Friday Night Covid Plotting #8

New Topic
This topic has been archived, and won't accept reply postings.
 wintertree 15 Jan 2021

A less gloomy update than of late.  The reporting lag seems to be up a bit, so the right most points on these plots are definitely provisional.

Cases at the England level are showing a convincing decay both on Plot 6e and in terms of crossing to a negative exponential rate constant or halving time on Plot 9e.  They appear to be plummeting at a faster rate than the last lockdown, but the peak from which they fall was very messed up by the sampling issues around Christmas and New Year's Eve.  I wouldn't read too much optimism in to this rate until next week's update.

Hospitalisations look to be levelling off in Plot 7.1e at a national level, and that can be seen by their exponential rate heading for a zero crossing in to the negatives, meaning decay.  The trend of cases then hospitalisations heading for decay on Plot 9e is similar to that at the start of the November lockdown so there's no reason in my mind to doubt that hospitalisations are likely to start reducing from now on.  Its notable that the lag between cases and hospitalisations (e.g. the number of days between where they cross the x-axis) looks longer now than in November.

Deaths are soaring on Plot 8e but they're clearly not on a fixed rate exponential - it looks more "linear", and you can see on plot 9e that their exponential rate is heading for 0 then hopefully decay (meaning a reduction in daily numbers) in sequence with hospitalisations and cases.

So, at a national level it looks compelling that things are improving - although it is going to be some time yet before the decay in hospitalisations translates in to a decay in hospital occupancy, as people on average take a long time to leave.

But, all of this has happened before in November, and the new variant blind sided everyone.  Still, this is more than I had hoped for.  It's critical that people don't relax their vigilance because cases are falling - they're still at a level I'd officially describe as "bonkers" and about 1/3ʳᵈ of all acute hospital beds in England are currently Covid patients.  This is an awful situation for everyone who works in or needs support from a hospital.

Link to previous thread - https://www.ukhillwalking.com/forums/off_belay/friday_night_covid_plotting_7-7...

Post edited at 19:50

OP wintertree 15 Jan 2021
In reply to wintertree:

The "4-nations" plots; cases are in decay in all nations.  You can see on plot 9x that this decay is much more synchronised than in the second half of 2020.  This I think is a step forwards - trying to lower cases in one nation or region whilst having them rising in another is not going to help - here the lockdowns and control measures are all working together, to the benefit of all.


OP wintertree 15 Jan 2021
In reply to wintertree:

Cases appear to be in decay in all the English regions.  All the "red" and "blue" regions now have negative exponential rates, and halving times for cases.  The North East has cases decaying more aggressively like the "red" areas that went through their spike in case rates earlier.  The South West is decaying the least (faintest blue on the right hand side of the cases heat map).  The South West has stood out on Plot 18 for about 4 weeks since it joined a "red" style path.  Likely related to prevalence of the new variant there.

Hospitalisations - I still need to fix this plot to use the NHS Regions not aligned to English Regions (as used by cases and deaths).  Si dH pointed out on plotting #7 that the NHS Regions do not always correspond perfectly to the English regions.  This must frustrate a lot of analysis.  All regions show are likely having a decrease in hospitalisations by today.  The red regions havre crossed in to decay on this plot, and the blue ones are on course to - and the plots ends about 5 days in to the past due to reporting lag.  Given the past 10 days of behaviour in the cases plot, I expect these to continue to drop to more negative exponential rate constants - meaning that hospitalisations halve in fewer days.

Deaths - the exponential growth rate is heading for zero in most regions meaning that an end to growth is in sight.  It'll take a while yet for this to happen though, but again given the previous two measures there's no reason to expect them not to level off and start decaying.

Like I said though, we've been here before.  There needs to be a lot more vigilance against hyper-local failure of lockdown and identification of the problems.  Screening travellers with qPCR negatives is probably not going to keep new variants out indefinitely as someone infected just before their test won't be shedding enough virus to be detected, and will become infectious once they are here.

It remains possible that some areas have currently low cases in growth that will come to dominate over the decaying areas and reverse the regional level trend, as happened last time. Lag in the data is worse at the moment and I can't usefully update my UTLA level plots (Plot 16 and the maps) until tomorrow.  They're one to keep a close eye on.

I've updated the figure on hospital occupancy.  It's bonkers, and what this plot means to me is that the people on the ground working well beyond anything most of them ever expected.   It's notable how the headroom created by reducing the rate of admission to ITU ~2x from the first wave was eaten up by just two weeks of exponential growth.


In reply to wintertree:

Awesome as ever. Thanks.
As I read this, I've just been reading the news articles saying that cases are falling nationally, and also saying R is estimated at 1.2-1.3.
Now, I mean, I know R is a relatively meaningless number pulled out of a repressed scientist's ass in the early hours under extreme duress, but how out of date are those estimates when they hit the headlines? Is there any reason to expect a contradiction like that? Other than it's a worthless number?

OP wintertree 15 Jan 2021
In reply to Longsufferingropeholder:

As far as I can tell the estimates of R given on the dashboard are nonsense.  The lower bound was generally 1 when cases were clearly rocketing and is now > 1 when cases are falling.  Perhaps it’s given for some period in the past and is lagged.  

I can’t see any reason for such a contradiction other than lag.

 MG 15 Jan 2021
In reply to wintertree:

I read something  20mins ago on  this. It does lag as it takes into account various data sources not all.of which are immediately available.  I'm not sure how useful an accurate but historic measure is.

 Si dH 15 Jan 2021
In reply to MG:

I saw that too. Apparently it accounts in some way for hospital data as well as cases. So not my understanding of what the true r is. I get the impression it is mostly about messaging - it's the number they want the media to report.

Post edited at 21:17
 Si dH 15 Jan 2021
In reply to wintertree:

I confess to a healthy level of scepticism this week. How confident are you that the combination of deweekending, xmas/ny shunting and filtering is dealing adequately with the behaviour over the last 2 weeks? Do you have any of the heavily affected dates* on the edges of the period you use for calculating the exponential constants in a way that could bias the outcome? And if you include the 11th January data in the calculation (which will now be virtually complete) how much does it change the answer? It would be interesting to see the sensitivity to this for the England cases graph and exponential constant graph.

*24/25/26/1 specimen date cases were all artificially low and 29/30/3/4/5 were all artificially high to my eyes. Some obviously more so than others.

Post edited at 21:25
OP wintertree 15 Jan 2021
In reply to Si dH:

All appropriate questions IMO.

>  And if you include the 11th January data in the calculation (which will now be virtually complete) how much does it change the answer?

All these plots use data from Monday the 11th January - that's necessary to backfill the weekend immediately before, as otherwise that's under sampling would dramatically bias cases/day down.  The 11th is not yet in the UTLA demographic level data, hence the absence of plot 16 and plots D1-D3 until tomorrow.   It's this wait for a Monday's data that largely fixes Friday as the plotting update day.

> Do you have any of the heavily affected dates* on the edges of the period you use for calculating the exponential constants in a way that could bias the outcome?

The exponential is fit to a symmetric window around each data point, and depending on the plot does some polynomial filtering before or after the fit, so they're pretty delocalised.  I'm least confident in the values around Christmas, but we're far enough away from that now that it's not having any effect on the most recent measurements; these are always provisional though as the fitting window is by necessity truncated to be asymmetric without the future days, so measurements can and will change.

> How confident are you that the combination of deweekending, xmas/ny shunting and filtering is dealing adequately with the behaviour over the last 2 weeks?

It's not great, but it's a lot better than without out.   If you look at the final plot, you can see there's still an increase in residuals from a trend line around the Christmas period, and Christmas Day in particular is under-sampled - I think there was a net reduction in tests over the period, so it can't be shunted over.  But, for the most recent dozen data points or so, the residuals look pretty typical, so I'm not worried about them - other than noting they remain provisional until data for what is currently the future provides them with a full symmetric window for fitting.

I've put a plot in below that shows the shunts.  They're done by re-assigning a fixed fraction of cases at whatever level (National, Regional, UTLA).  One plot below shows the shunted positions, with purple and orange lines showing where cases were moved from or to.  Note that when shunting to a Monday, I'm trying to re-create the typical reporting spike and not land it directly on the trend line, as then the normal de-weekending will do its thing. 

I think I could improve the shunts a bit more with some effort, but it mainly affects understanding of the shape of the peak, and that's rapidly becoming the past.  

The final plot shows each of raw data, shunted data and shunted and de-weekended data with trendiness.  The residuals are against the trend-line and decrease with each step - and the trend-line becomes smoother.  Key point - each level of improvement (shunting and then deweekending) slows the rate at which cases appear to be falling compared to raw data, so they're not leading to an over-representation of the improvements to cases.   You can see this flip-booking between the two plot 9s below.

Ultimately though I've a high confidence in this week's plots - the sequence of behaviour across the various measures on Plot 9e is sane and corroborates what the individual curves are saying.  I'd like to propagate the noise on the data through to confidence measurements on the final curves, but the non-gaussian and time correlated nature of the noise defeats my skills there for a proper noise analysis. 

My hunch is that the exponential rate of case reduction is going to slack off as the easy wins run out of give, and the more stubborn transmission pathways remain.  I could convince myself I see hints of that in the more provisional counts for Jan 12th which aren't used in these plots.  So by next Friday I'll be pleasantly surprised if the halving time for cases actually hits 10 days as it looks to be on course for now.

I remain kind of professionally ashamed over the approach to de-weekending, but we have a qualitative understanding of the problem, a phenomenological understanding of the consequences, and theoretically a convolutional approach to filtering is not appropriate for time-correlated noise or noise with an x-axis dependancy.  I'm largely out of the paper writing game now but it would be interesting to try and put this through peer review along with a push to focus on exponential rates and not "R" for analysing the effects of policy in real time.  It would be better yet to have the labs enter the date on which the sample was taken rather than ingested by the lab...  (Which I still think is the problem.). As grubby as I feel taking this approach to the noise, flip-booking between the two plot 9s below convinces me - look at how one has all these jagged edges and asymmetric peaks in the blue cases curve, and the other is much smoother and more symmetric.  The effect is minor in most places but becomes critical in the leading edge where there is less data to fit to, so noise and bias are worse problems.  It also becomes more important at UTLA and demographic level where the statistical noise is worse.

Post edited at 21:54

OP wintertree 15 Jan 2021
In reply to thread:

A new plot - each of the measures (cases/hospitalisations/deaths) re-scaled so that the height of the peak after the November lockdown is the same.  The mid-height horizontal line shows the value of each peak.  The second lines shows the value of each curve at the turning point of the new peak in cases after the start of the January lockdown.   

This is a vey noddy plot to go with the mental maths I know various posters have been doing to try and scale the recent cases to deaths.  It's not a happy plot, but it looks better than the numbers various people were - reasonably - arriving at.  The key I think is that the unusually noisy data around Christmas presented some individual daily numbers of cases that were exceptionally high, but these don't accurately represent the trend.

Looking at this, I'd be surprised if deaths/day top 1,100 for England if the current trends continue, although this is highly dependant on hospitals holding up to an insane amount of strain for another 3-4 weeks, as occupancy will likely continue to rise for some time yet until patients from the period of peak admission start to be discharged in large numbers.

Post edited at 22:14

mick taylor 15 Jan 2021
In reply to wintertree:

No optimism with me. Hi cases reported today (56k+)  Cases in NW still going up in many areas (slowly comin down elsewhere) people have had enough. Local news: people visiting beauty spots, high car usage, high lockdown fines. Rates will come down but very very slowly. Gut instinct: acquired immunity and vaccines will start to have a bigger impact than lockdown. 

1
 Si dH 15 Jan 2021
In reply to wintertree:

Thanks for the detail on the shunting etc, I appreciate you taking the time and think you've probably done the best job possible with what data is available.

Outside of London, SE and East of England I am still hesitant about whether I believe cases are really dropping much yet. It all comes down to how much you believe the data needs shunting and deweekending (and I appreciate your approach is based on data from past weeks, and that as you say it gives a less surprising result than the raw data.)  I have just lost a lot of confidence in data across quite a few of the dates due to the Xmas and New year issues. 

To illustrate: cases from Monday 4th through to Sunday 10th show a continuous decline almost everywhere. That's great. However, we know that Monday 4th and probably the 5th had a load of extra cases in them left over from the 1st. Looking at a cases graph just for the north west (link below) it is easy to postulate that if all the cases by which the 1st was below the trend line got added in on the 4th and 5th, so therefore need removing from those dates' totals, then suddenly you are in a situation where the cases on the 11th look higher than the true number of cases on the 4th.

https://coronavirus.data.gov.uk/details/cases?areaType=region&areaName=...

I feel more confident things are really dropping having seen your data and analysis than I did before from the basic case data and week-on-week changes in average cases. (Those data are totally skewed at the moment.) I think I'll feel truly confident once the data is complete for Wednesday 13th (in about 2 days time) at which point we'll have a direct week-on-week comparison with the first date (6th) that was mostly unaffected by holiday issues. If the case numbers for Wednesday 13th are lower than those on Wednesday 6th across all regions, I'll be happy . Probably wouldn't put any money on it just yet though.

OP wintertree 15 Jan 2021
In reply to Si dH:

> think I'll feel truly confident once the data is complete for Wednesday 13th (in about 2 days time) at which point we'll have a direct week-on-week comparison with the first date (6th) that was mostly unaffected by holiday issues. If the case numbers for Wednesday 13th are lower than those on Wednesday 6th across all regions, I'll be happy . Probably wouldn't put any money on it just yet though. 

Indeed; I’m planning an update once that is out probably on Sunday.  Tuesdays are used when available by deweekending at something like 1/3 of Mondays (based on patterns in the data).  

> However, we know that Monday 4th and probably the 5th had a load of extra cases in them left over from the 1st.

Yup.  I’ve tried to shunt those back in time and the trend is still there - but a much more moderate halving time.  

I’ll update the UTLA and demographic plots tomorrow when the 11th clears through in to that data and update the plot 9 variants once Wednesday is through.

As with last week’s updates, picking apart the details to decide if cases are falling a lot or not at least means they’re definitely not rising.  Right now I’m grateful for every break we get...

 Si dH 15 Jan 2021
In reply to wintertree:

> As with last week’s updates, picking apart the details to decide if cases are falling a lot or not at least means they’re definitely not rising.  Right now I’m grateful for every break we get...

Fair point - it's a better debate to be having than whether the exponential increase rate is still going up!

Also - the vaccine programme hit over 300k/day today. That's what they need to hit their target. Which would be a pleasant surprise.

 Misha 15 Jan 2021
In reply to Si dH:

All eyes on next week’s numbers. It will be the first time in a while that we could compare the week on week numbers from two ‘normal’ weeks. 

 Toerag 15 Jan 2021
In reply to Longsufferingropeholder:

> Awesome as ever. Thanks.

> As I read this, I've just been reading the news articles saying that cases are falling nationally, and also saying R is estimated at 1.2-1.3.

> Now, I mean, I know R is a relatively meaningless number pulled out of a repressed scientist's ass in the early hours under extreme duress, but how out of date are those estimates when they hit the headlines? Is there any reason to expect a contradiction like that? Other than it's a worthless number?

Cases* can fall even if R is above 1 if there are more people recovering than being infected.

*I'm assuming they're talking about live case numbers.

Post edited at 00:16
In reply to wintertree:

I know it was mentioned at the end of #7, but I feel it's worth repeating...

Toby Young and Telegraph censured for "failing to take care not to publish inaccurate and misleading information".

In July last year... Shame this process took so long to find what we all bloody knew full well...

https://www.bbc.co.uk/news/entertainment-arts-55676037

Post edited at 00:22
OP wintertree 16 Jan 2021
In reply to captain paranoia:

Indeed.  

What really disgusts me is that a gobshite “journalist” - and I use that term very loosely - is being held to a much higher standard than two professors from the world’s second oldest university.

 Toerag 16 Jan 2021
In reply to neilh in part 7:-

> I know when very well placed business man who seems to fly in and out of Jersey regulalry. How on earth does he get away with it?

> He goes for the day and has meetings at the airport. Suggesting that there are great big holes in the strategy for those with the right connections.

I suspect he's using a 'business tunnel' which essentially quarantines him in the airport then lets him go again. We had a similar thing except people using it could go wherever they liked as long as the local business had planned everything out in advance and had their plan okayed by government.  Some people used it to view houses, they had to wear masks and use strict social distancing. Any rooms entered had to be left alone for 2 days afterwards.

In reply to wintertree:

> What really disgusts me is that a gobshite “journalist”

I don't think Johnson is being held up to any standards...

 Punter_Pro 16 Jan 2021
In reply to wintertree:

Things are definitely ''starting'' to look up from a case point of view, the Cambridge Uni Biostatistical team have also reported a drop in R numbers in various regions and they are predicting deaths to have dropped by the 28th of January.

https://www.mrc-bsu.cam.ac.uk/now-casting/

I think this coming week will be the decider on how things are going, the talk of this new Brazil variant is slightly worrying, just because they haven't found it yet doesn't mean it isn't here. 

The Vaccination program also seems to be motoring ahead which is great, they seem to be well on course for 2m a week. If the J&J single shot one gets approved and is shown to be effective (which it looks to be), we could start seeing 500k a day if they can nail the logistics of it.

Re: Hospitals. My partner came home last night and said that things aren't improving hospital wise, Addenbrookes are now redeploying doctors to Ipswich & Colchester as it has now flared up in those regions. There is talk of her and her team being redeployed from their outpatient roles but we shall see over the coming week, she has her Pfizer Jab booked in for next week also so that is a bit of a relief...

Post edited at 06:38
In reply to Toerag:

They're usually taking about positive tests reported on a given day (not by sample date which doesn't help)

 Dr.S at work 16 Jan 2021
In reply to wintertree:

Re SW - certainly cases still rising very recently here - ‘points west’ do a useful regional case summary each evening, and whilst the rise of cases appears to be slowing, it’s still on the up last time I saw it.

To a certain extent I wonder if the fact that the SW has generally had a ‘good’ pandemic thus far contributes to that? Maybe some complacency?

 Wainers44 16 Jan 2021
In reply to Dr.S at work:

I think you are right,  there has been a little bit of covid seemingly happening somewhere else. It does look to be slowing down with only 3 of the 9 or so Westcountry areas still increasing.  Worst by far is still Cornwall,  remember *lucky* Tier 1 Cornwall.  Doesn't seem like being Tier 1 was such a good thing now.

 BusyLizzie 16 Jan 2021
In reply to wintertree:

Thank you again. You are a hero.

mick taylor 16 Jan 2021
In reply to Wainers44:

A fundamental flaw (of which there are lots) of the govt is that they come from such privileged backgrounds they don’t understand ‘normal’ behaviour. Chuck in Johnson’s need to be Mr Popular and we have a deadly mix - shit decision making.  The ‘low tiering’ was a disaster. 

mick taylor 16 Jan 2021
In reply to Punter_Pro:

Vaccinations are impressive: 803k in last 3 days. Not sure if they do as many over weekends, but I reckon they will hit the target. Last chance saloon for Johnson - they will throw everything at it. 

Oh, and hats off to Oldham for vaccinating homeless people. Couple on TV who live in a tent who coz they have no TV never really knew what was going on. I could have cried watching them get jabbed, they were very appreciative. 

Respect to the Vaccinators !!

 Wainers44 16 Jan 2021
In reply to mick taylor:

> A fundamental flaw (of which there are lots) of the govt is that they come from such privileged backgrounds they don’t understand ‘normal’ behaviour. Chuck in Johnson’s need to be Mr Popular and we have a deadly mix - shit decision making.  The ‘low tiering’ was a disaster. 

Correct but initiall it looked like it could make sense from here....sunny Devon....as the rates were all low around here. But if you gave it any thought at all it was a very high risk strategy. 

OP wintertree 16 Jan 2021
In reply to mick taylor:

> Hi cases reported today (56k+)  

It's worth keeping in mind yesterday was mainly reporting on a Monday, which contains a reporting spike after the weekend depression in samples processed.

> Cases in NW still going up in many areas (slowly comin down elsewhere) people have had enough

Edit: Looking at the NW it does look concerning - very close to levelling off.   It's hard to tell as the weekend effect is quite pronounced; unless they start rising or plummeting, a few more days data is probably needed to get a good idea, possibly until next Friday.  As you say though, if they're not coming down, it's not good.

> Gut instinct: acquired immunity and vaccines will start to have a bigger impact than lockdown. 

Immunity in some demographic bands of the more transmission-prone in some areas (that's a lot of "somer") may be getting quite significant; I'd be interested to see any reasoned study of this.  

In reply to Punter Pro:

> My partner came home last night and said that things aren't improving hospital wise

Yes; even as admissions start falling - assuming that develops as a clear and persistent trend - it's going to get worse as it takes a long time to discharge the people from the peak.  I worry that the messaging on this isn't very clear - cases and admissions have to keep falling for a few weeks yet to keep things from getting worse and worse in hospitals.

It's good news that your partner is getting the vaccine soon - the healthcare people I know offline have started receiving theirs.  In general, the vaccination roll out seems to be going very well.  It's great that we're going in to the best future from the range modelled in the LSHTM pre-print on the new variant.  Fingers crossed.

In reply to Dr.S at work:

 > To a certain extent I wonder if the fact that the SW has generally had a ‘good’ pandemic thus far contributes to that? Maybe some complacency?

If I didn't have enough to do in life, I'd want to work at unpicking what's social/behavioural and what's intrinsic to the mechanics of the virus with a mixed risk population.  One silver lining is that the studies coming out of this pandemic are going to produce so much advice for the next one.  Needs a government willing to read it of course...

Post edited at 16:57
 Offwidth 16 Jan 2021
In reply to wintertree:

Something weird is going on around London. A lot of area's cases in NE London and suburbs seem to have halved in the week to Jan 11 from the gov.data pages. On the surface this is great news so much so the R rate predicted for the new variant must wrong or some other data issues boosted Jan 4th numbers or herd immunity is kicking in (or a combination of factors).

https://coronavirus.data.gov.uk/details/interactive-map

Post edited at 17:27
OP wintertree 16 Jan 2021
In reply to Offwidth:

> Something weird is going on around London

If you look at plot 17 above, the reduction in cases started before the new lockdown, and the improvements can be seen on a similar time scale in plot 18 for all regions. 

I wonder if the schools closing for Christmas has had a large effect - this will be lagged as it involves transmission to the child then from the child to someone else in the household.

> or some other data issues boosted Jan 4th numbers

The Monday Jan 4th numbers were probably boosted by including additional samples from the 1st which was a bank holiday Friday, with a bigger "weekend effect"

OP wintertree 16 Jan 2021
In reply to thread:

A couple more plots now data for the 11th is through to the demographic download.  

Plot 16 - I should probably change this to show markers for the peak case rate and not the historical rate at tiering.  The green dot with the highest rate per 100k is Knowsley, a place that's frequently stood out.

Plots D1-D3 - it's notable that ages 5-15 have the fastest exponential decay rates.  The slowest decreases are seen at over 75 years of age.

I'll update plots 17 and 18 tomorrow when data through to Wednesday has cleared the majority of reporting lag.  That'll put a bit more space between the last bank holiday weekend and it's sampling weirdness and the leading edge of the plots - for now I still consider the right hand side of all of these as quite provisional.


OP wintertree 16 Jan 2021
In reply to thread:

A map of exponential rate constants and characteristic times.  These are all from the most recent week of data to give an idea of the trend, but the cautionary rider is that these measurements are provisional.  The North West and some of the areas with the softest recent tiering are amongst the last to tip to apparent decay.  I'm not reading much in to the growth in Rutland - it has a very low number of cases, so the noise is quite significant.  


 MG 16 Jan 2021
In reply to wintertree:

Possibly of no interest to you but I imagine you could sell your analysis of covid data. It is  by far the clearest and most rigorous I have seen.

 Si dH 16 Jan 2021
In reply to Offwidth:

Jan 4th, and also the 5th to a lesser extent (in Liverpool area also the 6th, presumably because of high case numbers causing a bigger backlog to work through) were all artificially high and Jan 1st artificially very low. Essentially a large number of the Jan 4th cases were actually Jan 1st cases. The week on week changes presented are currently erroneous because the days that offset each other like this are not all in the same week. I haven't looked to see how significant this is in East London but that was the initial hotspot along with south Essex and bits of Kent, and was the first to start falling a couple of weeks back.

Post edited at 18:24
 rurp 16 Jan 2021
In reply to wintertree:

Thanks for this work. It really helps to inform me and others in my line of work as we plan and deliver the vaccine rollout. I share your plots with our local primary care teams. 
They are clear, concise and visually simple to understand. Please continue if you can. 

 Si dH 17 Jan 2021
In reply to wintertree:

> > think I'll feel truly confident once the data is complete for Wednesday 13th (in about 2 days time) at which point we'll have a direct week-on-week comparison with the first date (6th) that was mostly unaffected by holiday issues. If the case numbers for Wednesday 13th are lower than those on Wednesday 6th across all regions, I'll be happy . Probably wouldn't put any money on it just yet though. 

> Indeed; I’m planning an update once that is out probably on Sunday.  Tuesdays are used when available by deweekending at something like 1/3 of Mondays (based on patterns in the data).  

Ok, I'm a believer now. Unusually few cases have been added to the 13th today, for a day that is only three days before the reporting date. As a result, cases on the 13th are definitely going to be lower than on the 6th (which was relatively little affected by New Year issues) in all regions. That means two things to me: (1) cases are definitely dropping week on week in all regions, although faster in some than others, (2) the testing system is working better (faster) as case numbers drop.

Thanks again

Post edited at 16:47
 dwisniewski 17 Jan 2021
In reply to wintertree:

Apologies if posted before/elsewhere, but..... Some good news on B.1.1.7 lineage front.

A new preprint out on medRXiv (15/01/21) from the ONS Covid-19 Infection survey team gives us a little bit of optimism: https://www.medrxiv.org/content/10.1101/2021.01.13.21249721v1

They've conducted the largest representative sampling of the UK population with respect to infection with the B.1.1.7 lineage of SARS-CoV2. 

Some of the main findings that give cause for hope:
 

  • Compared with initial reports, they found no higher viral load compared with other lineages (those which were predominant before the emergence of B.1.1.7). Explanation is that early reports were biased by increasing rates of test positivity. qPCR Ct values stabilising more recently, in line with previous lineages.
  • Growth rate is ~6% greater compared with other lineages (daily increase) - although was 9% probably sometime around early-mid Dec and now more like 4%. Good news is that again compared with earlier estimates of a 50-70% increase in transmissibilty, it now looks more like a 24-40% increase.
  • Infection appears to be similar across all age groups. Thus not disproportionately infecting <18 year olds.

So in all seems to be quite optimisitic compared to where we were a few weeks ago. Compared to reporting on B.1.1.7 a few weeks ago we now observe lower viral loads, growth rate appears to be slowing, and it's not disproportionately infecting young people.

Raises a few questions of what's actually going on with this lineage though.

OP wintertree 17 Jan 2021
In reply to dwisniewski:

Thanks.  

This is a very different take to the other analyses.  I don't view it as a "downgrade" in the findings so much as another piece of information, confusingly expanding the range of possibilities - although expanding them in a generally less disastrous direction moves the most likely reality that way too.

  • Pro for this paper vs previous: it's analysing more directly relevant data and doing less model fitting than the LSHTM piece
  • Con for this paper vs previous:,it was performed from the separate ONS study which is arguably less relevant than the P1/P2 data - and a lot smaller in sample size.

It would be good to see the same access given to the Lighthouse labs data so that this study could be replicated on the P1/P2 data.

If I understand right, they use a broader set of symptoms than those that qualify someone for P2 testing.  It would be good to see them test if a larger fraction of people are asymptomatic under the P2 qualifying criteria, to see if that backs some of what they ponder on asymptomatic transmission - they seem to want to say it's more of a driver with this variant, but qualify what they say to stop short of that.  Perhaps I over or mis interpret.  This seems like a missed opportunity.  

I am not sure about their Ct comments in the results; this would appear to be very different for random sampling surveys like theirs (ONS) and symptomatically targeted ones like P2.  There's no contextualisation on that and I don't trust my own interpretation. 

> Growth rate is ~6% greater compared with other lineages (daily increase) - although was 9% probably sometime around early-mid Dec and now more like 4%.  Good news is that again compared with earlier estimates of a 50-70% increase in transmissibilty, it now looks more like a 24-40% increase.

As I understand Figure 4B, the difference in the daily growth rate for the south east regions  ("red" regions on my regional plots) was more like 15% (the middle of their confidence interval, extending up to >20%).   Corresponding transmissibility increases would then be more like 50% to 80% for those regions, which tallies with the LSHTM analysis - which would have been largely biassed towards values from the regions then driving cases, hospitalisations and admissions - which were the "red" regions, where-as this study balances all the regions relatively evenly to get its average 6% day-on-day increase.   The key context here is that many of the regions contributing to the weighted 6% average in this paper were under Tier 4 at the time (lower growth rate), where-as the bulk of case growth driving the LSHTM analysis was under Tier 2 at the time (higher growth rate).  Diving in to the details of Figure 4B in this paper I think it pretty much confirms earlier estimates of the potential for this new variant to be terrifying, but it also offers evidence that control measures still work.  

The question arrises, why was there such a pronounced difference in the growth rates for the new variant in their findings.  As I understand it, they define "epochs" in a time-dependant way that is different for each region based on some modelling derived critera for when the new variant starts driving cases.  This means growth rates from different calendar times are being compared - when different control measures are in place.  Each variant will have a growth rate and transmissibility that decreases differently with increasing control measures.  Mixing different calendar periods in to the epocs in Figure 4A and 4B scrambles / looses the information on control measures and makes it hard to understand if those high day-on-day rates were unique to the "red" regions or to the calendar dates.

> Raises a few questions of what's actually going on with this lineage though.

I hope I've given a clear explanation of why I don't think there is actually a conflict in their findings on the growth rates and transmissibility with the LSHTM pre-print.  It definitely still raises questions to me over the age dependancy and the Ct value vs some of the other findings and speculation.

It's great to see the sheer amount of work being pulled together so quickly and openly on this from the scientific community.  It's just as well there is such a strong and diverse academic and medical community in the UK that was ready to drop everything and get on this; I hope the government appreciate how we'd be stumbling in the dark without this scientific community, and looks at future RCUK budgets accordingly.

Post edited at 22:00
OP wintertree 17 Jan 2021
In reply to MG:

> Possibly of no interest to you but I imagine you could sell your analysis of covid data. It is  by far the clearest and most rigorous I have seen.

I'd never really considered selling analysis of other people's data; it's very odd doing this with data I don't know everything about, because I didn't build the systems taking it .  I'm trying to have fewer jobs in my life rather than more, and couldn't really sell this as I'm not going to break with the Friday night tradition and don't have time to add much more on top of it.

I would however be delighted if I got a consulting gig of PHE for a few months to talk to them about the power of well presented data, and how to do it...

What I should really do is tidy this lot up and put it on GitHub.

Post edited at 21:55
OP wintertree 17 Jan 2021
In reply to rurp:

Thanks rurp.  Some of this analysis has made it to some pretty interesting places by another channel; it's good to know others are finding it professionally useful to.

> They are clear, concise and visually simple to understand

That's the plan.  Do let me know if there's a particular plot you'd like.  I was going to use tonight's Covid Data Half Hour to re-work Plot 16 to show levels vs those when the January lockdown started, but I ended up reading the preprint dwisniewski posted...

OP wintertree 17 Jan 2021
In reply to Si dH:

> Ok, I'm a believer now

Yes, today’s data release was good.  Hospitalisations look for be levelling off too.  No updated plots from me though, the weekend effect was strong and remains even after my algorithms and is clearly biassing the plots; it’s probably need to wait for Friday again; maybe this coming Tuesday.

> (2) the testing system is working better (faster) as case numbers drop.

Interesting. I stopped doing the daily download needed to measure the lag distribution function of testing; more direct fish to fry.  The scale of testing now is phenomenal; sequencing too.  If those labs remain hardened by the experience of this winter, by the time we’re down to 2000 infections/day, hopefully they can turn results around with very little latency and sequence all cases; with a low enough latency this enables “backwards” contact tracing which would help massively to detect and mop up super spreader events.

 Offwidth 18 Jan 2021
In reply to dwisniewski:

I don't see how age distributions being unchanged ties up with reports of proportions of young adults hospitalised nearly doubling and the extra proportion hospitalised above that age gradually declining with age.

 Toerag 18 Jan 2021
In reply to Offwidth:

> I don't see how age distributions being unchanged ties up with reports of proportions of young adults hospitalised nearly doubling and the extra proportion hospitalised above that age gradually declining with age.


Behavioural effect I suspect. Lots more 'young' people in infectious scanarios compared to more elderly being protected better than last year.  A higher proportion of young people going into hospital is due to either a) a change in the virus infecting them better b) a change in the virus making them sicker or c) lots more of them being infected than old people.

OP wintertree 18 Jan 2021
In reply to Toerag:

> Behavioural effect I suspect

I was going to suggest something similar; closing schools also lowers transmission in 4-18 (vey low hospitalisation rate) and the age range of their parents (generally 30s upwards), which lowers prevalence in the 30-50 band, meaning that 18-30 is higher relatively speaking.

It's really hard to tell though as the hospitalisation data is published in such coarse age bands.  

 Si dH 18 Jan 2021
In reply to Offwidth:

> I don't see how age distributions being unchanged ties up with reports of proportions of young adults hospitalised nearly doubling and the extra proportion hospitalised above that age gradually declining with age.

What data do we have at a trust or regional level to support this? I haven't seen anything.

 dwisniewski 18 Jan 2021
In reply to wintertree:

> Thanks.  

> This is a very different take to the other analyses.  I don't view it as a "downgrade" in the findings so much as another piece of information, confusingly expanding the range of possibilities - although expanding them in a generally less disastrous direction moves the most likely reality that way too.

Yes, I agree. More update than 'downgrade'.

> Con for this paper vs previous:,it was performed from the separate ONS study which is arguably less relevant than the P1/P2 data - and a lot smaller in sample size.

> It would be good to see the same access given to the Lighthouse labs data so that this study could be replicated on the P1/P2 data.

> If I understand right, they use a broader set of symptoms than those that qualify someone for P2 testing.  It would be good to see them test if a larger fraction of people are asymptomatic under the P2 qualifying criteria, to see if that backs some of what they ponder on asymptomatic transmission - they seem to want to say it's more of a driver with this variant, but qualify what they say to stop short of that.  Perhaps I over or mis interpret.  This seems like a missed opportunity.  

I'm not entirely sure what you mean here RE the pillar 1/2 testing?
To me one of the strengths of this surveillance study is that it's not biased by symptomatic testing. Also regarding what they say about asymptomatic transmission, I read it as saying that asymptomatic transmission accounts for a large proportion of cases as is the case for other lineages.

"Importantly, rates of SGTF infections with and without self-reported symptoms were similar, consistent with the higher prevalence of asymptomatic infection reported in defined populations (30%19) and community surveillance (e.g. 42%20, 72%21). Asymptomatic infections may therefore be contributing substantially to B.1.1.7/VOC202012/01 spread, and are not currently captured by the national testing programme, which focusses on symptomatic cases and their contacts."

> I am not sure about their Ct comments in the results; this would appear to be very different for random sampling surveys like theirs (ONS) and symptomatically targeted ones like P2.  There's no contextualisation on that and I don't trust my own interpretation. 

I think the differences are likely due to the period of time that testing was done. New exponential growth of a lineage, more cases that have newer infections and therefore higher viral loads. They do report the drop in Ct values in early Dec, that had first been reported around 20th December.

> As I understand Figure 4B, the difference in the daily growth rate for the south east regions  ("red" regions on my regional plots) was more like 15% (the middle of their confidence interval, extending up to >20%).   Corresponding transmissibility increases would then be more like 50% to 80% for those regions, which tallies with the LSHTM analysis - which would have been largely biassed towards values from the regions then driving cases, hospitalisations and admissions - which were the "red" regions, where-as this study balances all the regions relatively evenly to get its average 6% day-on-day increase.   The key context here is that many of the regions contributing to the weighted 6% average in this paper were under Tier 4 at the time (lower growth rate), where-as the bulk of case growth driving the LSHTM analysis was under Tier 2 at the time (higher growth rate).  Diving in to the details of Figure 4B in this paper I think it pretty much confirms earlier estimates of the potential for this new variant to be terrifying, but it also offers evidence that control measures still work.  

> The question arrises, why was there such a pronounced difference in the growth rates for the new variant in their findings.  As I understand it, they define "epochs" in a time-dependant way that is different for each region based on some modelling derived critera for when the new variant starts driving cases.  This means growth rates from different calendar times are being compared - when different control measures are in place.  Each variant will have a growth rate and transmissibility that decreases differently with increasing control measures.  Mixing different calendar periods in to the epocs in Figure 4A and 4B scrambles / looses the information on control measures and makes it hard to understand if those high day-on-day rates were unique to the "red" regions or to the calendar dates.

Yes their regional analysis for the south east in the pre-epoch align with the earlier estimates on transmissibility, which were a good warning. But given we're interested in estimating the biological transmissibilty of the new lineage I think the more recent matched regional growth rates (up to 2nd January) give a good indication that it's not *actually* as high as early estimates. Obviously different control measures will affect the apparent transmissibility, and it's always going to be difficult to disentangle the contributions of such measures from inherent transmissibilty.

Interestingly though to support the lower transmissibility of B.1.1.7 suggested by this preprint, the average increase in secondary attack rate for the new lineage is rougly 37% compared to other lineages as reported by PHE contact tracing teams.

> I hope I've given a clear explanation of why I don't think there is actually a conflict in their findings on the growth rates and transmissibility with the LSHTM pre-print.  It definitely still raises questions to me over the age dependancy and the Ct value vs some of the other findings and speculation.

I agree, I don't think there's any real conflict in the findings. I just think it raises further questions about the actual biological transmissibilty of the lineage, how much was it's early estimated value influenced by other factors in early-mid Decmeber, and if it is more transmissible how does it manage this?

On the latter point, the viral load doesn't seem to be significantly different compared with other lineages so increased transmission not likely due to increased shedding. The prepint suggests it may be due to the increased binding of ACE2 receptors therefore requiring a lower amount of viral particles to cause infection. Who knows though.

> It's great to see the sheer amount of work being pulled together so quickly and openly on this from the scientific community.  It's just as well there is such a strong and diverse academic and medical community in the UK that was ready to drop everything and get on this; I hope the government appreciate how we'd be stumbling in the dark without this scientific community, and looks at future RCUK budgets accordingly.

I agree it's been great to see, but I won't hold my breath over the future funding though especially with regard to basic/fundamental research. We can hope though.

 dwisniewski 18 Jan 2021
In reply to Offwidth:

I wasn't aware that this was the case? At least I've not come across anything that would suggest this.

OP wintertree 18 Jan 2021
In reply to dwisniewski:

> I'm not entirely sure what you mean here RE the pillar 1/2 testing?

These authors use many more self-reported symptoms to determine a "symptomatic" case than are used to select people for pillar 2 testing (12 vs 3).  So, their definition of "symptomatic" is different.  They could re-process their data with the definition used for eligibility for pillar 2 testing and see if the ratio of asymptomatic cases by that definition has changed - which is important as that is the definition used to select people for test/trace/isolate.   

> To me one of the strengths of this surveillance study is that it's not biased by symptomatic testing.

I think there is value in having both sorts of testing - and it's clear that the ONS and P1/P2 data does not have a fixed correspondence over time.  I see the random sampling as a great strength for information on societal prevalence, but the lack of that fixed correspondence makes it harder to use to understand what's going on with testing, which is a key part of what's going to help release restrictions.  I also wonder how well they can really control for the selection bias of those who choose to participate.

> Also regarding what they say about asymptomatic transmission, I read it as saying that asymptomatic transmission accounts for a large proportion of cases as is the case for other lineages.

Yes, that was my understanding too, but it felt to me like they were hinting that it could play more of a role.  As I said, perhaps I was way over- or mis- interpreting that - other parts as well as the part you quoted.  It read to me a bit like they thought it was a more serious driver of transmission but couldn't support that.  Probably all in my head.

> I think the differences are likely due to the period of time that testing was done. New exponential growth of a lineage, more cases that have newer infections and therefore higher viral loads. They do report the drop in Ct values in early Dec, that had first been reported around 20th December.

Yes, that was my understanding.  But this mechanism for Ct starting higher and lowering only applies to a random sampling approach in a rising then level/decay phase - which this study was.  I don't believe that this mechanism would translate to symptomatic pillar 2 testing, where people all go for testing at a similar, early stage in their infections.  So, whilst this theory explains the higher early on Ct value in their data, that doesn't explain the higher Ct reported by NERVTAG from symptomatic P2 data.  (Edit:  When, unless we assume there is a significant random sampling component to P2, which is possible given that the symptoms used for entry capture > 20x as many people as are detected as infectious, so people coming in with non-covid induced symptoms and non-symptomatic covid could be effectively randomly sampled...?)

I'm not aware of data of Ct vs time from infection for either variant, so it's all scrabbling around in the dark, but I do think they could contextualise their observation on random sampling data to qualify how compatible it is with symptomatic sampling.  But I regard Ct values and PCR in general as a bit of a dark bimolecular art. 

> Yes their regional analysis for the south east in the pre-epoch align with the earlier estimates on transmissibility, which were a good warning. But given we're interested in estimating the biological transmissibilty of the new lineage I think the more recent matched regional growth rates (up to 2nd January) give a good indication that it's not *actually* as high as early estimates. Obviously different control measures will affect the apparent transmissibility, and it's always going to be difficult to disentangle the contributions of such measures from inherent transmissibilty.

Yes, even with the best visualisations of the data and the timelines, it's like trying to read digital tealeaves to unpick all of this.  Having slept on it, I just don't like their production of an "average" value from data from different regions with each region being represented in the average by a different period in time.  I think that's an over-reach.  What I'd really like to see is a doubling time vs data estimate for old and new variants on a regional basis.

My take is that with lax control measures it really is a lot more transmissible, but that it responds strongly to them - although this is called in to question by how Kent and Medway remained exponential under Tier 4.  

> I agree, I don't think there's any real conflict in the findings. I just think it raises further questions about the actual biological transmissibilty of the lineage, how much was it's early estimated value influenced by other factors in early-mid Decmeber, and if it is more transmissible how does it manage this?

Totally agree.  This is going to be really important to understand for the unlocking process which I imagine will start in earnest when vaccinated down to 60 years of age or so.  

> On the latter point, the viral load doesn't seem to be significantly different compared with other lineages so increased transmission not likely due to increased shedding. The prepint suggests it may be due to the increased binding of ACE2 receptors therefore requiring a lower amount of viral particles to cause infection. Who knows though.

Yes; this was a  point of unrecognised conflict I thought I saw in the prerint. They also discussed how the lack of worse health outcomes is consistent with their not being a higher viral load, but to my mind the health consequences a product of (viral load ingested) x (probability of an individual virus particle binding) (*) as it's a numbers game and both affect the number that invade the host.  

However, we found no evidence that Ct values (a proxy for viral load) were intrinsically substantially lower in SGTF-positives [...] consistent with observations that B.1.1.7 infection is not more severe 

So more binding affinity would have a similar effect to a larger viral load ingested, so there's a contradiction to my mind between increased binding affinity and lack of more severe infection.   I suppose this critically depends on what the non-linear saturation points are in the infection pathways and if viral load actually matters a lot or not to the severity of an infection that makes it past the initial endothelial cells. 

(* - okay, really it's a more involved probabilistic calculation...)

> I agree it's been great to see, but I won't hold my breath over the future funding though especially with regard to basic/fundamental research. We can hope though.

After the original SARS, a shiny new CL-3 lab was built at great expense nearby.  Interest and funding dried up and it was mothballed.  It seems to have been stripped down, which is a shame if you're in the market for a CL-3 facility...

Post edited at 13:45
 dwisniewski 18 Jan 2021
In reply to wintertree:

>.....  I also wonder how well they can really control for the selection bias of those who choose to participate.

This is an important point. In lots of situations during the pandemic it seems that certain demographics are continually underrepresented.

> Yes, that was my understanding.  But this mechanism for Ct starting higher and lowering only applies to a random sampling approach in a rising then level/decay phase - which this study was.  I don't believe that this mechanism would translate to symptomatic pillar 2 testing, where people all go for testing at a similar, early stage in their infections.  So, whilst this theory explains the higher early on Ct value in their data, that doesn't explain the higher Ct reported by NERVTAG from symptomatic P2 data.  (Edit:  When, unless we assume there is a significant random sampling component to P2, which is possible given that the symptoms used for entry capture > 20x as many people as are detected as infectious, so people coming in with non-covid induced symptoms and non-symptomatic covid could be effectively randomly sampled...?)

> I'm not aware of data of Ct vs time from infection for either variant, so it's all scrabbling around in the dark, but I do think they could contextualise their observation on random sampling data to qualify how compatible it is with symptomatic sampling.  But I regard Ct values and PCR in general as a bit of a dark bimolecular art. 

Yes I wonder how symptomatic the symptomatic pillar 2 testing is? The initial Nervtag report using the sequecing data to infer viral load obtained samples through contact tracing so again might skew things a bit. But, you're right in that it's definitely worth investigating how compatible the new infection hypothesis is with symptomatic testing in general.

> Yes, even with the best visualisations of the data and the timelines, it's like trying to read digital tealeaves to unpick all of this.  Having slept on it, I just don't like their production of an "average" value from data from different regions with each region being represented in the average by a different period in time.  I think that's an over-reach.  What I'd really like to see is a doubling time vs data estimate for old and new variants on a regional basis.

I think there is something like this in the supplemental tables but using different time periods for SGTF or non-SGTF per region based upon their 'changepoint'.

> My take is that with lax control measures it really is a lot more transmissible, but that it responds strongly to them - although this is called in to question by how Kent and Medway remained exponential under Tier 4. 

Yeah very confusing.

> Yes; this was a  point of unrecognised conflict I thought I saw in the prerint. They also discussed how the lack of worse health outcomes is consistent with their not being a higher viral load, but to my mind the health consequences a product of (viral load ingested) x (probability of an individual virus particle binding) (*) as it's a numbers game and both affect the number that invade the host.  

> However, we found no evidence that Ct values (a proxy for viral load) were intrinsically substantially lower in SGTF-positives [...] consistent with observations that B.1.1.7 infection is not more severe 

> So more binding affinity would have a similar effect to a larger viral load ingested, so there's a contradiction to my mind between increased binding affinity and lack of more severe infection.   I suppose this critically depends on what the non-linear saturation points are in the infection pathways and if viral load actually matters a lot or not to the severity of an infection that makes it past the initial endothelial cells. 

My intuition on this points to something like you mention in your last sentence here. I suspect there are enough rate-limiting steps that the dynamics of the host immune response are sufficient to prevent runaway replication/infection of other cells (at least in those who were competent to mount such a response anyway). Although, I'm not sure this would still hold if someone was initially infected with a large dose of the virus.

Also I think it's been shown pretty well there is a good correlation between viral load and disease severity, which was also another reason why B.1.1.7 seemed a bit odd, as disease severity isn't worse compared with other lineages.

Post edited at 15:10
OP wintertree 18 Jan 2021
In reply to dwisniewski:

> This is an important point. In lots of situations during the pandemic it seems that certain demographics are continually underrepresented.

There's both that angle, and there's also a social one.  It seems how seriously one takes the threat of the virus and - to a frustrating degree - even really believes in the science around it - is pretty uncorrelated to demographics.  I'd guess that the ONS get more engagement from one end of that spectrum of emotional buy-in to the situation, and that this means they tend to get more careful people.  Then again, I imagine engagement with pillar 2 testing is similarly biassed - but hospital admissions and deaths are not.  Very wooly words from me, and they're absolutely not a criticism of the ONS team and their work.

> Also I think it's been shown pretty well there is a good correlation between viral load and disease severity, which was also another reason why B.1.1.7 seemed a bit odd, as disease severity isn't worse compared with other lineages.

Yes; all the different pieces are not making a compelling narrative here.  Perhaps there's a moving part nobody has identified yet.

> I suspect there are enough rate-limiting steps that the dynamics of the host immune response are sufficient to prevent runaway replication/infection of other cells (at least in those who were competent to mount such a response anyway)

Quite a few organoid groups are starting to get results out on Covid. Being able to do repeatable tests on this sort of thing is going to be really interesting/informative.

 Misha 19 Jan 2021
In reply to wintertree:

On a related topic, impressive to see vaccination at 300k+ a day on weekdays and still 200k+ on Sunday. There's more vaccination capacity coming on stream so these numbers could get higher. I fully expected the vaccine roll out to be a pig's ear but it's looking promising. I suspect it will be harder work getting the last 10-20-30% of each priority group (people who are less mobile, people who live in the sticks and so on), plus a few won't have it of course, but it's a good start.

 Offwidth 19 Jan 2021
In reply to Si dH:

Pretty sure the change in age demographic was from Speigelhalter on BBC news or R4. Obviously double a small number is still a small number.

Some more data on why vaccination of the oldest wont help as much with ICU admissions as some people think (even if it does cut deaths). Thanks to spidermonkey for the link on the other channel.

https://mobile.twitter.com/chrischirp/status/1350416428025962498?prefetchti...

Post edited at 10:54
 Offwidth 19 Jan 2021
In reply to Misha:

I think its going to be harder than the government realise to hit their target. I think they need to urgently flex boundaries unless supply is about to become a serious issue. Its better in my view to start vaccinating the keen and mobile over 70s rapidly than cause delays chasing the over 80s who are hard to access or reluctant to be vaccinated. Using up spare vaccine from no shows will get harder when everyone in the local NHS/care front line has been vaccinated from earlier no shows. Someone copied info from a tweet on the other channel showing a worrying graph of vaccinations this last week versus levels needed to hit targets.

 Si dH 19 Jan 2021
In reply to Offwidth:

> Its better in my view to start vaccinating the keen and mobile over 70s rapidly than cause delays chasing the over 80s who are hard to access or reluctant to be vaccinated.

Isn't this exactly what they announced yesterday? Over-80s and care home residents still the priority but moving on to over-70s and those who are clinically extremely vulnerable as well.

Post edited at 13:01
mick taylor 19 Jan 2021
In reply to Misha:

Sure is impressive. I reckon a conservative estimate: 350,000 midweek vaccines for 20 days = 7 million, 300,000 weekend vaccinations = 2.4 mill, 9.4 total. Not too hard to get this up to 11 mill and they would hit the mid Feb target of 15 mill. 

OP wintertree 19 Jan 2021
In reply to mick taylor:

Yup.  I'm not sure sufficient information is publicly available to properly evaluate different vaccination strategies.  At some point one has to be picked and run with.  I'm quietly hopeful that we're going to see the effort ramp up significantly beyond the 2 m per week target - this is kind of implied by the need for double that rate in 12 weeks time to allow first rounds to continue whilst second rounds also go ahead.  

It's nice to have something to be universally positive about - amazing what happens when money is poured in to existing national, regional and local infrastructure rather than going to some new company brought into existence within the chumosphere.  

1
mick taylor 19 Jan 2021
In reply to wintertree:

> .....amazing what happens when money is poured in to existing national, regional and local infrastructure rather than going to some new company brought into existence within the chumosphere.  

When the time is right (which may include now if Kier uses his charm), this ^ needs to be bigged up massively by Tory opponents. 

1
 lithos 19 Jan 2021
In reply to Misha:

> On a related topic, impressive to see vaccination at 300k+ a day on weekdays and still 200k+ on Sunday. 

numbers dropping a bit, i guess the logistics are getting complicated. Still impressive

Sun : 225K

Mon: 204K (208K inc 2nd dose)

https://coronavirus.data.gov.uk/details/vaccinations

mick taylor 19 Jan 2021
In reply to lithos:

The dip is a bit odd. Let’s see how things develop over next few days. 

 Misha 19 Jan 2021
In reply to mick taylor:

The mid Feb target is just a staging post. If they hit it by the end of Feb, that's still good going at this early stage and dealing with a largely OAP clientele. The real challenge will be sustaining and hopefully growing the vaccination rate to get at least one jab to everyone who wants it by the end of the summer. That would still be a great effort (considering first round jabs would need to have been given to a load of people by then).

Unless they start massively delaying 2nd doses, 2m a week would me the 25m most vulnerable jabbed by Easter, then it's the same people getting 2nd doses by the end of June. The remaining 25m adults (allowing for c. 10% who won't or can't have it) would then be done by the end of September, with second jabs by the end of the year. Any improvement on the 2m a week would accelerate this.

Only 200k reported today but it seems that there are some reporting delays going on. It's not clear from the brief explanation provided on the site whether the numbers lag by a day or not. Let's hope tomorrow's numbers are better.

Post edited at 19:12
In reply to wintertree:

> Yup.  I'm not sure sufficient information is publicly available to properly evaluate different vaccination strategies.

It's interesting that the Tories have been demanding the SNP explain 'failing' on vaccination and asking for transparency but when the SNP published their plan with the vaccine delivery schedule they immediately demanded it was taken down citing 'commercial confidentiality'.

Fortunately, deleting stuff off the web after its been up for a few hours doesn't work so the vaccine delivery schedule for Scotland is available.

https://twitter.com/fatweegee/status/1351583237454389249

The thing that stands out for me is 450k initial doses of Pfizer vaccine but if you count forward 12 weeks to when the second doses for those 450k people will be needed (according to the UK government's extended schedule, Pfizer says 3 weeks) there's only 78k doses of Pfizer per week arriving.

So if we go ahead and shove the initial 450k doses into arms as fast as possible how do we get the second doses on schedule.

I suspect the people who devised the Scottish and Welsh vaccination programs were considering future deliveries along with JCVI recommendations on the care home residents and over 80s first and building a program which is scaled to deliver vaccinations at a rate which allows for the second dose as well as the first.   

When you look at the delivery schedule and think about second doses it's a bit more complicated than getting the initial supply into arms as fast as possible.

4
 lithos 19 Jan 2021
In reply to Misha:

the site says for the 18th so in effect yesterday until midnight.

Todays(19th) not available until tomoz.  I guess there maybe reporting delays (i would hope it was largely automated - barcode/qr scanners etc)

Number of people who have received a COVID-19 vaccination, by report date.

Data are reported daily, and include all vaccination events that are entered on the relevant system at the time of extract. Data are presented for vaccinations carried out up to and including the end of the report date.

 Offwidth 20 Jan 2021
In reply to Si dH:

That's not what I heard. They said they would be moving onto that demographic very soon but were concerned with some parts of the UK being ahead of others and they urgently needed to finish the older cohort in those areas first (because constituents were moaning). Now the information I posted about the decline in vaccinations in the last three days is headline news in some of tomorrow's papers. I really thought this was a promise they would struggle to cock-up, as it's largely being run by the public sector. I now wonder.

Every day John Crace's humourous columns become less satire and more reportage.

https://www.theguardian.com/politics/2021/jan/19/uk-proves-world-beating-at...

The 46% increase in care home deaths a month after we have a vaccine is plain dreadful.

https://www.theguardian.com/world/2021/jan/19/covid-related-deaths-in-care-...

The other big news is the government have abandoned mass testing in schools. Dom's grand plans of moonshots turns to dust (along with the money spent on it).

Post edited at 00:27
 Misha 20 Jan 2021
In reply to lithos:

Yes, so the question is whether there is a reporting delay. Data for the 18th (presented on Tues 19th) could actually be Sunday's data. It's not clear what "all vaccination events that are entered on the relevant system at the time of extract" really means. I doubt it's entirely automated. There could be a reporting lag due to someone manually inputting data from multiple locations (it wouldn't surprise me) or at least checking it. 'Entered on the relevant system' suggests manual entry to me. If tomorrow's numbers go back up to 300k+ then it's just a reporting delay.

 Offwidth 20 Jan 2021
In reply to Misha:

What the plot below doesn't show is the planning was reportedly around 400,000 a day by now when you factor in the second shots can't be left too long. Even at first dose we need 2 million a week (300,000 a day). To be clear even if we miss the target we will have done a very good job overall (with provisos like those care home deaths maybe could have been reduced quite a bit) but the information on plans doesn't look very transparent and Tom is right to highlight what happened when the Scottish government did publish additional planning information.

https://coronavirus.data.gov.uk/details/vaccinations

mick taylor 20 Jan 2021
In reply to Misha:

It’s plain weird to go from 300k+ a day to well under 300k BUT with more vaccination centres (10 more super hubs opened Monday). 
OR it may be that the large figures reported end of last week were because of the lag (similar to high deaths reported Tuesday after the weekend).

Im opting for a reporting issue and 300k+ have been jabbed yesterday. There has been database issues. 

Post edited at 09:51
 neilh 20 Jan 2021
In reply to tom_in_edinburgh:

Anybody involved in manufacturing or assembly would want to know the reported wastage rate.

I wonder how they are measuring that.( it will never be publically available at the moment)

To me that is the most useful number.Particularly how they measure it. I have heard of one small vaccintation surgery who forgot to turn their fridge on , and had to destroy all the vaccines stored  as a result.So 1 days appointments were scrapped.

These things build up nationally and need to be nailled to the floor to eliminate wastage. Very hard to do with a huge rollout. A bit of automotive lean processing would help.

 neilh 20 Jan 2021
In reply to mick taylor:

Not weird. Just in time logistics on a national rollout with a variable manufacturing output would suggest an interesting logisitical and stretched excercise.

Post edited at 10:10
 elsewhere 20 Jan 2021
In reply to neilh:

Assuming wastage rate = (doses-jabs)/doses  the wastage rate of 10-20% is negative!

Pfizer-BioNTech is in 5 dose bottles but there's usually enough for 6 jabs.

Oxford-AZ is in 10 dose bottles but there's usually enough for 11 jabs.

More seriously I heard 1% wastage being discussed and they are using the extra dose in every bottle.

 neilh 20 Jan 2021
In reply to elsewhere:

Wastage encompasses a huge range of sins from delivery to the wrong address, fridges not being turned on, vehicle breakdown and so on.The list of scenarios goes on.Every failed event where a phial of vaccine cannot be injected as it is scrapped is waste.

Bet its higher than 1%

 Offwidth 20 Jan 2021
In reply to neilh:

I agree entirely.... yet as its inevitable, too much focus on wastage is dumb and potentially counter productive. All the information I have implies very sensible local use of spare vaccine.

Just been chatting to someone in the Lakes where all the over 80s in the combined GP vaccination area were done so they offered remaining supply to slightly younger demographics. If supply is available it's better in my view to let them carry on rather than making them wait. Most people in hospitals are under 80 and avoiding hospital overload is supposed to be important.

Post edited at 11:37
In reply to neilh:

The wastage is one thing but for me the two biggest questions are:

1.  If you vaccinate n people per week at the start then to keep the same number of unvaccinated people being vaccinated per week after 12 weeks you need to vaccinate 2n people per week (n second doses and n first doses).   If you start out at max capacity and vaccinate 2n people a week all that happens is after 12 weeks your entire capacity is giving out second doses.

2. Your vaccine supply needs to provide second doses as well as first.  But if you look at the data published by the Scottish Government, particularly for Pfizer, it doesn't look like that.  There's a huge initial delivery, a few largish deliveries and 12 weeks out there's a weekly much smaller delivery.  That looks like a reasonable schedule for the original plan of two doses 21 days apart and holding back vaccine for the second dose.   It doesn't match at all with sticking all the initial supply into arms as fast as possible.   There's not enough supply 12 weeks out to provide the second doses for the initial surge.

https://twitter.com/fatweegee/status/1351583237454389249

The particularly hypocritical aspect to this is that the Scottish and Welsh governments are getting slagged off by the Tories for 'falling behind' England at the same time as the Tories are insisting the delivery data is confidential.  It actually looks like the Scottish and Welsh campaigns have tried to match up with the delivery schedule and allow for doubling capacity later taking account of second doses while the English campaign is just putting vaccine in arms.  Which may well mean that second doses don't happen on time or are of a different vaccine.   That might be a reasonable trade off given the dire situation in England but it is a trade-off and they need to be honest about the downside of their big initial surge and definitely not crow about being 'world beating' and Scotland and Wales 'falling behind',

Post edited at 11:39
4
 Offwidth 20 Jan 2021
In reply to tom_in_edinburgh:

I agree... it's dirty politics and should be about science and logistics.

If there are supply issues the government need to own up and explain it may be longer than they thought before people get their second dose. If not, tuning the logistics by stopping supply to faster areas and providing it to slower areas, where vaccination capacity may not yet exist to use it, is not helpful. It's much better to tune delivery a bit more thoughtfully based on capacity elsewhere speeding up first without stopping or significantly slowing the faster areas.

1
mick taylor 20 Jan 2021
In reply to neilh:

The drop from Friday 15th to Monday 18th is massive, especially considering we had more vaccination centres on Monday. It’s a 30% reduction when it should have been an increase. And we saw just about steady day on day increase until this weekend so the logistics did appear to be working.


In reply to mick taylor:

Repost from t'other thread:
I don't know about you but what I'm gleaning from the actual information hidden between paragraphs of overhyped sensationalism, is that there is a huge supply of the AZ product but the batch testing and release is taking a long time. So right now it is a supply issue, but if what I understand is correct we should see a lot of product coming out the end of that process in the next week or two. Given that we have seen a lot of the infrastructure built up and people put in place ready to go, we could see things pick up immensely.
Or I could have interpreted wrong and we're hosed. One or other.

 neilh 20 Jan 2021
In reply to Offwidth:

Tuning delivery when there are probably random deliveries will not be easy logistically even with a real time delivery network..Its not easy to simply switch a consignement even with logistics being really good.

And Pfzier have to contend with European demands as well.I would not like to be in their logistics shoes.I can imagine alot of heated phone calls etc all the way to the top.

 neilh 20 Jan 2021
In reply to mick taylor:

That is nothing. You want to see how big companies manage just in time flows.

Not easy.

You are only getting a small picture of what is going on. That reduction was probably built into the system 2 or 3 weeks ago ( maybe longer) and its now only showing up. You need to back track the flow alot more to get the true picture.

OP wintertree 20 Jan 2021
In reply to thread:

Tuesday evening's data release is the last chance to get a good look at plot 17 until Friday, as the last weekend is working it's way through the reporting lag and has too-low sample numbers, introducing bias, and there's not the information to unpick it until the following Monday's data clears through.

As I said in the first post about the rate of decay in cases... "I wouldn't read too much optimism in to this rate until next week's update"...  I think the update this Friday will probably refine the rate to a somewhat lower - but still appreciable - level.  It looks like the "weekend effect" is particularly strong at the moment, with the delayed sample reporting from the weekend then being fed out most on a Monday and least on a Friday, creating an additional downwards ramp on top of the real decay.  You can see this flipping between the "shunted" and "shunted, de-weekended" plots below.  The first tries to fix the significant reporting lags around the recent bank holiday weekends and the later tries to fix the sat/sun under-sampling and mon/tue over-sampling.   You can see that the giant jagged edges of the weekends are still somewhat present - but muffled - in the "best" version, so they're still having a biassing effect on the rates. As we get another week of data on the real decay that biassing will be largely removed.

All the "red" regions are still clearly in decay despite this effect; some of the "blue" regions are worryingly close to level behaviour - notably the East and West Midlands.  We'll see what Friday brings...  The next barometer will be compering successive Saturday's then Sunday's data on the government dashboard for the regions

Also below is an update to the hospitals occupancy plot.  You can see the curve bending towards the vertical - this is where hospital occupancy starts levelling off, but ITU admission increases for another week or so as people already in hospital move towards more critical care.  

I'm thinking that sequencing capacity should be prioritised to areas that are seeing the slowest exponential decay rates, along with boots on the ground efforts to understand the problem.  We watched Kent and Medway behaving out-of-character on here for over a month before the new variant was identified; with the number of "worse" variants rising internationally and the exceptionally high prevalence in the UK right now, this is the time for the precautionary principle writ large. 

Post edited at 13:37

In reply to Offwidth:

> If there are supply issues the government need to own up and explain it may be longer than they thought before people get their second dose. 

I don't think what is happening with regard to Pfizer is really 'supply issues'.  The supplier is doing what was agreed.  The problem is the Tories switched the strategy completely and the purchasing and delivery schedule doesn't match the new strategy. 

The sh*t hits the fan in 12 weeks when they have to find vaccine and vaccinators to handle the second doses while still giving out first doses to unvaccinated people.

2
mick taylor 20 Jan 2021
In reply to neilh:

Good points

For folks info: 300,000 vaccinated yesterday. 1185 hospital deaths reported (highest figure).

Im guessing a week or two of increases in deaths.

mick taylor 20 Jan 2021
In reply to tom_in_edinburgh:

Regarding the supplier: I think Pfizer are revamping their factory and have stated ‘expect delays’ so don’t know if this will Impact future deliveries and vaccinations. 

 neilh 20 Jan 2021
In reply to tom_in_edinburgh:

Mmm....you are not involved in any form of manufacturing and logistics are you. The complete supply chain on this is quite mindboggling complicated.At least in say putting an aircraft together you are doing it on one site.Try working it through from starting point of manfacturing through to injection in arm at multiple sites..Even just having PPE in place for vaccinators is part of the whole process on the basis its new gloves etc on every jab as an example.

Post edited at 15:40
 elsewhere 20 Jan 2021
In reply to neilh:

On the programme I saw they reckoned wastage was normally assumed 5% but thought it was closer to 1%.

https://www.ukhillwalking.com/forums/off_belay/friday_night_covid_plotting_8-7...

https://twitter.com/fatweegee/status/1351583237454389249

Looks like the official assumption is exactly 5% (for the week that I checked).

Vaccines split within 4 nations by population so UK supply should be twelve*  times Scottish supply.

*66.65M/5.454M

Edit: adding up the doses, there's enough for everybody by late May - one dose only.

Post edited at 16:31
In reply to neilh:

> Mmm....you are not involved in any form of manufacturing and logistics are you. The complete supply chain on this is quite mindboggling complicated.At least in say putting an aircraft together you are doing it on one site.Try working it through from starting point of manfacturing through to injection in arm at multiple sites..Even just having PPE in place for vaccinators is part of the whole process on the basis its new gloves etc on every jab as an example.

Why are you telling me this?   I don't disagree with it but the point I was making is much simpler.

I am looking at the official schedule for when vaccines will be delivered in Scotland (I assume the English situation is the same except 10x larger) and it is totally obvious the delivery schedule from week 12 on does not match up with using the initial supply for first doses. It would be surprising if it did because the UK government completely changed its strategy.

It is a totally simple argument.   They get 450k doses of Pfizer right at the start and then a few 130k dose deliveries but by week 12 it is 78k doses a week.

If they chuck 450k doses of Pfizer as first doses into people's arms as fast as they can find people to do it they do not have matching supply of Pfizer for the second doses 12 weeks later.   They need to moderate the rate they do the injections and probably hold back some vaccine for second doses or they need to delay more than 12 weeks or use a different vaccine for dose 2.

Post edited at 16:10
 neilh 20 Jan 2021
In reply to tom_in_edinburgh:

Yep and Pfizer have basically thrown a spanner in the works( the  official schedule)  by knocking back deliveries( across Europe as well), which shows that no matter what it can all go pear shaped.

Post edited at 16:14
In reply to neilh:

> Yep and Pfizer have basically thrown a spanner in the works( the  official schedule)  by knocking back deliveries( across Europe as well), which shows that no matter what it can all go pear shaped.

I agree, it's a brand new product and a factory trying to scale up and there could easily be delays to the schedule to deal with.  One way to react to uncertain supply is to hold some stock as a buffer.

The thing I object to is the way the Tories are claiming 'world leading' and implying the Scottish and Welsh governments are failing by not injecting at the same rate as England when, if you look at the details of the delivery schedule and consider second doses there's actually very good reason for going slower initially.

There's a reasonable argument to be had about which strategy is the best trade off for England given the extremely high infection rate.  What isn't acceptable is the Tories claiming 'commercial confidentiality' on the delivery schedule so the people they are attacking can't defend themselves with the data and point out the consequences of using all the initial supply of Pfizer for first doses.

I suspect what they want is to bully the Scottish and Welsh governments into taking the same path as England because if they don't in a few months when it is time for second doses the media which is shouting about Scotland and Wales 'falling behind' is going to be shouting about people in England not getting a second dose of the Pfizer vaccine on schedule when people in Scotland and Wales are.

5
mick taylor 20 Jan 2021
In reply to mick taylor:

1820 deaths with 28 days of positive test. Grim. 

 neilh 20 Jan 2021
In reply to tom_in_edinburgh:

I am not sure that holding stock as a smoother buffer really helps in any shape or form at the moment. You need the AZ vaccine on line as well that is the real game changer.

Anyway if you think its bad here there are alot of European countries up in arms over delivery delays.And considering where we stand in terms of rates of vaccination compared with others we are overall in a reasonable place irrespective of ongoing glitches.

We have months to go on this.I am not going to get aggitated over the current position.Day to day watching of the vaccination number is not a good practise for the good of the sole so to speak.

OP wintertree 20 Jan 2021
In reply to mick taylor:

> 1820 deaths with 28 days of positive test. Grim. 

Bolstered by reporting lag from a weekend, but not a good number.  I think that's going to be the biggest number that we see in this "wave", with actual deaths/day maxing out at around 1100.

In reply to mick taylor:

> 1820 deaths with 28 days of positive test. Grim. 

Absolutely appalling. Words fail me. 

In reply to neilh:

> I am not sure that holding stock as a smoother buffer really helps in any shape or form at the moment. You need the AZ vaccine on line as well that is the real game changer.

Obviously the AZ vaccine is the one the UK is banking on and has ordered most of so it is the bet which needs to come off.

But the Pfizer situation is separate, it's the one that arrived in volume first and it's the one that's driving the big vaccination numbers for the last few weeks.   Without matching supplies of Pfizer they have a problem in 12 weeks.   One part of the solution could be to hold back some stock of Pfizer to smooth supply and provide for some of the second doses.

Obviously, people far more competent than me will have been planning for this and all the countries England, Scotland and Wales presumably had well thought out plans a few weeks ago to balance vaccinations with deliveries and allow for second doses.  England has thrown away that playbook by going for maximum doses in arms as fast as possible.

In reply to Deleated bagger:

> Absolutely appalling. Words fail me. 

It will get worse; we're still working the peak cases through infection/hospitalisation/death pipeline.

Hospitalisation numbers just seem to have passed the peak. Two weeks or so to peak deaths.

At least the cases are falling; the BBC map has only two 'old blood' coloured areas (cases > 1000/100k); Slough and Knowsley.

Post edited at 17:13
In reply to captain paranoia:

> > Absolutely appalling. Words fail me. 

> It will get worse; we're still working the peak cases through infection/hospitalisation/death pipeline.

> Hospitalisation numbers just seem to have passed the peak. Two weeks or so to peak deaths.

> At least the cases are falling; the BBC map has only two 'old blood' coloured areas (cases > 1000/100k); Slough and Knowsley.

My friends who work in adult ICU are shattered and traumatised. They can see no end to it.

 Si dH 20 Jan 2021
In reply to wintertree:

> I'm thinking that sequencing capacity should be prioritised to areas that are seeing the slowest exponential decay rates, along with boots on the ground efforts to understand the problem.  We watched Kent and Medway behaving out-of-character on here for over a month before the new variant was identified; with the number of "worse" variants rising internationally and the exceptionally high prevalence in the UK right now, this is the time for the precautionary principle writ large. 

Do we know how long it takes from a positive test being recorded in a lab and the individual being notified, to that case being fully sequenced such that the specific variant could be identified and added in to a database of some kind for a given area or region? I don't know whether the system is currently fast enough to do what you suggest. I agree it sounds sensible. Would also like to see it focused on areas that have seen exceptionally high spikes in recent weeks or that have been showing high overall growth rates despite low SGTF rates.

mick taylor 20 Jan 2021
In reply to captain paranoia:

> > Absolutely appalling. Words fail me. 

> It will get worse; we're still working the peak cases through infection/hospitalisation/death pipeline.

> Hospitalisation numbers just seem to have passed the peak. Two weeks or so to peak deaths.

I agree. And my gut feeling is that due to the health inequalities of areas outside the high infection areas of London/SE we may see high death rates in these areas (Merseyside, G Manchester, NE etc etc ).

Post edited at 17:24
OP wintertree 20 Jan 2021
In reply to Deleated bagger:

> My friends who work in adult ICU are shattered and traumatised. They can see no end to it.

Hospital admissions have just about turned into decay at national level I reckon, so ITU admissions and then occupancy should follow over the next couple of weeks.   Hopefully this will show through in Friday’s plot updates.

I can’t imagine the experience of working there now.  I hope your friends and their colleagues have as much support as possible, and that ongoing, longer term support is being put in place for the come down from this. There’s going to be a lot of very traumatised medical professionals coming out of this.  I tried to spread awareness of the potency of this new variant in channels beyond UKC; I don’t think it made a jot of difference.  

 Si dH 20 Jan 2021
In reply to wintertree:

> Bolstered by reporting lag from a weekend, but not a good number.  I think that's going to be the biggest number that we see in this "wave", with actual deaths/day maxing out at around 1100.

Help me with this. At https://coronavirus.data.gov.uk/details/deaths it is clear deaths by date of death based on the 28-day criteria are just about passing the peak seen last April (I ignore the reporting date numbers, which are completely useless.)

However if you look at the graph of deaths where covid was recorded on the death certificate, it suggests that we are still at this point well below the current peak of deaths that occurred in April. Unfortunately, the data is only available on website by date of registration. Do you know if the download includes this data by date of death?  It's not clear whether the big difference between behaviour is entirely down to lag and noise in the death registration date data or whether it's because there is also a significantly smaller proportion of cases being missed by testing this time and hence the 28 day numbers will be much closer to the real death tally than they were last spring. I'm sure both contribute to an extent.

Post edited at 17:35
OP wintertree 20 Jan 2021
In reply to Si dH:

It’ll be quite late before I can read thought the data and get back in both posts.  The death certificates update just out looks like it’s got a big historic release in it that eclipses the last peak height without redistribution.  I’ve never looked in to the structure of this data before so can’t comment more...  screenshot below 


 Si dH 20 Jan 2021
In reply to wintertree:

> It’ll be quite late before I can read thought the data and get back in both posts.

Don't worry about it too much on my behalf!

> The death certificates update just out looks like it’s got a big historic release in it that eclipses the last peak height without redistribution.  I’ve never looked in to the structure of this data before so can’t comment more...  screenshot below 

That's weird, when I go to the link it still shows that latest peak topping out at 6k rather than 12k.  Must be an issue with the website or my device.

 SouthernSteve 20 Jan 2021
In reply to Deleated bagger:

https://www.bbc.co.uk/news/uk-55738174

> Absolutely appalling. Words fail me. 

I agree - so awful. We have the highest death rate per capita in the world. We mock the USA at our peril.

mick taylor 20 Jan 2021
In reply to wintertree:

This is something I don’t understand. The last seven days we have had an average of 1223 deaths/day within 28 days of a test (today’s being 1860). The figure of 1223 balances out the weekend reporting lag. And i get that this figure includes those deaths that have just been reported but occurred earlier and missed off the stats. So even if we subtracted these deaths from 1223 then there will be deaths reported in the next 28 days that occurred yesterday, adding to the 1223 figure.  If anything, the actual people dying could be higher than 1223 yet you say it’s less. What am I missing?

In reply to wintertree:

> I can’t imagine the experience of working there now.  I hope your friends and their colleagues have as much support as possible, and that ongoing, longer term support is being put in place for the come down from this. There’s going to be a lot of very traumatised medical professionals coming out of this.  I tried to spread awareness of the potency of this new variant in channels beyond UKC; I don’t think it made a jot of difference.  

The footage on the BBC taken in an ICU of staff proneing a patient brought it back to me. I've worked in PICU where it would take 7-8 staff to do this procedure. That's commitment. 

​​​​

In reply to Deleated bagger:

> My friends who work in adult ICU are shattered and traumatised. They can see no end to it.

Send them my love and thanks.

The end is coming, provided Johnson and co don't f*ck it up again.

1
In reply to Deleated bagger:

> The footage on the BBC taken in an ICU of staff proneing a patient brought it back to me

Clive Myrie's reports from the Royal London? Yes, heartbreaking. I posted a link on the 'anti-virus' thread.

I'd do a clockwork orange job on Johnson, Mogg, etc; strap them in a chair with their eyes peeled and force them to watch it over and over. Trouble is, I don't think it would register with Mogg, privileged sociopathic parasite that he is. 

I'm still bloody furious. Can you tell...?

1
In reply to captain paranoia:

Someone dislikes love and thanks ..?

OP wintertree 20 Jan 2021
In reply to mick taylor:

> This is something I don’t understand. The last seven days we have had an average of 1223 deaths/day within 28 days of a test (today’s being 1860). The figure of 1223 balances out the weekend reporting lag. And i get that this figure includes those deaths that have just been reported but occurred earlier and missed off the stats. So even if we subtracted these deaths from 1223 then there will be deaths reported in the next 28 days that occurred yesterday, adding to the 1223 figure.  If anything, the actual people dying could be higher than 1223 yet you say it’s less. What am I missing?

I agree with you on the 7-day average by reporting date.  I haven’t looked at the lag distribution of recent reporting numbers; my suspicion is that the lag is worst around peak deaths which is about where we are now.  Could be more like 1200 than 1100 for the peak but I’m hoping that’s about it.  Maybe I’m wrong.  But I think deaths are about at their peak as of now.  Often reporting numbers higher than actuals means there’s a rise in actuals coming, sometimes it means lag.  

mick taylor 20 Jan 2021
In reply to wintertree:

Thanks for this. Might get my friend SummerShrub to do some science n that and do one of his graphs

 minimike 20 Jan 2021
In reply to mick taylor:

https://medicalxpress.com/news/2021-01-safrica-virus-strain-poses-re-infect...
 

this looks like particularly bad news. Can’t see an actual article yet but indicates 8 fold reduction in antibody efficacy with the SA variant (not the uk one), which is the threshold for seasonal flu vaccine tweaking apparently.. looks more and more endemic. Combined covid/flu vaccine each winter??

OP wintertree 20 Jan 2021
In reply to minimike:

Bad not not unexpected news given the number of changes to the receptor binding domain and that likely being the most effective location / epitopes for neutralising antibodies.

How long do you think it’ll take to go native here with our highly porous borders consisting of a single, randomly timed PCR test (50% false negative rate?) and a self-enforced quarantine?

We need hard borders, now, until the risk is better bounded and cases are low enough that we can detect and mop up outbreaks of any strain.

In reply to wintertree:

> We need hard borders, now, until the risk is better bounded and cases are low enough that we can detect and mop up outbreaks of any strain.

That is still only ever a delaying tactic. The answer to get on with that tweaking of the vaccine that's supposed to be not too hard......

Hopefully it doesn't need the full set of trials, otherwise.... same procedure as last year?

 Si dH 20 Jan 2021
In reply to wintertree:

> Bad not not unexpected news given the number of changes to the receptor binding domain and that likely being the most effective location / epitopes for neutralising antibodies.

> How long do you think it’ll take to go native here with our highly porous borders consisting of a single, randomly timed PCR test (50% false negative rate?) and a self-enforced quarantine?

> We need hard borders, now, until the risk is better bounded and cases are low enough that we can detect and mop up outbreaks of any strain.

In the short term, the key question here is presumably whether the South African variant is more transmissible than the UK variant. If it isn't, then we shouldn't see it taking hold over that variant in any significant numbers (if at all) until most of the population have been vaccinated and hence have protection against the UK variant but which is weaker against the SA variant. That is probably well in to summer. Hopefully, that is long enough to develop a modified vaccine solution to deal with the SA variant if required.

(From a selfish UK perspective of course.)

Post edited at 19:26
 minimike 20 Jan 2021
In reply to wintertree’s

> How long do you think it’ll take to go native here with our highly porous borders consisting of a single, randomly timed PCR test (50% false negative rate?) and a self-enforced quarantine?

I’d be amazed if that time is still in the future tbh.

 minimike 20 Jan 2021
In reply to Si dH:

I’m not sure about that. If the cross immunity is significantly reduced (and it may not be because T cell immunity is different to antibody which is what was measured int hat study) then they may act as ‘separate’ pandemics rather than competing significantly. That’s not a nice idea. Wish I hadn’t gone there now!

OP wintertree 20 Jan 2021
In reply to Longsufferingropeholder:

> That is still only ever a delaying tactic. The answer to get on with that tweaking of the vaccine that's supposed to be not too hard......

I'm not sure.  If we can't get this under control globally, then the variants are just going to keep coming; this is not like the flu in that respect.  We know this family of viruses has the ability to be at least 10x more lethal at all ages, and it looks like it can become more transmissible.  There's no wisdom in anthropomorphising a virus, but still it feels to me like this one is trying to be more like the original SARS. 

In reply to Si dH:

> In the short term, the key question here is presumably whether the South African variant is more transmissible than the UK variant

I partially agree, but as soon as we start vaccinating we alter the relative growth rates of the two variants - and even a small relative change makes a big difference to exponential growth down the line; I reckon after 90 days of vaccinating against one of two identically transmissible, equally prevalent strains, prevalence could be tipped to to the other one by 5:1 or so with lockdown measures.  Currently, the SA strain isn't highly prevalent but that can change quickly if/when lockdown is released as the vaccine brings the other strain under control.  Growth of the new strain could be masked by falling cases - very much a repeat of how control was lost in and after the last lockdown.   

In reply to minimike:

> I’d be amazed if that time is still in the future tbh.

Yes; with some cases identified here already there are almost certainly more - but when it's still in small numbers, stopping more importation events makes a big difference.  Further, it's almost certain this isn't the most worrying variant that's going to emerge somewhere in the world in the next year.  Best keep them out till we figure all this out.

Edit: I was so slow replying that you answered Si dH's post rather more succinctly than me

>  That’s not a nice idea. Wish I hadn’t gone there now!

Indeed.  It does feel like the seriousness of this is getting through to cabinet more.  Richard J suggested on another thread today that the MP Neil O'Brien is taking up some of the advisory vacuum left by Cummings.  Perhaps these things are not unconnected.

Post edited at 20:04
 Misha 20 Jan 2021
In reply to wintertree:

>  We know this family of viruses has the ability to be at least 10x more lethal at all ages, and it looks like it can become more transmissible.  

That would keep the yoof at home...

 Misha 20 Jan 2021
In reply to minimike:

Article here https://www.biorxiv.org/content/10.1101/2021.01.18.427166v1

They only considered antibodies so perhaps T cells would still be effective but this is way beyond my knowledge of immunology...

OP wintertree 21 Jan 2021
In reply to BusyLizzie:

> Thank you again. You are a hero.

Sorry; I meant to reply last week.  Thanks are always welcome, but...

I'm really not a hero.

I'm doing this sat here on my sofa, living my relatively isolated life, where I'm risking nothing to do this, beyond the criticism of the endless sock puppet accounts of the misinformation poster.  There's nothing heroic about what I'm doing.

A lot of people who trained for -  and took on - tough but bounded jobs are now being turned in to heroes, and their workload isn't going to drop for a while yet.  They're facing awful working conditions and the personal responsibility and long term emotional consequences of making decisions far beyond their pay grade, and far beyond what they ever expected to be responsible for when signing on.   Some of them are dying because of it, and more will yet die.  They're heroes because they keep doing it, even as the mental and physical toll racks up.  I don't think I could do that.

The situation that is turning good, decent people in to unnecessary heroes didn't have to be.  I failed to make a difference to what has unfolded since Dec 14th; I'm closer to a failure than a hero on the spectrum.

 David Alcock 21 Jan 2021
In reply to wintertree:

> I'm really not a hero.

Chin up. You're doing a grand job doing what you're doing. 

In reply to wintertree:

> I'm not sure.  If we can't get this under control globally, then the variants are just going to keep coming; this is not like the flu in that respect.  We know this family of viruses has the ability to be at least 10x more lethal at all ages, and it looks like it can become more transmissible.  There's no wisdom in anthropomorphising a virus, but still it feels to me like this one is trying to be more like the original SARS. 

I'm less convinced that it will go that way. The strongest evolutionary pressure is towards transmissibility, which, like with many respiratory viruses, should hopefully push towards less severe or even absent symptoms.
I'd like to think it's trying to be more like the common cold. Maybe that's just because I recently re-read the theories about Russian Flu, but I'm optimistic that the endemic phase (i.e. what we call 'the future') will be tolerable.

Post edited at 07:27
 BusyLizzie 21 Jan 2021
In reply to wintertree:

> I'm really not a hero.

You are doing a lot of work in the pursuit of truth, and we are all grateful.

Whether any of us is a success or a failure is a different question, and the answer is rarely seen in the short term.

OP wintertree 21 Jan 2021
In reply to minimike:

> I’d be amazed if that time is still in the future tbh.

6 cases of the SA variant sequenced from inmates in New Zealand’s MIQ:

https://www.rnz.co.nz/news/national/434911/more-cases-of-covid-variants-con...

A lot more people travelling in to the UK than NZ and our quarantine isn’t managed.

 Offwidth 21 Jan 2021
In reply to wintertree:

Typical of the self-effacing modesty of heros (in great contrast to the all too common anti-scientific idiots self identifying as heros).

 mik82 21 Jan 2021
In reply to wintertree:

https://spiral.imperial.ac.uk/bitstream/10044/1/85583/2/REACT1_r8a_final.pd...

Possibly some evidence that infections stopped reducing last week (although confidence intervals quite wide). Will be interesting to see this week's ONS survey (for some reason last week's wasn't released) to see if there's correlation. I seem to remember the REACT1 study being at odds with ONS and observed cases once before.

OP wintertree 21 Jan 2021
In reply to mik82:

It’s hard to know as REACT’s down period coincided with the P1+P2 peak.  The data doesn’t look great for measuring a rate within their most recent month but it’s not plummeting like I’d hoped to see.  

OP wintertree 21 Jan 2021
In reply to Longsufferingropeholder:

You could be right with your vision of the future.  I’m also optimistic that various other research into how the SARS and MERS viruses mess with the immune system is going to improve the treatment a lot more than the already impressive work to date, eventually taking the sting out of the tail of any new, immunity evading variant.

But that future feels an awful long way away with a lot of difficulties between here and there.

In reply to wintertree:

I think I have to be, don't I? I mean.... if that's not what the future looks like then no one will be here to tell me I'm wrong!

OP wintertree 21 Jan 2021
In reply to thread:

Looking at the latencies in the data, thread #9 will happen on Saturday night instead of Friday night.  This well let me get the demographic and UTLA plots out at the same time, and will reduce the weekend biasing in the cases date I think.

So don't panic at the absence of a new thread tomorrow...

 Toerag 21 Jan 2021
In reply to elsewhere:

> On the programme I saw they reckoned wastage was normally assumed 5% but thought it was closer to 1%.

The Pfizer vaccine is supplied with an assumption of 20% waste. We had 0.7% here from our first batch of 975 doses

https://guernseypress.com/news/2020/12/22/second-vaccine-doses-arrive-as-fi...

OP wintertree 21 Jan 2021
In reply to Toerag:

Interesting times ahead for your bailiwick - pressure to open up as the vaccine rolls out effectively, vs the threat posed by the SA variant and likely others having lost key epitopes reducing cross-immunity.

Can your runway take anything that can make an air bridge to New Zealand?

 Tallie 21 Jan 2021
In reply to Toerag:

I do wonder whether we’ll have to adopt a Guernsey / NZ / Australia / China total suppression approach before this is over?

Although threads like this: https://www.mumsnet.com/Talk/coronavirus/4142493-how-long-will-people-agree...

(which I’m sure exist on most SM platforms) make me doubt whether it would ever be achievable in the UK ?

Lots of examples of people who haven’t understood the arguments for lockdown and are actively advocating a ‘let it rip’ approach.  I suspect their views would change long before we reach Spanish flu level of deaths but by then it may be too late?

Post edited at 16:01
 Toerag 21 Jan 2021
In reply to wintertree:

> Interesting times ahead for your bailiwick - pressure to open up as the vaccine rolls out effectively, vs the threat posed by the SA variant and likely others having lost key epitopes reducing cross-immunity.

Our equivalent of SAGE (the 'CCA') is quite risk averse, so as long as things still look bad elsewhere they'll keep things tight.  We can all see how shit things have been this winter for Jersey and the UK so the pressure to open up isn't kickign in yet.  Once the 'at risk' people have been vaccinated things will undoubtedly change.  It seems that finally other jurisdictions are realising that they cannot risk letting the virus trundle on in the community due to its capacity to become more infective and/or more deadly, so hopefully travel restrictions will ease as they get things under control.

> Can your runway take anything that can make an air bridge to New Zealand?

Not in a million years. It'll manage an A320 to Gatwick as long as it's not completely full up and that's it. Barcelona is in reach of our Embraer 190.

 Toerag 21 Jan 2021
In reply to Tallie:

> I do wonder whether we’ll have to adopt a Guernsey / NZ / Australia / China total suppression approach before this is over?

> (which I’m sure exist on most SM platforms) make me doubt whether it would ever be achievable in the UK ?

> Lots of examples of people who haven’t understood the arguments for lockdown and are actively advocating a ‘let it rip’ approach.  I suspect their views would change long before we reach Spanish flu level of deaths but by then it may be too late?


I think the recent 'bad' variants and lockdowns 2/3/4* will focus governments minds on that strategy.  As I've been saying for months, 'living with the virus' is bloney difficult - even before the new variants kicked in it was obvious that restrictions equivalent to tier 3 were required to keep things under control (live case numbers static). Tier 3 for the sustained periods of time required simply isn't palatable for the economy and society, and as the new variants have shown, the risks of leaving the virus loose in the community are massive. If the SA variant reduces vaccine / antibody efficacy to 1/8th of it's normal value that is bad news. It there's another deletion mutation rendering PCR tests ineffective that's bad news, if a more deadly variant emerges that's bad news.  No sane government would risk further lockdowns and worse variants.

 Toerag 21 Jan 2021
In reply to mik82:

>  Possibly some evidence that infections stopped reducing last week (although confidence intervals quite wide).

Quite a bit in the news about cases may have stopped declining today, along with calls to not relax behaviours from politicians.  The government seems to be worried.

 mik82 21 Jan 2021
In reply to Tallie:

Well, looks like mumsnet are planning on testing the effectiveness of the vaccination programme at Easter then! Hopefully the South African variant isn't widespread by then...

I think someone linked a thread from there way back in the first wave, where people were justifying going into work with covid symptoms, saying they didn't care if someone random died as they put their family first. Looks like nothing's changed, it's all "me me me".

 Toerag 21 Jan 2021
In reply to tom_in_edinburgh:

> I agree, it's a brand new product and a factory trying to scale up and there could easily be delays to the schedule to deal with. 

All hands to the pumps to protect a vaccine factory in Wrexham overnight.

https://www.lbc.co.uk/news/emergency-services-scramble-to-protect-astrazene...

 Toerag 21 Jan 2021
In reply to Longsufferingropeholder:

> That is still only ever a delaying tactic. The answer to get on with that tweaking of the vaccine that's supposed to be not too hard......

It's still a wise tactic from a risk perspective. Why risk importing a bad variant into your community whose healthcare system is on the brink when you don't have to?

OP wintertree 21 Jan 2021
In reply to Tallie:

My peril sensitive sunglasses won't let me view Mumsnet Covid threads.

> I do wonder whether we’ll have to adopt a Guernsey / NZ / Australia / China total suppression approach before this is over?

To my mind, something close to elimination is the only approach that's not laden heavy with risks known and unknown.  The case level needs to be low enough that we can jump on outbreaks successfully and not have them snowball in to the next wave.

To go for an elimination strategy now, given the near endless disruption, deaths and the current healthcare situation, that would be quite the back-pedal and would make it clear that we didn't have to have all that disruption, death and overload.  

The population finds itself between an immovable force and irresistible object.  

Post edited at 17:04
OP wintertree 21 Jan 2021
In reply to Si dH:

> Do we know how long it takes from a positive test being recorded in a lab and the individual being notified, to that case being fully sequenced such that the specific variant could be identified and added in to a database of some kind for a given area or region?

I think the notification is prompt (next day) - excepting weekend postal issues, but entry to test and trace corresponds to appearance on the government dashboard so has quite a lag.  I don't know this - it's based on inference from some of the reporting over the ".xls failure mode" a few months ago.  Oxford Nanopore claim their reader can turn around a sample in to a genome in 7 hours [1], including the reverse transcriptase and PCR steps.  I don't know how much expert human effort is required to reassemble and validate the genome from the DNA fragments.  The forensic lab in NZ doing their sequencing processes up to 100 genomes / week with a 24 hour turnaround on priority samples [2].  I'm not reading much in to the low throughput of the Kiwi lab other than that they're blessed not to have needed more...

Given that expert human input is going to be needed to reach inferences in the contact tracing process, influenced by sequencing, I struggle to see how this work for more than a few hundred cases/day which would need significantly lower prevalence than last summer - entirely possible if we don't allow immune evading variants to spread unchecked until then - the catch 22 being we can't do the kind of sequencing and intervention needed to prevent that whilst cases are high.  

> I don't know whether the system is currently fast enough to do what you suggest. I agree it sounds sensible. Would also like to see it focused on areas that have seen exceptionally high spikes in recent weeks or that have been showing high overall growth rates despite low SGTF rates.

Yes; there're several things that suggest themselves from the data on priorities for this.  I think port areas should probably be high priority as well.   

[1] https://nanoporetech.com/covid-19/overview

[2] https://www.esr.cri.nz/our-expertise/covid-19-response/new-news-page/

Post edited at 17:02
 Toerag 21 Jan 2021
In reply to Misha:

> They only considered antibodies so perhaps T cells would still be effective but this is way beyond my knowledge of immunology...


Apparently most infections result in antibodies:-

"The possibility that some patients develop T cells but not antibodies does complicate the calculation of how many people have actually been infected by Covid-19 and are immune. But it is, nevertheless, a leap to claim one-third of the United Kingdom has been exposed and is immune. This is because most people who get infected do develop antibodies: between 91.1% and 100% of cases who tested positive by PCR then developed antibodies; studies show that antibodies last for several months at least. Thus, only a small proportion of people would be missed by serological surveys, by only having a T-cell response to the virus."

https://unherd.com/2020/11/the-trouble-with-covid-denialism/

So the whole T-cell thing hinges upon the effect T-cells have - do they reduce the time period it takes for the body to wind up its defences and thus result in less serious infections? Do they kill the virus off themselves?

mick taylor 21 Jan 2021
In reply to Toerag:

> >  Possibly some evidence that infections stopped reducing last week (although confidence intervals quite wide).

> Quite a bit in the news about cases may have stopped declining today, along with calls to not relax behaviours from politicians.  The government seems to be worried.

I believe it was the React study, swabs taken 6th to 15th Jan. 

mick taylor 21 Jan 2021
In reply to mick taylor:

And 363,500 got first vaccine dose, which is bloody good news!

OP wintertree 21 Jan 2021
In reply to mick taylor:

Two record days in a row for vaccination.  Seems like the logistics machine hasn’t finished getting up to speed yet as well.

 minimike 21 Jan 2021
In reply to wintertree:

Fascinating discussion at work over (zoom) coffee. We’ve started getting vaccinations as frontline staff and it seems there’s a strong correlation between having tested positive last year (which I did) and getting some noticeable side effects (viral symptoms, chills lasting 24h, nothing serious). Those who aren’t aware of having had covid don’t seem to report this. Applies to pfizer and AZ. No actual science of course but it would make sense I think k that if you have some natural T cell memory from infection your response to the vaccine would be rapid and strong. Vaccine effectively acting as a booster? Immunologist needed...

edit: having mine on Tuesday so I’ll report then whether this holds for me.

Post edited at 18:47
 mik82 21 Jan 2021
In reply to minimike:

I had a bit of a sore arm for a day (Pfizer), having had confirmed Covid in March. Then again, my covid was a bit of a non-event with a runny nose, mild dry cough and about 30 minutes of sense of smell disturbance.

 joem 21 Jan 2021
In reply to mik82:

From what I’ve heard it’s the second jab that’s the one you notice.

 minimike 21 Jan 2021
In reply to joem:

Maybe that’s why they’re so keen to prevent us getting it in 3 weeks.. bad publicity?!

 RobAJones 21 Jan 2021
In reply to minimike:

> Fascinating discussion at work over (zoom) coffee. We’ve started getting vaccinations as frontline staff and it seems there’s a strong correlation between having tested positive last year

Interesting. I know 30+ people who have had the jab, mainly elderly. Only two have had any sort of reaction and both of them had tested positive last year. Mrs J's mum was one, symptoms were exactly as you described, I had put it down to the fact that we had had to wait outside for 45 minutes in the freezing cold. I has been surprised on the other thread by the number of people reporting reactions to the vaccine, but they were often health care workers, so possibly more like to have ben exposed.   

OP wintertree 21 Jan 2021
In reply to thread:

With today's data release it looks like deaths in England have probably turned the corner in to decay.  The polynomial filter to the data (Plot 8e - Update 1) is trending down, and the exponential rate constant measured from this has crossed the axis in to decay (Plot 9e - Update 1). 

The noddy plot I made where all 3 measures are normalised to the size of their November peaks (Rescaled Plot - Update 1; explained in the post this one replies to) has the turning points all aligned to the same level.  The size of the December/January peaks has changed a bit as the last week's data affect the filtering behind the polynomial filtered trendlines (which are what's shown).  This set of plots gives me good confidence that hospitalisation will continue to fall for at least the next couple of weeks as this is "locked in" from past infections already detected as cases.  Deaths will hopefully fall with the relevant lag, but hospital occupancy is going to keep increasing for some time yet and that brings increasing pressures with it.

Edit: The last week or so of the blue "Cases" curve on Plot 9e is not trustworthy - updating cases data on a Thursday brings significant bias from the weekend effects in data sampling.  The next plotting thread will have an update of this from what I consider the least biassed point in the data.

Post edited at 23:17

 Misha 21 Jan 2021
In reply to wintertree:

Regarding zero Covid - in retrospect, that would have been the best strategy. We weren't that far just as they started to progressively lift the restrictions. A bit more effort and it could have been reduced to a sufficiently low level for (local) test and trace to keep on top of it. However hindsight is a wonderful thing...

With the vaccines on the way and way more testing and tracing capacity (the tracing is a bit iffy but at least there is capacity), it is perhaps possible to go for zero Covid now without a long, hard lockdown - just keep things as they are till about Easter, then open up very slowly until the level of vaccination is high enough to be able to open up more without cases getting out of hand again. Would need to communicate that this is the strategy, pretty much close the borders for the foreseeable, pay people properly to self-isolate if they can't WFH and test cases aggressively. A tall order and it's not going to happen...

Of course another school of thought is that it's not really necessary to go for elimination (which might not be feasible anyway) once most people have been vaccinated. Depends on your view on the risk of vaccine resistant strains and whether the vaccines could be tweaked in time. Trouble is, we just don't know.

 Misha 21 Jan 2021
In reply to Toerag:

I may be wrong but my broad understanding is T cells attack and destroy cells which have been infected by the virus. The immune system is vastly complex and every time I try to read up about it I give up pretty quickly...

 Si dH 22 Jan 2021
In reply to Misha:

I'm not convinced we could ever have achieved zero covid last year (once the first peak was underway) without a harsher set of restrictions. I don't think it was a case of just holding on to what we had for longer as many areas flat lined in cases at levels that were relatively low but still nowhere near zero. Of course it's difficult to know for sure due to the lack of community testing until the peak had already subsided and many restrictions relaxed.

Other than stopping international travel (a no-brainer in hindsight), personally I am very conflicted about whether additional restrictions over/above what we had last spring would ever be worthwhile or justifiable (eg, the sort of thing they did in Spain where kids weren't allowed out for 6-8 weeks - I don't think that is justified.) So I'm not sure I would ever have wanted to support a totally zero covid strategy in that scenario where the infection is new and there is no pre existing immunity. However, if things go well with the vaccine for the next 3 months then I agree with you it may well be the best strategy to adopt this summer. 

Post edited at 06:51
 Si dH 22 Jan 2021
In reply to Misha:

They do a few different things:

https://www.britannica.com/science/T-cell

 AJM 22 Jan 2021
In reply to wintertree:

> The last week or so of the blue "Cases" curve on Plot 9e is not trustworthy - updating cases data on a Thursday brings significant bias from the weekend effects in data sampling. The next plotting thread will have an update of this from what I consider the least biassed point in the data.

I was just looking with some dismay at that >30 day halving time and then read this! Phew.....

 neilh 22 Jan 2021
In reply to Si dH:

If you have a look at the list of exempt occupations on travel restrictions it illustrates the amount of movement that is required to keep the U.K. ticking over. Apart from professional sports or elite  athletes  the list is pretty reasonable. 
 

After that what you need to do is strictly enforce the 14 day quarantine period for the rest us. That means hotels paid for by travellers as in other countries and follow the Australian example. 
 

we are not that far away from it  , the gov just needs to get the infrastructure there and face the political backlash .

 SouthernSteve 22 Jan 2021
In reply to Toerag:

Some background:

T-cells and antibodies do not operate independently most of the time. For antibodies to be made there is a cytokine (chemical) environment from T-cells which promotes antibody production and often directs the type of antibody produced (e.g. IgG in infections and IgE in allergy). Coronavirus (in the past) has been shown to reduce antibody production and antibody levels may then also fall, but memory T-cells should remain. Whether the immune response can, in this semi-prepared state, ramp up when new infection is encountered in the absence of good antibody levels might make the difference between, no disease, subclinical disease or repeat obvious clinical infection.

So T-cells provide this 'helper' function importantly, but also confusingly may kill infected cells directly when foreign antigens (proteins usually) are expressed on the cell surface (cytotoxic function).

From the common GCSE and A-level understanding of immunology using the Gel and Coombes model of hypersensitivity (a really good way of getting into the subject) it is really easy to consider cells and antibodies as different effectors, rather than parts of a complicated network of interactions.

 mik82 22 Jan 2021
In reply to wintertree:

Just noticed that there is some new additional NHS England data available on admissions by age, slightly more stratified than the data off the dashboard, as it gives 10year age bands down to age 55

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

https://www.england.nhs.uk/statistics/statistical-work-areas/covid-19-hospi...

Remains to be seen how effective vaccination will be at preventing admissions. No-one got "severe" covid in the vaccinated groups, so you could take that as not needing admission (Highly likely to be incorrect in the elderly as any kind of mild infection can be debilitating to the point of them being unsafe remaining at home.  There's also the delayed second dose effect...).

Using an ideal scenario of 80% take-up rate and full protection from admission if vaccinated we get to about 50% admission reduction once over 65s done and about 75% after the over 55s

OP wintertree 22 Jan 2021
In reply to Si dH and Misha:

Re: Hindsight - back in March before the first lockdown, some posters were suggesting we could run cases hot in a "balance" against the economy.  One of the first points I raised there was that more cases ⇒ more probability of harmful mutations emerging.  I know that's not quite the context of your post.  But for any of this to be hindsight, leadership needs to look at what has happened and recognise that other approaches could have worked better, and to consider that going forwards...

I think by the March lockdown last year, it was too late to achieve zero covid with measures that were conceivable in the UK.  Perhaps at the tail end of summer 2020 there was a way, but I'm not sure and the past is in the past now.

> personally I am very conflicted about whether additional restrictions over/above what we had last spring would ever be worthwhile or justifiable (eg, the sort of thing they did in Spain where kids weren't allowed out for 6-8 weeks - I don't think that is justified.)

Other than hard inbound border closures with only MIQ as a route in, I'm not convinced that we needed more restrictions, so much as more public buy in to, and more enforcement of existing restrictions, combined with boots-on-the-ground teams and "flash" (hard, brief)  local restriction to allow the super-spreader events that lost control to be contained.

I don't think this would have got us to zero covid, but it might have held cases low enough that the new variant didn't emerge (if it emerged here) or didn't spread unchecked (if it was an importation).

> So I'm not sure I would ever have wanted to support a totally zero covid strategy in that scenario where the infection is new and there is no pre existing immunity. However, if things go well with the vaccine for the next 3 months then I agree with you it may well be the best strategy to adopt this summer.  

The vaccine significantly changes the equation on elimination.  If we get to the summer without an immunity evading variant having risen to prominence, I hope to see a more concerted push than last summer.

Post edited at 11:27
OP wintertree 22 Jan 2021
In reply to AJM:

> I was just looking with some dismay at that >30 day halving time and then read this! Phew.....

Yes, I should really have deleted that part of the blue lines.   We'll see what the next round of plots shows - I'm delaying them until tomorrow to get a more complete data set.

It looks a bit like hospital occupancy may also have maxed out although it probably needs a few more days to be really sure.  It can't fall fast enough given what the working conditions must be like for the staff - physically and mentally exhausting. 


OP wintertree 22 Jan 2021
In reply to mick taylor:

409,855 people vaccinated yesterday.

This is good.  Very good.

Post edited at 16:38
 Si dH 22 Jan 2021
In reply to wintertree:

> Yes, I should really have deleted that part of the blue lines.   We'll see what the next round of plots shows - I'm delaying them until tomorrow to get a more complete data set.

There seems to be quite a bit of variation in rate of fall. Midlands seem to be falling slower than most. The fall in the NE seems to be slowing too. Other regions generally falling quicker. I'll be interested to see if your graph tomorrow shows the same.

> It looks a bit like hospital occupancy may also have maxed out although it probably needs a few more days to be really sure.  It can't fall fast enough given what the working conditions must be like for the staff - physically and mentally exhausting. 

Would it be an easy job to produce these graphs on an NHS regional basis (or for a couple of select regions)? This would give a better idea of when pressure might be starting to relent in each area.

Post edited at 17:06
 Offwidth 22 Jan 2021
In reply to wintertree:

I agree its good news and the local centres have been bloody fabulous, but from the Twitter post on the other channel 400,000 seems to have been the average daily target all this last week. They are now saying slow down to some areas yet won't admit there is any supply issue beyond lumpiness (Hancock) and the minister in charge  says they haven't taken vaccine from Yorkshire (when they clearly seem to be). I'm crossing everything hoping this government doesn't miss an open goal.

1
 mik82 22 Jan 2021
In reply to wintertree:

Do I remember you doing an IFR plot before?

Any evidence of increasing IFR as per the 30% increase for the new variant suggested today?

OP wintertree 22 Jan 2021
In reply to mik82:

I used to, using the ONS estimates of daily infection.  They’ve been absent since mid December.  It takes a couple of weeks of data to be able to run that.

I have a CFR analysis I can run but over longer periods of time, comparison is fraught without a baseline for testing efficiency.  I’ll run it this week for tomorrow’s update but I doubt it’ll be very useful.

The longitudinal data for demographic IFR vs sliding time window analysis must exist within the NHS but I wouldn’t know where to ask about why outputs from there apparently aren’t published.  This is an absolutely key output that could be put together from the NHS data corpus.

If the variant is spreading more in the young - mixed messages on that from the pre prints - that would go some way to explaining it.

In reply to Si dH:

> Would it be an easy job to produce these graphs on an NHS regional basis (or for a couple of select regions)? This would give a better idea of when pressure might be starting to relent in each area.

Two minute job when sat at the laptop. I’ve got a lot of other 2-minute jobs queued up by watch this space...

Post edited at 18:40
In reply to Si dH:

Rate constant by age group will be a gripper this week too. a) Can't wait b) anything can do to help?

 minimike 22 Jan 2021
In reply to Longsufferingropeholder:

Busy day at work so I haven’t been keeping up, but did spot this:

https://www.theguardian.com/world/2021/jan/22/mhra-vaccine-makers-covid-var...

Critical point in the final paragraph that MHRA anticipates some testing required for modified vaccines but NOT full reappraisal. Flu vaccines usually require 2-3 month testing and approval cycle each year so if imaging similar for covid updates.. not great, but not terrible either. I’d feared worse..

In reply to minimike:

Slight issue in that the modified adenovirus vectors might be a one-shot deal; it's possible we'll all end up immune to the vector. So less hopeful the cheap and easy ones (AZ, sputnik V) can be modified indefinitely. So all eggs might be in the mRNA basket. Whether that means anything different for reapproval remains to be seen.

Post edited at 20:46
 Misha 22 Jan 2021
In reply to neilh:

Plenty of hotels going empty. Just need some security guards... I know it’s more than that but it’s totally doable. Especially as hardly anyone would go on holiday abroad if that’s what you’d need to put up with.

Some noise in the papers about it. Usual ministerial testing of the water. No smoke without fire? BoJo said today we might need to do more on the face of the new strains.

It would ruin the aviation sector for a while but that’s better than ruining the whole economy any longer than necessary.

OP wintertree 22 Jan 2021
In reply to Si dH:

> Would it be an easy job to produce these graphs on an NHS regional basis (or for a couple of select regions)? This would give a better idea of when pressure might be starting to relent in each area.

Okay; it was about a 5 minute job to hack the code about to a different layout.  It's the same data as plot 18 but laid out differently.  I think both have pros for assimilating it all.

I'd consider the last week of the blue "cases" curves to be less-than-perfect - they should improve as last Tuesday's data firms up in tomorrow's release; the weekend effect seems very big at the moment.


 minimike 22 Jan 2021
In reply to Longsufferingropeholder:

Oh.. I didn’t know that. Do you have a reference, or is it common knowledge (among experts!)?

In reply to minimike:

Can't really explain it better than xkcd does:

https://m.xkcd.com/2406/

 Si dH 22 Jan 2021
In reply to wintertree:

Thanks - it looks like all the regions are just tipping in to decay for hospitalisations, which is great news.

Although, I actually meant a different graph I had meant your curve of hospital occupancy Vs patients on mechanical ventilation. I think that's quite powerful. Only do it if you think it would be useful.

Post edited at 21:13
OP wintertree 22 Jan 2021
In reply to Si dH:

Gotcha.  Plot below.  

This is for data from 2020-09-01 onwards.  It's not the best plot and needs some work to make it fully self-descriptive.  

Good call on a nice regional plot - you can see that the East of England has turned the corner, the other "red" regions are turning it and the rest are just about it.

Edit:  I've added a second version.  In the first, the two axes auto-range to make the best use of space.  In the new one, I fix the aspect ratio to be 10 (general occupancy) to 1 (ITU occupancy).  So, if the ratio between them was the same for all regions, all lines would have the same sort of gradient on this fixed aspect set of plots.  (This would happen automatically if they were on a shared axis, but when I did that it just looked like a bunch of lines scrawled all over each other).

London has way more admissions to ITU, proportionally speaking.  There's something interesting to dive in to and try and understand.

Post edited at 21:36

OP wintertree 22 Jan 2021
In reply to Longsufferingropeholder:

As I understand it, the adenovirus that's used is very stripped back and is expressing almost no transmembrane proteins other than the spike protein, so whilst I understand the basis for your concern I don't know if it has other epitopes to allow it to be rejected.

Any of the immunologists reading this far down the thread?

 SouthernSteve 22 Jan 2021
In reply to wintertree:

> As I understand it, the adenovirus that's used is very stripped back and is expressing almost no transmembrane proteins other than the spike protein.

There is a response to the adenovirus backbone. In previous circumstances, this system was used for primary vaccination and followed by an alternative viral vector (e.g. modified vaccinia). 

 Misha 22 Jan 2021
In reply to wintertree:

BoJo today: “Currently, the rate of infection is forbiddingly high, and I think we have to be realistic about that. I think we will have to live with coronavirus in one way or another for a long while to come. I think it is an open question as to when and in what way we can start to relax any of the measures. Obviously, we want to do everything we can to open up but only safely, only cautiously.” Better late than never, I suppose. 

OP wintertree 22 Jan 2021
In reply to SouthernSteve:

Thanks.  

Begrudging thanks as I’m less happy than I was an hour ago, but thanks.  

Edit: I asked a question on the virus but it was a stupid question.  I have a lot of reading to do about nonenveloped viruses.

Post edited at 22:21
OP wintertree 22 Jan 2021
In reply to Misha:

There does seem to have been a big shift in the government’s messaging and what it suggests about how they view the situation. Reassuring but the CRG continues to concern me.

 SouthernSteve 22 Jan 2021
In reply to wintertree:

DOI: 10.1093/infdis/jiw244.  This explains the strategy. MVA gave poor initial immunogenicity. Using adenovirus first avoids this. I fancy having Oxford first followed by Pfizer, but this is not really my field!

OP wintertree 22 Jan 2021
In reply to SouthernSteve:

Thanks.  I’ll read that when the sun is up.  I keep finding myself drawn to immunology.  I’ll hopefully be due a career change in about 6-8 years time...  

From what you’ve said I can see why the AZ as a first jab and Pfizer as a second makes sense.

In reply to wintertree:

It's also the basis for the ongoing study into giving one dose of AZ followed by sputnik, or vice versa, to leverage the different vectors. And also as I understand it the rationale for choosing a vital vector that few people are likely to have been exposed to.

Sleepy time now so cba to find a link but I definitely read something about it, honest.

Post edited at 23:07
 Offwidth 23 Jan 2021
In reply to Longsufferingropeholder:

Thought I'd post this here as a revent discussion on age demographic was on a pub thread so doomed to die

This week's Indie SAGE presentation had.a good bit starting at 10.30 that illustrates various age demographic factors...why vaccination won't slow cases for quite a while...it wont cut pressures on ITU for quite a while,  it should reduce deaths quickly and hospitalisations  fairly quickly.

https://www.youtube.com/watch?v=gKTHqyFfzFs&feature=youtu.be

I'd recommend the first 15 minutes on the data (interesting stuff on regions and on school & workplace outbreaks).

 Si dH 23 Jan 2021
In reply to Offwidth:

Thanks Offwidth, I learnt a few things there. Particularly how much of ICU capacity is taken up by under 75s - I knew that the oldest generally didn't get put on mechanical ventilation but hadn't mentally translated that to the result that ICU occupancy will be brought down much more slowly than overall hospital occupancy (or deaths) by the vaccination strategy. That might be a piece of knowledge we could use with Wintertree's graphs of ICU vs hospital occupancy to see when vaccination is starting to have an effect that is distinguishable from the effect of restrictions (in lieu of, or additionally to, well broken down demographic hospitalisations.)

The link you posted was to the presentation from 15/01, the data slides in the one from 22/01 are also worth a look. Unfortunately indie SAGE do seem to have a bit of an agenda and in particular I thought for the 22/01 presentation it was best just to look at all the data presented and ignore what the guy was saying about it, there were several dodgy interpretations. Nevertheless, some interesting data I hadn't seen before.

Post edited at 07:48
 Paul Evans 23 Jan 2021
In reply to Offwidth:

Thanks for that. Really informative, Subscribed.

Paul

 Dr.S at work 23 Jan 2021
In reply to Misha:

I think since Christmas BJ has been pretty much on it communication wise. 
 

it’s a shame it’s taken him so long to get consistent - a lot of his messaging has been good, but he lapses into bombast so easily...

 mik82 23 Jan 2021
In reply to wintertree:

https://assets.publishing.service.gov.uk/government/uploads/system/uploads/...

Here's the NERVTAG paper on the possible increased fatality rate of the new variant

 lithos 23 Jan 2021
In reply to Offwidth:

thanks for that.

the age bit (10:15 .. 13:30) is so clearly explained.

ps  thats some microphone Alice Roberts has - guess you need it if you are omnipresent

OP wintertree 23 Jan 2021
In reply to lithos:

> the age bit (10:15 .. 13:30) is so clearly explained.

Indeed.  I really like their visualisation.  

There's a lot of confusion caused by hospitalisation rates commonly being reported as per 100,000 people and people taking the numbers as absolute - which pretty much flips the picture compared to actuals.   


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