UKH

I've been downloading the daily "Cases by specimen date" data from the government dashboard for a month or so now.  This lets me see how many days it takes for the dashboard to report cases.  

I've made a plot for up to 5 days ago showing, for each given day, how many of the cases are reported 1, 2, 3, ... days after the date the specimen was taken.  I picked 5 days as the cut-off as most specimen data is reported within 5 days; still the right-most 3 days remain a bit provisional as there could always be an unexpected large dump of data more than 5 days after the specimen date.  However, any more data release can only make the story worse, not better.

The numbers drawn on the shaded regions are the "n" values from the y-axis. 

So, as of 5 days ago, we have a record breaking low of only 15% of cases being reported within 2 days.  

I think that this "reporting lag" corresponds to how many days it takes for cases to be put in to contact tracing, given the press coverage of the "Excel debacle" which lead to both under-reported cases and cases not going to contact tracing.  Given the statement in today's news article [1] that "figures showed just 15.1% of people who were tested received their result within 24 hours" I think it's likely the reporting through the dashboard happens a day after test results are returned.

Shambolic.

The last few days (which aren't shown in the plot below as the data is provisional) look to be releasing this backlog.

What this suggests to me is a system that was not built with scalability in mind, and as the number of cases increase the system keeps hitting new barriers to scaling, reporting (and I think entry to the test and trace system) backlogs, then the barrier is addressed and the backlog released.

Shambolic.  If the system was not designed to be scalable it's not fit for purpose.  

An analysis of variance (about a polynomial fit) in the data suggests there's > 10x the day-to-day variation one would expect from statistics, and indeed > 10x more than is in cases or admissions data.  This is data by specimen date, so it suggests serious problems taking tests at a consistent rate - and this extends well beyond the "weekend effect".

Shambolic.

At this point I'm tempted to say we'd have been better off spending £12 Bn paying anyone with symptoms to self-isolate, and contact tracing them and paying their closest contacts to self isolate as well.  Go back in time a month or so and we had 30,000 people being detected as positive per week.  Pay them and their 5 closest contacts £1,000 per week to self-isolate for 2 weeks and you have a cost of £0.36 Bn per week (as two weeks of isolating people overlap).     This would probably have been far more effective than this shambolic system, keeping rates from rising anything like this and stretching the £12 Bn out until next year.    Okay; I only gave this idea about 2 minutes thought, but this system appears to be pissing money away for no real benefit.  If it's currently taking 3 days from a positive to contact trace 60% of cases it's  almost too late given the incubation times and timing of the pre-symptomatic infectious period. 

End Of Rant.

[1] https://www.bbc.co.uk/news/uk-54650029

I appreciate that this could be interpreted as a set up for a “get a life” type comment, so I’d like to clarify in advance that it’s honestly not as I’m finding your analyses really interesting and helpful. 

I was thinking while I was reading this that a hell of a lot of work must be going into these analyses. How much time is it taking to gather and analyse the data?

No need for the preface.

• Downloading data files from the dashboard - 2 minutes a day.  They need renaming as - apart from the date - every file has the same name.
• Writing the script to analyse daily changes in reported number - about half an hour
• Writing the script for the main plots with doubling time analysis - about an hour
• Writing the script for today's "greys of shame" plot - about half an hour
• Cranking the handle on the accumulated scripts to update them each day - a few minutes.  It would be quicker but I have to manually pick the two y-axes ranges so they neatly share gridlines.  I should probably automate that...
• What takes the most time is trying to write a coherent post to go with the data - I have to review everything I write as it tends to be garbage quality until I review it, and I review my own work best after taking a break to do something different. 
• I'm not putting anywhere near enough time in to this to properly contextualise it all - if I didn't have a life I'd be pulling in a lot more demographic and regional stuff, although this is frustrated by the different demographic age buckets used by different sources...

Can you expand a bit on this - you think cases were effectively over reported the last few days to catch up with some that were delayed even more than usual, or have I misunderstood?

That's simply not true. I'm sure there are friends and associates of government ministers who are benefiting considerably from such public expenditure.

This is my crude daily graph, for what it's worth. Surely scary.

https://bit.ly/2FQgU32

Thanks - not as long as I imagined. You've been providing a sterling service for sure! 

On a more sombre note if you're right about the link between the dashboard reporting and test and trace data, then that lag is really alarming.

Yes, that is what I think.  It's a hunch - it can't be directly evidence based as the complete "by specimen date" data is never out of the most recent 5 days.  I've attached my latest plot showing specimen date data coloured by reporting lag.  The giant orange bars on the right suggest that either cases are about to get really bad, really fast, or there's a batch of reporting lag being drained out - or less probably both.  Given the period immediately preceding this had the worst reporting lag to date, this seems likely that it's the lag buffer draining. 

The other reasonably consistent effect in the data is that fewer specimens are taken on weekends - the bars - once all reporting lag is resolved - tend to fall under the trend line on weekends and above it on weekdays - so there is a "specimen sampling lag" as well as the reporting lag.  So I think there's double whammy occurring this Monday just gone.  The "weekend effect" can be seen in the residuals from the smoothed curve.  The normalised residuals are the residuals divided by the square root of the smoothed curve as an estimate of the standard error on the mean expected from statistical fluctuations; that the normalised residuals have exploded in the last two weeks suggests some sort of wild fluctuation in testing availability to me...   Before the variance when really bananas, a 7-day cadence was visible in the residuals and a Durbin Watson statistic suggested they were pairwise correlated.  

It's really useful to see the data broken down by region.  I haven't had the wit/patience to collect the geographic breakdowns at a granular level.  If anyone else has such plots I hope they can share them here too.

To me, that's the point about taking a bit of time to make a script, using a data analysis friendly language to analyse the data.  Instead of spending 10 minutes a day re-doing it in Excel, adding up to 10 hours over two months, I can spend 10 seconds a day handle cranking and the rest of the time building on the scripts to do more useful things.  One of the reasons it doesn't take very long is that I do a lot of this sort of analysis on other data for other reasons, and I use the same toolkit of parts - plotting library, polynomial filtering, function fitting to measure parameters, making plots for other people to read.  

Gordon - this is absolutely not a dig at you, please don't take it that way - I don't expect many UKC readers to be well practices in writing programs to analyse and plot data, far from it I think it's great when other people get in to tracking and presenting the data too!

It is meant as a massive dig at PHE whose weekly surveillance report is an utter embarrassment of badly cobbled together Excel plots manually copied and pasted in to a Word file each week (pasted as low resolution bitmaps, not PDFs, to compound the failure).   I reckon on the order of 50 man hours must go in to manually doing the same thing, over and over again, in Excel, every week to make that steaming heap of crap.  Over the last few months, that's maybe 3,000 man hours that could have gone in to developing ever more useful, insightful analysis that updates the report daily in moments from a button press, and that eliminates the kind of mistakes that creep in when doing routine tasks by hand with a lot of copy/paste.  It horrifies me to think that the surveillance report is being looked at by people making policy decisions.  

I know someone who was in the local public health observatory and did this sort of thing well.  When it was merged into the then-new PHE, it didn't take long under the new management culture before they decided to leave and go somewhere less oppressive.  I think a lot of the talent left back then and it seems the analysis team now consists of some recent school leavers clicking away in Excel. 

I've been concerned from the outset that the 'weekend effect' on reporting hasn't been addressed. The dates that people are reported as being tested, or admitted to hospital, or dying, must surely be accurate to a day (barring a few unfortunate cases). The fact that the 'weekend effect' is not corrected when those results finally get spat into the reporting system shows that someone is simply counting incoming reports each day, rather than running the  raw data through a proper analysis pipeline, which would pull out the report date, and automatically correct all graphs to present the actual numbers, giving a better view of the trends.

If I were the PM's chief advisor, allegedly a 'data science guru', sitting in my data cave, I'd have my pipeline automatically checking for case data updates, and process and spit out the figures into my immersive data presentation suite...

It bothers me a lot.  

There's a clear weekend effect in almost all "by reporting date" data.  This is always more current than the actuals data (by specimen or by time of death).  There just should't be a weekend effect in this because it's a f*****g crisis situation and having the most timely and accurate data is key to making decisions quickly.  

There are two different possible interpretations for the weekend effect in the "by specimen" date data which is actuals and shouldn't have such an effect - one is that less testing happens on weekends, and the other is that the data is just handled to a garbage level precision.  I'm leaning towards less testing on weekends but I don't know if that's due to deficiencies in the system or to the circumstances of individuals when it comes to going for tests.  Either way it's systemic and should I wish have been addressed long ago.  

If I knew anything about underground bunkers in County Durham (and I'm not saying that I do), I'd probably be too busy digging another land drain to try and keep the water out, rather than immersing myself in data... 

I can forgive Worldometer for this because they're doing it for free (well, advertising revenue), but it really grinds my gears that the BBC etc. can't be arse to put together an actuals plot.  To be fair to the government, the latest incarnation of their dashboard does have an actuals plot as well as a reporting date plot, although the entire media still goes with the headline daily number and not the actuals as their driver of "OMG disaster" headlines, and the unnatural rises and falls in the reporting date data lie at the bottom of quite a few bits of deliberate and accidental misinformation.  

It could be worse. I have seen people adding up numbers in excel....with a calculator. 

Speaking of data pipeline implementations and the difference between using a bunch of spreadsheet files and a properly specified database...

Done right, one household wouldn’t get 30 to 60 separate, legitimate (not accidental duplicate) phone calls as a result of the household being infected.

https://www.bbc.co.uk/news/health-54559228

Because a proper system would recognise that people who live together get infected together a lot and would de duplicate the line items spread across the household prepared for the contact tracers to minimise the number of people and phone calls involved.  This all suggests to me a process driven approach using some really dumb processes and not the situationally aware human judgement based approach recognised as being critical to effective contact tracing. 

Personal privacy, innit? 

"Can I speak to little Johnny, please?"

"What's this about, please?"

"I'm sorry, I'm afraid I can't discuss this with anyone other than little Johnny..."

Ive been obsessing with local and GM data, but no graphs. However, a few charts down this link and you will find an animated bar chart that shows how areas in Grter Manc have gone up/down/switched positions.  It’s very good and worrying at the same time....

https://www.manchestereveningnews.co.uk/news/greater-manchester-news/latest...

I tend to do it in my head unless it's really complicated; then I use a calculator. I can never remember how to get Excel to do anything for me! But then I get students sending me stuff in word docs which is full of atrocious spelling mistakes - it actually seems like they've either managed to manually turn spell check off, or haven't quite worked out what the wiggly red line under the "werd" means. 

We all have our crosses to bear.

GDPR doesn't stop you from speaking with multiple people in a household on one call, you just have to go through your ID verification process with each person you want to speak to.

A bit of a tangent, but will the contact tracers leave a message if you don't answer? I don't answer unknown numbers, I just call back if they leave a message. How would I know it's them?

It's nice graphics but because of the problems with the testing system it's not as useful as it should be.

A similar graph for Covid hospital admissions or Covid deaths would be much more informative.

Phone apps, test and trace, wearing masks etc..  I'm not sure if these are distracting people or giving false confidence away from the basics of distancing and good hygiene. Practice the last two and the first three don't matter so much.  

Arrgh!! Just select the cells and look bottom right!

But if "sum" isn't enabled on the bottom bar...

(I don't think I've used the correct terms)

The ability to use a spreadsheet should be regarded as a core skill, along with reading, writing and arithmetic.

It is by default.....if you are able to disable it you are already better than a calculator user.....

And for all the wrong reasons, it’s nice to see Wigan climb up a table....

Its been obvious for some time that there are issues. I found it disturbing that some planning had not been out into place to at least outline what was required. But that is old  ground now.

I think sometimes we underestimate how difficult it must have been to set up a national test and tracking system. Just the logistics alone must have been mindboggling. Deciding on which post boxes for example are dedicated to returning samples alone must have been quite a process. I suspect the actual lab testing was the easy bit ( once they had made a decision to create these big labs). Did you ever find out the lead time on those automated machines.

People think its easy to set up a national distribution system for sending out and collecting samples. It is not.You only have to be in a non city area to understand the vagaries of the capabilites of different logistics companys to cover a particular post code.This was on top of all the other distribution and online stuff that was going on. Then finding suitable locations for testing centres and so on.

For all this I believe the big issue has been not involving the local authorities in contract tracing. Its been evident for some time that this is a flaw. And now it is only as part of Tier 3 that govt is dishing out money for local contract tracing.That should have been addresses  back in the summer.

I read in Private Eye that there are 15 people sitting on the Board for the Track and Trace Operation of which only 1 person has a Public Health background.Not sure if the others are logistics or IT, but it is time for a radical shake up at the top.

It is old ground that there are issues in the system, and that this evidences a lack of any suitable planning and scoping, but it's still news that it's still getting worse.

Yes, the logistics of the samples isn't trivial - but we do have a lot of logistics experts and business in the UK.

I totally agree on the role of local public health teams - their gutting and creation of PHE was a very dogmatic politics decision IMO with Andrew Lansley as the hatchet man and it's a decision the Tory party have a lot of credibility vested in; that it's taken until now to start stepping back from it and recognising what health experts have long been saying about the importance of local public health teams shows just how invested they were.

Re: large PCR machines - no better idea on the lead time for a large one.  We got a couple of small manual ones within a week mid-pandemic but that's not a useful data point.  

Re: the board - how can there be 15 people on it?  That seems far too large for an emergency system in a crisis that needs to be responsive.   Some details given here [1].  Call me stupid but I'd put a public health expert in charge, give them a highly competent t staff command rather than a board and make sure they have an advisor who is a logistics ninja. 

[1] https://www.lgcplus.com/services/health-and-care/revealed-top-leadership-te...

I'd describe that bar chart as "doing the Rochdale Shuffle".  The university effect on Manchester is quite stark in there. 

It is.

I’ve been chatting with colleagues about ‘why the rise in Wigan’ as we where tenth in England the other day yet mid table in early August. Some of the other Grter Manc areas had themes that could be part attributable to their increase: poverty factors (eg multi generational over crowded housing), low wage economy, food processing, students, yet Wigan has low or average numbers of all these themes. Possibly points to behaviours and attitudes and/or ‘its just the way these things work out’, which points to many other areas following suit. 

I believe our hospital is starting to struggle. 

The student effect in Nottingham is a lot worse but it has sadly spread into the general population now. 4 students were fined £10,000 each last week for pretending an illegal house party was over (30 people found hiding in the bouse).

Compare the average age of Pillar 1 and Pillar 2 and look at the map.

http://documents.nottinghamcity.gov.uk/download/8759

Whilst I was pretending to be the voice of the T&T provider, I suspect a simplistic approach to privacy, and dealing with individual cases in turn, is probably at the heart of the problem. Whilst the whole point of tracing is to find associations, I suspect the system isn't set up to find, and deal with, direct associations, as in people living at the same address. So they will blindly take each case individually.

I wouldn't be surprised if there's a contract that pays by the number of calls made...

I did my ONS/Imperial/IPSOS MORI/NHS random test Sunday morning. It was picked up by a courier about 1300 Sunday. I got my result back yesterday at about 1100 by email. So I guess 3.5, almost 4 really, days from pickup to result. I don't know if that's good or bad, or whether those random test are de-prioritized compared to ones from people with symptoms?

BTW, I noticed in the logbooks last night you had climbed in Goddard's Quarry a few weeks back. Thinking of going this afternoon - what did you think? OK or bag of shh...?

Yes we do have a lot of logistics experts in the UK. But it is still fragmented in terms of actual delivery. For example just coordinating one driver to go round a series of pick up points for collection by a certain time to hit a distribution centre is very fragmented across the entire country.I know for example being stuck in the wilds of Cheshire that some logistics companys do it very well in our post code. But then you speak to other people in other parts of the country and the same company cannot do it.Try say Ilfracombe ( I know this because of a supplier there). These things are not simple despite expertise.And it works both ways you have to get the sample kits out and then back in for every postal code throughout the UK.Then you also have to have the sample kits in the right place and in sufficient quantities in the first place. A logistical challenge.Amazon does it well, but only in certain areas. Post Office same. Etc

I also fail to understand why everything falls down at the weekend.This system should be working 24/7.But there is probably not a logistics network to support that.

Something not right at board level.

On first glance, that report is absolutely damning.   

I was pleased with that news.  It's time the kid gloves came off with the undergraduates.  Yes it sucks that they can't have parties.  But everything else is about to suck worse and it looks like they're not absolved of blame for that.  Hopefully highly publicising the scale of the fines will help convince others it is in their own interests to follow the rules.  Might even give them a hint that treating the police like total mugs and fibbing about having 30 people in your wardrobes isn't very smart.  

My test kit was delivered by Amazon. That's the norm, I understand.

Seems like a good idea. You could also add block-booking some hotels as isolation locations, so people had the option to stay somewhere away from their families. Would help boost the hospitality sector and the pandemic fight - win-win

Nottingham also had a major outbreak in a prison which wasnt helped by them messing up testing and having to redo it.

The prison is run by Serco. They cant even competently run tests in probably the most controlled environment available.

Aye, quite likely the call handlers will be in a target driven environment with quotas for calls, rather than quotas for closed cases. These companies like that sort of micro-management, and there will be no incentives to prioritise "customer satisfaction" over ticking the boxes.

There was one route- one of the first as you go in on the left- whuch was Ok. the rest is appallingly bad.

It maybe in your area, does not mean that Amazon covers everyone.

The issue will be collection in areas outside citys.If you have one delivered Saturday , and you take it to your post box, it may not be collected until Monday. And as for the tracking stytem on parcels within the Royal Mail, all I can say is that it leaves something to be desired compared with others.

The logistics  infrastructure on this is quite amazing and fraught with issues.

Do you mean Lowdham Grange as that is in really pleasant countryside, not the city (I did a 13k circular walk around it yesterday). 100+ cases

Nottingham Prison had a more minor outbreak mainly affecting staff as far as I was aware.

In contrast Nottingham Uni had 1500 new cases in a rolling ten day period at the peak (Nottingham Trent are being rather secretive but rumours are their numbers were not far behind).

https://www.nottingham.ac.uk/coronavirus/covid-19-active-case-statistics.as...

'my area'? I'm in Reading. The sample had to be sent to Scotland... That suggests something of a nationwide approach.

Amazon have a global distribution logistics system. I think they can handle a few hundred thousand extra packages a day.

You haven't had a test, have you? There are specific instructions on when to take a test, to check the times of last collection at the priority boxes, to post an hour before that time, and not to take a test swab if you can't meet it.

Amazon dropped it off then you presumably put it in a Post Box and then it was whisked upto Scotland or did Amazon collect it for you.Ours go to the Post Office.Amazon may be Ok for part of it , not the whole lot.

I assume that each day samples go to different testing centres depending on capacity.

You multipy that across the whole of the UK and you can see its not simple.

And they must not lose a sample.

Yes I have taken a test.. Have you any logistics expereince?

A postal test? If so, why do you not know what the posting instructions are?

No, I haven't. But Royal Mail do have about 170 years of delivering mail. And Amazon aren't too shabby, either.

Delivered to me by Amazon. Returned by Royal Mail. With no experience of logistics, I'd venture the former is the harder part: distributing from one or more regional supply and distribution centres to random addresses. The latter is collection from a far smaller number of collection sites, priority extraction of samples from general mail (easy as they're obviously recognisable), and sending to a small number of test centres.

Yes it was pretty easy. But at the weekend the system fails as you have to wait until Monday.That is why I said earlier on they need to up the service to 24/7.So that has a knock on effect on the test results and people being in isolation etc etc.

Royal Mail is not exactly at the forefront of technology.Try tracking a barcoded parcel  or letter with them,its hopeless and is a reflection of their systems.Just imagine what its like following that sample as it goes through their system. Not good.And if a postie is late on collecting priorty ( stuck in traffic or they have a vehicle accident etc, these things happen) it all compounds an issue.

And Amazon does not  do a same day priority delivery service across the whole of the UK.Most but not all.If you ask for a kit after a daily cut off time ( there will be one for every postal code) , this means you are bumped into the next daily slot.Again adding delays.The only cut off time that you know off  is the one for the daily priority pickup at the post box.

As I said right at the start its not easy getting this right for the whole of the UK continuously every day so that delays are kept to a minimum and this improves the testing reults which then helps the contract tracing.

I find it fascinating.

Cant remember which one it was offhand but was 300+ cases.

The universities are always going to be difficult and it is amazing that anyone thought it would be a good idea to send them back. Who would have thought that sticking lots of young kids, away from home for the first time in most cases, together in cramped accomodation would be highly likely to result in large numbers of outbreaks.

You can tell how completely out of their depth these people are by the fact that only a couple of months ago they thought they had over capacity and were thinking about downsizing the number of employees.

They react like a bunch of accountants with experience of running steady-state businesses. 

The other telling thing for me was the document I posted a few weeks ago with instructions not to do tracing on cases within schools.  That suggests people wo a. know they have nothing like enough capacity and b. don't want bad numbers from schools/unis which might lead to questions about why they are still open.    Not collecting or obfuscating data which might look bad is another dickhead accountant tactic which has no place in a pandemic.

Yes; if you look at the board member most are from government departments / XAs or big businesses with a pretty steady state operating model.

I missed the document you posted - do you have a link?

Just imagine the VC at the university with one of the worst student prevalences telling his staff that there's no evidence of transmission in face-to-face teaching when it turns out that the institution is not collating data on contact tracing hits from staff that occur within face-to-face teaching settings.  Presenting an absence of evidence when you're not even looking for evidence as evidence of absence should be a sackable offence for someone claiming academic credentials.   mean it's government SOP these days (Russian interference, Hatt Mancock and transmission in schools etc.) but it'd be nice if university senior management wasn't just more of the same.

Two in this household doing the ONS/Imperial/IPSOS MORI/NHS random test, taking up to a week to get results back. I assumed / hoped that symptom triggered tests and positives from the survey  are being prioritised over negative survey tests.

There is a private testing company just set up locally, charging businesses £30+VAT for a test with a 15 minute turn around on test. I did have the opportunity to have a tour of the facility and a free test the other day, but had a few other commitments and couldn't go.

Whereabouts in Manchester is that?

Just looking back at the email I received. I assumed they were Rochdale based, but it looks like they are mobile and where being hosted by a Rochdale engineering company. Trelleborg was the host company if you want to follow it up.

Must be Lowdham. The public information there from Serco is dreadful but several news sources have reported 215 total infections (including Private Eye). Cant find any info of 300+ so any links to that much appreciated.

In Stoke the NHS system  is rolling out and testing one of these 20 /30 minute tests for free.

Last time I looked local companys ( occupational health types) were offering swab testing at about £170 a pop.

Hey, cut them some slack, what can you really expect for £12Bn.

Putting a sprinkler head on the money siphon 

jk

I thought Hogwarts was strange but Eton house names are weird. Logistics! 

jk

https://twitter.com/Parents_Utd/status/1314972467648245763

This is an example of where the system falls down on itsel over tracing. 
 

 I live inTier 2 about to go into 3. In this example I do not have access to car so it’s a home test kit. I apply for one Friday and it gets delivered Saturday( if it was delivered Friday , then there is no issue). The last collection time on Saturday is 10 am from the priority post box. There is no collection on Sunday. ( fundamental flaw). Next collection is on Monday at 5 pm ( weekdays it’s 5 pm and Saturday 10 am).Send of the kit. Result Tuesday. ( or Wednesday).

Meanwhile in that gap of 5/6 days my wider “contacts” have been wandering around spreading the virus. 

Compound this logistical flaw across the country and you have real issue. 
 

I have read they need 24 hour test results to improve the position. 24 hours from the time  you apply for the test should be the standard. 

Calls today from Tory central to give Dido Harding a “well deserved break” and hand control to the military.

https://www.bbc.co.uk/news/uk-54678809

Hopefully the military can dismiss the whole board, put a staff command in and get some people who understand the health aspect in at the top - perhaps drawn from regional public health teams and not PHE.

The military and Spitfire fly pasts are the Tory solution to everything these days.   These guys are going to be busy given that they are also supposed to be doing logistics for no-deal Brexit, staffing Nightingale hospitals and quelling anti-Brexit riots.

The reason they want Dido Harding or the military running it rather than appointing somebody competent with a public health background is that wives of Tory MPs and military officers can be trusted to STFU when the cabinet do stupid or self-enriching stuff but a public health person can't. Military have the secondary benefit is that anybody who complains can be cast as unpatriotic.

Egotistics House is the most popular, of course...

Maybe I’m wrong but I don’t think it should be run by a public health person - the large scale logistics are so critical that I think it should be led by a logistics person with a public health person playing a very close second fiddle.

None of the current board look appropriate - so the real question is perhaps who would be best to lead it?

Which is why most countries have seen an explosion in cases, they switch between Lock down and near normal behaviour. The population needs to modify how they live, then you could potentially spread it to far fewer people and your more limited wider contacts will in turn spread it to far fewer people. 

Face masks and hand washing are great, but they should still be secondary to avoiding unnecessary contact with anyone, any age etc... in the first place. 

I think the basics are being lost on ever increasingly complex tiers systems. 

Don't think it matters which is pre-eminent as long as both are prominent enough to stop the other one taking a false turn.

That's certainly arguable.  I wouldn't complain if they had someone with a logistics background running it either.   Even if they got someone ex-Army who'd been involved in the logistics for Afghanistan or Iraq.   But I don't think that's what's going on here.  They want people in uniform for the same reason they want Spitfires.  As a distraction and a shield for their own incompetence.

I don't know, not being from either public health or logistics.   But I'm sure the Tories are not the right people to choose the best person to lead it.

Seems like centralising the testing in mega labs has just created a distribution problem.

Maybe they have fundamentally the wrong structure and they'd be better off with a much larger number of labs distributed across the country.  Like a lot of logistics works with local depots for each city with vans going out from them.

People who have been doing contact tracing for decades for HIV, STDs and public health or local/national government officials. Every* country with a functioning track and trace did so by building on existing systems and existing expertise. It doesn't even need to be scalable, the functioning systems use boosted local capacity.

*there may be an exception where a country has an outsourced national system to that works but I am not aware of it.

I'd not seen this site before today (posted on the other channel) but this has mobile phone data used to assess how mobile the population is amongst other useful plots. The number of deaths look odd for the UK (not quite ONS).

https://covid19.healthdata.org/united-kingdom?view=total-deaths&tab=tre...

What matter is not phone data, but just how many get with 3 or 4m of each other. When the England or Wales exits tier 3, or Scotland tier 27(or however many more sturgeon has added by then)... then folk will just be out making up for lost time and any minor reduction 2 or 3 weeks makes will be lost within a month. 

If folk want to have any kind of social Xmas they should have radically changed their behaviour the whole year. A few weeks stopping buying kettles isn't going to change anything. 

More localism - in Germany the police visited to check my 11 year old nephew was at home when he had to quarantine for 2 weeks despite negative test. A schoolmate had Covid.

Agree. It’s a logistical excercise. The actual testing is probably the easiest part. But distributing and collecting all those kits from individuals, mobile and fixed test centres is the complex challenge.

It’s not easy. 

Mobile data shows what it shows. Movement dropped to 30% of pre-covid levels in lockdown and has streadily increased back to 70%. It also showed the mass middle class  exodus from Paris and Milan to escape lockdowns there early in the crisis (maximising efficient spreading of the pandemic).

The real problem is in two areas: for our population meeting indoors with no essential reasons or precautionary measures (not being briefly less than 1m outdoors in passing); for our government terrible messaging and examples and no working TTI system, collectively amounting to incompetence.

I'd argue that T&T is less important if folk just practice extreme social distancing, there's always an excuse why they need to do x and y, even through the alleged strict Lockdown folk had freedom to interpret exercising, shopping and dog walking how they wished. 

The rule in sweden is kids must have 2 days clear with no cold, flu or covid systems to attend school. So far since this all kicked off in Jan/ Feb ours have lost 5 days of schooling, but because every kid has a school laptop any work is sent to them, unless they are clearly too ill to do it. Most parents just add a few days on for luck, because the government pays their wages for 2 weeks if they can't work because of covid/childcare reasons, not their employer.

Even the CMO and CSO of England reckon that Tier 3 is probably not enough: we already had a lockdown that was far more than Tier 3 so it is total fantasy to think Tier 3 is as far as things can go.  Equally, we've been in a situation over summer (and some of us still are) which is less restrictive than Tier 1.   

Basically all that she's done is given a number to the 'less than Tier 1' and 'more than Tier 3' states which makes more sense than pretending they don't exist.

I do wonder Dee how ma y people choose to get a test kit as it is easier instead of going to a drive centre even though they could do. 
 

As most people test negative perhaps it might be simpler to say you get your results back quicker with a drive in.That means you come out of quarantine quicker. Better to sell that so that people understand it  

The logistics of getting all those test kits back for the whole country in a reasonably quick time are horrendous. There is not a national infrastructure to support it irrespective of Amazon or the Royal  Mail s best efforts imho.

So posters go for a drive in every time as long as you can arrange it. 
 

Let's hope Boris doesn't have 5 or 6 tiers, otherwise the Scots are going to have to understand 10 or 11 tiers. 

Yes if we all went home and locked the doors the epidemic would shrink pretty fast. But so too would lots of other really important things. The point of (working) tech is to avoid this hair shirt nastiness and minimise nett harm.

Jk

I'd agree. It's a matter of having measures that are clear concise, don't change weekly etc.. but also limit human contact as much as possible. Much of what has been given the green light in the UK in July and August wasn't really essential at all. Now the population are paying the price, sadly many with their lives. 150+ per day... All for the sake of some shopping, getting smashed in the pub, crowded beaches, eating out for free... 

But it was in check in July and August. Mostly people were working and socialising at a level we could live with medium term. Something changed after that. 

Maybe, maybe that's what it looks like from and perhaps in Sweden but from here in the middle of the UK it doesn't really look like that's what actually happened. It looks more like the season changed subtly changing how we do what we were, for most of the summer, doing safely. Government messaging, doubtless driven by donors taking a beating on their London property portfolios went a bit mad pushing people back to offices nationwide whether or not the covid had been knocked back locally and shutting down furlough. Tired of it all and credulously willing to hear the message of hope and we all let standards slip a bit. In the middle of that the schools went back and R stepped up. Then the unis, the covid exploded and the students were marooned in their petri dish while we hoped it wouldn't leak out of campuses. It did. Maybe the grotty weather helps the virus more directly too, who knows.

Jk

Having recently noticed a suden increase in 0ur local town of 'Test-and-Trace I asked about the manual records the cafe I was in were ontained.  I was told people volunteered the information voluntarily.  It wasn't a surprise when she showed me that one of their customers was Micky Mouse, with an address in Holywood.  Another the previous day, was Robin Hood and he lived in Nottingham.  Several of the names were likely to be made up - unless of course Angus Mac Coatup or Ben Dover were real names.

Either way the more rubbish that gets put into data the more rubbish comes out.  I wonder what happened to our "World class test and trace system"?

It doesnt seem to matter what name you put down, it wouldn't be requested anyway.....

https://www.theguardian.com/world/2020/oct/24/pub-and-restaurant-check-in-d...

TT&I is required if you're going to try to 'reopen the economy', especially hospitality. If you can't socially distance, infection will occur, so you need to try to find out who is infected, and stop them passing it on further.

Given the asymptomatic, but infectious period, it's a bit of a flawed concept, really... Compounded by the poor implementation.

Chuck in a significant false negative rate from testing and the way it’s run - isolating people and contacts based on a positive test - and it doesn’t make all that much sense.

Someone else posted an article on the role of superspreaders, making the case that contract tracing should go “up” to find the source of infection and quarantine them; as that way you find the minority of people apparently causing the majority of the spread.  

You’re looking at it from the wrong angle. 
 

They’re not trying to control it to the degree you would like them to. 
 

Paying people to stay at home is one thing, but it’s not practical to shut down the economy. Especially with the level of false positives. 
 

The government’s aim is to look at trends and head them off at the pass. They need to prevent the local NHS departments from being overwhelmed by admissions. 
 

It’s no longer about deaths, because we have better treatments and are having better outcomes. 
 

Everything is about balance between economy and health. 
 

The test and trace system is doing two things. It’s limiting the spread and keeping the R number reasonably low, and it’s working as an early warning system in areas where the cases are increasing. 

You haven't got a bloody clue what's going on in my head but you spent so much time telling me how I'm looking at things and what I want.

My OP wasn't about the level to which I want the government to control things so I'm not sure why that's where you're telling me that I'm wrong there.

Let me spell my point out.  I am worried that test and trace is so shambolic and is build on some bad premises (clearing people to return to social contact with a 30% or more false negative rate), that some responses and entry in to contact tracing is so slow, that contact tracing is so ineffective, and that isolation orders are so partially complied with, that it's as effective as pissing in to the wind to put out a fire half a mile away.  

It was never directly about deaths.  I don't think I've ever claimed otherwise.  The problems is healthcare overload and what that means for everyone.  Having said that, deaths are at almost 200 per day and rising exponentially with a doubling time of a couple of weeks, with perhaps 400 deaths per day locked in.  For all the better treatments, the outcomes isn't looking much better with perhaps a 0.5% IFR in the UK right now.  Many of the people going in to hospital are > 75 and aren't being put in ITU going off the figures so it's death just like before.

Well, they've screwed the pooch royally on that one.  If TTI had been fit for purpose it would have nipped the super-spreading events in the bud that kicked off the current trajectory 6-8 weeks ago.  It didn't.  Because it's so shambolic as to be ineffective I think. 

Hospitalisation numbers are up at over 1,000 people a day and growing exponentially with a doubling time of a couple of weeks.  The healthcare system can't sustain that level of increase for much longer.  It's not looking good.

Is it?  That's highly debatable going off what I can put together.

Er - perhaps it was 8 weeks ago but they've continued rising everywhere since.

Wow. Thanks for jumping on the offensive straight away. 

I have said nothing offensive to or about you.

Let me spell it out.  I pointed out that you haven’t got a bloody clue what’s in my head so it’s presumptuous and rude of you to start with two personally aligned comments, one of which is putting the wrong words in my mouth.

I was pretty offensive about test and trace but that’s because I think it’s a bloody shambles that’s totally failed.

I don’t say you were being offensive. I said you were jumping on the offensive. 

A good article out today touching on another failing points of test and trace (record low fraction of people contacted), “tracing up vs tracing down” with a super-spreader driven mechanic, and the importance and resurgence of local health teams where the central service is failing.

https://www.bbc.co.uk/news/health-54666568

Maybe hospitality just needs to change for a year or two. It's just not a priority. Yes there are jobs at risk, but if global travel is massively curtailed, these constant efforts to get hospitality back to normal is not helping foriegn tourist money flow anyway, it's just scoring an own goal infecting more of the uk's own workers, which has a knock on effect to many other industries and education. Going to a pub, restaurant, cinema etc.. just isn't essential at all. 

Well no matter how good your system is, if people are providing duff info in the first place, it makes no difference.

I'd suspect quite a bit of it has already trickled through a shit filled matryoshka of shell companies to yacht and property brokers in the Cayman islands.

jk

I have been going to both pubs and restaurants over the past few months.I find most are operating in a very safe way, some have been outstanding.As long as they are operating safely there is no reason for them to be closed imho.( I have no issues with the 10 pm closing, that is sensible)

The ones that are not you can tell easily and you just do not visit.They are the ones wrecking it for everyone else.

I have already booked from my list of " safe" restaruants for the next 4 weeks on either a Friday or Saturday night.

I also look at is as helpful to keep people employed and business going.

I agree. It's sustainable. Profit will be way down of course, prices may have to rise, but it's better than a lock down and having nothing at all. 

They should be closed down by the authorities. The option of still using them shouldn't exist. Publicans and food licence removed, hefty fine etc.. there is no point in being nice about it. 

...and this is the problem with all the restrictions in the UK - so few people are being punished for breaking them there's not much deterrent effect - people think there's a high chance they'll get away with it so they do it.

The UK was only doing alright until mid-June, the low point for live infections. After that the number of live infections slowly but surely increased. so in a nutshell, the level of restriction in mid-June should be the minimum level of restriction now, and more likely the level needs to be more restrictive due to the whole being indoors and cooler, duller weather thing.  The problem is that I believe that level of restriction will be more than the economy, government and community can stand in the long term, hence my belief that the only sensible option with no definite sign of a vaccine is to aim for elimination and tough border controls.

£40,000 fine for 4 students in Nottingham for having a house party and trying to hide they had 30 other students in the house.

Sure. You shut the schools and the parents can’t work. Everything grids to a halt. That was the big problem when it came to trying to open everything back up. And why the cases didn’t really start to rise exponentially until October after the schools had been back a few weeks. 

It is wise to do so occasionally to check that you are constructing your formulae correctly.  Only a fool would assume that they always get it right.

Any sign of expulsions yet?  There was a lot of hard line talk from universities before the start of term about this, but not so many signs of action from what I can tell.  As well as the law, many institutions (including Nottingham) are having students sign a pledge which means there is no way than can claim ignorance of the institutional rules or the law.

From what I can tell there is a veritable bonfire of good will between local communities and the leading outbreak universities.  This good will has long been tentative and will be harder to win back than to loose I fear.

They were NTU... The VC there is very strict on such things so it would not be a surprise.

I don't think formulas featured in this lady's Excel universe.

Long term living like this doesn't work but we'll have some key answers coming in over the winter, we're only really talking a few months before we have a clearer route ahead.

Looking at the decay rate during the last lockdown, effective local eradication will take longer than we'll have to wait for the vaccine trials to start returning results, most of that would be the demoralising long tail of infections in people who can't fully isolate: shoppers, key workers, hospitals etc. Doing that in anything like a timely manner would require superb leadership to instil discipline, share the burden and build a sense of collective endeavour we're sorely lacking.

The vast majority of cases would decay away in the first few weeks (as previously) so a short sharp shock to buy another period of relative normality makes sense to me but after that, with several vaccine options now on the horizon a long hard, increasingly slow slog toward 'zero covid' doesn't look like the best use of scant reserves (ok, more debt) right now. It did in Feb but we choked.

It's an option at least if none of the early vaccines deliver but at that point the benefit more obviously outweighs the cost.

jk

Boris and the cabinet keep talking of repeat 'circuit breakers' as if they are a bad thing. I think if we get this current outbreak under control a sequence of 2 week hard brakes, where necessary on the numbers, look to be the best way to keep the worst effects of the virus at bay in terms of public health and minimised economic damage, until medical intervention improves. Their lack of flexibility in options under consideration are depressing: it is possible to have a national circuit breaker excluding regions that clearly have R less than one or a circuit brake approach with intermediate local lockdowns as required.

If we had followed SAGE advice in September we wouldn't be facing up to what might be in excess of 500 deaths a day in a couple of weeks (the peak in April was only about twice that) with possible long term lockdowns with severe consequences for business.

That is all well and good but unless they have the manpower resources to actually police people so they do not travel around I reckon you are p###g in the wind.That is the fundamental flaw in having circuit breakers excluding the low rate regions. Its just not realistic no matter which way you spin it.In turn its not realistic to have a ciruit breaker in low risk areas.

In place like France and Italy you have a different police force structure( national and local) which gives a bit more flexibility to do this.

I think they should just ditch the contract tracing and concentrate resources on those who have tests and make sure they all isolate. Strip it back to its basics and build up form there. Get it right. Do the contract tracing for specific places like food processing and car homes etc and narrow the scope down.

Trying  to do an all singing and all dancing approach is clearly not working ( even in Germany they are starting to think that the virus is so widespread that it just overwhelms capacity  as I understand it)

What concerns me here is that we’re about to lock in that sort of death rate, driven largely by areas that have been under T2/T3 for a while.  Infection rates in some of those areas seemed to be levelling off last week but I’m suspicious that this was the university cases levelling of and that the community cases are rising still (masked in the regional reporting for a while by university ones) and will take over driving growth soon; these community cases being the ones that feed through to hospitalisations and deaths.  Meanwhile the T1 areas all appear to be on the same sort of exponential curve, just a few weeks behind.  Carry on as we are for another couple of weeks and by the time people start talking about more significant restrictions, we’ve locked in 8,000 hospitalisations and 800 deaths a day with the result that many hospitals are cancelling routine operations again, leading to a late lockdown.  Time spent in a late lockdown will be the same or greater as an early lock down, more people will die from covid and more people will suffer and die from missed hospitalisations.  The usual mix of idiots and agenda wielding sorts will then be back to prattling on about how lockdown killed people by stopping routine hospital treatment.

I suggested this a few weeks ago. I actually wrote to my MP about it last night, in response to his regular 'update' email.

I think the regular, finite nature of such firebreak lockdowns would encourage compliance; people will know they will end, and when. And they will know when they will start, so everyone can plan their lives and businesses around them.

It isn't impossible. If it were of sufficient economic and or political value it could be done bearing in mind it doesn't have to be perfect, just not a laughing stock. Define clear rules and make them law. Physically close some roads to restrict the problem, use civilian security contractors with limited warranted support to roadblock trunk roads. Random checks on minor roads and rail. Explain the situation clearly and why it is being done. Fund the forces' overtime, temporarily relieve them of other responsibilities. Significantly fine people breaking rules or impose even less appealing penalties like community service.

That only makes sense if the resources you pull out of contact tracing can be redeployed effectively (home working phone operatives now doing swabs, delivery driving and hazardous lab work. Nope) or sacking them frees up money to do other work. Maybe but T&T looked for most of the summer like a thinly disguised parallel welfare program so maybe you don't free up as much money as you'd think and is money really the bottleneck here?

T&T is clearly at best marginally effective when it's this poorly run and this heavily overloaded but addressing both of those things is possible and makes more sense than simply binning it. Even just 60% of contacts traced, late, is still a lot of people potentially asymptomatically spreading covid for several days each. If you want people to isolate they don't need tests, they need income support and job security, something this government and its predecessors have spent a decade stripping away from the most precariously employed. If they really want to fix this mess they're not all cretins, they know how but the political cost is an unacceptable admission of that policy's failure to create a robust and stable society.

jk

Is it just me being an idiot or have the government pulled their interactive case mapping? They changed it a little while back to exclude reporting areas with <=3 cases (if only) but now I can't find a link anywhere. It was still available Monday.

jk

I think like with the school meals mess they've almost accidentally backed themselves into a bit of a corner with this for fear temporary changes to bad but ideologically on message policy will be hard to unwind. That bumbling into corners and blind alleys is of course helped by a fair bit of pushing and prodding from lobby groups.

I expect it'll take a big round number of deaths or another devastating leak/publication to trigger the next almost inevitable U-turn.

jk

They've changed to a new one:

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

You have not got the police manpower to do it on top of everything else.There is no point in doing a pipe dream.

On contract tracing mosts people  do not need contacting to tell them they need to isolate. Alot of people just take responsibilityand suggest to their friends etc that they do so. Its what happens anyway, people tell each other and people do it without thinking.I am surprised by the number of people I know who have just isolated without having been told as they have come into close contact with somebody. Its an instinct to do because in turn you do not want to spread the virus to your family and wider close circle.

The other side is those who ignore it, they will just continue to do so unless they are confined 24/7 with security. And that is just not going to happen.

My message is be real about what can be achieved. Its like a sales target, no point in setting it if its not achieveable.

Thanks.

That's the one I found the other day using a redirect from the old dashboard (now broken). It's supposed to be linked from their dashboard front page (according to their blog) but I can't see the link anywhere and google (driven by me) can't find it either.

Annoyingly the timeline slider rolls back the map colours but not the pop-up data when you click on an area, that stays for the most recent 7 day window.

jk

As I said: Shrink the problem: relieve them of other duties (many of which reduce anyway in a lockdown). Use them sparingly on a subset of open roads to support civilian contractors doing the actual  checks. Randomly elsewhere. Use other warranted forces. Fund the overtime to make the existing resource go further.

I think that's a rather privileged take on this. You're the boss, you're probably a good boss but not everyone has one of those and how long will your patience really hold with employees expecting pay to 'isolate' for weeks at a time at their mate's suggestion and if you're not paying them do you believe they will isolate, perhaps repeatedly when they really need to. Frankly in a world where lots of people took it on board early on that "it's just a cold" or "it's all a hoax" or "it only kills sick people" we can't rely on the good judgement and responsibility of others to do their own contact tracing in the national interest. Certainly not where those 'contacted' lack the job security, welfare access and savings to weather isolation.

If you could convince me by binning the tracing service you'd genuinely be freeing up capacity to do something else more valuable I could be swayed but I don't see it.

I presume, like me, your social group is not as broad as it could be.

So what? That isn't a reason to stop tracing everyone else, it's a reason to look into why some of those people don't comply and address it (some will just be dicks of course but probably not the majority of people who don't comply fully)

I just don't see what you think giving up gains nor what you think is unrealistic, 80% of contacts called within in 3 days of a swab? 70/2? What?

jk

People are making their own call on isolating, nothing to do with me. They weigh up their own risks. Let them carry on. Alot of busineses are also making decisions for their people anyway to protect the business.

Contract tracing for small local infections clearly works and .has been used very effectively.

But this concept of mass contract tracing working is just ridiculous. It clearly is not performing. Ditch it , focus on tracing only small local infections in certain sectors, getting the testing done and paying people if necessary to stay at home who have positive tests( after having test done in 24 hours from start to finish)

We do not have an unlimited infrastructure.

This isn't meant nastily but if you are using our free time to plot graphs about covid and it's not your day job, ask yourself if you are ok. 

We are all obsessing over covid a little bit but picking apart statistics in free time for me personally would be a warning sign for my wellbeing. Unless you are being paid to do it or you're in a position to act on this information, overreading and tracking covid is going to be a losing and frustrating game that results in anxiety, paranoia, frustration, and depression

Turning off the news some days is like using a rope on a route you know you can onsight - it's a game of prevention. even if doing this stuff doesn't feel like it's an issue or could be an issue, you wouldn't gamble with your physical health so protect your mental health too by taking a break from this.

I'm fine - trust me.  The thought it appreciated however.  Taking 5 minutes to update my plots each day is a nice bit of mental downtime once the house is quiet.

I can post it on UKC and see what other people think of it...  I also act on it directly - analysing the rise in different places in February/March left me under no illusions about what was coming.  I lobbied my then-employer hard to stop face to face teaching - as did others, and they gave in and did so about 10 days before lockdown.  I used my predictions to prepare my household and my new business for what was coming and we all made it though happy.

Not for me.  It's how I understand what is going on.  Getting in to the data and analysing and plotting it is how I get a better understanding than the crap conveyed in the news.  Take yesterday's papers for example - "OMG 384 deaths! panic" - I'm not panicking because I know that half of that number is the reporting backlog stemming from the weekend.  The only thing from your list I have had is frustration, but it's not worsened by eyeballing data.     

Different people, different coping strategies.  I'd rather take the time to understand what's going on around me than to burry my head and hope it all passes.  

Fine but they shouldn't have to and it very clearly isn't sufficient to rely upon people to do so even to this extent, let alone to a greater extent.

It works when the case load is small, as the national system would. The key is the small caseload, not the localism.

No but we do have the capacity to do far more if we choose and we have a lot of ineffective infrastructure. I'm still not sure what dismantling some of it improves?

What you're advocating now looks more like making TTI work, somehow, not as it earlier seemed, dismantling it.

How does closing down (distributed) contact tracing call centres (doubling as a welfare program) speed up test results? In what way are they the bottleneck or how would they provide either relevantly skilled staff or free sufficient funds which couldn't be better sourced elsewhere to increase test availability and speed up returning results?

It seems to me the problem with the current system is (glitches notwithstanding) it only just works. As load goes up so does latency and error eroding trust, efficacy and buy-in. Fix that. The most obvious and effective way (with other benefits) is by rapidly reducing the case(over)load.

That said, most of these problems really stem from ineptitude and corruption at the very top and the lack of consequences and accountability which flow from it further down the chain of command. Perhaps we should just give up because we're stuck with that for at least four more years.

jk

This is the central failing of the UK government.  They are dealing with a phenomeon with exponential growth and inertia/latency.   To avoid overshoot a control system for a phenomenon like that needs to use the current measured data as an input to a model which predicts the future level of the variable to be controlled.  The control system takes action - in this case imposes lockdown restrictions - when the prediction exceeds the acceptable level *not* when the current measured data exceeds the desired level.   

They don't have the intelligence or the political will or the communication skills to explain this and as a result we are going to get an overshoot which in this case means potentially thousands of deaths.

Control theory is a good lens through which to look at this situation.  Add in the latency of waiting for the press to whip people up into a scared mob before government act, as was happening the first time, and you have an endless series of unstable oscillations. 

Yup.  As I put it on a previous thread I see it as an abrogation of responsibility to communicate the need for timely action when times appear good, rather taking the cowards's option of waiting until the situation is so bad people demand it.  

Clear, consistent messaging from day 1 bringing the population on board, as with NZ's "Team of 5 million".  

People used to handling and analysing data will do that to get an independent view of what is going on.

Other people will watch the news.

Other people will listen to government bullshit.

Other people will go out of their way to hear anything at a about the situation: blissful head in the sand approach (a couple of posters here have already said that is their approach).

Personally, I like to have a good idea what is going on, so I know what measures to take. I'm grateful wintertree takes those few minutes to run his data processing scripts and publish his graphs; they're better than anything I've seen on news or government communications.

Here's an example of pouring over the numbers cheering me up a bit.  A crude measurement of doubling times and CFR (with a 14-day lag) from Worldometer data.

• Spain is the odd one out as they're apparently seeing a levelling off from stricter control measures.  Everywhere else is going exponential in cases and deaths.  
• The UK is doing better than France and Germany in terms of doubling time on deaths, and considerably better than Italy.  
• The high CFR for Italy suggests that they're failing to detect a large proportion of infections.  
• The doubling time on UK cases is currently very long - much longer than that of deaths.  I think this is partly an artefact of undergraduate cases temporarily driving a growth in cases and now that's decaying, it's masking the shorter (faster) doubling time in the underlying cases in the wider population.  

So, despite the doom and gloom, the UK appears to be doing a better job of slowing the exponential phase down than neighbouring countries.  Cases are still rising, but it could be much worse.  My latest doubling time plot is below.  The right side of the is very twitchy to retrospective case and death updates and to changing apparent trends from the "weekend effect" of sampling and reporting lulls over weekends. 

What I like about this is that there's a doubling time increase in cases - with a big blip form the universities perhaps, and then later on there's a similar increasing trend to the doubling time in hospital admissions and now it looks (looks - beware retrospective data updates) like the doubling time for deaths is increasing as well.  The reason I like it is that it feels pretty real - often a slackening in just one curve is an artefact of jitter in the data.  So maybe T2/T3 is having a big effect; although the areas currently under T1 are rising to take over driving cases.

Still, it paints a picture that isn't rosy, but is several notches below continental Europe in pucker factor.

I think it would be instructive to see doubling time calculated for the UK areas excluding greater Manchester, Liverpool region, NE and the city centres that have been hit hard by students in local case terms. Ie, those areas that infection is primarily community-spread but haven't had any significant restrictions until very recently. It would give an indication of whether and how quickly we might expect those areas to catch up the worst hit areas. I appreciate that's a lot of work!

If you can point me to easily ingestible datasets I can feed them through the script...

I'm convinced that the university cases should have been covered under separate reporting units.

I'm not sure what counts as easily digestible. I assume you are aware of the MSOA download link - that's all I have unfortunately. It shows cases each week in every MSOA in the country since the start of the pandemic (by specimen date.)

In fact since the dashboard was recently reconfigured I can't even find the link to that. The map still says all the data is public domain and available through the website somewhere.

Not very helpful I know.

I'd seen the maps but not looked for a download before.  After some interminable trawling of pages I found this - https://coronavirus.data.gov.uk/downloads/msoa_data/MSOAs_latest.csv

I've exceeded my downtime plotting limits for today but I'll look at this over the next couple of days.  What would be useful is if anyone wants to post a list of areas from the file that are a university area and the surrounding ones; then I can make plots for a few of those...  Although they're probably going to be a bit crappy, as we're into smaller number statistics and I can't filter the data that egregiously when looking for time lags between neighbouring areas.

I don't know that information to MSOA level. I could have a go for Liverpool but I'd definitely miss a few areas. At local authority level, you could exclude the vast majority of the student effect by taking out the city centre local authorities, which outside of London are all obvious. I think this would be much easier than trying to do it for MSOAs.

Annoyingly that file doesn't name the local authority in each entry, but the entries are ordered by local authority, moving around the country in some geographical sense. 

New react study in the news this morning suggests R is much higher in areas with low rates, averaging 2. Figure of 2.86 for London!

I see Merkel is perhaps coming round my view that mass contract tracing does not really work as there is not the state capacity for it to work. Well that is my interpretation of it.

Get the testing to 24 hours from start( when you first reguister for a test) to finish as the No 1 goal.Everything flows from that.It will be tough and in some areas almost impossible to do. But in city areas etc, the logistics by now should be there. Work at doing it 24/7 for deliveries/collections.

Yes, it's not looking good is it?

Mass testing and tracing in Germany worked well throughout summer, when most cases could indeed be traced back and contacts isolated. The capacity limit for this was about 50 cases per week and 100.000 people.

Since cases broke that barrier, the federal government has just managed to convinced all state governments to agree to uniform rules* for a four week "lockdown light" circuit breaker, in order to be able to open up again using a test and track regime, and thus allow travelling and visits over Christmas. Schools, work, and shopping will remain open, but travelling, going to bars and restaurants, and meeting others in private will be strongly reduced.

Too bad for my training (judo was out of the question anyway, but no we cannot even meet for outdoor circuit training in the local athletics stadium) and for my planned birding weekend at the North Sea coast, but I agree with the measures.

CB

* now there is something I never expected to see, must be a very catholic miracle!

It did, but the comments from Merkel about capacity suggested that this is history now,.......

I had the same problem but somehow found it

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

Ed: just seen it was posted up thread. 

It suggests exactly what I was saying yesterday, that the systems don't work well when overloaded. There are two ways to fix that, increase capacity or decrease the load. Decreasing load comes with additional benefits (fewer dead people for starters). The load we impose on the system is a choice, we know how to decrease it (see Germany, France et al accepting that reality a few weeks ahead of the UK again).

A significant part of having a formal contact tracing system is the anonymisation it provides, it deals with the shame people may feel over unwittingly having put other people and businesses at risk. In the near future we're going to have to go back to living in social circles where some of us catching this killed others or their business. Your belief that people will diligently do their own contact tracing simply isn't borne out, if that were the case the state's effort would be additional but even then we see the sum of those two efforts, state and personal is still inadequate at this case density, removing the state's effort only makes that worse.

jk

Not testing, that is still fine and has sufficiently quick turnaround times, but tracing. With too many cases you simply cannot tell where an infection came from, and thus which persons to quarantine. Hence a four week circuit breaker: Let every single cluster fizzle out (realistically as many as you can catch), as you cannot tell who is at risk of infection, and prevent transmission out of these clusters by drastically reducing contacts.

If the UK government had any sense they should have done thae same one or two months ago, as soon as cases started to rise, rather than the half arsed tier 1 2 3 business. Note that the relevant parameters (incidence, hospital and ICU admissuions, and deaths) are  all still much lower in Germany, and still the government acts more decisively. As much as I hate Merkel for other reasons, credit where credit is due.

CB

In the UK I'd estimate that contact tracing is currently reaching at most 10% of people who are getting infected, and that this is getting rapidly worse.  At this point it just can't achieve much with reported R values of 2 and above it can at best make things 1/5th better and in reality probably a lot less - which maybe shaves a few days of the time until we hit healthcare overload.  

My preference would be to use a lockdown to reset cases to a level of a few thousand a day and then to stop the central contact tracing and to pour all resources into ground pounding local public health teams - a lot of outbreaks stem from a s social circle it seems and so you need an experienced human getting in to that social circle and learning it.  What we appear to have is a blind process driven system that treats each person as an individual and every contact as equal.  The problem is funding local public health teams over a central service flies in the face of the dogmatic decisions behind PHE and puts cash into a lot of labour councils rather than into companies that by sheer coincidence have a lot of blue party connections.

My understanding in the UK is that the system tries to quarantine downstream contacts rather than upstream sources of infection.  Is that a good match to a pandemic apparently driven by super-spreading events?  Answers on a postcard...

I said some people wil do their own contract tracing and some people will then isolate where they can.Let that carry on. Its not a policy, people can figure it out themselves.

My view remains unchanged focus on the testing and make sure those people get fast results and isolate. I gave earlier in this thread an example of the logistical delays in getting testing right and then this working all the way through on contract tracing.. Surely you have done work in your engineering environment on lean processes ( mapping and the like )and getting down to the root causes.Its the same thing in logistics.

Mass contract tracing just is not working....becuase we are not getting the test results right.

It needs more improvements.. I appreciate you know the science far more than me, but logistics is the issue and here is a repeat of the example I gave earlier in the thread.:

In this example I do not have access to car so it’s a home test kit. Its based on a North west town.I apply for one Friday and it gets delivered Saturday( if it was delivered Friday , then there is no issue). The last collection time on Saturday is 10 am from the priority post box. There is no collection on Sunday. ( fundamental flaw). Next collection is on Monday at 5 pm ( weekdays it’s 5 pm and Saturday 10 am).Send of the kit. Result Tuesday. ( or Wednesday).

Meanwhile in that gap of 5/6 days my wider “contacts” have been wandering around spreading the virus.
Compound this logistical flaw across the country and you could do with some improvements.

Should have been clearer, of course you primarily look at downstream contacts: Who might a patient have infected?

However, you anyway cannot simply quarantine every single person a patient may have met in the five days or so before their positive test. That needs to be restricted to their immediate, close contacts at work or at home, which are normally readily identified (especially when pub or concert venues are closed, which is the main rationale for such measures, reducing anonymous contacts).

At the same time you also have to check whether infections are clustered, say to a specific school, factory, or church congregation, to focus testing on these populations.

If there are too many positive tests that upstream analysis will become impossible, as people could have caught the virus anywhere. Note that we are still talking about a small minority of the entire population, so tracking both ways SHOULD be possible.

CB

I am in awe of the approach in NZ where they rapidly sequence every detected case to get hard data on where it came from.  I gather that has been very useful to them, but neither taking the data nor understanding it are anywhere near viable for the UK right now; I'm not sure it could have been even at the best time in the summer.  The Oxford NanoPore devices could help roll out more wide spread sequencing perhaps.  Digital DNA readers, I want one.

The argument I've seen put forwards for tracing up in preference to down is that perhaps 10% of the infected are driving 80% of the spread, and there will be many links "up" to such a person, but only ever one link "down", so the probability of finding them is much higher when tracing up.   However, you're tracing against the flow of time so it's critical to get the latency right down, and the kind of detective work that finds them can't be distributed across low paid, barely trained call centre operators working from home.  

The daughter of my colleague just tested positive last week. Symptoms on Monday, test on Tuesday afternoon at the local test centre, result by email and phone early Wed afternoon. This made her a primary contact, resulting in obligatory quarantine and testing, and my entire lab secondary contacts, not having to fully isolate or stop working, but asked to be extra careful until her test result was in that would, if positive, made us primary contacts

She was tested Thursday, and had her negative result by Friday evening. As a precaution I had, however, taken my work computer and all relevant file home with me on Friday afternoon so I could have isolated and worked from home this week.

Obviously, such a system cannot work if the turnaround time for a test approaches one week.

Yes, the reason that R just gives you an average, and the distribution of the true numbers infected by each patient is highly skewed (there are "superspreaders", and many cases that fail to infect anyone even in their closest, private environment) is the reason why you want to test upstream.

This is the way superspreading events e.g. in abattoirs were intitially flagged (and subsequently confirmed by targeted mass testing).

I agree that this is a professional job for public health administrations that should not be sourced out to unemployed roadies or waiters (as much as I understand why someone working in hospitality and event managment needs any kind of job!).

CB

But you said it in conjunction with the idea contact tracing by the state should be wound down because we don't have sufficient resources. Being left to figure stuff out for ourselves where failure to do so heavily negatively impacts others represents a clear abdication of government's primary responsibility, to protect citizens.

That's fine but you don't explain the link between abandoning contact tracing and improving testing availability and latency. How does one facilitate the other?

I don't disagree with that* though it's clearly not the only problem with the contact tracing program. How does abandoning tracing improve testing?

*assuming by not right you mean tests are too hard to access and results too laggy.

https://www.theguardian.com/commentisfree/2020/oct/28/england-coronavirus-c...

Almost unbelievable shit like this, Serco quietly 'promoting' £6.45 minimum wage teens into band 6 clinical roles really can't be helping. I guess recruiting and paying the clinical staff they advertised for would hit the bottom line.

jk

I don't know what the granularity of the data is like but a possible university hot-spot to look at would be Univ of Sussex which is in Falmer just outside Brighton.

That sounds like a pretty effective system. Decent turnaround and an ambition to find secondary contacts. The UK system has neither - your colleague would not have been given a test unless she herself developed symptoms, and you would have been none the wiser to any of it. 

I’ve been busy on other projects so haven’t done any plots yet; the data is all quite noisy as small number statistics and isn’t amenable to a doubling time analysis with that and with the weekly reporting.

I did find this treasure trove of relevant plots though:

https://mobile.twitter.com/carlbaker/status/1321418588020985856

I still can't believe some of the comments on there "Many younger groups affected which means it probably not covid."

I would say "don't read the comments" but one of them noted the source - a new download on the dashboard that gives 5-year age bins (an improvement over the 10-year ones, but still not aligned <18 and >18)  and county by specimen date.  Fantastic.  

Plots from the new demographic breakdown for several UTLAs with big university outbreaks.

I've plotted the data as heat maps and used a funny colour map - the colour changes more rapidly for smaller values; this helps to see the low level effects when a simple plot is dominated by the high values - it's a good way of visualising data with a wide dynamic range.

There is a rising diagonal motif to the data for Co. Durham and Nottinghamshire and less so in East Sussex that could be interpreted as "bleed through" of cases from the university to similar ages, gradually working its way up into older groups over time.  It could be coincidence, but it's there in several datasets.  I'll do some more tonight.  

The Manchester plots look suspicious; 25-44 and 45-64 look suspiciously similar, as do the 65-80 and 80+. So similar I would guess there is some reporting error.

I’m going to guess the twitter author is using a manual pipeline to make the plots and forgot to change a title.  This is very easy to do and is why I use a script with a list of inputs to loop over - it’s really hard to do the same thing over and over manually and not make mistakes...

If only they would publish raw data: age (years), date of test result. Then sane analysis could be performed without pre-binning hiding the information.

They're similar, but not identical. So I suspect data sources overlapping, sharing a core set of data, with the small variations caused by the non-overlapped data.

Of course, it may be genuinely close tracking between those age groups...

The Daniel Howdon thread within that thread is very good for seeing the effect of universities.

Grimly interesting that you can see what are presumably individual care home outbreaks starting and ending along the top of each of those plots where the age bands below 80 seemed pretty well insulated until recently.

I wonder if the tailing off of cases in the 20 somethings is real or have they just stopped getting tested because they're all quarantined together anyway in infected blocks. Sort of doesn't matter so long as they are quarantined and for long enough. The worry would be one case locks a block down then everyone goes out to celebrate the end of quarantine a fortnight later, half a dozen of them still infectious.

jk

The latest stats from Warwick Uni shows the other way round. halls really low, out in the community high.

What's really low, positive tests or infections? That was the core of my question.

The opposite is clearly true here with ludicrous case density in the reporting region containing two campuses (old fashioned corridors and big 'houses' of students, 10-20 per unit) and just handful of rural villages. Perhaps the difference is a function of where the test sites are, York has one on campus, Warwick uni from memory is in the middle of nowhere, does it have a test site?

jk

I suspect the amount of transfer from students to rest of the population will depend to some extent on how much of the student accommodation is campus based and whether that campus is within an urban area or outside it.

So I would expect "transfer" to be high for Manchester Uni(s) but lower for Unis like Keele, Sussex & Warwick.

Please remember Notts stats are distorted by the significant Lowdham Grange young adult offender unit outbreak (200+ cases). Pillar 1 compared to Pillar 2 look diferent in the city and the county. I think the 'bleed' from student infection is mainly in the city.

http://documents.nottinghamcity.gov.uk/download/8807

https://www.nottinghamshire.gov.uk/media/2896449/weeklycovid19surveillancer...

Here're the plots for the relevant counties for those places.  I want to put one of these plots together with the MOSA vs time plots to give more context.  Keep your eyes posted.  Manchester has a bigger spike event, hard to infer much about transfer outwards without the geographic data too.

Still not there - I've loaded and visualised the MSOA data for areas around the universities in Nottingham.  It's noisy, coarse grained in time and isn't going to yield much in the way of meaningful doubling times as the spike is not well sampled in time, so I'll probably park the effort here.  Once I remember how shapefiles work I'll try and automatically make plots for the MSOAs surrounding universities with big outbreaks, and do the same for some negative controls.