The maths of it all and what it means to Ireland

yosemitesam1

ShineOn7 wrote: »

But just upto a few days ago I thought the consensus was that upto 50% was asymptomatic

This changes the mortality rate hugely

And also changes the hospitalisation rate massively

wakka12

wakka12 wrote: »

That is many more than originally thought, so hardly a bad thing?
The studies in Santa Clara and LA showed 2-5%, they are very urban areas, when you take into account rural and suburban areas then WHO's estimate of around 3% is probably bang on

The significance of the Californian findings isn't in the absolute numbers it's in how large they are compared to confirmed cases at the time.

There may be issues with the research (not for me to judge) but the conclusion is that prevalence is more like 50x than 10x confirmed cases. Might be a reflection on testing also of course.

But there are now a few findings that suggest many more people than we think may have got it - but they are tentative and of course don't say anything about immunity or how long-lasting it is.

The WHO guess is very conservative but it is their job to be cautious

wakka12

Deleted User wrote: »

The significance of the Californian findings isn't in the absolute numbers it's in how large they are compared to confirmed cases at the time.

There may be issues with the research (not for me to judge) but the conclusion is that prevalence is more like 50x than 10x confirmed cases. Might be a reflection on testing also of course.

But there are now a few findings that suggest many more people than we think may have got it - but they are tentative and of course don't say anything about immunity or how long-lasting it is.

The WHO guess is very conservative but it is their job to be cautious

I dont think its very conservative at all, in fact saying 3% of the entire world has been infected is very bold given that mass testing in iceland, Russia, UAE and South Korea has uncovered a relatively low level of the virus within the community.With only 1-2% of tests coming back positive in all of these countries, among hundreds of thousands up to millions of tests. It probably means WHO imagine a very large percentage of the population infected in epicentres like Spain, Italy, Belgium,UK, New York

normanoffside

If they were only testing for Antibodies in that Californian study would it not mean they have already recovered?
That means they had it maybe up to 2-3 weeks before.

The Antibody studies wouldn't be picking up the people currently infected at the time of the tests and given that blood samples were taken almost 3 weeks ago we can surely assume that a lot more now are infected or have already recovered.

ShineOn7

normanoffside wrote: »

If they were only testing for Antibodies in that Californian study would it not mean they have already recovered?
That means they had it maybe up to 2-3 weeks before.

The Antibody studies wouldn't be picking up the people currently infected at the time of the tests and given that blood samples were taken almost 3 weeks ago we can surely assume that a lot more now are infected or have already recovered.

I was thinking the same. There's so many variables to this

To illustrate how difficult modelling can be, especially with incomplete or inaccurate data, putting the current Irish data into a crude model we will be at zero new cases a day by May 11th topping out at 26,000, but will be at 20,000 deaths per day by November. I think my model needs a bit more work

ShineOn7

Deleted User wrote: »

To illustrate how difficult modelling can be, especially with incomplete or inaccurate data, putting the current Irish data into a crude model we will be at zero new cases a day by May 11th topping out at 26,000, but will be at 20,000 deaths per day by November. I think my model needs a bit more work

Yes.

Yes it does.

Cordell

Deleted User wrote: »

To illustrate how difficult modelling can be, especially with incomplete or inaccurate data, putting the current Irish data into a crude model we will be at zero new cases a day by May 11th topping out at 26,000, but will be at 20,000 deaths per day by November. I think my model needs a bit more work

Wow, you think?
Your model is wrong even considering incomplete data, it looks like you're modelling new cases and deaths independently, which is wrong no matter how incomplete the data is.

Cupatae

hmmm id question alot of these models... some of the numbers seem miles off alot with missing data... hmm ill crunch the numbers and post a more detailed post

normanoffside

Cupatae wrote: »

hmmm id question alot of these models... some of the numbers seem miles off alot with missing data... hmm ill crunch the numbers and post a more detailed post

You've changed your tune.
I thought you said that people crunching numbers are weird?

Cupatae wrote: »

The bottom line is we (People on boards) dont need to know most of the statistics. I dont get why people obsess over them, its weird tbh.

Cordell

Cordell wrote: »

Wow, you think?
Your model is wrong even considering incomplete data, it looks like you're modelling new cases and deaths independently, which is wrong no matter how incomplete the data is.

Thats my next step...
Even so, the case data from 29th March to 18th April is incorrect looking at the data visually. From 29th to the 9th, daily trend in case numbers deviates down significantly from what you would expect based on prior data. This is likely due to the testing backlog. After the 9th it swings the opposite way as German results come in. Only for the 19th and 20th does data appear to once again be conforming to some semblance of a pattern. Without knowing where the excess should filter back into date wise, it makes modelling difficult. Once we get a couple more days of "accurate" data, model should behave

Cupatae

normanoffside wrote: »

You've changed your tune.
I thought you said that people crunching numbers are weird?

What ever about that... You keeping tabs on me...how very stalker-ish and creepy...

Cordell

Deleted User wrote: »

Thats my next step...

IMHO if your daily deaths model is independent of the daily cases then it doesn't need any more refining, all it needs it's a new beginning
As in use the existing data to model the relationship between them, model the new cases and derive from it the daily deaths, it's pretty much obvious that a descending trend in new cases must be followed by a descending trend in deaths. You have it in the exactly opposite way...

IAMAMORON

The perils of forecasting.

Anecdotal inconclusive trend shaming, the pain of it.

ShineOn7

From the Reddit user who makes all the data very easy to digest:

_______________________________


New Cases: 388 Germany: N/A
Total Cases: 16041
New Death: 44
Total Deaths: 730 (*1 death de-notified)
Male: 18
Female: 26
Median Age: 87
Underlying conditions: 33

• Deaths: 730
• Hospitalised + died in hospital: 363 50%
• Died in ICU: 47 6%
• Underlying conditions: 601 (82%)
• Median age: 83

Of 318 in ICU

• Remain in hospital: 143 (45%)
• Discharged: 127 (40%)
• Died: 47 (14.80%)
• Underlying conditions: 263 (83%)
• Median Age: 60

As of Sunday 19th

• Cases: 15464
• In Hospital: 2323 (15%)
• In ICU: 315 (2.0%)
• Deaths: 719 (4.60%)
• Clusters: 491
• Account for 3447 cases
• Median age infected: 48
• Healthcare workers: 4180

Clusters

• Residential Homes: 287
• Nursing Homes: 176
• Death Nursing Homes: 427 - 303 lab confirmed
• Deaths Residential: 502 - 394 lab confirmed

Testing

• 111,584 tests - Increase in 20822
• 4025 positive - 19% positivity rate

Hospitalisations

• 816 Currently in hospital
• 856 Discharged
• 295 Deaths in hospital (And never in ICU)

ICU

• 156 currently in ICU
• 105 discharged
• 47 deaths

FOR EVERYONE WHO ALWAYS ASK FOR RECOVERY NUMBERS
Recoveries in community (never in hospital)

• 8377 recovered (55%)
• Balance are still within the first 14 days

Questions

• 9000-10000 capacity for testing a day. 24-36 hours once a sample reaches the lab for a result

Cordell

Cordell wrote: »

IMHO if your daily deaths model is independent of the daily cases then it doesn't need any more refining, all it needs it's a new beginning
As in use the existing data to model the relationship between them, model the new cases and derive from it the daily deaths, it's pretty much obvious that a descending trend in new cases must be followed by a descending trend in deaths. You have it in the exactly opposite way...

Improving my model - Now tops out at 21,366 cases with no new cases from the 5th May. Deaths topping out at 2,057 with no new deaths from the 16th May. A bit more refinement needed on the relationship between cases and deaths, and also, the fall off in cases is faster than I anticipate it will be, driven by late artificial peak due to the backlog making the negative slope week to date higher than actual.
What I'd like to have is case data by date of test not result and death data by date of death not reporting.

Blut2

IAMAMORON wrote: »

Debenhams were always closing. There is also an entire argument still to be had over the future of the high street.

But McDonalds will not be going bust, and people still eat out if they are on the dole.

Tourism will struggle for sure. But it will recover.

This differs to the credit crunch because prior to the property crash we had a debt problem insofar as the banks were propping up all the spending and then turned off the taps suddenly. This shut down has come about for different reasons. Some people will lose jobs, but there will also be a rebound. Companies will also be looking to achieve their targets. It is likely they will use this as an opportunity to shuffle staff, but they will need to employ people in certain areas to achieve desired productivity.

If you really want to have an economic crank I would be looking at the long term survival of the high street or for financial services employees. The computer is winning both these particular games. The banks do not need half the staff anymore. Their long term model is computerisation. This is not virus related however and was always coming down the line. Banks may use the crisis to validate redundancies they had always been planning, other firms too.

Slightly late replying here, but:

Irish GDP to fall by 10.5% this year due to Covid-19

He said the country will go from a state of near full employment to a peak unemployment rate of 22% in the second quarter, before averaging out at 13.9% for the year.

Approximately 220,000 jobs will be lost

He also said that the country's budget deficit is expected to be 7.4% this year, or €23 billion.

"Economy will take two years to come back from coronavirus crash"

https://www.rte.ie/news/business/2020/0421/1132910-stability-programme-update-2020/

It looks like the government, and every financial expert in the country, disagrees with your optimistic assessment. This dire predictions are only going to get worse the longer this lockdown goes on, too.

IAMAMORON

Blut2 wrote: »

Slightly late replying here, but:



https://www.rte.ie/news/business/2020/0421/1132910-stability-programme-update-2020/

It looks like the government, and every financial expert in the country, disagrees with your optimistic assessment. This dire predictions are only going to get worse the longer this lockdown goes on, too.

Yes but a lot of the current government speak is PR designed to facilitate the reduction in lockdown measures which will be happening over the next 4-5 weeks.

So when Pascal O'Donoghue states 7 minutes into yesterdays main evening news that we are entering into an unprecedented recession he is not only giving you the opportunity to think you might be right about something.

This is also allowing a very tender coalition to softly gaslight the entire population into believing the economy is phucked. It is a win win piece of political manoeuvring. Tell the public they are doomed but give them the caveat that if you stick with them they will make things better. If the economy bounces out they did a great job. If it continues to struggle they can blame the negative impacts of the pandemic.

Politicians lie for a living, don't forget that. Just because Pascal O'Donoghue is covering his political ass does not or should not get in the way of a genuine economical debate. His job is to get votes, not solve the economy. Civil servants in the department of finance will be doing that, I would ask them if you happen to know any.

I never denied that the Irish economy was going to struggle, it you read my posts. What I was stating was that any recession will not be as detrimental as 2008-2011 and that it is highly likely that we will bounce out of one quicker as there are considerably less confining elements driving the current recession. It will look bad, but will not be as long as our last recession.

yosemitesam1

IAMAMORON wrote: »

Yes but a lot of the current government speak is PR designed to facilitate the reduction in lockdown measures which will be happening over the next 4-5 weeks.

So when Pascal O'Donoghue states 7 minutes into yesterdays main evening news that we are entering into an unprecedented recession he is not only giving you the opportunity to think you might be right about something.

This is also allowing a very tender coalition to softly gaslight the entire population into believing the economy is phucked. It is a win win piece of political manoeuvring. Tell the public they are doomed but give them the caveat that if you stick with them they will make things better. If the economy bounces out they did a great job. If it continues to struggle they can blame the negative impacts of the pandemic.

Politicians lie for a living, don't forget that. Just because Pascal O'Donoghue is covering his political ass does not or should not get in the way of a genuine economical debate. His job is to get votes, not solve the economy. Civil servants in the department of finance will be doing that, I would ask them if you happen to know any.

I never denied that the Irish economy was going to struggle, it you read my posts. What I was stating was that any recession will not be as detrimental as 2008-2011 and that it is highly likely that we will bounce out of one quicker as there are considerably less confining elements driving the current recession. It will look bad, but will not be as long as our last recession.

Pascal's being very optimistic about the future, things could get a lot worse with the added bonus of much more limited emigration options than before

ShineOn7

From the Reddit user doing the useful daily digests. All below the line is their words


_____________________________


New Cases: 631
Total Cases: 16671
New Death: 49

• Of these 49 deaths the most that occurred on a given day is 14, which occurred on the 20th

Total Deaths: 769

• The largest number on a given day looks to be 42 - was difficult to read the graph. I'll try confirm this data.
• 10 deaths have been de-notified

Male: 26
Female: 18
Median Age: 48
Underlying conditions: 33
Deaths

• Deaths: 769
• Hospitalised + died in hospital: 386
• Died in ICU: 50
• Underlying conditions:
• Male: 415
• Female: 354
• Median age: 82
• Mean age: 69
• Range of ages: 23-105

Age Category Breakdown of Deaths

• 0-9 - 0
• 10-19 - 0
• 20-29 - <5 (won't give exact number to protect anonymity. Last I had seen, this number was 1)
• 30-39 - <5
• 40-49 - 12
• 50-59 - 29
• 60-69 - 53
• 70+ - 667

Of 327 in ICU

• Remain in ICU: 147
• Discharged: 130
• Died: 50
• Underlying conditions: 271
• Median Age: 60

As of Monday 20th

• Cases: 15871
• In Hospital: 2387
• Total In ICU: 327
• Currently in ICU: 132
• Deaths: 757
• Clusters: 507 *Account for XX cases
• Median age infected: 46
• Healthcare workers: 4393%

Residential Homes (Which include nursing homes)

• Clusters: 302
• Total cases: 2604
• Hospitalised: 281
• Deaths: 412

Just Nursing Homes

• Clusters: 179
• Total cases: 1944
• Hospitalised: 190
• Deaths: 348

Far more people are recovering in nursing homes, than those dying of Covid-19
HIQA - rapid tech detection system
Different Test methods

• Pathogen detection - rt-PCR - long turn around times, approx 6 hours
• Antibody tests
• New tech which we can leverage - some of them are associated with batch testing and have some degree of automation
• Some test methods with shorter turnaround times, which can alleviate the pressure in the labs

ANTIGEN DETECTION

• Antigen detection tests (another type of pathogen detection tests) - have had them for years for flu and RSV
• Less complex than PCR, faster, easier to use, but less sensitive than PCR.
• No performance data on antigen detection - research only until test type is validated
• May be used in conjunction if labs under pressure. Used as a rule in - aka if positive, that’s it. If negative, needs to be confirmed by PCR

ANTIBODY

• Antibody test - population level immunity or exposure
• Will identify those with a sub clinical presentation of the disease, to help modelling
• Knowledge gaps relating to antibody testing - many limitations
• Don't know yet the adequacy and duration of immunity
• Don't know if reinfection can occur
• Don't know if it will test positive to other coronaviruses in community
• Antibody drift
• HIQA endorsing ongoing monitoring for timely and accurately result
• Need clinically independent validation of tests, need to see that they perform as expected in Ireland, and ongoing QA program for patients
• Comprehensive strategy needed for rapid deployment of tests when they meet the standards, are validated and come available for use.

Questions

• If decision was made today, there would not be a recommendation of easing of restrictions? Tony is very concerned about people slacking off in advance of 5th May
• 2nd wave of infection? We haven't really had a first wave, because of how well we've managed the diseased. If no mitigations, we would have seen 33% growth in cases day on day. We didn’t see that because we had significant measures in place to keep the disease at bay, and eventually get the reproductive rate below 1
• 631 new cases - how many of these came from residential homes? More detailed info on pathway of nursing homes will be later in the week Encouraging early notifications
• Denotified deaths - what does that mean? What is reported daily is provisional, as it is just in. More detailed data comes in after 36 hours. This data is cleaned up, and some of the notifications can change their status. It's not that the death is no longer a death. The person may have been swabbed, died, and upon review of the result they were negative

Hmmzis

Median age: 48

Is that for infected or the daily deaths? If for the deaths then smething very abnormal must have happened, given the total fatality numbers by age groups.

trapp

Hmmzis wrote: »

Median age: 48

Is that for infected or the daily deaths? If for the deaths then smething very abnormal must have happened, given the total fatality numbers by age groups.

Median age for those who had the virus.

I believe median age for deaths is 80 +

Jurgen Klopp

TWO PEOPLE WHO died in California in early and mid February have been confirmed through autopsies as coronavirus cases – weeks before the first reported COVID-19 fatality in the United States.

The deaths, on 6 February and 17 February, push back the timeline for the entry of the virus on US soil, experts said.

Given what we know about how long it takes for the virus to kill people, the disease began to spread in the US sometime between early and mid January

What's more officials say they are unaware of any travel that would have caused them to pick it up

https://www.thejournal.ie/us-coronavirus-spread-january-california-5081499-Apr2020/

wakka12

I am so interested to see the result of the community antibody tests that many countries are undertaking...will provide such insight into how this pandemic is really spreading. Like those California deaths..what, just crazy

DeVore

Beware of antibody tests... they are quite inaccurate at times.

If you test 100k people and have a "false negative" rate of 95% accurate you are still giving 5k people the wrong result.

I'm working on a piece about this now but its complicated and hard to boiled down to easy communications

ShineOn7

“99% of Those Who Died From Virus Had Other Illness, Italy Says”

https://www.bloomberg.com/news/articles/2020-03-18/99-of-those-who-died-from-virus-had-other-illness-italy-says?sref=RKhDoZKr&fbclid=IwAR2lBEPXB-zeZnHmP5uYj9oNG9_GbZXL-s7HdU4DcES1_GEGFEB3GXiZ2Jw

Is Bloomberg a reliable source? It’s hard to know who is these days

2ndcoming

ShineOn7 wrote: »

“99% of Those Who Died From Virus Had Other Illness, Italy Says”

https://www.bloomberg.com/news/articles/2020-03-18/99-of-those-who-died-from-virus-had-other-illness-italy-says?sref=RKhDoZKr&fbclid=IwAR2lBEPXB-zeZnHmP5uYj9oNG9_GbZXL-s7HdU4DcES1_GEGFEB3GXiZ2Jw

Is Bloomberg a reliable source? It’s hard to know who is these days

99% is nearly always used as an expression rather than an actual percentage. Even Putin doesn't get 99%.

Our ICU cases in the stats above bear out that a huge proportion of those cases have underlying conditions. Very few over 70s don't have some kind of long term ailment either.

wakka12

ShineOn7 wrote: »

“99% of Those Who Died From Virus Had Other Illness, Italy Says”

https://www.bloomberg.com/news/articles/2020-03-18/99-of-those-who-died-from-virus-had-other-illness-italy-says?sref=RKhDoZKr&fbclid=IwAR2lBEPXB-zeZnHmP5uYj9oNG9_GbZXL-s7HdU4DcES1_GEGFEB3GXiZ2Jw

Is Bloomberg a reliable source? It’s hard to know who is these days

99% sounds very high. Nearly everyday sky news announce how many deaths in the U.K. had no underlying condition and it's almost always around 5%

2u2me

DeVore wrote: »

Beware of antibody tests... they are quite inaccurate at times.

If you test 100k people and have a "false negative" rate of 95% accurate you are still giving 5k people the wrong result.

I'm working on a piece about this now but its complicated and hard to boiled down to easy communications

Can they also return 'false positive'?

yosemitesam1

DeVore wrote: »

Beware of antibody tests... they are quite inaccurate at times.

If you test 100k people and have a "false negative" rate of 95% accurate you are still giving 5k people the wrong result.

I'm working on a piece about this now but its complicated and hard to boiled down to easy communications

False positives and false negatives are problems with the PCR test also.
The antibody testing shouldn't be about giving a definitive answer as to whether or not someone is immune but showing that the virus is nowhere near as dangerous as was originally suggested.
The initial surveys are showing 10-30+ times more spread than the official figures and false negatives are more likely than false positives.
So if they're scaled up further and continue to tell us that the virus mortality/hospitalisation rates are much lower than originally assumed, we could just let the virus run it's course in the general population