Of Tests and Confirmed Cases

Guest Post by Willis Eschenbach

I’ve been saying for some time now that the number of confirmed cases is a very poor way to measure the spread of the coronavirus infection. This, I’ve said, is because the number of new cases you’ll find depends on how much testing is being done. I’ve claimed that if you double your tests, you’ll get twice the confirmed cases.

However, that position was based on logic alone. I did not have one scrap of data to support or confirm it.

Max Roser is the data display genius behind the website Our World In Data. He has recently finished his coronavirus testing dataset, covering the patchwork quilt of testing in various countries. The data is available here.

Being a ‘Murican myself, I first looked at the US daily new testing versus number of US daily new confirmed cases. I have to confess, when I saw it, I did laugh …

Figure 1. Scatterplot, daily new tests versus daily new cases, United States. Yellow/black line is linear trend.

Just as I have been saying, in the US, new cases is a function of new tests. For every one hundred additional tests that we do, we find an additional nineteen confirmed cases of coronavirus. 

Of course, when I looked further there were other countries which were nowhere near as linear as the US. Here’s Australia, for example:

Figure 2. Scatterplot, daily new tests versus daily new cases, Australia. Yellow/black line is linear trend.

However, there are also plenty of countries that are just as linear as the US.

Figure 3. Scatterplot, daily new tests versus daily new cases, Turkey. Yellow/black line is linear trend.

Poland shows the same type of mostly linear relationship.

Figure 4. Scatterplot, daily new tests versus daily new cases, Turkey. Yellow/black line is linear trend.

So … how about for the whole world? Glad you asked. Here’s that chart.

Figure 5. Scatterplot, world total daily new tests versus total daily new cases. Units are thousands of tests and thousands of cases. Yellow/black line is the linear trend. Black “whiskers” show the uncertainty (one sigma) of the individual mean values for the various days.

One item of interest is the difference in the rate of discovery of new cases in various countries. In the US there are nineteen new confirmed cases per hundred new tests; Turkey is 13/100; Poland is 4/100; Australia is 1/100; and globally, there are eleven new cases for every one hundred new tests.

I suspect that this variation depends directly on at least a couple of things — the underlying number of cases in any given country, and exactly which subgroup is being tested.

For example, in the US we’re still short of tests. So the tests are being reserved for people who are showing obvious symptoms … and as a result, the US tests would be expected to come up with more new cases than the global average.

This leads to a curious situation. In addition to being a function of the number of tests, confirmed cases can also be a function of the scarcity of tests …

Conclusion? Don’t use confirmed cases as a metric of the spread of the virus—the number of cases is indeed a function of the numbers of tests. 


PS—When you comment, please quote the exact words you are discussing. It avoids endless misunderstandings.

PPS—While I’m here, let me shamelessly recommend the Watts Up With That Daily Coronavirus Data Graph Page. I create the daily graphs and maintain the page. I’ll also recommend my own blog, Skating Under The Ice. I note that it’s been one full month since I publicly called at my blog for an end to the American Lockdown. Finally, I’m on Twitter here. Enjoy.

5 1 vote
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
Steven Mosher
April 22, 2020 10:19 pm

Look at state testing data.

Anyway the problem is that as tests grow, the disease may also be spreading.
Further problem is no consistent criteria for testing.
Testing only the symptomatic or only severe will also undercount.
Undercount of asymptomatics is all over the map.
From .4% in some samples to well over 50% in others.

No controls.
No consistency
Not science
No conclusion.

Reply to  Steven Mosher
April 22, 2020 11:46 pm

Sort of backing up what you said Steven as to why Australia doesn’t fit the conclusion is because the amount of testing per day has remained somewhat constant (the test kits are rationed on a per day basis).

Here is the link for Australian testing … its the ninth one down
As you can see it wobbles between 4K and 18K
The telling part is the next graphic number 10 showing the daily tests done versus the new cases found.

Ok now the explaination.
In the beginning Australian testing was limited to only International arrivals and those who had contact with a known covid carrier because there was no known community transmission. There was excess tests so the criteria was widened to anyone in frontline services who had flu like symptoms. Finally there were again excess tests so it was widened to all and anyone with flu like symptoms. So basically what you are seeing now at the end there is a lot of testing being done on people who have the flu but few have covid.

I suspect all the linear relationship shows is countries that don’t have effective control on the countries infection.

Greg Goodman
Reply to  LdB
April 23, 2020 2:44 am

Yes, this whole issue is not a simple a Willis seems to suggest in this analysis.

This, I’ve said, is because the number of new cases you’ll find depends on how much testing is being done.

Now suppose that you have a limited number of tests so have to reserve them to A&E admissions where the need to identify the pathogen is the most urgent.

The more admissions you need to treat, the more tests you perform. The more test you perform, the more +ve you get. This correctly reflects the REAL state of the epidemic in that population. The correlation does not prove the stats are spurious. As we are constantly reminded CORRELATION IS NOT CAUSATION.

What Willis seems to have missed is an uncontrolled variable, the link between the number of tests done and the number people showing symptoms. What his US graph probably shows is that the number testing positive is a fairly consistent proportion of all people showing flu-like symptoms, reporting or taken to hospital.

I have done a detailed analysis of Italian case data and it shows pattern that is NOT consistent with the idea that the number of available tests is the CAUSE of 90% of the variation in case data:

comment image

It certainly may be a perturbing factor. There are many others.

The Italian data is already showing the first signs of the expected up tick due to relaxation. Following this analysis may help gauge the usefulness of simply delaying COVID cases into future months at enormous social and economic costs which will last for years.

Farmer Ch E retired
Reply to  Greg Goodman
April 23, 2020 4:28 am

Some simple questions:

If someone is confirmed positive for C-19, how many more times is that person tested on average before full recovery? Do those extra tests get added to the total test count? How does this very between countries and medical groups within countries?

Number of tests conducted may not equate well to number of people tested. This could also be skewed if the medical system exceeds capacity.

Reply to  Farmer Ch E retired
April 23, 2020 5:58 am

For Australia there is no average we had one minor celebrity Richard Wilkins dragged on for 4 weeks and used up 4 tests but he was unusual because he was asymptomatic and only tested because of a close contact. Due to how the test were rationed most had to have symptoms and were never retested until the symptoms cleared and most cleared 1st time.

No idea if the re-tests count towards the total test but the numbers are so small it wouldn’t really matter we only had 6661 positives while the tests done is 466,659.

Farmer Ch E retired
Reply to  Farmer Ch E retired
April 23, 2020 8:40 am

In addition, medical workers are likly tested multiple times. Are those C-19 tests included in the total count?

Reply to  Farmer Ch E retired
April 23, 2020 8:59 am

Yes I would say frontline worker tests would be in the number because they are the single most likely to return a positive test but again they have to have symptoms or a very good story :-).

Reply to  Farmer Ch E retired
April 24, 2020 2:48 pm

According to a local radio host who has been following and researching the local/state testing data, the testing numbers are based on test performed not individuals. Some people have been tested multiple times, ie one tested positive twice and negative once. All three results were included in the totals.

If this is true, the total number of cases could be lower, possibly much lower than reported.

Paul Jenkinson
Reply to  Greg Goodman
April 23, 2020 4:59 am

What if it is shown that the demographic of totally asymptomatic, untested pos rate is anything from 10-50 times the expected rate,apparently assumed to be close to zero? The denominator of mortality as fraction of number positive for corona changes dramatically if so.
Not sure what it does to the linear nature of the number of tests performed to number of positives as we are only testing symptomatic people at present who are therefore much more likely to have a positive result
If the graph is still linear then,wow!

Reply to  Paul Jenkinson
April 23, 2020 8:20 am

Can only speak for Australia and that is impossible, we would have random outbreaks of people who weren’t asymptomatic crop up in the public … we don’t so by default it isn’t right.

Crispin in Waterloo
Reply to  Paul Jenkinson
April 23, 2020 10:28 am

In support of LdB, if there is a pre-selection mechanism (ad hoc filter) of “obvious symptoms” the result isn’t going to predict much about the whole population, only the tested population+the filter rules.

It is interesting that testing of those with the single symptom of “being dead” before any suspected cases were around has turned up people (bodies) testing positive a) before there was a test or b) before anyone thought to consider a novel virus.
Unless there is random, representative testing (which is certainly not the case in Canada) we have no idea what is going one.

Everything done along that line so far (precious little) shows the number of positive responses to be far above any modeled spread. Now with word coming from the W USA and Italy that there were cases as early as the second week of January – before “official spread” we may have to revise the whole infection map, starting with testing all available possible deaths from say, mid-December.

Patient Zero in China certainly was not in the wet market of Wuhan, which is a sea food market. The fact that it spread from there means nothing. It was already spreading by other paths, as far back as October at least. The patient identified on 17 Nov was not patient Zero and had no connection to the Wuhan market. It may have been circulating and adapting in pets, snakes or pangolins for months before that.

Ian W
Reply to  Greg Goodman
April 23, 2020 6:31 am

In many areas it is precisely as Willis suggests. Testing stations will not allow testing of anyone that has not already been symptomatically diagnosed as having COVID-19 and testing stops each day when the number of test kits run out. Test kits often ran out only a couple of hours after testing started. Increases in the number of test kits increased the number of positive tests minus the very few that were incorrectly diagnosed by another means. Patients are not allowed into the COVID-19 wards unless they have received a positive test.

Testing needs to be far more rapid and far more widespread. At the same time more concentration needs to be put on increasing resistance to infection for example by avoidance of zinc deficiency and increasing Vitamin D levels. It is not feasible to lock down the world.

Those areas in lockdown seem to be following a similar infection profile as those not in lock down except that the locked down countries/regions are suffering extreme economic damage that is already leading to deaths due to other causes and will eventually lead to inability to afford health care for any patients. I realize that there is a race to become rich as Croesus from selling vaccines but that goal may be illusory as the vaccines may need to be handed out ‘free’ as the dead economies will not support paying for them.

The initial intent was to avoid a peak load that the health systems could not handle, and that has probably been achieved. A return to work while avoiding large gatherings and parties and perhaps wearing masks most of the time is necessary. Continual house arrest while being driven into penury by well paid politicians will not be tolerated for very much longer.

Reply to  Ian W
April 23, 2020 7:42 am

I’m with you In
I no longer care much about the testing. It is becoming clearer every day that the lockdown is doing more damage than the disease, except possibly in areas of high population density.
This despite the efforts to inflate the mortality numbers.
Obviously the Governor of NY is a far weaker breed than the likes of Patrick Henry and so many others and totally unable to relate to their American values.

Greg Goodman
Reply to  Ian W
April 23, 2020 8:20 am

Indeed, the priority now is unraveling this mess they have got us into as quickly as possible.

That is why I took the effort of providing an analysis of the progress of relaxation efforts in Italy. They are playing the role of crash test dummy once again in this.

We are starting to see the effects there NOW, so this can give some guidance on how to go without taking baby steps and waiting two weeks to see what happens at each step.


comment image

Clyde Spencer
Reply to  Ian W
April 23, 2020 8:51 am

You said, “Testing stations will not allow testing of anyone that has not already been symptomatically diagnosed as having COVID-19 …” Therefore, another way of looking at the correlations is how often the symptoms are NOT diagnostic. That is, for countries testing based on symptoms, less than 20% of the suspected cases are actually COVID-19! Symptoms alone give a very high rate of false-positives, reinforcing the need for reliable testing.

Reply to  Ian W
April 23, 2020 10:21 am

Ian and Clyde,

Well said.

Reply to  Greg Goodman
April 23, 2020 10:27 am

Response to Greg G:

Now suppose that you have a limited number of tests so have to reserve them to A&E admissions where the need to identify the pathogen is the most urgent.

Then you are testing only the most urgent cases, the most likely instances of getting a positive test result, leaving the less urgent population unaccounted for, including asymptomatic carriers of the virus or carriers of the virus with minimum symptoms, especially minimally symptomatic people who just brush it off and never seek out a test (I would be one of those types, if I came down with it).

This speaks directly to what Willis was writing about. On other non-COVID issues, I can be in serious disagreement with Willis, but on this issue, I find myself in complete agreement with him, and my own plots of my own state’s data are directly in line with his plots.

The more admissions you need to treat, the more tests you perform.

And that’s precisely the point — your tests are allocated ONLY for ADMISSIONS, thus overlooking those who are not admitted, who might carry the virus with minimal or no symptoms, thus, who NEVER seek out a test.

The more tests you perform, the more +ve you get. This correctly reflects the REAL state of the epidemic in that population.

No it doesn’t — it reflects the state of the epidemic in a limited, biased sample of the population, … biased in favor of those people who are most symptomatic, most worried about their symptoms, or who died and the virus was present at death. What about all the other people who are not worried about their symptoms, not symptomatic, or minimally symptomatic?

The correlation does not prove the stats are spurious. As we are constantly reminded CORRELATION IS NOT CAUSATION.

It does not prove that they are not either. Again, consider the previous points above.

What Willis seems to have missed is an uncontrolled variable, the link between the number of tests done and the number people showing symptoms.

Sorry, but that’s exactly what he is showing — the number of people showing symptoms who are tested, who turn out positive. Who are the people being tested? — people showing up at hospitals, people who are symptomatic and worried about it. Where exactly are the tests being done? — in locations specifically designed for symptomatic people or people worried about their symptoms, who actively seek out or are actively referred to testing.

What his US graph probably shows is that the number testing positive is a fairly consistent proportion of all people showing flu-like symptoms, reporting or taken to hospital.

What his graph shows is that the number testing positive is a fairly consistent proportion of people being tested in HOSPITALS — you say it yourself — and yet you say you do not agree. Your statement seems to be an exact agreement with what he is saying. So, I’m not understanding your objections.

Reply to  Robert Kernodle
April 24, 2020 1:57 am

Thanks for the detailed reply Robert. I think you are rather misunderstanding my point.

This post was about Willis’ straight line graphs which he claims prove case data is almost totally determined test numbers and thus is not worth even looking at. That is certainly not consistent with the highly structured nature of the Italian case data:

comment image

I am not saying that case data reflects the number of non symptomatic or low symptomatic cases in the wider community and what that implies about overall death rate. That is a different issue.

If A causes B and A causes C , B and C will likely be correlated. That does not mean that B causes C. Willis has plotted B and C , gets a strong correlation and concludes that B causes C and thus B is nothing to do with A.

He set out with a pre-declared opinion that case data was useless, found what looked like proof so obvious it made him laugh. He failed to question his own bias. That reminds us of Feynman’s famous ” the easiest person to fool is yourself”.

Matthew R Marler
Reply to  Steven Mosher
April 23, 2020 12:02 am

Steven Mosher: Not science

Don’t forget to add unknown false positive and false positive rates.

However, it is the “beginning of science”, rather than “Not science”. Hardly anyone in science has gotten anything right the first time.

We may face a “second wave”; the practice now will improve results during the second wave. That is not a negligible achievement, imo.

Reply to  Steven Mosher
April 23, 2020 12:25 am

Quite right. The testing is not random, and despite numerous requests, I cannot get the UK authorities to tell us what the sensitivity and specificity of the test is either.

So we have numbers that are completely meaningless.

Steven Mosher
Reply to  Phoenix44
April 23, 2020 1:38 am

Not completely meaningless.

think harder

Independent George
Reply to  Steven Mosher
April 23, 2020 4:47 am

Geez, it must be tough for you; being so much more intelligent than everyone else. We’re lucky we’ve got you here to enlighten us all.

Clyde Spencer
Reply to  Steven Mosher
April 23, 2020 8:58 am

BUT, it is not science. I have that on the best of authority! So, the philosophical question becomes, “Can the answer to a scientific question have meaning if the answer is derived from a non-scientific method?” think hard

Robert of Ottawa
Reply to  Phoenix44
April 23, 2020 1:55 am

But very useful to those who wish to scare; not just politicians, some people must be doing alright, the èxperts`and `researchers`for example.

Reply to  Phoenix44
April 23, 2020 12:24 pm

as far as I can see the UK authorities don’t even seem to want to volunteer who’s doing the ‘testing’ either – of was supposed to be run by Public Health England ( which is a semi-detached NDPB bureaucrat stuffed “quango” ) – but beyond some numbers being tossed about – no details of who’s doing them or which tests are being run seems to be getting out.

Not impressed…..

Reply to  Steven Mosher
April 23, 2020 1:25 am

Thank you, you highlighted a major factor that I have not yet seen adjustment made for. ” … as tests grow, the disease may also be spreading”.
A second confusion is that not all tests are the same. As a non-epidemiologist, I am quite confused by what is actually being tested for, in various exercises.
Then there is the complication of false positives and negatives, There might be standard stats to cope with these, but Ihave not seen much more than a mention in passing. The measurement of false readings feeds back into what you wrote.
Geoff S

Steven Mosher
Reply to  Geoff Sherrington
April 23, 2020 1:43 am

from everything thing I have seen the test accuracy is not a first order impact.
Say it may lead to some single digit errors.

while test protocol… only test people with fevers who have travel to china histories
( the first criteria the CDC laid out)
is likely to give you order of magnitude errors.

with state data I do a chart of

tests per million
Positive rate.

quite illuminating about two different extreme results:

high positivity & low test penetration
low positivity and low test penetration

It doesn't add up...
Reply to  Steven Mosher
April 23, 2020 4:14 am

Try scattering test rate against death rate. There are no cases of high death rate and low test rate. A high test rate however means nothing about the death rate.

Low test rates occur because tests cannot be afforded or procured or are simply not really needed in populations where the prevalence is very low. Testing (as opposed to quarantine) does nothing to suppress deaths. It’s a WHO sticking plaster because they have nothing better to say.

Reply to  Steven Mosher
April 23, 2020 2:24 am

Let’s not just beat up on Mosher when he didn’t say anything wrong.

Reply to  Hivemind
April 23, 2020 8:10 am

I think the point is that in his climate science posts, Mosh takes the opposite position in regards to “no control, no consistency”.

Steven Mosher
Reply to  MarkW
April 23, 2020 5:28 pm


I recognize however that people here do.

Reply to  Steven Mosher
April 23, 2020 4:11 am

Holy smokes, i agree with Mosher 100% !

The true mortality rate is overestimated without representative random sampling. And the first patient to die with covid-19 in the US died Jan 19th (9th?), if memory serves, as a local news release in California just showed.

The idea that this virus was not already here this past winter is fiction.

Reply to  RW
April 23, 2020 5:56 am

personally know 2 people who had symptoms last week October and first week November 2019.
flu like symptoms that were really rough and…heres the kicker…loss of smell.
both middle aged fairly healthy and survived.

Reply to  dmacleo
April 23, 2020 6:42 am

Loss of smell in upper respiratory infections is not unique to C-19.

Reply to  Scissor
April 23, 2020 11:22 am

its the combo of all the symptoms…the exact same symptoms used to diagnose today.

thought that was clear in my post guess I need to make sure to clarify every concept when posting for obtuse and/or speed readers.
or for those who want to purposely muddy the waters.
and FWIW I am a speed reader and often make errors.

all kinds post here.
this is not an attack on you but a general statement about everyone that posts.
just to clarify…

Reply to  dmacleo
April 23, 2020 10:03 am

Living in Palo Alto, CA (silicon value central), I had a really tough flu in early mid-November 2019. Four days of fever peaking at 102 F (38.9 C), about 10 days of general malaise, and two weeks to recover after that. No coughing, not much sinus or lung involvement.

Some local siblings and relatives came down with the same symptoms in December. It was just as tough on them. My brother had been on a commuter flight to SF a few days before he got sick.

Was it early covid-19? I don’t know. But I’d never had a flu like that before, either.

We all are middle-aged or past and recovered.

Reply to  Pat Frank
April 23, 2020 11:32 am

my friends had 102 fever, dry cough, very weak and tired (took 10 days for one and 3 weeks for other to feel somewhat normal) and hurt all over. they also had the nasal component, runny nose (was oct/nov in Maine…) with sneezes and loss of smell and taste. should have clarified taste loss in org post I failed there.
they are unable to get tested to see if they actually had it or not.

I had a really bad cold (I thought) Nov 30 or so that lasted long time. fever was only 100 (for whatever reason I generally don’t get fevers like most) with dry cough and nasal issues. did not have smell/taste issues, but I was weak for weeks. however due to multiple disabilities I am often weak so I have just attributed my issue then to general crappy week(s). who knows. whuhan knows… 🙂

Dr Deanster
Reply to  Steven Mosher
April 23, 2020 6:54 am

With a 50%+ asymptomatic rate, and test being reserved for patients showing symptoms, the confirmed cases-test relationship is rubbish for determining the actual number of infected individuals. Period, end of story.

Reply to  Steven Mosher
April 23, 2020 7:58 am

“No controls. No consistency Not science No conclusion.”

LOL… given the state of climate science and the models, this really is serving up a softball.

Crispin in Waterloo
Reply to  mcates
April 23, 2020 10:42 am

The parallel between the modeling of SARS-Cov-2 spread is quite similar to the efficacy and skill of climate models, particularly those produced in Canada (U-Vic).

Reply to  Steven Mosher
April 23, 2020 10:30 am

Still waiting for the CDC to do perhaps 10 randomized or representative tests of ~1000 randomly chosen people in random parts of the country. 10,000 test kits. Real data of the actual cases, non-symptomatic cases, cases with symptoms, # of hospitalizations, and type and incidence of serious or lethal complications.
Any of the major drug companies could fund this out of pocket change in their budgests.

Reply to  Steven Mosher
April 24, 2020 10:45 am

[bad email address – mod]

John Andrews
April 22, 2020 10:21 pm

My logic says the number of new cases vs the number of tests is a function of the maturity of the development of the disease in the population. If and when a vaccine becomes available the number should drop because of fewer new cases. I guess we will see what happens.

Reply to  John Andrews
April 22, 2020 10:37 pm

My logic says that if the proportion of people in the population who have been infected increases over time, then (assuming random and accurate sampling) the proportion of tests that are positive should increase over time.

Is there enough good data to be sure about anything?

Reply to  Willis Eschenbach
April 23, 2020 2:03 am

Perhaps you misconstrued, Willis.

Random – a truly random sample, rather than those who turn up in a hospital, or volunteer to be tested.

Accurate – testing done by professionals, and not self administered.


Reply to  ralfellis
April 23, 2020 7:17 am

We’re certainly not there (yet), but self reliable administered tests is where we need to go.
And there is no reason we can’t get there – several established examples in circulation for decades, think diabetes.

Greg Goodman
Reply to  Willis Eschenbach
April 23, 2020 2:55 am

Willis is correct, one thing this is not is random testing.

Under the sever lack of test kits available testing is mostly to hospital admissions, where they test those displaying “corona-like” symptoms to detect if they are dealing with a COVID case.

That preselection bias is almost certainly the reason for the straight line graphs. The more suspected case you have turn up, the more tests you do. Number of tests is not the independent variable it is the dependent variable.

The proper conclusion to draw from Willis’ graphs is probably : correlation is not causation.

However, having a critical eye and having these questions thrown up is very valuable. The data are messy and may often be biased by such confounding variables.

It doesn't add up...
Reply to  Mike Jonas
April 23, 2020 4:23 am

Since the epidemic reaches a peak and then falls back, the proportion testing positive will fall once the peak is passed unless the testing regime is changed. If spare tests are used to test more widely, the peak in acute cases gets masked by the rise in detected mild and asymptomatic cases caused by the change in test regime. But that too will peak (or may have already peaked). So the answer is no.

Following acute (i.e. hospitalisable) cases is probably as good an indicator as we have, in that it clearly leads deaths by some days, and given that effective total population testing to try to identify infections as soon as they occur is simply impossible.

Reply to  John Andrews
April 23, 2020 12:27 am

Not true. You can model an “infection” rising and then falling simply by modelling your testing. It’s a simple exercise on Excel. Try modelling say people with a Y chromosome through non-random testing that increases each day. Lo and behold theres a Y infection!

Clyde Spencer
Reply to  Phoenix44
April 23, 2020 9:04 am

You said, “Lo and behold theres a Y infection!” Militant feminists have maintained this for years. 🙂

Steven Mosher
Reply to  Willis Eschenbach
April 22, 2020 11:16 pm

Theres a csv
At lunch now

Zig Zag Wanderer
Reply to  Steven Mosher
April 23, 2020 12:04 am

Why is the CSV at lunch?

Reply to  Zig Zag Wanderer
April 23, 2020 8:16 am

Would your csv like a blt?

Steven Mosher
Reply to  Willis Eschenbach
April 23, 2020 1:35 am

Here csv


sorry I was at lunch and phone was running low

Steven Mosher
Reply to  Willis Eschenbach
April 23, 2020 1:50 am
Craig from Oz
April 22, 2020 10:24 pm


“Conclusion? Don’t use confirmed cases as a metric of the spread of the virus—the number of cases is indeed a function of the numbers of tests. ”

Can you expand on this? I accept your conclusion that the more you look the more you find, but having made the suggestion that confirmed cases is not a useful metric, what IS a useful metric? Or is there a useful metric at all?

Not questioning your analysis, more looking additional ‘useful’ information for my own (selfish and greedy) reasons.


Reply to  Craig from Oz
April 22, 2020 10:37 pm

Confirmed antibodies seems like to be the most useful.

Between the antibody tests which have been done, dubious as some of them are, and the examination of sewage for the virus, it seems to indicate that a large percentage of the population have probably had this with no serious effects.

Reply to  Willis Eschenbach
April 22, 2020 11:23 pm

“Now, COVID-19 deaths include those where it is SUSPECTED that the virus MIGHT have CONTRIBUTED to the death “

Snopes looked at this.
“Are CDC Guidelines for Reporting COVID-19 Deaths Artificially Inflating Numbers?”

The relevant CDC guideline is:

“In cases where a definite diagnosis of COVID–19 cannot be made, but it is suspected or likely (e.g., the circumstances are compelling within a reasonable degree of certainty), it is acceptable to report COVID–19 on a death certificate as ‘probable’ or ‘presumed.’ In these instances, certifiers should use their best clinical judgement in determining if a COVID–19 infection was likely. However, please note that testing for COVID–19 should be conducted whenever possible.”

Reply to  Nick Stokes
April 23, 2020 12:02 am

I think people forget the reporting can only be done by doctors who have no interest in a political agenda. They don’t get extra pay or incentives for declaring someone died from something they didn’t.

Zig Zag Wanderer
Reply to  LdB
April 23, 2020 12:11 am

I believe they do in the USA, or at least the hospital does.

Reply to  LdB
April 23, 2020 12:12 am

Well then it’s a US problem stop trying to make it out as something world wide.

Reply to  LdB
April 23, 2020 12:32 am

Nonsense. Many doctors are political. Why would they not be? We have numerous activist doctors in the UK. Are there none in the US?

And that’s before we get to unconscious biases, which have always been an issue for death certificates – numerous studies show that doctors de late a cause of death based on defaults and diseases that are prevalent at that time.

Paul Jenkinson
Reply to  LdB
April 23, 2020 5:09 am

LdB,I work in the business and I can tell you the cause of death is often a guess made by the certifying doctor.

Reply to  LdB
April 23, 2020 6:10 am

Its one the main reasons for the difference between say German numbers and those of other countries. Most doctors left to decide themselves whether or not to use the new advise from WHO will use the previous method, ie a heart attack caused death irrespective if CV-19 was present ( Germany). If the national health body insists their employees use the new definition ( ie UK) they will. US hospitals get paid a significant $ for each cv-19 patient and more for ICU patients, so guess what that means…
The only statistic that is stable, without manipulation is the overall morbidity stat. This should be used for strategic planning.

Reply to  LdB
April 23, 2020 6:45 am

April 23, 2020 at 12:02 am

It is very simple.

The “doctores” do not do tracking and monitoring of every freaking flu virus out there during a season.
Aka flu death numbers listed as seasonal flu, non specified.
“Doctores”, can not check for every type of flu…only do follow only the prevalent one during the season.
This season the seasonal flu very suddenly closed shop, boom…globally.

Because COVID-19 became the prevalent infection-disease to follow… globally.
The overall season flu deaths will be listed specifically to COVID-19…
as the rest can not be tracked efficiently under the load… globally.

It is not about politics and agendas in principle,
is about how it is done and how it works… the rest is only exploitation on top of it.

So if there is no any considerable overall death increase for this season, it will be like a seasonal flu death, even when the scale of infection wider, very specific even in global term,
and quite concerning and “alarming” due to it’s sudden appearance allover the place as a new unknown infection-disease.

There is no any infection-disease, that ever had or has the record of such a sudden allover the place appearance globally, like this one…
zero to hero globally in a “boom”, within 2 months period, every where.


Reply to  LdB
April 23, 2020 8:31 am

But it isn’t working like that we test thousands of cases of flu a day, death from covid19 is a tiny number. Second what the death rate is or isn’t in US, UK or Italy have no bearing on us we were locked down when we had near zero deaths it simply isn’t a factor. If things were better than thought they would simply lift some restrictions and that would be welcomed. I can’t speak for all countries but it is a bit of stretch to suggest that is universal.

I guess even if I took countries I don’t know I can’t for the life of me figure what forging figures to lock down a country that doesn’t need to be achieves. Call me always a skeptic but I would need proof and at least some plausible reason.

Reply to  LdB
April 23, 2020 9:03 am

FWIW – Here’s a comment from a Minnesota physician and republican state senator


“Right now Medicare has determined that if you have a COVID-19 admission to the hospital you’ll get paid $13,000. If that COVID-19 patient goes on a ventilator, you get $39,000; three times as much. Nobody can tell me, after 35 years in the world of medicine, that sometimes those kinds of things [have] impact on what we do.

“Some physicians really have a bent towards public health and they will put down influenza or whatever because that’s their preference,” Jensen added. “I try to stay very specific, very precise. If I know I’ve got pneumonia, that’s what’s going on the death certificate. I’m not going to add stuff just because it’s convenient.”


Reply to  LdB
April 23, 2020 10:38 am

Salute !

Thank you rickk

If your $ numbers are verified, then its another reason why our “semi-socialized” Medicare system is breaking the bank. I cannot see where the extra effort to use one of those “magic” machines can cost many times per hour of medical worker costs.

Gums wonders….

Zig Zag Wanderer
Reply to  Nick Stokes
April 23, 2020 12:10 am

When you pay people per Covid 19 case, you’ll get a lot more Covid 19 cases. The same happened with skin cancer in Oz, and everyone thinks we’re all dying of skin cancer. Quote from an Australian doc at the time “It’s the easiest way for me to earn $50.”

Reply to  Nick Stokes
April 23, 2020 12:12 am

The same holds here in the U.K. where patients in hospital are tested positive and subsequently die, then CoViD19 is put on the death certificate as a contributing cause.

It doesn't add up...
Reply to  John
April 23, 2020 4:29 am

The total mortality data are showing excess deaths exceeding the number of Covid attributed deaths. It is as yet unclear whether that is due to under reporting particularly outside hospitals, or whether it is a rise in deaths from other causes because people are not getting treatment.

Reply to  Nick Stokes
April 23, 2020 12:29 am

So…? The criteria could be inflating the numbers. Doctors are human, they will have a bias towards COVID at the moment.

Ben Vorlich
Reply to  Willis Eschenbach
April 22, 2020 11:32 pm

Is that because they need the highest death toll possible to justify draconian measures?

Reply to  Ben Vorlich
April 23, 2020 12:11 am

take a couple of countries Australia only has 78 deaths, New Zealand 16, so lets play your game and they faked them all. Yet both countries are in lockdown.

Your argument doesn’t work so perhaps it’s only true in your country 🙂

Reply to  LdB
April 23, 2020 3:16 am

I am not sure I follow your argument either. The 2 countries you cited have draconian measures. Ben just stated they needed the “highest Death toll possible.” 2 could be the highest possible 🙂

Reply to  LdB
April 23, 2020 4:00 am

So 2 is all you need to justify a lockdown then, so USA was justified in the lockdown make sure you tell Willis that I am sure he will agree.

Reply to  LdB
April 23, 2020 4:24 am

and almost all were OS travellers on cruise ships students and a very few community transfers.
and regional areas are almost unscathed as we banned tourism caravan parks etc very quickly so coty folks didnt spread it out into the boonies:-)
in 6 weeks the very large west vic has 15 or less intotal from Ararat outwards, even horsham had only 3, and the poor hospital staff there were having severe anxiety as modellers….predicted 300 a day could be incoming.
most of the people here apart from community sport dont do the social large crowd things very often, local shows field days and fishing comps would have been the other risks and they got canned as well.

Farmer Ch E retired
Reply to  LdB
April 23, 2020 4:48 am

LdB – is it reasonable to compare Southern Hemisphere C-19 case data with Northern Hemisphere data? They are 6-months apart related to flu season.

Reply to  LdB
April 23, 2020 5:41 am

Derg writes

The 2 countries you cited have draconian measures.

I dont think I’d class Australia as having draconian measures. People (eg Tradies) are still going to work. People (eg Office workers) who can work from home, do so. The draconian part would be – there is no “unnecessary” travel, no going to friend’s places and no large gatherings, not even weddings or funerals.

People accept social distancing for the most part and lots of sanitizing happens.

Australia acted early and has the right balance for controlling the spread of the virus. It was also probably lucky to control it so well.

Without a vaccine or effective treatment, Australia has backed itself into a virus free corner.

Reply to  LdB
April 23, 2020 6:08 am

Farmer Ch E retired we are entering our flu season the flu vaccine for this year is out and being administered.

Now to put that in perspective you can only get a test done if you have flu like symptoms or are a front line worker … we have no random testing. So numbers for yesterday 14,218 tests and 12 positive. Even if you allowed some stupidly huge number like 4000 for frontline workers that means over 10,000 people with the normal flu fronted for testing.

I don’t know how that compares to countries in Northern Hemisphere but we are definitely testing massive numbers of ye olde flu.

Reply to  LdB
April 23, 2020 6:53 am

April 23, 2020 at 12:11 am

Summer time in AUS and NZ… not the flu season… not yet.

If these two nations really facing the COVID-19 wave now, then you see how dangerous is this new disease…78 or 16 deaths during this new disease wave.


Farmer Ch E retired
Reply to  LdB
April 23, 2020 8:29 am

LdB – We’ll need to wait and see once SH flu season is in full swing. Fewer deaths may be related to SH summer season.

Reply to  LdB
April 23, 2020 8:37 am

It’s Autumn and it’s the start of our flu season .. we can hardly get this wrong the government has been running the flu shot adds for weeks now.

Clearly you don’t live in Australia and must think we make this stuff up … so here this is the program

As per the announcement all age care workers MUST be immunized by 1st May so 8 days from now.

Reply to  LdB
April 23, 2020 8:44 am

Farmer Ch E retired I would be very surprised if we don’t have similar numbers of deaths to Flu as per last year around 1000 … so I guess that is our benchmark. So we fully expect it to dwarf our Covid19 deaths unless we lose control. Explain what you think should happen?

Clyde Spencer
Reply to  LdB
April 23, 2020 9:12 am

Yes, the Aussies and Kiwis may be singing a different tune in 2 or 3 months.

Reply to  LdB
April 23, 2020 9:24 am

Clyde what do you think will happen make a prediction?
I am trying to get a feel for what you think will happen?

My only personal concern is when we lift the lock down up people go silly and we then get slammed back into a level 4 lock down like happened with Singapore.

Reply to  LdB
April 23, 2020 9:34 am

April 23, 2020 at 8:44 am

So we fully expect it to dwarf our Covid19 deaths unless we lose control. Explain what you think should happen?

There is still a problem there you not realizing.
The rest of the world, is trying a get out of the lock downs and heavy restriction.

If there is sign or indication of considerable increase of infection and COVID-19 disease
in AUS, the world will not care at all about the control and dwarfing of COVID-19 deaths.
It will isolate and restrict your country, regardless, unless there is no indication of infection risk.

There better be no increased infection risk, either because it will not happen,
or better you get a firm control on hiding it, if it does.

Same I think goes for NZ too.

The vaccines will not help you much there, as vaccines cannot actually stop the infection or the disease… especially the overall seasonal one.

Flu vaccines have not stopped any season of flu.


Reply to  LdB
April 23, 2020 9:54 am

Whiten we fully expect to be almost isolated from the world for quite some time you won’t be entering Australia anytime soon without doing a 14 day isolation unless there is a vaccine. Tourism is the only main industry we have that will be in the air going forward.

Most people in Australia take the flu to try to reduce severity and time off work (often paid by employer) and not kill vulnerable people (many age home care facilities demand it). It isn’t sold to us that it will somehow stop a flu season. The prediction is the same as last year 13.5 million people will have the needle so close to 1 in 2 and no we don’t expect a mild flu season.

Clyde Spencer
Reply to  LdB
April 23, 2020 1:17 pm

You asked for a prediction. With the caveat that if COVID-19 turns out to be strongly seasonal, as other similar diseases are, then I see a strong possibility that someone (or several) will re-introduce CV in the Fall to one of your larger cities with international airports, and it will spread undetected for a few weeks because of the long incubation period, high frequency of asymptomatic carriers, and lack of immunity in the general population. After a couple of confirmed deaths, it will be firmly entrenched and the politicians will panic and reimpose restrictions to strangle the economy.

Australia and New Zealand may get lucky and an efficacious cure or effective treatment of symptoms will have been developed by then, allowing an early lifting of lockdowns.

If a cure or effective treatment isn’t available early, Australia and New Zealand may then see an epidemic not unlike what Canada, Norway, Denmark, and Finland are experiencing. In any event, without a vaccine, the lockdowns will mean Australia and New Zealand will remain vulnerable to future flare-ups of the virus.

The development of an effective coronavirus vaccine is problematic. If it turns out to not be possible then, as island nations, you have a couple of choices: Isolate and forego tourism, or allow a ‘controlled burn’ to develop immunity.

Reply to  LdB
April 23, 2020 6:47 pm

I agree without a cure Autralia and New Zealand will have to keep 14 day isolation for anyone entering … that is definitely a given.

Farmer Ch E retired
Reply to  Ben Vorlich
April 23, 2020 9:08 am

LdB – imo what should happen is that we should wait and see – and prepare. I’m not saying I agree or disagree w/ the AUS and NZ approach. In late January, Trump’s travel ban from China was very controversial. Not so much anymore now that the MSM has developed a severe case of amnesia about the China travel ban.
In the SH, time will tell if the actions were too much or not enough as our understanding of C-19 continues to evolve. Yogi Berra was spot.

Reply to  Farmer Ch E retired
April 23, 2020 9:37 am

I really have no view on US, your country social safety nets and your whole economy is very very different to ours. At the end of the day all you hope is that governments act in the interests of the country. I certainly don’t think US should mindlessly follow the lock down path but I also think it is wrong to criticize countries who decide that path.

Farmer Ch E retired
Reply to  Farmer Ch E retired
April 23, 2020 10:02 am

“I certainly don’t think US should mindlessly follow the lock down path but I also think it is wrong to criticize countries who decide that path.”

Ldb – I agree – my initial comment was related to comparing C-19 cases and responses in the NH with the SH since they are 180 degrees out of sync related to seasonal flue.

Reply to  Farmer Ch E retired
April 24, 2020 12:33 am

My impression is that Australia wasn’t trying for eradication. It may be happening (this is not guaranteed because of the asymptomatic cases) because we were too effective at tracing imported cases and isolating. Now we seem to be stuck waiting to see if a viable treatment develops.

So, maybe only NZ tourists for a couple of years, but presumably antibody testing will soon be reliable enough to admit tourists from elsewhere. I’m not sure how governments here are going to be able to enforce lockdown outside of the large cities much longer.

Reply to  Willis Eschenbach
April 23, 2020 12:52 am

Ah well. Sometimes big numbers lead to a bigger grant.

Robert of Ottawa
Reply to  Willis Eschenbach
April 23, 2020 2:03 am

Probably because they were seeing their predictions of imminent disaster falter as time went by, so they loosened the criteria. I think there must be many people, the panic mongers, the wolf-criers, the chicken littles, the charlatans, who are hoping it is worse than is slowly becoming apparent.

Where have we seen this behaviour before. Climate catastrophy anyone.

Reply to  Willis Eschenbach
April 23, 2020 4:21 am

Willis is correct. I’ve read the CDC published PDF Stokes takes his quote from via Snopes. The issue is that the CDC is including those probable and suspected cases of cocid-19 listed under “underlying cause of death” and is including them in their covid-19 death counts. The CDC also says that their definition is just one of many different definitions that are out there and that their numbers will likely differ from other sources. That said, it seems fair to assume that the models use the CDCs mumbers, which is very xoncerning! (Which is really what Willis’s point was about)

Reply to  RW
April 24, 2020 3:57 pm

Willis is asserting what has been the experience of numerous contagions. For example, in the US after the Swine flu in 2009, analysis was done to estimate that infections were many multiples of confirmed cases. Carrie Reed et al. reported at CDC Estimates of the Prevalence of Pandemic (H1N1) 2009, United States, April–July 2009.
“Through July 2009, a total of 43,677 laboratory-confirmed cases of influenza A pandemic (H1N1) 2009 were reported in the United States, which is likely a substantial underestimate of the true number. Correcting for under-ascertainment using a multiplier model, we estimate that 1.8 million–5.7 million cases occurred, including 9,000–21,000 hospitalizations.”
“Using this approach, between April and July 2009, we estimate that the median multiplier of reported to estimated cases was 79; that is, every reported case of pandemic (H1N1) 2009 may represent 79 total cases, with a 90% probability range of 47–148, for a median estimate of 3.0 million (range 1.8–5.7 million) symptomatic cases of pandemic (H1N1) 2009 in the United States.”
Paper is here: https://wwwnc.cdc.gov/eid/article/15/12/09-1413_article#tnF1
My synopsis is https://rclutz.wordpress.com/2020/04/23/crash-course-in-epidemiology/

It doesn't add up...
Reply to  Willis Eschenbach
April 23, 2020 4:34 am

Deaths are a lagging indicator. Whilst you have clearly demonstrated that there are big differences in testing regimes, within a particular locale, so long as the testing regime is kept reasonably consistent, positive tests at least allow peaks in the local epidemic to be picked up much sooner. Of course, it says nothing about general prevalence, and we know even less about the circumstances and progression of mild and asymptomatic cases than those that become more serious. I’ve not seen that anyone is researching them.

Reply to  Willis Eschenbach
April 23, 2020 12:08 pm

Willis, can’t you just look at total mortality for the US?

The number of deaths above the average for the week has got to be pretty close to the number caused by Covid-19. The UK has this data released weekly (ONS data) and runs about 2 weeks behind. ie for the UK the week 3rd-10th April had 8000 more deaths compared to the 5 year average and the week 26th March-3rd April had 6000 deaths above the 5 year average. I would guess we won’t see more than 8000 excess deaths in the UK now.


Where do you get the same stats for the US?

Steven Mosher
Reply to  Willis Eschenbach
April 23, 2020 5:30 pm

“However, sadly the CDC recently changed the definition of a COVID-19 death. Now, COVID-19 deaths include those where it is SUSPECTED that the virus MIGHT have CONTRIBUTED to the death …

Nope they added a second category

Reply to  Craig from Oz
April 22, 2020 10:54 pm

Politically, and operationally from the perspective of managing the outbreak, the only metric anyone cares about is the death rate.

We can all be armchair academics interested in theoretical models of the disease, and bemoan the lack of clear data to validate them, but the crunch figure for anyone in charge of medical facilities or government policy is how many are dying, which way their families vote, and what can be done to lower the figure.

We may not be able to test the all living, but we should be able to test the newly dead.

Reply to  Leo Smith
April 22, 2020 11:04 pm

You can’t tell the death rate unless:

1. You know how many people caught it.

2. You have accurate death numbers for people who died of Chinese Flu, and didn’t just happen to die while infected with Chinese Flu.

Given that deaths from other diseases like cancer and heart disease have dropped off the charts in some areas with heavy Chinese Flu infections, it seems pretty clear that many have been counted as Chinese Flu deaths instead.

Reply to  MarkG
April 23, 2020 1:09 am

One way round that, for statistical purposes at least, is to do a separate attribution that assumes that the heart disease/cancer deaths are in line with averages and reduce the severity of the pandemic numbers accordingly.

This obviously does nothing for the actual number of deaths but might, if the numbers are properly crunched, provide useful information on the severity of the pandemic virus. Would these same people have died in a bad ‘flu year for example? And if so would influenza have been recorded as the cause of death?

If a man has a severe heart attack while swimming and drowns what goes on the death certificate??

Reply to  Newminster
April 23, 2020 4:23 am

“If a man has a severe heart attack while swimming and drowns what goes on the death certificate??”

An infant in Conneticutt drowned and tested positive postmortem. The governor of the state publically lamented how the infant died from covid and that no one was safe. He was exposed as a liar after the family talked with a social influencer. The coroner subsequently refused to register the death as being from Covid.

Reply to  Craig from Oz
April 23, 2020 1:31 am

Heart and everything else burnt out plus old age just might ,repeat MIGHT change all

Reply to  Craig from Oz
April 23, 2020 8:56 am

“Can you expand on this? I accept your conclusion that the more you look the more you find, but having made the suggestion that confirmed cases is not a useful metric, what IS a useful metric? Or is there a useful metric at all?”

We are sort of looking for Rolling Stones fans outside the stadium where the Stones are playing tonight…you are going to get lots of positives. You will get some non fans, say those invited along for the evening. You will also get those just walking by in the neighborhood. But this is not a measure of how many Stones fans there are in that community. You need to test ‘randomly’ in a ?10 block radius – then see what the results are of how many Stones fans there are in that community.

Current Wuhan virus testing of those that show in front of a healthcare provider with slight symptoms, or with real symptoms and perhaps associated family thereof, will reveal more positives. You need to test wider – say a 20 block radius – all comers – then you get a better idea of prevalence in the community. You need to do this across your county, state, province etc to have meaningful numbers (denominator).

That said, you also need an accurate test that tells you who really has when they really have it. Right now, some of the estimates say our tests are only 70% accurate (meaning we are telling 3 people out of every 10 that don’t have the virus when they really do).

John F. Hultquist
Reply to  Craig from Oz
April 23, 2020 9:19 am

Craig from OZ asks: “. . . what IS a useful metric? ”

Early this year in Washington State an elderly care facility presented a tragic beginning to this issue. Three to 7 people per month routinely died. Medical responders, staff, family and friends were coming and going as usual. A facility-wide Mardi Gras party was held. The next day, from various means, it was recognized that the virus was the cause of a spike in deaths above the normal 3 to 7.

In Italy a Champions League match (Feb. 19th) – Atalanta versus Valencia – brought many thousands of soccer fans to Milan and 40,000 in the San Siro Stadium. Many watched from crowded bars. Parties followed. Then people went home.

Such situations ought not to be used as templates for disease progression.
We live in a small population, large area, county. To date, there have been only a few “cases” (under 20) and Zero deaths.
The ancients found Zero a difficult concept.

April 22, 2020 10:40 pm

https://www.youtube.com/watch?v=MuuA0azQRGQ Willis I’m with you 100% one of the few sane voices. Lockdowns are not working anywhere this virus spreads like the flu re Sweden. Why do we not do lockdowns for the flu??. My guess is that when the antibody tests comes out nearly everybody in NY will test positive meaning that the mortality rate will be well BELOW the common flu. However Cuomo will not allow the release of that data if true for political reasons my 2 cents worth cheeers!

Reply to  Eliza
April 22, 2020 11:40 pm

Usually, when facing the seasonal flu, a part of your population already have some immunity and there might be vaccines available to those who have a compromised immune system.

In this case, with whole countries shutting down due to overflowing hospitals, we are no longer facing similar conditions as a typical flu season.

You mentioned Sweden. Well, here we haven’t shut down schools and people are still allowed to move freely about, BUT: People behave differently. People work from home (when possible) and the kindergartens are on high alert (you got the sniffles? Then both you and your siblings will stay home for a couple of days!). Our kids now come home washing their hands like they are about to perform open heart surgery. Our consumption of soap has risen sharply.

I live close to the Norwegian border. Our closest shopping center shut down their parking garage and reduce opening hours to roughly half of what it used to be. With Norway shut off, there are no plagueridden Norwegians riding into town anymore. Hence no business. The only Norwegians present are those of us who live here permanently (and now cannot visit our relatives on the other side of the border). Despite the empty shops, we still struggle finding fresh yeast in the store shelves. (again: people here act differently than they used to!)

People are encouraged to keep their distance when socializing. I haven’t visited a bar or restaurant in ages, but I’m told most people respect this advice.

Most of these precautions were introduced at an early stage.

But to do a complete shut-down is ridiculous. Norway went too far when they closed the schools. But OTOH Norway, I guess, have less hospital capacity than Sweden. Sweden has a reduced capacity compared to 20 years ago AFAICT, but is still ahead of Norway. It is my belief that Norwegian politicians have a guilty conscience about this, and are overcompensating as a result.

Anders Valland
Reply to  Rune
April 23, 2020 1:52 am

Sweden is hit relatively hard when it comes to the elderly part of the population. There is an interesting situation developing where Sweden has chosen a less restrictive policy on Covid-19 compared to Norway. At the same time, the current numbers in Norway are 1/10 of the Swedish numbers when it comes to mortality. The testing regime is fairly similar, and Norway has about 1/2 the confirmed cases of Sweden. This is not surprising, given that there are twice as many people in Sweden than in Norway.

The difference between the policies will be apparent in the long term. It might end up with similar numbers, or it might stay the way it is. At this point in time, no-one knows.

The policy of shutting down society is being debated in both countries. One of the many fall-outs of the Norwegian restrictive policies is now showing when restrictions are slowly lifted on schools and kindergartens. The population has been told that shutting these down was an important part of stopping the virus. Now they are opening, and that is because there is no rational, medical evidence that schools and kindergartens are important arenas for the spread of the virus (there never was any such evidence). But the anxious will not believe this, and according to polls about 10% of parents will not send their kids back to school or kindergarten.

When you scare people there will always be those who remain scared. And we are more and more a society of the anxious – I’ve always thought that we are not in the (I know it was not accepted) anthropocene but the anxiouscene – the age of the anxious.

I guess (hope) you were toungue-in-cheek when you wrote of the “plagueridden Norwegians”, since currently it is the exact opposite ;-).

Reply to  Anders Valland
April 23, 2020 2:38 am

It is my understanding that in Sweden, when discussing mortality rates, they count everyone who had covid-19 at the time of death.

In Norway they try to only count those who likely passed away due to covid-19.

So that accounts for some of the difference and further invalidates the strict approach to a lock-down. Someone else commented further down that a typical old folks home in Norway houses, on average, less people than a typical old folks home in Sweden. It sounds plausible. Ten on average sounds a tad low, but 20 feels about right. (I do not currently have any old relatives at this point, my dad is 70 and quite fit)

“The population has been told that shutting these down was an important part of stopping the virus.” Indeed! Actually, my recollection is that at the time, the health authorities said “meh” while the politicians said “we must act now!” (and “think of the children!”).

My impression is that Norway took a more emotional approach to their decision making, while Sweden approached this much more rationally.

Norway still has one advantage: A sizable pension fund that can be squandered on rebuilding the country. We could afford the experiment. (but queue the climate activists who argue Norway should remain locked down and not resume oil production)

Reply to  Rune
April 23, 2020 3:21 am

Norway is interesting. Enormous wealth due to oil. I wonder what impact there is now due to a lower price for oil?

Reply to  Rune
April 23, 2020 3:58 am

@Derg: The highest unemployment rate since the war.

I believe Norway will soon be hit with a collapse in the real estate market. Many unemployed people will have to reevaluate their living situation and find cheaper housing.

My hope is that the government will take this opportunity and invest in infrastructure. Plus move more government departments and offices out of the crowded capital. If there was ever a time to do that, it is now. And if I may be so bold: Reduce the focus on public transport. Let people ride in their own private (and virus free) cars. It is time to forget the socialist dream of living in one giant anthill.

The tourist industry is rotting away. Many hotels were closed before what should have been the high point of their main season.

Newspapers focus on hairdressers being shut down, but IMO the focus should be to preserve the kind of industries where it is difficult to get wheels rolling again when we return to a more normal situation.

Reply to  Rune
April 23, 2020 10:12 am

Sweden’s policy may lead to early herd immunity at the cost of more deaths in the short run, but with far greater preservation of their economy and way of life. This may be a wise choice. Time will tell.

Reply to  Rune
April 23, 2020 6:00 am

Actually Norway has more beds (3.6) per 1000 persons than Sweden (2.2), which has fewer than the UK (2.6).

Steven Mosher
Reply to  Eliza
April 23, 2020 5:32 pm

“My guess is that when the antibody tests comes out nearly everybody in NY will test positive meaning that the mortality rate will be well BELOW the common flu. However Cuomo will not allow the release of that data if true for political reasons my 2 cents worth cheeers!


Cuomo released. you are wrong

April 22, 2020 10:41 pm

Well done Willis.

Testing sick people showing symptoms tells us only so much. To find the number of sick people in our country, and to reveal the fatality rate, two other things should happen:

1) Random testing, which will show how many are infected with this disease which frequently is asymptomatic, and,

2) Deaths, from the disease, not those with co-morbidities, in other words Really Sick People who happened to die, after becoming infected, are not a Covid 19 Death. The Death Certificate is a hugely important thing.

Stanford study, USC study, showed maybe 40 to 80 times more with the infection than confirmed tests. This means maybe 40 to 80 times LESS fatalities. Of course, I am frantic to get off my couch, but, this is the real science, not the “science” from risk-averse Public Health officials who will only state worst-case.

Apparently CDC has mandated, if you Had it when you died, you died From it.

This is a lie. THIS IS A LIE!

The lock-downs, 22 Million Americans on Unemployment from the Lock-Downs, are based on this decision by the CDC.

The disease is bad enough, let us not make it far Worse because the medical professionals give us Worst-Case, instead of the Truth. I am 61, quickly becoming an at-risk male, but no co-morbidities, feel fine, still working out daily.

Let us quarantine the at-risk, elderly, elderly plus diabetes, obesity, Hyper-Tension, heart disease, and as you and I so desperately want, Put Healthy People Back To Work!

Solidly on your side tonight…


Reply to  Michael Moon
April 23, 2020 4:03 am

The antibodies tests produce significant rates of false positives, depending on the particular test at a rate somewhere between 1 and 10%. Even at the low end of this error rate the results of the Santa Clara study are meaningless, across 3300 tests they got 50 positives, even at a false positive rate of just 1% they’d get 33 false positives, at a false positive rate of 1.5% there might not have been a single true positive in the entire study.

Reply to  Andrew_W
April 23, 2020 7:31 am

Stanford and Southern Cal both checked carefully for false positives, found none.

Reply to  Michael Moon
April 23, 2020 9:51 am

Michael Moon April 23, 2020 at 7:31 am

From the Abstract: “We also adjust for test performance characteristics using 3 different estimates: (i) the test manufacturer’s data, (ii) a sample of 37 positive and 30 negative controls tested at Stanford, and (iii) a combination of both.”
(i) is not a scientific test, it’s an act of faith, (ii) is ridiculously as a check on a test that needs to be better than ~99.5% accurate on not giving false positives.

Reply to  Andrew_W
April 23, 2020 7:49 am

April 23, 2020 at 4:03 am

Extraordinary math.

Since when 33 happens to be 1% of 50, when 10% happens to be 5!!!!!
Is false positives, not false negatives.

Why did you not consider the 10% false positive of the test instead?
According to your math you would have proved once and for all, beyond any doubt, with your brilliant math, that antibody test of any kind is just irrelevant and a joke… and not applicable or useful.


Reply to  whiten
April 23, 2020 9:55 am

whiten April 23, 2020 at 7:49 am

Extraordinary reading comprehension.
” . . across 3300 tests they got 50 positives, even at a false positive rate of just 1% they’d get 33 false positives, at a false positive rate of 1.5% there might not have been a single true positive in the entire study.”
All very simple.

Reply to  Andrew_W
April 23, 2020 11:28 am

What do you think is the false rate for any test at all out there?
It ain’t zero, mate.

Do you think the virus test for COVID-19 has 0% falsity?
Or do you know of any test at all that has a better than 99% accuracy?

1% there is the test error itself mate, same or maybe even better than the viral test error.

I am sure you do not understand how such tests work.

Virus test is a diagnose test for the disease, not 100% accurate also there.
No any direct possible connection to the infection disease clause, not any need to validate it for or in relation to the infection… still misses a lot of infected outside the window of disease detection… misses a lot of asymptomatic that have passed the disease.
But still very good in what it supposes to do if at 1% falsity.

The antibody test a detection test for infected, also not 100% accurate there.
Due to the test needing to be validated by the disease condition, we get the relation of the disease to infection, the 10% false positive.
Accuracy of diagnosing for the condition of the disease 80-90%, when accuracy for detection of the infection 98-99%.

The primary function for antibody test, detection of the infection,
secondary, disease relation detection in consideration of the infection.

Virus test only one function, diagnoses of the disease, more accurate for confirmation of the disease, kinda of 98-99% at the best case… no good at all for the infection…
especially in the case of a new infection-disease.
100% miss of infection outside the disease window… 100% miss of the disease also, outside the disease window.

In consideration of a virus test, the number of infected will be a guess estimate, also the true number of people that have passed the disease, depending on an assumed factor, which in the case of a new infectious disease will be like pulling a rabbit or two or more from a hat.

But in proper application of an antibody test, identifying the scale of infection is quite very accurate, and the scale of disease still directly assessed, by a real established factor… not a purely assumed guess estimate.

Application of both tests properly, gives a much much clear accurate real picture.

Your math really bizarre, as there no any 100% accurate test in existence, unless in the consideration of Planet B of Billy Nay.
All very simple on that Planet, I am sure of it, but you see, I do not live there.


Reply to  Andrew_W
April 23, 2020 1:36 pm

whiten April 23, 2020 at 11:28 am
“What do you think is the false rate for any test at all out there?
It ain’t zero, mate.”

I’m glad to see you agree with me on that point, No doubt you realize that because the Santa Clara study was dependent on there being a zero false positive rate that that study is junk.

Reply to  Michael Moon
April 24, 2020 7:56 pm

“…Let us quarantine the at-risk, elderly, elerly plus…”

Are you promoting elder or comma abuse in the above suggestion?

April 22, 2020 11:00 pm

There’s a parallel relationship between the acceleration and deceleration of the daily cases, and the daily death curves:

comment image

One is a test, one is a corpse.

Harry Gee
April 22, 2020 11:13 pm

The stated purpose of the lockdowns – in all countries – seems to have been to slow the spread to avoid overwhelming the country’s health service. (If I’m wrong and some country decided to lock down for another reason, then please tell me).

Is there is an effective treatment for the disease? I HAVE READ THAT THERE IS NOT? There seems to be no agreement on what product(s) to use. There seems to be no agreement that ventilators actually help – I’ve read opinions that say they are actually worse than leaving the patient untreated.

So, if we can’t cure the infected, what is the sense in trying to “avoid overwhelming the health service ?”. Why not lift the lockdowns, let those who are going to die expire (hey, we can’t help them !) get the deaths over with and stop trashing the economy? Just asking…

Willis, I agree with your statement that the number of confirmed cases is a direct function how many you have have tested. It seems bleedin’ obvious. I seems to recall that Lord M disagreed with that idea, about 7-10 days ago. If he is reading, maybe he will re-iterate his thoughts. I’m willing to be shown to be an old fool – my wife calls me that all the time.

April 22, 2020 11:19 pm

As some of the commentator pointed out, the ratio positive_tested/test_number can be misleading if the sample is not randomly chosen; in many countries the people tested are already those with some symptoms. In the Czech Republic (we are dealing with the disease relatively well, at least so far, similarly as our neighbour Austria) we are now starting a big testing of ~30 000 people randomnly chosen to better estimate the real ratio of the COVID19 penetration. The results should be available in ~14 days.

Reply to  Milan
April 22, 2020 11:56 pm

Correct that is why his Australian data doesn’t work. To get what Willis is trying to do you would have to have a truely random selection of people for testing and no country is really going to waste what is a limited number of available tests. We have the flu going around in Australia at the moment so there is a hell of a lot of testing going on per day like 10000 tests for 8-12 detections of covid19.

Scott W Bennett
Reply to  LdB
April 23, 2020 5:36 am


They are not the same test! The most important test (Antibody/serology) testing is available en-masse from Australian and international companies right now.

It is illegal in Australia to test for antibodies (Serological testing) $20,000 fine in South Australia; just ask yourself why… If you really have to! ;-(

Two tests are are available, one can only tell you if you have the virus at the moment (PCR and that is best, well before 10 days after infection) and one can tell you if you’ve had it (Serological, that works best 10 days or more, after infection) neither test can tell if you will be sick, are sick or have been sick!

The current figures are legisltatively designed to sex-up / bias death rates. It is very well documented in long standing literature that testing only the symptomatic with single PCR tests leads to both to more false negatives and a greater than zero false positive figure, which inflates the death rate of any cohort (sample group).

To be clear, it is of very great importance that random tests** of a large sample of the asymptomatic population are done as soon as possible, in order to fully establish the virulence of this new “disease”.

It has been clear from the earliest date that we are all being gaslighted by the supposed “rates”of infection and death! ;-(

** Firstly using antibody testing and including PCR if available.

Reply to  Scott W Bennett
April 23, 2020 6:23 am

Initially I was scratching my head thinking you were making the slightly crazy claim that hundreds of thousands of Australians were infected and we never noticed 😉

However I assume I just don’t get what you are saying so are you saying the 78 deaths is sexed up? Given 90% of the cases we know the source (all from outside Australia) you can basically say we have had close to zero deaths from community transmission. I don’t get how changing the death number changes anything?

You probably need to have another go because I don’t see what you are saying.

Scott W Bennett
Reply to  LdB
April 23, 2020 7:25 am

Yours words:

To get what Willis is trying to do you would have to have a truely random selection of people for testing and no country is really going to waste what is a limited number of available tests. -LdB

Tests are limited only by government regulation not but availability.

Is that clear enough you gutless nameless troll.

Scott W Bennett
Reply to  LdB
April 23, 2020 7:42 am

Initially I was scratching my head thinking you were making the slightly crazy claim that hundreds of thousands of Australians were infected and we never noticed – LdB

Yes, hundreds of thousands of Australians have been infected and we have not noticed!

Why? Because we have not tested the infected we have only tested the symptomatic.

It amazes me that “you” can claim not to see this but perhaps my appreciation of the average IQ is too optimistic or perhaps you are just a bot.

Reply to  LdB
April 23, 2020 8:47 am

Ok you do believe that .. I don’t have a tinfoil hat .. no comment … leave you to it.

Reply to  LdB
April 23, 2020 8:57 am

Ok you do believe that .. I don’t have a tinfoil hat .. no comment … leave you to it.

Clyde Spencer
Reply to  LdB
April 23, 2020 9:31 am

What do you mean by “waste?” What good does it do to know if a particular patient has COVID-19 when there is no cure, only palliative care for flu-like symptoms? The available test kits would be better used with random testing to follow the development of the disease over time and get a better understanding of the percentage of asymptomatic carriers.

April 22, 2020 11:20 pm

May be, ingenius, but really not so much.

The linear shows that the medicine is well organized in the country.
If the plot is scattered, the medicine is a havoc, like in Australia.

The German medics have a clear narrative that the number of confirmed cases in testing must be less than 20% and more than 10%.
If you get more than 20% positive, you test more people with mild and irrelevant symptomatics.
If you get less than 10% positive, you have to test only those who were exposed.

They do not want to waste tests. Each test costs $300.
It is that simple.

Still wondering about the linear correlation?

Izaak Walton
April 22, 2020 11:39 pm

The graphs are linear if the rate of growth of COVID 19 is matches the growth in testing. Look at
Australia and NZ and you will see that the opposite. The more they test the few cases they find.
Over the past week NZ has substantially increased the number of tests it has performed and yet
the number of cases has dropped to 2 as of today.

Reply to  Izaak Walton
April 22, 2020 11:57 pm

Isolation works. Australia’s last high peak of daily new cases was 528 new cases, on the 28th of March.

There are 8 new cases today, 12 new cases yesterday and 20 new cases the day prior.

Sufficient Isolation for long enough ends it.

Reply to  WXcycles
April 23, 2020 12:05 am

They won’t believe you but look at the bright side we get armchair seats for the covid19 olympics.

Reply to  LdB
April 23, 2020 12:08 am

COVID-19 Para-Olympics


Reply to  WXcycles
April 23, 2020 4:16 am

What concerns me about the very “successful” Australian approach is .. what happens when we reopen our borders? Will Australia go through it all again while the “unsuccessful” countries are down to very low rates? IOW, is the success an illusion? Are Donald Trump and Jair Bolsanaro actually getting it right?

Too many unknowns.

Reply to  Mike Jonas
April 23, 2020 6:56 am

Amazes me how few people understand what quarantine is, how simple it is, and how reliable it is. I really would like to know why people presume we would be so silly as to let people into the country without serving a sufficiently long quarantine to prove they don’t have the disease before being allowed entry. Especially given almost everyone here is convinced testing will never work well enough. But quarantine sure will.

Somehow ‘critics’ never seem to have any reply, but will insist on repeating the exact same fake ‘question’ and play of consternation the next day, and pretend they never got a perfectly adequate answer the first time.

Reply to  WXcycles
April 23, 2020 8:08 am

You want to isolate Australia from the rest of the world indefinitely, fine. We might get lucky and come up with an especially effective vaccine and problem solved. Or we might not.

In that case the only way we get through this is it passes through the population, those especially susceptible succumb; those fortunate with the health and immune systems to fight it off become immune to it and life continues on. How many times has this happened in the past 2000 years? In this scenario, quarantine only moves the time you have to endure the disease out a ways. The virus will always be lurking out there. Probably the best scenario in this case will be better treatment protocols, but even that isn’t guaranteed.

Australia becomes a North Korea. No one in, no one out. Standard of living declines and everyone can’t go past their city limits without getting dragged back in. Good luck.

Clyde Spencer
Reply to  WXcycles
April 23, 2020 9:39 am

You asked, “… why people presume we would be so silly as to let people into the country without serving a sufficiently long quarantine …” There goes your tourist industry when most people only get two or three weeks vacation. They may not want to spend it all indoors looking for red-back spiders in their room.

Reply to  WXcycles
April 23, 2020 10:42 am

Yeah, if the native Americans could have just practiced effective quarantine for the last 400 years they wouldn’t have been decimated like they were.

If this virus becomes one like influenza, and continually lives and cycles through in resistant populations, then your proposal sounds like indefinite isolation.

Reply to  WXcycles
April 23, 2020 5:10 pm

I see all three of you wish to take an extremist nonsense position to avoid the real possibility of completely defeating it, with minimal disruption to life and economy, rather than a practical intelligent adaptive one, which works.

“… The virus will always be lurking out there. Probably the best scenario in this case will be better treatment protocols, but even that isn’t guaranteed. …” – rbabcock

So modern medicine just might not work. Right! OK, let’s just write that off as useless time-wasting, and give up sooner, so we can also fail, as there are tourists and airlines who’s priorities and balance sheet is so much more important.

I’m convinced – NOT.

We will be stopping it at the borders and investigating every opportunity to defeat it, and we’ll succeed.

April 22, 2020 11:45 pm

The UK’s Covid-19 latest update:

Reply to  Vuk
April 23, 2020 2:07 am

In the UK you cannot get a test until you have a fever. In which case the assumed infection rate is going to be far higher than it should be – it takes no account of asymptomatics.

I know loads of people who have had suspicious colds after visiting European Cov-19 hotspots, but none have been tested because they did not have a fever. (I do lots of international travel.)


Izaak Walton
April 22, 2020 11:50 pm

You say that “it’s been one full month since I publicly called at my blog for an end to the American Lockdown.”
So I am curious to know what you make of the recently published article
claiming that the lockdown and social distances has saved the US over 5 trillion dollars.

Reply to  Izaak Walton
April 23, 2020 10:52 am

Don’t know what Willis would say, but I say its a MODEL. Bet they used a case fatality rate of over 2%, and ignored the evidence of a large number of cases that are mild or asymptomatic.

April 23, 2020 12:03 am

Thanks, Willis. Australia’s scatterplot’s scattered pattern is unlikely to be because we’re all rugged individuals. I think, off the top of my head, that it could be some big outbreaks hitting a small population with a low background infection rate, when the number of daily tests has been variable.

A single ‘event’ here can make a big difference.

Matthew R Marler
April 23, 2020 12:04 am

Willis Eschenbach, thank you for another fine focused essay.

John Finn
April 23, 2020 12:08 am

I’m not sure if this is a general problem but analysis of UK test data is that the criteria for testing has changed over time. There have basically been 3 changes which have at least partly been driven by testing capacity, i.e

(a) Contact tracing – this involved testing suspected cases and all known contacts. While not random this covered a broader range than (b)
(b) Hospital attendances – resulted in an increased positive to test ratio for obvious reasons
(c) As (b) + Healthcare workers

(a) to (b) was a fairly abrupt change. (b) to (c) was more gradual.

It’s probably possibly to disentangle the data to get more consistent trends but I can’t be bothered. I suspect the UK new infection numbers peaked some weeks ago. Exactly when is a point of real interest. Given that there is a ~5 day incubation period it’s quite possible that the peak in new infections occurred BEFORE the big lockdown.

Hand washing & self isolation of those with symptoms might have actually been working .

It doesn't add up...
Reply to  John Finn
April 24, 2020 4:52 am

The peak in identified cases across the UK occurred on 4th April, as measured by taking a centred 7 day moving average to get rid of weekend effects, implying that the peak in infections was at least 5 days earlier – probably more, since only hospital cases were being tested, so perhaps more like 7-10 days after infection that 5. I have estimated when the peak occurred (or is yet to be certain) across England and mapped the result here:


Carl Friis-Hansen
April 23, 2020 12:09 am

Thanks Willies, I Eliza and Rune got me thinking of what a pensioned Swedish virologist said some time ago. He said that difference in size and isolation of nursing homes is significant.
For example in Sweden most nursing home count about 100 elderly and much traffic of personnel in and out of the institution. I Norway the average nursing home count more like 10 elderly and most personnel live more less integrated with the facility.
The virologist also assumed that about 50% of the general population at the time would have been infected. He therefore advised to pay serious attention to the nursing homes to protect the vulnerable elderly and regarded the lock-down as counterproductive, both economically, socially and health vise.

If anybody here have an idea how to get data about nursing homes in various countries, we might be able to see if the virologist is correct or not.

April 23, 2020 12:11 am

I used to work at Biofire, and they have a cool site that tracks infection rates.


I’ve watched that website for years, and it’s never done what it’s doing now. Usually their respiratory panel detects something 40 – 60% of the time it’s run, in the last couple months that has plummeted to 10%. The same is true of the gastro intestinal panel, plumbing new lows in detection frequency.

If we have 11 covid positives out of 100, and only about 10% with anything else for a respiratory or stool sample, there must be an enormous amount of hypochondria out there, or physicians ordering these very expensive tests have lost their minds.

Reply to  Ktm
April 23, 2020 12:15 am

Sorry, 19 covid positives out of 100 tests. Note that the BioFire respiratory panel does not yet detect covid, but it does detect other garden variety Coronaviruses.

Reply to  Ktm
April 23, 2020 8:13 am

My thoughts exactly … well sort of. My mind didn’t leap to hypochondriacs … but to Flu. 81 people out of a 100 presented with Flu … not COVID? They were tested because they had symptoms, including high temperature (not impossible, but tough to fake). That’s a LOT of Flu relative to COVID. A lot. Kinda makes me want to never touch another human for the rest of my life … hahaha … *cringe* … we’re headed for that dystopian future, aren’t we? Hopefully … we will eventually test EVERYONE, and discover just how HEALTHY we all are (for the most part).

Let’s get back to work!

And, Jeffrey Epstein did NOT commit suicide … neither should we.

April 23, 2020 12:15 am

If the tests are not randomized, we can’t conclude anything.

Robert B
April 23, 2020 12:31 am

There is probably a linear relationship with deaths and number of tests. STOP TESTING NOW!

Or, in countries where number of daily cases is growing linearly, there is a correlation with tests. In Australia, we were very fortunate that isolation stopped community spread so number of cases grew linearly for only a short time.

Of the half dozen countries with most tests, 4 have had negative trend in cases per day for about 3 weeks. Doubtful that tests have tapered off.

Before having a dig at my lack of data, I’ve rattled this off in a few minutes.

It doesn't add up...
Reply to  Robert B
April 23, 2020 4:51 am

There isn’t. High rates of testing are associated with both high and low death rates – compare e.g. San Marino and Singapore or the Faroes. There are also low rates of testing in places with low death rates, because essentially they remain largely uninfected – e.g. Indonesia and most of Africa. The Singapore/Indonesia comparison is interesting. Essentially next door, with very different health services, yet identical death rates. Perhaps their shared climate has more to do with it.

Robert B
Reply to  It doesn't add up...
April 23, 2020 12:40 pm

You might have missed my point. Just look at countries where deaths are increasing linearly.

April 23, 2020 12:35 am

I grabbed cases/1m against test/1m for European countries – for UK, France, Switzerland, Italy and Sweden, the R2 was over 99%!

But a word of caution – the number of TESTS in the UK is significantly higher than the number of PEOPLE tested. A significant percentage of people are being tested multiple times. I assume the same is true in other countries.

It doesn't add up...
Reply to  Phoenix44
April 24, 2020 5:02 am

I know for certain that the statistics for Wales only report one test for each individual (a positive one replaces any prior negative) for each six week period. A new period starts 6 weeks after the first test.

See the interpretation notes here:


I’ve not seen as clear a statement for England or Scotland or Northern Ireland, but it does seem likely that they have adopted a similar standard.

April 23, 2020 12:50 am

The Veneto Region in Italy has the highest world ratio between the number of throat culture per million inhabitants (at 04/22/2020 270,000 throat culture per 4.8 million inhabitants). The small town of VO EUGANEO has been fully tested twice in February 2020. The tests performed indicate: 1) in the first test 2.6% of the people were infected; 2) in the second test 1.2% were infected; 3) the infections occurred before the containment measures (lockdown); 4) infections occurred due to asymptomatic people with whom they lived together. 43.2% of positive swabs are asymptomatic. Researchers at the University of Padua did not find great differences in the viral load between symptomatic and asymptomatic people. See pre-printer (https://www.medrxiv.org/content/10.1101/2020.04.17.20053157v1).
Crisanti’s study, with which Imperial College of London collaborated, shows not only the effectiveness of measures of social distancing in the interruption of the transmission chain of contagion, but also the need for timely tracing of cases and their cases and their contacts, followed by possible isolation.
I completely agree with dr. Eschenbach when writing not to use confirmed cases as a metric for the spread of the virus. The real link is “number of cases” vs “number of tests”. In addition, tests are essential to identify asymptomatic people, otherwise the epidemic will never be stopped.

April 23, 2020 12:51 am

Are you comparing ‘like’ with like’ as I thought different tests were being used by diff. countries

Steven Mosher
Reply to  stan
April 23, 2020 1:46 am

diferent tests
different protocols

dogs breakfast.

Reply to  Steven Mosher
April 23, 2020 2:10 am

As long as the tests are consistently in error, and therefore different from each other, they will produce meaningful data.
It will not be precisely accurate data, but you can track trends from it.


April 23, 2020 12:54 am

When you first called for an end to the lockdown on 21st March, you gave a properly caveated forecast of only 670 deaths in US.

The current forecast is two orders of magnitude larger.

It seems a reasonable time to ask what number of deaths *would* justify a lockdown? If an earlier lockdown in NY had saved lives there, would that have been worth it?

Does your call still stand despite the increase in deaths? If so, waht number of deaths would cause you to change your all? Do you think a localised lockdown in NY implemented earlier would have been worthwhile?

April 23, 2020 12:59 am

I would have thought this would be of more use


In the UK “Scientists at the University of Oxford have been working on a vaccine to prevent people from catching Covid-19”


If many of the five and a half thousand test volunteers have had the virus without knowing it there’s likely to be problems with the results???

Reply to  mwhite
April 23, 2020 1:17 am

In the UK “Scientists at the University of Oxford have been working on a vaccine to prevent people from catching Covid-19”

Forget about that. Dr Whitty was crystal clear yesterday during daily briefing: no reasonable chance for vaccine till end of next year.

Robert B
Reply to  Paramenter
April 23, 2020 1:40 am

There is a good chance. The US allowed a vaccine to go out before proper testing in the late 70s. It would be reluctant to repeat this but this might be a time for it to happen again.

Reply to  Robert B
April 23, 2020 5:46 am

Not sure about that. Primum non nocere still stands and this vaccine business may be quite risky. Unless we will be in total desperation vaccine may not come for the good 12 months or so.

Robert B
Reply to  Paramenter
April 23, 2020 12:49 pm
Steven Mosher
April 23, 2020 1:33 am

For interested readers


Glad someone did this.

Some things I have seen looking at Korean data.

Superspreaders drive the damn thing. case case case case cluster, cluster, boom.

80/20 rule. 80% of the cases driven by 20% of the cases.

Spreading doesn’t happen in just any old social situation ( like being at the beach)

in Korea

Family, Friends, Cramped work space (call centers), Hospitals, mental wards, Nursing homes,
Gyms, Internet cafes, Singing rooms, churches.



Super spreader events:

lesson: there are specific types of situations that are KNOWN TO BE WORSE than anything else.

druken parties, funerals, church services, face to face business networking.


spread is highest in cultures that kiss on the cheek, next in those that shake hands and lowest in
bowing cultures.

There might be a surgical approach to social distancing. The hammer works, but it crushes the economy
as well

Robert B
Reply to  Steven Mosher
April 23, 2020 2:48 am

You said something sensible.

Reply to  Steven Mosher
April 23, 2020 3:01 am

Hey if you are in certain countries who you hold a religious festival … always good for a cluster or two.

Reply to  Steven Mosher
April 23, 2020 4:56 am

Excellent link.

Reply to  Steven Mosher
April 23, 2020 5:12 am

Thank you.
Her an “Amateur Scientist” have more of substance than WHO and diverse National Health Agencies. I think it is a shame on the scientific society.

Reply to  nobodysknowledge
April 23, 2020 7:13 am

Agree mosher Re: super spreaders. Any indication that school kids actually infect?. Ie teachers not at much risk?

Reply to  Steven Mosher
April 23, 2020 10:34 am

The human breath coming out of the mouth is characterized by a temperature of around 35-36 ° C and a relative humidity of around 80-90%. Coronavirus infection is greatly facilitated by external environmental conditions. As the temperature of the outside air decreases and above all the relative humidity of the outside air decreases, the ideal conditions are formed for the spread of the infection. The main mechanism to consider is the expansion of water vapor which in turn regulates the expansion of the viral load in the air (see https://valedo.com/umidita-temperatura-e-sars-cov-2/ .. use the google translator as the post is in Italian).

April 23, 2020 1:46 am

Thanks for this report however I sligthly disagree with you with this part “For every one hundred additional tests that we do, we find an additional nineteen confirmed cases of coronavirus.”

In fact testing, and the percentage of positive case, is a good way to control whether the country has the situation under control.

2 examples: Italy. around the 18th of March they start to increase the number of daily tests. from 12.000 to 16.000 then 25.000 by 24/03, then consistently above 30.000 from 1st of april and then regularly above 50.000 and even 61.000 yesterday.

During that time the % of positive tests has been going down. Between the 11th of March and 24th of March it was between 20% and 31% positive tests. then from the 25th up to the 6th of April it was between 19% and 10% but going down. since the 7th it is always below 10% and was at 5.3% yesterday. This I believe shows that most of the cases are identified.

The same for South Korea, when the epidemy started on the 18th of February, the positive tests were around 10% and some 2.500 tests per day, South Korea increased then their testing capacity and by the 4th of March it was performing some 17.000 daily tests and around 3% were positive. Now since end of March positive tests is below 1% and now even below 0.5%

Well just checked the USA for yesterday , 311.000 tests and 27.000 positives so less than 9%, en encouraging figure which needs to be confirmed in the next couple of days. It might mean that most of the case are identified.

Robert of Ottawa
April 23, 2020 1:50 am

That`s in line with the cruise ship Princess Diamond where 85% of the peopl didn`t have it. But these American tests are still biased towards the infected as they are not testing randomly so the number is probably lower. And it doesn`t tell us how many people actually HAD the virus.

Reply to  Robert of Ottawa
April 23, 2020 4:57 am

Take a look at the French aircraft carrier Charles de Gaulle which had 60% infection rate.
Perhaps isolation in a cruise liner works after all. Is this proof that a lock-down reduces infection?

April 23, 2020 1:58 am

I have said from the very beginning that we need random testing, before we can assess the true infection rate or death rate. How can anyone formulate a medical or political strategy, before we have that data?

Wild guesses have been made of 5% mortality, when we have no idea how many are really infected. And this has been talked up and inflated by the alarmist media, who sell newspapers and TV news through wild and exaggerated stories. And some irresponsible medics chimed in, estimating millions of deaths, in order to get their 15 minutes of fame.

But when a random sample was at last taken in Santa Clara, it unveiled infection rates that were 50 to 80 times higher than thought. And that suggests the true death rate may be 50 to 80 time LESS than thought. In which case, the Cov-19 virus is nowhere near as deadly as has been advertised – so why did politicians shut down the economy?


Has this unprecedented shut-down of the world been a media-led waste of time and money, with politicians being hounded into making poor judgements and decisions based upon the baying howl of ignorant reporters?


John Finn
Reply to  ralfellis
April 23, 2020 3:05 am

But when a random sample was at last taken in Santa Clara,

It wasn’t a random sample. It was a sample recruited via a Facebook ad so will almost certainly be biased in favour of those who believe they have been infected.

And that suggests the true death rate may be 50 to 80 time LESS than thought.

Not True. Chief Medical Officer made it clear that he (and colleagues) expected the true fatality rate to be below 1%. This was in a Parliamentary Select Committee meeting in February. Neil Ferguson (the modeller) has since suggested that 3-4 million people had probably been infected several weeks ago.

The Fatality rate was not the issue of most concern. Lack of immunity means that at least half the population would need to be infected before there was a significant decline in the spread of the disease. Even that might have been acceptable if it took place over, say, a 12 month period. If the virus were allowed to spread freely and assuming a fatality rate of 0.2%, there would be at least 320k deaths in the US within a 3 month period. But that assumes the US healthcare system could manage to treat the 16 million patients who require hospitalisation.

Mitigation was necessary to prevent healthcare systems becoming overwhelmed.

Reply to  John Finn
April 23, 2020 4:22 am

That is NOT the scare story that has been propagated by the media, both of the responsible and sensationalist kind. The story that had resulted in millions cowering and imprisoned in their homes, is that the mortality rate was between 2% and 5% – and thus this is a very dangerous pandemic that could kill millions.

Please give some links to popular press articles, that give the likely death-rate as being just 0.2%. Most people could live and work with a 0.2% mortality rate, without crashing the economy. While most elderly and infirm people could self-isolate. But that us not what we have ended up with – the scare stories have locked down entire nations.


John Finn
Reply to  ralfellis
April 23, 2020 10:45 am

Please give some links to popular press articles, that give the likely death-rate as being just 0.2%.

I don’t read the “popular press”. I know that Neil Ferguson used a fatality rate of 0.9%. It could be lower than that but not as low as 0.2%. I used 0.2% to show that even that figure could result in over 300k deaths in the US.

Most people could live and work with a 0.2% mortality rate

Even at that rate Healthcare systems could be overwhelmed.

Clyde Spencer
Reply to  John Finn
April 23, 2020 9:50 am

You commented about “biased in favour of those who believe they have been infected ..” Also, Facebook is going to be biased in favor of a younger group, demonstrably more active socially, and less susceptible to the virus. Facebook recruitment is anything but random.

Reply to  ralfellis
April 23, 2020 3:51 am

In addition to the point John Finn makes, the antibodies tests throw out false positives, depending on the particular test at a rate somewhere between 1 and 10%, even at the low end of this error rate the results of the Santa Clara study are meaningless, across 3300 tests they got 50 positives, even at a false positive rate of just 1% they’d get 33 false positives, at a false positive rate of 1.5% there might not have been a single true positive in the entire study.

Steve S.
April 23, 2020 2:12 am

“Don’t use confirmed cases as a metric of the spread of the virus—the number of cases is indeed a function of the numbers of tests. ”

And the number of tests is a function of the number of people who have severe enough symptoms to go to the hospital where they are subsequently tested. Hence, it is a good metric for the spread of severe cases in the population. Mild cases, no doubt, parallel the severe cases, but the ratio of mild to severe is still in question.

Ian Coleman
April 23, 2020 2:20 am

The statistic I am most interested in, but is missing so often from reports on the progress of the pandemic that I suspect that reporting of it is being suppressed, is the median age of the dead. I have never understood why anyone would place any value on the number of cases when the testing for them is not random. Also, what is a case? A twenty year-old who is asymptomatic and a 75 year-old with emphysema are both cases. The criteria are so broad that the category has no useful meaning.

I realize that this question is profoundly offensive to many people, and the fact that I would even ask it could be evidence of moral deficiency, but why are we ruining our economies to limit the spread of a disease that mostly kills old people?

I abhor the unreasonable lengths the medical profession goes to prolong life in the elderly. My parents died of diseases of old age, which is to say, they died slowly and painfully, and each at least a year after he or she would have died if medical professionals had not compulsively and unreasonably prolonged their lives. Dying of old age is the equivalent of being held down and slowly smothered, after being stripped of every capacity that makes life worth living as a human being. My own plan for happiness in old age (and I’m 68) is, avoid doctors unless you have something they can fix. When you’re dying (and you’ll know when) don’t let them treat you. Better to die in a week and be done than over the course of a year.

It doesn't add up...
Reply to  Ian Coleman
April 23, 2020 9:12 am

There’s plenty of reporting of ages in the UK. Try the latest total announced deaths spreadsheet from here, which breaks it down by date of death as well as age ranges:


Over half are 80+

John Finn
April 23, 2020 2:45 am

Covid-19 Tracker for 20-69 year olds in UK.


Contributors use mobile phone app to record symptoms. Tracker estimates probability of Covid-19 from symptoms based on learned data from actual tests. Extrapolations across regions then provides number of likely Covid-19 cases.

According to tracker there were 2 million ‘probable’ cases in UK on April 1st. Most recent estimate is 437k.

It doesn't add up...
Reply to  John Finn
April 23, 2020 5:07 am

The interesting number would be the total number who have been infected. I wonder if they have the means to estimate that from their cumulative data.

John Finn
Reply to  It doesn't add up...
April 23, 2020 10:52 am

If there had been enough available tests regional samples could have been tested which would have confirmed (or otherwise) the reliability of the apps data.

John in Isolation in Oz
April 23, 2020 3:04 am

How do they account for people who have a negative test but then are infected afterwards?

Being tested does not provide immunity so they are possible future positive cases.

If infected post-test, are they counted in both the negative AND positive numbers?

They could be tested again if they showed symptoms and would appear in the data but, if asymptomatic, would not be counted.

April 23, 2020 3:43 am

There are two ways to test; use the tests to confirm suspected cases, or use the tests to track and trace the spread of the disease. The countries using the test for the former have high positive test rate, often over 20% (UK, US, Spain, France, Belgium, the Netherlands), the countries using the test for the latter have low positive test rates (South Korea, Taiwan, Iceland, Australia, New Zealand). The former countries are stumbling around in the dark, the later countries have the lights on, can see what they’re doing, and are knocking this virus over quickly.

If you want any data about what the true fatality rate of this virus is look at the data for the latter countries and the Diamond Princess and Theodore Roosevelt. It’s somewhere between 0.8 and 2% in most Western populations.

It doesn't add up...
Reply to  Andrew_W
April 23, 2020 5:13 am

Testing solves nothing unless it leads to rapid and effective quarantine. Speed of quarantine imposition of case contacts is much more important than detecting every infectious person. Delay in quarantine allows all contacts to be spreaders. Missing the odd spreader contact will likely be picked up by infection of someone else, and further spread suppressed through that.

Reply to  It doesn't add up...
April 23, 2020 10:02 am

It doesn’t add up… April 23, 2020 at 5:13 am
Which is why lock-downs were used, everyone gets quarantined (Which is not a position on whether their use is morally justified), and obviously those with positives results know it’s important that they adhere to their quarantining.

It doesn't add up...
Reply to  Andrew_W
April 24, 2020 5:16 am

The reason for imposing lockdown was that in urban environments, and e.g. on London’s packed tubes, it was impossible to trace contacts in any meaningful way. On the other hand, it was probably quite unnecessary in less urbanised environments where people do not go to work on public transport, and where there social interactions are much more limited: the same locals at the pub even if they still go to one, for instance – not random encounters in the queue for a sandwich at Prêt à Manger.

Just look at the enormous differences in infection rates across Wales:


Clyde Spencer
Reply to  It doesn't add up...
April 23, 2020 10:02 am

It is of little value to know that someone has COVID-19 because there is no known cure accepted by the medical community. Those advocating HCQ say it can only be used on those who have high probability of recovering on their own, and it doesn’t work on those who clearly are in need of a prompt cure. As to quarantine, anybody who has flu-like symptoms should stay at home anyway, and practice social distancing and enhanced hygiene. Until such time as there is universal testing, with repeat resting, we cannot identify the asymptomatic carriers. Testing has the greatest value in determining the prevalence of the disease in the population and the rate of growth. However, it isn’t being used in that way.

It doesn't add up...
Reply to  Clyde Spencer
April 24, 2020 5:20 am

The point is that quarantine needs to be enforced around identified cases. Their contacts may be asymptomatic spreaders. That is stopped by quarantining them, regardless of test results. If you wait until they also show symptoms they will have spread the virus. It is doubtful whether testing is good enough at 30-40% false negatives to let them bypass quarantine.

April 23, 2020 4:01 am

People are not being tested at random. People are being tested because they are symptomatic, so in that case, tests are going to reflect the incidence of the disease in question, and if more people are getting sick, the number of confirmed cases is going to go up.

Tom Kennedy
April 23, 2020 4:02 am

The CDC has stated the following on (https://www.cdc.gov/flu/weekly/index.htm)

“Key Points

Nationally, influenza activity is now low.
With ongoing declines in influenza activity and the continued effects of the COVID-19 pandemic, FluView will be abbreviated for the remainder of the 2019-2020 season.”

In addition the government announced and that it will pay hospitals for the treatment of anyone contacting Covid-19 . This was in a response on how the uninsured will get treatment.

This incentivizes hospitals and health care providers to report everything especially deaths using the “Covid -19 codes”.

This has the potential to distort Covid -19 death statistics.

Reply to  Tom Kennedy
April 23, 2020 8:26 am

It’s beyond potential. It’s kinetic.

Farmer Ch E retired
April 23, 2020 5:05 am

Willis – many moons ago when I was a young researcher at Battelle Pacific Northwest Laboratories, I took a course called Strategy of Experimentation which was developed by E.I. DuPont. Using statistics, it taught how to design an experiment with minimum number of test trials based on how many variables were being studied. Science is seldom two-dimensional so we need to remain weary of conclusions drawn from two-dimensional analysis.

April 23, 2020 5:16 am

The hypothesis is that the virus spreads widely because some carriers get the SARS-CoV-2 virus but not COVID-19 disease. If this is the case we need to know! #antibodytesting would tell the business owners which employees are safe to get back to work. #antibodytesting would let the epidemiologists make better predictions. Testing whether or not an individual has an ‘active’ case is medically correct, but not so useful to decision makers and paid prognosticators.

April 23, 2020 6:02 am

“Don’t use confirmed cases as a metric of the spread of the virus—the number of cases is indeed a function of the numbers of tests. ”

That is only true if tests are not exceeding the number of cases by far. Like in South Korea, Austria, Germany, Taiwan. Worldometer is suggesting a rough estimate of an at least 10-fold overtesting is able to show you the real development of cases. Otherwise there would be no curves bending.

Reply to  Ron
April 23, 2020 2:48 pm

Seems the curve has bent a bit in the US without the 10-fold overtesting.

Reply to  Bob
April 23, 2020 4:16 pm

Maybe there is now sufficient overtesting as growth rate declines and testing capacity increases. Hard to say from the available data.

Earl Rodd
April 23, 2020 6:03 am

Australia’s health department site (https://www.health.gov.au/news/health-alerts/novel-coronavirus-2019-ncov-health-alert) says they have done 466,000 tests with only 1.4% positive rate. They are seeing single digits of new cases many days. Their site also says they can trace the source of nearly all cases, mostly to overseas.

What have they done so differently?

Steven Mosher
Reply to  Earl Rodd
April 23, 2020 6:54 am

“What have they done so differently?

maybe nothing.

No superspreaders.

Reply to  Earl Rodd
April 23, 2020 6:42 pm

They’ve used the test to track and trace, America and most European countries have mostly been using it to test suspected cases, hence the difference in the positive test rates.

April 23, 2020 6:04 am

Willis, many thanks for your most interesting article and graphs.

I downloaded into Excel the data that you recommended, from Our World in Data (I agree they are an excellent source). I can see why you wanted to do a scatterplot! If you try to plot the ratio of daily new cases to daily new tests against date, the data is horribly jumpy. Not to mention that the reporting of a test and the reporting of the resultant case may happen several days apart. Particularly in those countries (such as France and Germany) in which they only seem to be reporting tests on a weekly basis. To get around this, I’d previously been looking at the evolution of the ratio of cumulative total confirmed cases to cumulative total tests over the time frame of the epidemic.

If you look at the US numbers on that basis, you’ll see that the total cases per test to date is indeed 19%. But I think the linearity of your scatterplot is caused by the fact that, in the US, that ratio has been unusually constant compared with other countries. In fact, it has been between 18% and 19% every single day since April 8th. If you go back towards the beginning of the epidemic, the ratio of cumulative cases to cumulative tests started at 14.1% on March 7th, dropped to 8.2% on March 18th, and has since risen towards its present value of 19%. That little cluster of points at the bottom left of your graph, which is clearly below your trend line, is actually a feature of interest. It represents data from the early stages of the epidemic, when the tests were both lower in number, and showing a lower incidence of confirmed cases per test, than now. Indeed, if I disregard your trend line and that outlier at the far right, and squint hard enough, I think I can just about see a hockey stick!

I looked in a bit more detail at the UK figures. BTW, Our World in Data reports tests for the UK as per person tested, which explains the differences between these and the figures from Worldometers (which seems to be using total number of tests). The ratio of cumulative confirmed cases to cumulative tests started on January 30th at 1.1% (2 out of 177). It reached a minimum of 0.1% on February 22nd. It then climbed, relatively gently, back up to 0.3% by March 4th. On March 5th it started to accelerate upwards. By March 16th (the day the first stage UK lockdown was announced) it was up to 3.2%. By March 23rd (second stage lockdown) it had reached 6.8%. Today it is 31.4%, and still rising; though the rate of rise did begin to slacken about 10 days ago. The ratio of daily confirmed cases to daily tests has been north of 30% every day except one since March 28th.

Looking at Austria, a country which has all but beaten the epidemic, the ratio of cumulative confirmed cases to cumulative tests started out at 0.6% on February 26th. It reached a peak of 18.4% on March 31st – coincidentally (?) the same day the daily new confirmed cases started a precipitous decline after the third of three peaks. Today, it is down to 7.4%, and still falling. Individual days’ confirmed cases to tests ratios have only been above 30% twice in the whole epidemic. Since 2nd April no day has been above 10%, and the maximum in the last week has been 2.3%.

What might account for these differences? Population density is one thing; or, at least, the numbers who live, work or travel in high-density environments. The possibility of greater immunity according to how far the virus has already spread through the population is another. Shortage of test kits is a third; if you can only test those with significant symptoms, then as long as the epidemic is still expanding, you would expect the cases to test ratio to go higher and higher. I’m interested to know what others here think might have been the causes of the huge divergences in the path of the epidemic between these three countries.

Earl Rodd
April 23, 2020 6:07 am

You note that Australia sees only 1.4% of tests positive (of over 400,000 tests in a population of 21 million). They are seeing very few cases and their health department site also says they can trace the source of nearly all cases, mostly to overseas. From what I know, Australia had some very early cases and put in “social distancing” type measures in the same time frame as the US.

What have they done so differently to have so few cases – which seems to be for real given the amount of testing done and the low positive rate.

Steven Mosher
April 23, 2020 6:09 am

“Conclusion? Don’t use confirmed cases as a metric of the spread of the virus—the number of cases is indeed a function of the numbers of tests. ”

Sweden, denmark, Finland, US, Germany, Norway would make an interesting panel

final figures

Norway 27,421 tests per million 5% positive
Germany 24,738 tests per million 7.5% positive
Denmark 20,134 TPM 7% positive
Hong Kong 17,579 TPM 0.7% positive
USA 13,071 TPM 19.7% pos
Finland 12,814 TPM 6% pos
Sweden 9,357 TPM 18% pos

Germany 3.4% of urban pop tested 7.5% positive
Norway 3.3% of the urban population tested 5% positive
Denmark 2.2% of urban pop tested 7% positive
Hong Kong 1.8% of pop tested 0.7% positive
USA 1.6% of urban pop tested 19.7% pos
Finland 1.5% of urban pop tested 6% pos
Sweden 1% of urban pop tested 18% pos

Reply to  Steven Mosher
April 23, 2020 7:28 am

23 April 2020 – Queensland has two new confirmed cases of novel coronavirus (COVID-19) raising the state total to 1,026. “… Dr Young said a total of 92,699 tests for COVID-19 has been undertaken in Queensland, with 2,539 tests undertaken in the previous 24 hours. …”

Source: https://www.health.qld.gov.au/news-events/doh-media-releases/releases/queensland-novel-coronavirus-covid-19-update-2020-04-23

Yet QLD had 2,137 tests on the 28th of March that produced 111 new cases.

Source: https://www.abc.net.au/news/2020-03-17/coronavirus-cases-data-reveals-how-covid-19-spreads-in-australia/12060704

But 2,539 tests during the past 24 hours produced 2 positive cases. So more testing does NOT produce more cases. More virus being present produces more cases, while less virus being present due to weeks of isolation, produced 55.5 times fewer cases, from 19% more tests.

Chris D.
Reply to  Steven Mosher
April 23, 2020 10:01 am

“Hong Kong 1.8% of pop tested 0.7% positive”

Was this a typo in omitting “urban”?

Steven Mosher
Reply to  Chris D.
April 23, 2020 8:39 pm

I had no figures for rural pop in HK.
having been there many times I’d say its close to 100%
I could calculate it from source data but was lazy

Clyde Spencer
Reply to  Steven Mosher
April 23, 2020 10:07 am


“USA 13,071 TPM 19.7% pos”
“USA 1.6% of urban pop tested 19.7% pos”

Shows that the US statistics are being driven by urban environments. It is almost as if NYC were another country.

It doesn't add up...
Reply to  Clyde Spencer
April 23, 2020 10:13 am

It’s not only the US. I’ve posted maps of case rates for Wales, England and Germany below.

April 23, 2020 6:29 am

The use of active cases is being pushed to alarm Australians to keep them compliant with pretty tough measures by highlighting the risk of a second wave in Singapore. Yes, active cases shot up when they relaxed borders by bringing in foreign workers who live in hostel type dorms. Why understanding is that infected workers were in home detention but their families weren’t and there was community spread. But what is interesting it’s now at least two weeks and even though active cases have indeed exploded but deaths have barely moved. It may be that the number of deaths will also explode but I’m watching it closely. The number of deaths is really the clutch number . Singapore currently has the grand total of 12 deaths but are being used as an example of why we have to keep everyone locked up in Australia where as you point out we have a ridiculously rate of active cases / tests and deaths. The other thing is that also confirmed cases is not active cases and the acknowledgement of recoveries is often overlooked by the media.

April 23, 2020 7:05 am

Which is why the most important data point is hospitalizations.

That data point directly relates to hospital bed usage. It reflects the amount of sickness in a community and the community’s ability to deal with it. The death/cure rate of those hospitalized is easily counted and can show effectiveness or the lack of effectiveness in treating the virus.

Yet, while every newspaper in the country shows new cases and deaths, almost none show the hospitalization number.

Reply to  Dave
April 23, 2020 7:12 am

I agree with @Dave.

Alternatively look at how much oxygen hospitals consume.

One hospital in Italy reported that they had an annual consumption of about 150 liters of oxygen. The first three and a half months this year, they spent more than 3000 liters of oxygen.

That could be a useful metric.

Reply to  Rune
April 23, 2020 8:20 am

O2 use depends on the treatment and is not constant. High flow devices like HFNC use vastly more O2 than ventilators, and the volumes change frequently depending on physiological response.

Reply to  Dave
April 23, 2020 7:05 pm

I think hospitalization rates are being skewed in areas of high rates of infections by people choosing not to go to hospitals when they should. Either through fear of being stuck in a ward with lots of C19 +tive people or because they don’t want to burden the health system when they see others requiring the services more than they.

Mike Smith
Reply to  Dave
April 24, 2020 6:49 pm

I agree, the number of hospitalizations (where available) is probably the best metric we have at this time.

The # of deaths would be interesting but it seems we can’t even count those that arise outside of a hospital setting (i.e. in homes, managed care facilities etc).

April 23, 2020 7:08 am

So along with testing those who are “at risk” or may already be showing symptoms, we will have false positives/false negatives guaranteed. Even with very reliable testing, if someone shows “negative” today how do we know they won’t catch the Wuhan virus tomorrow? Are we supposed to test everyone every day?

April 23, 2020 7:10 am

Willis wrote:
“Conclusion? Don’t use confirmed cases as a metric of the spread of the virus—the number of cases is indeed a function of the numbers of tests.”

I agree. I came to this same conclusion about one month ago, and have been dismissive of much of the public data since then. Most Covid-19 testing is “chasing positives” to limit contagion and thus grossly overestimates the severity of this virus.

The best data I’ve encountered is Iceland, where random testing has been performed on over 10% of the country’s entire population and the Total Fatalities/Total Infections at less than 0.1% – SIMILAR TO A TYPICAL FLU.

As I wrote in March:
[excerpt- posted 21Mar2020]

“This full-lockdown scenario is especially hurting service sector businesses and their minimum-wage employees – young people are telling me they are “financially under the bus”. The young are being destroyed to protect us over-65’s. A far better solution is to get them back to work and let us oldies keep our distance, and get “herd immunity” established ASAP – in months not years. Then we will all be safe again.”

Have we wasted many trillions, harmed billions of young people and trashed our economies for nothing? Seems so.


In Iceland THE DEATH RATE [Total Deaths/Estimated Total Infections] IS LESS THAN 0.1%, SIMILAR TO OTHER SEASONAL FLUS.

Covid-19 appears relatively mild, often showing no symptoms among younger people, but is dangerous to the elderly and the infirm.

I’ve also suspected that Covid-19 arrived in North America sooner than generally assumed. As I recall, the earliest mention of its occurrence in China was 17Nov2019 – I expect it hit North America by no later than Dec2019. A newly-discovered Covid-19 fatality in California on 6Feb2020 supports that hypo.

New test results confirm the novel coronavirus was responsible for two deaths in February
By David Debolt and Kerry Crowley
Published: April 21, 2020

Two individuals who died in Santa Clara County in February tested positive for COVID-19, health officials learned Tuesday, revealing the novel coronavirus was responsible for deaths in the Bay Area more than a month earlier than initially reported.

The individuals who tested posthumously for COVID-19 died at home on February 6 and February 17, according to autopsy results released by the county late Tuesday. The deaths appear to be the first confirmed coronavirus fatalities in the country — occurring weeks before the first U.S. death was publicly reported on Feb. 27 — adding to mounting evidence that the virus was spreading here far earlier than once believed.

April 23, 2020 8:04 am

Thanks to Mosh for posting this excellent video elsewhere.

This elderly Swedish gentlemen is obviously mature, expert and brilliant – and he agrees with me. 🙂

April 23, 2020 8:36 am

By David Menzies April 23, 2020

“In fact, at her hospital, personnel are so bored because on many days they have nothing to do as they await for a tsunami of COVID-19 cases that never materializes…”



Here in Alberta, the Covid-19 lock-down has resulted a debacle.

Most of our deaths are in nursing homes – our policy seems to be “Lockdown the low-risk majority but fail to adequately protect the most vulnerable.”

The global data for Covid-19 suggests that deaths/infections will total 0.5% or less – not that scary – but much higher and clearly dangerous for the high-risk group – those over-65 or with serious existing health problems.

“Elective” surgeries were cancelled about mid-March, in order to make space available for the “tsunami” of Covid-19 cases that never happened. Operating rooms are empty and medical facilities and medical teams are severely underutilized. The backlog of surgeries will only be cleared with extraordinary effort by medical teams, and the cooperation of patients who die awaiting surgery – patients who were too impatient…

This may look like 20:20 hindsight, but I called it this way in ~mid-March.

Regards, Allan

April 23, 2020 3:42 pm

In Iceland 10 deaths out of 1791 cases is 0.56% but 10 deaths and 1509 recoveries (which is a better metric) is 0.66% which is 6 times as high as the 0.1% you quote above. You could take their random testing results which say that 0.6% of people are actually infected and infer that their total population of infected are 364,000*.0061 = 2,223 people and that would get you to an IFR of ~0.4% but I’m not seeing how you get to 0.1%. Would love to be wrong though. Either way much less than the scary numbers in the headlines. Iceland certainly has the best data but maybe not applicable to non island, non homogeneous populations.

Reply to  dwestall
April 23, 2020 7:23 pm

dwe wrote:
“In Iceland 10 deaths out of 1791 cases is 0.56% but 10 deaths and 1509 recoveries (which is a better metric) is 0.66% which is 6 times as high as the 0.1% you quote”

Your stats are simply a function of the number of tests – NOT MEANINGFUL.
You need a large random testing of the population and you get 43,000 cases, ~13% of their country’s population.

Active cases peaked on 5April2020. On 4May2020 the Icelandic government will begin relaxing COVID-19 restrictions in Iceland in general. Icelandic preschools and elementary schools will return to regular operation; salons, massage parlours, and museums will reopen; and gatherings of up to 50 people will be allowed. Swimming pools, gyms, bars, and slot machines will remain closed for the time being.

Iceland Total Tests to 21Apr2020 ~43,143
Confirmed infections 1773
Population of Iceland 341,250
Total Tests/Population 12.6%
Infections/Tests = 4.1%

Extrapolating to Iceland’s population = (341,250/43,143) * 1773 = 14,023 estimated total infections in Iceland

Ten deaths have been recorded to date.
10 deaths/1773 confirmed infections = 0.56%
10 deaths/14023 estimated total infections in Iceland = 0.07% = LESS THAN 0.1% MORTALITY RATE IN THE GENERAL POPULATION OF TOTAL ESTIMATED INFECTIONS
10 deaths/341,250 population = 2.9*10^5 = 0.003%

April 23, 2020 9:54 pm

“Extrapolating to Iceland’s population = (341,250/43,143) * 1773 = 14,023”

You’ve totally screwed that up.
“Stefansson said Iceland’s randomized tests revealed that between 0.3 to 0.8 per cent of Iceland’s population is infected with the respiratory illness, that about 50 per cent of those who test positive for the virus are asymptomatic when they are tested, and that since mid-March the frequency of the virus among Iceland’s general population who are not at the greatest risk – those who do not have underlying health conditions or signs and symptoms of COVID-19 – has either stayed stable or been decreasing.”

April 23, 2020 10:27 pm

Your mistake is that you’ve rolled the results from targeted testing into the results of the random testing.

Reply to  Andrew_W
April 24, 2020 3:44 am

Let’s assume you are correct. NUHI testes are targeted, deCode are random.


Infected/Tested NUHI*= 9.57%
Infected/Tested deCODE Genetics= 0.61%
Iceland Population= 341,250

Total Infected Est.= 2084 (using deCode 0.61% only)
Deaths/Infected= 0.5%
7/10 deaths over 70 years of age, 9/10 over age 60, 1/10 age 30-39.

Based on the data, the Swedish strategy of limited contagion (not full lockdown) is much better.

Reply to  Andrew_W
April 24, 2020 3:57 am

This elderly Swedish gentlemen is obviously mature, expert and brilliant – and he agrees with me. 🙂

Reply to  Andrew_W
April 24, 2020 4:55 am

“Let’s assume you are correct.”
We don’t need to assume I’m correct, the links you provided state what I’ve said. deCODE Genetics was doing the random testing and the results were infection levels in the wider community of 0.61%.

Prof. Johan Giesecke and Anders Tegnell are the architects and main advocates of the Swedish policy, in terms of controlling Covid-19 the results say it’s not a very successful policy, especially when you can compare Sweden’s progress in controlling the virus with its Scandinavian neighbors.

It doesn't add up...
Reply to  Andrew_W
April 24, 2020 5:07 am

Given that the famous Imperial model was predicting 250-500,000 deaths in the UK absent a lockdown, I think the fact that Sweden (with its highly cosmopolitan cities being its main centres of population) I think we need to compare against the pro-rata 35-70,000 deaths implied for Sweden absent a lockdown. It seems that deaths are way, way short of those levels – which must throw lockdown policies into question at the very least.

Reply to  Andrew_W
April 24, 2020 10:54 am

Andrew, [snip – mod] All you did was quote a newspaper article.

Sweden has lower total deaths than many countries that chose full lockdown
See Mortality Monitoring in Europe

This Swedish expert explains the higher mortality among the elderly in Sweden vs Norway. https://youtu.be/bfN2JWifLCY

I suspect that most of these differences in total mortality in Europe are due to worse vs better control of Covid-19 infections in nursing homes. Some countries like the UK are doing a great job of killing off their elderly – you’d almost think it was deliberate. https://drmalcolmkendrick.org/2020/04/21/the-anti-lockdown-strategy/

Reply to  Andrew_W
April 24, 2020 1:01 pm

“Sweden has lower total deaths than many countries that chose full lockdown.”

Given the later arrival of the virus in Sweden her position in 10th place in the world in deaths/million (7th if you exclude densely populated micro-states) is nothing to crow about.

The challenge is to get R0 below 1 and keep it there until the enemy has been vanquished.
The ability of countries, states and communities to do that will come down to both the threats imposed by the state and the social culture of the people in a country.

At the moment we can see that some East Asian countries have gotten R0 without the use of government imposed measures, Western cultures are different to East Asian cultures, we’re more supportive of individuality – people having the right to do their own thing, that’s a good thing when it comes to economics and things like creativity and individual freedom. But when it comes to defeating this sort of enemy it’s not so effective, a few rogues spreading the virus can defeat the efforts of the majority to contain it.

In military terms it only takes a few to break the ranks under pressure for the front to collapse. It’s the people that are on the front line in this one and Western Society, with its individuality isn’t so well equipped for this fight. The US least of all.

Posting the same youtube video again doesn’t advance your argument, as I said Prof. Johan Giesecke is one of the architects of Sweden’s policy on this, he also advises WHO – which might explain so many of WHO’s mistakes in not taking the threat this virus poses seriously.

April 23, 2020 7:12 am

What we have here is nothing more than a simple sampling problem. To determine the number of cases, we need to use the antibody test, not the test which only indicates CURRENT affliction. Using the antibody test, one can create a randomized sampe stratified by the affecting things, take a healthy sample – a couple of thousand would be plenty, and then calculate the total for the country. But this is specific as to the time of the tests.

April 23, 2020 7:13 am

I just read, that smokers are less hit by Corona, seems, the virus can’t hook on, because of nicotine is placed there.
They will start studies with nicotine pavement for medical personal and patients too.

Reply to  Krishna Gans
April 23, 2020 7:59 am

Holy smokes, Batman!

The literature presented in this review strongly suggests that nicotine alters the homeostasis of the RAS by upregulating the detrimental angiotensin-converting enzyme (ACE)/angiotensin (ANG)-II/ANG II type 1 receptor axis and downregulating the compensatory ACE2/ANG-(1–7)/Mas receptor axis, contributing to the development of CVPD.

Nicotine and the renin-angiotensin system

Reply to  icisil
April 23, 2020 8:16 am

But if infected you would want downregulation of ACE due to its inflammatory role in RAS, and because viral infection downregulates ACE2 and its anti-inflammatory response. So maybe good as a infection prophylaxsis, but not so good once infected.

Terry k Anderson
April 23, 2020 7:37 am

I have been tracking Tennessee USA cases, deaths, and testing for two months. We have averaged about 250 cases a day with 3,000 to 10,000 tests. Wouldn’t anyone who feels poorly and anyone who came in contact with that person or a person that has tested positive and those that came in contact with them go get a test? So, if that assumption is true, why would increased testing lead to increased cases?
When an area reaches the top of the bell curve or starts down the slope, increased testing won’t lead to increased cases? Straighten me out, if you will.
Terry Anderson