The Corona Princess: Learning from a petri dish cruise (i)


Q. Was it a mistake to quarantine the passengers aboard the Diamond Princess in Japan?

A. The original statement, which is not unreasonable, was that the best thing to do with these people was to keep them safely quarantined in an infection-control manner on the ship. As it turned out, that was very ineffective in preventing spread on the ship. So the quarantine process failed. I mean, I’d like to sugarcoat it and try to be diplomatic about it, but it failed. I mean, there were people getting infected on that ship. So something went awry in the process of quarantining on that ship. I don’t know what it was, but a lot of people got infected on that ship. (Dr. A Fauci, Feb 17, 2020)

This is part of an interview of Dr. Anthony Fauci, the coronavirus point person we’ve been seeing so much of lately. Fauci has been the director of the National Institute of Allergy and Infectious Diseases since all the way back to 1984! You might find his surprise surprising. Even before getting our recent cram course on coronavirus transmission, tales of cruises being hit with viral outbreaks are familiar enough. The horror stories from passengers on the floating petri dish were well known by this Feb 17 interview. Even if everything had gone as planned, the quarantine was really only for the (approximately 3700) passengers because the 1000 or so crew members still had to run the ship, as well as cook and deliver food to the passenger’s cabins. Moreover, the ventilation systems on cruise ships can’t filter out particles smaller than 5000 or 1000 nanometers.[1]

“If the coronavirus is about the same size as SARS [severe acute respiratory syndrome], which is 120 nanometers in diameter, then the air conditioning system would be carrying the virus to every cabin,” according to Purdue researcher, Qingyan Chen, who specializes in how air particles spread in different passenger crafts. (His estimate was correct: the coronavirus is 120 nanometers.) Halfway through the quarantine, after passenger complaints, they began circulating only fresh air–which would have been preferable from the start. By then, however, it is too late: the ventilation system is already likely filled with the virus, says Chen.[2] Arthur Caplan, the bioethicist who is famous for issuing rulings on such matters, declares that

“Boats are notorious places for being incubators for viruses. It’s only morally justified to keep people on the boat if there are no other options.”

Admittedly, it is hard to see an alternative option to accommodate so many passengers for a 2 week quarantine on land, and there was the possible danger of any infections spreading to the local population in Japan. So, by his assessment, it may be considered morally justified.

The upshot: As of 19 March 2020, at least 712 out of the 3,711 passengers and crew had tested positive for covid-19; 9 of those who were on board have died from the disease (all over the age of 70). As I was writing this, I noted a new CDC report on the Diamond Princess as well as other cruise ships; they state 9 deaths.[3] A table on the distribution of ages of passengers on the Diamond Princess is in Note [4].

So how did the Diamond Princess cruise ship become a floating petri dish for the coronavirus from Feb 4-Feb 20?

The Quarantine

It was their last night of a 2-week luxury cruise aboard the Diamond Princess in Japan (Feb 3) when the captain came on the intercom. He announced: a passenger on this ship who disembarked in Hong Kong 9 days ago (Jan 25) has tested positive for the coronovirus. (He was on board for 5 days.) Everyone will have to stay on board an extra day to be examined by the Japanese health authorities. A new slate of activities was arranged to occupy passengers during the day of health screening–later mostly dropped. But on the evening of February 3, things continued on the ship more or less as before the intercom message.

“The response aboard the Diamond Princess reflected concern, but not a major one. The buffets remained open as usual. Onboard celebrations, opera performances and goodbye parties continued”. (NYT, March 8)

The next day, as health officials went door to door to screen passengers, guests still circulated on board, lined up for buffets, and used communal spaces. But then, the following morning (Feb 5), as guests were heading to breakfast, the captain came over the intercom again. He announced that 10 people had tested positive for the coronavirus and would be taken off the ship. Everyone else would now have to be quarantined in their cabins for 14 days. The second day of the quarantine (Feb 6) it was announced that 20 people more had tested positive, then on day three, 41 more, then 64 more, and on and on. By the end of the quarantine on February 19 at least 621 on the ship had tested positive for the virus.

Adding to the stress, “we quickly learned that our tests were part of an initial batch of 273 samples and that the first 10 cases reported on day one were only from the first 31 samples that had been processed” from the passengers with highest risk. (U.S. passenger, Spencer Fehrenbacher, interviewed on the ship)

As the number of infected ballooned, passengers were not always informed right away; some took to counting ambulances lined up outside to find out how many new cases would be announced at some point. I wonder if the passengers were told that the very first person to test positive was a crew member responsible for preparing food. In fact, by February 9, around 20 of the crew members tested positive, 15 of which were workers preparing food. Crew members lived in close quarters, shared rooms and continued to eat their meals together buffet-style. They had no choice but to keep running the ship as best as they could.

“Feverish passengers were left in their rooms for days without being tested for the virus. Health officials and even some medical professionals worked on board without full protective gear. [Several got infected.] Sick crew members slept in cabins with roommates who continued their duties across the ship, undercutting the quarantine”. (NYT Feb 22)

Passengers in cabins without windows (and later, others) were allowed to walk on deck, six feet apart, for a short time daily. Unfortunately, presumed infection-free “green zones” were not rigidly separated from potentially contaminated “red zones”, and people walked back and forth between them. Gay Courter, a writer from the U.S. who, as it happens, situated one of her murder mysteries on a cruise ship, told Time “It feels like I’m in a bad movie. I tell myself, ‘Wake up, wake up, this isn’t really happening.’” (Time, Feb 11). This is the same bad movie we are all in now, except our horror tale has gotten much worse than on Feb 10.

At some point, I think Feb 10, the ship became the largest concentration of Covid-19 cases outside China, which is why you’ll notice the Diamond Princess has own category in the data compiled by the World Health Organization (Worldometer).

In a Science Today article, a Japanese infectious disease specialist regretted the patchwork way in which passenger testing was done:

Japan has missed a chance to answer important epidemiological questions about the new virus and the illness it causes. For instance, a rigorous investigation that tested all passengers at the start of the quarantine and followed them through to the end could have provided information on when infections occurred and answered questions about transmission, the course of the illness, and the behavior of the virus.

(They were only able to test people in stages.) A similar paucity of testing in the U.S. robs us from crucial information for understanding and controlling the coronavirus. However, there is a fair amount being gleaned from the Diamond Princess, as you can see in the references below. (Please share additional references in the comments.) More is bound to follow.

Estimates from the Diamond Princess

“Data from the Diamond Princess cruise ship outbreak provides a unique snapshot of the true mortality and symptomatology of the disease, given that everyone on board was tested, regardless of symptoms”–or at least virtually all. [link] The estimates (from the Diamond Princess) I’ve seen are based on those from the London School of Hygiene and Tropical Medicine, in a paper still in preprint form,”Estimating the infection and case fatality ratio for COVID-19 using age-adjusted data from the outbreak on the Diamond Princess cruise ship”.

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%) [among symptomatic] and 1.2% (0.38-2.7%) [all cases]. 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. (PDF)

(For definitions and computations, see the article.) These are lower than the numbers we are often hearing. They used their lower fatality estimates to adjust (down) the estimates from China data. The paper lists a number of caveats.[5] I hope readers will have a look at it (it’s just a few pages) and share their thoughts in the comments. (Their estimates are in sync with an article by Fauci et al., to come out this week in NEJM; but whatever the numbers turn out to be, we know our healthcare system, in many places, is being overloaded. [6])

Another study takes the daily reports of infections on the Diamond Princes to attempt to evaluate the impact of the quarantine, as imperfect as it was, in comparison to a counterfactual situation where nothing was done, including not removing infected people from the ship. They estimate nearly 80%, rather than 17% would have been infected. [link]

We found that the reproductive number [R0] of COVID-19 in the cruise ship situation of 3,700 persons confined to a limited space was around 4 times higher than in the epicenter in Wuhan, where was estimated to have a mean of 3.7.[7]

The interventions that included the removal of all persons with confirmed COVID-19 disease combined with the quarantine of all passengers substantially reduced the anticipated number of new COVID-19 cases compared to a scenario without any interventions (17% attack rate with intervention versus 79% without intervention) … However, the main conclusion from our modelling is that evacuating all passengers and crew early on in the outbreak would have prevented many more passengers and crew members from getting infected.” [link]

Only 76, rather than 621 would have been infected, they estimate. [8]

Conclusions: The cruise ship conditions clearly amplified an already highly transmissible disease. The public health measures prevented more than 2000 additional cases compared to no interventions. However, evacuating all passengers and crew early on in the outbreak would have prevented many more passengers and crew from infection.

These studies and models are of interest, although I’m in no position to evaluate them. Please share your thoughts and information, and point out any errors you find. I will indicate updates in the title of this post.


I leave off with the remark of one of the U.S. passengers interviewed while still on the Diamond Princess:

“Being knee deep in the middle of a crisis leaves a person with two options — optimism or pessimism. The former gives a person strength, and the latter gives rise to fear.” (link)

He, like the others who were evacuated, faced an additional 2 weeks of quarantine.[9] He has since returned home and remains infection free.



[1] As a noteworthy aside, Fauci was able to assure the interviewer that the “danger of getting coronavirus now is just minusculely low” (in the U.S. on Feb. 17). What a difference 2 weeks can make.

[2] In a 2015 paper, Chen and colleagues found a cruise ship’s ventilation spread particles from cabin to cabin. They found that 1 infected person typically led to more than 40 cases a week later on a 2000 passenger cruise. By contrast, the coronavirus, with a reproductive rate of 2 cases per infected person, would only lead to 3 new cases during that time. Planes rely on high-strength air filters and are designed to circulate air within cabin sections.

[3] In a March 23 CDC report: Among 3,711 Diamond Princess passengers and crew, 712 (19.2%) had positive test results for SARS-CoV-2. Of these, 331 (46.5%) were asymptomatic at the time of testing. Among 381 symptomatic patients, 37 (9.7%) required intensive care, and nine (1.3%) died (8).

They found coronavirus in Diamond Princess cabins 17 days after passengers disembarked (prior to cleaning).

[4] A table from the Japanese National Institute of Infectious Diseases (NIID) (Source LINK):


“There were some limitations to our analysis. Cruise ship passengers may have a different health status to the general population of their home countries, due to health requirements to embark on a multi-week holiday, or differences related to socio-economic status or comorbidities. Deaths only occurred in individuals 70 years or older, so we were not able to generate age-specific cCFRs; the fatality risk may also be influenced by differences in healthcare between countries”.

[6] In a March 26 article by Fauci and others, Covid-19 — Navigating the Uncharted, we read:

“If one assumes that the number of asymptomatic or minimally symptomatic cases is several times as high as the number of reported cases, the case fatality rate may be considerably less than 1%.”

[7] R0 may be viewed as the expected number of cases generated directly by 1 case in a susceptible population.

[8] The number in the most recent report is 712, but that would be after the quarantine ended on Feb 19.

[9] I read today that one of the U. S. evacuated passengers just entered a clinical trial on remdesivir. This would be over a month since the end of the first quarantine.



  • Giwa, A.,  LLB, MD, MBA, FACEP, FAAEM; Desai, A., MD; Duca, A., MD; Translation by: Sabrina Paula Rodera Zorita, MD (2020). “Novel 2019 Coronavirus SARS-CoV-2 (COVID-19): An Updated Overview for Emergency Clinicians – 03-23-20”; Pub Med ID: 32207910; (LINK)
  • Japanese National Institute of Infectious Diseases (NIID). “Field Briefing: Diamond Princess COVID-19 Cases, 20 Feb Update” (LINK)
  • Rocklöv, J., Sjödin, H., & Wilder-Smith, A.  “COVID-19 outbreak on the Diamond Princess cruise ship: estimating the epidemic potential and effectiveness of public health countermeasures”, Journal of Travel Medicine, (Feb 28, 2020) [link]
  • Russell, T., Hellewell, J.,Jarvis, C., van-Zandvoort, K.Abbott, S.,Ratnayake, R., Flasche, S., Eggo, R. & Kucharski, A. (2020). “Estimating the infection and case fatality ratio for COVID-19 using age-adjusted data from the outbreak on the Diamond Princess cruise ship.” MedRXIV: The preprint server for the Health Sciences. (March 9, 2020). (PDF)
  • Zheng, L., Chen, Q., Xu, J., & Wu, F. (2016). Evaluation of intervention measures for respiratory disease transmission on cruise ships. Indoor and Built Environment, 25(8), 1267–1278. (First Published online August 28, 2015 ). (PDF)
Categories: covid-19 | 10 Comments

Post navigation

10 thoughts on “The Corona Princess: Learning from a petri dish cruise (i)

  1. Thanks for posting this discussion. It was enlightening.

  2. Thank you for this interesting article and analysis.

    It is very regrettable that we don’t know when each passenger contracted the virus. It would be helpful to know how many were infected prior to the “lockdown” and how many more contracted it (other than from their cabin mate) during that period. That would be tremendously interesting and also help quantify the concern about the recycled air and inadequate filters.

    It is interesting that passengers with symptoms were matched by only one infected/asymptomatic extra case.

    Another surprise is that the asymptomatic passengers didn’t skew to the younger age groups. The sample sizes are low, but it almost seems as if younger passengers may have been more likely to be symptomatic than older ones. That’s a curious contrast with the morbidity rates.

    I’m also surprised by the suggested R0 rate of 3.0 in Wuhan. To be blunt and also polite, none of the Chinese data is very robust; I tried to trace the references that evidenced this claim back, but came to a dead end with a non-online reference. My sense, and I’ve not tried to model any of the data, is that the R0 “felt” like less than 3, although of course a lot depends on the serial interval as well.

    But the massively greater R0 claimed for the ship really just points that a ship is not a good model to use for an “ordinary” environment. And that, at last, is really my point. That we lack adequate data to fully understand the Diamond Princess case, and that in any event, the special nature of the cruise ship environment does not translate reliably to regular populations ashore.

    • David:
      Thank you so much for your comments. You point out a number of things I hadn’t noticed, e.g., about ages of asymptomatic cases.

      “It is very regrettable that we don’t know when each passenger contracted the virus. It would be helpful to know how many were infected prior to the “lockdown” and how many more contracted it (other than from their cabin mate) during that period.”

      It would be great to know this. If everyone had been tested at the start, it would have been clear, or at least clearer. The Hong Kong passenger had symptoms on Jan 22, and there was another passenger who did on Jan 23 (who also tested positive). In the “official” reports it is surmised that the majority of the infections occurred before the quarantine. That would mean during the 5 days the Hong Kong passenger was on board. I can’t help feeling that the health officials are being defensive in presuming this, since otherwise, they’d have to admit transmission occurred during the quarantine (and not just between the crew members who lived and ate together). Fauci, on the other hand, in saying the quarantine failed, is clearly saying there was transmission during the quarantine. But he claims to be mystified as to what went awry.

      “But the massively greater R0 claimed for the ship really just points that a ship is not a good model to use for an “ordinary” environment. And that, at last, is really my point.”

      True, but it reflects what could be the case since, after all, it really was the case on the ship. I wonder that there isn’t more discussion today of the possible spread through recirculated air. As I was reading about this, I began walking outside here more (I live in the mountains in the country), and will open windows when it’s warm enough.

      I wonder if they contrasted infections among those in windowless cabins and those with outside decks. (Of course there are confounders).

  3. I noticed in the “related posts” that WordPress automatically generates, that there was a 2014 post on Ebola, quarantines, and Arthur Caplan saying that quarantines were a bad idea (then). Here’s the link:

  4. James T. Lee

    ED Note: This comment refers to a different post to which I allude in my comment: I brought it over here so that people would find it.

    Language is powerful. In the present hornets’ nest atmosphere both precision and accuracy of speech need our closest attention. Dr. Caplan is a philosopher and brilliant thinker but he is neither a physician nor public health expert. He seems confused about both the meaning and medical importance of “Quarantine”.

    He betrays gross misunderstanding in his very first bullet point, “Quarantining people without symptoms makes no scientific sense”. What really makes no sense at all is Caplan’s very first sentence!

    Quarantine is a public health tool used for observing people who have had a credible exposure but NO SYMPTOMS. Caplan makes a layman’s mistake in getting confused by two facts: (1) quarantine involves isolating plausibly well people from society and (2) SICK people are also isolated during their treatment/recovery. Sick people with any contagious disease absolutely must be isolated from open society during their recovery period to prevent disease spread. Deep confusion about the meaning of “quarantine” is very widespread right now — watch CNN or MSNBC for 15 minutes and you can see for yourself.

    The utility of quarantine as a combined detection / prevention tool goes back many years in the history of public health and the term refers to the process of removing ASYMPTOMATIC people who have been credibly exposed from all contact with open society for a defined period of time (T) in order to observe these isolated people for reliable signs/symptoms of some disease X. Quarantine does reduce the spread of contagious diseases that are transmissible by human-to-human contact. The value of T is determined from empirical knowledge of the maximum incubation period for disease X as recorded for large numbers of past cases. Incubation period is defined as the number of days between exposure and the date of disease X development. Some exposed people develop a disease very quickly and obviously; others may not develop disease that promptly.

    For COVID19, T = 14 days and the median incubation period is known to be between 5 and 6 days. The nasty skunk at the picnic here is that people who have COVID19 disease can be highly contagious virus shedders, yet feel just peachy during their early incubation period. Not all contagious diseases are like that.

    In sum, if quarantined person Joe Smith does not become ill during T days of isolation from society, we can conclude with high confidence that Mr. Smith has not contracted disease X. On the other hand, if a quarantined person DOES develop findings that pin down the diagnosis of X during time T, the process of quarantine has not only allowed diagnosis of X at the earliest possible moment, but it has also prevented that person from exposing other people. Quarantine breaks the chain of transmission from persons who are asymptomatic but shedding dangerous pathogens during their incubation period.

    I understand that Dr. Caplan’s “points” were made in the context of a discussion about Ebola virus infections. Nonetheless, in the present COVID19 disaster it is absolutely certain that we are dealing with a new disease that is (a) Highly contagious, and (b) Definitely transmissible from persons who are carriers of the virus even though not yet symptomatic (they look well, feel well etc.). There is one more thing to note — Ebola is much more lethal than COVID19 and the irony here (or is it a paradox?) is that COVID19 arguably is far more insidious than Ebola because COVID19 victims can “stick around” much longer while shedding active SARS-CoV-2 virus.

    The lay public may be getting unwittingly hoodwinked in hearing from our government that “80 percent of you will be just fine if you get this illness”. Mr. Pence keeps sending that basic message although in not so many words. Well, yes maybe, contingent on the meaning of “just fine”. But persons who have “mild” COVID19 disease do also shed the active virus. COVID19 is a highly contagious disease that, unlike Ebola infection, does not incapacitate its victims rapidly. Quarantine is an absolutely essential tool in the present chaos.

    (Comment originally posted today under:

    • James: I moved the comment you made on my 2014 post, mentioning Caplan) to this post since it is relevant, and no one would find it there. I doubt Caplan feels that way about today’s quarantines. On the “paradox”, I think that’s precisely what we expect from natural selection, too lethal and you kill all your hosts.
      I’d be interested to hear more of what you think regarding today’s covid-19 policies.

  5. A tweet by Ron Kenett:

  6. This link says 10 deaths, but the CDC report says 9. This is 2 months out of course.

  7. rkenett

    Mayo is handling this as an applied statistician would. We are facing an unstructured problem. The first step in the life cycle view is to do some problem elicitation so that is can be handled in an analytic space. She then considers all dimensions of information quality. Given data, X, goal, g, analysis methods, f and utility, U, the InfoQ she generates has high: InfoQ=U[f(X\g)}.

    The three points worth expanding on are: 6) Generalisation of findings, 7) Operationalisation of findings and 8) Communication of findings.
    i) How can the Princess cruise data analysis findings be applied to other situations
    ii) How can these findings turn into policy decisions
    iii) How can this interesting analysis be communicated to the right persons at the right time.

    Stronger responses to these would make the information quality in Mayo’s blog even stronger.

    The corona virus global crisis is also a crisis of information quality. There seems to be a herd effect in policies implemented by governments. The aggregated data we get in more and more fancy graphs do not account for local effects. We are information poor, and this carries a terrible price.

    For more on reviewing applied work with information qualty dimensions see 10.3233/SJI-160967

    For the evolution of quality, from product quality to information quality see

    For a life ccycle view of statistcis see

    For a ten questions on statistics and data science see

I welcome constructive comments that are of relevance to the post and the discussion, and discourage detours into irrelevant topics, however interesting, or unconstructive declarations that "you (or they) are just all wrong". If you want to correct or remove a comment, send me an e-mail. If readers have already replied to the comment, you may be asked to replace it to retain comprehension.

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.

Blog at