[Sidehaas]

Really!?  A quote from that news article.

"Someone who stays in intensive care with Covid-19 for five weeks and dies would not be counted as a coronavirus death, for example."

Anyone here think that is honest?

Of course not, but no more so than someone who had a positive test in April and was hit by a bus last week.

If you want the gold standard for covid impact on death rates then excess deaths is the way to go - and that's been negative for weeks. But you are comparing with other nations, and in England until this revision, no one ever recovered from covid, so you couldn't compare.

You mean, no one who subsequently died after a positive covid test “recovered” before death.
This was a pretty stupid system.

There are medical professionals who post here, aren’t there? A death certificate will have a cause of death listed, right?
 A medical opinion on the reason for death. Surely this should be the primary criteria for determining who has died of Covid or complications arising from that infection, not an arbitrary four week cut off.

The ONS death data uses death certificates. They also do excess death stats. The point of the PHE data is not to catch everyone (they openly admit it doesn't, as can be seen from the ONS data), but rather to get as accurate as possible picture of the trend in deaths (and cases) as quickly as possible. This can then inform actions to control the disease. In this context, the change they have made was the right thing to do imho.

Ps not a medical professional.

[Offwidth]

Really!?  A quote from that news article.

"Someone who stays in intensive care with Covid-19 for five weeks and dies would not be counted as a coronavirus death, for example."

Anyone here think that is honest?

Of course not, but no more so than someone who had a positive test in April and was hit by a bus last week.

If you want the gold standard for covid impact on death rates then excess deaths is the way to go - and that's been negative for weeks. But you are comparing with other nations, and in England until this revision, no one ever recovered from covid, so you couldn't compare.

Except we know pretty quickly those who die from covid related death in hospitals and care homes; it's not rocket science.
This is gerrymandering. The ONS add the slower public death certificate information to hospital and care home deaths and say at least 55,000 covid related deaths so far. Those who die non-covid related deaths after testing positive are a small number (probably in the hundreds) compared to slow covid related deaths (in the thousands that we know of); covid increases mortality in those with cancer and heart disease (our two biggest baseline killers).

Excess deaths are not even bronze standard now we are away from the peak as you don't know what the baseline average would have been. At the peak such variability in the average was noise.

The point of PHE is to firefight as they have lost too many vital resources during austerity and have to do the best they can with the little they have. Their leadership should be working for the health of the nation with honest data, not proping up the government. That's the factor you miss: as well as utility the headline numbers need to be basically honest enough to retain public trust. I didn't realise the other countries of the Union used such flawed headline data until now. Why 4 weeks and not 6 or 10, given intensive care covid deaths are often seen to be drawn out?

[Oldmanmatt]

Really!?  A quote from that news article.

"Someone who stays in intensive care with Covid-19 for five weeks and dies would not be counted as a coronavirus death, for example."

Anyone here think that is honest?

Of course not, but no more so than someone who had a positive test in April and was hit by a bus last week.

If you want the gold standard for covid impact on death rates then excess deaths is the way to go - and that's been negative for weeks. But you are comparing with other nations, and in England until this revision, no one ever recovered from covid, so you couldn't compare.

Except we know pretty quickly those who die from covid related death in hospitals and care homes; it's not rocket science.
This is gerrymandering. The ONS add the slower public death certificate information to hospital and care home deaths and say at least 55,000 covid related deaths so far. Those who die non-covid related deaths after testing positive are a small number (probably in the hundreds) compared to slow covid related deaths (in the thousands that we know of); covid increases mortality in those with cancer and heart disease (our two biggest baseline killers).

Excess deaths are not even bronze standard now we are away from the peak as you don't know what the baseline average would have been. At the peak such variability in the average was noise.

The point of PHE is to firefight as they have lost too many vital resources during austerity and have to do the best they can with the little they have. Their leadership should be working for the health of the nation with honest data, not proping up the government. That's the factor you miss: as well as utility the headline numbers need to be basically honest enough to retain public trust. I didn't realise the other countries of the Union used such flawed headline data until now. Why 4 weeks and not 6 or 10, given intensive care covid deaths are often seen to be drawn out?

For once, I agree with Offwidth.
This is a blatant attempt to massage the figures.
If what you said was true, Duma, about the down and dirty nature of PHE historical numbers, then there is no need to revise those figures now, because they are already moderated by the more accurate, certificate based, ONS data.
Many of the deaths removed from the “died of Covid” category this week, are going to be put back into it by ONS.

This is a a cynical headline grab to try and minimise damage to the Government. Greater than 90% of the public will simply view it as “almost nobody is dying of Covid anymore and all the reported figures were wrong and probably will continue to be”, which is a gross miss-representation.

[Offwidth]

Another way to look at this is only about 300,000 people have tested positive so far, that's below 0.5% of the population and supposedly distributed evenly across the population. So of non-covid deaths there would be a max of 60 of the 12,000 baseline non covid deaths a week over a few months (a massive overestimate as death rates are not uniform with age). Nearly all those 5000 removed deaths are covid related.

The latest estimate of population infection is out today. It estimates 6%, so more than 9 in 10 have not been detected; a well known historic indictment on past test, track and trace. The scary factor is this shit performance on TTT is continuing from the analysis of current worldometer and ONS data. Death rates and hospitalisations are way too high for the low numbers of positive tests a few weeks back. We are missing between 3 in 4 and 9 in 10 of the infected with the testing systems currently in use unlike almost every other western country.

https://www.theguardian.com/world/2020/aug/13/up-to-6-of-englands-population-may-have-had-covid-study-suggests

[Oldmanmatt]

Aaaanyhoooo...

I take it everybody is in the mood for another ‘bad news preprint” ?

Yeah, me too:

Aerosol transmission looks highly likely.

https://www.medrxiv.org/content/10.1101/2020.08.03.20167395v1.full.pdf

Does leave me wondering why it’s not massively more wide spread than the 6% population penetration in the UK (July).

[Johnny Brown]

Yeah, not much too add there to the discussion we had months back. They've detected the virus from air samples and proved it's viability. Big wow. Clearly, either this is not the significant mode of transmission (otherwise we'd be in much deeper shit), or if it is infection must be far more widespread than thought and therefore not as deadly. Same as those scare-mongering cyclist 120m slipstreams we were warned about.

I can see why this sort of lab-based bottom-up research is easy-pickings for scientists but imho the emphasis in this pandemic needs to be top down: track, trace, identify the mechanisms of spread by how they are actually happening, not theoretical potential.

[Paul B]

Same as those scare-mongering cyclist 120m slipstreams we were warned about.

This was just bad science IMO (and I spent 4 years of my life staring into that software modelling solute transport).

[Oldmanmatt]

Yeah, not much too add there to the discussion we had months back. They've detected the virus from air samples and proved it's viability. Big wow. Clearly, either this is not the significant mode of transmission (otherwise we'd be in much deeper shit), or if it is infection must be far more widespread than thought and therefore not as deadly. Same as those scare-mongering cyclist 120m slipstreams we were warned about.

I can see why this sort of lab-based bottom-up research is easy-pickings for scientists but imho the emphasis in this pandemic needs to be top down: track, trace, identify the mechanisms of spread by how they are actually happening, not theoretical potential.

The problem being, it leads to this kind of thing:

https://www.independent.co.uk/news/world/americas/coronavirus-latest-airborne-aerosol-outbreaks-social-distancing-a9667706.html?utm_medium=Social&utm_source=Facebook#Echobox=1597301815

[Sidehaas]

I can see why this sort of lab-based bottom-up research is easy-pickings for scientists but imho the emphasis in this pandemic needs to be top down: track, trace, identify the mechanisms of spread by how they are actually happening, not theoretical potential.

You need both.
If you only have the bottom up theoretical model, then you have no integral evidence your model is valid from which to draw conclusions.
If you only have the top down empirical version, then you have no grounds to assume it is correct following a small change in input assumptions or conditions.
In an ideal world, you should develop a model with as much theoretical knowledge as available, then fill in the gaps with empirical models based of empirical evidence, then use a separate set of overall empirical data to validate the complete model, with a given uncertainty band.

Edit: obviously you could expect an individual scientific paper to be doing any one of these things, as long as they aren't drawing over ambitious conclusions about how their work should be applied.

[tomtom]

Inductive and deductive.

[IanP]

Really!?  A quote from that news article.

"Someone who stays in intensive care with Covid-19 for five weeks and dies would not be counted as a coronavirus death, for example."

Anyone here think that is honest?

Of course not, but no more so than someone who had a positive test in April and was hit by a bus last week.

If you want the gold standard for covid impact on death rates then excess deaths is the way to go - and that's been negative for weeks. But you are comparing with other nations, and in England until this revision, no one ever recovered from covid, so you couldn't compare.

Except we know pretty quickly those who die from covid related death in hospitals and care homes; it's not rocket science.
This is gerrymandering.

This seems like a very simplistic knee jerk response, and we have a enough simplistic responses from the government with out resorting to same sort of approach on here.

This was I believe a fairly well known issue with the differences in the way England was counting covid deaths compared to other UK countries - it was certainly covered by More or Less a few weeks ago.  Are you suggesting that the Scottish figures have been gerrymandered by the Scottish goverment for the whole of the epidemic?
   
Obviously there some difficult questions to be answered around how covid deaths are counted and I don't think the way PHE was doing it was obviously correct and it did appear to potentially give some peculiar figures as the epidemic went on.  To give one ancedotal example a friend of my wife's father who has long term illness went into hospital a couple of months ago, caught Covid while he was in there, recovered from Covid while in hospital, came back home with continuing long term illness, a month later is back in hospital and things aren't looking great - previous rules would say that if the worst happens he would be a Covid statistic when there is absolutely no evidence of that.

Also, interestingly the change to recording methods seems to provide an explanation for the unusual death rates to cases ratio you observed for the UK earlier in the thread - possibly a less exciting but potentially plausible explanation rather than some sort of exceptional coronavirus transmission pattern in the UK compared to the rest of Europe? 

[Offwidth]

No it doesn't account for the UK anomaly as it only takes about 10% off the covid deaths and yes Scotland et al are gerrymandering the data in my view as intensive care deaths beyond 4 weeks after testing positive are quite common.

[IanP]

No it doesn't account for the UK anomaly as it only takes about 10% off the covid deaths

Have you looked at the updated figures? Cases to deaths ratio for the UK look to fit within range of other similar countries:

e,g for 20th Jun / 10th July (based on 7 day average)

       Jun-20       Jul-10   Mortality   Cases / Deaths
France   468       15   3.2%   31
UK           1173     35   3.0%   34
Italy           289       15   5.2%   19
Sweden   1080    10   0.9%   108
Belgium   91        3          3.3%           30
Spain   334        3          0.9%           111

More recent figures for UK have mortality rate down to approx 2%, cases/deaths rations up to 50

[mrjonathanr]

Gerrymandering means messing around with constituency boundaries. Massaging or fiddling is what you do with figures.

Sorry to be a pedant. I can’t help it 

[Oldmanmatt]

Massaging or fiddling is what you do with figures.

Only with mutual consent.

No means no.

[mrjonathanr]

 

[SA Chris]

Gerrymandering means messing around with constituency boundaries. Massaging or fiddling is what you do with figures.

Sorry to be a pedant. I can’t help it 

Not really pedantic, it's a clear catachresis                                                                                                       

[Offwidth]

No it doesn't account for the UK anomaly as it only takes about 10% off the covid deaths

Have you looked at the updated figures? Cases to deaths ratio for the UK look to fit within range of other similar countries:

e,g for 20th Jun / 10th July (based on 7 day average)

       Jun-20       Jul-10   Mortality   Cases / Deaths
France   468       15   3.2%   31
UK           1173     35   3.0%   34
Italy           289       15   5.2%   19
Sweden   1080    10   0.9%   108
Belgium   91        3          3.3%           30
Spain   334        3          0.9%           111

More recent figures for UK have mortality rate down to approx 2%, cases/deaths rations up to 50

I was out yesterday so I didn't get a chance to look at how the changes affected the worldometer data until this morning. My ratios were based on 7 day rolling averages of deaths now and cases three weeks before.

For the UK the ratio using my methodology is now 50 and France 72, so yes the numbers are way more comparable on the new more dishonest UK data.

Wintertrees analysis upthread is unchanged and shows the UK is still missing 80% of cases with its current testing at the point of the most recent ONS estimates a few weeks back.  This means actual infections in the UK are about 5 times higher than indicated, something like 5000 new cases a day, and the missing data being so large has to be mostly those avoiding tests for whatever reason. These missing positives are the most likely people to be spreading the infection. 5000 a day means we are a lot worse than France, where from tomorrow we insist on quarenteen. My most serious concern is this, after all these months UK TTT is still failing. If we don't know anything about 80% of those infected we can only respond weeks too late when those missing positives lead to local outbreaks as visible from the more responsible people who do get tested.

[andy popp]

The little I've heard so far (e.g. from one dept) suggests clearing was a shit show yesterday. A very long and difficult day with many stressed and anxious students and parents, some so emotional and upset they were unable to make coherent decisions. Many students with predictions of AAB showing up on the phones with BCC and lower. And for the moment zero clarity as to where final numbers will end (though it seems certain they will be down).

Anyone else heard anything? TT, teachers here, or any parents?

[tomtom]

Nope. Will send some whatsapps as I was wondering.

Unusually no "we've had a great / challenging first day of clearing" missive from the VC either...

[Paul B]

If anyone else is looking at their currently booked holidays thinking they're not likely to go ahead it's worth noting that many airlines are allowing people to move their flights at little to no cost (this includes EasyJet which is free and you pay any additional increase in fare).

[Offwidth]

Gerrymandering means messing around with constituency boundaries. Massaging or fiddling is what you do with figures.

Sorry to be a pedant. I can’t help it 

Not really pedantic, it's a clear catachresis

The boundary moved in this case is the number of days to count as a covid death, after a positive test, to 28, a point where a significant proportion of those seriously ill on a ventilator (who subsequently die) are still alive in intensive care... I think it deserves largesse given it is a defined boundary the moved and English is a living language based on common usage.

[tomtom]

Anyone else heard anything? TT, teachers here, or any parents?

Word is "its gonna be a long slow clearing until people are sure of appeals etc.." & "This is going to drag out"

[andy popp]

Anyone else heard anything? TT, teachers here, or any parents?

Word is "its gonna be a long slow clearing until people are sure of appeals etc.." & "This is going to drag out"

Exactly as I heard just a few moments ago.

[Offwidth]

As I've retired it will be a week or so before I get news from my old place but we were overfull before clearing.