Guest Post by Willis Eschenbach
OK, here are my questions. We had a perfect petri-dish coronavirus disease (COVID-19) experiment with the cruise ship “Diamond Princess”. That’s the cruise ship that ended up in quarantine for a number of weeks after a number of people tested positive for the coronavirus. I got to wondering what the outcome of the experiment was.
So I dug around and found an analysis of the situation, with the catchy title of Estimating the infection and case fatality ratio for COVID-19 using age-adjusted data from the outbreak on the Diamond Princess cruise ship (PDF), so I could see what the outcomes were.
As you might imagine, before they knew it was a problem, the epidemic raged on the ship, with infected crew members cooking and cleaning for the guests, people all eating together, close living quarters, lots of social interaction, and a generally older population. Seems like a perfect situation for an overwhelming majority of the passengers to become infected.
And despite that, some 83% (82.7% – 83.9%) of the passengers never got the disease at all … why?
Let me start by looking at the age distribution of the Diamond Princess, along with the equivalent age distribution for the entire US.
Figure 1. Number of passengers by age group on the Diamond Princess (solid) and expected number of passengers given current US population percentages (hatched).
When as a young man I lived in a port town with cruise ships calling, we used to describe the passengers as “newlyweds and nearlydeads”. Hmmm … through some improbable series of misunderstandings and coincidences, I’m in the orange zone now … but I digress …
In any case, Figure 1 shows the preponderance of … mmm … I’ll call them “folks of a certain distinguished age” on the Diamond Princess. Folks you’d expect to be hit by diseases.
Next, here’s the breakdown of how many people didn’t get the virus, by age group:
Figure 2. Percentage of unaffected passengers on the Diamond Princess. “Whiskers” on the plot show the uncertainty of each percentage.
In addition to the low rate of disease incidence (83% didn’t get it), the curious part of Figure 2 for me is that there’s not a whole lot of difference between young and old passengers in terms of how many didn’t get coronavirus. For example, sixty to sixty-nine-year-old passengers stayed healthier than teenagers. And three-quarters of the oldest group, those over eighty, didn’t get the virus. Go figure. Buncha virus resistant old geezers, I guess …
Next, slightly less than half the passengers (48.6% ± 2.0%) who got the disease showed NO symptoms. If this disease is so dangerous, how come half the people who got it showed no symptoms at all? Here’s the breakdown by age:
Figure 3. Percentage of Diamond Princess passengers who had coronavirus but were symptom-free. There was only one illness among the youngest group, and they were symptom-free. As in Figure 2, the “whiskers” on each bar of the graph show the uncertainty.
Again, a curious distribution. Young and old were more likely to be symptom-free, while people in their 20s, 30s, and 40s were more likely to show symptoms. Who knew?
There were a total of 7 deaths among those on board. All of them were in people over seventy. So even though the generally young were more likely to show symptoms if they had it, it hits old people the hardest.
Finally, according to the study, the age-adjusted infection fatality rate was 1.2% (0.38%–2.7%). Note the wide uncertainty range, due to the small number of deaths.
For me, this is all good news. 83% of the people on the ship didn’t get it, despite perfect conditions for transmission. If you get it, you have about a 50/50 chance of showing no symptoms at all. And the fatality rate is lower than the earlier estimates of 2% or above.
It is particularly valuable to know that about half the cases are asymptomatic. It lets us adjust a mortality rate calculated from observations, since half of the cases are symptom-free and likely unobserved. It also gives a better idea of how many cases there are in a given population.
To close out, I took a look at the current state of play of total coronavirus deaths in a few selected countries. Figure 4 shows that result.
Figure 4. Deaths from coronavirus in four countries. Note that the scale is logarithmic, so an exponential growth rate plots as a straight line. Blue scale on right shows the deaths as a percentage of the total population.
At this point at least, it doesn’t appear that we are following the Italian trajectory. However … it’s still early days.
Finally, a plea for proportion. US coronavirus deaths are currently at 67, we’ll likely see ten times that number, 670 or so, might be a thousand or three … meanwhile, 3,100 people die in US traffic accidents … and that’s not 3,100 once in a decade, or 3,100 per year.
That’s 3,100 dead from auto accidents EACH AND EVERY MONTH … proportion …
My best to all on a day with both sun and rain here, what’s not to like?
w.
As Always: When you comment please quote the exact words you are referring to, so we can all understand who and what you are discussing.
Terminology: Yes, I know that the virus is now called 2019-nCoV, that it stands for 2019 novel CoronaVirus, and that the disease is called Covid-19, and that it stands for COronaVIrus Disease 2019 … so sue me. I write to be understood.
Data: For those interested in getting the data off the web using the computer language R, see the method I used here.
Other Data: A big hat tip to Stephen Mosher for alerting me to this site, where you can model epidemics to your heart’s content … Mosh splits his working time between Seoul and Beijing, he’s in the heart of the epidemic seeing it up close and personal, and he knows more about it than most.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
“Seems like a perfect situation for an overwhelming majority of the passengers to become infected.”
What isolation measures were taken? Did the crew and staff split up ? were different kitchens used for the infected etc?
Willis,
Good job!
JimG1
Hi Willis. On a hunch I googled: “diamond princess” “no smoking” policy
Found this:
“We recently completed a cruise on the Diamond Princess. There was no smoking permitted inside the ship, including the Casino and Churchill Lounge. I’m not sure if there was a smoking area topside by the pool area. There was a small designated area on the Promenade Deck. On rough weather days the doors to the Promenade Deck would be roped off. Smokers would go under the barriers.”
https://boards.cruisecritic.com/topic/1492522-discuss-implementation-of-smoking-policy/page/36/
I have all along felt strongly that for a disease that exploits unhealthy lungs the U.S. should NOT be taking mortality rates from China, Korea, Iran or Italy at face value. A very high percentage of older men in all of those countries are lifelong smokers, and a lot of the women smoke too.
My guess is that the numbers in those countries will turn out worse than the Diamond Princess numbers but that U.S. numbers will look a lot like the Princess numbers, at least as regards mortality.
Their numbers wrt transmission are at odds with the one local incident I know about. A friend of mine from the East Side of Seattle discovered that he and about a hundred others had all been exposed when a friend they had all been regularly playing contact sports with was an early WA Corona death.
He says that everyone he talked to in the group had gotten sick about the same time. Knowing they had been exposed they all figured they had Corona and self quarantined. As testing came available all who got tested (about 10) came up positive.
Most had a mild to ordinary flu. Some got a sicker but all got over it reasonably quickly while the fellow who died was known to have some serious health problems.
His sense of it in retrospect: hyper-transmissive. If you get exposed your going to get it. Second, its just an ordinary flu. It’s not a big deal. Their group is not young and except for the guy with pre-existing problems most of them breezed through it.
They are all very active with probably close to zero smoking.
Smoking was apparently a very significant factor in China.
However, I’m not sure it is simply “unhealthy lungs”. One of the main effects of habitual smoking is that it incapacitates the hemoglobin which gets permanently bound to carbon monoxide and thus useless for oxygen transport. A tobacco addict will only have about 40% functional red blood cells.
Thus any infection affecting the lungs hits a system which is already severely compromised in its capacity to transport oxygen. This quickly means you need an oxygen supply and a ventilator.
This sounds not just unlikely but flat out untrue.
For one thing, the amount one smokes and the way the tobacco burns must affect CO content, and how one smokes (inhale deeply, shallowly, etc?) will have a huge effect on the amount of carbon monoxide in the lungs and hence the blood.
But just from a basic knowledge of blood chemistry, I can say for sure this is untrue.
A normal hemoglobin count is a fairly wide range, 14-17 for men and 12-15 for women (due to menstruation), and anything below 10 is called anemia.
I have had severe hemolytic anemia several times, and I can tell you, anything below 10 for a man is considered a life threatening condition.
Only 40% of red cells being functional would be equivalent of a hematocrit below the level that would sustain life. Ascending a mountain would kill you if you were not already dead.
Unlikely anyone could survive, let alone be walking around on their feet, with hemoglobin reduced to 40% of normal.
Some smokers are athletes, and in fact it used to be common.
And I wonder…why do you single out tobacco?
Is that the only leaves that produce CO when burned?
I do not think so.
Someone who smokes marijuana absorbs over 5 times as much carbon monoxide as a heavy cigarette smoker.
For the record, I do not smoke, never have touched a cigarette, take no drugs like alcohol or pot.
But facts are facts.
Smoking is bad for you, but 60% of RBCs being useless is malarkey.
And it would be 5x worse for a pot smoker because the smoke is more acrid/dense and is held deeply in the lungs for as long as possible, typically. Also 3x the tar, and that is compared to unfiltered cigarettes.
Just the facts.
Call me Joe Friday.
re:
The cilia , you left out ANYTHING about the cilia …
After all the cilia are responsible for ___________ and _________________. Right?
https://medlineplus.gov/ency/imagepages/19533.htm
Willis
Very nice article with some important information.
I think by definition those on the cruise, whilst elderly and often on medication will not be severely ill otherwise they would not be going on the cruise or allowed to join it. A cruise ship only has certain medical facilities that are suited to common illnesses not the severely ill.
It is the over 70’s with severe underlying health conditions who seem to be succumbing, but even then it is not everyone. So locking down everyone because of their age profile seems counterproductive as you will destroy the mental and physical health of those who get our and about every day and need the social aspect as well as the exercise.
Everyone says you shouldn’t compare corona virus and flu, let alone traffic deaths but why not? Plenty of lives could be saved by making flu jabs compulsory, preventing smoking and heavy drinking, making obesity a thing of the past, dropping the speed limit etc.
It is surely highly relevant to point out that during the flu season in the UK some 180 people will die every day in a bad season/ Latest figure today is 14 deaths from corona virus since yesterday.
Ps. To me, going on a cruise would be purgatory so I was surprised at figure 1 which you say are only passengers and yet includes a surprising amount of under 50’s. Definitely No crew are included in the figures, who you would expect to be in a younger age group?
tonyb
The UK appears to be in the very early stages of the exponential growth so it’s difficult to be project the number of deaths at the present. Looking at the UK figures there was an increase in cases of 400 today, based on other countries the doubling time was 3-4 days so could be 1000+ cases per day soon. One of the differences with flu seems to be the impact on the hospital and ICUs, seems like that COVID-19 has a much bigger impact/patient.
Yes. If you do the same basic math on the UK numbers, there might be, right now, about 50,000 cases in the population, of which only about 2,000 have been confirmed through testing. So in about 18 days time this would be expected to lead to around 1,000 cumulative deaths. The current lockdown measures would only lead to a lower death rate a couple of weeks later as they lower infections.
Numbers subject to very large margin of error, but I think this rough calculation and the Italian experience is a much better guide to what may happen in the US than extrapolating from the special environment of a cruise ship.
Today’s increase in the UK ~700 so new cases likely to exceed 1000 in a day or so.
Has the nCOV fatality rate exceeded the City of Chicago homicide number yet?
There’s another cruise ship data point: The Grand Princess had passengers infected with nCOV – who later debarked and some died. But 60 of them went on the subsequent cruise – I am still waiting to see what the 1st (original) cruise infection rate was, but the 2nd cruise infection rate was apparently really, really low.
Of course, the 2nd cruise – they were trapped for a lot less time…
Am I reading this right: 100% of the children on the DP had the virus but had no symptoms? If so then they were the primary carriers and spreaders of the contagion?
According to the document that Willis linked to, there were 16 children in the 0-9 age group. Only one of these children tested positive and that child did not exhibit any symptoms producing the 100% result in Figure 3.
No error bar is shown for that result, but if it had been included, a 95% bar would be one-sided reaching down to the value .05.
I would guess that the 20-50 age groups are basically the “crew”. Though it is misleading to think of them a the ship’s crew. They are hotel staff.
It seems cruise ships have about 3:1 or 4:1 guest to staff ratio.
I like R direct call method for data. Have to remember to locally cache a hash of the dataset used for configuration/data management.
Spread of the new flu (Coronavirus) is being countered with a lot of publicity and intense preventive measures. It will be interesting to see how much the intense preventive measures reduce the mortality rate of the old flu which has been averaging about 50 a day.
“For me, this is all good news. 83% of the people on the ship didn’t get it, despite perfect conditions for transmission.”
About that “perfect conditions for transmission”, we all “know” that cruise ships are like that, but are they really? For an industry which targets the older and wealthier, the industry provides all sorts of services, from cooks to entertainers. I would hope they also provide doctors and nurses.
And for an industry which is linked in the public’s minds with norovirus, I would also expect that they are quite serious about containing the spread of viruses.
How did the Diamond Princess measure up? Was the 83% good news in any way the result of something they did? Getting people to stay in their rooms?
For another study of the mathematics of epidemics, this paper is not so optimistic:
https://doi.org/10.25561/77482
https://spiral.imperial.ac.uk:8443/bitstream/10044/1/77482/2/Imperial%20College%20COVID19%20NPI%20modelling%2016-03-2020.pdf
“In the (unlikely) absence of any control measures or spontaneous changes in individual behaviour, we would expect a peak in mortality (daily deaths) to occur after approximately 3 months […] we would predict approximately 510,000 deaths in GB and 2.2 million in the US, not accounting for the potential negative effects of health systems being overwhelmed on mortality”
From their Table 1 “Current estimates of the severity of cases”
In the 80+ age group, “% symptomatic cases requiring hospitalisation”: 27,
“% hospitalised cases requiring critical care”: 71,
“Infection Fatality”: 9.3%
It deals with health services becoming overwhelmed. The worst case is when the doctors and nurses become infected, then paraphrasing someone else, “hospitals become hospices, places you go to die”.
The proper quote, from Dr. Mehmet Oz
https://www.foxnews.com/media/dr-oz-coronavirus-hospitals-weak-link
The referenced paper offers no data about infection and/or lethality by gender. I’m curious if the Wuhan virus differentiates significantly in infection rates, seriousness of the infection, and lethality between males and females.
@ur momisugly Willis
Thanks for supplying data and evaluation. Maybe, this can be of some help for your work:
https://www.euromomo.eu/
“Mortality monitoring in Europe
Welcome to the EuroMOMO website. We publish weekly bulletins of the all-cause mortality levels in up to 24 European countries or regions of countries. The weekly bulletin is published every Thursday around noon.”
As someone who lived and worked in Asia for 20 years the China curve presented, while based on the official available data, has a high likelihood of being in accurate in terms of total infected, total dead and any ongoing cases. What is correct is China’s ability to enforce quarantine even if involves welding people into their apartment buildings. The China figures are about as accurate as the BOM adjusted Darwin ones.
Having been on one in Dec there is also some skewing for cruise ships. Every time you enter a food area the staff have you use an anti-bacterial hand wash. Internal railings are always sticky with cleaning and disinfecting solutions so the likelihood of infection from surfaces is reduced. None of that stopped me from getting a bad cold for the last few days however.
As an Australian I fear we are just at the beginning of our exponential growth curve because our government has not closed the borders and allowed tens of thousands of people from early affected areas to cross the borders, without tests and without any even recommended isolations. They are still allowing children to go to school while simultaneously telling people not to congregate in any numbers. We are running out of test kits and our health services have very little ability to cope with potential numbers.
Aa aside that I’ve been a keen Willis-ite for some time and enjoy your data driven position.
Did you consider the current thesis that there are two strains of Covid-19 and that the Cruise Ship version could have been the milder one?
Yes we tend to become complacent about constant background risks and then get spun up about something new which is by the numbers anyway less dangerous. There’s also the perception that we’re at least somewhat in control of our traffic risks, while for COVID-19 people are demanding that somebody else “do something”.
Having this in a Presidential election year probably skews perceptions even further away from reality.
The numbers from the Diamond Princess are indeed not what I would have expected from all the other reporting.
From perusing the daily WHO reports, it seems South Korea has had the most effective response. Italy alas has not, and Spain is not too far behind.
https://www.dailymail.co.uk/news/article-7978695/Journey-cruise-ship-passenger-infected-coronavirus.html My initial thought was behavior and outings/ports of call may explain some of the data (when did he get it, when was he shedding, and when did he leave the ship). According to this article (anyone have anything that contradicts?). Patient zero was only on the ship five days and reported symptoms on the 6th day after he left (1/26). He did go on a bus ride with others. He may have gone back to his room feeling a little off, thus limiting exposure. He may not have infected people until about the last day of his trip. Incubation time unknown. Once people were confined to their rooms, the outbreak would have been limited due to quarantine.
Interesting discussion!
I also found that authorities determined patient zero had only been in contact with under 300 people and the first 61 infected were removed to hospital immediately. If exposure to infected people were limited, people were quarantined to rooms the infected had not accessed, and no infectious material got into ventilation, I can imagine this was not the perfect petrie dish but a dodged bullet. Some of this is really about assumptions and unknowns…
facts are a bit scarce at a time like this.
we know when it started and the populations of countries. we know the deaths,
the rest is modelling or sparse testing data.
According to the best guesses, the pinch point will be icu beds
According to the best guesses the people in the beds will be over 70 with previous problems
by now, we should know one thing. are the icu beds being filled with over 70’s covid patients ??
that is where i would be looking
Diamond Princess Cruise Ship
Data source: https://www.medrxiv.org/content/10.1101/2020.03.05.20031773v2.full.pdf
Patient 1 was on board from 20-25Jan2020
By 20Feb2020 there were 634 cases, or which 328 were asymptomatic (3063 tested) out of 3663 passengers and crew.
Only 17% of the people were infected, and only half of those (8.4% showed symptoms.
There were 1231 people over age 70.
By 1Mar2020 there were 7 deaths, all over age 70 (0.6% mortality of those over 70)
Assume this was a fully uncontained situation, such that infection and death rates will be much less with good containment.
DIAMOND PRINCESS CRUISE SHIP
Data source: https://www.medrxiv.org/content/10.1101/2020.03.05.20031773v2.full.pdf
ABSTRACT
Adjusting for delay from confirmation-to-death, we estimated case and infection fatality ratios
(CFR, IFR) for COVID-19 on the Diamond Princess ship as 2.3% (0.75%–5.3%) and 1.2%
(0.38–2.7%). Comparing deaths onboard with expected deaths based on naive CFR
estimates using China data, we estimate IFR and CFR in China to be 0.5% (95% CI: 0.2–
1.2%) and 1.1% (95% CI: 0.3–2.4%) respectively.
______________________________________
If I understand this paper correctly, projections from the ship’s total population averaged :
CFR – Case Fatality Ratio (those who died / those who showed symptoms) ~= 2.3%
IFR – Infection Fatality Ratio (those who died / those who were infected, including those who showed no symptoms) ~= 1.2%
The ship was placed in quarantine on or about 3Feb2020 but it appears that containment was not effective, since by 20Feb2020 there were ~634 infections of which ~half were asymptomatic.
“The quarantine hasn’t worked, and this ship has now become a source of infection,” said Dr. Nathalie MacDermott, an outbreak expert at King’s College London. “There’s no reason this should not have worked if it had been done properly.”
Deaths commenced 20Feb2020 and by 1March2020 totaled 7, 2.4% of the 288 total infected over age 70, and 5.6% of the 124 total who showed symptoms over age 70.
Total projected deaths for over-70’s who were on this ship is projected at 12 (5 more) with another 3 deaths projected for ages 60-69.
Need coffee…
Could be you’re thinking too much. You can say there are broadly 2 types of quarantines in play right now. The draconian Russian type that results in very few infections or deaths and the other type in play in most places that is soft voluntary quarantines that results in much higher infection and death rates.
The Princess ship instituted russian style quarantines and they did it quickly… which is the secondary key to success.
By comparison, Italy instituted draconian quarantines, but far too late.
My completely unscientific 2 cents. Cheers
According to WHO 712 caught the disease and 7 died. Roughly 1%
https://who.maps.arcgis.com/apps/opsdashboard/index.html#/c88e37cfc43b4ed3baf977d77e4a0667
First off, those who died and those who didn’t had some of the best care available. Quite a few still haven’t recovered and some are still critical. Some may have been connected to ECMO machines which completely take over most of the main functions of the body. The deaths would be far higher if the health services of the World were full. Which they will be very soon. Even basics like oxygen supply might struggle to keep up.
Next the assumption that those who didn’t catch the disease were never going to. The ship went into lock down ie, the kind of ‘panic’ that you’re concerned about. This wasn’t normal cruise ship precautions. Thus the spread of the virus slowed to a trickle but, because of the confinment of the ship, didn’t stop until they allowed people to leave. With no efforts to limit spread, it’s expected that 60-80% of populations would catch it (assuming no mutations and the more people who catch it, the more chances of that). In the UK alone the lower figure would be about half a million. Equal to an entire year of deaths from all causes. Ok, natural changes in behaviour might reduce that to 250,000.
Those deaths might not be statistically scary but in reality they’re terrifying. OK, the assumptions might be wrong, maybe high numbers don’t catch it but it’s too dangerous to test the theory.
How much intervention is necessary to stop the spread is a guess. The UK is reacting to the failure of early interventions, to stop rapid rise in cases. If only to build more ventilators, work out if antivirals and anti malaria pills help and get a bit closer to a vaccine. This isn’t panic, this it catchup.
Robert Turner, referring to this as a mere cold, or even the flu, runs completely ignorant of actual medicine. As for the car analogy, we utilize safety measures every single day in our vehicles, therefore most of us end up okay in the end. However, this doesn’t negate the number of deaths every year. The same concept is true with this virus, but which we currently possess no safety measures against it aside from avoiding it in the first place. This is why we need to practice some serious self-responsibility, and help out our neighbors when necessary. If we go about this the right way, then a normal summertime could still be a possibility for us here.
From a medical professional with whom I shared this article, he took issue with the “good news” of a 2% infection rate saying:
1: The virus has been spreading for months unchecked in US. Not weeks, which is the incubation period.
2: Ten times the people dying is the best possible scenario. If our efforts were to fail completely, 2% of 300 million is 6 million. This is the exact wrong time to diminish the severity.
“he took issue with the “good news” of a 2% infection rate saying:”
Pretty sure that’s not what was said. What was actually written was:
“For me, this is all good news. 83% of the people on the ship didn’t get it, despite perfect conditions for transmission. If you get it, you have about a 50/50 chance of showing no symptoms at all. And the fatality rate is lower than the earlier estimates of 2% or above.”
Meaning, this showed LESS fatality rate than the 2% initial estimates, which IS good news (assuming it’s not outside the range). Not to mention the fact that 83% did not get infected and of half of those infected were not effected (other than the real concern of being unknowing carriers).
Also, the medical professional’s math is off. Assuming the numbers from this study could be generalized as a worse case scenario of near universal exposure:
17% infection rate of 300 million is 51 million people infect.
With a fatality rate at 1.2% (range of 0.38%–2.7%) would be 612,000 deaths (193,000 to 1,337,000 taking both ends of the range).
Willis this is probably to far down the post for you to see. However, thought it worth a post
Without going into detail the way the specific immune system works might explain the apparent low rate of infection. the mighty macrophages break down foreign invading proteins into short lengths of the peptide chain of which the foreign antigen is composed. These shorter bits are presented to t-cells and trigger both antibody and killer t-cell responses. Thus a foreign antigen is broken down into bits and it these bits that produce the immune response. Because the corona virus is one of a family of similar viruses cross reactivity of the immune system to several similar viruses is possible. So it is possible that many of the passengers on the cruise ship may have been infected with similar, but not as pathogen viruses in the past and have sufficient immune cross reactivity to provide protection. That would explain why some older passengers do not show symptoms. The only way to examine this would be to check if the individual has been infected, nasal swabs and also check the patient for virus antibodies. Of course, that hasn’t been done so this is just speculation, but could have profound implications for assessing the amount of infection in the population as a whole
Tony, I follow my posts to the bitter end … and yes, cross-reactivity of our immune system from one corona-virus to another may indeed be in play.
w.
Does anyone know what was put on the death certificates of those that died that had tested positive for this corona virus?
Dang … I made it onto Instapundit here. First time ever.
WUWT ROOLZ!
w.