UKC

Today has me feeling a bit like Punxsutawney Phil. I  haven't seen my shadow.

The Cold Shoulder?

Last week, the decay in cases in all 4 nations looked to be slackening off - in exponential terms.  There was a lot of confusion in the data I thought, and a lot of good discussion on plotting 13 - thanks all.   

• This week, it looks like the decay rate is picking up again in England, Scotland and Wales (Plots 6e, 6s, 6w) and all are returning to falling cases.  
• Northern Ireland looks to be remaining level still which is a concern.  A look at the raw dashboard data [1] including the provisional edge not used in these lots looks like it might about to start dropping slowly.  I wonder how tied it is to the situation south of the border? 
• My best guess is that the worst of the cold spell caused a couple of problems - behavioural changes spreading more covid (more inside time, decreased ventilation in buildings and on public transport seem likely candidates), and possibly delays to processing of samples from the period of transport disruption from the drifting power snow in eastern locations.
• Things won’t really be clear for another week, but I hope to see exponential decay continue at rates more like those from before the cold spell in England.
• Scotland had a hairy moment around Feb 14th; it’s long had slower exponential decay than the other nations and this cold shoulder pushed it right to the brink of growth in cases (blue curve touching y=0 on plot 9x).
  • Contacts in Scotland and discussion on UKC make it clear the rules on travel are more restrictive there than England.  It makes me wonder: are restriction and enforcement targeted in the right places to make a difference?  What else is different about Scotland?  There's very little margin for error, and if they continue to have slower exponential decay it's eventually going to pose a problem for south of the border.
• The cold shoulder is visible in the cases plots as the right hand side of the decaying curves levelling off for a week or so before returning to decay.  It's visible in plot 9x as the exponential rate constant rising to a higher plateau for a bit (slower decay, longer halving times) before dropping again to faster dacay on the far right.
  • This drop back to faster decay is in the provisional zone, days for which the next week's data release will refine the estimates of case numbers and exponential ratec constants.  
• So, whilst everything looks good for a return to decay in cases, and with the help of discussions on the previous thread (thanks all), I feel like there's a good qualitative understanding of what's been going on, and what we see in the provisional zone agrees with that understanding, and whilst I'm confident and optimistic, I can't justify dismissing the matter as closed until another week of data showing continued decay is in....  
  • Also, Scotland could really do with getting decay rates a bit baster; this was close to growth for them...
• If this really is as simple as a ventilation issue in cold weather, it's a bloody strong hint about control measures and the couple of months ahead.  For starters I'd remove all ventilator windows from school busses and fit safety grilles instead - especially upstairs on double deckers, and I'd institute spot checks of window opening on schools and workplaces.  It feels like a simple win...

Link to previous thread:

https://www.ukclimbing.com/forums/off_belay/friday_night_covid_plotting_13-...

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

In reply to wintertree:

Thanks, as ever, for the effort you've put into the plots and analysis. But also bravo for the "cold shoulder"! Inspired bit of naming.

In reply to wintertree:

Daily rates and exponential rate constant plots for England.

• Plot 6e - deaths is clearly not exponential - under exponential decay, the curve becomes ever more shallow, never quite meeting the axis at y=0.  This plot looks to be on a linear descent towards zero deaths in a couple of weeks.
  • This is also visible as the red curve on Plot 9 dropping to ever more negative exponential rates (faster decay, shorter halving times).
  • This is now clearly the effect of the vaccine on the older age ranges causing deaths to fall off rapidly there; as the curve approaches 0/day I’d expect it to return to a more conventional exponential decay as the lower but still significant number of deaths from younger people take over driving the curve, and they're still following a more fixed exponential course (see demographic post below).
  • Having said that, vaccination continues down younger ages at pace, and many of the more vulnerable ave been vaccinated at younger ages.   
• It’s notable that the hospital admissions plot is not decaying as sharply as deaths - the implications of this on the time patients spend in hospital (more, or less?) is not clear, which cuts to hospital occupancy - which is where one of the key society wide risks posed by the pandemic.  
  • One interpretation is that for the oldest people the vaccine is tempering the effects of infection but not fully preventing it.  This fits well with the qualitative take of an immune system weakening as the years pass.  It looks to me like hospitalisations are about to start falling faster than exponentially 
• As vaccination works down into working age adults, I'd hope that hospitalisations and cases in those ages are moderated more by the vaccine than for older people.  We'll see.

My take on the efficacy of the vaccination program is that it exceeds any tempered predictions and is the best news we could have hoped for.  Combined with cases returning today, things are looking up considerably, but they only do so whilst everyone continues to play their part in keeping R<1 until vaccination is substantially complete.  The difficulty for those in charge becomes managing the increasing number of good news stories against the need for people to hold the line on control measures for now, in order to give the best shot at more good news coming our way.

In reply to wintertree:

Decay of all measures continues in all regions ( Plots 17, 18.1,18.2).

• The formatting on Plot 17 is really broken now.  I think think this plot has served its time and will be retired for next week, there's no longer a great asymmetry between what I split as the "red" and "blue" regions; not surprising given the prominence of the Kent variant now being spatially homogenous and isotropic. 
• The effect of the vaccine is really visible on the right hand side of the bottom heat map (lots of dark purple) and the bottom exponential rate constants plot which is ever more negative than those for hospitalisations and cases on the right hand side.
• The "cold shoulder" is stronger than some regions than others (the elevated bump in the right side of the blue curves in plot 18.2).  I'm taking this to mean that those of us in the North East are good at opening windows in the cold... 
  • I might try and correlate weather data with the size of the shoulder at UTLA level when another week's data is out.  

The decay of general and ITU hospital occupancy continues in Plot 22.  The absolute numbers of Covid patients in both class of beds remains exceptionally high - since last week we now also have fewer patients in general beds than when we locked down in November for a while.  The regional plot 22r shows nothing very concerning to my eye, although the steeper gradient for London (larger fraction of patients admitted to ITU) still stands out as unexplained.

In reply to wintertree:

Plot D3 - time evolution of rate constants

• This compares the average value over the last 7 days of data with that from two weeks previous.  
• Everything is up except for 90+, presumably from the "cold shoulder".  I hope to see it dropping down again next week.  The rise is greatest in working age adults which is part of what I think it's about transport and workplace ventilation during the cold weather.  Speculation, really.  

Plot D4 - the demographic extrapolation of English cases.

• This is not a prediction. It extrapolates the current situation forwards to March 8th, the confirmed day for reopening English schools.  It's not a prediction as things will change between now and then - vaccination, the weather, variants, behaviour.  It's done as a way of understanding changes to the current situation, by asking "if this goes on as it is...".  
• Michael Hood offered this alternative explanation which I like:  For each date ("from date" - "to date") on the x-axis, the "point" shows the number of cases there would be on 8/3/21 (y-axis value) using the latest data to determine the exponential decay rates at that date (on the x-axis) and maintaining those rates through to 8/3/21 - so every day the curve will approach one "point" closer to 8/3/21.
• The "cold shoulder" and perhaps a slight, general slackening of decay rates have pushed the extrapolation up to ~6,000/day.  I'm hopeful that it'll go down to sub 5,000/day by next week's update as the faster decay fully establishes itself in the data (the analysis of exponential rates is a bit delocalised to minimise effects of noise).  But I no longer hope for much lower than that.

In reply to wintertree:

Yes. Good messaging is required. Let's hope they manage to do that. Just a bit more patience, or we will blow all the sacrifices we have all made.

There are still nearly 15000 people in hospital. That's only just under half the peak. The NHS is still under significant pressure, and far from 'returning to normal'.

In reply to wintertree:

Looking at demographic deaths and cases data.  

Plots D5.lin and D5.log

• The left side of the linear plot shows actual values vs time for various age bins
• The left side of the log plot shows each age bin normalised to the same maximum, and has a log-y axis.
  •  It's increasingly clear on this plot that 80+ is falling faster than 70 to 79, itself falling faster than 60 to 69.  This matches with the age ranked decreasing order of the vaccine program.
  • 50 to 59 is a lot noisier from the smaller numbers and doesn't stick to one place in the precedence.
• The right side of both plots is identical - the exponential rate constants
  • 80+ is just decaying faster and faster exponentially speaking.  
  • 60 to 69 and 70 to 79 had been in a plateau meaning fixed exponential rate decay, much like the corresponding cases data.  I think about a week ago they started the move to ever more negative values (shorted halving times, faster decay) which seems to be the signature of the vaccine.
• By the time the vaccine is established enough to offer significant protection in adults less than 60 years of age, I hope that the case rates will be sufficiently low that the infections aren't there to cause enough deaths to actually analyse and look for this effect.  IMO this is a really positive state of affairs compared to anything we've seen for a long time.

The demographic effects of the vaccine are clear in D1.x with the oldest ages being further left (faster case decay) and further down (faster deaths decay).  The "cold shoulder" is visible as a rightwards trend in the case rate becoming less negative; hopefully that'll end up as a small excursion on the road to faster decay.

In reply to wintertree:

When do the data runs in 22r start wintertree?

I've been watching the South West one, which I had assumed was where Bournemouth would sit (mind you, having just had a look, it might be South East) - but regardless, that's going to be back at the start pretty soon...

In reply to wintertree:

Last data plot for today.  The UTLA Watch list. 

• Plot 16.1 - UTLAs are ranked horizontally by their case rates when we entered lockdown on Jan 5th
• Plot 16.2 - UTLAs are ranked horizontally by their case rates two weeks before the date of the plot

Plot 16.2 gives insight in to more recent changes in the data.

A reminder of the annotations that not UTLAs showing concerning trends:

• ▲ - Cases have risen more than 1.3x above their minimum since the ranking date, suggesting that cases could be on the rise - or it could be noise.  The minimum level is shown by a horizontal line under the data marker.  
  • I've raise the threshold from 1.1x to 1.3x as the "cold shoulder" was pushing a lot of places up that are likely to drop again.
• ▲▲ - Cases have risen above their level on the ranking date
• ▼ - Cases have fallen by less than 0.1x their level on the ranking date

Most of the UTLAs showing here seem to be effects of the "cold shoulder - those that were falling the slowest, exponentially speaking, are sometimes triggering a warning.  I would't read too much in to them this week given the weird behaviour of cases data recently.  By next week hopefully we'll have clear decay for another whole week at the national level and warning flags here could be more worthy of attention.

• A lot of the North West urban areas have been some of the slowest to decay in the last two weeks.  Areas there keep flashing up on these maps.
• North East Lincolnshire has appeared a lot in these plots, looking at the government dashboard case plot, it hasn’t decayed in over a month and now appears to be in gradual growth [1].  Given how areas around it have sometimes been sluggardly this looks increasingly concerning - one way or another, whatever is happening there is going to end up limiting decay over a wider area.  I hope PHE are using their sequencing capacity and expert sequence interpretation heavily in the area.

https://coronavirus.data.gov.uk/details/cases?areaType=ltla&areaName=No...

In reply to AJM:

Sorry, that should be on the plot!  2020-09-01.

In reply to wintertree:

Thanks!

Looks like generally good news especially as far as deaths go, as long as the cold shoulder resolves itself. And if people are tempted to relax, I hope that the good weather many places seem to have this weekend will tempt them to do so outside where potentially the effects are least. It was glorious on the beach today!

In reply to captain paranoia:

Indeed.  It's bonkers.  

Both the baseline for falling cases and the effects of the vaccine on working age adults (faster decrease of cases, hospitalisations falling faster still, hopefully) are going to help, but I imagine a "long tail" of the most dependant hospitalisations continues to build up in the occupancy figures, and is going to take some time to clear, even as the inputs hopefully fall off a cliff starting from mid-March due to the vaccinations.

In reply to wintertree:

Thanks again. It's great how fast deaths are dropping.

My nextdoor neighbour in his early eighties thankfully got home from hospital the other day. He had been in there since early Feb, having tested positive on the 18th Jan before being offered a vaccine. It shows just how much lag there can be between vulnerable people getting infected and either recovering or dying, which puts those death figures in an even better light.

Re: messaging, Van Tam was certainly on point, his "don't wreck it" took all the headlines from yesterday's press conference. I also noticed various local news outlets have started reporting on the flattening rates from last week and my wife's uncle was on the phone earlier worrying that rates are going up in Kent (they aren't.) So I think the less informed part of the public currently have a pessimistic view of the reality in their heads, rather than the opposite.

In reply to AJM:

NHS regions are quite confusing (I especially like the fact that half of Cumbria is in the North East region). They also appear to have changed over time but for now you are in South West. The best map I've found is half way down this page.

https://www.hee.nhs.uk/our-work/population-health

In reply to Si dH:

Thanks Si. In which case our hospital occupancy looks (on the assumption the squiggle is basically November) like it's back in October and closing in on late September, or something like that...

In reply to AJM:

Yes although I think most of the squiggle is probably Bristol as they had very high case rates there just as we entered lockdown 2.

If you want to see a better picture of current occupancy in your local area it would be more informative to look at your local trust if you know it. There are three trusts with "Dorset" in the title and all seem to have some covid patients.

https://coronavirus.data.gov.uk/details/healthcare?areaType=nhstrust&ar...

(Edit - although the trust level data is slightly older)

In reply to Si dH:

I must admit that I only really listen to the briefing now properly if Van Tam is speaking.  He seems to nail the real issue every time. 

In reply to wintertree:

Yeah. I have heard that some poor souls are spending weeks, or months in ICU.

In reply to wintertree:

Thanks, generally seems good news. 

I'm worried our approach to schools reopening fully is a bit "just carry on like we were in December and hope for the best"

Schools have been open in Italy for a few weeks and cases now seem to be taking off there worryingly quickly.

In reply to Si dH:

One of the first bits of advice I received, when moving to the West of Cumbria was " If there is anything serious wrong with you, get someone to drive you to Newcastle" To be fait I've had a couple of minor ops. at Whitehaven and both went really well.

In reply to RobAJones:

The data ramblin dave posted a link to on the previous thread shows much lower car usage than the November lockdown.  It’ll be interesting to what fraction of that difference fades when schools are back.

Schools will only be back for a couple of weeks before the Easter holidays, and when they reconvene the vaccine program should be ramping up its effect of moderating transmission rates in working aged adults; so hopefully we’re never going to see a large or sustained rise in cases...    Where as they went back in winter not spring in Italy, and with much less vaccination.  

Make sure you open all your windows though!

https://www.ukclimbing.com/forums/off_belay/friday_night_covid_plotting_13-...

In reply to wintertree:

shows how much attention I've been paying, that hadn't registered with me

and when they reconvene the vaccine program should be ramping up its effect of moderating transmission rates in working aged adults; so hopefully we’re never going to see a large or sustained rise in cases...    Where as they went back in winter not spring in Italy, and with much less vaccination.  

Winter in Rome v Spring in Cumbria? but they did go back with case rates similar to ones we had a week or so ago,  hopefully ours should be significantly lower

Don't think Mrs J will allow me back, I'm hoping the weather allows me to see the benefits of all this lock down training.  

In reply to wintertree:

Regulations for German schools are to open windows for 5 minutes every 20 minutes.

In reply to wintertree:

Do you have a view on the recent changes in the proportion of PCR Vs lateral flow and the impact of this on the trends. The last week showed an increase in the number of tests being performed but a drop in the number of PCR, suggesting an increase in the number of lat flow tests being used. If the PCR is still the gold standard for those showing symptoms and the lat flow used for asymptomatic community testing. My limited stats experience would say it's important to maintain the same sampling regime to make the trends comparable. Would be interesting to see the trends if you were able to just use the PCR tests, but I can't see a way of stripping this data out.

In reply to wintertree:

Hospital occupancy is obviously a key metric and its great to see that it’s still falling quickly although at a slowing rate. In England it reduced by 16k in the last four weeks, 7k in the last two weeks and 3k in the last week. Assuming it keeps dropping by an average of 2k a week in March, we would be down from 12k currently to about 4k, which for context is the level from mid June. Not bad, considering.

I think there are two big unknowns. First, the impact of schools. We should have an initial indication in 3 weeks’ time. Second, the level of vaccine uptake in the under 50s and the impact of that on transmission. We won’t know that till the summer.

The reason I say not till the summer is that it looks like they can crank through the 12m remaining people with health conditions and over 50s by the end of March. From that point the number of people needing second jabs will be substantial - roughly equal to today’s vaccination capacity. So whether there would be any significant scope to vaccinate the younger age groups depends on whether they can increase vaccination rates (primarily a supply issue). Otherwise it would take a good 10-12 weeks to administer second doses, so the under 50s won’t get started in earnest till early June. So I suppose the third big question is how much this could be speeded up. To be fair, even if they keep going at the current rate it would be an amazing effort. 

In reply to Moomin.williams:

With fewer cases you'd expect to see fewer people noticing symptoms and going to get a PCR test. It'll be interesting to see whether we get a change in the rules of engagement as PCR capacity becomes underused.

There's going to be a huge jump in LFTs used as schools go back, but whether they'll be reported is less certain (how many are going to bother logging 2 a week and for how long?).

Tl;dr That ratio of tests will get more wonky over the next few weeks.

In reply to Longsufferingropeholder:

I think wintertree has done a great job of cleaning what I always thought would be data too messy to get any reliable trends from. From the start my guess was that the cases data would be too greatly impacted by the changes in testing regimes and people's behaviour wrt getting tested, I was too pessimistic.

Occasionally my desire for cleaner data gets the better of me and I start pondering the what ifs. Or maybe just hoping the situation is even better than the data is showing and the current flattening in case rates is being caused by the change in test ratio.

In reply to captain paranoia:

"There are still nearly 15000 people in hospital. That's only just under half the peak. The NHS is still under significant pressure, and far from 'returning to normal'."

Don't forget current ITU figures as a percentage of the peak level are, as expected, quite a bit higher (the link is in the Indie SAGE data presentation but they didn't show the data this week)

https://www.icnarc.org/Our-Audit/Audits/Cmp/Reports

The Friday Indie SAGE presentation is here. There is a welcome focus on inequality (putting things much better than my own clumsier attempts on these threads) and a critical look at the government plan (could do much better):

youtube.com/watch?v=6YuU52uqzmM&

Response to the government roadmap 

https://www.independentsage.org/strategy-for-covid-19-maximum-suppression-o...

Wintertree can have another pat on the back as Indie SAGE haven't yet spotted the cold shoulder.

In reply to Moomin.williams:

It's easy to get carried away with media hype and overblown perception of risk where we currently see lots of people (outdoors, in cars and in shops). The primary factor right now in population risk is key workers who cannot afford to self isolate: either not getting tested with clear symptoms and continuing to work; or testing positive and continuing to work (not self isolating as required). Some major risk might move to secondary schools next Monday if the grand reopening fails to run well (quite likely in my view). Not focussing enough on inequalities might come back to bite us all.

In reply to Offwidth:

I think that statement applies on a global level as well.  European countries stockpiling unwanted (at the moment) AZ vaccine, when it could be used on front line staff in less affluent countries, doesn't seem the most effective strategy.

In reply to RobAJones:

I agree on the global point but can forgive the UK side of that a bit more than most. We are the hardest hit of the big economies (due to government incompetence) and getting the UK population vaccinated quickly frees resource and vaccine to help the poorer parts of the world. Providing of course the plans to run at higher covid levels than ideal and with very leaky borders doesn't put us back to near square one with some nasty variant.

In reply to Offwidth:

Leicester cropped up on the #13 thread. The mayor claimed on the East Midland version of the 10:00am  BBC1 Poliitics show today that some local mass vaccination centres were at times down to 10% capacity: hence more flexibility is required to speed up vaccination and help GPs with that, in communities like those in his city.

In reply to Offwidth:

I'm certainly not at risk of getting carried away with any perception of risk. I think the need to stay at home has been overblown, being outside has to be about the safest place you can be. The government's statement that schools will be last to close has forced them to unnecessarily shut down a lot of safe outdoor activities. This has also scared a lot of people into staying indoors watching too much daytime TV!

In reply to Moomin.williams:

It's hard to reach a view because as far as I can tell, insufficient information is published to understand everything about the position of LFTs in this data.  I assume (but don't know) that the widening gap between P1+P2 sample counts and PCR tests run is the increased rollout of LFTs.

Yes; given the lower sensitivity of LFTs when deployed en mass and asymptomatically, and given the falling prevalence I wouldn't expect them to be making much difference to the case number.  It's hard to know for sure but I'd put the number of cases they generate right now at a hard upper bound of ~1,500 per day and a likely value of perhaps <200 per day.  Crude estimates though.  I'm also not clear at all on when a positive LFT is followed by a confirmatory PCR test - as with the Liverpool trial - and when it isn't.

I don't see anything in the API documents to suggest there's a way of operating it, but the list of metrics hasn't been updated since August before the LFTs came online, and I know it's incomplete as I've hit on at least one undocumented key...

The way I see it, it's a matter of timescales.  Changes in sampling (ongoing LFT rollout, supply vs demand between peaks and toughs of our "waves") tend to be progressive rather than sudden, so the longer the separation in time between two dates, the more dodgy the comparison of cases/day; over the timescale of a week or so I'd expect changes in exponential rates are largely "real".   But cases remain a measure of limited certainty unlike hospitalisations and deaths.

There is only so much give that "better testing" can have in terms of increasing the number of cases, masking the fall of infections.  The rollout across schools in just over a week might confuse things for a bit.

In reply to wintertree:

Approximately 20% of cases in Liverpool continue to be found through lateral flow according to the link below. This number has stayed fairly steady right through from last November (I don't have ready figures but I don't think it has ever been above 30% or below 15%, despite big case rate swings from under 100/100k to over 1000/100k and back down again in that period.)

https://liverpool.gov.uk/communities-and-safety/emergency-planning/coronavi...

The LF testing rate is ROM 1% of the population of the city per day. 

Vast majority of the above should be confirmed through PCR though, and as I understand it the cases recorded on the dashboard are cases, not tests, so where someone tests positive for LFT and PCR they only show up once.

Across England nationally the LF testing rate ramped up in January and since then has been up and down but broadly in the range of 0.5-0.7% of the population per day. So if you were to extrapolate, you might expect 10-20% of national positive cases to be being found through LFT. Whether the extrapolation is valid, too low or too high I'm not sure! I would expect the LFT positivity to be higher in people going to work or school* but equally I think quite a few LFT cases just visiting local test sites voluntarily are probably people who have things like sore throats or snotty noses and don't have the right symptoms to "qualify" for PCR, ie I do not think it's necessarily a completely truly asymptomatic selection.

* Interesting point although I don't know if the data is stored anywhere - during the Liverpool LFT trial stage the council were tweeting results updates daily and provided separate results for Liverpool residents Vs non Liverpool residents (ie, people who travelled in to the city for work.) The latter were consistently between 50-100% higher positivity.

In reply to wintertree:

I was wondering by the way, is it possible to do a sensitivity analysis on any of your cases graphs to including an extra 2 days of data in your curve fitting? Ie, use data up to and including (current date minus three days) and removing a couple of the oldest days in the range used by the calculation?  For the moment the dashboard data is essentially complete down to regional level by that point, because of the testing programme getting faster. This is consistent across all English regions and has been for several weeks at least. It's possible that one or two UTLAs take longer but if so they are in a small minority. I think at times of significant uncertainty/ change in case rates it could be really useful to see how sensitive your leading edge trends are to using the extra data in determining the exponential rate. And obviously it could give slightly earlier warning of problems. However I'm not sure if the dashboard data download includes anything/everything you need for those "incomplete" days.

In reply to wintertree:

I hadn't considered the "repeat" test confirming a positive LFT with a PCR. I'd also assume that in that case there would be a single case recorded but possibly 2 tests, but we know how dangerous assuming anything can be!

The use of LRT in schools feels more like being seen to do something to make schools safer rather than actually anything supported by the testing science. The high false negative rate for LFT when self administered has got to render the test at just slightly better than useless.

In reply to Moomin.williams:

I'm pretty sure they have said that cases are individuals, regardless of how many tests they have had.

In reply to Moomin.williams:

I've said this before, but it's always worth circling back to...
Yes, one LFT is pretty close to useless. But two a week is a lot harder for an infection to hide from. Should help. Maybe not a lot, but it should help.
A shotgun pellet is pretty likely to miss.....

In reply to wintertree:

Been looking at ‘cases in last seven days’ for Grter Manc, where things are all over the place but generally looking OK. We’ve had 5 (out of 10) LAs show increases in the last few weeks, most recently Stockport (least deprived area) whereas Oldham (most deprived) has the second lowest numbers with a trend of -16%. In fact, 6 LAs are at about the England average trend but all our case numbers are higher than the average. All going in the right direction though. 

In reply to Longsufferingropeholder:

Yes, this.  PCRs would be fairly useless for this application because you need the result straight away. Schools are somewhere we are going to open anyway...so the addition of two LFTs per week per pupil can only provide benefit. If their family is all tested too, the chances of missing positive cases is further reduced, so this looks like good policy to me. It's about having defense in depth, once you have already determined that community case rates are low enough to open and have put in place baseline measures in class. Need to ensure the parents don't see their own LFT results as a free pass to ignore restrictions - but that's about communication.

In reply to Moomin.williams:

As Captain Paranoia notes, the reported numbers are cases - which are positive test results de-duplicated across people - some also have more than one PCR e.g. perhaps when being tested then admitted to hospital.

I think in terms of directly reducing the amount of infection in a school at the point of use of a test, you're probably right.   But as Longsufferingropeholder says, multiple tests improve things somewhat.  More generally, if the LFTs remove 10% of infectious people it might not sound like much but in removing that person from circulation and starting contact tracing, it goes on to speed up the exponential decay process, and small changes to exponentials have big effects...  

In reply to Si dH:

Thanks for the detailed calculations and notes on the Liverpool case; so my estimates are in the right ballpark probably...  It's a shame more isn't published on the results by test type.

I've not been reassured by the people I've spoken to about their LFT results and how they interpret them - despite there being good, clear comms. I think people are sick of restrictions and will look for loopholes to assuage their guilt rather than to actually make it safer, hence I worry that mass LFT that's not well targeted in some sense is potentially playing with fire.

In reply to wintertree:

Things are changing rapidly in the deaths data in the two days since I posted plots D5.  It looks like the vaccination is now showing through strongly in the age range 70 to 79 as it has been in 80+ for a couple of weeks, with the halving time for 70 to 79 no longer being a plateau (indicating consistent exponential behaviour) but getting ever shorter indicating that the deaths are falling faster than exponentially.  

I have expanded the size of the next lowest age bin to ages 40 to 59 inclusive, as there are few enough deaths here that the analysis becomes noisy. 

In reply to Si dH:

I'm generally cutting off at the point where including another day would introduce a consistent trend for a false decay; I think the cutoff has crept forwards a day at some point in the last few months but I'm not keeping a lab book on all this stuff so I can't say when I did it.  The exception is the demographic and UTLA level cases data I do which comes from the "unstacked" download which has a 5-day cutoff built in.  If I find the time I'm going to look in to demographic downloads from the API to see if they're a bit more current.

Edit:  Update of new plot I did on the last thread when discussion the slowdown of decay with Wicaomi.   It looks like cases are back on to the pre-cold decay of ~ 0.7x per week.  One way of looking at the cold is that it has introduced a net decay of 3-4 days in to the decay of case numbers. 

This plot uses the API which answer the question "Does the API return more recent data than the manual downloads?".  The answer is no, no it doesn't.   On

In reply to wintertree:

Ok, I didn't realise you had brought it forward a day from the usual 5-day cutoff. What I was explaining was that you could now bring it forward two days pretty confident that there would be no false droop (at least in England, I haven't looked at the data elsewhere), but if you have already brought it forward one day I don't think the sensitivity check to one extra day would be worth it.

Lots of good news in the press conference today as long as this unknown person with the P1 variant doesn't do a super spreader on us. Hancock mentioned off hand that they are now sequencing a third of all cases, which I think is really good.

Cases continue to drop fast too. I think it will get below your old 5000/day hope before 08/03. Weather seems to be really helping. Graph for England as of today - there won't be much more movement in data for the 26th and before.

In reply to Si dH:

Baring any more unexpected surprises I think so.  

Today's data update has it the extrapolation just over 5,000 but as discussed up thread, the demographic data cuts off (from the source) 5 days in to the past and things are looking better in the most recent data.  Although there's not so many days left for the difference in rates to play out over...

Edit:  I've been staring that my latest cases plot and wondering if we're about to see the effects of the vaccine kicking in significantly.  It won't be the apparent plunge to y=0 as it nearly is with deaths as the demographic spread of cases is not so heavily aligned with the vaccination as the deaths are, but we might see a more negative exponential rate...

... or it could be the weather.

In reply to Si dH:

It's concerning that a case slipped the net but that will always happen since there's no perfect net.

I'm relieved that we have one (known) seed case of it at a time when R<1*. Could have been a lot worse, in many ways. Still could be, I guess. Or it might not affect vaccine effectiveness. We don't know. 

* - I hate that metric as much as you but can't be bothered to write the long version.

In reply to Longsufferingropeholder:

Given that we're catching perhaps half of infections, that means odds are on another importation event and downstream chain of infections.  I'm taking some solace in the fraction of cases now being sequenced; presumably that's going to rise to near unity soon enough with the decay in daily case numbers.   An infectious chain shouldn't be able to hide for long assuming enhanced contact tracing goes up as well as down.

Is it <1 for this variant though?  

Infection and concomitant naturally induced immunity will be concentrated within the sub populations that are most at risk of catching and transmitting the virus, and after the most recent wave this is going to be making a big difference I think to transmission within those groups.  Add in vaccine induced immunity in most front line social care and health workers and I think immunity could be making the difference between decay and spread.

So, if the new variant evades immunity enough, it could well turn from decay to spread.  Pre-print on the subject due out tomorrow... https://www.ft.com/content/51cf718d-e701-4292-a9dd-dd36c1b1c5ea

In reply to wintertree:

I did wonder that. But then after a second thought it's really a coin toss in this case whether that individual infects more or fewer than one other person. Given the state of society you'd hope <1. Given they couldn't be arsed / choose not to fill in the forms properly you might think >1. Well find out.

As for transmissibility, we're already awash with N501Y, and i didn't think E484K conferred any particular transmission advantage (just possible immune escape), but of course I could have the wrong end of the wrong stick there.

In reply to Longsufferingropeholder:

I don’t think there’s any reason to think it’s more transmissible in a susceptible population.

But a lot of people in the UK have some natural immunity, and that’ll be concentrated in the sub populations most likely to catch and transmit the virus including during lockdown.   I think the concentrated, naturally acquired immunity could be moderating R a lot in these sub populations, and therefore at the top level given their role in driving cases.

If the Brazil variant partially evades immunity, it could be more transmissive in a population with some immunity, especially when that immunity is concentrated in the people most likely to catch and receive the virus.  

There doesn’t seem to be a consensus on the cross immunity; I imagine the Kent variant might produce less weakened cross immunity than old ones, which is something.  

Unless we properly harden the borders, importation events will keep happening, and one way or another I suspect we’ll find out re:cross immunity.

In reply to wintertree:

I think this is still the only thing we disagree on: You can't harden borders enough. There'll always be one no matter what you do, and the cost overtakes the benefit.

Lower case numbers and near universal sequencing should mean we can keep on top of it though.

In reply to Longsufferingropeholder:

I certainly recognise that hard borders aren't a long term solution and aren't viable for long.  

I recognise that they're not perfect - NZ have just put Auckland back in to lockdown after another case slipped through, and their measures are much more robust that ours.  

On the other hand, I think a harder border than we have is sustainable for the next couple of months, which increases the odds of us getting to...

... before it all goes a little caca.

As always, it's trading off certain costs - financial and political - now with future costs that are impossible to predict accurately, and that probably won't occur if we take the expensive measures now.

I'd be happier if we had more stringent measures on travellers from all countries, as testing let alone sequencing is far from universal on many of the countries of origin people are coming from.  

I'm beating a broken drum, and so far it's not gone disastrously wrong, so that's something.   The SA variant continues to pop up in the updates in low numbers.

Interestingly the latest report on variants [1] has 3 cases "not yet linked to individuals" including on a Brazil variant, and has this footnote on the SA cases:   Excludes 36 VOC-202012/02 sequences reported from devolved administrations and 22 sequences not yet linked to individuals and under further follow-up (12 confirmed, 10 probable).

This is panting a picture that quit a lot of samples don't tie back to individuals.  Sequencing isn't much use if you don't know who the sequence came from.

https://www.gov.uk/government/publications/covid-19-variants-genomically-co...

In reply to wintertree:

There was brief discussion of this topic in the press conference today. I was only half following but they said the numbers of tests for whom they had lost the identity was very low* for general community testing, but slightly higher for home tests and from surge testing where they were delivered door to door. I think they said this slightly higher rate was 0.3%.

* It doesn't necessarily mean someone didn't provide their details. My son was one of this small number (maybe bigger then) when he tested positive last summer. They lost somehow his registration (IT problems were mentioned) but fortunately we managed to reconnect the dots as we had kept hold of his test card and eventually managed to re-register him for that test ID over the phone, 5 days after he had taken the test (this took us three attempts to test/trace on the phone before they listened properly.) Another day later we got a text saying it was positive. End rant!

In reply to wintertree:

Also there's hope in the presumption that vaccine immunity and recovered immunity are different; in the former there's only the spike protein to go on. In the latter there's a couple of other targets, notwithstanding the apparently shorter duration (??? confirmed or postulated?) of protection. 

Not sure if it helps much or not, but if the lockdown R is dominated by the doorhandle-lickers* then maybe cross immunity is likely to be higher??

Edit: this post is bollocks. cite{Manaus}. I should go to sleep now.

* - need a better term, since this implies recklessness where in reality it's a blend of that and occupation.

In reply to Longsufferingropeholder:

Indeed, but the receptor binding domain of the spike seems provides epitopes for neutralising antibodies that stop cells from being infected on account of binding first and preventing receptor binding, even before the clean up process of antibody tagged stuff happens; I don’t know if the membrane fusion protein can be neutralised or not? The capsid etc probably won’t provide neutralising antibodies but will help T-cells spot and mop up infected human cells; so with all the RBD mutations it seems naturally acquired immunity might moderate severity of illness but not prevent infection.  Where does transmission fall on this?  If it can still happen, what you might get is a lot of people in the high risk of transmission, naturally immune sub population having mild, non symptomatic but transmissive infections which rapidly spread to everyone else.   I think that’s a worst case scenario made by combining several worst case assumptions, so probably not realistic.  Still...

Quite.  My worst case scenario above goes away once most people have been immunised - then any new variant is hopefully tempered by partial cross immunity and spreads without a follow on wave of serious illness, hospitalisation and death, and new immunity is generated tracking the variants.  With elimination being very unlikely this is probably the best we can hope for, along with improved therapeutics.  Until immunisation is complete, soft borders let variants in that at worst could really mess this transition up.  I’m not sure how accurate my interpretation of all the immunology is mind.

Those who can’t isolate due to work and/or caring responsibilities is my go to description.   

In reply to Si dH:

Thanks for that insight.  You’d think they’d print a matching QR code on the sample return paperwork...  

In reply to wintertree:

Seem to recall reading yesterday that one of the Brazilian cases was someone travelling from Brazil to Paris to London to Scotland. Perfect example of insufficiently hard borders... I wonder whether the test sample with incomplete details was someone similar - perhaps they got swept up in surge surfing so wanted to make it look like they were playing along but didn't want to be traced because they'd been a bit naughty. Although I'm probably being too cynical and it was a bona fide mistake...

The 'cold shoulder' is coming through in the 7 day average cases graph on the official website, particularly for the 'date cases reported' graph. Nice to see but, as ever, you were ahead of the curve (literally)!

Hopefully the nice weather last week will translate into fewer cases being reported this week and we'll squeeze the case count as low as possible before the inevitable uptick with the schools going back. Anecdotally, there were definitely more people milling / sitting around last week (I head out for a lunchtime walk most weekdays including weekends), including in small groups in some cases. The way I see it, they may be breaking the rules but it's better that they gather in small groups outdoors than indoors.

Which makes me wonder whether that's the other reason nice weather helps, in addition to public transport and workplace ventilation which you've pointed out. Small gatherings which might otherwise happen indoors are more likely to happen outdoors. People have a natural urge to socialise with friends and family but when the weather is nice they are more likely to do that outdoors, partly because they know that it's safer and partly because they just want to be outdoors anyway. The more I think about it, the more the whole 'keep crowds away from beauty spots' thing is a bit silly (especially if people have arrived by car as opposed to public transport), as if you chase people away from outdoor environments, some of them will just congregate somewhere indoors instead. Of course it's not as simple as that (nothing Covid related is) but it's a consideration... 

In reply to wintertree:

I’ve been keeping a track of the demographic on hospitalisations in England since the start of the year and just over the past two weeks there does seem to be a vaccine effect kicking in. The rolling 7 day % of hospital admissions for the 85+ age group has fallen from just over 20% to just over 17%. For the 65 to 84 age group it’s fallen from just over 39% to just over 36%. In the same time the % for the 18 to 64 age group has risen by about 6% to just over 44%. Not huge changes yet but certainly encouraging and backing up what was said at the government briefing today.

On a sad note I read in my local paper today that my barber of many a year had died recently after contracting coronavirus in hospital. He was 78 and a great local character of Italian origin who had not long retired after running his tiny one roomed barber shop in town for over 50 years. He’ll be sadly missed by his many friends. I guess it just shows that even with the vaccination effort people should still be wary and keep their guard up for a while longer.

In reply to wintertree:

Thanks again, really positive news all around and well done for spotting the ''cold shoulder''. 

I didn't watch the briefing last night as I just assumed it would be all about the new variant of concern, I have just caught up on the summary and yeah, it's all really positive actually, the vaccine news is really great and matches up with the snippets of trials data that have been leaking out over the past few weeks. I wonder how Europe are feeling about it all now, very interesting how that story died out as quickly as it came...

Also, lots of hope in this thread from the Furry Molecular Biologist who has ties to Moderna, with regards to P1 and vaccine effectiveness.

https://twitter.com/sailorrooscout/status/1366075542097825792

There are also reports now of vaccine supply considerably increasing over the next week or two

https://twitter.com/jburnmurdoch/status/1366471684493086727

I am feeling quite optimistic about it all now...

In reply to Punter_Pro:

First million jab day isn't far off now

In reply to thread:

Incidentally, despite great news in the latest case data overall, rates in Lincolnshire are still holding surprisingly flatish right up to 26/02. I don't think their weather was much different from elsewhere last week. Their rates are mostly modest but it does seem strange.

In reply to Si dH:

Something about that region, isn't there? One of the first (If not the first? Memory hazy) into local lockdown, and always leading the way into higher tiers.
Is it just a stubborn region or is there something going on? And is it possible to tell the difference?

In reply to Si dH:

North East Lincolnshire and now Lincoln seem to have a real problem with cases either level or rising.  Kingston upon Hull also looks to have levelled off.  These are places that have been popping up in plot 16 a lot of times.  It’s definitely not just slow falling and noise.

In reply to wintertree:

Seems to have plateaued near me, but it's actually a small rise in 0-59, and continued fall in 60+

In reply to Punter_Pro:

That’s not quite how I read the Furry link.  

Cross immunity is no longer neutralising (as the key binding sites where antibodies can block virus/cell binding have changed) so infection is no longer prevented, but the vaccine does prevent serious disease.  Neutralising activity is needed to contribute to R<1 against weakening restrictions; with both naturally induced and vaccine induced immunity likely not neutralising this variant, it has the potential for more spread under current conditions.  The vaccine tempers its effects, moderating illness, but that’s only of use to vaccinated people.

Once the vaccine is rolled out, it’s probably okay - hopefully it’ll spread with mild illness, and will update immunity as it goes, and new vaccines can come out.

But until we get everyone immunised, there’s potential for things to badly wrong with a strain that’s changed enough not to be neutralised by current immunity.

From the linked pre-print: “To understand the impact of globally circulating variants, we evaluated the neutralization potency of 48 sera from BNT162b2 and mRNA-1273 vaccine recipients against pseudoviruses bearing spike proteins derived from 10 strains of SARS-CoV-2. While multiple strains exhibited vaccine-induced cross-neutralization comparable to wild-type pseudovirus, 5 strains harboring receptor-binding domain mutations, including K417N/T, E484K, and N501Y, were highly resistant to neutralization.” 

The Twitter poster starts with “We know the vaccines can neutralize this variant” which doesn’t to my reading tally with the pre print they link.  I can’t see all the replies but it looks like they get bit of a kicking and backpedal to “What I try to express is that the vaccines are effective even with decreased neutralization.”  Which is quite different - still really good news but it potentially changes everything when it comes to the risks on a partially vaccinated population.  Thankfully we’ll all have been offered the vaccine soon enough.  If the Brazil variant gets loose and rises to prominence here, we could worst case see a lot more mild and asymptomatic spread in the vaccinated and so a lot more infection and death in the unvaccinated - with the potential for another healthcare overload if this is allowed to happen to fast.  I don’t think it’s going to go down well asking people to continue restrictions because of people who have chosen not to be vaccinated.  Decisive and ugly, so I’m hoping the reality turns out a lot better than my attempt at bounding the worst case.  Missing from the pre print is an understanding of the scope for antibody neutralisation of the ‘M’ protein which could mean naturally acquired immunity is still neutralising.

The political side to all this is well beyond my ken.  

In reply to wintertree:

https://www.dw.com/en/covid-france-allows-elderly-to-get-astrazeneca-vaccin...
"Only 240,000 of the 1.45 million AstraZeneca doses that were delivered to Germany had been used by February 23."

Probably needn't have bothered with all that unpleasantness over "best reasonable efforts" then.

In reply to wintertree:

This is, so far, a thankfully, splendidly, wonderfully, magnificently, unexpectedly, gloriously low number. Sure, we're less likely to see 90% all the way down the ages, but if the trend of higher-than-we-ever-imagined continues then this is should be moot.
'Haven't been vaccinated yet' is worrisome for the set of reasons you laid out, so.... jab faster!
 

In reply to Longsufferingropeholder:

This is a problem of their own making potentially precisely because they couldn't get the supply they wanted. I think the "leaks" about it's lack of effectiveness were a deliberate attempt to deflect from the supply issues but has resulted in low uptake. 

In reply to joem:

Yep. This. Couldn't have seen this coming. Only we all did. Weeks ago. Forget which thread it was now but it's one of those times, as happens all too often, when UKC was like a junior school nostradamus.

In reply to wintertree:

Not sure whether it's old news or new news or just plus ça change, but there's a new new variant, b.1.526 or "The New York variant" as it will tediously be known, for us all to shit the bed about.
How's that red list looking?

In reply to joem:

So it seems, although (much to my surprise) it seems public buy in to vaccination has been much lower in France and Germany for a long time.

As problems go, it’s a problem for the UK has well; the pandemic phase needs closing out globally to get this under control.  Having near neighbours, with whom we have a lot of travel, unable to vaccinate effectively is a problem for us.  It’s not encouraging in terms of global vaccination if developed European nations can’t accomplish vaccination to significant levels.  Still, it’s early days in the vaccination era.

In reply to Longsufferingropeholder:

Yes I remember the discussion. 

In reply to wintertree:

It's a problem, but it's not without its silver lining. Turning to self-interest, the first country back to work is the 'winner'. If we can be paying taxes for a few months when the rest of Europe isn't, then we're, relatively, better off? Less worse off. Let's go with less worse off.
That's only a good news story for a very short while though; it's not long like you say before we really need our neighbours, and soon after the world, to get back to work too.

Europe should, really, be able to get on top of this soon and stay at most a couple of months behind, you'd think. Won't be long until that's a dot in our memories.

In reply to Longsufferingropeholder:

Death traded in his white horse for a white airplane some time ago.

The news headlines on the Brazil variant are generally not optimistic either.   

There’s a lot of preliminary work out there that says “worry”.  I hope the fast vaccine roll out blunts any worst cases consequences for the UK but if something resembling the worst case is true, this is really bad news for many places - especially as I suspect like us they’re implicitly relying on naturally conferred immunity in the sub populations most at risk of spreading the virus.

On the bright side, none of the variants have developed a much higher lethality, and therapeutics continues to come out of clinical trials into use, with more in the pipeline.

In reply to wintertree:

Various thoughts...

Plot D4 - have you got a next date in mind once we get to 8th March?

Plot D4 could be done for hospital admissions and deaths as well - although I'm not sure if there's any point/need except that hospital admissions and deaths are fairly reliable quantities that have been fairly consistently measured, whereas cases seem (to me) to be much less reliable with all the different testing regimes, and with people less likely to go for testing as overall numbers decrease, etc. Obviously, the main benefit of looking at cases data is that they precede the other measures, so they're able to give earlier warnings, even if the lesser reliability/consistency means they're more prone to false alarms.

And, assuming no nasty variant springs out on us (like P1) and either wins the race against vaccination or totally avoids any existing immunity, should we have a sweepstake on when most (*) of your graphs become "useless" (beyond saying "nothing significant to see") because the numbers have dropped down so much that they're all hidden by noise - a day to look forward to even if we can't get to total elimination 😁

(*) - eventually, only the "number" v date plots will remain, all the rate constant, rate doubling/halving, lissajous graphs will close in on their origins.

In reply to Michael Hood:

The next nominal English unlocking date - March 29th.

It can, but hospital admissions have a coarse granularity - too coarse to accommodate the vaccine roll out and so I think extrapolations of that will be less accurate, and I hope that one for deaths will be irrelevant within a couple of weeks.

That's potentially when a really clear, best possible analysis becomes most important because that's when the misinformation people start to use the confusion caused by the signifiant noise in the data to push dangerously wrong nonsense with a veneer of plausibility.  It'll be different this time round because we should be safe to drop most restrictions - I'm not convinced that it's going to be safe (in terms of healthcare provision) to truly drop them all this summer.  So it's hard to see what they're going to use misinformation to argue for, over and above what we're doing already, my suspicion is that it will turn to an anti-vaccination effort intended to undermine the need for future booster rounds.

But there won't be much purpose to weekly updates; perhaps a set of plots a month from July onwards unless anything starts to change...

In reply to wintertree:

 

I reckon the anti-vaxx movement will become increasingly insignificant:  Johnson and his govt will be viewed as heroes because of the vaccine and the gap between UK death rates and other comparable Euro countries will get smaller. UK hospitals will have a high proportion of non vaccinated Covid patients.  Johnson and the media  will use his increased credibility and Covid in none vaccinated people to press home the ‘vaccines are our way out’ message  

In reply to mick taylor:

They'll pivot to opposing stuff like the data harvesting potential / privacy issues / traceability / liberties infringement that potentially comes with the vaccination certification schemes that'll be implemented to permit travel.
Or at least I'd rather their energy goes into pissing into that gale than some of the other things it could.

In reply to wintertree:

With all plots waiting on the side-lines to be resurrected on any upturn in numbers.

Hopefully, the numbers will all be so low, and the direction of travel so well set in the right direction, that any misinformation will be insignificant. I suppose the danger is the "it's gone so we don't need to vaccinate any more" misinformation.

In reply to wintertree:

Can't thank Wintertree enough for this work.

Re the Brazilian variant, speaking to medical colleagues more in tune with this matter there seems to be a lot more concern in the background that is coming through in the media. 

Graph on page 2 of https://www.thelancet.com/action/showPdf?pii=S0140-6736%2821%2900183-5 is how one colleague suggested 2021 will go in the UK.

A paper referenced by the BBC today ( https://github.com/CADDE-CENTRE/Novel-SARS-CoV-2-P1-Lineage-in-Brazil/blob/... ) added to a worsening picture. Now I know vaccines can be tweaked but can we keep up?

In reply to Toccata:

Personally I'm not clear why we aren't worrying more about home grown variants than the international ones.  Our cases are already dominated by a variant that appears more transmissible than most if not all others given that it's apparently gradually becoming dominant in both Europe and the US (ref various news reports). And we already know variants of that most transmissible strain have now emerged in the UK with the E484K mutation that people seem to think is the primary cause of immune escape in the SA and Brazilian variants (to the extent of my limited understanding.) We can't keep home-grown mutations out so why aren't we hearing more in the news about updates on the spread of them and findings from local surge testing that has been taking place?

In reply to Toccata:

You're a medical professional aren't you?  No thanks needed - I am lucky enough that I've been able to isolate my household pretty robustly and remain usefully employed; a lot of people have had it far worse one way or another.  It's nice to be able to contribute in some small but useful way to others.

If we can get the vaccine in to everyone over 18 in the next few months, I hope it will give enough cross immunity that relatively mild control measures and perhaps local lockdowns can take the sting out of variants like this, and keep the wheels on society until updated vaccines are out.  But I think the risk is vey serious until that point is reached.  The societal equivalent of taking a half dose of antibiotics.  The best way to keep variants under control until vaccination is complete is to keep R<1 and to make the borders as hard as possible until vaccination is complete.

There's been a lot of talk about how vaccines can track changes like they do with the flu, but IMO this is misguided as the change mechanisms are very different (point mutations in Covid, gene swapping between influenza viruses) and we have processes to identify the likely season flu in advance, using the Southern Hemisphere, which gives time to pre-emptivly spool up vaccine product.  With Covid, it will always be reactive.  As long as there are parts of the world with uncontrolled spread, the variants will I expect keep happening, and we'll be forever chasing them.  

I was thinking about other ways the UK can help with this internationally - the lack of substantial sequencing in many places is a real barrier to understanding what the heck is actually going on when cases are high.  If we succeed in getting our case rate down and holding it down, we will have a phenomenal unused diagnostic capacity for sequencing which could be donated to countries for use in understanding their problems - and this will benefit us immensely, as the better everyone understands the problem, the better it can be managed, and its an interconnected world.   Used wisely this could have more beneficial effect than the plan of donating purchased but unneeded vaccine doses.

The other thing I'm interested in is if the membrane fusion protein can be neutralised by antibodies, if so, a separate "M" vaccine could significantly reduce the probability of a variant escaping a neutralising immune response.  This paper Dave Garnett shared is on my reading list - https://www.nature.com/articles/s41577-020-00480-0

In reply to Si dH:

I think both deserve a lot of caution, but we can do more about the International ones - at least initially.

Yes, there seems to be a lot of convergence going on in the variants.  Once the next stable configuration is reached, I wonder where mutations out of that will take us?   Given the convergence the distinction between variants is quite blurry.

A very good question.   The weekly update here [1] gives a window in to what's going on, but there's not much being shared beyond summary findings, from what I can see.

[1] https://www.gov.uk/government/publications/covid-19-variants-genomically-co...

In reply to wintertree:

This. This is on my mind.
It seems pretty clear that the three or four mutations that crop up again and again are advantageous. But why haven't we seen any others yet? Is that it? Is it maxed out?
What happens when we eradicate the vaccine-susceptible strains? Then we're optimising for vaccine escape and it's a new playbook.

In reply to wintertree:

I'm wondering about this even harder today.  Unless there's a sudden jump in reporting latency, it looks like the 0.7x week-on-week factor is headed for 0.6x or lower.

Or it could be the weather.

In reply to wintertree:

Thanks, I hadn't come across that page before, only a summary but would imply those variants aren't spreading much at the moment if we are looking at a third of cases being sequenced.

In reply to Longsufferingropeholder:

Isn't Lincolnshire quite agricultural with the workforce often sharing accommodation? Not the only region with this demographic but this could be one of the issues - then again, it's not the right season. 

In reply to wintertree:

Interesting point. I struggle to understand the motivations of dedicated anti-vaxxers in any case (well, apart from the ones making money out of 'alternative' treatments, where the motivation is pretty clear). What would be the point of pushing the anti-vax argument against booster jabs once 90% of the population has had the original jabs? I suspect they will still try...  

In reply to mick taylor:

Hopefully there will be stats to demonstrate this. Of course anti-vaxxers don't believe the stats (except their own) and don't really do common sense...

In reply to wintertree:

You may enjoy reading this from the Economist ( maybe behind paywall)

https://www.economist.com/britain/2021/02/27/how-british-science-came-to-th...

In reply to Misha: and others

Hospital deaths tolls are falling rapidly: 204 reported today which is down by 44% on last week, Tuesday was 315 which was down by 28% on week before. I remember a few weeks ago they were dropping by about 20%, so things going in the right direction and at pace.

Misha: us two shared some thoughts a about 3 weeks back on numbers - they are better then we guessed at. 

In reply to wintertree:

Or it could be cases data that has been misplaced?  Latest update is delayed with this message on the dashboard: Owing to an issue with the processing of cases data, the update for 3 March 2021 is delayed.

In reply to wintertree:

PHE give the daily number - by reporting date - in a Tweet - https://twitter.com/PHE_uk/status/1367143680696610819 - it doesn't include an unexpected rise, so perhaps cases really are about to start plummeting.   Dashboard still not updated...

In reply to wintertree:

It's all up on the dashboard now. They've added a few deaths that were previously missed in Scotland and I noticed it says yesterday they added a few cases that were previously missed in SE England. Only in the low hundreds.

The numbers are really great considering today's update includes almost all the cases for this last Monday. On current trend I think you might even see next Monday's raw figure being around about or below 5000 on the day by specimen date, never mind the weekly average or your data from the SG filter (Monday being the weekly peak in raw figures.)

The dashboard map now also looks pleasingly shorn of dark blue. Leicester is the only UTLA left above a weekly average 200/100k and will drop below it when they update the calculation tomorrow.

In reply to Si dH:

Yes, it looks like the cases plot is about to start an ever increasing decay (exponentially speaking), like deaths did a while back. Makes sense as vaccination is now taking effect down the age scale in both front line health workers and independent adults of ages where it’s more likely to prevent transmission. R should be dropping by the day.

I hope it’s that and not the weather.  It feels right, but it’s a modelling problem well beyond what I’d consider well boundable to predict.

I’ve got a plot prepared for this increasing decay; hopefully get it up here in an hour or so.

This is going to deliver ever more decay to deaths.

A critical time for the government to hold their nerve on control measures despite the positive news ahead, to get to the point everyone has had their first dose.

In reply to mick taylor:

Yes, deaths, admissions and cases are are falling well but particularly deaths, which just goes to show the impact that the vaccines are having. A bit concerning that the vaccination numbers haven't yet got back up to the levels seen in early Feb, although the massive day to day variation depending on the day of the week makes comparisons harder. I think the only meaningful comparisons which can be made on vaccinations at the moment are week on week and in the scheme of things even 200k / day on the quieter days is still pretty good going. It's just that there was some noise about numbers going up in March so I've got an underlying expectation of upping the ante or at least getting back to the early Feb levels. In practice, it's only 3 March so the expected increase probably hasn't happened yet or at least hasn't fed through into the numbers.

In reply to wintertree:

Even if it's partly the weather, still great to see the drop. Hopefully this drop will stick, so even if the decay rate slows again, it will be slower based on significantly lower absolute numbers.

Going by 5 days on average to symptoms appearing and then a couple of days for people to get a test, we'll see the last of the 'good weather' benefit coming through in the numbers being reported this time next week, give or take.

What makes me a bit concerned is what's the real incidence rate with these lower case numbers - could it be proportionately higher compared to when cases were high? What I'm wondering is could there be a certain minimum number of cases (in absolute terms) within demographics which are vulnerable to infection and aren't likely to get tested, partly due to being asymptomatic and partly due to financial considerations. For example, badly paid / zero hours people working in customer facing roles, youngsters going to all night raves and so on. The official cases might keep dropping but this rump of real cases might remain at a similar level. Perhaps I'm being overly pessimistic.

In reply to Si dH:

Here's some digital tea leaf reading.  

This is the plot of raw, unfiltered, cases by specimen date numbers.  It's wrapped module 7 so every week overdraws the previous.  If cases fall at a constant rate but there is a weekly redistribution of cases in a ratio basis, then the part of each bar "peeking though" on a log-y plot will have the same vertical size.  A bigger bar on a specific day means a larger week-on-week drop, a smaller one means smaller.  The blue ones are the week I'm calling the "cold shoulder".  The legend gives the date of the Mondays on the plot.  This is the best plot I can come up with for thinking about changes to trends on a day-by-day basis given the weekly structure superimposed on the underlying mechanic.

The final dark grey data for the 25th and 26th of Feb (Thu and Fri) aren't in the API download, so I've taken the provisional values off the dashboard and scaled them up by 1.002x and 1.006x, that being a stab at estimating the number of cases yet to be reported for those dates, going off other recent dates.  The red bars peeking through in that week are generally larger.

The second plot shows the height of the bars peeking though, which is the ratio of cases on a given date to those 7 days before.  This is another take on the exponential rate constant, but expressed as a week-on-week decay fraction.  The secondary y-axis gives the corresponding halving times.  

The blue period is what I've been calling the "cold shoulder".  The week on week decay for the week following this is clearly getting more aggressive every day, with the boosted provisional decay rate numbers for Thursday and Friday being the best we've seen in a month.  Looking at these plots and the even more provisional data for the weekend, I think this is going to continue for a bit.

It'll be really interesting to see where this accelerated decay is landing demographically - I'm not convinced it will be that segregated along vaccination lines as people's contacts are not strictly segregated along age lines.  

I think this way of plotting things is really quite powerful, but I'm not sure I've explained it at all well.

In reply to wintertree:

The falls are really impressive....

It is worrying that there is still talk (emphasis) (Matt hancock yesterday) about relaxation being irreversible...I worry that if something bad happens the irreversibility will be too hard wired... and we will again see delays in changing direction.

Early on there were some elegant graphs of I think incidence of new cases (might be more managable with a rate contant) vs prevalence, which produced looped tracks showing how as prevalence fell and restrictions were relaxed transmission increased ....can't find the link at the moment. They allowed quite useful annotations of interventions....

In reply to wintertree:

For folks info, just been looking at the 7 day infection rates for Grter Manc (on Manchester Evening News) and they are plummeting with an average of -26% and all negative. Still have higher than average cases though. 

In reply to mick taylor:

Update to the plots on daily changes.  

Very much hoping this is the vaccine and not just the warm spell, but either way it represents more absolute decay in case numbers which is no bad thing.

In reply to mick taylor:

The latest update to the demographic projections; it's turning in to a real roller coaster given the cold spell and now the vaccine effect coming online (or the warm spell).  

In reply to Si dH:

That ship has sailed with the raw number currently sitting at 5729, but assuming it doesn't go up much it still represents a much larger week-on-week drop than we've seen in a long time.

On the other hand:

This looks really promising and about as close to certain as I'd like to call something that's still 8 days away in the data, but if the halving time stabilises to its current value of around 9 days, we could see cases down to around 2,500 per day by the return of schools.

A lot hinges on if this effect is the vaccine (permanent), the warm week (transient) or a mix of the two.  The next 14 days or so of data will be very interesting.  Then it gets harder to interpret as the direct and indirect consequences of school reopening start to show through; if what we are seeing is the effect of the vaccine, I'd expect the halving time to continue becoming shorter over the next 14 days.  

In reply to wintertree:

Check these rates for Grter Manc LA’s !! (today’s, from Manc Eve News). Worth remembering 5 of those areas went +% in the last 4 weeks. 

In reply to wintertree:

Interesting thread on seasonality, I am not really knowledgeable enough to fully understand it so I welcome your thoughts.

https://twitter.com/DevanSinha/status/1367684859423035392?s=19

In reply to Punter_Pro:

I think they’re six to thirteen days behind UKC... It’s the same signal visible in Plot 9x and Plot 9e up thread and this much simpler plot [1].  It’ll be clearer in tomorrow’s plot 9 updates but is very clear already in the simpler plot.

The alignment of the change in rate constants (or R in the case of your link) and the cold spell is compelling.  There’s no previous excursion to slower decay in this wave.  I can’t offer solid grounds to put a P value on the excursion being random (as the noise in the data is well weird), but it’s highly improbable.  I’ve not seen any other compelling explanations.

As to what it means - I suspect it’s as simple as people closing more doors and windows when it’s cold out side.  As we know most transmission happens inside the weather doesn’t relate much to this, other than humidity.

If my noddy interpretation of the mechanism is right, it’s a very powerful insight in to what controls this virus...  Ventilation.

[1] https://www.ukclimbing.com/forums/off_belay/friday_night_covid_plotting_14-...

In reply to wintertree:

I meant below 5000 on Monday 8th, not Monday 1st. That would have been very optimistic! I'm guessing that below 5000 raw data on the 8th would be equivalent to 3-4000 weekly-average or using the SG filter.

In reply to Si dH:

That makes a lot more sense.  I agree we’re likely to see data for the 8th below 5000 - raw data with the reporting spike and all.

5000 per day for the 1st would have been optimistic - but not far out from how things are shaping up.  A good news week.

In reply to wintertree:

Re ventilation: I didn’t realise until recently the importance of ventilation and good to see BBC highlighting the schools issue and good advice;

https://www.bbc.co.uk/news/health-56268188

In reply to mick taylor:

Various studies of individual outbreaks (in a bus, restaurant, etc) were done ages ago, and some Spanish (?) scientists put great graphics onto them which really showed how the amount of infection depended on ventilation.

All comes back to the importance of viral loads.

In reply to wintertree:

"Owing to a delay in the submission of cases data for England to PHE, today's update is delayed."

ps. it does amuse me a little that the government covid map is using OpenStreetMap mapping. Don't we have a national mapping agency we could use...?

In reply to captain paranoia:

Yup, it's Wednesday all over again.  I'm once again left wondering if a lot of .XLS files have been lost behind a metaphorical sofa and this miraculous fast decay is going to be taken away by the next delayed update...

In reply to wintertree:

Maybe they could ask for help here...

https://www.ukclimbing.com/forums/off_belay/excel__formula-731897

In reply to captain paranoia:

Still awaiting the update...

I can understand this - trying to get geographic boundaries for UTLAs, LTLAs, MSOAs, English Regions, NHS Regions and NHS Trusts  is maddening.  If someone has tagged it all in OSM that's a lot of integration work taken care of...

I had this dastardly plan to re-project hospital data to an English Region basis by decomposing NHS Regions in to NHS Trusts, partitioning ones straddling region boundaries and putting it back together.  I haven't yet found a boundary map for NHS Trusts.  Postcodes can be manually scalped off the web, and with the PAF, LTLA level mid year population estimates and some water shedding I can see a way to something that tastes almost but not entirely unlike tea...

In reply to wintertree:

I think people meeting friends and family outdoors rather than indoors is the other significant factor.

In reply to captain paranoia:

It's been made really easy to overlay data on OSM. OS not so much.  

I'm satisfied with the off the peg solution. There would have been a news story like "government wastes millions paying coders to draw blobs on map" if they'd gone the other route.

In reply to wintertree:

Normal flu down by 98%  !!!!! 

You know the one we have every winter for decades!  Kills hundreds of thousands of people every year for decades !  

In reply to Monkeydoo:

Easy on the exclamation marks. Where on earth did you pluck that figure from? Around 10,000 deaths are caused by flu each year in England and Wales. https://www.greenwichccg.nhs.uk/News-Publications/news/Pages/Around-10,000-...

In reply to Longsufferingropeholder:

As I said, I'm just mildly amused. It would still seem appropriate that the national mapping agency would be best placed to have definitions of administrative boundaries; they have distinct legal uses (e.g. voting, healthcare, local council payment & services, etc).

Maybe as you point out, it's just that their mapping system isn't as versatile as an open-source one. Then again, I probably shouldn't be too surprised, comparing the quality of their mapping apps with those written by single-person, spare time developers...

In reply to wintertree:

I take that back with apologies.  Sinha suggested the weather pretty early on - Eric9Points linked a twitter post from them on plotting #13; this was 4 days after it was discussed on here and I expect they were thinking about it some days before they tweeted it.