COVID-19 Coronavirus Containment Efforts

[bob012345]

Well, you'd be wrong. See https://ourworldindata.org/coronavi...testing-compared-to-the-scale-of-the-outbreak For example, croatia and Portugal are more than an order of magnitude apart.

Well, I'm wrong about a lot of things but this chart is daily tests vs. daily confirmed cases. I'm not sure what this has to do with my point?

 

[Vanadium 50]

That's just the first plot. Scroll down and you will see "How many tests are performed each day" and "
total tests performed relative to the size of population".

 

[bob012345]

That's just the first plot. Scroll down and you will see "How many tests are performed each day" and "
total tests performed relative to the size of population".

What does this have to do with my original suspicion? I didn't mention testing.

 

[mfb]

The European Union has a population of 445 million. You are only prepared to compare Sweden with three countries with a combined population of 16 million. Excluding 96% of the data is fundamentally unscientific and designed only to support an a priori conclusion.

The list had more than three countries.
Why stop at the EU? Why not take Earth?
Excluding 96% of some larger dataset is perfectly fine and even necessary if these 96% are different.

If you really believe what you say you must be prepared to absolutely condemn Spain in comparison with its neighbor Portugal: 580 deaths per million against 144. That's appalling. What was the Spanish government doing to lose so many people?

Act too late, among other problems. Why is this even a question? Does anyone doubt that Portugal did better than Spain?

I didn't mention testing.

Yes, that's the problem. You compared people tested positive without taking into account how many people were tested. No tests, no confirmed cases, perfect outcome?

 

[Vanadium 50]

What does this have to do with my original suspicion? I didn't mention testing.

Here's what you said.

I suspect that the number of actual cases tracks the number of reported cases fairly consistently in European countries. If so, using reported cases is still a better benchmark than deaths.

The way you get to be a "reported case" is to test positive. That means there is a strong dependency on how many people you test.

Fatalities don't have the same problem. Dead is dead, whether or not you've been tested. They do have the problem that ascribing a single cause is difficult if the patient has multiple conditions, and we know that these are the people for whom Covid is most deadly. But this is perhaps tens of percent, not the order of magnitude of the testing rate.

 

[bob012345]

Here's what you said.
The way you get to be a "reported case" is to test positive. That means there is a strong dependency on how many people you test.

Fatalities don't have the same problem. Dead is dead, whether or not you've been tested. They do have the problem that ascribing a single cause is difficult if the patient has multiple conditions, and we know that these are the people for whom Covid is most deadly. But this is perhaps tens of percent, not the order of magnitude of the testing rate.

Ok, I made an assumption and used a misleading word when I said it 'tracks'. What I meant to communicate was that I believe the number of actual cases correlates to the number of official confirmed cases. Obviously not one-to-one. For a large population size and roughly equivalent social dynamics like European countries I thought that was reasonable but I would not compare Belgium to Belize. Sorry for the confusion.

 

[PeroK]

Fatalities don't have the same problem. Dead is dead, whether or not you've been tested.

It's funny you should say that. The Spanish suddenly went from an average of 179 deaths per day for the week up to 26th May to an average of 1 death per day for the week up to 2nd June. They recorded 280 new deaths on May 26th and no more than 5 on any day since. This is a pattern not shown in any other European country. It's not credible that one day the deaths from COVID-19 would almost suddenly stop.

I found this in the Financial Times, which is also skeptical of the Spanish figures. They quote that 17 deaths were recorded in the regions but never made it to the national figures:

https://www.ft.com/content/77eb7a13-cd26-41dd-9642-616708b43673

 

[bob012345]

It's funny you should say that. The Spanish suddenly went from an average of 179 deaths per day for the week up to 26th May to an average of 1 death per day for the week up to 2nd June. They recorded 280 new deaths on May 26th and no more than 5 on any day since. This is a pattern not shown in any other European country. It's not credible that one day the deaths from COVID-19 would almost suddenly stop.

I found this in the Financial Times, which is also skeptical of the Spanish figures. They quote that 17 deaths were recorded in the regions but never made it to the national figures:

https://www.ft.com/content/77eb7a13-cd26-41dd-9642-616708b43673

Why assume the number of deaths per day are anything other than accounts of the numbers recorded that day, not the actual deaths that day. The accuracy and consistency of reporting systems might vary geographically and temporally.

 

[Evo]

V50 asked to have this thread merged with the Covid-19 containment thread, so I have done so. The posts are now here https://www.physicsforums.com/threads/covid-19-coronavirus-containment-efforts.983707/

 

[Vanadium 50]

Thanks!

 

[Vanadium 50]

The Spanish suddenly went from an average of 179 deaths per day for the week up to 26th May to an average of 1 death per day for the week up to 2nd June.

That's not what the black plot in Wikipedia shows. https://en.wikipedia.org/wiki/COVID-19_pandemic_in_Spain

My guess is you (well, FT, which is behind a paywall) was looking at a point in time snapshot and things hadn't caught up. For a while the US had exceeded 100,000 fatalities, but the sum of all states and territories had not.

 

[PeroK]

That's not what the black plot in Wikipedia shows. https://en.wikipedia.org/wiki/COVID-19_pandemic_in_Spain

My guess is you (well, FT, which is behind a paywall) was looking at a point in time snapshot and things hadn't caught up. For a while the US had exceeded 100,000 fatalities, but the sum of all states and territories had not.

The up-to-date figures are here:

https://www.worldometers.info/coronavirus/country/spain/

 

[Vanadium 50]

Here is a larger metastudy:
Physical distancing, face masks, and eye protection to prevent person-to-person transmission of SARS-CoV-2 and COVID-19: a systematic review and meta-analysis
Not surprisingly, they find that all three help.

Every time I start softening my stance on "meta-analyses", and start thinking "maybe it's not so bad". a paper like this comes around.

What they are saying is

1. In a population with an unknown amount of physical distancing and face mask wearing, eye protection helps.
2. In a population with an unknown amount of physical distancing and eye protection wearing, face masks help.
3. In a population with an unknown amount of eye protection and face mask wearing, physical distancing helps.

That's very different from "all three help".

The problem is that the three actions are correlated. If you made a 2x2 of physical distancers and face mask wearers, you would find a lot of people who do both, or neither, and relatively few who only do one. To see the problem, suppose the correlation were 100% but only one is actually effective: all three would still show up as effective, because everyone doing one is also doing the other two.

Of course the correlation isn't 100%, but the problem is still there, just at a lower degree.

 

[sophiecentaur]

That's very different from "all three help".

Where the three populations totally unrelated?
I don't know a lot about meta-analysis but I thought it was supposed to take care of your dilemma as much as possble. I thought it dealt reliably with multiple factors. If not then I am as skeptical as you have been about the system. I mean, it brings into doubt an awful lot of information that's obtained about populations and health. But medics do have some strange attitudes to statistics in general.

I agree that there has to be a certain amount (or a lot) of correlation. It's pretty amazing that details about the 'other two' were not considered important enough to include in any questionnaire / analysis. You'd need to be in the business if you wanted to do any survey of that kind so wouldn't you know?

I have a personal interest in this due to my age so its hard to be dispassionate.

 

[BillTre]

I just read on the local PBS (OPB (Oregon Public Broadcasting)) website some of the criteria by which Oregon counties are allowed to go to phase 2. (We (Lane) county) did this on Friday.)

Each county has to spend a minimum of 21 days in Phase 1, and see no significant increase in positive cases. The Oregon Health Authority analyzed county-level metrics to see if counties have sufficiently contained the virus, and if local health departments are adequately testing and tracking cases. State epidemiologist Dean Sidelinger said that the state wants to see that county case numbers over the last seven days aren’t above where they were the week prior.A minimum of 95% of all new cases must be contact traced within 24 hours, and a minimum of 70% of new COVID-19 positive cases must be traced to an existing positive case.

The phase 2 step seems to be: larger numbers of people allowable in groups, things open later.

Being able to contract trace cases (and therefore knowing where the infection came to be), seems like it would be a pretty laborious task (assuming not using the cell phone tracking solution (which ain't happening now anyway!)).
This means that there is a tradeoff between the size of the infected population and the ability of society to successfully undertake the task (ability to get it done), and the investment society wants to put into the task (to fund the required labor).
Thus reductions in numbers of cases should be important to feasibility of doing the tracing.

 

[Vanadium 50]

I have a personal interest in this due to my age so its hard to be dispassionate.

Then do all three. How can it hurt?

Consider a simple case - we're looking at some effect vs. clothing. We see a strong effect with red hats and blue ties. If everyone who wears a red hat wears a blue tie and vice versa, there is no amount of meta-analyzing that will determine if the important factor is the red hat, the blue tie, or both.

 

[mfb]

The up-to-date figures are here:

https://www.worldometers.info/coronavirus/country/spain/

Where do you get single-digit death counts from that source? They adjusted their total deaths down by 2000 on May 25, but that doesn't mean people were resurrected that day, it's just changing how past deaths were counted.

----

15 days without new case in NZ. One active case remains. New Zealand could be the first major* country to eliminate the disease. Iceland was ahead for a while, but keeps finding a few cases, the last one a week ago.

*Not counting places like Fiji: 18 reported cases, the last one late April, all recovered

 

[kyphysics]

Australia is lifting restrictions. Tons are screaming - lift them entirely - we have beaten it. Not so fast - we are in fact tetering as the growth factor (r0 is 1 right now) shows:
https://www.abc.net.au/news/2020-04...ustralia-growth-factor-covid-19/12132478?nw=0

We need to lift restrictions slowly and be prepared to clamp them on again. I try to explain this in forums out here, but to no avail. Personally I think we are lifting them too fast.

Thanks
Bill

a.) A lot of people don't even know what R0 means, nor are very literate about the virus. They know it's like the flu that's killed a lot of people and has locked the world down, but aren't very literate about the stats and concepts behind it.

b.) Isn't Australia going into winter? Some have said that warmer/summer countries like Australia haven't really been tested yet, as the virus spreads less easily in the summer months (which Australia was in, no?).

 

[PeroK]

Where do you get single-digit death counts from that source? They adjusted their total deaths down by 2000 on May 25, but that doesn't mean people were resurrected that day, it's just changing how past deaths were counted.

The page for Spain (like all countries) gives the daily figures in a graph. But, for Spain they are now so low you can't see them on the graph. Lower down the page there are the figures for each day - you have to expand the data for each day. From May 26th the deaths in Spain are:

280, 1(*), 1, 2, 4, 2, 0, 0, 1, 5, 1, 1

And, for the record, the new case numbers are:

859, 510, 1137, 658, 664, 201 (**), 209, 294, 394, 334, 318, 332

(*) The deaths in Spain suddenly almost stopped on May 27th.

(**) The cases in Spain suddenly went down on May 31st.

 

[mfb]

From May 26th the deaths in Spain are

That's a reporting delay, almost certainly. Wait a week for these numbers. Wikipedia probably uses a Spanish source and has plausible two-digit numbers.Tweet chain about COVID-19 and protests in the US

 

[PeroK]

That's a reporting delay, almost certainly. Wait a week for these numbers. Wikipedia probably uses a Spanish source and has plausible two-digit numbers.

There are several news items about this now.

 

[Vanadium 50]

The up-to-date figures are here:

You can see -1915 deaths on one day. That's probably not what happened. You also see the total number of deaths go down. That's probably not what happened either. And what's really off is those events happened on the same day.

It's likely a redefinition, either of cause of death, or geographically. (something double-counted) These things happen. They are annoying when they do.

 

[PeroK]

You can see -1915 deaths on one day. That's probably not what happened. You also see the total number of deaths go down. That's probably not what happened either. And what's really off is those events happened on the same day.

It's likely a redefinition, either of cause of death, or geographically. (something double-counted) These things happen. They are annoying when they do.

I'm not looking at the one-off readjustments where a large number is added or subtracted on a single day. There are several news items reporting that Spain recorded zero deaths on consecutive days and is generally recording single-figure deaths per day now. Some of these (mostly the ones from 5-6 days) are positive. The most recent ones (including El Pais) are more skeptical. This is from the Telegraph:

Spain’s government has been accused of hiding the death toll from Coronavirus amid changes made to its method of reporting new cases that saw the country’s fatality toll plummet from around 50 per day last week to zero on both Monday and Tuesday.

The accusations of undercounting the impact of Covid-19 come as Spain announces plans to welcome international tourists back to the country from July 1, or possibly during the second half of June.

Make of that what you will.

 

[anorlunda]

The news is saying that the mortality rate is well below 1%, such as 0.4%. Today's news also said for the USA, 110K dead, 500K recovered. If everyone uses the same definitions, shouldn't we have this?

$mortality=\frac{dead}{recovered+dead}=\frac{110}{500+110}=0.18$

It makes my head hurt because those two stats seem to differ by the ratio 45:1.

One way to make both stats true is to assume a huge number of asymptomatic (and not tested) who are never counted as recovered or dead but who are counted when calculating 0.4% mortality. That mean means assuming 27 million asymptomatic. That is 7% of the population, which I guess is possible. But if that is the assumption, how could they not mention it when figures are released?

$mortality=\frac{dead}{recovered+dead+asym}=\frac{110}{500+110+27000}=0.004$

Do they publish their assumptions and methods of counting mortality?

 

[PeroK]

The news is saying that the mortality rate is well below 1%, such as 0.4%. Today's news also said for the USA, 110K dead, 500K recovered. If everyone uses the same definitions, shouldn't we have this?

$mortality=\frac{dead}{recovered+dead}=\frac{110}{500+110}=0.18$

It makes my head hurt because those two stats seem to differ by the ratio 45:1.

One way to make both stats true is to assume a huge number of asymptomatic (and not tested) who are never counted as recovered or dead but who are counted when calculating 0.4% mortality. That mean means assuming 27 million asymptomatic. That is 7% of the population, which I guess is possible. But if that is the assumption, how could they not mention it when figures are released?

$mortality=\frac{dead}{recovered+dead+asym}=\frac{110}{500+110+27000}=0.004$

Do they publish their assumptions and methods of counting mortality?

There must be a large number of uncounted cases in most countries - in any case. We have estimates in the UK that up to 7% (5 million people) may have had the virus, going by random antibody tests etc.

If this is true and we have 40,000-50,000 deaths from COVID-19, that puts the mortality rate at about 1%. And, obviously, if the 0.4% is true, then we must have had more like 12.5 million cases in the UK.

I don't know what the resolution to this is.

That said, in addition to the mortality rate itself, there are two additional factors. 1) How the mortality rate varies across demographics (especially age). There are huge variations across the world in the number of cases to deaths. Generally in western Europe it's 10-20%. Elsewhere it's generally much lower. Russia is about 1%, India about 3%, and the Middle East is recording almost no deaths (e.g. Qatar at less than 0.1%).

And 2) the additional excess deaths caused if the outbreak gets out of control and not only people with COVID-19, but also people requiring emergency hospital treatment of any kind, may die when they otherwise might not.

Who knows what's really going on here.

 

[Vanadium 50]

shouldn't we have this?

No, because it does not include people who have the disease but have neither died nor recovered, i.e. still sick.

If you are using the CDC's 0.4%, that's the rate of fatality given that there are symptoms. They also say 1/3 of the people never develop symptom, so the IFR is actually less. 0.27%. This is for the USA as a whole - numbers would be different at the Boston Nursing Home for Men than for the Missoula Women's Junior College.

Does 0.27% make sense? 110,000 fatalities implies 40 million infected. Of those, 14 million showed no symptoms, 2 million were sick enough to get tested, which leaves 24 million who got sick, but not sick enough to see anyone about it. (We have a forum member who says he is in this category).

Is this possible? One way of looking at it is, if infected, there is a 33% chance there are no symptoms, 62% the symptoms are mild, 5% the symptoms are severe, and a 0.27% chance it's fatal. That sounds plausible. Another way is that 24M people are feeling sick in the middle of a pandemic, bombarded from all sides with the message "You're going to die! You're going to die!" and yet they do nothing about it. This seems unlikely. So there's a tension here.

The only way I can reconcile this is that the vast majority of the 24M have only weak symptoms and they attribute them to something else. If you spend all day working around the house and have fatigue and muscle aches and no other symptoms, your first thought is probably not "I have Covid".

 

[anorlunda]

The only way I can reconcile this is that the vast majority of the 24M

My question is more directed at informing the public instead of confusing the public.
I would write the headline for the TV news as follows.

110 Thousand Dead, 500 Thousand Recovered, 24 Million Infected

Failure to do so feeds the feeling that the real facts are concealed from us. It breeds conspiracy theories.

 

[PeroK]

My question is more directed at informing the public instead of confusing the public.
I would write the headline for the TV news as follows.

110 Thousand Dead, 500 Thousand Recovered, 24 Million Infected

Failure to do so feeds the feeling that the real facts are concealed from us. It breeds conspiracy theories.

One problem is what happens if the CDC is wrong? If the UK decided to gamble on the fatality rate being 0.27, we would be looking at a maximum of 150,000 deaths in the UK if we had gone for minimal lockdown measures. You might argue that even that is too many, but let's say the UK decided to gamble. What happens if the CDC is wrong? And, after 3 months we are looking at 150,000 deaths with the health service overrun and no end in sight?

 

[Vanadium 50]

110 Thousand Dead, 500 Thousand Recovered, 24 Million Infected

People are in one of six categories: Unexposed, exposed, infected, sick, recovered or dead. The easiest to measure are "recovered" and "dead" (and even these have some problems). We know testing positive undermeasures "sick", and way undermeasures "infected".

Here's a plot where I plot case rate vs. inferred infection rate (deaths/0.027%).

You can see things are all over the place, although there is a trend that places without a lot of infections do a better job of counting them than places with a lot of infections (which makes sense). The three highest states are UT, SD and NE, and the four states out to the right are NJ, NY, CT and MA.

Failure to do so feeds the feeling that the real facts are concealed from us. It breeds conspiracy theories.

Others here have argued the reverse - that it's good to make things look worse than they are in order to maximize public compliance.

 

[Ygggdrasil]

Regarding the data from Spain, another way of estimating deaths from COVID-19 that does not rely on COVID-19 testing is to track the total number of deaths in a region (from any cause), and compare to the historical baseline. This measure of "excess deaths" can help see how many deaths are potentially being missed.

Of course, there are some limitations to excess deaths. There are some arguments that excess deaths could overstate COVID-19 deaths. For example, hospitals have seen admissions for heart attacks and strokes sharply decrease during the pandemic, so some excess deaths could be due to people avoiding medical care and dying from preventable conditions at home. However, there are also reasons to think that excess deaths could understate COVID-19 deaths. For example, with fewer people on the roads, there have been fewer fatal car accidents during the pandemic, which should decrease the baseline number of expected deaths.

With these caveats in mind, here are some data from The Economist's site that is tracking excess deaths across various locales:

These data fairly clearly show excess deaths leveling off to baseline around May in Spain, so reports of a low number of COVID-19 deaths in Spain are plausible. Remember that Spain's lockdown measures were fairly strict (the Financial Times has a headline saying "Spain's reopening is stricter than America's Coronavirus lockdown"), and other countries with very strict lockdowns have managed to reduce cases to near zero (e.g. China).

Regarding whether Spain is under-reporting cases:

The number of COVID-19 deaths reported vs the total excess deaths is not overtly different than other similar countries. However, the time period for the data in Spain (Mar 10-May 18) do not seem to include the time period for which the numbers were revised (around May 27-31 according to @PeroK's post), so we may have to wait a few weeks to see whether Spain's revisions to their death counts were appropriate.

 

[Bystander]

Has anyone examined mortality as a function of previous polio vaccination? Age, respirators, think iron lung, Salk/Sabin/sugar cubes in childhood in the "fifties," versus adulthood in the "fifties" who are today's geriatric crowd?

 

[Vanadium 50]

The Economist is using Euromomo, which is pretty good. You've outlined many of the issues. Others are:

1. There are "excess deaths" every single year, which suggests that the expectation needs some adjustment.
2. Excess deaths counts things we don't want to include, such as suicides. I don't know why suicide stats lag 4-6 months - I see no reason why they should.
3. Excess deaths is now down to zero (slightly below, in fact) but the number of Covid deaths in Europe is still ~700 per day.

#3 surprised me. What I think is happening is that since Covid strikes the elderly particularly harshly (the median age of fatalities in Massachusetts is 82) the surviving European population is statistically younger and healthier than pre-Covid. Put another way, people who otherwise would have died in June died back in April. This didn't just start happening now - it's just that we can see it now - so even "excess deaths" underestimates Covid.

I'm not saying we should disregard the "excess deaths" metric. I'm just not prepared to conclude that this is the One True Way to calculate things and anyone who does things differently is covering up.

 

[bob012345]

Has anyone examined mortality as a function of previous polio vaccination? Age, respirators, think iron lung, Salk/Sabin/sugar cubes in childhood in the "fifties," versus adulthood in the "fifties" who are today's geriatric crowd?

Are you suggesting a link between Covid-19 deaths and polio vaccines?

 

[Vanadium 50]

One problem is what happens if the CDC is wrong?

If you are talking about science, I think need to be more specific than "I just think it's wrong". If we are talking about public policy, we should probably weigh the costs and benefits and recognize that these are not borne equally.

If you are arguing that the US number doesn't apply directly to the UK, I agree with you (as would the CDC). The UK population is older (average 40 vs 38) so I would expect the relevant number for the UK to be higher. Looking at the high end, the UK has about 6% of their population 80+ and the US it is more like 4%.

The four US states with the highest mortality rates per thousand are NJ, NY, CT and MA, varying from 1.04 to 1.36. All four have had a very large fraction of deaths (around half, up to 60% in MA) in nursing homes. How elder-care facilities respond will make a huge difference to the outcome.

 

[Ygggdrasil]

Has anyone examined mortality as a function of previous polio vaccination? Age, respirators, think iron lung, Salk/Sabin/sugar cubes in childhood in the "fifties," versus adulthood in the "fifties" who are today's geriatric crowd?

Here's what a document from the Global Polio Eradication Initiative says:

There is no evidence that oral poliovirus vaccine (OPV) protects people against infection with COVID-19 virus. A clinical trial addressing this question is planned in the USA, and WHO will evaluate the evidence when it is available. In the absence of evidence, WHO does not recommend OPV vaccination for the prevention of COVID-19. WHO continues to recommend OPV as part of essential immunization services to prevent poliomyelitis and as part of global polio eradication efforts.

http://polioeradication.org/wp-content/uploads/2020/03/Use-of-OPV-and-COVID-20200421.pdf

IMHO, there does not seem to be good scientific evidence to suggest that the poliovirus vaccine would provide protection against coronavirus, though if you have seen some compelling scientific publications to suggest otherwise, I would be open to taking a look.