UKC

A lot of the discussion. on the previous thread has been about specific areas with slow decay or seeing growth.

The growth and decay of the pandemic is very fragmented, with different behaviours in different regions, age groups, employment types and many other differentiators.  When we start from a very high number of daily cases and the introduction of broad control measures, we see rapid fall across the board; yet if somewhere within those fractured sub-populations, one or two have growth, as small in number as it may be, that growth will inevitably come to dominate the daily figures as the other areas all decay.

I think perhaps today's update is on the cusp of where the no-decay and growth areas start to take over driving top level cases. The last couple of days of data for England look like growth, as does Scotland.  With Scotland we've seen this before, and then it turns to decay again - a lot of short term variation against a background of little decay.  With the exponential rate constants in my plot being measured over longer time scales, they don't yet reflect this possibility of a return to growth.  Really, it's too early to say for sure and we need to wait another week, but I'll try and dig in to the demographic and regional data a bit to see what insight there might be.

It's an uncomfortable time to be looking at growth with the rising concern over one of the Indian variants, and with the age range 40-44 not yet all at least 3-weeks post their first vaccination.  On that note I had my first jab yesterday and have been sleeping it off today, so I may well be less coherent than usual.

Wales is the only nation to continue to hold clear, unambiguous decay over the last week.

Link to previous thread: https://www.ukclimbing.com/forums/off_belay/friday_night_covid_plotting_24-...

In reply to wintertree:

As with last week, I'm putting the England plots on a log-y axis as we have such a range in values.  The up-tick of the filtered trendline on cases for England wouldn't look so dramatic without the lowest data point, so it could be that noise is having quite an effect here.  This sort of exceptional low/high can happen around a bank holiday due to changes in testing, so an extra dollop of caution goes on the interpretation until we have another week of data through.

Regardless, cases are not falling at a decent exponential rate any longer, and I think that's likely related to growth in some areas reaching a tipping point against decay in others - and from that point on, growth can add more more than decay can take away.  

In reply to wintertree:

There's no "smoking gun" suggesting growth in any particular area in the demographic case rates in D1.c.  Notably, the exponential growth in cases associated with the return to schools is much smaller after easter - the second island of orange in the bottom right is fainter than the one to its left.  As this is in the rate constants not the absolute numbers, this implies that the return after easter went better, and cases multiplied less during the second return to schools.  Perhaps there's a bit of growth in cases in the oldest ages in the top right, but the absolute numbers are so low that the rate constant measurements are very noisy there.

I haven't show Plot D1.x for a while as not much is changing.   This does a lot of filtering compared to the 1D plots of individual rate constants, as otherwise it looks like a seismograph on a bad day.  Earlier on this had been showing signs of going to much more negative rate constants (faster exponential decay) but that didn't hold; perhaps this suggests the vaccination program is holding the line against the progressive relaxation of control measures.

The demographic daily deaths data in D5.lin is doing a convincing impression of heading towards 0/day.   Hopefully in another week or two it'll be lower than at any point since the start of the pandemic.  

In reply to wintertree:

Plot 16 has been on a break whilst the data jumped up and down with LFDs coming on and off as it really broke the concept of this plot.  It's back this week by request and seems quite informative.

This is the English UTLAs, ranked in x-axis position by their cases/100,000 people/day on 2021-04-11, with data markers showing their rates 3 weeks later.

There's a copy of the plot where UTLAs that have risen by more than 20% of their rate on the ranking date are annotated.  It's worth looking at the raw data on the UK government dashboard to help interpret these - this plot is really meant to guide attention and not to proclaim the existence of a problem in and off itself.  For example, Plymouth and Bedford have some pretty significant growth with more coming in the provisional window (not used in my analysis), where-as Bromley looks to be in the noise.

https://coronavirus.data.gov.uk/details/cases?areaType=ltla&areaName=Be...
https://coronavirus.data.gov.uk/details/cases?areaType=ltla&areaName=Pl...
https://coronavirus.data.gov.uk/details/cases?areaType=ltla&areaName=Br...

I've then done a version of D1.c - the absolute case numbers and their exponential rate constants broken down by age, for the "red" annotated UTLAs.  I've also put in D1.c so you can flip-book between them for context (this works much better on a computer than a mobile device, using arrow keys to swap images).

• Compared to D1.c this measures rate constants over ±14 days instead of ±7 to deal with the greater noise arising from lower numbers; this "blurs out" features in the plot compared to D1.c.
• I've put a ±14 day version of D1.c in for a fairer comparison, 
  Even with a wider measurement window, the upper right part of the rate constant plot, and the very bottom row on the right, have a lot of noise due to the very low case numbers when broken down by age and a small subset of UTLAs.  

What stands out to me on the demographic data for growth areas:

• In the absolute numbers (left plot), the decay in ages 45-60 from red on the colormap to black just hasn't happened in these regions.  It would be interesting to see demographic regional data on the vaccine uptake.
• Growth in secondary school ages is significantly more than nationally
• Growth is across the board demographically speaking.

I'm a bit uncomfortable at this plot - if we had genuine decay in all the UTLAs masked by random noise turning a few to apparent growth, and then I took the ones appearing to grow and added them together, all I'd really see is the largest of the random numbers embodied in the noise.  There are many scientific methods to test the possibility that a result is due to random noise rather than an actual underlying trend, but to use these methods fairly/appropriately/correctly, a decent understanding of the noise statistics is required and I've not achieved that for this data.   Still, my qualitative judgement is that this is very real - if it was the effects of randomness, it wouldn't be so clearly correlated over the different age bins.  

Another possible interpretation is that we're looking at the sum of all areas currently affected by what at the UTLA scale are sporadic outbreaks.  Only time will tell.

In reply to wintertree:

A grab bag of other plots.

Week on week method PCR only English rate constants - we have our first data point of growth (+ve rate constant) that isn't caused directly by a bank holiday.  It may be related to cases displaced from the bank holiday however.

Four Nations Absolute Numbers - Wales is standing apart from the other nations in terms of maintaining definite decay.

Vaccine plot - the most recent weekend low is deeper than usual, and not just down to the bank holiday Monday.  A small difference however, and progress continues at pace.

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

• As ever, lots of qualifiers that we don't know a lot of things about this data - what fraction are from travel vs community transmission etc.   Si dH linked to some SAGE papers with more detail up to a couple of weeks ago on the last thread [1].
• This all got a bit political last week with reports in the media that PHE had delayed release of the data until after the local elections.  If true, I'm very disappointed; one thing experts have been very clear on from the start is the importance of openness and transparency.
• The UK variant that had been seeing large growth, VUI-21FEB03, is levelling right off.  Perhaps the apparent growth was caused by enhanced contact tracing chasing down existing live infections (a bigger pool than daily "new" infections found by symptomatic testing) generating apparent growth, and now it's caught up to the new infections that growth is gone.  Lots of other interpretations are also possible.
• The SA variant is dropping below a fixed rate exponential as well, again perhaps this is a sign that enhanced contact tracing has caught up to the leading edge of infections?
• The red curve is the India variant B.1.617.2.  Although I've measured a doubling time, there are only two data points and so basically any form of curve can be fit to them, so the measurement in no way implies that the data is rising exponentially, it's just asking the question "if the rise is exponential, what is the doubling time?".  Some suggestion from PHE via the BBC that much of this rise was down to people rushing to fly in to the UK after India was announced as being red listed, but before the date on which it came in to force.  Hopefully this variant isn't particularly bad news, but if this rise is due to a madcap travel rush it perfectly illustrates the pitfalls of our approach to red listing - if a genuinely bad news variant arrises, encouraging people to rush in with it doesn't seem smart.

[1] https://www.ukclimbing.com/forums/off_belay/friday_night_covid_plotting_24-...

In reply to wintertree:

Its been mighty cold of late with less sunshine - cases data maybe fits this with your weather based hypothesis?

looking at the demographic rate constants plot the data is really 'stuttery' - just a mark of the small numbers in play now?

In reply to Dr.S at work:

Indeed.  It's bonkers.  We woke up a couple of days ago to lying snow, and I nearly got brained by giant hail getting out of the car to do the gates on the Oxnop Scar road coming back from my jab.  I heard on the radio that it was the coldest May bank holiday on record, and there's a lot of rain about as well.  Not what we need right now, I'd been hoping for a repeat of last year's glorious spring to help drive rates down before May 17th.

Yes, it gets to the point there are so few cases that the measurements start to reflect the noise more than the data in the upper right of D1.c; it's a lot better in the 14-day version in a later post.

In reply to wintertree:

This has worked really well at picking out ongoing outbreaks, thanks. It's interesting to delve into demographic data for each and get a idea of what is causing it. The main headline though seems to be that it's not consistent. Eg, in Derbyshire (which is mostly due to a big outbreak in Long Eaton) it all seems to be school age (10-14).  In Sefton it's all around Formby and seems to be focused on young adults (local news was reporting a gym/pool had been told to close after cases were found there a couple of days ago, so presumably that's related). The rise in Bedford all seems to be in the 15-19 age group. Some are more demographically spread out.

It's really striking how the frequency of these localised occurrences has increased a lot in the last 1-2 weeks.

In reply to wintertree:

Thanks as ever. At a high level, it makes sense that what we’re seeing is the net effect of decay in some areas and growth in others. This is illustrated by changes in the dashboard map. Cue the discussion on the previous thread. Your graphs have highlighted additional areas to watch - although absolute case levels are still relatively low there, making the data more susceptible to noise (eg isolated outbreaks which might be contained but take a while to drop out of the 7 day averages). As you say, it’s a waiting game.

In reply to wintertree:

The PCR graph outlier is BH testing displacement I think. 

In reply to Misha:

For us, it's a waiting game. For people in office, I hope it's a time in which they are putting a lot of thought and sufficient resources into suppressing these outbreaks rather than letting them spread. The better we can do that the more people can be fully vaccinated before we relax further and the fewer get ill. It's a time when test and trace should be showing it's worth.  However, it's difficult to know whether this is happening because, apart from where variants have been found and surge testing is implemented across whole areas, none of the local response is being reported. I'm hopeful that active responses are taking place - we know they are in some areas because we know the local council/school was able to give PCR tests to whole year groups in Accrington, and we know the government have now stepped in in Bolton. But I don't know if anything much in happening in the rest. Fingers crossed!

In reply to Si dH:

Thought this report is worth linking for posterity within these threads.

https://www.theguardian.com/world/2021/may/08/uk-government-failed-to-consi...

In reply to wintertree:

Surely everyone's had enough of these fear mongering shitehawk made up stats !

In reply to Monkeydoo:

Yeah...made up from....erm...the DATA!  Clearly not everyone has had enough as week in week out there is considered and thoughtful discussion about them from a good number of engaged and knowledgeable posters.  

But, yeah good point glad you took the time to make it.

In reply to Si dH:

So what we see in the demographic plot of outbreak areas is the sum of all their fears...

I don't know how much is increased visibility of them, due to the general fall in case rates, and how much is a genuine increase in frequency of events.  Top level cases aren't falling that fast, so presumably more of the later.

In reply to Misha:

I think so, but I'm not going to declare it off the back the last data point in the set, I'd have to hand back the cornflake packet with my degree on it if I did that.

In reply to Offwidth:

The gender aspect of Covid illness and death has - to my reading - had much less air and political time than the minority communities aspect, yet the gender difference in death rates is stark.  I don't know how much is biological, and how much is down to gender bias in employment roles, but it seems Covid ruthlessly exploits inequality in any society it come to.

In reply to Monkeydoo:

I expect everyone had enough of these numbers 6 months ago.  For some people, trying to understand them is a good way of allaying fear otherwise driven by contradictory and confusing reporting.  For some other people, ignoring it all as much as possible is their way of coping.  We're all different after all.  However, for someone to think it's "made up", let alone to think it's done  for the purposes of fear mongering, well, I don't think that's a productive or helpful way of coping.   Inspired by your derogatory "shitehawk" though, I've decided that if I return to my super-hero alter-ego, he's getting re-imagined as The Black Kite.

In reply to wintertree:

Wow. Yes, obvious now you put it that way, and very troubling

Absolutely. There are a lot of us who are extremely grateful for your analysis. I've said that before but am happy to say it again.

In reply to Si dH:

Absolutely - and if T&T can’t cope with local outbreaks at the moment then we really have no chance of keeping this under control until we get to a high level of her immunity. I’ve been watching the dashboard map for the West Mids closely and it’s encouraging that whilst there have been a few very local outbreaks (1-2 MSOAs going blue or dark blue), these haven’t spread and have reverted to light green or white over time. There was a city wide flare up in Cambridge a couple of weeks back but that has also been contained.

So there’s reason for cautious optimism on the whole. The situation ‘up North’ is more concerning though as the base level is higher and the outbreaks are more significant. I think there’s a case for not moving to the 17 May reopening in those areas.

In reply to Monkeydoo:

Thank you for your valuable contribution to the discussion.

In reply to wintertree:

No Covid by August. This is a very odd statement to make and he should know better really.

https://www.theguardian.com/world/2021/may/08/clive-dix-claims-covid-not-in...

In reply to Misha:

That is optimistic indeed.  Spitballing numbers don’t agree.  I wonder why he thinks that?  Be interesting to know...

In reply to wintertree:

Due to the vaccine rollout presumably. However, given that vaccines aren’t 100% effective and an as yet unknown % of people won’t have them, plus the fact that variant are being imported, it seems rather optimistic. 

In reply to Misha:

"Outgoing vaccine chief claims Covid will not be circulating in UK by August"

I think that means outbreaks (imported cases?) peter out due to high levels of vaccination so not "circulating in UK" rather than no Covid in UK.

In reply to elsewhere:

Indeed, but that's still a very optimistic take IMO, unless he knows something we don't, e.g. about authorisation of the vaccine for under 18s.  Even with that authorisation, it feels optimistic given the estimates I've seen of herd immunity thresholds for the Kent variant.

We'll see.

In reply to wintertree:

If you squint quite hard at the right hand side of Demographic rate constants for Plot 16 RED there seem to be three bands of slightly brighter red, each separated by 25-30 years, it's more apparent some days than others but at the back end of April there is one more obvious near vertical stripe with a slight rightward rising slope suggesting maybe infections (or at least T&T driven detections) move from young to old so I'd guess that's school and childcare related in 3 generation families. Or noise and cognitive bias.

I can't see anything obviously tying the regions apparently in growth together, they actually seem quite disparate but I don't know England that well. Could be mostly due to a small number of super-spreader events/clusters now the absolute numbers are low.

jk

In reply to wintertree:

The expected US authorisation has been in the news for a while, but I was astonished to hear that my 12yo niece is expected to be vaccinated within a few weeks. I guess this fits in with the talk of US states struggling to find further people to take up the vaccine at the moment.

In reply to elsewhere:

I was being a bit loose with terminology but so was he, to be fair (is there a definition of an outbreak anyway?). Anyway, it seems rather optimistic. 

In reply to Misha:

Updated plot with a couple more days of data below.

Not looking a lot like a BH effect now.  The day itself is likely depressed, giving a false drop on the datapoint for day/(day-7) and a false rise on the datapoint for (day+7)/day, where "day" is the BH.  Cases shunted to day+1 could give a "false rise".  Three rising days in a row looks unassailably genuine, and this is in the PCR only data.

There's also growth in a 4th week-on-week period, the "right hand" day for which, May 7th, is still provisional and so isn't in my plot.  Numbers below from  "Cases by test type and specimen date" below in [1].

⬆  27 Apr / 4 May (1129 > 1271)
⬆  28 Apr / 5 May (1114 > 1209)
⬆  29 Apr / 6 May (1034 > 1099)
⬆  30 Apr / 7 Mau (998 > 1004*)

* - provisional number

What's missing to understand this properly is a public release of data broken down by test type, specimen date and age.  I've noticed a few such not-requests made on here appearing on the dashboard days later, so if there's a dashboard backend person reading this, pretty please? 

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

Hopefully it's just a blip correlated with the highly and unseasonably cold and crappy weather...

In reply to jkarran:

I could buy that as a theory.  I wave my hands incompetently at trying to test it as a hypothesis.

Here's a plot showing the (rate constants for red list UTLAs) - (rate constant for all UTLAs).  A different colour map to break the perceptual link to absolute rate constants.  There also a rate constants plot for the red regions so you can flip-book between them viewing on a computer. It's looks like these regions beat had a larger growth during the Nov-Jan rising phase, and had a more +ve rate constant (later and slower decay) during the subsequent falling phase, followed perhaps by later but faster decay.

Perhaps if you squint you can see generational banding in this; ignoring the noise ridden upper right (low absolute numbers), the brightest red is in secondary school ages. 

Perhaps these are canary regions.  Or harbinger regions.

In reply to Misha:

https://www.theguardian.com/world/2021/may/08/clive-dix-claims-covid-not-in...

It's the info in the URL, "clive-dix-claims-covid-not-in-uk-by-august-vaccine" is deceptive or just plain wrong compared to the content of the article.

I'm optimistic (or determinedly keeping my head in the sand) - all UK/EU approved vaccines appear to be good enough (not perfect) against all variants so far. That appears to be true in countries with far more cases, fewer vaccinated and different variants in circulation.

My reason for optimism is....

"The seven-day death toll of 67 was down by 39.1% compared with the previous seven days"

From the same Reuters URL, a reason for pessimism...

"with the seven-day total of 14,659 cases down by 4.3% compared with the previous seven days" - barely changing.

Is it really that slow? If so, then you are right to be pessimistic about "not circulating in August". 

https://www.reuters.com/world/uk/covid-19-cases-deaths-fall-further-uk-vacc...

In reply to jonny taylor:

Good point; it makes the absence of comment here a bit odd; I can see reasons to defer making and publicising the decision until there's vaccine available for that purpose; I'm quite thankful in some ways that our problem is supply limited compared to the demand limited situation developing in the USA.

In reply to wintertree:

Yes, I suppose the UK has more time due to mostly doing things by age, and I suspect our Pfizer supplies are going to be stretched a lot more thinly than in the US. Presumably they're not going to let any Oxford/AZ anywhere near children.

If I were a clinically vulnerable 15 year old, though, I'd be starting to wonder if the government had forgotten about me, though...

In reply to wintertree:

Tbh I don't think that's really a valid analysis from which to draw any conclusions. Like you said in your original post it's best to go and look at the dashboard data for each individual area. The outbreaks that have occurred definitely do not have common cause, there seems to be a mixture of school age outbreaks, working age outbreaks and in between (student age).  Some of these areas have had high case rates in the past but others haven't. And at the moment, even within these areas the behaviour is very heterogeneous, in most of them the increased cases from an outbreak are all concentrated in to < 5 MSOAs. I think it's largely random events occurring but with an obvious biasing effect that areas with higher existing prevalence have a higher probability of suffering an outbreak on any given day. My suspicion is the number of them will go up fast come late May.

In reply to Si dH:

I think the difference in rate constants is quite interesting and "real" - whatever differences are manifesting now as susceptibility to outbreaks seem to have caused a coherent (over time and age)  shift from the national average over the last 6 months.   Quite what that means is another question.  I'll see of I can do a "difference from national level" plot for each individual UTLA but the numbers are a bit sparse to get sufficiently good rate constant measurements perhaps.

I take the point thought that looking for inter-generational banding in the ages likely isn't  valid, as the different fast-growth bands are drawn from different UTLAs.  I should have joined the dots on that before my last post...  I'm having an exceptionally slow few days right now as the vaccine / response works itself out.  I think I made the mistake of taking some time off to rest and now the last 3 months are catching up with me...

That's a good point, and it biasses the contributors to my plot towards areas that have had less decay, so the "differences" plot may be something of a self-selecting prophecy.  

If it does, it won't be long before they coalesce as the individual identities are lost to the general rise.

In reply to jonny taylor:

My nephew who is 16 and has Asthma, had his first jab early in March, and I think it was AZ.

In reply to wintertree:

Was just going to say the same thing but less scientifically. Certainly stopped dropping and rising a bit in absolute numbers (not insignificant in % terms). I suspect what we're seeing is the impact of the various local outbreaks feeding through - cue all the discussion on the last thread. Some areas are continuing to drop or holding level, others are rising. With the absolute numbers fairly low, a few local outbreaks have a not insignificant impact. As long as they're controlled, that's ok. If the numbers keep rising, that's bad news.

Of course as the 12 April unlock has continued, coupled with the colder weather, we can expect base cases to bubble up gradually in areas outside very low prevalence (i.e. a gradual increase in the base line, in addition to local outbreaks). All bets are off after the 17th... As such, I suspect the end of April saw the lowest number of cases  we'll see for a while yet - might not get as low again for another year?

The impact of the 17th will start feeding through into the numbers from about the 24th. At a guess, we could be averaging 1,500 by the 23rd. If we can keep it under 10,000 through the summer, I reckon that would be good going. Hospitalisations and deaths should of course be much lower compared to the levels previously seek at those case loads. This is all very finger in the air, just making an idle prediction.

In reply to Misha:

Dashboard map suggests another outbreak at Bham Uni... and another local outbreak across 3 neighbouring MSOAs in a relatively deprived area. Otherwise generally good and a rate of 17 for a city of 1.1m people is not bad at all (the rest of the West Mids is similar or slightly higher, again not bad for a conurbation of about 3m people).

In reply to wintertree:

So I was saying stuff up-thread about how these outbreaks had no apparent common cause because they were differently distributed among different age groups in different places.

These pages (linked from a guardian article) have put me in to minor panic mode.

(1) shows a map of b.1.617.2 prevalence as of 25th April. The correlation with areas that have since seen outbreaks (Blackburn, Bolton, Bedford, Sefton at least) is striking. Apparently (from the Guardian article) these figures exclude those from people who had travelled and those from surge testing.

https://covid19.sanger.ac.uk/lineages/raw?show=B.1.617.2&colorBy=p&...

(2) shows that in the four weeks to 7th May, there have now been nearly 1400 cases with b.1.617.2, which is over 6% of total UK cases in the period, up from 520 cases declared up to 28/04 on gov.uk and included in the analysis on this thread. I understand the nearly-1400 number does include travellers and can include some duplicates where the same person has been tested more than once, but I don't think that would be many?

http://sars2.cvr.gla.ac.uk/cog-uk/ (and click on table 3)

Couldn't really have come at a worse time with indoor mixing from Monday, especially while we aren't completely sure yet about vaccine effectiveness on this variant.

In reply to Si dH:

Someone pointed me at the next release from SAGE after the one linked last thread [1].

To quote from the summary (my emphasis): VUI-21APR-02 (B.1.617.2) was escalated to a variant of concern on 6 May 2021 (VOC- 21APR-02). It is assessed as having at least equivalent transmissibility to B.1.1.7 based on available data (moderate confidence). There are insufficient data currently to assess the potential for immune escape. There has been a steep recent increase in the number of cases identified (N=509 genomically confirmed) of this variant of concern in the UK, which includes both imported (n=157 confirmed after travel) and domestically-acquired cases. Postcodes of residence are most frequently identified as London and the North West.

This suggests 157/509 = 30% of cases come from travel importations.   That's not a happy statistic.  Table 6 (cases and travel data by region) suggests it would be over 50% which is slightly less worrying.  I don't immediately understand the disparity or what is wrong with either the report or my understanding.  Figure 10 suggests the headline number of 509 cases likely includes a lot of recent detections, the origin of which (local/travel) is yet to be determined, perhaps more of those resolved to travellers between Figure 10 and the summary. 

It's a tense couple of weeks, for sure.  I am very much awaiting the next update to the variants data, and the next technical briefing.  It does indeed seem like bad timing, but we're going to have to cross the bridge at some point.  I'd much rather we were a month ahead with first doses and two months ahead with second doses, but we move forwards regardless.  I suspect it won't be long before we have a good understanding what level of protection the vaccine offers to people of different ages and health.  I still suspect the blip in cases and death rates a few weeks back to be not unrelated, taken with the news coverage of a care home outbreak amongst vaccinated residents.

If the measured doubling time remains sub 7-days by this Thursday I'm going to be joining you re: panic mode.

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

Edit:  That Sanger institute map is fantastic, thanks for sharing.  I just pressed the "play" button on the time-lapse animation.  I wish I hadn't.

In reply to wintertree:

Playtime 😁

In reply to Si dH:

Today's update is out.  I've updated my week-on-week plot.  I've pulled the end window forwards from -3 days from the end of the data to -1, as there's already week-on-week growth in this "highly provisional" leading edge; it's almost certainly not going to go down.

There's a bit of bank holiday displacement in there perhaps, but it can only affect one or two data points max.  

So, it looks like we've had a solid 7 days of growth in PCR cases.

Edit:  Added an update of plot 18 for regional cases exponential rate constants; this plot has been on a break for a few weeks.   Unlike the week on week plot, this includes LFT data.  Let's hope the vaccine breaks almost any correspondence in the hospitalisations signal.

In reply to wintertree:

Before I join you in panic mode, which I suspect I will, I'd like to see the numbers broken down by imported cases and people who live with imported cases. Household transmission is considerably less worrisome than genuine community transmission.

In reply to Longsufferingropeholder:

Indeed - it may well be that the Venn diagram of "households where this variant is spreading" and "households where vaccine refusal is higher than average" has quite a lot of overlap as well.  Some press quotes from the EMA today that neutralising immunity is induced against this variant by the mRNA vaccines and that they await data to review on the viral vector ones.

So, I'm not changing the light bulb just yet, but I know where the different colours are.

In reply to thread:

I tried a different way of doing the vaccines plot.  As with before, the cumulative number of first and second doses are lined up as if given in sequence to estimate the time between doses.  Rather than plotting it directly as a value however, I change the colour for first doses that have been followed up to green on the plot.   Now, the 11.5 week gap from first to second dose is visible as the distance from the end of the green region to the right edge of the data on the plot.

What I like about this is that it makes it visually clear how long there is to go, if weekly dose rates remain the same, before we return to giving mostly first doses.  This is the length of the period in the middle of the graph that is mostly un-followed-up first doses - the light blue colour between the green on the left of followed-up first doses and the dark blue on the right where most doses given are second doses, so there are few first ones to swallow up.  

I like this because it's really quick to read off the quickly made plot "by eye", where-as measuring it with an algorithm would involve quite a lot of effort for minimal gain, and to a level of detail not really justified given that we can't really predict future dosing rates.

So, about 5.5 weeks until the focus of the campaign might switch back to first doses.

In reply to wintertree:

I was more thinking of average # people in a household x number of imported cases. If 157 infected everyone they live with, or even just the unvaccinated ones, how many cases would that add up to? And T&T would be right there looking for them so they'd almost certainly be counted.

If that's what we're counting, and *if* they isolate properly, I can stay at brown alert for now.

In reply to wintertree:

I like the vaccines plot. I like it even more when combined with Paul Mainwood's work. A lot of good stuff he's piecing together on vaccine rollout.

In reply to Longsufferingropeholder:

If only he used a proper platform to disseminate his work like UKC, instead of the UI monstrosity that is Twitter.  It's been great to see much scientific analysis come together on there but by the gods it can be a struggle to follow it, and it doesn't half attract the lunatics.  

Gotcha.  Given the red listing, hopefully if that's the case it'll by falling back down by this Thursday's variants update.

In reply to wintertree:

I imagine the unlock on the 12th contributed to the spread of the ‘new’ variant beyond travellers. Coupled with the prevalence in India, it’s not surprising that this has started spreading significantly here in the last few weeks.

Or it could be travellers to their households as discussed above  you’d think they would split travellers, households ans ‘genuine’ community but it wouldn’t surprise me if the system isn’t capable of matching addresses… It’s probably a bit of both. After all, households of travellers don’t have to self isolate as long as they don’t have symptoms. Crazy.

In reply to wintertree:

Just got to day this is one of the most informative threads on here. keep up the good work

In reply to wintertree:

To go with the updates for the week-on-week PCR numbers and the regional PCR+LFT numbers, here's an updated UTLA watch plot and demographic plots for all English UTLAs and for those with > 20% growth from baselining on the weatchplot.  This filters the UTLA plots more than the usual demographic one, to make up for lower absolute numbers and so more noise in the plot for growth UTLAs.  That has - for now - reduced the intensity of growth on the leading edge of the England plot.

The absolute numbers are still very low - much lower than when the "Kent" variant started to spread late last year - so even if the worse comes to pass with the Indian variant, we're in a much, much better place to face it than when control was lost of the Kent variant.  For now, I think it's a week or two too soon to jump on the doom-and-gloom waggon over this; it's definitely a time for caution but not enough is known publicly about precisely how and where this growth is happening down at the level of individuals to understand if this is a rather specific "flash in the pan" that'll have no choice but to burn out, or the start of a genuine new wave of infection.  I certainly hope it's the former - but with the round of unlocking due next Monday I'd have expected PCR cases to start rising soon anyhow, which will rater mask the understanding of what's going on here.

The UTLA watch plot is somewhat unprofessional in the layout of the annotations - a point driven home by the great discussion of the layout of route lines where intersections occur in this thread - https://www.ukclimbing.com/forums/ukc/rockfax_digital_deep_dives_part_1_-_t....  It turns out that getting a bunch of annotations on a plot without obscuring anything is a non trivial problem to automate...

In reply to wintertree:

Scary cases graph in this article on Bolton. I like Burnham’s idea of extending vaccination to all adults in the area. 
 

https://www.bbc.co.uk/news/uk-england-manchester-57075618

In reply to Misha:

There’s worse already in the provisional window - not in my plot nor the BBC one - https://coronavirus.data.gov.uk/details/cases?areaType=ltla&areaName=Bo...

Worth doing I think, but it wont touch the next couple of weeks of growth.  At the very least, locally delaying the next set of unlocking steps would seem wise, as well as potentially reversing some past opening steps for a couple of weeks.

Some of the other annoyed areas in my watch plot above (eg Shropshire, Plymouth) look to be contained outbreaks but others (eg Bolton, Bedford) have more growth lurking in the provisional window.

It’ll be interesting to see tomorrow’s variant numbers.

In reply to wintertree:

It would, but what do you think is the chance of 'data not dates' Johnson going for that...?

In reply to captain paranoia:

Some bad news in tomorrow's papers:

https://www.independent.co.uk/news/uk/home-news/covid-indian-variant-surge-...

In reply to wintertree:

You’d hope some of that growth is due to the surge testing but who knows (what I mean is this might be picking up some asymptomatic cases so the numbers might not be entirely like for like).

If it were you or I in charge, we’d have implemented all of the above and more by now. Instead the government is just going to wait until it gets totally out of control. They just don’t learn...

You’ve been ahead of the curve once again. Initial reports of this variant being highly transmissible and cases tripling within a week. https://www.theguardian.com/world/2021/may/13/uk-covid-scientists-variant-f...

The Independent article which Offwidth linked above has the numbers. 520 to 1,723 in a week (it says there could be some duplicates in there). By eye, that’s a doubling time of about 5 days, so your finger in the air estimate about a week ago was pretty much bang on. At that kind of doubling time, it’s not going to be just travellers’ households... Well, I suppose it could be if it infects most people in the household including children and they all get surge tested but I bet that’s not the case. 

In reply to Offwidth:

Bad news on the spread but the silver lining is that delaying the 21st is a pretty sensible thing to do. Meanwhile The Sun’s front page proclaims the end of masks on the 21st. Madness. I actually wonder what proportion of people would continue to wear masks in shops etc anyway. I reckon about half of those who currently wear them. 

In reply to Misha and wintertree:

A few random responses:

- the first data link I gave above apparently excludes travellers and surge testing, and still shows high Indian variant prevalence in some areas. So surge testing is not the main cause.

- the second data link I gave had updated variant numbers to 11th May so they're a flavour of what you'll get in tonight's (?) gov.uk update

- in some of these areas, detected cases are rising fastest in younger groups but there is also an increase in people in their 60s who should mostly be double jabbed by now, as well as everyone below that age. However, no signal in covid hospitalisations yet, Bolton is still running at 1 or 2 per day, so that's good.

- I am quite strongly against localised restrictions now (and, although the latest legislation is designed to allow for them, the Government messaging has been that we aren't going back to them). The mental health impact is strong and they will probably lead to a lot of resentment, unnecessary travel between areas and/or low compliance (depending on the nature), especially if the local residents feel they aren't being supported well enough. If we need to slow down then it must be done nationally, together. As to whether that is necessary, I think that will depend on hospital data. I'm personally concerned (being in my late 30s) about the prospect of rates rising an awful lot in the next few weeks, but I don't expect any policy change on behalf of people my age.

In reply to wintertree:

The public messaging at the moment is very odd. Radio 3 news at 0630 today said only that cases are at their lowest since August. The Guardian has several articles mentioning concern about the Indian variant. Boris says it's a concern. Boris says no more masks after 21 June ...

DOH!!!!!

(Bangs head on wall)

In reply to BusyLizzie:

To be fair (?) Boris did also emphasise to MPs yesterday when talking about people going back to work on 21st that the data was not certain yet and would depend on what happens over the next few weeks. Personally I doubt he'll delay it without a bi rise in hospital occupancy though.

In reply to BusyLizzie:

“Max Headroom - 20 minutes in to the future”; these days it’s about a week in to the past for the media.   Or more so that there’s a lot of desynchronisation in what is reported.

I’m quite happy that the PM seems a lot more informed of the most recent situation than in the past.  We’ll see how much that does (or does not) affect policy over the next two weeks.

In reply to Si dH:

Its a strong point, but if the whole country is held back because one area in particular is going very badly, I think the same issues affecting mental health apply but in the other direction outwards and affect more people.  The big difference is to out of area travel which as you say could be incentivised by local restrictions, and is the last thing we’d actually want.  Hopefully this is all just a large flash in the pan as returnees from India trigger a suppressible burst of activity on their return...  The SA variant has been tailing off lately.

In reply to wintertree:

Who could have predicted a surge of returnees rushing back to best the deadline and avoid £1750 quarantine?

It is not fair on those who stay home but making the quarantine free for those returning immediately would save a lot of NHS and lockdown costs of treating those infected by returnees beating the deadline.

In reply to thread:

No sign of the variants data being updated yet today.  All seeming quite political - not surprising giving the timing with regards next Monday - and we're back to "sources" telling stuff to the BBC and fore-warning of an announcement.  The announcement doesn't make for happy reading, but one way of squaring the two parts I've hi-lighted in bold could be that significant chains of infection were established before the risk was identified and surge testing was brought to bear, and that once surge testing eventually catches up, control will be rapidly regained.  There are of course other interpretations.

From https://www.bbc.co.uk/news/live/uk-57069295: (my emphasis)

Surge testing in areas where the Indian variant has been found is not working, sources have told the BBC. 

The strategy is identifying lots of cases but is not stopping the spread of the variant, according to sources.

The variant is now being seen in lots of places, with few cases linked to travel, and case numbers have been "grossly underestimated", they said. 

Extra meetings of government scientific advisers - including the health secretary and chief medical officer - are taking place to consider the growing number of cases of the Indian variant, the BBC has been told.

It is believed stage three of the roadmap for easing lockdown in England will go ahead on Monday but consideration is being given to the question of whether social distancing rules could be relaxed. 

The BBC understands that Prime Minister Boris Johnson has already signed off on a reduction of social distancing measures – but a politically difficult rethink may now be under consideration.

There are now also questions being raised about the possible need to delay stage four of the easing - planned for 21 June - which would see all legal limits on social contact lifted. 

One of the key tests for unlocking is the emergence of variants - and the increasing number of cases of the Indian variant could have an impact. 

It is believed an announcement of a greater focus on outbreaks of the Indian variant could be made some time after 14:00 BST.

In reply to elsewhere:

Or just enforce the quarantine immediately especially given they tend to sit on their arses about it for several weeks before realising that yes it might be a problem.

In reply to mondite:

Definitely enforce the quarantine immediately rather than say "get home by Friday and you don't need to quarantine".

Costs to individual health, economic damage of longer restrictions and the cost of NHS treatment of those infected by those returning are such that free quarantine might save lives and money.

Alternatively say "screw it, you've had a year to know quarantine might be imposed at any time" and announce immediate requirement for quarantine paid for by the traveller.

In reply to wintertree:

Numbers for B.1.617.2 in the link I shared previously above...

http://sars2.cvr.gla.ac.uk/cog-uk/

(see table 3)

... have in the last 24 hours or so gone from 1401 to 1649 in the UK in the last 28 days, and 1728 in the UK all time. The number for B.1.1.7 in the last 28 days is 17274. 8.33% of total cases in that period are now B.1.617.2, 87% of the total are still B.1.1.7 and the rest are a mixed bag of other variants. Obviously B.1.617.2 is increasing rapidly and therefore would presumably be something over 10% (maybe a lot over) in, say, the last week. Unfortunately it looks to me like it could soon be taking over like B.1.1.7 did, once we open up and allow greater spread*. The quarantining has surely come too late.

* I suppose it's possible that this variant somehow depends less on indoor contact than B.1.1.7 and therefore gets less : benefit' when we open up indoors again but that seems unlikely.

In reply to Si dH:

Indeed; thanks.  Sorry I could have been more specific; the government summary page [1] hasn't had it's Thursday refresh yet.  Last week the "bad news" refresh was apparently delayed out of consideration for the elections, and the BBC source this week suggests an announcement is coming, and again the page is not yet updated.  Pause for thought.  I haven't yet taken enough daily sets of numbers from the COG source to understand if I can drop it in as a replacement for the numbers from the summary table or not for my noddy plots; not that I necessarily need to keep these up given their visualiser.

Edit: You updated your post whilst I was typing to more precisely reference what the 8.33% means; as it's a rising phase from next to nothing over the last 28 days against a very modest fall in cases I suspect that means quite a bit more than 10% of new cases are now this variant.

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

In reply to wintertree:

Re: Vaccine plot - Do you know what percentage of first doses each day are Pfizer / AZ? If so, you could add a predictive line for 2nd doses based upon the standard intervals between doses. It would be interesting to see if the real second dose numbers match the predictions.

In reply to Longsufferingropeholder:

The problem is the UK isn't making households isolate well enough, so household transmission = community transmission sooner or later.

In reply to Toerag:

As far as I know, there's no regularly updated, detailed breakdowns available on the vaccine uptake.  

In reply to  Si dH:

The government variants page has been updated this evening. It gives a total count of 1313 cases for VOC-21APR-0 (B.1.617.2).   This is smaller than the 1649 number in the table you linked - there is a rider on one of the other tabs on that link which states "NB Number of genomes is not equal to number of COVID-19 cases as data have not been deduplicated." this presumably explains the difference.

1313 cases is still an alarming enough number, mind.  I've put a plot updated with today's table's data in below.  A reminder for all that this is cumulative numbers, not daily or weekly numbers.  However, so long as a measure is increasing at a fixed rate exponential as indicated by a straight line on this log-y plot, the cumulative and daily measures are both exponentials with the same doubling times.

This all raises lots of questions, not many answers are in the public domain right now it seems.  There are two key measures I'm looking for to suggest how this is going to develop.

1. Does the exponential rate remain this aggressive over the next 7 days?
2. Do hospitalisation signals start appearing in the next 5-15 days in the regions where we are already seeing growth in cases?

In terms of grading the answers:

• 1: yes, 2:yes - very bad news
• 1: yes, 2: no - interesting news, badness to be determined 
• 1: no, 2: yes - bad news 
• 1: no, 2: no - that was close

I very much hope the exponential rate starts to moderate soon - throughout the second wave when we had good measurement of infections detected as cases, we'd see an initially aggressive exponential growth rate in a local area, and then see this moderate apparently in response to rising local prevalence of the virus.  Poster minimike in particular commented on this at the time IIRC.  Hopefully whatever "automatic" locally moderating mechanisms are at work there (behavioural responses to rising cases? saturation of the limited size high risk of transmission network?) also kick in here PDQ. 

What's been said so far about this variant suggest to me that vaccines reduces the severity of health consequences but offer less protection against catching and transmitting the virus.  What that means for individuals is clearly that they're in a better position for this variant than a few months ago.  What this means for healthcare is a subtly different matter, and one that remains to be seen.  Perhaps there is a beginning of an up-tick in the hospitalisations for London; it's hard to know just yet.

In reply to wintertree:

Tc~5 days??! 

YIKES.

as you say let’s hope the purely empirical rate limiting effect we saw last time isn’t purely empirical after all..

In reply to wintertree:

Amazing (and terrifying) just how consistent most of those exponential rates are.

In reply to captain paranoia:

The plot below shows the residuals between the exponential fits and the data, normalised to (divided by) the square root of the model fit.  (it would look basically the same divided by the square root of the data).  This is normalising it to the expected magnitude of statistical variation from a Poissonian process. 

The normalised residuals are about the magnitude expected (±1) so there's no great support for saying the curves are a form other than exponential.  However, the residuals for both Manuas and the other India strain are well correlated with each other - this to me suggests some of the micro structure in the data is coming from factors pertaining to the set of processes including contact tracing and sequencing.

The other curves (not included on this plot, but on the previous one - they cluttered this one) have not been on a fixed rate exponential and are both now slowing down significantly.  I have wondered with the "SA" variant in particular if it's some sort of flash-in-the-pan effect that we're going to see play out here too.  It would be nice to think so.

In reply to wintertree:

Mitigations to present to the bulb-change committee:

i) We're still 14 days since the epic super-mega-turbo-uber-multiple seeding event. If you dropped a fleet of 747s full of this variant into the country, started running round with swabs, and wrote up your findings after 2 weeks, what would that look like even if this variant was identical in virulence to 1.1.7?

ii) Decide how you feel about stereotyping then apply as many regional/cultural factors as you're comfortable with to estimate household spread and the Venn diagram of vaccine uptake / lockdown compliance.

iii) That's three points you've drawn a line through. I know the third appeared right on the line, right where it would do if this were really bad news, but that only means it's *very probably* really bad news. It's not definitely really bad news until the 4th point is bob on the line too.

In reply to Longsufferingropeholder:

I mean, 5 days is just nuts. That's off the charts. To the point that it just can't be real in the sense of transmission alone driving it. Something in the sampling is pushing that apparent rate up. 

In reply to Longsufferingropeholder:

I too am waiting for the 4th point; that’s about the point I think where surge testing playing catch up on live infections would have to start running out of steam when it comes to generating a faster rise than actual infections.

The other mitigations all have merit, although the sub committee in favour of changing the bulb wish to present information on rising PCR cases over the last week; that can be rebuffed with suggestions  of household transmission from travellers, or could have until lunchtime today...

However there is the unidentified government source in the BBC link above who says “The variant is now being seen in lots of places, with few cases linked to travel, and case numbers have been "grossly underestimated", they said.”  

Edit: thinking on that, “have been grossly underestimated” could imply that there were a lot of live infections previously under the radar that are now being found, so this could imply that much of the rise is due to surge testing; the exact opposite of what I first thought it to mean.  An ambiguous quote from an anonymous source.  Hard to clarify...

In reply to wintertree:

I mean, completely aside from the variant debate, cases are definitely not falling any more. We've been in the "Are they? Aren't they?" region for long enough now to say for sure that the big picture is not going the right way. But it's come later than I had thought; I at least have a vaccine appointment booked, so the bus I was expecting my (our?) generation to be thrown under is late. 

Sucks to be a millennial. Again.

In reply to wintertree:

They also threw in an "according to sources" for extra journalistic credibility. 

Needs both decisive action and more data.

In reply to Longsufferingropeholder:

We've just had Eid at the end of Ramadan. I don't think that's going to help at all. There's likely to have been more close contact amoungst an ethnic group with lower vaccine take up. Will take a couple of weeks to see the effects.

Overall, I think the main thing is "are hospital admissions going up"? Must be about time we started seeing that following the cases increase (or not hopefully).

In reply to wintertree and longsufferingropeholder:

A few interesting points in a BBC article this morning about Bolton. Assistant director of public health there says that they had some travel cases originally but it is now in widespread community transmission. A doctor there said he had started to see increased numbers coming to a&e with covid but so far in younger age groups (hopefully that doesn't mean this variant affects younger people more.) And, there is a map in the article showing vaccine take-up across Bolton, which certainly shows some (not complete) correlation between low vaccine take-up and high infection; this is obviously concerning in terms of likely hospital impact. They are sending a vaccine bus in to those areas apparently.

On a local level I've been intrigued by Formby a few miles up the road, which is where most of Sefton's cases are coming from, and not primarily in school ages (although, according to local news one primary school has temporarily shut alongside a gym/swimming pool and a pub after outbreaks at each). It's the opposite of somewhere like central Bolton - the most affluent town in the area (Liverpool footballer central), lots of big houses, lots of open space and a beach, very un-diverse so no tradition of multi generational households, and by any measure I have seen it comes at the bottom of the nationwide deprivation indexes. However it is worth saying that the outbreak there is not accelerating as fast as the Bolton one now, it seems like it has crested a first peak (although surge testing is just beginning today so case rates may now go up again). Ironically the (very) deprived bits of Sefton at either end (Bootle and bits of Southport) have no problems at the moment. Just shows how this can affect anywhere and has a very large element of randomness. 

In reply to Si dH:

I guess what I'm struggling with is whether this would look any different if we'd seeded a few hundred cases of b1.1.7 in exactly the same way. All of the things you mention would happen. Cases would go up, vaccinated people would show up proportionally less in the hospitalised cases so anecdotes of more younger people in a&e would surface, the areas that always lead growth would lead growth. 

And still, after all this, nobody's doing anything to get ready for the giant fire breathing lizards. When will they learn.

In reply to Longsufferingropeholder:

Back of a brain cell says with gen time of around 10 days that makes R~4!

In reply to minimike:

Uncertainties on characteristic time and doubling time make me not want to think about an R, but if you do, and 5 days is right, better learn to love sitting in the house.

In reply to Longsufferingropeholder:

You might be right, but, I'm far from convinced we have seeded hundreds of cases in any one local area, that seems unlikely.  We saw data a couple of weeks back showing the number of cases of the first Indian variant seeded through travel in different parts of the country, and I thought they were pretty well distributed with a concentration on London. If this was all caused by seeding I think you'd have a larger number of more modest outbreaks rather than a few big ones? It looks more likely to me that a smaller number of seeded cases have resulted in random super spreading events in certain areas.

Whatever the answer it seems to me that it's obviously here in sufficient numbers now that we aren't going to get rid of it if we open up fully as planned on Monday without some pretty special action. We don't know if it will take over as fast as B.1.1.7 but I think we can be fairly sure it isn't going to be immediately out-competed based on what has already happened both here and in India where both variants coexist. So the remaining questions are how fast it spreads from then, whether it is more or less harmful and whether it evades the vaccine.

In reply to Si dH:

Those maps on Bolton offer hope that the worst of their outbreak could be self limiting and self contained, with bleed through to surrounding areas being moderated.  Also promising to see the photo of people queuing for the mobile vaccination unit. 

> Just shows how this can affect anywhere and has a very large element of randomness. 

Indeed.

In reply to Si dH:

Having grown up in Bolton, I can believe it. How can I say this objectively.... The inhomogeneity of intense concentrations of people who may have had cause to visit India is remarkable. And the areas listed on the BBC article were exactly those places.

Yep. This. This is what we don't have much evidence for. Early indications are that the vaccine will work. And the mutations look really similar to 117's. We'll have to see.

In reply to wintertree:

You can finally get rid of your deweekending algorithm!

https://coronavirus.data.gov.uk/details/whats-new#change_to_source_of_pilla...

In reply to Si dH:

Nice one government.  This should make it much easier for them to understand developments at the leading edge of the dataset with confidence.

Previously, the number of tests conducted was collated daily from individual laboratories in England based on the day the test was processed. These data are now supplied by NHS Test and Trace digital management information systems, using the date that the test was taken (specimen date). 

Also nice to know I correctly deduced the problem some months back - it was the undergraduate tests being assigned to their home address and not the one on the test's paperwork that gave it away.  We're getting the point where the testing capability is truly phenomenal in terms of the scale, reporting latency, associated sequencing capability and so on.  I wonder what can be done with it after this virus?  For now I hope we can find ways to effectively donate sequencing internationally (through samples posted to us) both out of charity and out of enlightened self interest.  

In terms of where we are today on the leading edge off the data, and trying to understand the ambiguous quote discussed above about a lot of undetected cases, the latest data point on the week-on-week derived rates constant plot is back in to decay; the data point is still a bit provisional and could go up, but it feels compatible with a period of "catching up" to live infections, which is compatible with one interpretation of the ambiguous quote.  If so, the 5-day doubling times on the variant were not "real" in terms of infections, and won't be sustained.   As always lately we have to wait and see what the next week brings, but I'm vey hopeful this is all just a flash in the pan - our largest yet, and one that really should be raising alarm bells over how we handle in bound travel.

Edit:  Briefing from the PM due later over the variant.  I'm hopeful it's not going to be cautious, concerned and towards the "good news" end of the spectrum.   

In reply to wintertree:

They seem to have dug out the red flashing bulb for Glasgow. Every time I look at the news today it reports a higher case rate. This had better be driven by increased testing, because  the pollokshiels numbers are pretty terrifying 

In reply to jonny taylor:

Gosh, just looked at Glasgow - what's happened there?

In reply to wintertree:

Doesn't a move to "date of test" from "date of processing" create a need for what in insurance world we could call an IBNR (incurred but not reported) - the last few day or days will be systematically underreported because (making up numbers) 70% of tests done on day X will be processed on day X, 90% by day X+1, 95% by X+2 etc. This may already overlap with your leading edge caveats, admittedly!

In reply to AJM:

This has always been the case anyway. Through most of last year there was a delay of up to 5 days before the specimen (processing) date data was considered complete for a given day, which is why whenever you see weekly average rates reported they are always for the week ending 5 days earlier. However, more recently they've been very quick and generally only the previous two days have any significant under reporting. You can look at all this on the dashboard if you know where to look.

In reply to AJM:

That is an excellent point.  The effect would I think be to create an artificial rise in cases for a day or two where old ones are counted by sample processing date and new ones by test date, where the dates are the same.  Ideally the switch would happen away from the weekend to minimise the effect.

Edit:  I am assuming that this change is being made going forwards from a point in time, reducing the reporting lag for days from that point on - creating a disinclination in the timeline.  The alternative assumption would be that the entire dataset is being overhauled to use the new time line, but it seems unlikely the information required to do this is all present. 

That would explain the two very fast growth rate points (k~0.06x / day) on my latest week-on-week plot.  The leading edge of that plot looks very "ugly" and that could explain it.  I'd been assuming it was down to surge testing catching up to the leading edge of infections, but I think you're on to something.

In reply to jonny taylor:

I went a bit myopic towards the English data when things were getting bad and haven't found the time to figure out regional Scottish data since.   The government dashboard for Glasgow looks worrying in the provisional window, although AJM's point could be behind some of that.

With all this uncertainty, I should be in the light bulb business.  I'm still hopeful it's going to be a set of limited duration outbreaks and that the apparent growth rates have been coming from surge testing.

In reply to Si dH:

That's effectively a second layer of reporting lag is it - between the processing of the test and getting it to the central dashboard?

I think this change retains that lag, and introduces an extra source of lag between the test being done and being processed.

Swings and roundabouts, obviously - you lose the weekend blip but increase the lag before a day is complete?

In reply to AJM:

What you actually need to know in order to analyse the progress of the pandemic is the date of infection. Obviously the date of test is closer to that than the date of processing is, so the change removes an unknown lag vs what we actually want to know. There is still an unknown lag (infection-->test) but it's smaller than it was before (infection--> processing.) The lag between testing and processing used to be unknown (and inconsistent, hence the weekend issues) but will now be known to be part of the observed lag between specimen and reporting dates.

NB I edited this paragraph and removed a paragraph I wrote about processing to reporting lag. We don't actually know what processing in the old system meant - I suspect when the sample arrived rather than when the result was known, but it could have been somewhere in between.

Anyway, it's clear the lags are all generally very short now, the system works pretty well.

In reply to wintertree:

Latest SAGE meeting minutes and variant papers are up. This paper dedicated to b.1.617.2 has lots of material available to support us in any proposed doom-mongering activities...

https://www.gov.uk/government/publications/phe-investigation-of-novel-sars-...

These three pages I have attached stuck out at me, they all basically lead to the same conclusion. S gene positive is now being used as a surrogate for b.1.617.2, it's a good job taqpath didn't change their process!

Time for bed.

In reply to Si dH:

Interesting that contacts of travellers are a fairly small proportion of cases. Admittedly a lot of cases are under investigation but you’d think that most traveller contacts would be identified pretty quickly. So it’s community spread - not surprising if it’s potentially 50% more transmissible than Kent.

The jury is still out re vaccine escape on this one but once again the government is chancing it. I’d have put back the 17th by 2-3 weeks (announcing that a week ago) to try to slow this down or at least get a better idea of what’s going on. Instead it’s Covid happy hour. Especially as the weather is poor so people will start socialising indoors.

Someone from SAGE was quoted as saying we can’t panic over every variant. I have sympathy for that view but we can’t afford to not respond to something with an apparent 5 day doubling time. A couple more weeks of current measures to assess the situation would be a price worth paying I think against the risk of completely losing control of the situation and potentially having to reimpose measures later on.

I’m told it’s still taking a couple of weeks from test date to sequencing data being fully analysed, so what was reported on Thursday is a couple of weeks old. 

In reply to Misha:

Reading between the lines of what was said yesterday, I think there's an acceptance the bird has flown on this one already and the spread can't really be stopped without more significant restrictions again (ie status quo not enough.) We don't yet know whether there is significant vaccine escape and/or whether the increased transmissibility is enough to cause a rise that threatens the hospitals in the next stage (the sage minutes discuss that.) So in terms of actions, it's just going to be a case of watching cases rise, trying to maximalise test/trace to slow it down and monitoring really closely the rate of spread and numbers of vaccinated positive cases with a plan to reintroduce restrictions if necessary. This is also the only logic I can see behind accelerating second doses to 8 weeks - you get fewer first doses done and the evidence suggests the 12 week gap gives better long term immunity (acc Whitty yesterday) - but the 8 week gap means far more people will have had their second jab and hence reduced hospitalisations if there is a big spike in the next 1-2 months.

Hope they're all wrong! Things might still move quite slowly if the weather warms up.

In reply to Si dH:

Another genius decision from our glorious leaders. Who would have thought of sitting on their arse for a several weeks about whether quarantine would be required for India and then giving the best part of a weeks warning for people to avoid it could possibly have any downsides.

In reply to Si dH:

The slightly inferior long term protection from an 8 week second jab is probably moot; we're already taking about updated autumn boosters. Shifting to 8 weeks doesn't necessarily shift the first jab completion back by much if at all, since we don't seem to be supply limited any more, but it does show willingness to minimise hospitalisations in >50s at the expense of infections in younger ages. It might be a forced decision since that ship may have basically sailed already. Certainly has in Bolton. There's an argument that every first dose given to someone who is or will be infected in the week either side is a dose wasted, which is why surge vaccination is a nuanced, preventative tool and a bit daft without an accompanying lockdown, which someone needs to tell Andy Burnham.

In reply to Longsufferingropeholder:

I didn't realise that - any further info? Seems surprising if so that we aren't doing more.

In reply to Si dH:

It feels like we’ve stepped through the looking glass this week.  As well as that, the study validating 12-week gaps as more antigenic is out for the Pfizer vaccine, and we’re shortening the gap.

https://www.theguardian.com/science/2021/may/14/delay-in-giving-second-jabs...

I’m still clinging to my “flash in the pan” hope, that this will moderate away from rapid growth as it runs in to the social and vaccine-immunity boundaries around the places the variant has landed.  But the 8-week news and the much promoter shift to discussions over how much more transmissible than “Kent” this could be are not encouraging.

In reply to mondite:

 Re quarantine, we’re currently “enjoying” the spring cleaning opportunity afforded by this.  They have got their act together to a degree, daily phone calls and a company hired to do home checks.  But, I don’t see how it can be effective if you are quarantining in a house with other people going about their normal life.  

(They also seem to have got their act together at border control, all paperwork checked and then on to now open e-gates,  job done in 10 minutes). 

In reply to Si dH:

Paul Mainwood, Twitter. Shame about the platform but good analysis.

In reply to Misha:

I'm a bit out of the loop on how effective track and trace currently is but the increase in the number of cases under investigation suggests to me that it is already overwhelmed by the case load. They can no longer keep up with demand and don't know where the transmission occurred for the majority of cases.

As you point out, finding travellers should be straight forward. But now that there are so many cases under investigation, I think you have to assume that they are no longer in a position to rule traveller contacts out for any cases.

In reply to SDM:

True What I meant was household contacts. If the traveller and their household self isolate, the cases are contained. If they’re going to work etc, it’s out in the wild and onward transmission will occur. There should really be separate categories for household and non household traveller contacts. The latter will be harder to trace as you say but effectively represent community spread anyway. So what I’m saying is it looks like it’s being primarily driven by community spread either directly from travellers or from onward transmission. Traveller related spread should slow down due to red listing but it’s too late as it’s roaming wild...

In reply to wintertree:

Re the 50% extra transmissibility. That would go some way to explaining the explosion of cases in India. I recall that was debated here a week or two ago.

The modelling graphs are scary. 40% more transmissible gets us to last year’s hospitalisation levels, peaking at the end of July. 50% would be well above the January levels, peaking in the second half of July. Here I was hoping for an ok summer with no lockdowns to worry about until autumn... It’s gone full circle: a few weeks ago there was still talk of a significant third wave in July / August; then consensus seemed to shift towards a ripple rather than a wave; and  now we’re looking at a wave again.

See p7 here https://assets.publishing.service.gov.uk/government/uploads/system/uploads/...

Here’s the Warwick paper. Worth a read for those interested to understand the assumptions and also the baseline prediction before taking into account the potential impact of the new variant. Stage 3 is 9k admissions and 1k deaths. Stage 4 is 35k admissions and 7k deaths. Obviously there’s a range behind these numbers.

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

In reply to Misha:

There are olive branches in there.

Point 21: SPI-M-O is therefore confident that B.1.617.2 has a significant growth advantage over the UK’s currently dominant strain, B.1.1.7. The difference in growth rates between B.1.617.2 and B.1.1.7 is consistent with the former having a transmission advantage of more than 50%; this is based on observed growth that has already happened and it is unclear whether this same growth advantage would apply to sustained wider community transmission regionally or nationally. Resolving this question of the applicability of this growth advantage to the wider population will be difficult while the number of cases are small and relatively focussed.

A couple of likely mechanisms here have been alluded to up thread and certainly suggest themselves.  I’m still hoping that these growth rates can’t be sustained - in the parlance of point 21x that this transmission advantage doesn’t exist in the wider community.  

I’m going to keep my positive - but deeply concerned - outlook for now; we’ll see what developments in the data and the understanding the next week brings.  This will I think be seriously confounded by the effect so of the next round of unlocking however.
 

In reply to wintertree:

“Doom mongering scientist in positive outlook shocker!”

someone call barmatt

In reply to wintertree:

Good to see this thread is keeping up with the experts!

If it’s 50% more transmissible, we’re screwed and shouldn’t be going ahead with Monday’s relaxation, never mind the 21st. Lockdown in early July - just what we need this summer... My point is I would put Monday back by a couple of weeks to figure out what’s going on. No need to stoke the fire.

Next week’s numbers won’t reflect the impact of the 17th, particularly as the variants data is over a week behind anyway. So if we see another doubling, that’s going to be bad news.

Given the relatively small proportion of travellers and contacts of travellers in the Indian variant case numbers, I think it’s roaming wild and so I’m not optimistic. But then I’m not an optimistic person by nature anyway.

In reply to minimike:

He’s busy hugging friends and family. 

In reply to wintertree:

Here is another glass half full perspective...if this variant turns out to be more transmissible but not hugely so, or if the vaccines take the edge off it to the better end of the range of predictions, then it might take over from b.1.1.7 with only a fairly moderate case spike. B.1.1.7 is 30%? more deadly than other previous variants and we have no comparable data for b.1.617.2 yet...so if we are lucky we might get a (slightly) less deadly variant taking over.  Fingers crossed!

In reply to jonny taylor:

I think this is the reason. You can now fly to the US and walk up and get a vaccine. But by opening up age groups we are keeping the vaccination numbers up. At our school around 40-50% of those over 16 are now vaccinated.

We're still vaccinating millions but we seem to be slowing with the groups that can. There are millions who will get vaccinated though but are scared to lose days at work. 

Selfishly I'm glad I live in MA, we're about 60% vaccinated with one dose and will likely hit herd immunity locally this summer. It's really affected my thinking about where I want to live next. All of New England is about 45% fully vaccinated.

In reply to wintertree:

Dashboard numbers for Bolton NHS Trust are quite a few days behind, especially for admissions, but number in hospital increased from 12 to 18 on Tuesday after plateauing around 12-13 for weeks. Last time it was above 18 was at the start of April.

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

In reply to Roadrunner6:

All my relatives in ME have both doses now. 

Planning to visit in July, for the first time in 18 months,  but not sure of it is going to happen......

In reply to Misha: Forgive my ignorance in this subject but when people talk of the increased transmissibility of this Indian variant, is that being measured from Indian data principally among unvaccinated people or have they the ability to accurately estimate its transmission among U.K. people  who potentially have been either fully or partially vaccinated.  

In reply to didntcomelast:

That’s not ignorance, that’s the critical question.  To my reading of point 21 in the expert document linked above, the current answer is “give us time and more data”.

Edit: sorry; I mis read your question.  See posts below with Jon Read.

In reply to wintertree: thanks for the response.  Perhaps it would be wishful thinking if this variant struggles and fails in the wider U.K. community due to vaccination levels but after conversion with my 25yr old who runs a vaccine clinic the conversations she has been reading on social media among people her age is that they won’t accept another lockdown. She doesn’t support that idea but can understand the frustration of a large proportion of people who have had enough. More time spent under restrictions would suit me but I suspect the government know they can only lock down for so long. 

In reply to didntcomelast:

The estimates I know of are based on case data from the UK. For example, figure 1 from this document: https://assets.publishing.service.gov.uk/government/uploads/system/uploads/... 

In reply to Jon Read:

Sorry I mis-read didntcomelast’s question as applying to unvaccinated vs vaccinated people in the UK.  I agree with you that the observations of higher transmissibility come from data from the UK. 

My reading of comments that they’re not able to link all cases back to travel and several points in that document, e.g. “this is based on observed growth that has already happened and it is unclear whether this same growth advantage would apply to sustained wider community transmission regionally or nationally. Resolving this question of the applicability of this growth advantage to the wider population will be difficult while the number of cases are small and relatively focussed” is that there isn’t enough longitudinal data on chains of infection to figure out how much transmissive advantage is rooted in the low vaccine uptake in certain areas vs social factors in those areas vs in the strain itself.  Given the location of important events from India (and subsequent outbreaks) and some areas of low vaccine uptake likely have a correlation, the data is confounded right now.

In reply to Jon Read: That’s not good  then.  

In reply to wintertree:

It's always biased, alas. So, very likely, yes. And there's always the possibility that this is a network effect playing out, the new variant having happened to find a sizable pocket of well- and inter-connected susceptibles (either through low vaccination uptake, larger households, workplaces/schools/colleges, etc for e.g.).

In reply to Misha:

All newspaper headlines today seem to be warning that the Indian variant could delay 21st June unlocking. Looks like the start of a softening up campaign to prepare the public for a delay.

In reply to alan moore:

I'm hoping things are OK by then. My parents are planning the fall but I think late summer could be the safest time. We're still going to have a good 25-30% of the country unvaccinated so a winter surge seems inevitable here. 

There are talks between Shapps and Buttigieg about a US-UK link. But as it stands I think both countries still don't allow it, not just the UK.

In reply to wintertree:

I suspect it’s more transmissible per se, among the unvaccinated. Open question whether there’s an element of vaccine escape as well.

My reason for thinking this is that the various other strains aren’t spreading as quickly among similar demographics. In fact the numbers were reducing until recently in many urban areas. It would also help to explain the disaster in India. 

In reply to Roadrunner6:

Plus the risk of vaccine resistant variants sending us back to square one. This global disaster has a long time to roll yet, given the slow vaccination drive in many countries. Unfortunately for everyone...

In reply to wintertree:

Good thread here: https://twitter.com/ThatRyanChap/status/1393593980944896000?s=19

Lots of information, he concludes the Indian variant is 40% more transmissible.

Otherwise, wait two weeks.

In reply to Misha:

I shudder to think where we'd already be with this Indian variant if it had got here at end of Jan (or pretty much the same thing - if our vaccination program was 3 months later than it's been).

At the moment everyone seems to be "it's very worrying but we need more time/data to be able to properly see what's going to happen".

3 months ago it would have been "oh s**t, here we go again".

Hopefully the vaccines are moderating the seriousness of this variant's illness, and our vaccination levels are already high enough to stop this developing into a "threatening NHS collapse" wave like the previous ones. Cases/hospitalisations/deaths are all no doubt going to go back up significantly, but hopefully not to the levels we've seen before.

In reply to Michael Hood:

To be fair, 3m ago we were in lockdown. It would have spread but nowhere as fast as it will do now.

Re your last comment, it depends how much more transmissible it is. See the graphs in the papers I linked above. 50% more transmissible = higher hospitalisation numbers than we had in January. 40% = same as last spring. 30% = bearable. Clearly there’s a modelling range involved and various assumptions, notably around vaccine effectiveness (they assume 90% effective against hospitalisation). Pause for thought. 

In reply to Si dH:

Reportedly, the shorter interval reduces transmission, and that's what they are going for at the moment, rather than absolute immunity. Given the talk of an autumn booster, this does seem to make sense.