UKH

Or is there something in this?  A great deal seems highly plausible to me.

https://lockdownsceptics.org/radical-uncertainty-and-government-innumeracy/...

Actually, it makes a lot of sense. I will be genuinely interested to her some refutation. 

I can't speak for the maths. However, as far as we know, the virus has not changed. The excess deaths figure in the first peak was massive and there are a large number of people currently living with Long Covid. Hospital admissions for Covid are rising, and we know people are doing more things which will enable Covid spread. So, maybe the numbers are right or maybe they aren't, but I see nothing in the surrounding information to suggest that it is going away and that we can go about life as we please. 

Anything that sounds too good to be true, usually is!

Reports of a large drop in UK testing rate due to logistics problems at the same time detected cases have more than doubled.  If that’s a false positive issue I’ll eat my hat.

Is it all bollocks? Dunno. What I do know is that local respiratory wards are filling up again and we're back to queues in corridors.

Aren't false positives going to be (crudely) balanced by false negatives? 

Anyway, sounds like cobblers to me given all the other evidence.

Except... By several measures, deaths from CV are tiny. Hospital admissions are tiny. WTF is going on?

People > 65 are barely getting infected.  People < 25 are getting infected quite a lot.  This can be seen in the breakdowns in the PHE reports for example, especially the hotspots ones.  There’s > a 50x difference between the two age ranges in fatality rate.

The current fatality rate is actually quite high; 0.7.% by reported detections and infections, perhaps 0.3% in reality given incomplete testing.  That’s just seems low compared to March/April because we were catching only about 2.5% of infections with testing - and only the most gravely Ill at that - whereas now we’re getting perhaps 50%.

Attached below is my latest update of various estimates for our fatality rate (left) and the first age breakdown from the latest LTLA "hotspots" report [1].

In short, almost no old people are getting Covid at the moment, and those that do seem to be making up a lot of the current fatalities.  As most people getting it are young and not dying, the overall fatality rate is quote low.  

0.3% fatality rate may sound low, but it's nothing to shout about if we project if forwards to a hypothetical winter with no covid risk control measures with perhaps 10% of the population have had it and may be immune; say 70% have get hit before herd immunity takes over; 67.8 m people x (70%-10%) x 0.3% = 122,000 people who will die.   In reality it would likely be higher as if the virus was that widespread, shielding of the more vulnerable would almost certainly break down and viral loads would rise.

When starting down the data-overload rabbit hole, don't forget that there's a lag between a detected infection and a death, perhaps 3 weeks.  There are various suggestions that the reduced viral loads associated with the current control measures and low general prevalence are also reducing the severity of cases.

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

Well looking at France or Spain, I'd say the calm before the (second) storm.

Exactly.....over 1100 in Spanish ICUs at the moment and 20% of all hospital beds Madrid are taken up by Covid-19 patients. 

The stuff about probabilities and the impact of false positives where incidence of the disease is low looks to be largely correct although the source of some of the figures is unclear. The conclusions they draw from it I'm less sure about and some of the comments later in the piece suggest a definite agenda.

The FNR is way higher than the FPR. Perhaps as high as 30% although it varies hugely.

It's not all bollocks, but as Tennyson said "a lie which is part a truth is a harder matter to fight.”

Given the various rule-breaking parties and raves, this may be more appropriate:

youtube.com/watch?v=OMIyNWav3-I&

This is interesting. Are Sweden on the right track (not in a libertarian nutty sense, just pragmatically)

https://amp.ft.com/content/5cc92d45-fbdb-43b7-9c66-26501693a371

So forget about FPR affecting numbers? Actually they are worse?

No particular reason to suppose false positives and negatives would be the same amount.  Also they don't cancel each other out as such.  

How so? One FP and one FN would result in the same numbers wouldn't it?

No. Nothing like. This is something that is very commonly misunderstood even, worryingly, by doctors. Suppose 1% of the population have a disease and the test gives the wrong result 10% of the time. Suppose 1000 people are tested. 10 of them will actually have the disease. Of these 9 will test positive and one negative. 990 of them will not have the disease. Of these 99 will test positive and 891 will test negative. So there are 99 false positives but only one false negative! And of the 108 positive results, only 9 will actually have the disease - the test will be overestimating the prevalence of the disease 12 times! Even if the test is 99% accurate about half the positives will be false (but there will be hardly any false negatives).

Of course, one would hope that the people doing the statistics are well aware of all this and know how reliable the tests are and are adjusting things accordingly.

In that sense, yes, if the numbers happen to be the same, but if people are being given false negative results then the spread will increase as they don't isolate.  

So there is clearly some truth in there, but I'm not sure of the implications either for policy or in terms of what we might expect this winter. Other than, i) project moonshot and any truly randomized population testing seems fundamentally flawed unless the FP rate is both known and taken into account and ii) perhaps things are not quite as bad as we might think. I certainly hope someone in a position of authority and influence is thinking this through properly (but have my doubts).

Time will tell.  So far, they've had 116 times the number of per-capita deaths as New Zealand, and both have largely kept the wheels on their economies.  Killing over a hundred times a larger fraction of the population isn't a great start for success; if we don't get decent vaccines within a year and if clinical care and risk control measures don't continue to improve in effectiveness, and herd immunity from mild infections turns out to be persistent, then Sweden may well be ahead of everyone else in the cull > immunity race.  A towering pile of uncertainty; one thing I'm certain of is that if the public in the UK were asked to followed the Swedish model of individual responsibility, things would have not gone like they have in Sweden.

I see what you are saying, yes. 

NZ is an outlier too, I think. I think the long term focus and wide view of public health of Tegnell is worth serious attention.

I agree that we shouldn't assume that FP and FN rates "balance" each other, but I disagree with the reasoning you give...  

That's an invalid assumption however - that the FP and FN rates are symmetric.  They're not.

• False negatives in qPCR are a result of the sample being to weak and the sensitivity of the detection kit, and so are linked to the positivity of the samples and their distribution of strengths.   The FN rate is somewhere between 30% and 100% depending on when the sample was taken with respect to when infection occurred, and on the strength of the infection.
• False positives in qPCR are a result of cross-contamination and so are linked to the quality of and adherence to the protocols of the lab, to the sensitivity of the kit, and to the positivity of other samples (from which the contamination arises), and to the infection status of employees.  

If we assumed no staff were covid positive, then if no positive cases came in to the lab, there would be no false negatives (can't be less than 0) and no false positives (cross-contamination not possible).  The FP and FN rates both depend on the incoming positivity but by different factors which is why under most circumstances that can't be considered to balance/cancel.  

Indeed.  A qPCR lab will systematically run control negative samples to monitor the effectiveness of QC with regards cross contamination.  I'd be surprised if a 2% FP rate was routinely tolerated.

As opposing outliers go, I prefer the one that doesn't rush to kill 115 times as many people as the other.  That's not a reversible choice.  I also don't think it's an approach that - if it turns out to work - could work in the UK.  Sweden effectively backed a relatively un-evidenced hunch and one for which there are counterpoints in terms of what's know about (lack of) persistent immunity to some other coronaviruses.   Arguably the main difference to the UK is in personal responsibility taking place of government mandated closures.

It's too soon I think by 6 months to make a good quality judgement on the Swedish approach; winter is coming; if they get through winter without another wave of deaths as bad as their first that would be some strong support.  I don't see how it could translate to the UK however.

The wider view is critical - but we were forced in to lockdown in the UK because we screwed the pooch on that to the point the health service had become practically a one-disease service; if we'd spent the previous two months messaging and preparing people to take effective measures to reduce the spread of the virus that might not have been the case and we could have been more like Sweden; but we'd already set ourselves on a path to a panic driven lockdown and reconfiguration.  

Come on!

I agree, a year's time is when we will know. Also, I think it *could* have worked in the UK but can't now given the government s behaviour

Sorry, I should have said I had made that assumption. I was giving a simple hypothetical example to show how the false positives and negatives certainly don't necessarily anything like cancel out and how false positives can counter-intuitively swamp true positives even when the test is fairly accurate.

If you gave me a choice of plane tickets today, I'd take NZ without a moment's hesitation. 

From the moment Boris talked about shaking hands, all hope was lost I reckon.

Fair enough.  The key difference to me is that the false positive rate can be and is routinely tested, where-as determining the false negative rate involves all sorts of navel gazing, longitudinal studies and statistical voodoo.

Sure but no one set out to kill people. A key point about NZ is it being a sparsely populated island at the end of the world.

One would certainly hope so and that the ONS or whatever are taking it into account when trying to work out what is going on!

Sweden knew what was going to happen with their approach.  Their authorities made the decision early on that they were going to take a course of action that would kill a lot of people, and that they were willing to accept it.  NZ knew what they hoped to achieve with their approach.  Sweden rapidly took an approach that they knew with good certainty would result in a lot of deaths - per-capita it's almost as bad as the UK.  

With most people living in cities where they have similar social, work and educational networks to many other places, and where they have similar numbers of close contacts, and with many flights to and from many other parts of the world.   Critically, they slammed the door on those flights early on and quarantined people.  

A compelling look at the data. Not project fear still being pushed by the media. Worth the watch. 

youtube.com/watch?v=8UvFhIFzaac&

Around 27 mins in looks like the same agenda wielding interpretation over PCR results as seen elsewhere, with slights of hand including:

• Comparing detected case and death numbers between March/April and now without noting that testing was running 50x less back then
• Ignoring the lag between infection/detection and death when looking at the early stages of an exponential rise (*) - any fatality rate calculation tends to look all roses under such inappropriate analysis
• Ignoring the data showing that infection is now confined largely to young people and ignoring that data showing the > 50x fatality rate difference from young to old
• Ignoring information on hospital admissions in other near by countries “ahead” of us in this process (*)

Having painted this as a mystery, some rather questionable explanations are offered to explain away what we now see.  Items marked with * are ones that were argued back in early March and that I rallied against them.

Make of that what you will.

If Trump did this stuff there would be a fight on. Go anti-pandemic guys. Anders FTW!!

The contrary view appears to be - SARS-COV2 is a virus like any other and follows the same pandemic trajectory, with some interesting arguments, unfortunately based on historic data and with few testable predictions.

My prediction would be that not many will be convinced, but time and the data will tell.

If hospitalisations and deaths fail to go up to match the case rate, they win. I think we should allow some kind of adjustment for age-related exposure, up to a certain point at which it becomes community-spread.

What effect will schools have - none in Sweden it appears.

Eek. Seems they're not even discussing this in SAGE.

It's something of a coincidence, but false positives and false negatives may actually balance out: assuming

- false positive rate = 0.5%

- false negatives rate = 20%

Then FNs = FPs when:

- proportion of population infected = 0.5/(20+0.5) = 2.4%

Aren't we approaching that now, at least in some regions?

What the author of that piece is mainly describing is base rate neglect - the failure to take into account the effect of a low base rate of a disease in a population making false positives more common than true positives. However, his conclusions are built on the assumption that the base rate is very low which is something of a circular argument as that's the point he's trying to prove! Basically his argument is; the base rate is low, so most of the positives we're seeing are false positives, so the base rate must be low. He also doesn't present much real evidence for a significantly lower base rate and surely it's going to vary widely across the country?

As others have said, one would hope (or even assume) those advising the Govt on this are clued up on all that.

Happy to be corrected on any of that. I know just enough about this stuff to get out of my depth!  

I only read the "Are you positive you are positive" section. In that, the bits that jump out are:

1. No mention of false negatives? There was a high false negative rate for PCR tests at the start of all this, due many reason, from the sensitivity of the test to the difficulty of getting people to self administer properly. This false negative rate seemed pretty large (30%?) from anecdotaly reports at the time so this would make the false positive numbers pretty insignificant. I've no idea if the PCR test has been improved, but I imagine the general public's skills in self testing have stayed the same. 
2. Pillar 1 vs Pillar 2 positives comparison. The author doesn't seem to know/understand who is tested by hospitals. Pillar 1 is testing loads of asymptomatic people, anyone going in for any procedure gets swabbed to show they HAVEN'T got it, prior to going to hospital for the procedure. If they got Covid at the hospital, it would probably get picked up by a Pillar 2 test two weeks later once they are back home.

I gave up after that.....

So I suppose it is saying that the data is not inconsistent with a low base rate.

Well, don't we have an interesting scenario here - we have a definite surge in identified cases, notwithstanding any anomalies or concerns about the accuracy of that figure - and that should translate into unambiguous hard data by the end of this week - hospital admissions and deaths. 

In fact, shouldn't we be seeing a significant uptick of those now?

Yes, but not in to very many admissions - with deaths probably another 1-2 weeks down the line.  The reason I don't expect it to be very many is because - for now - there is hard evidence to support that it is mostly young people getting infected, and there is evidence around them being less affected by infection.

Admissions appear to be rising here - https://coronavirus.data.gov.uk/healthcare

Maybe in a week or two's time - the case rate didn't really start to accelerate until the end of August, and then there's a reporting lag, so a month or so after that we should see something, or not.

Other posters have seen an increase in hospital admissions, however.

That would depend what the two rates and the true prevalence are.

I fear this is another case of the government's instinctive secrecy about what it doesn't know and or doesn't want to admit presenting those with a small state agenda an opportunity to undermine its bumbling but essential work. It's a well known problem, random screening for rare disease produces very very poor results. It's worth remembering pillar two results aren't random screening though, people presenting for tests report relevant symptoms. Yeah, there probably are quite a few 'false' positives in the current numbers but even what we mean by false positive deserves some scrutiny, broadly I'd say these fall into three categories:

1. Test kits that would test positive having never even been near a test subject and swabs with no CV19 that test positive because of contamination or misidentification

2. Tests that find the CV19 virus but where the subject subsequently fights the infection and doesn't get ill

3. Tests that find remnants of an infection in a person that is no longer a threat to society but who will as a result be asked to isolate

The anti-lockdown argument hinges on 1 being the dominant cause of reported positive tests. If this is the case then a lot of very competent qualified people who well understand this risk simply aren't doing their jobs and are very effectively being prevented from blowing the whistle. It's possible but incredible.

2 isn't a false positive.

3 is unfortunate but not a serious problem at this juncture though it probably will shoot down our 'moonshot'.

jk

Not if our society is effectively segregated and it's currently predominantly spreading in a sub-group at low risk of serious complications. If.

Sounds like we are in the hot-spot areas.

jk

'I fear this is another case of the government's instinctive secrecy about what it doesn't know and or doesn't want to admit presenting those with a small state agenda an opportunity to undermine its bumbling but essential work.' FWIW I don't have a small state agenda, and I'm a massive fan of the concept of the NHS.

I do however have a degree of scepticism about the way that the bureaucracy of the NHS works, about the scientific training of doctors, and about how clearly the government is evaluating the evidence as it emerges and refining its plans as time progresses. 

History is littered with the medical establishment getting things wrong, or raising concerns/creating panics, and intervening unnecessarily and often dangerously. I don't know wtf is happening  but it seems bizarre to me that we got to 50,000 deaths in practically a matter of weeks and then the number fell off a cliff and thankfully has remained there. Do I believe that was because people stayed at home? No, because if it was the number would have shot up again when we all started going out again which, let's not forget, was back in June. 

Current policies don't seem to be reflecting the relatively few hard facts we have to hand.

I didn't mean you, I meant the people authoring and publishing these contrarian pieces aimed at undermining the government's response, furthering a 'do less' agenda.

We're funny things aren't we, it doesn't seem at all odd to me.

For starters we didn't all start going back out. Some of us did but we're mostly behaving quite differently. Many are still living very small lives cocooned for now away from infection risk.

I suspect the massive first wave of deaths was the result of poor infection control (the PPE disaster) and the widespread institutional mixing of vulnerable people which occurred as wards were cleared en-masse to make space for sadly the very same people subsequently returned to die with covid a few weeks later. That isn't happening at the moment, that lesson has hopefully been learned even if it's not widely acknowledged (understandably given the shame some must feel). At the moment it's predominantly  young people presenting for tests and getting positive results, they're generally at very low risk of death.

No? As you can't have missed I'm no fan of our government but I think at the moment their covid priorities are broadly right, getting education back up and running and keeping as many people as possible in jobs for as long as possible. That strategy makes sense where we expect medical tech to deliver for us in the short to medium term, if there are setbacks we may need to reconsider effective elimination. The 'moonshot' is bollocks but I don't believe it's ever intended to amount to anything more than distraction tactics and another siphon dipped into public funds. FWIW I think we'd have been in a much better position if elimination was option 1 in February but from where we are with schools open, unis invested in re-opening, vaccines showing promise I think we ride this out for now.

jk

If not elimination, at least not letting community spread get so out of control it forced a panic-driven lockdown.  Like you I think a lot of harm was caused by the rapid discharge from hospitals back in to care homes; I think that goes a long way to explaining how so many died so quickly.  As you say there’s been very little acknowledgement or discussion of this.

I largely agree - but I'd be doing a bit more, now, in anticipation of what's coming.

Fair enough and I agree re. Christmas we will have to drop the caseload in advance to minimise deaths. How the fresher's covid impact towns and mortality is to be seen but if we agree it probably will rip through halls of residence this term then reducing the prevalence now seems moot, unis will get it either way and will become (at best) self contained hotspots. Containing outbreaks in institutions will be the trick, I suspect shutting bars may be required to limit cross contamination between uni hotspots and towns but it's drastic action, it will bankrupt many of them still barely breaking even after the spring-summer closure. Most clubs are still closed.

jk

There are those concerns about universities; my bigger one is that if they slow the burn rate enough that it sizzles through the population all term, it’ll still be going by the end of term and many asymptomatic carriers will return to their parents who are generally aged mid forties to mid sixties.  Not Good.

Pubs and bars seem like the main mixing points assuming proper care is taken with shopping; closely followed by Friday night A&E visits.  I can see many universities opening their bars given how risk averse they are once there’s a trail to show legal liability, and any attempt to partition town bars into “student” and “local” venues is going to cause massive town/gown problems.  Not as big as the ones that will happen when all the pubs get shut down after a student driven outbreak mind...

Testing and test/trace needs a marked improvement in throughout and latency, rapidly.

That's the Christmas problem I see too, that and the mixing of grandparents in those extended family groups for the special occasion where they may until that point have been maintaining some degree of separation or at least limiting contact.

The bars here in York are semi-segregated already, we may just see that degree of separation increase naturally. Hard to say given how weird everything is at the moment given the sector has already been shaken up a lot by the council turning over so much pavement space for outdoor seating (a nice change to the city IMO)

I guess you mean can't? Not sure if that's been decided yet.

I can see unis taking this upon themselves to protect their reputations and investment in re-opening, hopefully in close collaboration with local public health teams.

jk

I haven’t seen how much of our local city has been turned over to outdoor seating as I’ve not been in to the centre since February...  

You are correct.

Indeed, but the universities can’t fix the current issues with throughout and latency in the central labs.

Yes. And in fact he might be right about the low base rate and FPR leading to an incorrect impression of the number of infections. My worry about the piece was more that this potentially important or accurate observation was being co-opted to serve the 'it's all over/was no big deal anyway' agenda.

Is there any observation that can’t be selectively presented by those pushing this agenda? It’s been a persistent effort since at least March.

Probably not! 

Interesting they stop their analysis late August.

https://www.worldometers.info/coronavirus/country/uk/#:~:text=United%20King...

Kind of makes their argument totally redundant when we include the first two weeks of September.

True but if I understood right one of my local unis is putting its own sampling teams together, it wasn't entirely clear whether they planned to have the analysis performed privately or submit samples into the national T&T system. It may simply be intended as a shortcut on the postal test system (drive through testing being inaccessible to most students) but the issue seems to have been considered.

jk