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 - Modelling Aspects Only

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

Re: Coronavirus - Modelling Aspects Only

#387839

Postby scotia » February 18th, 2021, 9:06 pm

Another week of (English) data. To recap:- The Blue Points are the deaths by publish date, summed over the preceding week.
As I described in previous posts there has been a strong correlation between the deaths by publish date, and the hospital admissions of some days previous. The Red Points are the hospital admissions, summed over a week, multiplied by 0.3, and moved forward by 11 days - And these are renamed as being the projected deaths by publish date. The size of the vertical bars are the statistical standard deviations, assuming a Poisson distribution. These parameters provided a reasonable fit a few weeks ago, however going back a few months, a 13 day slip and a factor of 0.265 provided an optimal fit. There was speculation that the increase in the mortality ratio between those times could possibly have been due to over-stretched hospitals, or the new Kent virus variant. However - back to the present and more good news which may be vaccination related - and which makes the 0.3 factor look decidedly pessimistic.

Image

The actual death rate curve continues to deviate downwards from the projected rate based on the admissions. Over the past week, a factor of 0.26 gives a much closer fit than 0.3, and the factor appears to be reducing further over that week. I'm reasonably convinced that we are now seeing the effect of vaccinations for the elderly in Care Homes. So Care Home deaths are substantially decreasing, without have a similar effect on admissions.
The Scottish vaccination program targeted Care Homes as the first priority, so I think we may be ahead of England in this category. Yesterday the National Records of Scotland (NRS) stated that there had been a 62% reduction in Care Home Deaths in the last 3 weeks. Looking at one of their graphs, and accepting that the data was statistically very noisy, it looks like the ratio of Covid-19 Care Home deaths to Covid-19 Hospital deaths peaked at around 1 in 3, but is now back down to about 1 in 10. However I should stress that this is just a rough estimate.
Source of the Scottish information https://www.nrscotland.gov.uk/files//statistics/covid19/covid-deaths-21-report-week-06.pdf

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

Re: Coronavirus - Modelling Aspects Only

#390084

Postby scotia » February 25th, 2021, 9:01 pm

Another week of (English) data. To recap:- The Blue Points are the deaths by publish date, summed over the preceding week.
As I described in previous posts there has been a strong correlation between the deaths by publish date, and the hospital admissions of some days previous. The Red Points are the hospital admissions, summed over a week, multiplied by 0.3, and moved forward by 11 days - And these are renamed as being the projected deaths by publish date. The size of the vertical bars are the statistical standard deviations, assuming a Poisson distribution. These parameters provided a reasonable fit about a month ago, however going back several months ago, a 13 day slip and a factor of 0.265 provided an optimal fit. There was speculation that the increase in the mortality ratio between those times could possibly have been due to over-stretched hospitals, or the new Kent virus variant. However - back to the present and continuing good news which is probably vaccination related - and which makes the 0.3 factor look decidedly pessimistic.

Image

The deaths are continuing to fall well below the admissions curve - which last week I hypothesised as being due to the vaccination related reduction in deaths within Care Homes which did not significantly affect admissions. I'm a little surprised at the (recent) slight reduction in downward gradient of the admissions curve - but this could simply be due to reduced pressure on hospital beds. Its the death curve that really matters - and its downward progress certainly looks good.
I mentioned last week that the factor of 0.3 between the deaths and the time slipped admissions was now clearly too large, and I suggested that 0.26 was now more appropriate - so here is a display using 0.26

Image

It will be interesting to see what the next week of data brings - since we should soon be seeing the effect of vaccinations in the at-home oldies. Its now around three weeks since we (in our mid seventies) received our first vaccinations.

zico
Lemon Quarter
Posts: 2139
Joined: November 4th, 2016, 12:12 pm
Has thanked: 1074 times
Been thanked: 1086 times

Re: Coronavirus - Modelling Aspects Only

#392374

Postby zico » March 4th, 2021, 4:54 pm

I've tried doing some simple modelling for future UK deaths using the following modelling assumptions.

57 million UK adults at risk from Covid.
1% Death rate for those infected.
57m x 1% = 570,000
Assuming 100% of population become infected (it's 20-25% already).
We've already had 130,000 deaths (from 20-25% of population being infected) so (570k-130k = 440k) further deaths if everyone was infected (and unvaccinated).

There'll be a big difference between vaccinated and unvaccinated people, so I've split the remaining potential deaths into vac/non-vac.

Scenario 1
Assuming 80% vaccination take-up, there'd be 88,000 unvaccinated deaths.
Assuming 90% vaccination protection against death (based on latest analysis).
For the vaccinated group, their (unvaccinated) expected deaths would be 352k x (1-90%) = 35,200.

So total deaths would be 88,000+35,200 = 123,200

Scenario 2 - Assuming 90% vac take-up and 90% vac protection gives -
Unvac'd deaths = 44,000
Vac'd deaths = 39,600

Total deaths = 83,600

If you think only 50% of the population would be exposed to Covid infection, then you simply halve the above numbers.

What could be wrong with these above scenarios?

Firstly, the unvaccinated people might be only those who know/believe themselves to be very low-risk (e.g. 20-30 year olds). However, there's growing evidence that the more vulnerable sections of society are least likely to get vaccinated).

Secondly, the higher the vaccinated %, the less likely the virus is to spread. However, I believe that infections will inevitably increase when all restrictions are dropped after June 21st, so over the next 12 months, I think it very likely that the vast majority of people will be exposed.

Thirdly, maybe a new wave of infections wouldn't be as deadly as the previous two waves because all the vulnerable people have already died from Covid. But that logic was used by people to say the 2nd wave of infections would have much lower deaths than the 1st wave - and we all know that was wrong.

When restrictions are lifted, there'll be a lot of elderly people mingling with friends and family, and elderly people mainly do their mingling indoors, which is where the virus spreads much more easily.

Another 12 months of deaths at similar levels to that of the last 12 months would be pretty bad news for the NHS, although it would just about cope.

NB - The above assumptions don't allow for emergence of vaccine-resistant variants (either home-grown or imported) which I think is a pretty optimistic assumption.

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

Re: Coronavirus - Modelling Aspects Only

#392388

Postby dealtn » March 4th, 2021, 5:22 pm

zico wrote:
What could be wrong with these above scenarios?



Assuming at the outset there is a 100% chance of infection and the resulting 1% morbidity rate?

It is entirely possible 20% of the population have a "natural" immunity, from perhaps previous Coronavirus type infections and triggered T-cell responses. Those 20%, or whatever, will have a different set of responses to the other 80%, on which your assumptions of 1% mortality rate, and 20-25% being infected already are based.

Your model appears to be 130,000 out of 57 million, with 20-25% already infected, assuming 100% infection and the remaining 75-80% will be representative of the 20-25% already infected.

What if in reality 20% have natural immunity, a quarter of whom have been exposed. Your 57 million could be 11 million unable to to be infected, of whom 3 million have been exposed. Your 20-25% number is actually 25-30%. Of the remaining 40 million odd in the population 20% of them might be immune before vaccination effects are even taken into account. Your results will be hugely different if not everyone can be affected by it to start with.

zico
Lemon Quarter
Posts: 2139
Joined: November 4th, 2016, 12:12 pm
Has thanked: 1074 times
Been thanked: 1086 times

Re: Coronavirus - Modelling Aspects Only

#392394

Postby zico » March 4th, 2021, 5:34 pm

dealtn wrote:
zico wrote:
What could be wrong with these above scenarios?



Assuming at the outset there is a 100% chance of infection and the resulting 1% morbidity rate?

It is entirely possible 20% of the population have a "natural" immunity, from perhaps previous Coronavirus type infections and triggered T-cell responses. Those 20%, or whatever, will have a different set of responses to the other 80%, on which your assumptions of 1% mortality rate, and 20-25% being infected already are based.

Your model appears to be 130,000 out of 57 million, with 20-25% already infected, assuming 100% infection and the remaining 75-80% will be representative of the 20-25% already infected.

What if in reality 20% have natural immunity, a quarter of whom have been exposed. Your 57 million could be 11 million unable to to be infected, of whom 3 million have been exposed. Your 20-25% number is actually 25-30%. Of the remaining 40 million odd in the population 20% of them might be immune before vaccination effects are even taken into account. Your results will be hugely different if not everyone can be affected by it to start with.


Fair points, but I'm not relying on the 20-25% already infected figure for my modelling - I just included that to show the % already infected (or exposed if you prefer) is consistent with the number of deaths (as 130k deaths is 22% of the 570k deaths expected if the IFR is 1%).

Are you saying you believe that 20-25% have been infected and there's another 20% with natural immunity that were exposed, but not infected? So giving 40-45% of the population exposed? If so, where does your 20% figure come from?

I did say that if anyone believed only half of the remaining population could be infected, then just halve my figures to get the appropriate estimate.

Julian
Lemon Quarter
Posts: 1385
Joined: November 4th, 2016, 9:58 am
Has thanked: 532 times
Been thanked: 676 times

Re: Coronavirus - Modelling Aspects Only

#392401

Postby Julian » March 4th, 2021, 5:47 pm

zico wrote:...
What could be wrong with these above scenarios?
...
NB - The above assumptions don't allow for emergence of vaccine-resistant variants (either home-grown or imported) which I think is a pretty optimistic assumption.

One other thing you don’t allow for, understandably since we have no idea if it will happen and if it does then what its effect will be, is improvements in treatments lowering mortality rates. And on new variants it’s also at least possible that new less lethal variants could emerge although in a timescale of the next 6 to 12 month maybe that’s wishful thinking.

- Julian

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

Re: Coronavirus - Modelling Aspects Only

#392402

Postby dealtn » March 4th, 2021, 5:48 pm

zico wrote:
dealtn wrote:
zico wrote:
What could be wrong with these above scenarios?



Assuming at the outset there is a 100% chance of infection and the resulting 1% morbidity rate?

It is entirely possible 20% of the population have a "natural" immunity, from perhaps previous Coronavirus type infections and triggered T-cell responses. Those 20%, or whatever, will have a different set of responses to the other 80%, on which your assumptions of 1% mortality rate, and 20-25% being infected already are based.

Your model appears to be 130,000 out of 57 million, with 20-25% already infected, assuming 100% infection and the remaining 75-80% will be representative of the 20-25% already infected.

What if in reality 20% have natural immunity, a quarter of whom have been exposed. Your 57 million could be 11 million unable to to be infected, of whom 3 million have been exposed. Your 20-25% number is actually 25-30%. Of the remaining 40 million odd in the population 20% of them might be immune before vaccination effects are even taken into account. Your results will be hugely different if not everyone can be affected by it to start with.


Fair points, but I'm not relying on the 20-25% already infected figure for my modelling - I just included that to show the % already infected (or exposed if you prefer) is consistent with the number of deaths (as 130k deaths is 22% of the 570k deaths expected if the IFR is 1%).

Are you saying you believe that 20-25% have been infected and there's another 20% with natural immunity that were exposed, but not infected? So giving 40-45% of the population exposed? If so, where does your 20% figure come from?

I did say that if anyone believed only half of the remaining population could be infected, then just halve my figures to get the appropriate estimate.


We don't know yet what percentage of any population is susceptible to Covid. Most models assume the worst case 100%

The 20-25% number is (presumably) based on those showing antibodies in a population (or a small sample from it). However if 10, or 20 or 50% of that population can't get infected, won't produce antibodies etc then the implications are massively different.

To keep the maths simple assume a population of 100 million. If 25% show antibodies then with 100% potential that's 25 million infected (most survive some die) and 75 million waiting for their infection. But if 50% will never be infected and produce antibodies, perhaps through prior exposure to other Coronaviruses and/or T-Cell response, then that 25%, or 25 million antibody number, actually represents 3/4 of the population (when added to the 50% unable to be impacted by Covid), not 1/4. As such the model might be over representing the "waiting" population by a factor of 3.

zico
Lemon Quarter
Posts: 2139
Joined: November 4th, 2016, 12:12 pm
Has thanked: 1074 times
Been thanked: 1086 times

Re: Coronavirus - Modelling Aspects Only

#392413

Postby zico » March 4th, 2021, 6:08 pm

Julian wrote:
zico wrote:...
What could be wrong with these above scenarios?
...
NB - The above assumptions don't allow for emergence of vaccine-resistant variants (either home-grown or imported) which I think is a pretty optimistic assumption.

One other thing you don’t allow for, understandably since we have no idea if it will happen and if it does then what its effect will be, is improvements in treatments lowering mortality rates. And on new variants it’s also at least possible that new less lethal variants could emerge although in a timescale of the next 6 to 12 month maybe that’s wishful thinking.

- Julian


Yes, I'm assuming current mortality rates. Obesity in a society looks to be a big driver of mortality rates, and in the 12 months since Covid emerged, treatments have become a lot better, but 1% still seems about right for UK society. I'd be pleasantly surprised if any major new breakthrough happened to reduce mortality through better treatment, that sort of thing only happens in blockbuster films.

There have been a lot of new variants emerging - potential changes for a virus are infectiousness, virulence, and vaccine-evasiveness. Mortality rates seem very similar for all variants so far. A very few variants have become more infectiousness than Wuhan-standard (these are likely to dominant, by virtue of being more infectious) and a worrisome few are more resistant to the vaccine (again, if Darwin is right, these will also become more dominant in a vaccinated world, or a vaccinated first-world to be more accurate).

zico
Lemon Quarter
Posts: 2139
Joined: November 4th, 2016, 12:12 pm
Has thanked: 1074 times
Been thanked: 1086 times

Re: Coronavirus - Modelling Aspects Only

#392417

Postby zico » March 4th, 2021, 6:13 pm

dealtn wrote:
We don't know yet what percentage of any population is susceptible to Covid. Most models assume the worst case 100%

The 20-25% number is (presumably) based on those showing antibodies in a population (or a small sample from it). However if 10, or 20 or 50% of that population can't get infected, won't produce antibodies etc then the implications are massively different.

To keep the maths simple assume a population of 100 million. If 25% show antibodies then with 100% potential that's 25 million infected (most survive some die) and 75 million waiting for their infection. But if 50% will never be infected and produce antibodies, perhaps through prior exposure to other Coronaviruses and/or T-Cell response, then that 25%, or 25 million antibody number, actually represents 3/4 of the population (when added to the 50% unable to be impacted by Covid), not 1/4. As such the model might be over representing the "waiting" population by a factor of 3.


Personally, I'm not convinced about the T-cells immunity hypothesis (for that's all it is, unless you can point me to actual evidence) because it doesn't seem to be supported by what's happened in populations. If it was mainly correct, then we would have seen "herd immunity" responses in some badly-affected areas by now. However, one of the worst affected cities, Manaus in Brazil has now had at least 2 very deadly and widespread outbreaks, so even they don't yet seem to have developed herd immunity.

I'm not saying T-cells and natural immunity doesn't happen, just that I haven't seen solid evidence for it anywhere.

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

Re: Coronavirus - Modelling Aspects Only

#392496

Postby scotia » March 4th, 2021, 9:28 pm

Another week of (English) data. See the notes at the end on the model data. To recap:- The Blue Points are the deaths by publish date, summed over the preceding week. The Red Points are the hospital admissions, summed over a week, multiplied by 0.3, and moved forward by 11 days - And these are renamed as being the projected deaths by publish date. The size of the vertical bars are the statistical standard deviations, assuming a Poisson distribution. However - back to the present and continuing good news which is probably vaccination related - and which makes the 0.3 factor look no longer realistic

Image

The deaths are continuing to fall well below the admissions curve - which last week I hypothesised as being due to the vaccination related reduction in deaths within Care Homes which did not significantly affect admissions. On checking various sources, I believe that approximately 75% of Covid-19 deaths in Care Home residents take place in the Care Home. Only 25% of them die in hospital. Last week I also plotted the data with the 0.3 factor reduced to 0.26 - and here is the updated display using 0.26

Image

And this displays even more good news on the deaths front - the factor of 0.26 is now also too large to fit the recent data. So in my final graph I have reduced it to 0.2.

Image

I await next week's data with interest.

Now - some more comments on the model, and where it is going. Its basis is described in
https://www.lemonfool.co.uk/viewtopic.php?f=98&t=22737&start=960#p359126
It relies on a period of stability when the hospitals are not overloaded, there is no significant change in the treatment of Covid-19, and the infectivity and mortality rate of the virus remain constant. This worked well on data from the end of September until the beginning of December with a factor of 0.265 and a time slip of 13 days - then the death rate multiplier appeared to increase (to 0.3) at a reduced time slip from admissions (11 days). Which factor caused this change? Overloaded hospitals? Oldies determined to ignore advice and do Christmas shopping? Or the new Kent virus variant? Take your pick! However we once again had a reasonable fit - up until about a month ago - when I'm reasonably sure that vaccinations (particularly in Care Homes) have had a significant effect. And the vaccination program will have a continuing and changing effect on the model parameters. Hence I would suggest that its predictive capabilities are currently rather limited. However the graphs may provide some insight into what is happening.

servodude
Lemon Half
Posts: 8271
Joined: November 8th, 2016, 5:56 am
Has thanked: 4435 times
Been thanked: 3564 times

Re: Coronavirus - Modelling Aspects Only

#392503

Postby servodude » March 4th, 2021, 10:06 pm

zico wrote:
dealtn wrote:
We don't know yet what percentage of any population is susceptible to Covid. Most models assume the worst case 100%

The 20-25% number is (presumably) based on those showing antibodies in a population (or a small sample from it). However if 10, or 20 or 50% of that population can't get infected, won't produce antibodies etc then the implications are massively different.

To keep the maths simple assume a population of 100 million. If 25% show antibodies then with 100% potential that's 25 million infected (most survive some die) and 75 million waiting for their infection. But if 50% will never be infected and produce antibodies, perhaps through prior exposure to other Coronaviruses and/or T-Cell response, then that 25%, or 25 million antibody number, actually represents 3/4 of the population (when added to the 50% unable to be impacted by Covid), not 1/4. As such the model might be over representing the "waiting" population by a factor of 3.


Personally, I'm not convinced about the T-cells immunity hypothesis (for that's all it is, unless you can point me to actual evidence) because it doesn't seem to be supported by what's happened in populations. If it was mainly correct, then we would have seen "herd immunity" responses in some badly-affected areas by now. However, one of the worst affected cities, Manaus in Brazil has now had at least 2 very deadly and widespread outbreaks, so even they don't yet seem to have developed herd immunity.

I'm not saying T-cells and natural immunity doesn't happen, just that I haven't seen solid evidence for it anywhere.


It's fuzzy isn't it?

I've been trying to get my head around it (and it helps me try to help the elder offspring with her prep for studying biomed)
- here's some bits I've encountered recently

https://academic.oup.com/cid/advance-article/doi/10.1093/cid/ciaa1803/6017810
Although it is possible that there are rare instances of individuals possessing antibodies from prior seasonal HCoV infection may be able to also target SARS-CoV-2 S, our data argue against a broad role for preexisting protective humoral immunity against SARS-CoV-2


https://www.bmj.com/content/370/bmj.m3563
Formally, the clinical implications of the pre-existing T cell reactivity remain an open question. “People say you don’t have proof, and they’re right,” says Buggert


https://www.nature.com/articles/s41577-020-00460-4
Cross-reactive T cell memory may or may not affect COVID-19 disease severity in individuals. However, given the implications of the possibility, it is worth considering specific immunological and epidemiological scenarios. Four distinct immunological scenarios were considered herein. In each of the three scenarios that we consider plausible, the reduction in viral spread potentially afforded by pre-existing immunity is already accounted for by the empirical observational data available and factored into epidemiological models of spread and herd immunity

- which is kind of where my enthusiasm for digging deeper petered out as that bit I've emboldened pinged something in my brain

From a practical point of view (don't blame me... it's my default) the T-cell cross immunity fog is already "baked in" to the uncertainties of the herd immunity threshold
- a lot of that fog that has been burnt off by the serology results in Manaus indicating an infection level around 70%
- and then again by its subsequent massive reinfection stats

In summary:
I don't doubt have any that cross-immunity exists but on the evidence in Amazonia it would appear to be in the order of noise


- sd

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

Re: Coronavirus - Modelling Aspects Only

#392705

Postby scotia » March 5th, 2021, 12:37 pm

Thanks for your analysis, although I think you are being brave in view of the multiple factors which are involved with large uncertainties.
Could I ask about one of your assumptions:-

zico wrote:Assuming 90% vaccination protection against death (based on latest analysis).


Could you direct me towards the latest sources? Back in the initial published tests on the effectiveness of the vaccines, vaccinated deaths were (nearly?) non existent - and certainly too few to give any statistically sensible figure.

zico
Lemon Quarter
Posts: 2139
Joined: November 4th, 2016, 12:12 pm
Has thanked: 1074 times
Been thanked: 1086 times

Re: Coronavirus - Modelling Aspects Only

#392773

Postby zico » March 5th, 2021, 3:02 pm

scotia wrote:Thanks for your analysis, although I think you are being brave in view of the multiple factors which are involved with large uncertainties.
Could I ask about one of your assumptions:-

zico wrote:Assuming 90% vaccination protection against death (based on latest analysis).


Could you direct me towards the latest sources? Back in the initial published tests on the effectiveness of the vaccines, vaccinated deaths were (nearly?) non existent - and certainly too few to give any statistically sensible figure.


Not that brave - I've just made the model simple to be more comprehensible, and so that others can simply input their assumptions. I think of it as more a scenario tool (if a,b,c happens we'll get x,y,z) rather than a serious attempt to provide an exact figure.

90% vaccination is from page 2 of the link below - under "Results" they estimate the combined effectiveness of preventing death at 85% with a single dose, so I've assumed 90% from the 2-dose regime. (I've assumed their alphabet soup names for 2 different vaccines are generally known as Pfizer & AstraZeneca). I've picked 90% rather than 85% because all that I've read says there'll be a bigger effect from 2 doses, and also longer-lasting.

The second link is from Scotland which gives for Pfizer and Astrazeneca respectively, an 85% (76%-91%) or 94% (73%-99%) reduced risk of hospitalisation. Figures in brackets are confidence intervals - note that they are pretty wide, particularly for the AZ vaccine.
(I remember reading/hearing that Astrazeneca gives 99% protection against death and wondering where that came from - I reckon it came from the upper confidence interval in the brackets above, and the Scotland study is hospitalisation, not death. Intuitively I'd expect the protection against death to be even better than that from hospitalisation, but that's just a guess and not backed up by evidence. The death % might be worse depending on admissions policy in Jan/Feb (when hospital resources were stretched).

https://khub.net/documents/135939561/43 ... 4617945615

https://www.ed.ac.uk/files/atoms/files/ ... eprint.pdf

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

Re: Coronavirus - Modelling Aspects Only

#393014

Postby scotia » March 6th, 2021, 10:25 am


Thanks for the links. I'm trying hard to digest the English paper. Could I be allowed to suggest there is much (possibly unintentional) technical obfuscation?

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

Re: Coronavirus - Modelling Aspects Only

#393027

Postby dealtn » March 6th, 2021, 10:52 am

scotia wrote:

Thanks for the links. I'm trying hard to digest the English paper. Could I be allowed to suggest there is much (possibly unintentional) technical obfuscation?


Well I guess it's true they weren't prepare for the "lay person" as audience. However the conclusion is in pretty accessible language. (Substitute BNT162b2 with Pfizer/BionTech and ChAdOx1 with Oxford/AstraZeneca)

Conclusion

Vaccination with either a single dose of BNT162b2 or ChAdOx1 COVID-19 vaccination was associated with a significant reduction in symptomatic SARS-CoV2 positive cases in older adults with even greater protection against severe disease. Both vaccines show similar effects. Protection was maintained for the duration of follow-up (>6 weeks). A second dose of BNT162b2 provides further protection against symptomatic disease but second doses of ChAdOx1 have not yet been rolled out in England. There is a clear effect of the vaccines against the UK variant of concern.

Julian
Lemon Quarter
Posts: 1385
Joined: November 4th, 2016, 9:58 am
Has thanked: 532 times
Been thanked: 676 times

Re: Coronavirus - Modelling Aspects Only

#393079

Postby Julian » March 6th, 2021, 12:43 pm

dealtn wrote:
scotia wrote:

Thanks for the links. I'm trying hard to digest the English paper. Could I be allowed to suggest there is much (possibly unintentional) technical obfuscation?


Well I guess it's true they weren't prepare for the "lay person" as audience. However the conclusion is in pretty accessible language. (Substitute BNT162b2 with Pfizer/BionTech and ChAdOx1 with Oxford/AstraZeneca)

Conclusion

Vaccination with either a single dose of BNT162b2 or ChAdOx1 COVID-19 vaccination was associated with a significant reduction in symptomatic SARS-CoV2 positive cases in older adults with even greater protection against severe disease. Both vaccines show similar effects. Protection was maintained for the duration of follow-up (>6 weeks). A second dose of BNT162b2 provides further protection against symptomatic disease but second doses of ChAdOx1 have not yet been rolled out in England. There is a clear effect of the vaccines against the UK variant of concern.

Ironic, after Macron's idiotic comment that the AZ vaccine was "quasi-ineffective in the over 65 year olds"(*), to see the AZ vaccine actually demonstrating better efficacy that the Pfizer vaccine in preventing symptomatic infections in the >= 70s after a single dose, plateauing at 61% for Pfizer vs 73% for Oxford/AZ.

Unfortunately 2nd dose data isn't available for the Oxford/AZ vaccine yet, and even when it is I suspect any comparison with the Pfizer vaccine could be called into question since there is some pre-existing clinical trial data suggesting that the AZ vaccine's effectiveness might well be enhanced by extending the dosing interval whereas it is at least possible that the post-2nd-dose efficacy of the Pfizer vaccine was compromised by extending the dosing interval. Even on the basis of that first-dose comparison though it says to me just how irresponsible, unhelpful and ignorant Macron's unscientific comments were.

- Julian

(*) To single out just one person although in my view one of the most egregious of the actors in the anti-Oxford/AZ saga, and in fairness also assuming that his quote was accurately translated.

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

Re: Coronavirus - Modelling Aspects Only

#393235

Postby scotia » March 7th, 2021, 1:14 am

zico wrote:I've tried doing some simple modelling for future UK deaths using the following modelling assumptions.

As I mentioned I have been trying to extract from the reference you supplied, the effect on mortality of covid-19 in vaccinated and non-vaccinated persons.
https://khub.net/documents/135939561/430986542/Early+effectiveness+of+COVID+vaccines.pdf/ffd7161c-b255-8e88-c2dc-88979fc2cc1b?t=1614617945615
As far as I can determine, this information only appears in Table 5, and the text immediately above it. There is also a discussion of Table 5 above Table 4 - ignore it, it looks like a typo, and should be the discussion of Table 4, which does not cover deaths.
Table 5 relates to the 80+ age group only, and only to the Pfizer vaccine. Taken at its face value, the unvaccinated death total is 1063 in 8096 infections (13.1%), whereas the vaccinated death total (14 days or more after vaccination), is 51 in 750 infections (6.8%). If we divide 6.80 by 13.13 we get 0.518. When "survival analyses" is applied the "hazard ratio" drops to 0.49, with a range (95% confidence?) of 0.38 to 0.63. The vaccination is stated to provides an additional 51% protection against death - presumably using the 0.49 "hazard ratio". In the later Discussion section on the Pfizer vaccine, this figure of 51% is quoted for the entire age range (70+) under study, without any further justification.
I'm aware that the mortality rate is strongly dependent on age, but I was surprised at how high the values were in Table 5 - for both (Pfizer) vaccinated and unvaccinated deaths. I was also surprised at how small the additional 51% protection was.

Gersemi
Lemon Slice
Posts: 492
Joined: November 4th, 2016, 3:57 pm
Has thanked: 535 times
Been thanked: 222 times

Re: Coronavirus - Modelling Aspects Only

#393260

Postby Gersemi » March 7th, 2021, 9:37 am

scotia wrote:
zico wrote:I've tried doing some simple modelling for future UK deaths using the following modelling assumptions.

As I mentioned I have been trying to extract from the reference you supplied, the effect on mortality of covid-19 in vaccinated and non-vaccinated persons.
https://khub.net/documents/135939561/430986542/Early+effectiveness+of+COVID+vaccines.pdf/ffd7161c-b255-8e88-c2dc-88979fc2cc1b?t=1614617945615
As far as I can determine, this information only appears in Table 5, and the text immediately above it. There is also a discussion of Table 5 above Table 4 - ignore it, it looks like a typo, and should be the discussion of Table 4, which does not cover deaths.
Table 5 relates to the 80+ age group only, and only to the Pfizer vaccine. Taken at its face value, the unvaccinated death total is 1063 in 8096 infections (13.1%), whereas the vaccinated death total (14 days or more after vaccination), is 51 in 750 infections (6.8%). If we divide 6.80 by 13.13 we get 0.518. When "survival analyses" is applied the "hazard ratio" drops to 0.49, with a range (95% confidence?) of 0.38 to 0.63. The vaccination is stated to provides an additional 51% protection against death - presumably using the 0.49 "hazard ratio". In the later Discussion section on the Pfizer vaccine, this figure of 51% is quoted for the entire age range (70+) under study, without any further justification.
I'm aware that the mortality rate is strongly dependent on age, but I was surprised at how high the values were in Table 5 - for both (Pfizer) vaccinated and unvaccinated deaths. I was also surprised at how small the additional 51% protection was.


The additional protection does seem low, but I note that the death rates are as a percentage of infections. We would hope that the vaccines would prevent any infection in a reasonable percentage of people, so we need some measure of infection rate between vaccinated and unvaccinated people to get the whole picture.

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

Re: Coronavirus - Modelling Aspects Only

#393266

Postby dealtn » March 7th, 2021, 9:56 am

dealtn wrote:
scotia wrote:

Thanks for the links. I'm trying hard to digest the English paper. Could I be allowed to suggest there is much (possibly unintentional) technical obfuscation?


Well I guess it's true they weren't prepare for the "lay person" as audience. However the conclusion is in pretty accessible language. (Substitute BNT162b2 with Pfizer/BionTech and ChAdOx1 with Oxford/AstraZeneca)

Conclusion

Vaccination with either a single dose of BNT162b2 or ChAdOx1 COVID-19 vaccination was associated with a significant reduction in symptomatic SARS-CoV2 positive cases in older adults with even greater protection against severe disease. Both vaccines show similar effects. Protection was maintained for the duration of follow-up (>6 weeks). A second dose of BNT162b2 provides further protection against symptomatic disease but second doses of ChAdOx1 have not yet been rolled out in England. There is a clear effect of the vaccines against the UK variant of concern.


Better than anything I could do, by Dr John Campbell

https://www.youtube.com/watch?v=U3otWaBrvEc

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

Re: Coronavirus - Modelling Aspects Only

#393306

Postby scotia » March 7th, 2021, 12:26 pm

dealtn wrote:Better than anything I could do, by Dr John Campbell

https://www.youtube.com/watch?v=U3otWaBrvEc

Excellent summary for the lay person. Its a pity he swallowed the precise 51% figure from the paper's conclusions. I think it would have been preferable for him to have said that the additional protection against death was around 40% to 65% in the over 80s. The spread is almost entirely due to the poor statistical accuracy on the vaccinated deaths number - two standard deviations (95% confidence) is about plus or minus 14 in 51. And this statistical accuracy will only get worse at lower age groups, with significantly lower death rates. So I think its unlikely we will get an accurate estimate of the vaccine's protection against death in the overall community.
And a word of caution on your earlier stated reliance on the paper's conclusions. If you are acting as a referee for a scientific paper it is necessary to work through all of the text - no matter how indigestible it may be, and check if its conclusions are accurate.


Return to “Coronavirus Discussions”

Who is online

Users browsing this forum: No registered users and 5 guests