Donate to Remove ads

Got a credit card? use our Credit Card & Finance Calculators

Thanks to johnstevens77,Bhoddhisatva,scotia,Anonymous,Cornytiv34, for Donating to support the site

Coronavirus - General Chat - No statistics

The home for all non-political Coronavirus (Covid-19) discussions on The Lemon Fool
Forum rules
This is the home for all non-political Coronavirus (Covid-19) discussions on The Lemon Fool
swill453
Lemon Half
Posts: 7962
Joined: November 4th, 2016, 6:11 pm
Has thanked: 984 times
Been thanked: 3643 times

Re: Coronavirus - General Chat - No statistics

#359491

Postby swill453 » November 24th, 2020, 5:34 am

tjh290633 wrote:Quite why the 90% result was ignored by much of the media can only be construed as a wish to denigrate the Astrazeneca results.

Every report I heard or read yesterday reported the 90% figure.

Scott.

servodude
Lemon Half
Posts: 8272
Joined: November 8th, 2016, 5:56 am
Has thanked: 4435 times
Been thanked: 3565 times

Re: Coronavirus - General Chat - No statistics

#359492

Postby servodude » November 24th, 2020, 5:40 am

servodude wrote:
johnhemming wrote:
servodude wrote:
That date suggests infections peaked around the 4th of November
3-5 days median to symptoms from exposure
- 11-12 days subsequently to most likely death date


Scotia produced a good chart that linked admission to date of death. That used 13 days from that.

Admissions are (at least) 13 or 14 days from infection.

I would be interested to see your detailed calculations as to how you work back from 11 November to get to 4 November.


The spread of symptoms onset to death (from a meta-analysis paper) has been posted before (I'll have a look later and see if I can find the post) - peak of that is 11-12 days
- same with the infections to symptoms onset being median 3-5

Now where do you go to hospital between those is a lot fuzzier: but generally if you need hospital you'll be to the earlier side of the spread; and 3-4 days after symptoms is what the wards we've been working with expect - albeit our data set was a lot less noisy than yours

Worth remembering also that peaks implies the mode of the set, so I wouldn't be surprised if any scheme involving week wise "averaging" pushed the time out by a couple of days given the distribution of these sets
- which is partly why there's an expected few of days error on these things (other parts being spread, latency, lumpiness and measurement error)
- but anyways I'm pretty sure we'd have had identical posts if the lockdown had started a month ago instead of last week (or whenever it was) - the numbers would just be lower ;)

-sd


Hi John sorry for the delay TLF search fails me in finding my earlier post for now so here's some other references

have a look at
- https://www.medrxiv.org/content/10.1101 ... 20156307v1
its looking at data from Belgian cases
however it does mention https://www.medrxiv.org/content/10.1101 ... 20059972v2 from the UK which gives a 5.14 median days from symptoms onset to hospitalisation for your area - you can have a look at that for their working if you need a deeper dive (its also quite interesting)

what the Belgian centric paper does nicely is graphically display what I was mumbling about regards using medians to look at timing with respect peaks

Here's an extracted image from it which shows this clearly
Image
- notice how the peak is much earlier than the median due to the skew in the distribution

same with the length of stays in hospital
Image
- length of stay in hospital (death) is an interesting way to put it? sounds like a last holiday

which shows that if you're using with medians of time spread to work back from peaks you'll over count by quite a bit

hope that explains why I used the figures i did and why I think that claiming Admissions are (at least) 13 or 14 days from infection. is just a little bit wrong

- sd

redsturgeon
Lemon Half
Posts: 8912
Joined: November 4th, 2016, 9:06 am
Has thanked: 1309 times
Been thanked: 3667 times

Re: Coronavirus - General Chat - No statistics

#359494

Postby redsturgeon » November 24th, 2020, 6:26 am

zico wrote:
zico wrote: that they don't have any scientific logic as to why the half-dose full-dose regime would work more effectively, though there must be some reason why they switched to it.


Mystery solved. According to the Telegraph, the plan was to give two full-doses separated by a month, but someone messed up so about 3,000 people only got half the planned dose, but they decided to give the second full-dose and see what happened.
If they hadn't made the mistake, we'd be looking at 62% efficacy for this vaccine, which would have been pretty poor. As it is, something around 90% may be possible (though it's on extremely low positive infections for that group)

There have been lots of mistakes in the pandemic, but at last, a mistake with a very positive result! The AstraZeneca vaccine is more likely to be used worldwide (it's cheap, easy to store and non-profit making) so an improvement in its effectiveness could save hundreds of thousands of lives.


This is one of the reasons why this sort of process usually takes years and short cuts can lead to mistakes. I am a little baffled since my knowledge of drug development, though not of vaccines, would suppose that, at an early stage, dose response curves are calculated in order to optimise the therapeutic dose to give the correct balance between efficacy and side effects.

I agree that it seems odd that half the dose result in apparently better outcomes.

John

johnhemming
Lemon Quarter
Posts: 3858
Joined: November 8th, 2016, 7:13 pm
Has thanked: 9 times
Been thanked: 609 times

Re: Coronavirus - General Chat - No statistics

#359497

Postby johnhemming » November 24th, 2020, 6:50 am

servodude wrote:hope that explains why I used the figures i did and why I think that claiming Admissions are (at least) 13 or 14 days from infection. is just a little bit wrong

I got that information from somewhere. I think scotia's work is good for the relationship between admissions and deaths. I still don't see how you can get seven days although I accept the profile of range of dates is worth considering.

In the end if we are considering whether the second English lockdown was too late to have a material impact on the peak we need to consider a number of sources of data. The GP surveillance data has infections peaking earlier in October.

If you consider the chart of hospital admissions
https://coronavirus.data.gov.uk/details ... me=England

You will see it start plateauing (as a result primarily of levels of infection) in late October. We will see if there is a discontinuity as a result of the lockdown changes.

Hence even if you can substantiate the 7 day figure that I have not seen in the papers so far the substantive point (in that the lockdown was after the peak of infections - or after the point at which the peak was established by the start of a plateau) is well evidenced by other sources.

From the paper you refer to:
With the exception of Singapore, the sum of the mean
incubation period and mean onset-to-hospitalisation interval is never shorter than 9 days.


Looking at either lockdown we need to see the effect of the lockdown in outcomes. I think we can see that for the first English lockdown by comparing the deaths chart in England to that in Sweden which has the lockdown having an effect after about 4 weeks.

The substantive disagreement where there is one here is that I see the process of infection as being one which is as far as England is concerned primarily limited by levels of infection and only slightly affected by restrictions. This seasonal second wave arises because of a discontinuity in R0 caused probably by relative humidity shifts, but potentially having another seasonal cause.

However, I need to get on with my day (checking news (RNS) at 7am then taking children to school).

servodude
Lemon Half
Posts: 8272
Joined: November 8th, 2016, 5:56 am
Has thanked: 4435 times
Been thanked: 3565 times

Re: Coronavirus - General Chat - No statistics

#359510

Postby servodude » November 24th, 2020, 7:54 am

johnhemming wrote:The substantive disagreement where there is one here is

...using means or medians when they are absolutely the wrong metric ;)

I don't disagree with the measures you've posted; only that they've been used incorrectly.

-sd

johnhemming
Lemon Quarter
Posts: 3858
Joined: November 8th, 2016, 7:13 pm
Has thanked: 9 times
Been thanked: 609 times

Re: Coronavirus - General Chat - No statistics

#359511

Postby johnhemming » November 24th, 2020, 7:58 am

servodude wrote:
johnhemming wrote:The substantive disagreement where there is one here is

...using means or medians when they are absolutely the wrong metric ;)

I don't disagree with the measures you've posted; only that they've been used incorrectly.

johnhemming wrote:although I accept the profile of range of dates is worth considering.

dealtn
Lemon Half
Posts: 6072
Joined: November 21st, 2016, 4:26 pm
Has thanked: 441 times
Been thanked: 2324 times

Re: Coronavirus - General Chat - No statistics

#359531

Postby dealtn » November 24th, 2020, 9:24 am

zico wrote:I'm genuinely interested in trying to understand the anti-lockdown rationale ...

(There's an associated argument that cracking down on Covid deaths causes more cancer deaths, suicides and other forms of deaths. This argument states that hospitals should be more focused on treating non-Covid diseases. The big problem here is that if there's a Covid epidemic, the hospitals can't treat non-Covid diseases, because they'll be full of Covid patients. Even if there is capacity, as in the first wave, cancer patients will be deterred from visiting hospitals because of the risk of catching Covid. The only way around this is to prevent Covid patients from entering hospitals, but then they will die at home, and the death rate will be 4/5 times if they are denied medical care)



It's not true that hospitals can't treat non-Covid patients. There is/was a switch in resource to Covid treatment, but plenty of resource isn't compatible. In wave 1 many wards/operating theatres (and GP surgeries) were simply closed, but the operating theatres, GPs etc. weren't switched and used for treating Covid.

It also wasn't true that (most) cancer patients were deterred from going to hospital. Many Cancer facilities aren't even in conventional hospitals, and the ones that are were mostly segregated from the Covid parts (as was the case with most facilities where "zones" were created). I would suggest most cancer patients, and certainly those with first diagnosis would have welcomed the certainty and diagnosis of cancer despite the perceived risk of Covid transmission. Furthermore by closing GP surgeries, dentists, beauty salons etc a large number of early stage 1 detections will have been missed that become first diagnosed at stages 2 or 3, dangerous enough in itself, even more so with a backlog of treatments and queuing that might make them stage 3 or 4.

Of course when wave 1 accelerated with a disease that was "new" and "scary" mistakes will have been made. As more was learnt, and specifically as the casualty numbers, be that tests, cases, deaths, or some other measurement plummeted post April to July/August etc it was scandalous that so many NHS facilities remained closed. Not only were Covid wards nearly empty (or completely unused in the case of some Nightingale hospitals) but the closed non-Covid facilities remained mothballed.

I think it perfectly possible to hold a view that (some) lockdowns, to use that crude term, are necessary at different points in the "cycle", but that more effective or efficient use of facilities under that lockdown should have, and should be happening. To many advocates, and managers of the pandemic the "Covid" excuse for making decisions to close or limit things has been argued for or executed extremely badly. Wave 2 has shown a degree of learning from the earlier months, and schools are open for instance. But it is still the case that many GP surgeries are effectively closed, as are many dentists, and a host of health provision is both severely curtailed and subject to large waiting lists, queues and rationing.

88V8
Lemon Half
Posts: 5769
Joined: November 4th, 2016, 11:22 am
Has thanked: 4098 times
Been thanked: 2560 times

Re: Coronavirus - General Chat - No statistics

#359547

Postby 88V8 » November 24th, 2020, 10:12 am

Given that the Helsinki dogs can detect COVID five days before it shows in PCR, does this mean that the initial estimate of incubation time was wrong, and is nearer ten days than five?

Apologies if this has been covered.
It would obviously make a significant difference in interpreting the effectiveness of control measures.

V8

tjh290633
Lemon Half
Posts: 8209
Joined: November 4th, 2016, 11:20 am
Has thanked: 913 times
Been thanked: 4097 times

Re: Coronavirus - General Chat - No statistics

#359593

Postby tjh290633 » November 24th, 2020, 11:42 am

swill453 wrote:
tjh290633 wrote:Quite why the 90% result was ignored by much of the media can only be construed as a wish to denigrate the Astrazeneca results.

Every report I heard or read yesterday reported the 90% figure.

Scott.

I got notifications on my mobile phone from Bloomberg and the Telegraph, both of which highlighted the 70% figure in the headline.

TJH

Mike4
Lemon Half
Posts: 7088
Joined: November 24th, 2016, 3:29 am
Has thanked: 1638 times
Been thanked: 3796 times

Re: Coronavirus - General Chat - No statistics

#359748

Postby Mike4 » November 24th, 2020, 9:13 pm

zico wrote:
zico wrote: that they don't have any scientific logic as to why the half-dose full-dose regime would work more effectively, though there must be some reason why they switched to it.


Mystery solved. According to the Telegraph, the plan was to give two full-doses separated by a month, but someone messed up so about 3,000 people only got half the planned dose, but they decided to give the second full-dose and see what happened.


Y'know, that Telegraph explanation in the cold light of day really doesn't seem credible to me. With all the care poured into designing, manufacturing and testing this new vaccine (any new vaccine), I can't believe they are making mistakes like dosing 3,000 people with half the dose the trial design specified.

Or do they? If they do, I'm damned if I trust the trial results. What other sloppy mistakes might they have made and NOT noticed? What if they accidentally inject me with say, ten times the correct dose instead of half?

scotia
Lemon Quarter
Posts: 3561
Joined: November 4th, 2016, 8:43 pm
Has thanked: 2371 times
Been thanked: 1943 times

Re: Coronavirus - General Chat - No statistics

#359751

Postby scotia » November 24th, 2020, 9:40 pm

johnhemming wrote:
servodude wrote:hope that explains why I used the figures i did and why I think that claiming Admissions are (at least) 13 or 14 days from infection. is just a little bit wrong

I got that information from somewhere. I think scotia's work is good for the relationship between admissions and deaths. I still don't see how you can get seven days although I accept the profile of range of dates is worth considering.

A clarification - my analysis compared reported deaths against reported admissions, using the most up to date data from the UK government source.
See https://coronavirus.data.gov.uk/details ... pers-guide which describes how this data is selected
I used areatype= nation, areaname = England, admissions=newAdmissions, deaths=newDeaths28DaysByPublishDate
I believe the admissions data is reported at the date of admission, but the deaths data is reported at the date of its registration, which is likely to be spread for several days after death. So the predictions I made were all of the reported deaths at the date of registration.
I can extract deaths at the actual death date if I use the filter deaths =newDeaths28DaysByDeathDate, however this suffers from being a number which gets updated as the data is received. So it doesn't settle down to a final figure until a (variable) number of days have passed. In contrast the ByPublishDate is published each evening, and remains at that published value.
I must take a closer look at both death data reports to get some estimate of the time lag in between, but a cursory look suggests that the difference may be around 4 to 5 days. So if I were to predict the delay between admissions to actual date of death, I would trim this off my 13-day figure. Apologies for any confusion. I'll get back with a more scientific computation of the difference on the science board.
PS - the headline deaths reported in the media are usually (if not always) the Deaths by Publish Date.

1nvest
Lemon Quarter
Posts: 4326
Joined: May 31st, 2019, 7:55 pm
Has thanked: 681 times
Been thanked: 1316 times

Re: Coronavirus - General Chat - No statistics

#359775

Postby 1nvest » November 24th, 2020, 11:17 pm

redsturgeon wrote:This is one of the reasons why this sort of process usually takes years and short cuts can lead to mistakes.

Raw virus might affect 10% severely, perhaps half of that group going on to die. 90% majority endure no/mild/modest symptoms, maybe at worst a week in bed with flu like condition. A rushed vaccine rolled out across the majority could turn around in 5, 10, whatever years time with severe adverse side effects hitting the majority of those inoculated.

Personally I think the vaccine should only be given to those more likely to be in the 10% group, and not more broadly across the remainder 90%. Also why has the UK ordered 350 million doses of vaccines relative to a population of 65 million. Seems like yet a further case of political pals being gifted taxpayers money, in some cases for absolutely zero public benefit in return.

Mike4
Lemon Half
Posts: 7088
Joined: November 24th, 2016, 3:29 am
Has thanked: 1638 times
Been thanked: 3796 times

Re: Coronavirus - General Chat - No statistics

#359778

Postby Mike4 » November 24th, 2020, 11:33 pm

1nvest wrote: Also why has the UK ordered 350 million doses of vaccines relative to a population of 65 million. Seems like yet a further case of political pals being gifted taxpayers money, in some cases for absolutely zero public benefit in return.


I wondered this too, but surmised it was because the vaccine needs two doses to start with, and will probably only last a few months so will need re-administering annually.

Even so, 350m seems too much. On the other hand have AstraZenica not pledged to supply it at zero profit "during the pandemic"?

gryffron
Lemon Quarter
Posts: 3608
Joined: November 4th, 2016, 10:00 am
Has thanked: 551 times
Been thanked: 1587 times

Re: Coronavirus - General Chat - No statistics

#359790

Postby gryffron » November 25th, 2020, 1:00 am

1nvest wrote:Raw virus might affect 10% severely, perhaps half of that group going on to die. 90% majority endure no/mild/modest symptoms, maybe at worst a week in bed with flu like condition. A rushed vaccine rolled out across the majority could turn around in 5, 10, whatever years time with severe adverse side effects hitting the majority of those inoculated.
Personally I think the vaccine should only be given to those more likely to be in the 10% group, and not more broadly across the remainder 90%.

But we don't know who the 10% are. Young and previously healthy NHS staff have been dropping like flies. Most likely the quantity of initial viral dose is a big driver of disease severity.

Also, the main purpose of the vaccine is to prevent SPREAD. Which can only be achieved by using it widely. For example. Suppose the vaccine doesn't work on you, but does on your closest friends. Because they can't catch it, they can't pass it to you. So the vaccine saved your life even though it didn't work for you. This is why a vaccine doesn't need to be 100% effective to work. As long as it is effective enough it can reduce R below 1 so the disease dies out.

1nvest wrote:Also why has the UK ordered 350 million doses of vaccines relative to a population of 65 million. Seems like yet a further case of political pals being gifted taxpayers money, in some cases for absolutely zero public benefit in return.

Because we didn't know if ANY of them would work. And it was likely MUCH cheaper to order them up front speculatively, than to order them AFTER they have been proved to work. When, presumably, everyone in the whole world will be clamouring for them at any price.

Gryff

servodude
Lemon Half
Posts: 8272
Joined: November 8th, 2016, 5:56 am
Has thanked: 4435 times
Been thanked: 3565 times

Re: Coronavirus - General Chat - No statistics

#359794

Postby servodude » November 25th, 2020, 5:27 am

88V8 wrote:Given that the Helsinki dogs can detect COVID five days before it shows in PCR, does this mean that the initial estimate of incubation time was wrong, and is nearer ten days than five?

Apologies if this has been covered.
It would obviously make a significant difference in interpreting the effectiveness of control measures.

V8


This is my favourite covid research topic!

The "5 days before" thing really just means that they are more sensitive than the PCR test - as well as being a damn sight faster, and well... dogs!

The Finnish study should be published next month
- and they're already starting trials with dogs in other places:
https://www.afr.com/policy/health-and-e ... 125-p56hql
https://www.financialexpress.com/lifest ... 5500/lite/

I just think this is brilliant
-sd

GoSeigen
Lemon Quarter
Posts: 4351
Joined: November 8th, 2016, 11:14 pm
Has thanked: 1590 times
Been thanked: 1579 times

Re: Coronavirus - General Chat - No statistics

#359795

Postby GoSeigen » November 25th, 2020, 5:48 am

zico wrote:I'm aware there are a lot of intelligent and thoughtful people who are against lockdown, and I'm just trying to understand the rationale behind it.


You don't understand why people might be against governments telling you where you can and can't go, when and in what manner, and criminalise you if you infringe those rules?


GS

johnhemming
Lemon Quarter
Posts: 3858
Joined: November 8th, 2016, 7:13 pm
Has thanked: 9 times
Been thanked: 609 times

Re: Coronavirus - General Chat - No statistics

#359798

Postby johnhemming » November 25th, 2020, 6:40 am

scotia wrote:I can extract deaths at the actual death date if I use the filter deaths =newDeaths28DaysByDeathDate, however this suffers from being a number which gets updated as the data is received. So it doesn't settle down to a final figure until a (variable) number of days have passed.

Thank you for your earlier work on this. It would be good to do it again using the death date. I suggest you ignore the last week of deaths as they are figures which are most likely to be subject to change and that would skew the final calculation. It will still give you August, September, October and part of November to consider. It may be best to ignore August as well as you wish to be fully into the seasonal virus wave.

johnhemming
Lemon Quarter
Posts: 3858
Joined: November 8th, 2016, 7:13 pm
Has thanked: 9 times
Been thanked: 609 times

Re: Coronavirus - General Chat - No statistics

#359799

Postby johnhemming » November 25th, 2020, 6:43 am

GoSeigen wrote:
zico wrote:I'm aware there are a lot of intelligent and thoughtful people who are against lockdown, and I'm just trying to understand the rationale behind it.


You don't understand why people might be against governments telling you where you can and can't go, when and in what manner, and criminalise you if you infringe those rules?


Additionally there are questions as to whether there is any benefit from "lockdown" per se.

In Europe it appears that the countries who had the most effective "lockdown" previously are now having the largest number of seasonal wave deaths.

That, of course, would not be surprising as people would have been prevented from getting a milder version of the virus (the same virus, but a smaller viral load).

In England this lockdown started after the rate of infection had peaked and was going down. (primary source GP consultations).

redsturgeon
Lemon Half
Posts: 8912
Joined: November 4th, 2016, 9:06 am
Has thanked: 1309 times
Been thanked: 3667 times

Re: Coronavirus - General Chat - No statistics

#359803

Postby redsturgeon » November 25th, 2020, 7:07 am

johnhemming wrote:
GoSeigen wrote:
zico wrote:I'm aware there are a lot of intelligent and thoughtful people who are against lockdown, and I'm just trying to understand the rationale behind it.


You don't understand why people might be against governments telling you where you can and can't go, when and in what manner, and criminalise you if you infringe those rules?


Additionally there are questions as to whether there is any benefit from "lockdown" per se.

In Europe it appears that the countries who had the most effective "lockdown" previously are now having the largest number of seasonal wave deaths.



That, of course, would not be surprising as people would have been prevented from getting a milder version of the virus (the same virus, but a smaller viral load).

In England this lockdown started after the rate of infection had peaked and was going down. (primary source GP consultations).


You seem to be missing the fact that in England we have LINO, so in actuality we have run a very similar course to Sweden who many have been praising for their approach, while pretending we have put similar measures in place to the likes of France and Spain.

How does LINO work? The government announces various measures, usually too late and then in the vast majority of cases fails to enforce them. The end result being very similar to if the measures had never been announced but with the benefit to the government that they can always look back and point to the very prudent measures they put in place.

In the end this may prove to be a brilliant strategy, much like AZ's strategy on the vaccine dosing. Why rely on science when luck will get you through.

A graph I saw yesterday shows that the UK and Japan are the two countries likely to hit herd immunity first in mid 2021.Back of the net!

John

swill453
Lemon Half
Posts: 7962
Joined: November 4th, 2016, 6:11 pm
Has thanked: 984 times
Been thanked: 3643 times

Re: Coronavirus - General Chat - No statistics

#359818

Postby swill453 » November 25th, 2020, 7:48 am

redsturgeon wrote:A graph I saw yesterday shows that the UK and Japan are the two countries likely to hit herd immunity first in mid 2021.Back of the net!

I take it that depends in the main on rolling out an effective vaccination programme?

Scott.


Return to “Coronavirus Discussions”

Who is online

Users browsing this forum: No registered users and 5 guests