COVID-19

[KAJ]

I'm not even sure which data is more responsive to infection levels. It can take a few days from the emergence of symptoms to get a test result. How long does it take to go from first symptoms to hospitalisation?

CDC: Interim Clinical Guidance for Management of Patients with Confirmed Coronavirus Disease (COVID-19)

Clinical Progression
Among patients who developed severe disease, the median time to dyspnea from the onset of illness or symptoms ranged from 5 to 8 days, the median time to acute respiratory distress syndrome (ARDS) from the onset of illness or symptoms ranged from 8 to 12 days, and the median time to ICU admission from the onset of illness or symptoms ranged from 10 to 12 days.

[Bird on a Fire]

I'm not even sure which data is more responsive to infection levels. It can take a few days from the emergence of symptoms to get a test result. How long does it take to go from first symptoms to hospitalisation?

CDC: Interim Clinical Guidance for Management of Patients with Confirmed Coronavirus Disease (COVID-19)

Clinical Progression
Among patients who developed severe disease, the median time to dyspnea from the onset of illness or symptoms ranged from 5 to 8 days, the median time to acute respiratory distress syndrome (ARDS) from the onset of illness or symptoms ranged from 8 to 12 days, and the median time to ICU admission from the onset of illness or symptoms ranged from 10 to 12 days.

Ok, thanks. So hospitalisations probably lag about a week behind test results.

[lpm]

Can't be arsed right now to work out how many weeks too late this is. After Germany, before France, dunno compared to Ireland and Belgium.

Belgium's new lockdown starts this coming Monday. Ireland's is already underway, I think.

Yes, but need to calibrate it to where they are on the curve. In other words, France and Germany lockdown around the same day, but France is something like 6 weeks up the curve.

[AMS]

You'd think a government might expect the need for tests to go up when infections go up.

And, you know, plan for it.

The correct approach would be to have huge capacity during the summer lull that was used for surveys of entire towns - test 500,000 people and study exactly what the case distribution is. Or test every child in the week before schools go back. And then expand capacity further during waves to always keep ahead. If you're not wandering high streets asking passersby if they want a test going spare, you're not doing it right.

Evidence seems to indicate that Leicester/Tier 3 etc brings about an initial dip in cases. But as the weeks pass it never gets down properly and starts to go up again. Liverpool is going downwards at the moment.

Mass testing of whole towns in the infection hotspots would still be a good idea. But to be effective, people still need a financial incentive to isolate properly if they cannot afford to do so. That's still a big missing piece in the strategy.

[KAJ]

I'm not even sure which data is more responsive to infection levels. It can take a few days from the emergence of symptoms to get a test result. How long does it take to go from first symptoms to hospitalisation?

CDC: Interim Clinical Guidance for Management of Patients with Confirmed Coronavirus Disease (COVID-19)

Clinical Progression
Among patients who developed severe disease, the median time to dyspnea from the onset of illness or symptoms ranged from 5 to 8 days, the median time to acute respiratory distress syndrome (ARDS) from the onset of illness or symptoms ranged from 8 to 12 days, and the median time to ICU admission from the onset of illness or symptoms ranged from 10 to 12 days.

Ok, thanks. So hospitalisations probably lag about a week behind test results.

The quote above is times "from the onset of illness or symptoms". The same source says:

Incubation period
The incubation period for COVID-19 is thought to extend to 14 days, with a median time of 4-5 days from exposure to symptoms onset. One study reported that 97.5% of persons with COVID-19 who develop symptoms will do so within 11.5 days of SARS-CoV-2 infection.

So for tests on asymptomatic cases there would be an additional week or two. I guess (?) most tests are on symptomatic cases, but a substantial proportion may be on asymptomatic contacts, health workers, etc..

[Bird on a Fire]

There's between a third and a half as many Pillar 1 (NHS/PHE) tests as Pillar 2 (partner labs), depending on the week, and the latter are only given to people already showing symptoms.

This is a known issue in the data, btw, or at least people who've actually run the numbers have identified it too.

For example, one of PeteB's links a few pages back linked to this preprint https://www.medrxiv.org/content/10.1101 ... 20128777v1 comparing different data sources during the first wave via a data assimilation technique borrowed from geosciences. What they call 'experiment DH' was a model using deaths and hospitalisation data only to model the progress of the epidemic in England.

As discussed in Section 5.1, the reported number of cases testing positive for SARS-Cov-2
underestimates significantly the actual number of people infected by the virus. We can assess the
magnitude of this underestimation by comparing the number of cases predicted by experiment DH
to the reported number of cases. We find that as the number of people infected grew exponentially
before the lockdown started and the reproduction number declined to below 1.0, the percentage
of positive cases not being reported increased due to a severe lack of testing kits. On March
25th, this number peaked, with reporting of only 1% of cases. As more tests became available,
and simultaneously the rate of infection reduced, the percentage of unreported cases decreased.
From the beginning of May, about 2% of the accumulated cases are reported. The UK government
achieved its self-imposed target of 100,000 tests per day for the first time on May 1st. The estimated
percentage of reported cases can be compared to the percentage of asymptomatic cases. Estimates
of the proportion of asymptomatic cases vary widely between studies. However, WHO suggests
that 80% of infections are mild or asymptomatic [61], supported by [34]. Therefore, due to the
testing strategy in the UK (only testing people displaying symptoms), we could conclude that
approximately 10% of the accumulated symptomatic cases are being reported from the beginning
of May.

I know there's more tests available now, but still not enough, so it seems reasonable to assume that the proportion of cases identified varies as a function of the number of cases in a way that systematically underestimates case number.

[Woodchopper]

326 deaths today, which for a Saturday is especially bad.

[Little waster]

Johnson started well but he lost his upper frame in the Jive and his feet were all over the place by the end.

It’s a 4 from me.

Not sure the sequinned tux was a wise choice to break the news in TBH.

[Trinucleus]

Johnson started well but he lost his upper frame in the Jive and his feet were all over the place by the end.

It’s a 4 from me.

Not sure the sequinned tux was a wise choice to break the news in TBH.

Too much gapping

[sTeamTraen]

326 deaths today, which for a Saturday is especially bad.

I ran some historical numbers and Saturday turns out to be, on average, the "most average" day of the week for UK deaths, in the sense that deaths reported on Saturday are very close to 1/7 of the total. The percentage of deaths per weekday in the ECDC database are:

Saturday 14.18%
Sunday 8.35%
Monday 8.46%
Tuesday 19.10%
Wednesday 17.38%
Thursday 15.45%
Friday 17.10%

However, for cases, the most "typical" day is Tuesday:

Saturday 13.37%
Sunday 13.09%
Monday 11.94%
Tuesday 14.57%
Wednesday 15.93%
Thursday 15.89%
Friday 15.25%

This is interesting because one would expect the opposite: Your body doesn't care what day it is when you die and intensive care is presumably 24/7 (unless you believe Jeremy Hunt's stories of loads of extra deaths at the weekend from a few years ago), whereas it seems reasonable that testing is more likely to happen on a weekday. My guess is that maybe deaths are by when the death certificate was signed and tests are by some other criterion, or else there is just more regularity than I imagined in when people get tested.

[mediocrity511]

326 deaths today, which for a Saturday is especially bad.

I ran some historical numbers and Saturday turns out to be, on average, the "most average" day of the week for UK deaths, in the sense that deaths reported on Saturday are very close to 1/7 of the total. The percentage of deaths per weekday in the ECDC database are:

Saturday 14.18%
Sunday 8.35%
Monday 8.46%
Tuesday 19.10%
Wednesday 17.38%
Thursday 15.45%
Friday 17.10%

However, for cases, the most "typical" day is Tuesday:

Saturday 13.37%
Sunday 13.09%
Monday 11.94%
Tuesday 14.57%
Wednesday 15.93%
Thursday 15.89%
Friday 15.25%

This is interesting because one would expect the opposite: Your body doesn't care what day it is when you die and intensive care is presumably 24/7 (unless you believe Jeremy Hunt's stories of loads of extra deaths at the weekend from a few years ago), whereas it seems reasonable that testing is more likely to happen on a weekday. My guess is that maybe deaths are by when the death certificate was signed and tests are by some other criterion, or else there is just more regularity than I imagined in when people get tested.

Not all hospital deaths will be entirely random though. If you think of people on ventilators who aren't showing improvement and there comes a time when they switch the ventilator off. There may well likely be days of the week where this is more common, whether thats because of staffing, availability of relatives etc. If that was the case, as cases rise, I'd expect to see day of death evening out as decisions to withdraw treatment are reached faster in part due to the pressure of rising admissions.

[sTeamTraen]

Not all hospital deaths will be entirely random though. If you think of people on ventilators who aren't showing improvement and there comes a time when they switch the ventilator off. There may well likely be days of the week where this is more common, whether thats because of staffing, availability of relatives etc. If that was the case, as cases rise, I'd expect to see day of death evening out as decisions to withdraw treatment are reached faster in part due to the pressure of rising admissions.

I know that ventilators are sometimes switched off in case of brain death, and I can imagine it might become a thing even without that if ICUs started to fill up, but is that something that is happening now (or happened in the spring) for UK COVID-19 patients? I would have thought we'd have heard about it by now from outraged relatives. (I haven't even seen paranoid rumours that "that's why they're not letting relatives into the room with the dying patient - they're turning the ventilator off to make more money for Bill Gates!!!!1!" --- and I've seen a lot of other paranoid rumours.)

But in any case, all I can report is what the numbers show. I know that some countries report nothing on Saturday or Sunday because their national statistical people take the weekend off (e.g. Spain, where Monday's numbers are triple the other days of the week), but it's not always simply a case of "weekend numbers are less bad than the overall situation".

[AMS]

Is it possible that the Sunday/Monday dip for reported deaths is because some part of this information is coming in via the post? Unlikely for hospital deaths, but anything reported via GP surgeries could be reported this way.

[Woodchopper]

Thanks for that SteamTraen. Yes, as far as I know Tuesday usually has the most reported that day because it includes some delayed data for deaths that took place over the weekend. As far as I know the delay is due to staffing.

So a high number on a Saturday is more portentous than one on Tuesday.

[sTeamTraen]

Thanks for that SteamTraen. Yes, as far as I know Tuesday usually has the most reported that day because it includes some delayed data for deaths that took place over the weekend. As far as I know the delay is due to staffing.

So a high number on a Saturday is more portentous than one on Tuesday.

I'm assuming that Saturday is a Goldilocks day: Mostly weekday data, but with enough weekendiness to balance it out.

[Fishnut]

Not all hospital deaths will be entirely random though. If you think of people on ventilators who aren't showing improvement and there comes a time when they switch the ventilator off. There may well likely be days of the week where this is more common, whether thats because of staffing, availability of relatives etc. If that was the case, as cases rise, I'd expect to see day of death evening out as decisions to withdraw treatment are reached faster in part due to the pressure of rising admissions.

I know that ventilators are sometimes switched off in case of brain death, and I can imagine it might become a thing even without that if ICUs started to fill up, but is that something that is happening now (or happened in the spring) for UK COVID-19 patients? I would have thought we'd have heard about it by now from outraged relatives. (I haven't even seen paranoid rumours that "that's why they're not letting relatives into the room with the dying patient - they're turning the ventilator off to make more money for Bill Gates!!!!1!" --- and I've seen a lot of other paranoid rumours.)

But in any case, all I can report is what the numbers show. I know that some countries report nothing on Saturday or Sunday because their national statistical people take the weekend off (e.g. Spain, where Monday's numbers are triple the other days of the week), but it's not always simply a case of "weekend numbers are less bad than the overall situation".

From what I understand my neighbour had ventilation withdrawn. Technically he didn't have COVID as all the tests came back negative but this was early on in the pandemic (before lockdown) and they only had the tests for active virus so from what I can tell it was all the post-infection stuff that actually got him. They definitely treated him as a COVID case and he was in ICU on ventilation for several weeks before it was decided to end the interventions. They tried f.cking hard to get him better but it just didn't work. His niece was able to visit him before he went (in full PPE) and then had to self-isolate after (though we were had gone into lockdown by then anyway). They offered the family the option of continuing with treatment but it was clear by then it was a losing battle.

[Woodchopper]

Thanks for that SteamTraen. Yes, as far as I know Tuesday usually has the most reported that day because it includes some delayed data for deaths that took place over the weekend. As far as I know the delay is due to staffing.

So a high number on a Saturday is more portentous than one on Tuesday.

I'm assuming that Saturday is a Goldilocks day: Mostly weekday data, but with enough weekendiness to balance it out.

Probably.

As far as I remember the NHS weren’t registering deaths until the family had been informed. However this was taking longer at the weekend. So the backlog was cleared on Monday and reported on Tuesday.

[KAJ]

I ran some historical numbers and Saturday turns out to be, on average, the "most average" day of the week for UK deaths, in the sense that deaths reported on Saturday are very close to 1/7 of the total. The percentage of deaths per weekday in the ECDC database are:

Saturday 14.18%
Sunday 8.35%
Monday 8.46%
Tuesday 19.10%
Wednesday 17.38%
Thursday 15.45%
Friday 17.10%

However, for cases, the most "typical" day is Tuesday:

Saturday 13.37%
Sunday 13.09%
Monday 11.94%
Tuesday 14.57%
Wednesday 15.93%
Thursday 15.89%
Friday 15.25%

This is interesting because one would expect the opposite: Your body doesn't care what day it is when you die and intensive care is presumably 24/7 (unless you believe Jeremy Hunt's stories of loads of extra deaths at the weekend from a few years ago), whereas it seems reasonable that testing is more likely to happen on a weekday. My guess is that maybe deaths are by when the death certificate was signed and tests are by some other criterion, or else there is just more regularity than I imagined in when people get tested.

I don't know much about the ECDC data base. The UK COVID data is available by both reported/specimen date (cases) and death/published date (death). The different presentations show the differences you suggest. For cases the weekday cycle is more prominent by specimen date than report date. For deaths the weekday cycle is more prominent by published date than death date.

CasesvReport.png (32.72 KiB) Viewed 3065 times

CasesvSpec.png (33.97 KiB) Viewed 3065 times

DeathsvPublish.png (38.64 KiB) Viewed 3065 times

Don't seem to be able to add a 4th image. I'll put it in next post.

[KAJ]

4th image

Deathsvdeath.png (40.62 KiB) Viewed 3065 times

[sTeamTraen]

I don't know much about the ECDC data base.

For the UK data, ECDC strictly tracks the UK figures. For other countries it does a better job than, say, Worldometer. For example, today's headline number for Belgium was 20,056 new cases, but ECDC spreads them over the days when the test was recorded, because that data is available for Belgium. (As a result, Belgium's 14-day moving average curve always looks as if it is going slightly down, since the last day or two's numbers are quite small due to inevitable reporting delays; the other day a "skeptic" told me that this "proves" that infections have stopped rising, as if there are no human processes involved in bringing the numbers to us.)

[jimbob]

I ran some historical numbers and Saturday turns out to be, on average, the "most average" day of the week for UK deaths, in the sense that deaths reported on Saturday are very close to 1/7 of the total. The percentage of deaths per weekday in the ECDC database are:

Saturday 14.18%
Sunday 8.35%
Monday 8.46%
Tuesday 19.10%
Wednesday 17.38%
Thursday 15.45%
Friday 17.10%

However, for cases, the most "typical" day is Tuesday:

Saturday 13.37%
Sunday 13.09%
Monday 11.94%
Tuesday 14.57%
Wednesday 15.93%
Thursday 15.89%
Friday 15.25%

This is interesting because one would expect the opposite: Your body doesn't care what day it is when you die and intensive care is presumably 24/7 (unless you believe Jeremy Hunt's stories of loads of extra deaths at the weekend from a few years ago), whereas it seems reasonable that testing is more likely to happen on a weekday. My guess is that maybe deaths are by when the death certificate was signed and tests are by some other criterion, or else there is just more regularity than I imagined in when people get tested.

I don't know much about the ECDC data base. The UK COVID data is available by both reported/specimen date (cases) and death/published date (death). The different presentations show the differences you suggest. For cases the weekday cycle is more prominent by specimen date than report date. For deaths the weekday cycle is more prominent by published date than death date.
CasesvReport.pngCasesvSpec.png
DeathsvPublish.png
Don't seem to be able to add a 4th image. I'll put it in next post.

The ECDC data is offset by a day compared to the UK government's - so the UK government reports the data in the afternoon on the 25th and the ECDC will report that in the morning of the 26th as the data for the 26th. At least that's how I understand it:

ECDC data here:
https://www.ecdc.europa.eu/en/publicati ... -worldwide

[sTeamTraen]

The ECDC data is offset by a day compared to the UK government's - so the UK government reports the data in the afternoon on the 25th and the ECDC will report that in the morning of the 26th as the data for the 26th. At least that's how I understand it:

They release a new file some time between 13:00 and 14:00 every day. It usually includes the latest numbers from Belgium, which appear around 10:00, but not those from the Netherlands, which are usually available by 12:00 or so (but of course ECDC will have its own overheads).

Germany's figures change throughout the day - I presume each state reports separately, like the US and A.

[KAJ]

The ECDC data is offset by a day compared to the UK government's - so the UK government reports the data in the afternoon on the 25th and the ECDC will report that in the morning of the 26th as the data for the 26th. At least that's how I understand it:

They release a new file some time between 13:00 and 14:00 every day. It usually includes the latest numbers from Belgium, which appear around 10:00, but not those from the Netherlands, which are usually available by 12:00 or so (but of course ECDC will have its own overheads).

Germany's figures change throughout the day - I presume each state reports separately, like the US and A.

Comparing the ECDC data with UK governments:

• ECDC data is indeed offset by a day compared to UK govt; UK govt data for reported/published date 31 Oct is in ECDC data for 1 Nov
• ECDC data is by reported date (cases) and published date (deaths), with that offset. At first sight I can't see it listed by specimen date (cases) and death date (deaths), which are available in the UK govt. data.

[KAJ]

I reckon the UK new case rate has substantially declined since about 5 October.

chart.png (12.31 KiB) Viewed 2865 times

[KAJ]

I reckon the UK new case rate of increase has substantially declined since about 5 October.
chart.png

ETA "of increase"