The maths of it all and what it means to Ireland

bb1234567

Seweryn wrote: »

Well... that is basic mathematics. Do we really need to go through that?


As I said above, I have no interest in spreading numbers taken from a pie in the sky, but I will not publish an email here. If you do not believe in the base numbers, I will not try to convince you.

You are trying to convince us though. Tell me do you ever believe uncited information just because the person telling you the info says so in a convincing way ? For your sake Id hope the answer is no

Seweryn

bb1234567 wrote: »

You are trying to convince us though. Tell me do you ever believe uncited information just because the person telling you the info says so in a convincing way ? For your sake Id hope the answer is no

No, and that is why I wanted to obtain the base numbers and then do my own analysis rather than listening or reading official publications that are not always accurate. I did the same for Sweden in early May, but their statistic office has actually daily recorded deaths published in an excel file starting from 1st Jan 2015, all in one document, so very convenient. Here in Ireland the accurate death rate data base is not published and you need to to ask for it.

Anyway, I am not trying to convince anyone nor to spread false numbers. Even if I publish the base numbers, someone may ask if I counted the deaths myself...

martingriff

Seweryn wrote: »

Well... that is basic mathematics. Do we really need to go through that?


As I said above, I have no interest in spreading numbers taken from a pie in the sky, but I will not publish an email here. If you do not believe in the base numbers, I will not try to convince you.

So this person who happens to email you how do you know them, did you request the figures. You could black out the names in the email and scan and link it in. I thought if the GRO were to give figures to people who asked for them they would give it in an Excel sheet or a PDF otherwise its sounds amateurish

Seweryn

martingriff wrote: »

So this person who happens to email you how do you know them, did you request the figures.

No, I do not know them personally.

martingriff wrote: »

You could black out the names in the email and scan and link it in. I thought if the GRO were to give figures to people who asked for them they would give it in an Excel sheet or a PDF otherwise its sounds amateurish

Yes, I was surprised as well. The numbers are in the email body. Anyway, I will cut a small bit with the latest April numbers update as of today, and will leave it at that. As I said I have no interest in either publishing false numbers or in trying to prove them to be correct.

bb1234567

Seweryn wrote: »

No, I do not know them personally.


Yes, I was surprised as well. The numbers are in the email body. Anyway, I will cut a small bit with the latest April numbers update as of today, and will leave it at that. As I said I have no interest in either publishing false numbers or in trying to prove them to be correct.

Umm so why do you have deaths in April 2020 as 2787 in your earlier spread sheet when your source says April 2020 deaths are over 3000?

Seweryn

bb1234567 wrote: »

Umm so why do you have deaths in April 2020 as 2787 in your earlier spread sheet when your source says April 2020 deaths are over 3000?

Because that email was received today with the numbers updated - see my updated table as of today with 3015 for April in the table. The earlier number of 2787 is a figure for April which was up to date a couple of weeks ago. The final figure of 3015 as of today may rise by say another 20 or 50 in a few weeks time. But that will not make a significant difference.

ShineOn7

This post from Seamus in the main daily thread a couple of days ago was an excellent analysis I thought

Just to clarify, it's incredibly unlikely that this vaccine would give you narcolepsy specifically. That came with one particular brand of flu vaccine, where other brands that vaccinated for the same virus didn't cause narcolepsy. So they have a pretty good grip on why it happened in that case.

The rate was also very low - just over 3 cases per 100,000 vaccinations. The media created a perception that all of the swine flu vaccines caused narcolepsy, caused it in large numbers, and was due to the vaccine being rushed. And none of these things are true.

Of course, it's narcolepsy. It's horrible. There was a 0.003% chance of developing narcolepsy from a shot, and (in hindsight) a 0.00042% chance of contracting and dying from swine flu. So I'll forgo the vaccine, thanks.

We should absolutely be wary of a quickly-developed vaccine. To cover as many bases as we can. But we should also take all of the data together to make informed decisions.

Given today's figures, about 0.5% of the Irish population have contracted covid. And the fatality rate is 7%.

That means that your chances of contracting and dying from covid are 0.035%. Which itself is very low.

But if the odds of developing <insert disabling disease here> from the vaccine are 0.003%, then on balance it's a decent payoff. If you do nothing you're ten times more likely to die from covid, than you are to contract <something> if you get the vaccine.

Thoughts on the bit in bold above?

Blut2

The bolded part is absolutely awful analysis. Do you really think every single case of corona in Ireland has been caught by testing? Because thats what the bloded text suggests.

The fatality rate for corona in many many closed environments studies has averaged at 0.38% [1]. Which applied to the Irish figure of 1,754 deaths would suggest we've had approx 450,000 cases - or approaching 10% of the population.

[1] https://docs.google.com/spreadsheets/d/1zC3kW1sMu0sjnT_vP1sh4zL0tF6fIHbA6fcG5RQdqSc/htmlview?pru=AAABchM1MrM*aJ5Rxb9kousGZwDSuLVUvQ#gid=0

Which means the odds of actually contracting and dying from covid are even lower again, by a large large margin.

bb1234567

Blut2 wrote: »

The bolded part is absolutely awful analysis. Do you really think every single case of corona in Ireland has been caught by testing? Because thats what the bloded text suggests.

The fatality rate for corona in many many closed environments studies has averaged at 0.38% [1]. Which applied to the Irish figure of 1,754 deaths would suggest we've had approx 450,000 cases - or approaching 10% of the population.

[1] https://docs.google.com/spreadsheets/d/1zC3kW1sMu0sjnT_vP1sh4zL0tF6fIHbA6fcG5RQdqSc/htmlview?pru=AAABchM1MrM*aJ5Rxb9kousGZwDSuLVUvQ#gid=0

Which means the odds of actually contracting and dying from covid are even lower again, by a large large margin.

I dont know if we can estimate that based on our deaths given how many were in nursing homes. I think for the purposes of estimating community spread, deaths in nursing homes should be largely ommitted, they massively skew the infection estimate if you are basing it on death rates.

Say you have two countries of similar size of 1 million, and both had outbreaks that infected 10% of their populations. But one country didn't protect nursing homes and the other did. They might have the same number of community deaths but obviously one country will have much higher death rates because nursing home residents are so vulnerable, despite a relatively similar percent of the population being infected as the other country as people in nursing homes do not make up a big percent of the population, but if there are some infections in this group, it will lead to many deaths. And if we are talking about 1 death generally suggesting the potential of about 200 other infections in the community, in the estimate above, this concentration of cases, and hence deaths, in this vulnerably group due to negligence will grossly inflate the estimate of prevalence of the virus int he community. So these deaths are falsely paintaing a picture of widespread transmission when the virus may not have spread widely at all in the general population, it's just unfortunately spreading easily among a very very vulnerable group in a closed facility, and the inevitably high death toll will then wrongly suggest high rates of transmission among thousands of healthy/younger people if we use the method above of equating 1 deaths to roughly 250 other infections on average.

So, community spread is still the same in both countries overall ,despite the disparity in death rates.

Sorry, don't know if my point is clear. It's not that nursing home deaths aren't representative of community spread at all,if it's getting into many nursing homes it probably has spread widely.But it's just that if a country has had a lot of nursing home deaths, if they make up half of their deaths overall, it's a really poor means of estimating how many people in the community at large have got it, 99% of whom don't reside in nursing homes.

If you take out our nursing home deaths, the estimate would then be roughly 5% of the population, which is exactly what the recent antibody testing has shown.

Hmmzis

Blut2 wrote: »

The bolded part is absolutely awful analysis. Do you really think every single case of corona in Ireland has been caught by testing? Because thats what the bloded text suggests.

The fatality rate for corona in many many closed environments studies has averaged at 0.38% [1]. Which applied to the Irish figure of 1,754 deaths would suggest we've had approx 450,000 cases - or approaching 10% of the population.

[1] https://docs.google.com/spreadsheets/d/1zC3kW1sMu0sjnT_vP1sh4zL0tF6fIHbA6fcG5RQdqSc/htmlview?pru=AAABchM1MrM*aJ5Rxb9kousGZwDSuLVUvQ#gid=0

Which means the odds of actually contracting and dying from covid are even lower again, by a large large margin.

The preliminary serology data would put Ireland's infection rate at just under 5%, so that would mean our approximate infection count would be approaching 240,000 with an all age group IFR of 0.7% or thereabouts.

DeVore

It really depends on what you want to know from the numbers.

If you want to compare us with other infections/diseases CFR then they need to be on the same basis. Usually a CFR is "Number of cases known about" / "Number of deaths". There is no attempt to go out and find every single infected person, only the ones that come to light. For example, HIV, there is no widespread testing of the populace for HIV so there are plenty of cases we dont know about but we still have a CFR for it.

If you want to compare countries then again, they need to be using the same CFR calc and additionally the same reporting regime (all countries seem to have different ways of deciding if a death is "Covid" related or not.

If you want to know "what are my chances of getting Covid and if I do what are my chances of dying from it" then thats a different question altogether and it VERY much depends on your personal circumstantces. Age, medical history, job, attitude to masks etc etc.



Finally, I've been trying to convey something to everyone. "Not Dying" is about as low a bar as you can find for "success". Its only one of a few different outcomes, some of which are quite negative and SHOULD be calculated in if we are talking about cost/benefit analysis.

Blut2

Blut2 wrote: »

The bolded part is absolutely awful analysis. Do you really think every single case of corona in Ireland has been caught by testing? Because thats what the bloded text suggests.

The fatality rate for corona in many many closed environments studies has averaged at 0.38% [1]. Which applied to the Irish figure of 1,754 deaths would suggest we've had approx 450,000 cases - or approaching 10% of the population.

[1] https://docs.google.com/spreadsheets/d/1zC3kW1sMu0sjnT_vP1sh4zL0tF6fIHbA6fcG5RQdqSc/htmlview?pru=AAABchM1MrM*aJ5Rxb9kousGZwDSuLVUvQ#gid=0

Which means the odds of actually contracting and dying from covid are even lower again, by a large large margin.

Initial serology tests suggest its about 5%. Or a bit under 250k cases. And aligning with where the latest data seems to be pointing around the globe - 0.65% IFR. This is a number we will never know with certainty though.

For contrast it is about 0.1% for Flu. But this is based on an estimate on the total number of cases of Flu, taking confirmed cases and adding reports of ILI's from medical professionals where no confirmatory test has been completed. This does not however include the cases of Flu where the sufferer just goes to bed for a few days, and the real IFR is likely lower again.

ShineOn7

Deleted User wrote: »

Initial serology tests suggest its about 5%. Or a bit under 250k cases. And aligning with where the latest data seems to be pointing around the globe - 0.65% IFR. This is a number we will never know with certainty though.

For contrast it is about 0.1% for Flu. But this is based on an estimate on the total number of cases of Flu, taking confirmed cases and adding reports of ILI's from medical professionals where no confirmatory test has been completed. This does not however include the cases of Flu where the sufferer just goes to bed for a few days, and the real IFR is likely lower again.

Yes I think, given the reasons you listed, that Flu's IFR is actually a lot lower than 0.1%

Making Covid six times more deadly at the absolute minimum

Blut2

ShineOn7 wrote: »

Yes I think, given the reasons you listed, that Flu's IFR is actually a lot lower than 0.1%

Making Covid six times more deadly at the absolute minimum

That depends entirely on the year of the flu in question. Its IFR varies significantly depending on the strain. Some of the more deadly years in recent history have killed millions of people - the Hong Kong flu of 1968/69 killed up to 4 million for example. A number covid19 is still nowhere near - despite the world population then being less than half of what it is now. Covid19 would have to be approaching 8 million deaths worldwide by this time next year to be considered as deadly.

ShineOn7

Blut2 wrote: »

That depends entirely on the year of the flu in question. Its IFR varies significantly depending on the strain. Some of the more deadly years in recent history have killed millions of people - the Hong Kong flu of 1968/69 killed up to 4 million for example. A number covid19 is still nowhere near - despite the world population then being less than half of what it is now. Covid19 would have to be approaching 8 million deaths worldwide by this time next year to be considered as deadly.

Oh come on now

We're talking standard Flu to compare to here

Not a feckin' outlier mental Hong Kong flu that wiped out 4 million people over 50 years ago

bb1234567

Blut2 wrote: »

That depends entirely on the year of the flu in question. Its IFR varies significantly depending on the strain. Some of the more deadly years in recent history have killed millions of people - the Hong Kong flu of 1968/69 killed up to 4 million for example. A number covid19 is still nowhere near - despite the world population then being less than half of what it is now. Covid19 would have to be approaching 8 million deaths worldwide by this time next year to be considered as deadly.

Worldwide yes, because large countries like China managed to effectively stop an outbreak in 2020 but could not during that period of history. But in hotspots such as USA, COVID will soon have killed more than those pandemics did, even when adjusting for current population size.100,000 died in the states in the Hong Kong flu outbreak when the US population was 175 million, that was over the course of two years btw. USA will hit 180,000 in early Autumn, so by then it will have more deaths than that pandemic caused, in about 1/4 of the time. Excess deaths also suggest massive underreporting of COVID deaths in the United states currently, by as much as 30%.

Likewise 33,000 Brits died in that pandemic, COVID has already far exceeded that death toll in about a 3 month period, even when adjusting for population size.

So again, if COVID was unmitigated worldwide as those pandemics were the result would be far worse.
I also don't know where you've found that 4 million figure, almost all sources I've seen online commonly cite a figure of around 1 million deaths for both the Asian flu and Hong Kong flu pandemics. 4 million may be an absolute upper bound esimtate which you are deceivingly using to bolster your argument, but nonetheless if you are using that figure, it was certainly over the course of a two year period, not one, so come back in 18 months to judge properly .

DeVore

Yeah, that level of wilfully allowing someone to conflate "flu" with "really terrible, one time flu 50 years ago" is so borderline that saying "its the same as the flu*" is basically lying.


*really terrible, one time flu 50 years ago

Blut2

bb1234567 wrote: »

Worldwide yes, because large countries like China managed to effectively stop an outbreak in 2020 but could not during that period of history. But in hotspots such as USA, COVID will soon have killed more than those pandemics did, even when adjusting for current population size.100,000 died in the states in the Hong Kong flu outbreak when the US population was 175 million, that was over the course of two years btw. USA will hit 180,000 in early Autumn, so by then it will have more deaths than that pandemic caused, in about 1/4 of the time. Excess deaths also suggest massive underreporting of COVID deaths in the United states currently, by as much as 30%.

Likewise 33,000 Brits died in that pandemic, COVID has already far exceeded that death toll in about a 3 month period, even when adjusting for population size.

So again, if COVID was unmitigated worldwide as those pandemics were the result would be far worse.
I also don't know where you've found that 4 million figure, almost all sources I've seen online commonly cite a figure of around 1 million deaths for both the Asian flu and Hong Kong flu pandemics. 4 million may be an absolute upper bound esimtate which you are deceivingly using to bolster your argument, but nonetheless if you are using that figure, it was certainly over the course of a two year period, not one, so come back in 18 months to judge properly .

I'm not really sure what you're googling, the number one result on Google gives an up to 4 million deaths figure for me:

https://www.britannica.com/event/1968-flu-pandemic

Or if you want something more academic: Paul, William E. (2008). Fundamental Immunology. p. 1273.

Its not just Hong Kong flu as an extreme example, either. The 1957–1958 flu for example killed up to 1 million people globally at a time when the world population was under 3 billion.

Western death figures are obviously going to be much high now, even adjusting for population increases, given our population is much much older than it was 60 years ago - so far far more vulnerable to infection/death.

My overall point is corona's death rate isn't that unusual, historically speaking. Its numbers are still increasingly, but it very much remains to be seen if in 50 years time we'll be assessing it along the lines of a very bad flu year along the lines of 1958 or 1968 or not. It still has a long, long way to go.

bb1234567

Blut2 wrote: »

I'm not really sure what you're googling, the number one result on Google gives an up to 4 million deaths figure for me:

https://www.britannica.com/event/1968-flu-pandemic

Or if you want something more academic: Paul, William E. (2008). Fundamental Immunology. p. 1273.

Its not just Hong Kong flu as an extreme example, either. The 1957–1958 flu for example killed up to 1 million people globally at a time when the world population was under 3 billion.

Western death figures are obviously going to be much high now, even adjusting for population increases, given our population is much much older than it was 60 years ago - so far far more vulnerable to infection/death.

My overall point is corona's death rate isn't that unusual, historically speaking. Its numbers are still increasingly, but it very much remains to be seen if in 50 years time we'll be assessing it along the lines of a very bad flu year along the lines of 1958 or 1968 or not. It still has a long, long way to go.

Right, emphasis on up to. It could equally be 'as low as' 1 million. For the sake of the argument we would probably assume a figure of around 2.5 million. Or is there a reason you're only mentioning the upper bound estimate as if it is the most likely estimate?
If you are using that figure, then you compare like with like. An upper bound estimate of the current global COVID death toll is in the region of one million, based on excess deaths in nations globally.
https://www.ft.com/content/6bd88b7d-3386-4543-b2e9-0d5c6fac846c
Of course in reality, some of those deaths were not COVID related. But we don't know that yet. But likewise, it is most likely the Hong Kong flu did not kill 4 million people, it is more than likely in the region of 2 and 1/2 million based on the estimate provided. So you compare like with like .

But yes I agree with some of your other points, it may not turn out to be a defining pandemic in history apart from the social impacts the reactions had, could well be the case. Though overall I would say based on the deaths caused in areas where it is unmitigated, this does not look to be the case, and will be markedly worse than those previous pandemics in the middle of the century, even with a constant level of containment and control attempted within most regions of the world.

McGiver

Deleted User wrote:

For contrast it is about 0.1% for Flu. But this is based on an estimate on the total number of cases of Flu, taking confirmed cases and adding reports of ILI's from medical professionals where no confirmatory test has been completed. This does not however include the cases of Flu where the sufferer just goes to bed for a few days, and the real IFR is likely lower again.

Real IFR for influenza is 0.04% (not 0.1%) if you include an estimated number of asymptomatic or unreported cases which are thought to be in the region of 50-75%.

wait4me

Philip Nolan put up a series of graphs on Twitter this morning. Anybody know where these can be found online? I've tried the HPSC and HSE websites but cannot find them there. Can anybody give me a steer?.

ACitizenErased

wait4me wrote: »

Philip Nolan put up a series of graphs on Twitter this morning. Anybody know where these can be found online? I've tried the HPSC and HSE websites but cannot find them there. Can anybody give me a steer?.

Those are the modelling team’s graphs, Prof. Nolan’s twitter would be the only one with them until the modelling press conference on Thursday

wait4me

Ah okay. Do you know if the Modelling Team shares them anywhere online?

ACitizenErased

wait4me wrote: »

Ah okay. Do you know if the Modelling Team shares them anywhere online?

The only place is twitter as far as i know

Beasty

Threads merged

Beasty

It's easy enough to plug the figures from the sticky of the Ireland numbers into a spreadsheet

Unfortunately the published figures do not accurately reflect the dates of testing, but even if they did that would not accurately reflect the date of infection

The use of 5 day rolling averages is a very poor way of doing this due to the weekend distortions. 7 or 14 day rolling averages are much more meaningful

Coyote

Hi bb1234567

I don't believe they share there models
but there is other places that have interactive data
Covid-19 Charts
you can pick what places you want to show and a number of other options
https://corona.njoyard.fr/?r=BG-FR-GR-GC-IE-IT-SP-UI-SH&x=start&ya=true&yc=true&yl=true&yr=true


91-DIVOC
this shows countries and US States for total and cases/deaths per million
http://91-divoc.com/pages/covid-visualization/



and if you want to look at a Epidemic Calculator
the first link is to the default setting
the second link is for one that i did some adjustments to match a bit closer to what Ireland numbers looked like. but you would need to update it each day which the calculator is not able to do
it allows you to adjust all the different options at the bottom of the screenshot
https://gabgoh.github.io/COVID/
https://gabgoh.github.io/COVID/?CFR=0.0597&D_hospital_lag=7.92&D_incbation=5.2&D_infectious=1.95&D_recovery_mild=11.1&D_recovery_severe=28.6&I0=40&InterventionAmt=0.21999999999999997&InterventionTime=18.13333333333333&P_SEVERE=0.2&R0=3.4&Time_to_death=32&logN=15.54



and if you want to get right in to the data sets
COVID-19 Dataset: 2019 Novel Coronavirus COVID-19 (2019-nCoV) Data Repository by Johns Hopkins CSSE
https://github.com/CSSEGISandData/COVID-19/tree/master/csse_covid_19_data/csse_covid_19_time_series
Coyote

ShineOn7

5 months on since this thread began, and are we seeing any changes in CFR and IFR etc in Ireland?

This post from Thierry12 just caught my eye

If that's correct and we had say 1400 cases in 2 weeks, 4 in icu, 20 in hospital

0.28% case ICU rate, 1.42% case hospital rate and lets say 50% of the ICU patients die thats a 0.14% fatality rate

5 million people @ 0.14% IFR

About 7,000 people will die if its lets run wild with no immunity

Question: can anyone mathematically work out the hospitality rates, CFR and IFR for the under 40s and under 50s with no underlying conditions (two separate calculations) as they stand now?

ShineOn7

ShineOn7 wrote: »

5 months on since this thread began, and are we seeing any changes in CFR and IFR etc in Ireland?

This post from Thierry12 just caught my eye




Question: can anyone mathematically work out the hospitality rates, CFR and IFR for the under 40s and under 50s with no underlying conditions (two separate calculations) as they stand now?

Thierrys post isn’t maths. Based on his calculated IFR, 25% of the population would have had to had contacted the virus, which we know is not true. Also the assumptions on ICU and hospital admissions only count the increase and doesn’t allow for discharges therefore undercounts.

I don’t think there is enough publicly available data to answer your question on the under 40’s and under 50’s, however most data throughout the pandemic has converged on Covid doubling your risk of death in a given year, which for someone in their 40’s is about 0.2%. Obviously this includes some with pre existing conditions etc etc, so would imagine for the average healthy person in their 40’s, the risk due to Covid would be <<0.1%

Anyone know where the raw data is available from the Covid dashboard?