Regular WUWT contributor Willis Eschenbach always goes to data when questions and issues arise, he has been plotting the official death rate data from the Coronavirus almost daily, and will continue to do so. I’ve dedicated a permanent WUWT page to this. We will continue to add to this page as needed and as Willis makes updates.
Note that it is now a menu item in the left most section of the WUWT Menu bar, right under the header image.
Friday’s graph:
See the full page of graphs here: https://wattsupwiththat.com/daily-coronavirus-covid-19-data-graph-page/
Graphing deaths is not very useful.
1. There are two, maybe three, or even four fatality rates involved. As I see it, there is: fatality rate below ICU capacity, below hospital capacity, above hospital capacity, and after supplies (and staff) run out.
2. Deaths, by their very nature, are a lagging indicator, by about ten days, give or take. This is particularly problematic when the doubling rate is estimated at 6 days.
3. A far better, but much more difficult graph would be available hospital beds per thousand people. I’m sure that epidemiologists will be using such a graph by 2030.
Yeah, and some countries report death by Corona when the patient had the virus, regardless if he died of that, and others don’t even bother to test the deceased.
Willis’ plots and his analysis of the Diamond Princess gave me another angle on the likely endgame case fatality rate (CFR), using ‘personal’ information from a doctor in Wuhan.
The ultimate tested Diamond Princess result was 705 infected out of a starting total of 3711 passengers and crew. 19%. That is likely a bit skewed high by the ‘experimental’ circumstances— more elderly passengers, higher viral titer in the confined conditions.
Of the 705, 392 had symptoms defined as fever >100.4F. 10.6%.
The Wuhan doctor said 81% of his patients recovered after about 10 days, and 19% worsened. Of the 19%, 14% became serious (needing supplemental oxygen. 5% became critical, needing a ventilator. (These facts are the apparent source of Cuomo’s NY ventilator requirement estimate of ~27000 when the state only has ~3000–hence the need to ‘bend the curve’.)
The Wuhan doctor said that in his hospital’s experience, about 20% of serious and 80% of critical die, so about 6.8% of those admitted.
Now a 0.19 infection rate*0.106 symptoms*0.19 not recovering naturally*0.068 fatalities among those is (worst case) in the end 2 fatalities per 10000 people. UNLESS the curve is bent by social distancing, frequent hand washing, and not touching mouth, nose, eyes.
I think your math is messed up. Your are applying the 10.6 symptom rate of the total population to the infected population. The symptom rate of the infected population on DP was 392/705 or 55%. What am I missing here?
Is the “critical” percentage 5% of the “serious”, or 5% of who “worsened”?
I’ve written that question three different ways, and I’m still not happy with it, but it’ll have to do.
Using those numbers in various ways, if the ventilator requirement estimate is 0.19 x 0.14 x 0.05 = 0.00133, then 27,000 ventilators means 20,300,000 “worsening” cases. If it’s 0.019 x 0.05, then the number of “worsening” cases becomes 2,800,000.
Since ~ 19% of all cases “worsen”, then 20,300,000 of them means a total infected population of 20,300,000/0.019 = 107,000,000 total. If the number is 2,800,000 “critical” cases, then the total infected population wold be 2,800,000/0.019 = 14,700,000 total cases.
If this bug truly has an R0 of 2, those thresholds would have long been passed, wouldn’t they?
I’m over 60 with no underlying conditions. I’m happy to self-quarantine for as long as the government wants, along with all others of any age advised by their personal doctors to do so because they have underlying conditions. But, given the numbers that are coming out as we get more data, for goodness sakes, the rest of the population needs to get out there and work and have some fun. This is getting ridiculous.
It seems that some of the increase in mortality is associated with increased testing. This is counter-intuitive unless clearing the backlog of samples is being used post-mortem to re-classify deaths as Covid19. If that is the case, the numbers should be adjusted for actual date and not just included in “new” deaths. This may reshape the curve.
An absolutely sensational graph. Great work, Willis!
I’d love to see Germany added to this graph. They’re a large country, comparable to the U.S. in technology, and close to Italy and Spain, which are already on the graph.
I know the graph starts to get busy when adding another country, but if so, my suggestion would be to add Germany, and drop Iran. The U.S. and Australia, U.K., Canada, etc. are so different from Iran, that I think the situation in Iran provides no useful information as to where we (in the U.S. and the other countries I mentioned) are possibly headed.
In light of the recent WUWT story about the inverse correlation between coronavirus and malaria, I noticed another correlation, IQ.
Coronavirus struck first in countries with high IQ. So, what’s the experience been like in countries with low IQ? If we look at countries with an average IQ of 67 or lower, there are only 33 confirmed cases and 20 of those are in Cameroon. IQ by country Half of those countries had zero confirmed cases.
You could object that Equatorial Guinea has a tiny population but Iceland, with a quarter of that population, has 473 confirmed cases. Cameroon, with a population of 24 million is doing much better than Canada with a somewhat larger population.
So, is low IQ protective against coronavirus? LOL Maybe there’s a lurking variable.
I personally think it has to do with how tall you are.
Yes taller you are the higher mortality rate. This due to vast majority of children recovering, and children tend to be shorter than adults.
I’ve read that there’s a correlation between IQ and wealth. More wealthy, more tests, maybe?
As your IQ increases, more lucrative careers are available. link The graphs are at the very end of the paper.
On average, your MD is probably smarter than you. Having said that, I’m guessing that there are a few WUWT denizens who are smarter than the average MD, averages and distributions being what they are.
Most informative, Willis. The rise on the curve for the U.S. is quite variable. I would love to take the graph apart and compare it to factors, such as the weather in the places of highest death rate growth, some four to seven days prior to those wiggles that flatten out.
It also suggests that another 10 days of isolation will indicate pretty clearly whether the present strategy is succeeding, and hopefully the ultimate rate would be about 2 per million.
I also wondered about the strange shape of the US curve. Is it a matter of low initial testing and recent increase reducing the number of previously undetected cases? Or possibly a blending of multiple epidemics with different start dates?
I live in Canada and I follow WUWT daily. Would it be possible to provide a way for other countries to see the deaths versus days graphic? When this is all over, I think that these curves will be a key measure of how well each country reacted to the situation.
Steve Rowland
Willis … Please do a breakout on your new page separating out Hot spots like NY WA from the rest of US. WA state and NY are driving the curve for US. Would be comforting to the rest of the US showing how well we are actually doing.
Completely agree that breaking out the hotspot states would be very helpful. I’ve played around with this a bit and it looks like Washington has started to plateau whereas other states have not yet. The outbreaks are clearly regional and using stats from the entire US muddles the picture a bit.
That being said this looks like the best format I’ve seen by far. Deaths is the only meaningful statistic as number of cases is hopelessly confounded by testing disparities. Per capita deaths is also much more meaningful the absolute deaths.
One other note is that the increased slope of deaths in the last few days is also likely due to increased testing availability rather than being real. Almost certainly some deaths we being misclassified. Not to downplay the personal costs for those who lost loved ones but 250 deaths in a country of 330 million is a microscopic fraction. A dozen misclassified deaths could significantly change the apparent trend.C
Doc, the problem is the lack of data. The Worldometer has current data for the number of deaths by state here, but no historical data.
You are right about the skew, however. Of the current 281 US deaths, 163 (585) are in New York/Washington/California, which have 20% of the US population.
If you have historical data by state I’d take a look.
w.
Now 791 total, 271 NY, 44 NJ, 55 CA, 123 WA.
Diamond Princess is probably one useful source and should be very encouraging. 100% of people on that ship had to have had constant, close, extensive contact with the virus, yet less than 10% actually developed a symptomatic infection. That is basically the same as the estimate of flu cases in an average year in the US and lower than the swine flu rate. If you adjust the population characteristics of the ship for the population characteristics of the US, you get a mortality rate of .5% or less, and again that would assume that 100% of the population had the same close, extensive and constant exposure to the virus, which is obviously not the case. At the end of the day, when we have more actual data, there is absolutely no reason to think that this virus will have a fatality rate substantially higher than influenza viruses. That is the typical course of fatality estimates. Meanwhile we are destroying our economy, which in itself will cause in worse health harms, among other things. #thecureisworsethanthedisease
You think?
As soon as it became apparent that COVID19 was on the ship all of the passengers were put in to Isolation in their cabins ie quarantined.
Since when has that ever been done to the general public for the Flu?
Then they had air piped in to their rooms from other parts of the ship, and ate meal cooked and delivered by the likely infected crew …
Steven Mosher tweeted a link to a good study of the Princess, I reposted it on that thread … hang on … OK, it’s here.
w.
That is a very nice study.
However the serious patients were removed to Japanese hospitals that were not in a state of being overwhelmed like Italy.
The ship patients received the very best of care, those in Italy are IN ICU, open wards, on trolleys and on the floor. With life & death decisions being made on who they save.
The medics are now patients and dying, this was 2 days ago and it is much worse there now than then.
https://www.dailymail.co.uk/news/article-8129499/More-2-600-medical-workers-infected-coronavirus-Italy.html
Were you aware of how bad it actually is in Italy?
ps the Italian daily mortality rate has increased again today and now stands at 793.
They are desparate for the lockdown to start working, but they still have so many in the system from before the lockdown.
Other small towns are in a much better situation.
Anyone attempting to parse the disease or the epidemiology by using available data is bound to be wrong by the time anyone else can even read what they wrote, both because of incomplete or outdated info, or because of the rapidly changing situation.
No one can use math to prove anything when the numbers are bad and rapidly changing.
JMO.
This situation is too fluid to stand up to conventional analytical techniques, or so it seems to me.
Read what we were saying two weeks ago, one week ago.
People were talking about how except for one nursing home, there was not much of a problem.
Bill DeBlasio refused to close NYC schools until last week, even though schools are where diseases go to spread.
Public health authorities were still opining until recently that asymptomatic transmission was unlikely to be occurring.
Hardly anyone, possibly exactly no one, has a clear picture of the current real time situation, regarding any number of parameters.
We still have people right here who might have been considered fine logical thinkers, making statements that are in direct opposition to reality.
Public health officials are in many cases people who are vastly under qualified who are in positions due to political sinecure…or whatever the proper terminology is for people who only have a job due to supporting the right person in an election, or knowing someone who is in the position to make public job appointments.
One take away, for me, or this whole unfinished episode is…this was the time that our progression towards living in an Idiocracy really bit us in the @ur momisugly$$ hard for the first time.
For Brits
Daily Telegraph published (behind paywall) longish article
Coronavirus: The unintended consequences of the UK lockdown and why millions of people could already be infected
Article is discussing what various models say, they are presented to the UK government to consider when policy is formulated. I put it on my webpage strictly for my own use. OK!. Anyone who intends to read please obtain proper authorisation from the publisher. OK!
The international numbers, not all of which qualify as data, aren’t fit for comparative purposes. Even in countries with honest reporting of deaths, procedures vary. For instance, Germany doesn’t test the dead, while France, Italy and Spain do. There are no doubt other reasons as well for Germany’s lower death figures, but testing is a big part of it. Some people who died of pneumonia there, or other apparent causes, had the Wuhan virus.
Good summary.
Glad you liked it. Could have said even more about testing of the living as well, of course.
I’m not sure that procedures even within countries are sufficiently consistent for valid comparisons. But what are we to do, except make do.
Influenza Like Illness data probably has captured many of the early COVID-19 cases before specific testing started and indicates that the virus has been wide spread in the US since February and the peak has passed. https://www.amgreatness.com/2020/03/19/dangerous-curves/
That’s a very interesting article (and some good comments). One sentence especially caught my eye:
“But since the disease originated in China in December at the latest,
it’s highly unlikely the number of reported cases in the
United Statesall countriesbetween January 1 and late February is accurate.”
This is exactly what my bullsh!t-sensor has been telling me since the media hype started.
Willis/Anthony
Thank you for the continuing presentation of data and information.
Couple comments/questions:
Active cases appears to be highly variable data with lots of variables as to testing penetration. Lots of questions as to what the actual infection rate is at any given time.
Death Rates are all over the place. Extreme variability depending upon age/underlying condition, care provided etc. Regardless of specific treatments, as case management gets more informed, I would expect to see better outcomes (assuming there is any real chance of recovering from pneumonia). Couple this with more draconian protective measure for the most vulnerable, and I would expect the death rate to further decouple from the actual infection rate.
I am having a hard time connecting the concerns with hospital bed space, and specifically the most serious cases meriting ventilator use and actual case numbers. I see lots of comments, both “official” and MSM rumors that bed space and ventilator usage is under pressure. What I can’t find so far is actual numbers/data/statistics for actual ventilator usage for CV-19. I surmise that if we had good data for this metric, then we would have a better handle on the actual impact of CV-19. Pragmatically I understand that from a macro point of view it is somewhat of a binary concern: either you have enough ventilators or you don’t (presuming that lack of a ventilator significantly leads to worse outcome…ie death). At any rate, as I try to get some kind of coherent handle on how CV-19 is actually progressing, I’d really like to see solid data on ventilator usage….specifically how it relates to published infection rates, death rates and test results rates. I assume this data is out there…it really is the center of the most significant health impact of CV-19. Using the worldometer data (seems a decent data source), and assuming the “serious cases” include ventilator usage, I really can’t see any meaningful information. The values are all over the place.
I have been searching, reading and researching for this ventilator usage data, and I am coming up dry (expect for the pervasive…”not enough”).
At any rate I will continue to search high and low, but ask the incredibly talented wuwt community for assistance.
Specifically: What is the actual data for ventilator usage for CV-19 patients. Time and region. Duration. Demographics. Outcome.
Thanks for any help, I will keep looking.
Ethan Brand
Ethan, that is a problem, there does not seem to be any central repository for data on ICU COVID19 usage.
However one clue is to look at how many people have been infected and how many still are infected (active cases).
There is a lot of individual country’s data on ICU use including China, the Diamond Princess, plus quite a bit of anecdotal info from Medics.
Hi A C Osborn
Thank you. I can see the source of the “anecdotal” data (ie China, Diamond Princess, etc), but that is not dynamic data.
“However one clue is to look at how many people have been infected and how many still are infected (active cases).”
Same problem overall….the “have been infected” number has huge uncertainty.
I am betting the data I seek is out there…just a matter of finding it and assembling it.
Again thank you,
Ethan Brand
Followup, I found this link: “Penn Medicine – COVID-19 Hospital Impact Model for Epidemics”, https://penn-chime.phl.io/
Model (groan…:)).
It has some references, but I have not followed them yet to see what data is being used to support the default values used. I will email someone on their contact list and provide an update if I get something useful back.
Ethan Brand
I have been looking at the mortality rate since this epidemic started. What do you make of these facts. Italy: 4,032 deaths/47,021 cases= 8.6% mortality China: 3,253 deaths/81,304 cases = 4% mortality US: 260 deaths/19,624 cases = 1.3% Germany: 73 deaths/21,652 cases = 0.34% The US is late getting the infection wave but China was earliest, having the most cases. Germany and Italy were both infected at relatively the same time. There are many variables that affect susceptibility of populations to infection and death but something else is going on here. The virus should not be an order of magnitude more lethal in Italy as compared to it’s geographically close neighbor Germany. Your thoughts??
A small town in Italy (Prata) has the second largest Chinese population in Europe (Paris is first). There were regularly scheduled flights between Wuhan and Italy. Here’s an article that describes why so many Chinese are in Northern Italy https://www.newyorker.com/magazine/2018/04/16/the-chinese-workers-who-assemble-designer-bags-in-tuscany
People don’t understand graphs. Read this for more insight. Data is data.
I would like to see more comparisons to H1N1. Why is the daily data for H1N1 so hard to find. All I can find is a few months from 2009 with daily data. The final numbers are staggering for H1N1, but the daily data from mid 2009 through Obama’s emergency declaration in 2010 are missing. Anybody have the daily data for all of 2009 and 2010 for H1N1?
I became very ill with what I believe was H1N1 towards the end of 2009. The hospital had so many cases that they stopped testing people. As a result, a lot of very relevant data for epidemiological purposes is missing. Don’t trust data collection.
They may start to do the same with the WuFlu:
In hard-hit areas, testing restricted to health care workers, hospital patients
https://www.msn.com/en-us/news/us/a-new-message-on-coronavirus-in-hard-hit-areas-dont-get-tested/ar-BB11vpAa?ocid=spartandhp
As of July 2019, there were 74 countries having populations each less than 1 million people. Therefore, just a single COVID-19 death in any one of these would put that country at the 10 deaths in 10 million population mark, or at the midway point along the logarithmic ordinate axis of the graph in the above article.
It is a mathematical calculation, but not too meaningful due the error bars associated with a calculation based on a single occurrence.
Caveat emptor.
I’m trying to repost my comment.
For the latest report on italian deaths (2020 03 20), please go to this link of the italian Istituto Superiore di Sanità
https://www.epicentro.iss.it/coronavirus/bollettino/Report-COVID-2019_20_marzo_eng.pdf
This is a very bad news.
“The death toll from the coronavirus outbreak in Italy rose by 793 to 4,825 on Saturday, officials said.”
This is 1,500 more than the China’s total with 20 times more numerous population.
Only if you actually believe China’s data, which based on anectdotal data in February it is not very likely.
Things like photos of dead people in the street, people self isolating dead in their apartments, COVID19 deaths recorded as Pnuemonia, the dead being cremated 24/7 for weeks without autopsy etc.
@Derg
https://www.worldometers.info/coronavirus/coronavirus-age-sex-demographics/
Willis, I have shared your chart and the link to this article and to your chart page on a couple of blogs. One commenter asked how you can be sure of your data, especially the data for China. What are you doing to ensure the accuracy and integrity of your data sources?
Today’s UK numbers are bad, update here:
http://www.vukcevic.co.uk/UK-COVID-19.htm
More pain ahead, but the UK will turn the corner.
Log scale death comparison
I’m wondering what is going to happen with the various states here in the US.
Since they can’t print money, they have to run a balanced budget.
With the economy slowing dramatically, tax revenues have to be falling. At the same time expenses are going up.
What happens with state and local governments when the rainy day funds run dry?
Hand outs from above.
If that happens, total systemic collapse of government won’t be far off.
The current round of handouts is already dangerously close to unaffordable.
The bailouts are looking to be on par with the 2008 banking crisis, maybe worse. We’ll have to see what is passed before we know what’s in store.
Unaffordable is putting it mildly. It’s especially concerning that there’s little challenge against these executive orders by the President, governors and mayors. A bad precedent is being set.
Remember what “retirement planning” meant something other than binge watching the show “Doomsday Preppers”, or “Life After People”?
Let me say that again without the typo:
Remember when “retirement planning” meant something other than binge watching the shows “Doomsday Preppers”, or “Life After People”?
I wrote this on another blog about 8 hours ago:
“But let’s not be blasé about the scale of the problem. I was idly playing around with the virus figures yesterday (as former mathematicians tend to) and started doing some “what ifs” with the figures in one of the places (outside China and South Korea) which seems closest to over the outbreak. That’s the Faeroe Islands. They have had 92 confirmed cases so far, with no deaths. That represents about 1.9 per thousand of their population. Their new cases per day had been going down until today, when there were 12 new cases, but they have been ramping up the testing hugely in recent days, so that’s not unexpected. The data comes from here – https://www.worldometers.info/coronavirus/ – a useful source if you’re into this kind of stuff.
Eyeballing their “Gompertz curve” of cumulative confirmed cases at https://en.wikipedia.org/wiki/2020_coronavirus_pandemic_in_the_Faroe_Islands, I guess they’re probably around half way to their final count. Last Tuesday their chief medical officer said he thought that most of the people in the islands had by then been exposed to the infection, but I think he was probably over-optimistic. The reason the Faeroes are “ahead of the game” is that the virus seems to spread quicker among small, isolated populations. I discovered this when I idly sorted the figures by cases per population, and found that San Marino (the worst of all), Faeroes, Iceland, Andorra and Liechtenstein all came out in the top 10.
So my best guess right now at how many confirmed cases they’ll finish with is 4 per thousand population. The second part of the equation is how many of those confirmed cases die. This is a function of the quality of health care in the country more than anything else. The UK is currently running at about 4.5% of confirmed cases leading to death. That’s way below Italy, which has its own set of problems; and also below Spain, but above other European countries like France and the Netherlands. (Above China, too! And almost four times the rate in the USA.) That may be due to a low testing rate so far, or may eventually have something to say about the merits or otherwise of socialized health care.
Anyway, if I take both those numbers at face value, I guess we’re looking at 66.44M * (4/1000) * 4.5% = about 12,000 deaths when all is said and done. To put it in perspective, world-wide deaths from ordinary flu per year are between 300,000 and 600,000, and if 0.9% of those are in the UK (roughly pro-rata to world population) we’re looking at 2,700 to 5,400 flu deaths in an average year. So the COVID virus looks to be worse than normal flu by a factor of 2.2 to 4.4. Bad – which means Boris and co are right to take some action – but not the end of the world (alarmists have been bandying around figures like a quarter of a million).”
Willis, what do you think?