[Stabbsy]

They’ve covered this on More or Less in the past - 23rd Sept last year. Just relistened to it, it’s the second segment about 8 minutes in. Conclusion seemed to be that false positive rate is nearer 0.05% and that worrying about it probably a distraction.

[Jerry Morefat]

They’ve covered this on More or Less in the past - 23rd Sept last year. Just relistened to it, it’s the second segment about 8 minutes in. Conclusion seemed to be that false positive rate is nearer 0.05% and that worrying about it probably a distraction.

Sure, but even with zero prevalence, a 0.05% FPR gives 50 positives per 100k. So even with no prevalence of Covid in the population you're still hitting the 50 positive tests per 100k criteria. With increased prevalence the proportion of positive tests is only going to increase. I thought this was Fultonious's point.

[abarro81]

Went and had a quick look...
https://www.ons.gov.uk/aboutus/transparencyandgovernance/freedomofinformationfoi/falsepositivityrateofthecovid19pcrtest

Key points for ONS data:
2% is way too high, false positive is prob much lower
They don't adjust for false positive or negative
Even at low prevalence they seem to think that false neg outweighs false pos

I assume the above means the PHE and other main data is unadjusted too given the ONS stuff is.

I'm saying they'll do the numbers on people testing positive per 100k of population I.e. this main dataset. They're not aiming for prevalence of 50 per 100k of population, just positive tests of that... Your last point (and all of Jerry's) assumes they've tested everyone or scaled up for not having done that, but AFAIK only the ONS convert to estimate of actual prevalence - not the main test data.. bear in mind we were at about 10 when I went to Germany last Sept so clearly the data is not scaled or false positive is <0.001%  and there was no covid then. I'll bet £100 they don't scale the main data.

[abarro81]

They’ve covered this on More or Less in the past - 23rd Sept last year. Just relistened to it, it’s the second segment about 8 minutes in. Conclusion seemed to be that false positive rate is nearer 0.05% and that worrying about it probably a distraction.

Sure, but even with zero prevalence, a 0.05% FPR gives 50 positives per 100k. So even with no prevalence of Covid in the population you're still hitting the 50 positive tests per 100k criteria. With increased prevalence the proportion of positive tests is only going to increase. I thought this was Fultonious's point.

Standard metric is per 100k of population not per 100k tests. Also proportion of +ves that are false goes up in low prevalence not high (due to lower real +ves while false is static)

[Jerry Morefat]

I'm saying they'll do the numbers on people testing positive per 100k of population I.e. this main dataset. They're not aiming for prevalence of 50 per 100k of population, just positive tests of that... Your last point (and all of Jerry's) assumes they've tested everyone or scaled up for not having done that, but AFAIK only the ONS convert to estimate of actual prevalence - not the main test data.. bear in mind we were at about 10 when I went to Germany last Sept so clearly the data is not scaled or false positive is <0.001%  and there was no covid then. I'll bet £100 they don't scale the main data.

I think I understand now. I hadn't realised it's done like this. Seems like a strange way to do it. Can't you game the statistic by testing very few people per 100k?

[spidermonkey09]

Interesting briefing today that cases and the R rate are no longer the key metrics for the easing of lockdown, instead focusing on hospitalisations. Quite a step change but probably logical as we get smaller numbers as they will be all over the place.

[Loos3-tools]

An interesting article on vaccine corruption. Gotta love being awake at this time ugh

https://mg.co.za/coronavirus-essentials/2021-01-30-bill-gates-big-pharma-and-entrenching-the-vaccine-apartheid/#click=https://t.co/EAVdJQl1Qy

Ps the BMJ article and pandemic cronyism maps are really worthwhile reads

[remus]

Interesting briefing today that cases and the R rate are no longer the key metrics for the easing of lockdown, instead focusing on hospitalisations. Quite a step change but probably logical as we get smaller numbers as they will be all over the place.

I imagine there's a cost aspect to it as well. I'd speculate that you need to be doing a lot of testing to get reasonable estimates for cases and r rate.

[Stabbsy]

Interesting briefing today that cases and the R rate are no longer the key metrics for the easing of lockdown, instead focusing on hospitalisations. Quite a step change but probably logical as we get smaller numbers as they will be all over the place.

I imagine there's a cost aspect to it as well. I'd speculate that you need to be doing a lot of testing to get reasonable estimates for cases and r rate.

I’d linked this to the switch in strategy from trying to get to zero Covid to trying to manage the impacts, plus the impacts of the vaccine. I think the suggestion of empirical data was that the vaccine reduced hospitalisation significantly more than it reduced cases, so making whatever cases did occur less serious. So cases might be less of a good indicator as more of the population is vaccinated.

I’d agree with both comments on the R-number though - potentially more volatile as cases reduce (although I’d guess they’d have to be lower than we managed in the summer for this to be a real issue) and more testing needed to get an accurate view. However, disagree with Remus’s suggestion on case estimates, although this might just be mixing up cases with prevalence. We try and estimate prevalence in ONS survey, but I think the government target was based on actual numbers of positive tests.

[Ru]

An interesting article on vaccine corruption. Gotta love being awake at this time ugh

Quite apart from the fact that it's a shit, one sided, article with no proper analysis about whether the utopian view of vaccine and intellectual property distribution it espouses was practically, politically and economically possible, please explain where the corruption is. Corruption being dishonest or fraudulent activity usually involving the taking of bribes, not just things you don't like.

Dan, come on, if every view you put forward is predicated on everyone, and every organisation, being corrupt, power-seeking liars, it's not surprising that everyone gets a bit annoyed. Quite apart from that, you do your own cause a disservice, because if there is any truth in it, it gets drowned out by the hyperbole and exaggeration.

[sdm]

R has always been a terrible metric for communicating the situation and for making policy decisions.

Too many assumptions have to be made to work back to R, so it gets very noisy.

There isn't enough data to use it on a local authority level so a reliance on R masks rapid local changes until the problem has already got out of hand.

It isn't intuitive to the average person (or politician). Tell them that R is 1.3-1.5 and the average person doesn't know what to think. Tell them that cases are doubling in some London local authorities every 3.8 days and the situation sounds as terrifying as it was in the autumn.

It becomes even less useful in the situation of emerging variants where the time period for transmission may not be the same.

Simple periods of doubling/halving for cases/hospitalisations/deaths require less guesswork, paint a clearer picture, and are more sensitive to change, allowing for quicker decisions to be made when the situation changes rapidly on a local level.

Interesting briefing today that cases and the R rate are no longer the key metrics for the easing of lockdown, instead focusing on hospitalisations. Quite a step change but probably logical as we get smaller numbers as they will be all over the place.

Greater weight on hospitalisations makes more sense as vaccine numbers grow. But the lag for hospitalisations is too big to drive policy if (when) we return to situation where things are getting worse again.

If you wait until hospitalisations have already started creeping up, it is already too late.

Hospitalisations driving policy only makes sense while everything is decreasing.

[TobyD]

R has always been a terrible metric for communicating the situation and for making policy decisions.

Too many assumptions have to be made to work back to R, so it gets very noisy.

There isn't enough data to use it on a local authority level so a reliance on R masks rapid local changes until the problem has already got out of hand.

It isn't intuitive to the average person (or politician). Tell them that R is 1.3-1.5 and the average person doesn't know what to think. Tell them that cases are doubling in some London local authorities every 3.8 days and the situation sounds as terrifying as it was in the autumn.

It becomes even less useful in the situation of emerging variants where the time period for transmission may not be the same.

Simple periods of doubling/halving for cases/hospitalisations/deaths require less guesswork, paint a clearer picture, and are more sensitive to change, allowing for quicker decisions to be made when the situation changes rapidly on a local level.

Interesting briefing today that cases and the R rate are no longer the key metrics for the easing of lockdown, instead focusing on hospitalisations. Quite a step change but probably logical as we get smaller numbers as they will be all over the place.

Greater weight on hospitalisations makes more sense as vaccine numbers grow. But the lag for hospitalisations is too big to drive policy if (when) we return to situation where things are getting worse again.

If you wait until hospitalisations have already started creeping up, it is already too late.

Hospitalisations driving policy only makes sense while everything is decreasing.

Really the best way to drive policy is surely to have a really well functioning test and trace system so you can get a broadly accurate estimate of risk of infection risk to an individual in a given area.

Unfortunately we have someone whose main expertise appears to be horse riding, and main qualifications having been a friend of the prime minister for decades. 

[Loos3-tools]

An interesting article on vaccine corruption. Gotta love being awake at this time ugh

Quite apart from the fact that it's a shit, one sided, article with no proper analysis about whether the utopian view of vaccine and intellectual property distribution it espouses was practically, politically and economically possible, please explain where the corruption is. Corruption being dishonest or fraudulent activity usually involving the taking of bribes, not just things you don't like.

Dan, come on, if every view you put forward is predicated on everyone, and every organisation, being corrupt, power-seeking liars, it's not surprising that everyone gets a bit annoyed. Quite apart from that, you do your own cause a disservice, because if there is any truth in it, it gets drowned out by the hyperbole and exaggeration.

Yes I’m aware it wasn’t the best article, the BMJ one was more informative as is the crony diagram. With regards ‘my’ cause, let’s face it that was lost a long time ago. I’ll not pretend to try and save face on here but stand by my comment about being warm and genuine in person. Yes I do believe that in any and all organisations people who seek power rise to the top where corruption is inevitable, the nhs included. You mentioned politics and economics as factors influencing the potential for a ‘vaccine utopia’, the fact that politics and economics are involved is the problem.

[Oldmanmatt]

R has always been a terrible metric for communicating the situation and for making policy decisions.

Too many assumptions have to be made to work back to R, so it gets very noisy.

There isn't enough data to use it on a local authority level so a reliance on R masks rapid local changes until the problem has already got out of hand.

It isn't intuitive to the average person (or politician). Tell them that R is 1.3-1.5 and the average person doesn't know what to think. Tell them that cases are doubling in some London local authorities every 3.8 days and the situation sounds as terrifying as it was in the autumn.

It becomes even less useful in the situation of emerging variants where the time period for transmission may not be the same.

Simple periods of doubling/halving for cases/hospitalisations/deaths require less guesswork, paint a clearer picture, and are more sensitive to change, allowing for quicker decisions to be made when the situation changes rapidly on a local level.

Interesting briefing today that cases and the R rate are no longer the key metrics for the easing of lockdown, instead focusing on hospitalisations. Quite a step change but probably logical as we get smaller numbers as they will be all over the place.

Greater weight on hospitalisations makes more sense as vaccine numbers grow. But the lag for hospitalisations is too big to drive policy if (when) we return to situation where things are getting worse again.

If you wait until hospitalisations have already started creeping up, it is already too late.

Hospitalisations driving policy only makes sense while everything is decreasing.

Really the best way to drive policy is surely to have a really well functioning test and trace system so you can get a broadly accurate estimate of risk of infection risk to an individual in a given area.

Unfortunately we have someone whose main expertise appears to be horse riding, and main qualifications having been a friend of the prime minister for decades.

We also have a large section of the population, who are uncooperative, arrogant, dismissive of “experts” etc and (to my mind) have a “Toddlerish” attitude to personal responsibility (like to scream “free country”, “I know my rights” or ‘Well, that’s my opinion and you can’t tell me what to think” etc. Yet seem incapable of understanding any form of responsibility or consideration for others or society. Hence they bring to mind a toddler screaming “NO” or “I don’t wanna go bed” or “Why can’t I eat the dog poo? I wanna eat the dog poo!”)

The sort of people, who would write “Bugs Bunny” in the name section of any  official form they have to fill in.

So, I’m not sure test and trace would have been successful enough, even if the management and roll out had been competent.

In many ways, to me, I think the whole debacle is rather representative of modern British society and a large chunk of it’s population.
“We” voted for them, and all the other pending issues, after all. We even voted to keep our current system of elections in place.

I mean, look at how many people buy newspapers because the newspapers make snide comments about a woman of colour who dared to marry into the Royal Family. I feels as if such a large number of people are happy to suffer any indignity or economic hardship, as long as they get to titter and sneer at “the Other” race/gender/residents of a different region/foreigner blah blah blah blah (insert meaningless distinction between fellow humans here).

Anyway.

Happy Saturday guys!   

[abarro81]

I don't think that article really relates to corruption, there's no dishonesty. Protecting IP or not sharing process knowledge might be mean or immoral, but not sure how it fits "corrupt". Perhaps that's just semantics, immoral isn't really any better 

I don't know enough about the inner workings of making these drugs to have a strong view on how logistically easy or not it would be to assist smaller or non-specialist manufacturers in scale-up. In PV they usually send teams of engineers to new facilities for the ramp process, but I imagine they're all currently busy ramping the main facilities (AztraZenica already have a big Indian fab for their vaccine by the looks of it)... pity they didn't do more digging into those kinds of issues - their only expert basically said that they don't know how feasible it would be to ramp at non-specialistals. Probably best to separate out production questions from purchase questions around rich countries being able to buy to the front of the queue (not exactly new!)

[Stu Littlefair]

Dan’s posts make a lot more sense if you imagine that he regards any sort of compromise that may be needed to get something done, or any decision taken for practical rather than moral reasons as “corruption”.

[andy popp]

the fact that politics and economics are involved is the problem.

I would love to know how politics and economics would not be involved?

[Loos3-tools]

Yikes it’s team lattice world police with the cavalry of messers Popp and Davies in tow. Once you’ve finished with me ye can head off to save the environment by burning rubber chips and and mining battery components. A necessary evil debate 

[abarro81]

Still not got the hang of actually engaging with the points have you 

[Davo]

To be fair, that was a pretty funny comment he made.

[Loos3-tools]

I don’t believe it’s a particularly radical idea that a large scale health intervention can be delivered without  meddling cronies and fucking Bond villain philanthropists cashing in on the misery. The BMJ article I posted and the crony map which no one has commented on show to some extent the degree of this. Show me a (leading) politician that isn’t a liar a fraud a deviant or a stooge and I’ll eat my hat

[teestub]

Yikes it’s team lattice world police with the cavalry of messers Popp and Davies in tow. Once you’ve finished with me ye can head off to save the environment by burning rubber chips and and mining battery components. A necessary evil debate

Straw men, moral equivalence, slippery slope reasoning, ad hominem are all regularly used to rebuff arguments here

[tomtom]

I don’t believe it’s a particularly radical idea that a large scale health intervention can be delivered without  meddling cronies and fucking Bond villain philanthropists cashing in on the misery. The BMJ article I posted and the crony map which no one has commented on show to some extent the degree of this. Show me a (leading) politician that isn’t a liar a fraud a deviant or a stooge and I’ll eat my hat

My little crony has been posted here before (it’s 2 months old?) and maybe it’s more for the politics thread. Being ‘generous’ - at moments of crisis you reach out to those you know first to help. The PPE decisions in the first few weeks smacked of that - and if you’re drowning you don’t choose which hand you grab to pull you out. But after that it really looks like it’s taking the piss if not blatant corruption.

It is a bugbear of mine that it’s called favours for mates, or chumocracy or some shit like that when it is really corruption. Over-riding due process to award contacts to friends.

When paid up next year - TTI will cost the equivalent of £560 per person. TTI is vital for tracking the pandemic - but whether this represents value for money... Maybe last summer a decision should have been made to move this to the NHS and bring it in house instead of SERCO/dido based (they could have walked away at that point fairly easily). Instead we get the “double down” culture where we keep on digging...

[Loos3-tools]

Yikes it’s team lattice world police with the cavalry of messers Popp and Davies in tow. Once you’ve finished with me ye can head off to save the environment by burning rubber chips and and mining battery components. A necessary evil debate

Straw men, moral equivalence, slippery slope reasoning, ad hominem are all regularly used to rebuff arguments here

Yeah well one can have a joke, right?

[mrjonathanr]

Chumocracy -amiable sounding euphemism - agreed. Corruption is the correct term, but smacks of other, less civilised countries, or so we would like to think.

When paid up next year - TTI will cost the equivalent of £560 per person. TTI is vital for tracking the pandemic

I don’t quite agree that TTI is vital for tracking the pandemic, TT. I’d say it is vital for isolating newly infectious people and so shutting down transmission.

If it doesn’t do that effectively, including fast, it’s worthless.