How Many People are Dying from Covid-19?
Spoiler Alert: Nobody knows.
There is, right at the moment, a proliferation of data freely
available that would usually be hidden behind a paywall. Most of this is data
held by some of the world’s great medical journals such as Lancet and The New
England Journal of Medicine and as a result of the Covid-19 pandemic, most
journals and other databases have taken down their paywalls and made data
freely available.
There is also an awful lot of data being published by
various governmental and supranational organisations. Add to this the
aggregating being done by earnest geeks at github and Wikipedia and it is, for
once possible, that a nerd with a spreadsheet has a fighting chance.
So it is possible, for example, to determine which countries
have reported cases of Covid-19 and how many cases they have reported. It’s also
possible to check which countries have reported deaths from Covid-19 and how
many deaths they have reported. Unfortunately much of the reporting is inaccurate
for reasons that I’ll outline below, but lets start with some top 20s.
These numbers are the best I could get at around 18:00 on 2nd
April 2020.
Table 1 –
Top Twenty Countries by Number of Cases
Table 2 –
Top Twenty Countries by Number of Fatalities
Table 3 –
Global Totals
How do we measure how deadly it is?
A simple death rate for a disease, sometimes called a “Mortality
Rate” is calculated by dividing the number of deaths by the number of cases and
expressing the result as a percentage. So for example, if 10 die and 100 are
infected, the mortality rate is 10%.
This number is always going to be an approximation as not
all cases or deaths will necessarily be known. Some can be overestimated and
some under.
The scientists in the field, epidemiologists, tend to prefer
using the “Case Fatality Rate” (CFR). While this can never be known exactly,
there are various statistical tools that can be used to achieve a result that
is a good predictor for large populations.
One outcome of this is that while a CFR might be (say) 1%,
in a particular population the mortality rate might show as much higher or
lower. Examples of some CFR calculations are 9.5% for SARS, 34.4% for MERS and
0.1% for seasonal influenza.[i]
For Covid-19, the current mortality rate is about 5% worldwide.
The tables above show rates in other countries. However we know for a fact that
there are a great many cases that are undiagnosed, so we can be fairly
confident that the CFR will be some number lower.
One of the reasons we know that there will be a large number
undiagnosed is a lack of testing. Tests were not widely available until March
and remain costly, time consuming and difficult to complete. It means that
tests are not being done randomly in most countries as we simply don’t have
that luxury.
A second reason why there are undiagnosed cases is the
problem of asymptomatic carriers.
The Problem of Asymptomatic Carriers
One factor complicating the epidemiology of Covid-19 is the
relatively high rate of asymptomatic carriers. That is to say, a fairly large number
of people contract the disease but show no ill effects. This is a phenomenon
that is not unique to Covid-19 and many diseases will have carriers that are
asymptomatic.
Asymptomatic carriers create difficulty from a statistical
point of view because they are often not counted amount the infected. They create
an even bigger problem in containment of the disease as they can spread it widely
and unknowingly. In the early 20th century, a lady by the name of
Mary Mallon worked as a cook in a number of wealthy household that would later
suffer from Typhoid. While Mary herself was asymptomatic, she was a carrier of
typhoid and became the cause of several outbreaks. Today “Typhoid Mary” is part
of folklore.[ii]
For Covid-19, we know that there are some carriers that are
asymptomatic, but it’s hard to know how many. What would be really helpful,
would be a locked down community of people where we knew exactly how many
contracted the disease, how many remained asymptomatic and how many died.
Luckily enough we have exactly that.
The Natural Experiment of Diamond Princess
On the passenger ship Diamond Princess, there were 3711
people and 705 of those were diagnosed as positive and 7 of those died, giving
a mortality rate of just under 1%. Of those diagnosed, 17.9% were asymptomatic,
which is to say that they didn’t feel any symptoms of the disease.[iii]
Had these people not been on board a ship where a passenger had been positively
diagnosed, it is doubtful that they would ever have been tested.
We also know that almost 60% of the passengers were over 60[iv],
a figure that would be under 30% in even the oldest countries in the world. To
what extent this affected the rate of asymptomatic carriers is difficult to guess
but it almost certainly means that the final CFR will be higher on the ship (1%)
than it is in the general population.
There appear to be several other factors that are affecting
the local mortality rate for many countries although the science here is very
new and, as yet, there are no peer reviewed papers. One factor being reported is
that the local climate plays a role, with slightly cooler climates being
preferred.
A second factor is that the disease appears to be getting
less deadly with time. Early mortality rates in China ran between 4-5%, but
more recently the estimated CFR is running at around 1% outside of China.
It’s not unusual for a virus to become less deadly over time
through natural selection. From an evolutionary point of view, a “perfect”
virus is one that would actually make the host stronger and more likely to
survive. A very deadly virus is less effective because the virus depends on the
host moving around so that it can spread. Scientific inquiries have established
that a dead host tends not to move around as much as a live one.
How contagious is Covid-19
For any disease, initial figures for infections are likely
to be underestimated. When a disease is first discovered, it’s likely that the
first person discovered is not the index case. They have most likely caught it
from someone else. In this way, our initial number of cases is 1 and that value
is almost always underestimated.
This might seem like a trivial example, but it’s a good
illustration of a problem that will plague us for some time: there will be
people that have the disease that have not been identified yet. It’s a problem
that’s exacerbated by the asymptomatic carriers discussed above.
It became pretty clear early that Covid-19 has a high rate
of infection compared to other diseases. Epidemiologists refer to a number
called the “basic reproductive number” also sometimes called the R0 (pronounced
as “R nought”) value. The value of R0 is an approximation of the
number of people that will contract the disease from one carrier.[v]
This number is very important in determining how fast a
disease will spread and in estimating how many people need to have the disease to
create “herd immunity”.
The Basic Reproductive Number of Covid-19
The best estimate of the R0 at present is about 1.94.[vi]
It might be a little higher or a little lower, but the best research we have
right now is around that number. Earlier in the outbreak we thought it was much
higher, as high as 3.8 in some early reports.[vii]
In the early days of the outbreak, there was also a lot of wild
information flying around about how R0 of measles is up to about 15
(true) and so 3.8 is nothing to worry about (false). The R0 is but
one factor and anything above 1 is a problem. The R0 of the H1N1
influenza virus that caused the 1918 Spanish Influenza pandemic was between 1.7
– 2.0[viii]
and 50 million folks died.[ix]
For an influenza pandemic, anything above about 1.5 is high.
So the R0 for Covid-19 is high. It’s indisputably
a high number and by any reasonable measure, it’s a highly-contagious disease.
One impact of that is that there are likely to be a lot of cases in general
society when it is first discovered.
It also means there will be some huge problems putting the
genie back in the bottle by using widespread testing. We will eventually get to
the point where there will be random testing for Covid-19, but for now that’s
just not possible. I’ll discuss testing in more detail in a separate article.
What Next?
It’s not clear yet how the disease will end. There are many
experts in many parts of the world that are working on tests for the disease
and tests for antibodies to the disease. Progress is being made but it will be
several weeks before any country is able to do the testing on a grand scale
that we need.
There are also teams working on vaccines, treatments and
cures. As a card-carrying nerd, it makes me proud to see all the medical nerds
working furiously and without regard for trivialities like national boundaries or personal hygiene.
There has been some disgraceful behaviour by national and
international bodies in the last few months and I’ll write about this if there’s
enough interest. There’s also a lot of discussion among economic nerds about what
this will all mean for the world once the ‘rona is behind us.
I hope my scribblings have left you a little better informed
than you were when you started. If nothing else, I hope they’ve provided a distraction
for a while. I would find it very encouraging if you could like/share/comment on
the article.
Shane
[v]
This is a highly simplified explanation. For more detail see https://www.sciencedirect.com/topics/medicine-and-dentistry/basic-reproduction-number
or if that’s too much work, take a look at https://en.wikipedia.org/wiki/Basic_reproduction_number
[ix]
In total around 500 million people were infected and around 50 million died.
Records are not great as we were coming out of WWI at the time and we had other
things to worry about. For more detail, take a look at this - https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3291398/
Image by Pete Linforth from Pixabay
Tables: own work
I really hope I'm only the 1st commenter and not the 1st reader. Awesome post, I hope you continue the blog. Thanks!
ReplyDeleteI’m looking forward to more! Great guide for dummies. I’ve been sharing this with everyone who is finding it difficult to maths. It’s amazing how many people don’t understand numbers. Thanks for putting it into words!
ReplyDeleteHi Stefaling and thanks for your support!
DeleteYou wouldn’t happen to be in Sweden I suppose? I’ve had an oddly high number of hits from there.
Hope you enjoy the rest of my blogging efforts. :)