Regular WUWT contributor Willis Eschenbach has been plotting the official death rate data from the Coronavirus almost daily. We will continue to add to this page as needed and as Willis makes updates.
Read the note from Willis, or Skip to Latest Plot:
My thanks to Anthony for thinking of making this a page to follow the history of these plots. I’ll be updating them daily, unless mischance and coincidence intervene, sometime just before midnight GMT, which is when the Worldometers data page refreshes for the new day.
Note that the data changes throughout the day, and there are various sources of data, so there may be some variation between your data and mine at any given instant.
For those interested in how I’ve done it, I put the R computer code online, download it by clicking here. There are two files, “Willis Functions.R” and “Coronagraph.R”. The “Willis Functions” file loads a bunch of libraries, so you may have to download one or more. It also contains the code for a host of things I use over and over, such as the lines with the black background that I like to use for clarity in my graphs.
The R code goes and gets the HTML source document for the Worldometer page. It then parses the document to obtain the various historical figures. These don’t include the most recent day, so I download and parse another page for those figures and tack them on the end.
Finally, a plea to all those in charge. The economic damage from the current insane “shelter-in-place” regulations is going to be huge—lost jobs, shuttered businesses, economic downturn, stock market losses. This doesn’t count the personal cost in things like increased suicides and domestic and other violence. Think pissed off young men out of a job and on the street.
And on the positive side, as my graph clearly shows, the South Koreans have managed to contain the virus. How did they manage that?
First, they test widely, then use confirmatory tests to avoid wasting time on false positives.
Then they trace all contacts of infected people and test them, and identify and quarantine SICK people, not HEALTHY people.
My oft-infuriating good friend Steven Mosher is living in Korea at the moment. He said that when one person in an apartment building gets ill, they test the whole floor plus a couple of floors above and below where the person lives.
When a guy in a call center got the virus, he said, they tested all 250 people working there. Heck, those jokers even have phone-booth style testing facilities to increase the number of tests per day … they are on it.
In this manner, using testing, tracing, and quarantining the ill rather than the healthy, they’ve been able to control the spread very well. As of today (2020-03-20) they have only 94 deaths in the whole country and leveling out (see below) and they’re NOT locking down the entire population and destroying their economy like we are.
Let’s emulate success, folks. I don’t mind learning from experience, but generally, I prefer to learn from other people’s experiences, and we have Europe and South Korea to learn from.
Here’s the crazy, bull goose looney part no one is talking about. The US government is about to spend a trillion dollars of your and my tax money to prop up the economy whose wheels have just been taken off by the insane shelter-in-place orders of the US government. Sen. McConnell unveiled a roughly $1 trillion stimulus package on Thursday to help “mitigate the economic pain that tens of millions of Americans are already feeling”.
That trillion dollars won’t put the wheels back on. It won’t get us rolling again. It just pays us for the losses already suffered.
Do you ever think how many ventilators and hospitals and test kits and testing personnel we could buy for A TRILLION DOLLARS OF YOUR AND MY GAD FARKING TAX MONEY!!
Typical ventilator cost US$25,000, in normal times. Say you have to pay double in scarce times. Say we want a half million of them, big number, more than we’ll ever need, but why not? How much of our trillion pinche dollars of tax money remains?
Ninety-seven percent. We’ve bought a half million ventilators and have hardly dented the pile.
My point is simple. If we’re going to spend a trillion, let’s put out wartime prices with war-time high-speed bidding processes. Say that the government will pay double the peacetime costs for ventilators and mobile field hospitals and beds and the like. Focus on American made. Phone-booth testing sites? Koreans can make them? Americans can make them. Buy all that the Koreans willl sell, plus encourage US manufacturers can make them by putting tariffs on them.
Seriously … wouldn’t putting a trillion of our hard-earned dollars into that be far, far better than doling it out in dribs and drabs, in grants and loans, a bit here, an overhead cost there? Because here are two ugly truths.
Ugly Truth 1) Some good-sized proportion of the population worldwide is going to get the coronavirus. Only question is when.
Ugly truth 2) Remember that trillion dollars to pay for the losses occurred so far during the nationwide lockdown? You know how long the lockdown has been going on?
One week. One. Stinking. Week. And it’s already cost A TERABUCK OF OUR TAXPAYER MONEY. And the government is talking about it lasting a month?!? Madness of the highest order.
A trillion to prop up one week? What say we suffer an attack of sanity, cancel next week’s lockdown, and put the trillion we just saved in just one stinking week into ventilators and beds and field hospitals?
Because it will hit, and the only question is how prepared we’ll be when it hits. All this stick-your-head-in-the-sand is doing is delaying it. Why? Well, theoretically so that we can be medically prepared for it with enough beds and ventilators and the like. Which is a very good reason. Gotta have more beds and ventilators than you have sick people. Medical preparation is what we want to achieve.
Given that being medically prepared is the over-riding issue, how about we
a) stop this mad stay-at-home failed experiment,
b) get America back to work,
c) continue with all the precautions we spent all this time learning, wash my hands, don’t touch my face, no sex with fruit bats, go back to disposable plastic grocery bags, social distancing, and most importantly, spend that trillion we just saved on d) …
… you know … urgently, four-alarm urgently, wartime production urgently, getting medically prepared for the wave that we’re damn sure is going to break? Buy field hospitals. Pre-position them. Stockpile ventilators. How many field hospitals does the Army have? Put them all on standby to be rushed to an overloaded city. Buy test kits. Pay double pre-war prices for everything if some people can provide it in a crazy rush. GET READY … and critically …
END THE AMERICAN LOCKDOWN!!! We cannot afford a dead economy costing us a trillion a week.
Best to all,
PS—Further discussion of the economic aspect of the coronavirus epidemic, as well as of other extraneous and forbidden topics, is going on over at my blog, Skating Under The Ice.
CONTACTING ME: Twitter @weschenbach, or at my blog, “Skating Under The Ice“. Click on “Tips and Notes” at the top of the page.
Source of data: https://www.worldometers.info/coronavirus/#countries
IMPORTANT NOTE: I use a logarithmic scale for the deaths, because on that scale exponential growth, the scary fast kind of growth, plots as a straight line. This lets us know when the increase in deaths is slowing down.
So I thought I’d take a look once again at the most basic of numbers, the number of deaths. I find that there are several patterns. First, there are countries and states that went up fast, dropped fast, and are now near zero. Here are some examples.
(I’ve removed the one-day addition of ~ 1800 deaths to the NJ record, which also was showing up in the US record, as we don’t know when they occurred.)
Then we have some that peaked, dropped some, and have been either dropping slowly or running more or less horizontal.
Some seem to be at or just past their peak …
And finally, there are a few that are still rising … of course, these are the ones that get all the media attention.
And what about the planet as a whole? Here’s that graph:
Onward, ever onward …
Yes, it has been a while since I posted here. I have not been idle. Instead Ive been looking for some way to make sense of the corona numbers, and guess what?
I can’t do it. Here’s why. There are four important numbers in the game—positive tests, negative tests, hospitalizations, and deaths. Each has its own problems.
The positive and negative tests used to refer just to detecting active cases. Now, however, we have test for both active cases and antibody tests to determine if someone has ever had the virus … and unfortunately, there’s no distinction in the data between these totally different tests.
Hospitalizations were no problem when only corona patients were in the hospitals. But now that elective and other procedures are happening again, say some young guy goes in for a vasectomy. He gets tested, he’s not an active case but he has antibodies. So guess what? His hospitalization is now a COVID hospitalization. So they test his wife and two kids. They had it too, asymptomatic, but they’re now three more positive tests.
Then there’s the problem of repeat testing of people who have the disease. If someone gets six tests, then it’s reported as six separate cases even though it’s only one case.
Finally, the deaths. The CDC in its infinite stupidity has said to categorize a death WITH coronavirus the same as a death FROM coronavirus. They’ve also said that if a physician even SUSPECTS, not that corona caused the death but suspects that corona somehow contributed to the death … mark it as a death from coronavirus.
This has led to goofy results like positive test percentage going up while hospitalizations are going down, and the like.
And despite looking for some way around all of that, I’ve not been able to figure out how to overcome all of those changes that have occurred since I started posting the graphs. Garbage in, garbage out, as they say.
So I’ve decided to declare that, while my work may not be done here, I’ve done all that I can here.
My profound thanks for all of the support that I’ve received. If I can make further sense out of this data goatrope I may post more graphs, but until then … a couple of last ones.
Stay well, stay healthy, stay happy,
In which I take a look at a bunch of states that are supposed to be in trouble from easing lockdowns … but first, the usual.
Not seeing the dreaded second peak yet, but hey, I was born yesterday …
Not much interesting for a bit, but today, Arizona is a puzzle. First, the countries:
We now have four South American countries that have earned a spot in the top fourteen graph shown above. We’ll see where they end up.
Next, here are the states:
Still no sign of any jump in positive test percentage from the mostly peaceful beatings, arson, and riots … and since the median incubation period is only five days, we’re not going to see any jump.
Finally, here’s the Arizona mystery in a nutshell—positive tests rising, hospitalizations rising … but deaths falling or level. The positive percentage started rising around May 15.
The rise in positive tests is closely followed by the number of hospitalizations, viz:
But the deaths never got the memo …
Four possible causes, off the top of my head.
1) The virus is mutating to a less lethal version. Several epidemiologists have suggested this might happen.
2) The demographics have shifted, with more younger people getting infected. I’ve seen news reports to this effect but didn’t pay much attention.
3) Improvements in treatment regimens have improved outcomes.
4) Bad data. All corona data is always suspect.
Best to all, stay healthy,
A day off, a day on. Still no sign of any uptick in positive test percentages from the protestolooting …
And for fun, here’s Wyoming …
Today I looked at a variety of countries and states. First, as usual, the overviews:
I see that Peru, Brazil, and Ecuador have all made it up into the top scoring countries …
Next, some daily deaths curves. First countries. Ecuador, Mexico and Brazil look like they might be at or near the peak.
Peru and Colombia, on the other hand, not so much …
Next, states. There’s been a lot of talk about a second spike in deaths … might happen in a few places, but not in most states.
Here in California they’ve just decreed that my barber, a charming Thai woman, can go back into business … if she hasn’t starved by now.
Yeah, I know, I didn’t post for two days. So sue me, I was getting tired of fighting the good fight and people responding with “but muh cases!
Folks, cases are MEANINGLESS. Double the testing and you’ll double the cases. Halve the testing and … well, you do the math.
So here are some real numbers—deaths, and percentages of positive tests.
Sweden still refuses to play ball with the doom merchants …
There have been no spikes from the protestoriots …
And despite lots of hype about Arizona and Texas … deaths continue to decline.
And the beat goes on … here, the night wind off the ocean brings the smells of adventures past and adventures future … stay well, dear friends.
Still waiting for the second peak …
Still no “second peak” from the riots …
Sweden continues to confound …
And while most of Europe and the industrialized states have passed their peak, there are still countries out there who have not done so …
That’s all the news that’s fit to print …
Countries, states, individual countries, states with riots, the hits just keep on happening.
Next, a comparison of a couple of countries. First the UK:
Well past the peak in the UK, with no sign in positive test percentage, the leading indicator of any bounce in the hospitalizations and deaths.
You can see that leading indicator quite well in the data from Israel …
As you can see, the positive tests bottomed out on the 20th of May. Now, here are the Israeli deaths.
Following the bottoming of the positive test percentage on May 20th, the Israeli deaths bottomed out about a week later, on May 26th.
(However, before you start getting all concerned about Israel, look at the units on the left side. They’ve gone from one death per day on average, all the way up to 1.5 deaths per day on average …)
Finally, having seen how well the percentage of positive tests work as a leading indicator of deaths, here’s another look at some of the US states which had large riots.
No spike in infections after all this rioting with little pretense of masks and none of “social distancing”? Can we finally admit that the fat lady has sung and it is time to END THE AMERICAN LOCKDOWN NOW! The facts are obvious to everyone but politicians and idiots … but I repeat myself.
My best to everyone, apologies to Mark Twain,
The usual. And California.
Now, it seems to me that the countries that have done the worst, the ones that for whatever reason are hit the hardest by the virus, all peak out in the region of 4,000 – 7,000 deaths per ten million. The only places worse than that are the US states where the government (usually the Governor) mandated that hospitals transfer COVID-19 patients to the nursing homes. Hard to believe, but it happened. That’s pushed the deaths in New York, Connecticut, and New Jersey up around 15,000 per ten million, the highest I’m aware of anywhere.
Deaths have peaked in almost every state, as well as in the US as a whole.
End the American Lockdown NOW!
The usual, plus some thoughts on the elusive “second peak”.
The latest interesting turn in the coronamania is the emergence of the protests and riots involving tens of thousands of people in close proximity to each other with lots of screaming and yelling and chanting. Bizarrely, it’s lead to epidemiologists abandoning science by saying that social distancing is sooo last month when it comes to riots … I’ve heard of situational ethics, but situational science??
A county near me has issued official regulations. They limit “social gatherings” to 12 people, and “protests” to 100 people … like the virus could tell the difference. Seems like if you and 50 folks want to have a picnic, just tell the CoronaConstables that you’re protesting the lack of Vitamin D in the underprivileged …
In any case, lots of folks have said that there would be a “second peak” in the COVID cases due to the near-total lack of social distancing in big groups in dozens and dozens of cities across the US.
Me, I’ve said since before the start of the protests/riots that the US lockdown did little to slow the spread of the virus … and that as a result, breaking the lockdown wouldn’t make much difference. So we are getting to test that bigtime.
Now, the earliest warning that we will have of a second peak will be a change in the percentage of positive tests for the virus. If we’ve been getting e.g. five positive tests per hundred tests and we suddenly start getting seven positive tests per hundred tests, that’s not good. In some cases, this will show the disease even before the onset of symptoms.
So here is the percentage of positive tests for six of the states that have seen the largest public protests and raids.
There is a common misunderstanding that we wouldn’t be able to see any change in the tests earlier than fourteen days. But that’s not what the CDC says. I quoted it on the chart so folks won’t be confused. It says that by the end of the first five days, fully half of the people will show symptoms. And if that were the case, now that we’re twelve days out from the first protest, it should be clearly visible in the chart above.
And the beat goes on … stay well, dear friends,
Still looking for the dreaded spike in the rate of positive tests … JHU is late again, it’s the weekend.
Here are the rates of positive tests for a half-dozen states that had large riots/protests. Still nobody headed for the sky.
Here are the daily death totals and the positive test percentages for Florida and Georgia. People have been watching these states because they were the first to suspend the lockdowns. Deaths are a “trailing indicator” of the spread of the illness. On the other hand, the rise or fall of the percentage of positive tests is a leading indicator—it would be the first indicator of an upcoming rise or fall in the spread of the virus. First, Georgia.
No particular sign of any spike, and in neither state has the medical system been overwhelmed.
Seems the riots have rendered the coronavirus incapable of spreading … or at least we’re not seeing the dreaded “spike” in cases, hospitalizations, and deaths.
Like the cops say at the scene of yesterday’s accident, “Move along, folks, nothing to see here” … here’s another day’s worth of the states where the most intense rioting and protesting went on (and in some cases is still going on).
Not spiking that I can see.
With all the protests and riots, I take a look at the most sensitive indicator of an upcoming spike—the percentage of positive tests. First, here are the countries and the states.
Next, here are the positive test fractions for those US states where we’ve seen large riots/protests, with people jamming the streets. Here’s that data:
Ten days since the first riots/protests, and not a single twitch in the percentage of positive tests. If there’s gonna be a second peak, it’s not visible yet.
Onwards, stay well,
I finally got fed up with Worldometer and I changed my data source for my country data to Max Roser’s wonderful site, Our World In Data. Curiously, almost no visible difference from the Worldometer data. Here’s that chart:
Next, the Johns Hopkins data is below. Note the difference in the vertical and horizontal scales from the graph above. Nowhere are the deaths greater than a tenth of one percent of the population.
And the states. Here, New York, New Jersey and Connecticut are all above 1% …
And a couple other charts …
That’s all the news that’s fit to print, as they say … stay well, everyone.
The Worldometer site is giving me grief. They’ve changed the format, but only for certain pages. A week ago it was Netherlands and France. Then when I’d rewritten my code and got it running again, today it’s doing the same thing to Germany and Sweden. Grrr … fortunately, JHU data is back on line. So for the moment I’m down to two graphs about total deaths. However, there’s a bunch more, looking at the percentage of positive tests.
Not much new there, the inexorable march continues. However, there’s lots of interest in the percentage of positive tests. IF the country testing data is accurate, the percentage of positive tests will be the first warning of any “second peak”. Here, without much in the way of commentary, are a whole variety of different countries. Alphabetical, because why not?
The Czechs seem to be in a bit of a holding pattern.
Iceland has tested the widest. They also have the smallest percentage of positive confirmed tests of any country.
Unlike all of the previous countries, the spread of the virus and the load on the Mexican medical systems has not peaked.
To date, most of those countries, including the UK and the US, are not showing signs of the dreaded “second peak”. Me, I rate it as “infreqent” and “doubtful”. I predict we’ll see flareups in various places, but very few countries will see a significant second peak.
… but of course, only time will tell.
New month. Usual plots, plus a rumination about excess deaths.
People have been saying that we can use what are called “excess deaths” to see how many corona deaths have occurred. The Financial Telegraph just put out the first dataset of excess deaths. Here are excess deaths as a percentage of coronavirus deaths.
As you can see, excess deaths won’t help us because the numbers are all over the map, with excess deaths ranging from 60% to almost 300% of corona deaths.
Last day of the month. Corona news totally eclipsed by rioting and looting. Probably just as well. JHU hasn’t updated their data. Here’s country and state data …
Next, the most sensitive forerunner of any “second peak” would be an uptick in the percentage of positive tests. So here is that graphic for the top twelve states with the highest death rates:
No clear indication of any “second peak” in that data.
Next, are there countries that should not open up yet? Sure, those where the deaths haven’t peaked. Remember that deaths are a trailing indicator, so when they have peaked, it means we’re also past the peaks in cases and in hospitalizations. Here are some countries that perhaps shouldn’t open up quite yet:
Onwards, ever onwards … stay well, dear friends.
JHU data is late again, so here’s the countries and the states, plus a look at winners and losers.
Here are the top thirty countries by deaths per capita …
And here are the states, with a vertical black/yellow line showing the worst major country by death rate, Belgium.
Finally, here’s one of the reasons that the US has so many deaths … because no less that 41% of deaths are in nursing homes. Even in progressive California the government can’t get it together to protect nursing homes, and as a result OVER HALF of California deaths are in nursing homes.
I put this on the State Governors and governments. The Feds don’t regulate nursing homes, the states do. Here’s a look at the problem.
End the American lockdown NOW.
The covid story continues …
I’ve put in the updated projections for total deaths for New York and the US.
The graph above shows that Sweden (red/black line) and the US (orange/black line) are running right in parallel. However, this may not continue. The model that has tested best projects the following:
Note that because of the lockdown in the US and not in Sweden, the US is projected to stay at about the current number of daily deaths until September. In Sweden, on the other hand, daily deaths are projected to drop steadily down near zero by September.
And as a result, the projected deaths per million by September are as follows:
United States – 611 deaths per million.
Sweden – 612 deaths per million.
I guess we’ll know by September whether this model is any better than the others.
Not much tonight …
Get plenty of sleep, stay healthy,
The usual graphs, plus a look at what the death rate correlates with. First, the countries.
Sweden and the US have been running almost exactly parallel for six weeks … here are the US states.
No big changes, no “second peaks”.
Now, I said above that I’d been looking at the correlation of the coronavirus death rate per capita and a variety of other possibly related variables. Here are those results.
As you might imagine, the largest correlation is with total cases per capita. After all, without cases, you don’t have deaths …
Next is age. There are three categories: median age, 65 and over, and 70 and over. All three of them correlate with total deaths per capita. Not only that, but the older the group, the better the correlation. This supports the observed connection.
Next is tests per thousand. Again this would be expected to correlate with deaths—no tests, no cases, no deaths.
Next in amount of correlation is female smokers. That would make sense on its own, but curiously, male smokers are negatively correlated. Go figure.
Next, wealthy countries report higher death rates, and very poor countries report lower death rates … odd, huh? But note that this does NOT mean that wealthy countries actually have higher death rates. It means that wealthy countries have more tests, which leads to more reported cases and more reported deaths.
The one big surprise to me was that population density wasn’t a significant factor in the death rate. But then I realized that many countries have big cities and lots of country in between … with the deaths occurring mostly in the cities.
Finally, the huge elephant in all of this are the largely preventable deaths in the nursing homes. This is a new feature, in that flu preferentially infects and kills the young. Recent estimates are that in the US, a stunning 40% of all deaths have been in nursing homes. So however bad you think it is out in the larger world, it’s only two-thirds that bad …
To close out, here’s the correlation matrix for all of the above variables. Blue is positively correlated, and red is negatively correlated
Lots of interesting things in there … older median age is associated with more female smokers and more hospital beds. Hmmm.
Best to all, rock and roll, it’s cooled down after a warm afternoon and we’re going to take a walk in the forest around our house.
A closer look at California, along with the usual graphs:
I note that Peru has displaced Brazil in the graph above of per capita deaths.
Now, for California. Here’s the daily death toll.
As you can see above, deaths peaked at about the first of the month, a while ago now. And below is confirmed cases (positive tests) as a percentage of total tests.
If we were going to see a “second peak” here, the first sign would be that number, the percentage of positive tests, increasing. And despite the state removing restrictions starting on May 4th … still no sign of the second peak.
Finally, I kept seeing people saying “But Los Angeles County is still in trouble, so we can’t remove the lockdown”. So I got the county data. Here are the daily deaths for four major population areas—LA, Santa Clara, Sacramento, and San Diego counties.
LA County deaths peaked at about the same time as the statewide deaths. Next, here’s Santa Clara County, with one of the highest per-capita death rates in the state. It was one of the earliest places hit … and also one of the earliest dates of the peak deaths.
Finally, the city and county of San Diego and the capital county and city of Sacramento
Continuing the lockdowns in California at this point is all pain and no gain. We’re past the peak deaths everywhere; our economy is in the crapper; millions are unemployed; cities, counties, and the state are hemorrhaging cash; businesses are closing never to open again; and people are wondering why they can buy at giant businesses but not at small stores.
End the lockdowns now, wherever the deaths and the weight on the hospital system have peaked. And that’s most everywhere.
Stay well, wash hands, no hanky-panky with pangolins, and remember—the virus can only spread five and a half feet (1.85 m) at a time.
Well, I came up with what I think is a sensitive measure of the progress of the disease … more on that below. Here are the countries.
Sweden and the US continue to run in parallel … go figure. I note that Brazil has overtaken Germany in the top death rate countries shown above.
People keep claiming that without the anti-social distancing and shelter-in-place the US would have a million deaths. My guess is that the death rate in New York/New Jersey is the absolute worst case scenario. And that rate for the US as a whole would give us about 340,000 deaths (right scale).
Anyhow, like I said, I had an interesting idea regarding determining if we’re headed for a “second wave” of cases and deaths.
My thought was that we can look at daily cases, not in isolation, but comparison to the number of daily tests. So here’s that graphic for New York and New Jersey.
In both states, the percentage of positive tests started out around 50%, and over time it steadily dropped to around 5%.
Now, when the “suffer-in-place” regulations are removed, IF that releases a “second wave” of infections the first sign should be an increase in the number of cases found per test.
In that regard, here are two states that have removed a number of regulations, starting about a month ago. In Florida and Georgia people have been saying that we’re seeing high numbers of cases (positive tests) … but is that so?
Hmmm … some ups and downs lately, but it’s under 3% positive. Here’s Georgia:
Same as Florida. Here are some more states of interest: