring to the problem for where we are now, don't get me wrong.  I do wonder if the membrane fusion protein also provides epitopes for neutralising antibodies, but I don't know.  My career plan is (hopefully) pretty undefined after the next few years right now, but getting in to immunology on some level has a great appeal to it.

Yes, DG posts are awesome.
I presume we'll see more studies appearing on the relative effectiveness of the live-unattenuated vaccine soon. Will be interesting to compare to the current range of man-made alternatives.
The bit I don't get is how antibodies binding any of the membrane/nucleocapsid proteins could be neutralising, but that's for me to go and find out more about somehow.

 wintertree 09 Jun 2021
In reply to Longsufferingropeholder:

> Not sure how we could ever know the answer to this in the wild

It's a difficult one for the statistical types I think, and I doubt a clear cut answer will emerge.  Be interested to know what the immunology types think... 

> The bit I don't get is how antibodies binding any of the membrane/nucleocapsid proteins could be neutralising, but that's for me to go and find out more about somehow.

I wasn't suggesting that the capsid protein could be neutralising, but that its synthesis by infected cells presents a lot of targets for the T-cell mechanisms in particular, to moderate infection.

I was wrong about the "membrane fusion" protein; I thought the M-protien was involved in the fusion process and so could potentially be used for neutralising antibodies but that's guff memory on my part, it seems the membrane fusion machinery is in a sub-unit of the S protein.   the membrane protein does still present a lot of epitopes for the T-cells however; so an alternative vaccines based on that and the capsid would be less effective against preventing infection but would perhaps still prevent the runaway infective processes that pre-empt the eventual immune  over-responses.

 elsewhere 09 Jun 2021
In reply to Longsufferingropeholder:

> I'd query this part too. The classic HIT of 1-1/R0 suggests probably not that high. You'd need R0 of ~10 for it to be that high, which is pessimistic.

Neil Ferguson is estimating R0 for Delta is 5 to 8, compared to R0 of 2.5 for original in China.

18 mins into 6pm news on Radio 4 today.

In reply to wintertree:

some interesting papers about to drop:

https://p.dw.com/p/3ubOZ

In reply to Longsufferingropeholder:

Also the preprint linked here is on my reading list.

https://p.dw.com/p/3u7V8

Can't get past how much it reads like a software problem. Which, I suppose, it is.

In reply to Longsufferingropeholder:

Yep. Just got through it. You (wintertree) will definitely want to read that.

Direct link: https://www.researchsquare.com/article/rs-558954/v1

 wintertree 09 Jun 2021
In reply to Longsufferingropeholder:

Paper 1 - that rather pre empts the UK study, the power of doing an impromptu experiment nation wide…. A great result - not unexpected given the interest in running exactly that kind of trial, but great.  I’ll avoid political comment given this is a data thread.  

Paper 2 - I’m going to have to read it now…. if this is a mechanism of RNA expression of the spike protein occurring at the RNA stage, does this mean the mechanism is shared with the virus, and so whatever risk factors individuals have are present for the vaccine and the virus?

Edit: should have read the paper before pondering - crossed with your post.  Paper title:

“Vaccine-Induced Covid-19 Mimicry” Syndrome

Post edited at 21:40
In reply to wintertree:

> Paper 1 - that rather pre empts the UK study, the power of doing an impromptu experiment nation wide…. A great result - not unexpected given the interest in running exactly that kind of trial, but great. I’ll avoid political comment given this is a data thread.  

It's great news. If the final results follow then that's both the 12 week gap and the mixing of vaccines fully vindicated. Although this initial reveal part of the study does seem to conflate the two, I do slightly wish our ever-present many-faced tantrumist could be here to share these moments.

> Paper 2 - I’m going to have to read it now…. if this is a mechanism of RNA expression of the spike protein occurring at the RNA stage, does this mean the mechanism is shared with the virus, and so whatever risk factors individuals have are present for the vaccine and the virus?

Yes and no and maybe, but yes, but sort of no. Read the paper. It's a longish one but I *promise* you won't regret it. It's one of them ones that you just can't put down. I won't ruin the ending. It's a little bit speculative in places, but you wouldn't bet against it being the answer.

> Edit: should have read the paper before pondering - crossed with your post. Paper title:

> “Vaccine-Induced Covid-19 Mimicry” Syndrome

 MG 09 Jun 2021
In reply to Longsufferingropeholder:

Anyone know why the India cases have come right down? Has it just burnt through everyone?

 Misha 09 Jun 2021
In reply to Toccata:

True but it’s still not clear how much spread there is from children, particularly younger ones. When case rates among children went up in March, those in adults did not. Although given it was due to wider use of LFDs, it’s not entirely clear. 

 Misha 09 Jun 2021
In reply to Longsufferingropeholder:

If Delta is 40% more than Kent which is 40% more than the original with an R of about 3, current R would be around 6. I’m which case your formula suggests 85%ish. The big question is how (younger) children factor into this, ie is their R naturally lower somehow. 

 wintertree 09 Jun 2021
In reply to Longsufferingropeholder:

You’re right about the paper.   Thanks for avoiding any spoilers.

My summary if I got it right: Viral RNA designed (by evolution) for cytosolic expression, if converted to DNA for a viral vector delivery, can then produce different RNA by triggering intron stripping functions during DNA conversion in the nucleus.   This removes parts of the RNA that happen to have intron style motifs, eventually resulting in a different protein that can be soluble not membrane anchored, which triggers the rare side effects. Totally get the software/exploit analogy although perhaps it’s closer really to information theory.

What’s fascinating about this is that RNA viruses and synthetic mRNA carrying vesicles can produce proteins within a native cell that the cell will struggle to produce itself, because of its nuclear subroutines for intron stripping.

I’m having a proper mind blown moment.

Its amazing that introns are still such a mystery.

 Misha 09 Jun 2021
In reply to MG:

> Anyone know why the India cases have come right down? Has it just burnt through everyone?

That and introduction of restrictions / people self isolating voluntarily. 

In reply to wintertree:

Yep. That was what I took from it. My summary in different words if I got it right: There are splice options available in the nuclear expression of the DNA other than the desired one which can result in production of RNA that's missing a bit that can produce protein that's missing a bit that can get out of the cell, rather than just pop up on the surface, and float around and bind to the receptors that the viral spike would bind to and get hounded by antibodies.

What wasn't clear to me was whether they're saying infection with the real virus can or can't produce these soluble proteins. Need to read it again when it's not bedtime.

Post edited at 06:28
In reply to MG:

> Anyone know why the India cases have come right down? Has it just burnt through everyone?

Eh?!

 MG 10 Jun 2021
In reply to Longsufferingropeholder:

About 70% drop in cases from the peak. I didn't think their lockdown was very significant, and the vaccination rate is low.

In reply to Misha:

> If Delta is 40% more than Kent which is 40% more than the original with an R of about 3, current R would be around 6. I’m which case your formula suggests 85%ish. The big question is how (younger) children factor into this, ie is their R naturally lower somehow. 

https://en.m.wikipedia.org/wiki/Herd_immunity

Talking about R and herd immunity only works at a population level. And even then only vaguely.

Also I don't think there's any good evidence for lower transmission in young kids is there? Harder to swab properly and less likely to show symptoms, sure, but I've not seen anything that definitively says they transmit any less than adults. Happy to be corrected though.

 wintertree 10 Jun 2021
In reply to MG:

A poster on a different thread said the colleagues they spoke to daily in India were under a regional lockdown that was stricter than any from the UK.  The thread got nuked from orbit however. Imposed lockdown or not, people respond to bad news - and the news in India got far worse than here.  

I think what we’re seeing is that control measures work; for the virus to have gone near universal would need a much larger under-detection ratio than the already significant ones that have been mooted.

1
 MG 10 Jun 2021
In reply to wintertree:

I guess so.  I am bit surprised, having visited India,  that any sort of lockdown is possible.  Interesting contrast to Brazil too, where there seems to be some sort of steady state at a high level.

In reply to Longsufferingropeholder:

> Yep. That was what I took from it. My summary in different words if I got it right: There are splice options available in the nuclear expression of the DNA other than the desired one which can result in production of RNA that's missing a bit that can produce protein that's missing a bit that can get out of the cell, rather than just pop up on the surface, and float around and bind to the receptors that the viral spike would bind to and get hounded by antibodies.

> What wasn't clear to me was whether they're saying infection with the real virus can or can't produce these soluble proteins. Need to read it again when it's not bedtime.

I'm not sure either - here's helpful summary of the viral replication process.  There is a transcriptional step but it takes place in various odd perinuclear structures and I'm not clear whether they have access to nuclear splicing mechanisms (or even if that's required for this sort of splicing)  

https://www.nature.com/articles/s41579-020-00468-6

I was tipped off about the Kowarz preprint by a climbing friend who happens to be basically a professor of splicing while we were climbing in Pembroke last week.  He thought it was credible and it explains a lot of my thoughts about the role of spike protein in thrombotic events.

 Offwidth 10 Jun 2021
In reply to MG:

I've seen life in India from 5 star hotels through middle class living, to rural life and even had a tour of some Bangalore slums. Thinking about the proportion of indoor poorly ventilated space you see in India and class age demographics the middle classes were always by far the most vulnerable to covid.

As for how the numbers can drop that fast.. public reactions in middle class India to bad news on contagious disease has always been pretty strong.

On a different subject, we now have news of PHE trying to keep more pandemic simulation tests secret. This includes a MERs simulation.

https://www.theguardian.com/society/2021/jun/10/secret-planning-exercise-in-2016-modelled-impact-of-mers-outbreak-in-uk

1
 wintertree 10 Jun 2021
In reply to Dave Garnett:

Interesting differences noted in the paper between the prevalence of intron tags in the AZ and Janssen vaccines, along with the suggestion that the codon optimisation step may have increased their prevalence.  

I've been pondering this today; if mis-expression of the S protein is a step along the path to these rate events, how does ageing at the molecular level - particularly in older patients with worse stress responses during the illness - leading to worsening quality control of the expression process factor in to clotting events from live viral infections? 

Interesting that human proteins will have some selective pressure against their becoming dangerous under mis-expression but that perhaps viruses don't have that pressure; so one doesn't just have to consider deliberate mechanisms but failure modes, with there being less pressure for viruses to be fail safe.

Fascinating stuff.

Post edited at 15:01
In reply to Dave Garnett:

Thanks for that. Really good summary there.

The Kowarz preprint states:
"Most importantly, the RNA of this virus is translated and replicated only in the cytosol of infected cells, and thus, evolution of coronaviruses has always taken place in the cytosol of permissive cells, and in the absence of processes that are necessarily taking place when nuclear-encoded genes are being transcribed (capping, splicing and poly-adenylation)."
And that's not contradicted by that nature rev paper.
So to me that hints that there's no chance of the same mis-transcription in the case of viral infection. So does that mean

> I've been pondering this today; if mis-expression of the S protein is a step along the path to these rate events, how does ageing at the molecular level - particularly in older patients with worse stress responses during the illness - leading to worsening quality control of the expression process factor in to clotting events from live viral infections?

is moot in the case of sars-cov-2??? Still not sure I can reconcile all the statements in the paper, but it's definitely only because of my complete lack of knowledge.

Also states:
"Moreover, nearly all severe cases of SARS-CoV-2 infections (COVID-19) suffer from life-threatening thromboembolic events due to the many viruses with Spike surface protein in the blood stream. Even pseudoviruses with Spike protein on the surface cause strong inflammatory reactions in tissues and endothelial cells, indicating the danger of this protein when available in a systemic fashion"
Which possibly addresses the point on last week's thread about the dangers of the vaccines being fully and wholly enveloped by the dangers of the virus.

Post edited at 17:57
 wintertree 10 Jun 2021
In reply to Longsufferingropeholder:

> Which possibly addresses the point on last week's thread about the dangers of the vaccines being fully and wholly enveloped by the dangers of the virus.

To clarify, I said that as a likely point in relation to mRNA and RNA vaccines only, and then clarified to note they have very rare allergic side effects not apparently related to the virus.

I assume when an infection really gets going in a weakened patient more spike proteins end up in the blood, hence some of the reason for the demographic difference in the vaccine side effects and live infections…?

In reply to wintertree:

> > Which possibly addresses the point on last week's thread about the dangers of the vaccines being fully and wholly enveloped by the dangers of the virus.

> To clarify, I said that as a likely point in relation to mRNA and RNA vaccines only, and then clarified to note they have very rare allergic side effects not apparently related to the virus.

You did, but now I'm thinking it must be true even of the viral vector vaccines. So one of my edits was taking out the mRNA caveat on the last line.

> I assume when an infection really gets going in a weakened patient more spike proteins end up in the blood, hence some of the reason for the demographic difference in the vaccine side effects and live infections…?

So I'm not sure if they're implying it takes a weakened patient. It sounds like they're saying, and seems plausible that, it's a given that the spike protein will be found floating about in an infected individual. Seems very plausible when you think what has to happen to all the spikes on the cartoon virus that aren't on the side that sticks to the cartoon cell. They must end up floating around with all the other bits of discarded virus crap?

In reply to Longsufferingropeholder:

> You did, but now I'm thinking it must be true even of the viral vector vaccines.

if the mechanism in the preprint is the real culprit.

 Si dH 10 Jun 2021
In reply to Longsufferingropeholder:

> Which possibly addresses the point on last week's thread about the dangers of the vaccines being fully and wholly enveloped by the dangers of the virus.

I've been away this week so not really contributing much. Still following along where I can though. Without disputing the scientific detail, I would suggest the evidence directly contradicts this theory. The rare blood clot issues caused by the AZ virus albeit at quite low numbers, have occurred independently of other symptoms or hospitalisation in people who are in low risk groups from the virus itself. There is no evidence I have ever seen anywhere that otherwise-healthy people get covid and suffer similar clots, without having other severe symptoms that have first put them in a hospital bed and made them more vulnerable to those issues as a result. Unless there is a pool of people across all age groups who have contracted covid and died from a resulting brain clot independent of other symptoms, which has somehow gone unobserved, then the proposition is demonstrably false.

 Si dH 10 Jun 2021
In reply to Longsufferingropeholder:

Ps. I think you made a comment up thread about small kids and transmission. Independent of biological factors, whatever they may be, it's clear from a fluid dynamics perspective that viral spread via either droplets or aerosol from the mouth of someone 3 feet tall will on average be much less than from the mouth of someone 6 feet tall, partly due to reduced particle range but also further exaggerated as an effect if most of the surrounding uninfected people are adults with mouths at 5 feet up. The difference would be massive. I suspect it is at least enough to explain any measured differences in spread between younger and older kids, which have certainly been postulated fairly confidently by various scientists and the media. Just a theory though, and doesn't limit classroom spread between kids so much. I do also have sympathy with a perspective that says many young kids' infections are simply not picked up because they are either (1) asymptomatic, (2) have symptoms without obvious cough or fever so are not tested, or (3) have symptoms but are not tested anyway because they get ill all the time and their parents don't want to put them through it.

Post edited at 22:49
 girlymonkey 10 Jun 2021
In reply to wintertree:

Just watched Tim Spector's update this week and he is a bit less upbeat than usual! Still sounds positive for double vaccine doses but single dose far less effective.

youtube.com/watch?v=OHBua3aXQ7c&

In reply to Si dH:

This was about the third hit on Google

https://www.bmj.com/content/373/bmj.n1005

> I've been away this week so not really contributing much. Still following along where I can though. Without disputing the scientific detail, I would suggest the evidence directly contradicts this theory.

I'm not finding any evidence that contradicts it. I've had a medium good look but I'll keep going. What I'm finding is that CVTs definitely happen in covid patients. 

> The rare blood clot issues caused by the AZ virus albeit at quite low numbers, have occurred independently of other symptoms or hospitalisation in people who are in low risk groups from the virus itself.There is no evidence I have ever seen anywhere that otherwise-healthy people get covid and suffer similar clots, without having other severe symptoms that have first put them in a hospital bed and made them more vulnerable to those issues as a result.

Typically would be with other symptoms, because there more going on than just the spike protein. But that's exactly what the statement 'enveloped by' meant. Vaccine -> possible CVT. Covid -> more possible CVT along with extremely likely other bad things.

> Unless there is a pool of people across all age groups who have contracted covid and died from a resulting brain clot independent of other symptoms, which has somehow gone unobserved, then the proposition is demonstrably false.

It would be highly unlikely to show up without other symptoms in the case of viral infection. Not seeing why it has to be demonstrated to be independent for 'enveloped by' to be a fair statement.

Regarding kids, my intuition is that despite projecting any droplets less far, they lick everything, so it's not clear why or whether they would be more or less infecty. Would be great to find some evidence one way or other because it seems a big and important unknown.

 Si dH 11 Jun 2021
In reply to Longsufferingropeholder:

> Regarding kids, my intuition is that despite projecting any droplets less far, they lick everything, so it's not clear why or whether they would be more or less infecty. Would be great to find some evidence one way or other because it seems a big and important unknown.

Perhaps more likely to infect parents and close relatives but less likely to cause a mass spreading event?

You aren't going to persuade me on the other topic so we'll have to agree to disagree.The importance of independence between symptoms is that the act of going to hospital and lying in a bed for a long time puts you immediately at high risk of blood clots relative to the rest of the population; to demonstrate that the covid virus has the same blood clot risk as the vaccine you need to show me that similar numbers of healthy people outside of hospital have got them. It shouldn't be hard to do for the groups at low risk from covid because there are huge numbers of them who have been infected with no or mild symptoms.

Perhaps I'm instinctively suspicious because the argument that a vaccine can't cause any symptoms outside the scope of the virus itself is exactly the sort of argument people make without proper substantation when they want to avoid scrutiny or regulation.

Post edited at 07:53
In reply to Si dH:

I don't expect to change your mind on this, and actually I'm pretty grateful for the discussion and opportunity to understand a considered but different opinion.

My thoughts in the last few posts are fully predicated on the mechanism in that paper being correct, from which it follows that free spike protein floating around is a bad thing. And the inference that infection by the virus would lead to that situation is mine and something I can't directly verify but is implicit in a lot of the process explanations I've been reading.

I'm definitely slightly more confident in the safety of the mRNA vaccines having read these papers and improved my understanding a little. Seems like that bit of mRNA is getting into everybody's cells at some point, the choice as I see it is whether it comes with the rest of the package or not (and, I guess to some extent whether it's certain that it will be correctly expressed).

 wintertree 11 Jun 2021
In reply to Si dH:

>  the argument that a vaccine can't cause any symptoms outside the scope of the virus

I know I’m banging on but I was suggesting that seemed likely only for the Pfizer vaccine and I’m happy to extend that noddy thinking to other mRNA vaccines.

I was going to say that this argument would hold no water for licensing and that evidence is king there, but after this week’s developments in a different area, one could almost theorise that something else can influence the decisions.

https://arstechnica.com/science/2021/06/a-disgraceful-decision-researchers-blast-fda-for-approving-alzheimers-drug/

Lately I’d been taking the view that the emerging failure of the first wave of alt Alzheimer's drugs was shaping up as a very expensive error of judgement in treating the plaques as the primary cause of symptoms and so a valid therapeutic target, rather than as one of many separate downstream effects of the causal mechanisms, and one not critical to the progression of the disease.  There are animal models of the disease where no significant extra cellular deposition of amyloid beta plaques builds up…

But na, let’s just licence it anyway after destroying their analysis as having no statistical power, take it to market with a 40% chance of brain bleeds, and see what happens.

In reply to Longsufferingropeholder:

Digging a bit deeper: https://www.nejm.org/doi/full/10.1056/NEJMc2009787?query=RP

https://www.sciencedaily.com/releases/2020/06/200604095600.htm

And for interest: https://stm.sciencemag.org/content/12/570/eabd3876

Not really big enough numbers to shout about, but interesting.

> ...exactly the sort of argument people make without proper substantation when they want to avoid scrutiny or regulation.

You're right. I'm not trying to preach here, although I seem to be in danger of falling into that mode again. Just looks like there are some really interesting loose ends that could be tied together. If the O/AZ clotting mechanism is what that paper thinks it is, and that's in part common to one of the effects of the virus (my conjecture), that's an incredibly interesting development.

IRT wintertree:

Yeah that one is really sketchy. Shines a light on the American system and makes me want to pray we don't end up there.

Post edited at 09:24
 wintertree 11 Jun 2021
In reply to girlymonkey:

> Just watched Tim Spector's update this week and he is a bit less upbeat than usual! Still sounds positive for double vaccine doses but single dose far less effective.

Maybe I've blown a fuse but I think we might be going in the right direction - all be it a bit faster and sooner than I'd like.

 mondite 11 Jun 2021
In reply to wintertree:

> But na, let’s just licence it anyway after destroying their analysis as having no statistical power, take it to market with a 40% chance of brain bleeds, and see what happens.

Havent they been given nine years to carry out a follow up study to prove whether it works or not?  am sure they will carry out the study as fast as possible and not hold it back until 23:59 on the day it needs submitting.

 Offwidth 11 Jun 2021
In reply to girlymonkey:

Agreed, there was a point of concern though: deta variant symptoms have shifted a lot and don't match the government symptom lists very well anymore. A bit too close for comfort to symptoms for hay-fever and summer colds.

https://www.huffingtonpost.co.uk/entry/delta-variant-covid-symptoms_uk_60c31132e4b0af343e9e39c6

 wintertree 11 Jun 2021
In reply to mondite:

> Havent they been given nine years to carry out a follow up study to prove whether it works or not?  am sure they will carry out the study as fast as possible and not hold it back until 23:59 on the day it needs submitting.

As I understand it, the compound has been licensed for sale and use, and the company has been mandated to perform a follow up study as well to determine if the compound actually does anything beneficial.

> am sure they will carry out the study as fast as possible and not hold it back until 23:59 on the day it needs submitting.

This is not the issue, the issue is that the efficacy data is poor and there are strong side effects; many credible industry experts are suggesting this compound should not have been licensed for sale off the back of the evidence to date, and indeed licensing it for sale is not necessary or required in order to perform an additional study.  

This FDA decision is ruffling a lot of feathers.

 girlymonkey 11 Jun 2021
In reply to Offwidth:

Yes, I wasn't sure if that was because it was more prevalent in young people who tend to get more mild symptoms anyway or whether the virus has actually changed? These symptoms sound like they are milder, so does that mean that the virus is doing the shift to more transmissible but less deadly? I guess it's just to early to tell, but it would be useful if it did turn out to be the case!

In reply to wintertree:

There was a British-Irish council meeting this morning and there was a lot of talk about the Delta variant. As a result of the lowered efficacy of single vaccine doses against it, our (Guernsey) proposed opening up of travel on 1st July has been put back and a new condition of 2 doses +2 weeks is required to travel relatively freely within the UK Common Travel Area.  I would expect to see similar 'put back opening up' approaches being taken by the other UK jurisdictions in due course. Anyone keen can watch our government briefing from lunchtime on www.gov.gg .  We're currently covid-free and completely unrestricted on island with 14 day quarantine for most travel areas and less for parts of the UK (diminishing as prevalence increases).

Post edited at 16:42
 wintertree 11 Jun 2021
In reply to thread:

Spoiler alert for tomorrow - things are on the up - updated plot 18 below.

Doubling times are an odd one to interpret when a “conversion rate” from one measure to another is changing (vaccination) and it's clear to me I need to go back to cases and hospitalisation actuals and make some comparative plots for this wave and the last wave as well...

A brain teaser for the mathematically inclined out there:

  • If a fraction "x" of cases go to hospital with a fixed latency, both cases and hospitalisations will double with the same characteristic time - accounting for latency - regardless of the value of x; this fraction does not feature in the exponential term of hospitalisations.  
    • Say we have 100:200:400:800 cases each week and x=0.01 then 1:2:4:8 people going to hospital a couple of weeks later - both have a doubling time of a week.
    • But now say x=0.1, now it's 10:20:40:80 people going to hospitals- the doublings time is the same.
  • But, in the later case, 10 times as many people go to hospital, so starting from a period of level cases, the measured doubling time for hospitalisation must be shorter, as the measure has to rise more in the same period of time...
  • Eh?  (I hope I have explained this clearly enough)
  • There's no actual paradox but the solution lies in the many differences between idealised maths and the real world with things like quantisation, dispersion, noise and so on.  In a theoretical sense, doubling time is useless to visualise a change in hospitalisation rates (a worse variant could mean up, vaccination means down).  In a practical sense it has to capture the change, but interpretation is a jumble of all those factors separating real life from idealised maths.
Post edited at 16:58

In reply to wintertree:

> But, in the later case, 10 times as many people go to hospital, so starting from a period of level cases, the measured doubling time for hospitalisation must be shorter, as the measure has to rise more in the same period of time...

This is where you broke it. What do you mean by more? It's still doubling. That's not more. That's the same change. You can't just wake up from logarithm space like that. What do I win?
Like you've said, but in different words, it makes complete sense if the example numbers you start with are much bigger than the smallest amount they are allowed to change by.

Post edited at 17:30
 elsewhere 11 Jun 2021
In reply to wintertree:

> But, in the later case, 10 times as many people go to hospital, so starting from a period of level cases, the measured doubling time for hospitalisation must be shorter, as the measure has to rise more in the same period of time...

No. It's just 10:20:40:80 intersects with the static background 3ish doubling times earlier than 1:2:4:8. Same curve at intersection on log graph.

10=2**X where X is 3ish

Post edited at 17:35
In reply to Longsufferingropeholder:

Another point that seems relevant to throw in here is that average age of those hospitalised is dropping (bad because young people getting hospitalised, good because means old people aren't) the length of stay in hospital is also considerably shorter than it has been. Which is good for occupancy. But messes up the doubling time of yet another metric.

 wintertree 11 Jun 2021
In reply to Longsufferingropeholder:

> This is where you broke it. What do you mean by more? It's still doubling. That's not more. That's the same change. You can't just wake up from logarithm space like that. What do I win?

But, the higher the conversion rate from cases to hospitalisations, the more the actual values have to rise from their current level  - so they have to double faster from their previous level at the end of the last, unfinished decay phase (“starting from a period of level cases” - and hospitalisations) to get to their future, higher level.

The answer lies in...

In reply to elsewhere:

> No. It's just 10:20:40:80 intersects with the static background 3ish doubling times earlier than 1:2:4:8. Same curve at intersection on log graph.

Elsewhere nails it here - it's because we're not starting from zero cases and zero hospitalisations, but from a static background.  Where the rising component of hospitalisations with a fixed doubling time (deriving from cases that are doubling) crosses the static background of hospitalisation from the component of cases that are not doubling changes, occurring sooner for higher conversion ratios and later for higher baselines.

The higher the static baseline or the lower the conversion ratio, the longer for which the doubling time measurement is suppressed by (biassed towards the characteristic time of) the baseline component, meaning that it takes longer for the doubling time to show through when the ratio of baseline to conversion rate is higher.  Which is why in simple theory the doubling time doesn't encode the conversion rate, but in practice it will for a while as we transition away from the last decay phase.

Something similar is happening right now I think with the fixed baseline component of LFD data, PCR cases and top level doubling times.

So it's not a very useful measure for the next few weeks.

Post edited at 18:20
In reply to wintertree:

> > This is where you broke it. What do you mean by more? It's still doubling. That's not more. That's the same change. You can't just wake up from logarithm space like that. What do I win?

> But, the higher the conversion rate from cases to hospitalisations, the more the actual values have to rise from their current level  - so they have to double faster from their previous level at the end of the last, unfinished decay phase (“starting from a period of level cases” - and hospitalisations) to get to their future, higher level.

This is just going to be one of them things that we're all in agreement on but can't find a way to express it universally....
Hospitalisations aren't exponential. There's nothing that makes new hospitalisations proportional to the number of hospitalisations. They're a straight multiple of the number of cases. So if you flick a switch and cases change from a constant to an exponential growth, then you're just starting doubling from on top of a number that nothing is proportional to. Hospitalisations since you flicked the switch will still behave as you expect. Talking about doubling times when the maths changed part way though is a non-starter.

I'm going to stop just saying the same thing in different words now.

Post edited at 18:36
 wintertree 11 Jun 2021
In reply to Longsufferingropeholder:

>  Talking about doubling times when the maths changed part way though is a non-starter.

Exactly - so measuring them when some of the underlying maths is changing is of dubious utility.  

> I'm going to stop just saying the same thing in different words now.

If UKC supported LaTeX markup we'd get these things dispatched in moments...  I've been trying to make inroads with unicode superscripts and subscripts...

In reply to wintertree:

> >  Talking about doubling times when the maths changed part way though is a non-starter.

> Exactly - so measuring them when some of the underlying maths is changing is of dubious utility.  

One of the twitterers that shows up on Mainwood's thingy now and then seems to have had a punt at unpicking a hospitalisation ratio. @Bristoliver is his name-thingy but I still haven't figured how the **** you read stuff in any logical order on that pile of crap site. Does it get easier if you sign up? Actually don't answer that; I wouldn't anyway.

 wintertree 11 Jun 2021
In reply to elsewhere:

> No. It's just 10:20:40:80 intersects with the static background 3ish doubling times earlier than 1:2:4:8. Same curve at intersection on log graph.

Indeed.  But when there’s the static baseline, the later that intersection is delayed and/or the smaller the conversation factor, the longer a period of doubling is suppressed in doubling time measurements by the baseline.  So doubling time measurements of a compound system can be very misleading - as can simple plots of actuals.  

 Misha 11 Jun 2021
In reply to Longsufferingropeholder:

> Another point that seems relevant to throw in here is that average age of those hospitalised is dropping (bad because young people getting hospitalised, good because means old people aren't) 

I'd like to understand if that's because there are fewer older people being hospitalised, more younger people being hospitalised or a bit of both. I've seen a few media references to more younger people in hospital but not seen actual data.

 elsewhere 11 Jun 2021
In reply to Misha:

> I've seen a few media references to more younger people in hospital but not seen actual data.

https://www.travellingtabby.com/scotland-coronavirus-tracker/

Scroll down to "New Hospital Admissions by Age Group" - peak in ages 25-44.

In reply to Misha:

It's very hard to pick out what's expected and what's because of the variant. In a sense it's what everyone always expected to see when we reopened, i.e. if you protect the old then the small proportion of young people that are hospitalised becomes a bigger proportion of the hospitalisations. Whether it's more people per case, that's starting to get published but it's a bit early to say. Indications are probably yes. I'll edit this if I can find a good source.

Here you go. Page 50 https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/993198/Variants_of_Concern_VOC_Technical_Briefing.pdf

Post edited at 21:21
 Misha 11 Jun 2021
In reply to wintertree:

At the risk of boring people with local data, I've just noticed something interesting in the dashboard demographic case rates heat map. For both England and the West Mids, cases are high in the 10-19 age group and then tailing off up the age groups. Whereas in Birmingham (an area with relatively high cases and relatively low vaccine uptake), cases have been highest for longest in the 10-19 group but over the last few days they have also been high (almost 100) all the way up to 49, with a significant drop off above that (also curiously 30-34 are significantly lower - would be interesting to know if that's due to noise or if it's a 'too old to party, too young to be a parent of a teenager' sweet spot).

A reflection of double dose effectiveness or simply due to over 50s having fewer social interactions? Or a bit of both?

 Misha 11 Jun 2021
In reply to wintertree:

I think the answer to your conundrum is that the 'day 1' number of hospitalisations will be different depending on what the 'day 1' x is. In each example you're starting from level cases but that level will be either 10 or 100. If it doubles in a week to 20 or 200 respectively, the double time will be the same. Now if it goes from 10 to 200 in a week, clearly the doubling time will be a lot shorter but it would only do that if x suddenly increases massively.

Where it gets complicated is when x changes over time (hopefully it will reduce...).

 Si dH 11 Jun 2021
In reply to Misha:

Feels like obvious vaccine effect to me. The trend is stronger than it has been in the past. Haven't read the latest variant paper yet,so cut me some slack if it has a different answer.

I don't think you can read much in to age distributions in utlas yet, perhaps except those that have risen a lot for several weeks now. Those earlier in the wave still have specific weightings towards or away from specific age groups that can only be driven by where outbreaks just happen to have occurred in that region.

Another boring local example here - some cropped/zoomed charts for Liverpool and Cheshire. These are all in strong growth but still at the stage where they have small outbreaks in the midst of large areas with still-low prevalence. As you can see the age distribution of cases at that level is still quite noisy - its variable between these UTLAs with no underlying reason.

Post edited at 21:45

 Misha 11 Jun 2021
In reply to Longsufferingropeholder and elsewhere:

Thanks both. The Scottish hospitalisations by age paint a stark picture. However this would need to be overlaid against case rates by age, adjusted for vaccination status.

PHE document, page 50:

Using stratified Cox proportional hazard regression, there was a significantly increased risk of hospitalisation within 14 days of specimen date (HR 2.26, 95% CI 1.32-3.89, p=0.003), and emergency care attendance or hospitalisation within 14 days (HR 1.45,95% CI 1.08-1.95, p=0.015), for Delta cases compared to Alpha cases after adjustment for confounders (age, sex, ethnicity, area of residence, index of multiple deprivation, week of specimen date, vaccination status and international travel within 14 days prior to specimen date).

Table 6 in the PHE document suggests that around 1% of Delta cases result in hospitalisation with an overnight stay (I'm roughly taking an average of the last two rows, as the last row will have some people who stay overnight for another reason and happen to test positive as well as people who didn't / couldn't get tested before presenting at A&E). This 1% seems to be regardless of vaccination status. Clearly the numbers in the last 3 columns are smaller and hence more prone to noise and the % does vary but it doesn't vary significantly (i.e. it's not like it's 1% overall but 0.1% for second dose plus 2 weeks). In fact the % for second dose plus 2 weeks is higher! Presumably a reflection of that cohort being older. More data required...

Table 18, vaccine effectiveness against symptomatic disease (presumably an average for all vaccines given and I do wonder if this would vary with age as well): Alpha 1 dose 50%, 2 doses 88%; Delta 33% and 81%.

For simplicity, let's assume no one with a second dose goes to hospital and no children go to hospital either (the latter is mostly the case). 20% of adults, so 10m, haven't had the jab yet. A further 25%, so 12m, have only had one so far and of those, two thirds are susceptible, so 8m. Total 18m susceptible pool. 1% in hospital is 180k. Of course not everyone will get it even in an epidemic and the 1% might be on the high side for younger people where most infections will happen. Still, 10s of thousands - comparable with previous waves. Put it this way, it's not going to be in the low thousands...

 Misha 11 Jun 2021
In reply to Si dH:

Yeah, lower overall rates, particularly in UTLAs with relatively low populations (I imagine Sefton is under 100k?) will result in more noise and as you say this probably reflects where the local outbreaks have been most prevalent. Then as it goes it, it will spread out across the age groups, until the vaccine effect and reduced 'socialising' kick in in for those over 50ish. However the national picture suggests less spread by age due to masking, for now.

In reply to Misha:

> Put it this way, it's not going to be in the low thousands...

You'll like this:

https://mobile.twitter.com/JamesWard73/status/1401806768519405569

specifically #s 17&18

 Misha 12 Jun 2021
In reply to Longsufferingropeholder:

Interesting. I like the term ‘exit wave’. Although one of his scenarios suggests yet another wave next spring (later than I’d have expected). The main takeaway for me from all of this is we need more data and that requires more time, so that alone suggests delaying the unlock or at least most elements of it. I actually think we probably need to keep most of the current measures until around this time next year, by which point the global situation should be a lot better. If we do that without having to roll back things like indoor dining, that would already be good going. 

 Misha 12 Jun 2021
In reply to wintertree:

Graph half way down this article. 15k cases by 21 June and 40k by early July. That’s in line with doubling every 10 days as mentioned above. Even if it runs out of steam a bit due to burning through the low hanging fruit of the most susceptible population clusters, it’s going to be pretty dicey even without the unlock.

https://www.bbc.co.uk/news/health-57434493

I wonder whether the 21st would have been palatable without Delta. Back in autumn it seemed that Tier 3 May may have been sufficient to contain the original variant, until Alpha came along. The goalposts keep shifting. I hope I won’t learn the entire Greek alphabet (they’re already up to some letter I’d never heard of before).

Post edited at 01:06
 Misha 12 Jun 2021
In reply to wintertree:

I keep meaning to ask if it’s worth chopping the last day or two off those graphs as the rates tend to drop in the provisional window and that makes it harder to see the ‘real’ numbers.

 Si dH 12 Jun 2021
In reply to Misha:

> Yeah, lower overall rates, particularly in UTLAs with relatively low populations (I imagine Sefton is under 100k?) will result in more noise and as you say this probably reflects where the local outbreaks have been most prevalent. Then as it goes it, it will spread out across the age groups, until the vaccine effect and reduced 'socialising' kick in in for those over 50ish. However the national picture suggests less spread by age due to masking, for now.

Sefton is a reasonable size (275k population I think). Wirral and Cheshire West are both slightly bigger, Liverpool about 500k.  Data for Halton and to a lesser extent Knowsley and St Helens are usually more noisy because they are significantly smaller, so I don't tend to use them so much.

What I was trying to explain is that I think some of this type of 'noise' (in this case I mean biases towards or away from specific age groups dependent on where outbreaks have occurred in that area, rather than general noise obscuring trends) is currently present in all UTLAs with rising cases except perhaps those whose outbreaks are more 'mature' like "the three Bs" (edit - actually most of GM and Lancashire), so I wouldn't read anything in to a particular UTLA like Brum having a lower infection rate in a particular (eg 30-34) age group. The only trend I can see is that the worst affected groups are mostly 15-19 and 20-24.

Post edited at 07:56
 Misha 12 Jun 2021
In reply to Si dH:

Yes, agreed. Never imagined Sefton was as large, thought it was just a few footballers in Formby

 wintertree 12 Jun 2021
In reply to thread:

I'm just reviewing things before writing up the next update this evening.

The India case numbers on the variants page have increased by by 3.4× in the last week, which corresponds to a doubling time of 3.9 days.

Which I don't think can possible be real - neither PCR cases nor PCR+LFT cases have been rising anything like that rapidly, and the number of sequence cases reported exceeds I think the number of PCR cases reported in a week long period ending one week ago, corresponding to the average latency in sequencing.

My best guess is that we're seeing a rise in the rate of sequencing drawing down the latency of reporting, and that more like 1.5 weeks worth of sequences are included in this most recent update.

Now that the India variant is supplanting everything else I think this data source has for now served its purpose for these threads, and I'll put it to one side - for now...

I welcome other views on what I might have missed with this value not squaring off against top level cases.

https://www.gov.uk/government/publications/covid-19-variants-genomically-confirmed-case-numbers/variants-distribution-of-case-data-11-june-2021

Latest VOC report has a bunch of figures on outbreaks worth a look.

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/attachment_data/file/993198/Variants_of_Concern_VOC_Technical_Briefing.pdf

Post edited at 16:39
 wintertree 12 Jun 2021
In reply to Misha:

> I keep meaning to ask if it’s worth chopping the last day or two off those graphs as the rates tend to drop in the provisional window and that makes it harder to see the ‘real’ numbers.

The plots stop 4 days in the past which seems to work well with the way P1+P2 reporting is running off late.  I tend to keep an eye on what the appropriate cut off point for the plots is although I don't bore the thread with the details...  It's all a bit subjective mind you; hence the qualification that the leading edge of the doubling times plots is always provisional.

In reply to wintertree:

> I welcome other views on what I might have missed with this value not squaring off against top level cases.

Not waiting on sequencing any more
https://www.reuters.com/article/health-coronavirus-britain-variant/total-uk-cases-of-delta-variant-spike-to-42323-idUSS8N2LL04V 1

 Si dH 12 Jun 2021
In reply to Longsufferingropeholder:

Yes, this. The genotyping was discussed in the latest technical briefing and the inclusion of that data will make total numbers more representative but means this week's doubling time is artificial.


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