2.6
Ignoring time in processes
Specific claim
Since we cannot calculate mortality and morbidity for lack of reliable data, let’s put the dubious
ratios aside and use only the most accurate reports we have - the Diamond Princess numbers.
Coronavirus caused an unusual number of deaths, seven on a single ship. It cannot be compared to
the flu.
Diamond Princess quarantine wasn’t an instantaneous event. It was a process that took about
four weeks. It’s a relatively short time, so it might seem that adjustments for baseline mortality
are negligible. It turns out otherwise.
Let’s examine the numbers. From February 2nd 2020 to March 1st 2020, 7 people died on
Diamond Princess. How many people would have died in normal circumstances (in the absence of
coronavirus)?
Since the majority of the passengers were Japanese, for best approximation we use data from the
mortality database from the Japan National Institute of Population and Social Security Research.
[22, 23]
In Table 2 we use the crude mortality rate per age group to estimate the expected number of
deaths per year that would happen on Diamond Princess in the absence of special conditions.
Age group
0-9
10-19
20-29
30-39
40-49
50-59
60-69
70-79
80-89
90-99
Crude mortality rate
(actual data for Japan from 2016
that serves as an estimate for 2020)
0.002260
0.000276
0.000756
0.001122
0.002524
0.006265
0.015194
0.038053
0.124751
0.390717
Total:
Persons
aboard Expected deaths
per year
16
23
347
428
334
398
923
1015
216
11
3711 0.04
0.01
0.26
0.48
0.84
2.49
14.02
38.62
26.95
4.30
88.01
Table 2: Expected deaths per year for the Diamond Princess population
We limit our analysis to the quarantine period from February 2nd to March 1st which is 29
days. Assuming a uniform distribution we expect 7 (29/365*88) deaths during this period.
If we redo the calculations with the updated data on the resolved cases on March 28th, we get
12 actual deaths on Diamond Princess and 13 expected deaths in the period from February 2nd
to March 28th 2020
We need to remember that this number is encumbered by selection bias. We are applying
crude mortality rates of population of Japan to the subpopulation that is not matching in terms
of ethnicity and social status. In particular the cruise ship population is likely characterized by:
• higher than average socioeconomic status
• better access to health care
• healthy enough to be able to travel
Time cannot be ignored. We should consider the expected number of deaths in the same period
in normal (infection-free) circumstances. Although there is an unknown margin of error related
to selection bias, knowing that with the number of actual deaths on Diamond Princess we didn’t
8