A Diamond in the Sea.
In a blog post yesterday, I laid out the case that my good friend and co-author Charley Hooper made for the conclusion that the deaths from the coronavirus-induced Covid-19 disease will be comparable to the deaths from flu. One of Charley’s big pieces of evidence was the case of the Diamond Princess.
Another friend, Phil Magness, contacted me to give his input about why that was Exhibit A for Charley’s case. I am reporting this with Phil’s permission. Here’s Phil:
Charlie’s comment on the Diamond Princess is one of the most convincing I’ve seen to date.
I replied: Ditto me. I’m really excited that I might lose the bet.
Phil:
In addition to the relatively low infection rate despite being several weeks in close quarters, it’s basically the only case where we have a true controlled experiment where 100% of passengers [and he obviously means “and crew”] were tested and we know exactly who got it. It has everything all the other stats are lacking: a reliable numerator and a complete 100% accounted for denominator, plus known locations of all passengers.
READER COMMENTS
Karen Selick
Mar 17 2020 at 6:18pm
When they tested everyone, did they ask these questions:
1. What was their vitamin D status? Deficient or adequate?
2. Where had they been wearing their Princess medallions, around their necks or somewhere else?
3. What does their gut flora look like?
These would all be important clues as to who might be vulnerable, but I’ll bet they didn’t ask any of these questions.
KevinDC
Mar 17 2020 at 7:03pm
This is good and hopeful news. My knee jerk reaction is to wonder why these numbers are so much lower than what’s commonly being quoted. I’ve seen epidemiologists predicting anywhere from 25% to 60% of people in America likely to get coronavirus in the next year. Are the ones making the most alarming predictions just more likely to end up quoted in the media? Are they unaware of the numbers from the Diamond Princess? Or are they aware of these numbers, but have additional information or experience to use in their projections that a regular Joe like me just doesn’t know about, which justifies the higher assessment?
Charley Hooper
Mar 18 2020 at 4:03pm
Good questions. I suspect they are building complex mathematical models based on incomplete information and ignoring case studies.
Oleg
Mar 17 2020 at 7:17pm
About 700 infections out of about 3,500 passengers and crew. That’s an infection rate of 20%. Of the infected, 1% died.
There are 8 billion people in the world. That’s 1.6 billion infections, and 16 million dead.
The seasonal flu kills around 600,000 a year. That makes 15+ million deaths in excess of the seasonal flu in the world this year from a novel infectious agent.
What are we celebrating exactly?
Mark Bahner
Mar 17 2020 at 9:50pm
No, that would only be if we put all of the people in the world on a cruise ship (with its incredibly cramped quarters, meals and most activities in common areas, and limited medical assistance), and if all the people in the world were old geezers like the people on the Diamond Princess. (I can safely classify them as “old geezers” because the average age of the passengers was probably about my age or even slightly younger.)
There is simply no way that 16 million people are going to die–in the next year or two, anyway–from COVID-19.
You mention that “around 600,000” people per year die of flu. I’d be happy to bet you $40–and give you 2-to-1 odds–that the Johns Hopkins website will not record 600,000 deaths in 2020 from COVID-19:
https://coronavirus.jhu.edu/map.html
Oleg
Mar 18 2020 at 11:45am
You can have 10-to-1, if you’d like.
If it’s 600,000 or under, then COVID-19 was (or, rather, will have been) just a bad flu season. One wonders how it ever became a news item if that’s the case.
Mark Bahner
Mar 19 2020 at 9:14pm
Hi Oleg,
Sorry I never responded to you. You wrote:
Actually, I’d prefer 2-to-1. 🙂 What I meant by giving you 2-to-1 odds, I meant that if the Johns Hopkins coronavirus tracking website showed total worldwide deaths in 2020 from COVID-19 of more than 600,000, I’d give you $80, but if the number was less than 600,000, you would only have to give me $40. I’d rather not do 10-to-1, where I’d have to give you $400 if the total worldwide deaths were higher than 600,000. The reason I’d rather not do 10-to-1 odds is that I don’t have much knowledge or confidence about what could happen in developing countries like those in Africa. I just don’t know enough about their medical systems or societal adaptability.
So, that is my bet offer…over 600,000 deaths from COVID-19 worldwide by the end of this year, I’ll give you $80, if it’s under 600,000, you give me $40. (I probably was a little rash in making that offer, the more I think/worry about what might happen in developing countries.)
<blockquote>If it’s 600,000 or under, then COVID-19 was (or, rather, will have been) just a bad flu season. One wonders how it ever became a news item if that’s the case.</blockquote>
Yes, I was planning sometime to write a post on my blog about the U.S., titled, “Are we insane?” The post was going to deal with the U.S. response to COVID-19, and whether it is out of all proportion to the danger. (A bit of a spoiler alert…my answer will almost certainly be, “Yes, we are insane.” Our response has been totally out of proportion to the threat.) But I want to wait a few more weeks to make sure I won’t look bad if I write that our response has been out of proportion to the danger.
Best wishes,
Mark
Catherine Barrett
Mar 26 2020 at 3:51am
So, cruise ships are perfect laboratories in which to observe a real time epidemic, except when this produces number you don’t much like the look of, at which point you get to step in with all the reasons why they are *not* perfect laboratories.
This is a thin and rather dishonest argument.
It also ignores that the mortality rate is only part of the problem – overwheming ICUs with severe respiratory distress cases has enormous impact on treatment of other conditions. I find US citizens have trouble conceptualizing the wider impact on the health system, and I suspect this may be cultural.
In any case, this is now entirely out of the hands of armchair pundits and will run its course regardless.
Matthias Görgens
Mar 17 2020 at 11:29pm
Cruise ship demographics tend older (and richer) than the general global public.
We know that older people are hit harder by the virus. So take that into account when trying to extrapolate.
Of course, the people on the cruise ship also got better care than most people around the globe can expect to get.
Oleg
Mar 18 2020 at 9:59am
Yes, they’re richer, which means they all received excellent care. It’s early in the pandemic, so no one was denied a hospital bed. The ship had approximately one crew member for every 3 passengers. None of them died, of course, but this was a very small sample. Also, as you can imagine, cruise goers tend to be older, but they also are not infirm, and usually not seriously ill in other ways (they’re on vacation, after all).
Alright, let’s say out in the wild the death rate will be half what it was on the Diamond Princess. That’s 7 million. Seasonal flu kills 600,000 globally at the high end (I think the figures are 200,000 to 600,000, depending on severity).
Seems just a little bit early for high-fives to me.
Jens
Mar 18 2020 at 5:04am
There is at least some immunity and vaccination against the seasonal flu in the population. But attention for COVID-19 is much larger than for the seasonal flu. Both effects may neutralize more or less. Wicked thing about COVID-19 seems to be, that incubation period is quite long and hosts are infectious without showing symptoms for quite a while. But incentives to do smth is quite large in all sectors.
JFA
Mar 18 2020 at 8:07am
How much solace should we be getting from the Diamond Princess? The timeline for the number of cases is a bit worrying: https://www.pharmaceutical-technology.com/news/coronavirus-a-timeline-of-how-the-deadly-outbreak-evolved/. After 5 days into the cruise, the cruise ship knew that a passenger had had the virus. I don’t know what kind of measures were taken at that point, but there seem to have been some restrictions on movement that lowered the infection rate substantially: https://academic.oup.com/jtm/advance-article/doi/10.1093/jtm/taaa030/5766334 (they estimate it could have been 80% without quarantine measures). So there was a week of free movement, then movement was restricted, and you still had a 17% infection rate. It seems like all this optimism is coming after restrictions on gatherings and movement were put in place. I can’t find it right now, but there was one chart that showed the number of confirmed cases (measured by onset of symptoms rather than date of diagnosis) in Wuhan. The decrease in cases only came after restrictions on movement were put into place. Given experience in other regions (Iran, China, Italy, though there is lots of variation in other regions) we might have expected the death rate on the Diamond Princess to be much higher than it was (due to the age profile of the passengers). One hypothesis could be that those who tested positive from the cruise were whisked away to uncrowded hospitals and given necessary care.
One comment on Levitt’s optimism taken from exponential growth models: more sophisticated contagion models show that network structure drives the rapidity of disease spread. So while I may not be meeting new people everyday, there is almost certainly somebody in my network who is meeting someone new everyday. “Small world” networks (e.g. six degrees of separation) can lead to the quick spread of disease. At this point, pontificating (I’m speaking of Levitt) on the spread of disease without a model of how movement restrictions, the emerging population of immune people (due to previous exposure), etc., is, at the very least, premature.
I wish I could put more time into this, but my wife and I are currently working at home with two kids. I really appreciate the civilized/non-emotional discussion (for the most part) of this sensitive topic and this site. Keep up the good work.
Alan Goldhammer
Mar 18 2020 at 9:51am
Extrapolation of small group data to the general population is often problematic. We know that the elderly with other health conditions are most at risk so it would be useful to know the patterns of those who were infected and those whose illness progressed.
Yes, the mortality and infection rate appear to be lower than observed in other settings. Since aerosol transmission still appears to be the principal means of infection, it would be good to know whether those infected regularly sat with one another at meals or were close during activities.
There is a lot that we don’t know and most of my epidemiologist friends are not sanguine about small group data such as that being discussed here. There are a variety of circulating models about what might happen in the general population and I don’t think any of us know which one is correct.
One possible good note is favipiravir, an anti-viral that is easy to make at scale and can be administered orally is showing effectiveness in a Chinese trial.
Alan Goldhammer
Mar 18 2020 at 10:28am
I meant to add that I do a daily check of the WHO COVID-19 reports. These are interesting but as with any data needs to be looked at with some skepticism. They publish the breakdown for all countries. The data for EU countries is ‘probably’ reliable. Most have good public health surveillance, all of them with some form of national health care, and most with electronic medical records systems. The two countries that stand out as extreme contrasts (as of today; things can obviously change) are Germany and Italy.
We have all read about the situation in the Lombardy region of Italy and country wide there 27980 confirmed cases with 2503 deaths for a mortality rate of almost 9%. However, Germany is the exact opposite; 6012 cases and 13 deaths for a mortality rate of 0.2%, twice that of the 1957 and 1968 pandemics but much lower than the 1918 event. From my quick look, Germany has the lowest death rate of any Euro country. We will see if this continues.
Aleksander
Mar 18 2020 at 9:01pm
The Italian and German case numbers tell us nothing about their respective mortality rates, since both are obviously far from the real infection numbers. Germany may have identified a larger share of their infections than most countries, but Italy hasn’t identified more than one tenth (or even a hundredth!) of theirs.
There are some reasons to believe that mortality rate is higher in Italy than Germany, but the current death numbers isn’t among them. “Confirmed cases” numbers are almost never useful without extensive analysis
Aleksander
Mar 18 2020 at 9:33pm
I’m confused about what this claim means. Do 500 Italians die from flu every single day? Are they put on respirators to help them breath for several days? Are patients that lack a respirator left to fend for themselves, eventually dying unaided in a hospital bed because their lungs stopped working? Are there regular reports of fit and heathy people with the flu who end up in the ICU with breathing problems, eventually being put into an induced coma to avoid his constant coughing killing him outright, finally “recover” with a permanently damaged lung? Does the flu consistently kill off every fifth 80+ year close relative you have, every year?
Tyler Wells
Mar 19 2020 at 7:10am
Influenza is an accepted part of life and death, the novel coronavirus new and scary. We are about to pass 9,000 deaths from coronavirus. We are at 104,000 deaths from seasonal flu. For sure both will grow. The point is, people die from the flu all the time and life goes on, work gets done, kids get educated.
I cannot attest to any of your other numbers, as I don’t find them. Mine are below.
https://www.worldometers.info/coronavirus/
Mark
Mar 19 2020 at 6:48pm
Your link confirms the daily death rate of nearly 500 people in Italy as of now. By contrast, I believe there are just a few hundred deaths *per year* in Italy due to the flu. Surely you agree that Aleksander has a point here.
Tyler Wells
Mar 20 2020 at 7:17pm
I tried to do the math but failed, as influenza in the United States fluctuates highly depending on the season. Numbers peak around February and fall precipitously in the Spring, only to rebound in the fall. So you are most likely correct. We had 63,000 flu deaths last year in the US with a population 5 times that of Italy. So that is about 5,200 Italian flu deaths a year. They are already at 4,00 COVID 19 deaths, and so are very likely to have more deaths from COVID 19 than from influenza.
On the other hand, roughly 1,500 Italians die every day from various causes. We don’t know enough about COVID 19 to know how significant it will be in the death toll of Italy.
Really, all of this is irrelevant. No one is saying that no precautions should be taken, far from it. The question is whether the current attempts at suppression will, first, be successful, and second, considering the massive societal unrest, huge unemployment, and resulting societal problems (born primarily by those least affected by the virus) will not be worse than the virus itself.
Mm
Mar 21 2020 at 11:10am
The problem with the cruise ship data is apparently many passengers refused to be tested- so many more may have had a mild case of the disease. So it maybe even less of a threat than the 1% death rate. But that makes Italy’s numbers even more puzzling- the death rate there is horrific-even their demographics make it hard to understand why the death rare is so high. They must not be testing people with mild symptoms &/or their hospitals are totally overwhelmed & care must be severely rationed or substandard to explain the high death rate-or every old person in Lombardy got infected. If the Italian experience recurs in other Western countries we are in for a rough year.
Michael Sandifer
Mar 21 2020 at 12:46pm
There could easily be important selection bias going on with data from a cruise ship.
Bill Johnston
Mar 25 2020 at 8:41am
This discussion needs to incorporate the evolving details of the Diamond Princess infections. According to data from Johns Hopkins, as of March 25, there are 10 deaths among 715 infected passengers. There are 587 classified as recovered. This leaves 128 whose outcome is still undetermined. If some of them are still hospitalized, the final death rate may be considerably higher. Like all the other analysis and speculation associated with this outbreak, it is premature to construct towers of extrapolation from preliminary data.
Comments are closed.