Wednesday, April 8, 2020

The Diamond Princess - Estimating the COVID-19 Case Fatality Ratio in a Closed System

As I posted here, the global health community is dealing with an incomplete database when it comes to a statistical analysis of the current COVID-19 pandemic.  This makes it impossible to produce accurate statistics regarding the infection rate, mortality rate and transmissibility of the current novel coronavirus, particularly given that a significant portion of the population is either asymptomatic or presents with very mild symptoms that they may not seek treatment or testing.  That said, at this point in the evolution of the COVID-19 pandemic, there is one source of data that can give us a reasonable sense of the mortality rate; the Diamond Princess cruise ship which was quarantined at the Yokohama port facility in Japan from February 4th to February 19th, 2020.  At that point in time, those that had intense contact with a confirmed case-patient (i.e. sharing a cabin) were held in quarantine beyond the 14-day window.  The fact that the Diamond Princess was a closed system makes the infection and mortality rate more accurate than other estimates where testing is far from complete. 

At the time of quarantine, there were 2,666 guests and 1,045 crew working onboard the Diamond Princess with approximately half of the guests being from Japan.  The average age of those aboard the Diamond Princess was 58 with a 1,231 being 70 years of age and older.  The first confirmed case of coronavirus from a passenger was discovered on February 1, 2020.  This passenger, a guest from Hong Kong, travelled onboard the Diamond Princess between January 20th and January 25th, 2020 and tested positive for the coronavirus on February 1, six days after disembarking.   

Research by Timothy W. Russell et al released in a paper entitled "Estimating the infection and case fatality ratio for COVID-19 using age-adjusted data from the outbreak on the Diamond Princess cruise ship" in which they analyze the data collected from the Diamond Princess passengers and crew.  Among the 3,711 passengers, 3,063 PCR tests were performed, starting with the most elderly patients and descending by age, resulting in the following (data to February 20, 2020):

Confirmed cases - 634 or 17 percent of all passengers

Asymptomatic cases - 328 out of the confirmed cases or 51.7 percent of the total

When looking at age-structure data, the authors found the following corrected Case Infection Ratio and corrected Case Fatality Ratio using a non-truncated hospitalization-to-death distribution:

1.) All ages combined: 

Case Infection Ratio (CIR) - 1.9 percent (range 0.60 to 4.3 percent)
Case Fatality Ratio (CFR) - 0.91 percent (range 0.11 to 4.3 percent)

2.) 70 years of age and older:

Case Infection Ratio (CIR) - 7.3 percent (range 3.0 to 14 percent)
Case Fatality Ratio (CFR) - 14 percent (range 6.0 to 27 percent)

Here is a graphic showing the number of daily confirmed cases and number of daily deaths for the Diamond Princess:

Here is a table breaking down the crew and passenger age, infection and death data:

From this data and after adjusting for the delay from confirmation-to-death, the authors found that of the passengers that tested positive for the COVID-19 virus and were symptomatic, the fatality ratio was 1.9 percent.  When all asymptomatic passengers that tested positive for the COVID-19 virus are included, the fatality ratio was 0.91 percent.  When these numbers are extrapolated to China, the authors estimate that 1.1 percent of symptomatic cases resulted in death and when asymptomatic cases are included, the fatality ratio drops to 0.5 percent.

Let's close with this comment from lead author Timothy W. Russell.  He notes that, while there are still uncertainties in the data because some passengers may have become symptomatic or died after the end of February 2020, he feels that using a true fatality ratio of 0.6 percent to 0.7 percent is a "good ratio".  This is higher than the normal seasonal influenza fatality ratio of 0.1 percent but not as alarming as the 4 percent ratio that some politicians and decision-makers are using to create fear and submission.  

1 comment:

  1. It appears that the number of people who died has almost doubled. Depending on the ages of those who subsequently died, the age-specific death rates could be double or triple that in the paper. I haven't found an updated analysis.