Is the death rate from the Diamond Princess a reasonable worst-case estimate for the coronavirus death rate in the U.S.?

Apologies for the macabre subject, but with everyone freaking out about the coronavirus, I’m wondering if it makes sense to step back and ask why the Diamond Princess wasn’t a worst-case scenario. The ship held 3,700 people. The virus spread all around the ship before anyone knew what was going on. Then everyone on board was kept on board, all breathing from the same ventilation system, eating food from the same kitchen (almost surely prepared by at least some workers who had the virus, but didn’t know it). Out of 3,700 passengers and crew, 6 have died (Business Insider).

That’s a death rate of 1/6th of 1 percent (0.16 percent), and concentrated among people whose immune systems were weakened due to other factors (i.e., people who might have died a year later from the flu).

The U.S. overall is not more crowded than a cruise ship. Why should we expect more than 0.16 percent of Americans to die when this is all over? That’s unfortunate, of course, and a huge number when multiplied by 330 million: 535,000. But it is not an economy-ending or country-ending number. And, since our country is not in fact as densely populated as a cruise ship, the real number might be far less than this upper bound. It might be closer to the 80,000 who died from the flu in 2017-2018 (source). And there might not be that many additional deaths because the same people who get killed by the flu are also susceptible to COVID-19.

One factor that could explain how the death rate could be higher: as the disease spreads, hospitals and other health care resources will be spread thin. But, on the other hand, knowledge about how to treat the infection will improve. If these two factors cancel out, we’re back to the Diamond Princess being the worst-case scenario. Finally, consider that the cruise ship demographic is older and more fragile than the general population.


17 thoughts on “Is the death rate from the Diamond Princess a reasonable worst-case estimate for the coronavirus death rate in the U.S.?

  1. Philip said:
    “One factor that could explain how the death rate could be higher: as the disease spreads, hospitals and other health care resources will be spread thin.”

    Consistent with that, a friend in Rome reported today:
    “So far so good here. I think they are taking strict precautions to delay too many cases at the same time which with the elderly population could imply not enough hospital beds with respiratory equipment in case needed and infrastructure collapse. Schools are closed, parents are going crazy, we are all saying hello with our elbows and feet, movie theatre seats have white ribbons indicating where you can sit at a distance from the other, restaurants are less crowded, people are overall nicer.”

  2. The ship still has 445 active cases, with 32 critical. Too early to compute its CFR.

    • This is the correct answer.

      South Korea is showing a ~1% CFR, which is probably more representative of outcomes in a developed country with sufficient medical care capacity. Outside of Hubei China looks similar. Inside Hubei the surge of cases resulted in a lot of otherwise survivable cases dying for lack of care.

      In typical American fashion we seem to be divided between “this is a hoax” and hysterical panic buying of toilet paper and bleach.

  3. Attribution is important. Otherwise, we can easily report 0 cases, just like in North Korea. Some people could get sick or even die after the disembarkation, but who really cares if an old geezer dies (OK boomer? why so quiet?) and what does it all have to do with the virus?

    Furthermore, all those percentage points may be off as the epidemic is still evolving so they may not add up to 100%. And I would not trust the major news network with the numbers as their well-being depends on NOT understanding math.

    >> “The Strongest Evidence Yet That America Is Botching Coronavirus Testing” (Atlantic):
    >> “I don’t know what went wrong,” a former CDC chief told The Atlantic.
    What testing? The US doesn’t have a sufficient number of test kits and no means to produce them in the short-to-medium term. (What are we? Taiwan?) That seems to be the real reason of why the CDC restricts testing or even wearing face masks.

  4. This is the first widely transmitted pandemic of a disease with zero preexisting immunity in the modern era. The numbers you see today are based on very thin testing and reflect transmissions that occurred two weeks ago. It may or may not be as bad as the 1918 flu, but that’s a reasonable ballpark. We can thank our lucky stars that this one isn’t as severe as SARS or MERS (case fatality rates in the 5-30% range).

    It’s not a huge health risk for healthy, mid-age individuals, but we all have friends and family who are at higher personal risk. At a societal level, our health care infrastructure has no capacity for the number of severe cases or the length and intensity of care they will require. Substantial disruption is justified over the next few months to slow transmission through the world population to try to keep health care systems functional.

    • This one is more dangerous because of the lower fatality rate. Dead people don’t travel!

    • It’s the impact on the healthcare system that most worries me. If my local hospitals are full of mechanically-ventilated CoVID-19 patients, and the ER is full– both of which are well within the realm of possibility given Farr’s Law– people who have *different* kinds of medical emergency are at higher risk.

  5. A Ph.D. biologist estimates we will run out of hospital beds by the end of May at the latest if the spread isn’t contained, the growth rate is exponential (doubling time 6 days) and 2.5% of the patients require hospitalization. We will run out of masks, hospital beds, and eventually health care workers, at least according to her projections.

  6. Media coverage is almost non existent, compared to the HIV coverage we had 40 years ago. The reported statistics are next to worthless. Death rates are definitely higher than reported. Areas with jobs are about as crowded as a cruise ship. They’re hiding most information out of concern about causing a racism scandal.

    We do know the chances of dying from the virus are better than the chances of winning a lottery or surviving a marriage.

  7. I’m thinking the opposite idea, cruise ship might have better odds due to a lack of intensive medical intervention. Would anyone like the odds of surviving as the first Corona patients Mass General gets to play with? And if you don’t survive the care, the virus still gets credit for the kill.

  8. > the 80,000 who died from the flu in 2017-2018

    Don’t Believe Everything You Read About Flu Deaths. The CDC uses the term “influenza-associated death” (in 2010 – 36,000) vs actual “cause-of-death” (less than 500)


    • The CDC says this:

      “CDC estimates that influenza has resulted in between 9 million – 45 million illnesses, between 140,000 – 810,000 hospitalizations and between 12,000 – 61,000 deaths annually since 2010.”


      Dr. K, can you please point me to a CDC link that discusses what you are claiming?

  9. Seems to me that there are at least three cruise ship health disasters every year – legionnaire’s disease, novovirus, gastrointestinal illness, etc. Nothing makes this one any scarier.

    The really sad thing is that no one will go back and do the “root cause analysis” of the bad statistics, bad policy and bad information and improve our systems.

    Our only hope is that we don’t have a real global problem.

  10. Amesh Adalja of Johns Hopkins is putting the best guess upper bound on worst-case death rate at 0.6%. He said this just now on the Sam Harris podcast. He bases this on the South Korea numbers as South Korea, at this point, has the most aggressive testing. As I write this, S. Korea’s numerator is 54 deaths and denominator is 7755 for a rate of 0.7% deaths. Note that this might be high as those seeking out testing might tend toward the symptomatic.

    • Actually, Italy reports a somewhat higher mortality rate: closer to 5%; but what’s a 4.3% difference between friends? it’s only 10.5 million peasants, give or take; and besides, it’s mostly old geezers. (I didn’t realize some folks would hate their parents so much.)

      Even at 0.7%, the death toll would top at 172K at the saturation point of, say, 75% of the total population (when the virus runs out of hosts). However, unlike flu it’s not going to be equally distributed because a median estimated time from infection to death is less than 3 weeks, so there are going to be death clusters, and 5000 deaths in your county matters more than 100K far away.. But I guess not everyone likes big cities and their inhabitants.

      Here is another point to consider. The epidemic seems to be spreading exponentially for now, and it will temper out eventually. But before that point, the mortality rate is not a particularly interesting number: the rate of infection is, since any variation of fatalities can be erased in just a matter of days (which is a nature of exponential growth).

  11. The deaths are up to 11 now and still have 82 cases that are unresolved. 10 are still in critical condition. Seeing that most of these people received medical care and died on dry land, then I don’t think we can give too much emphasis to the medical staff on the ship. Also death rates are calculated from infected people, which is 619. It doesn’t go by the entire ship population. So the death rate as of now is 1.5%, not .16%. And it may go up higher if any of those people in critical condition pass away.

Comments are closed.