About this site
Who are you?
Hi! My name's Alex Waygood, and I'm a freelance journalist. You can visit my main website here.
What's this website all about, then?
This website auto-generates graphs comparing excess mortality rates between countries over the course of the pandemic.
What's ‘excess mortality’??
Excess mortality looks at the total number of deaths a country experienced over a given time period,
and compares it to that country's long-term trend over the past few years.
The "excess mortality" figure is the total number of deaths
that exceeds the "expected" number of deaths for that timeframe.
By calculating this figure for the period since the COVID-19 pandemic began,
we can get to a way of measuring the impact of the pandemic that is truly comparable between countries.
Simply showing the raw number of excess deaths, however, is somewhat unfair to larger countries.
These countries would normally have far more deaths in any given year—
because they have larger populations—
and so, it stands to reason that they would have more excess deaths during a pandemic.
1,000 excess deaths would be far more shocking in Ireland,
for example, than it would be in the United States,
as there are just far fewer people living in Ireland than in the United States.
As a result, the graphs on this website show excess death figures
expressed as a percentage of the "expected" number of deaths,
rather than the raw number of excess deaths.
This adjusts for the population disparities between countries.
Why not just look at confirmed COVID deaths rather than ‘excess mortality’??
Most countries keep a count of all deaths officially confirmed to have been due to COVID-19.
However, these counts are often inaccurate, and usually unsuitable for comparisons between countries.
Countries vary widely in the number of COVID tests that they do relative to their population,
and there is no internationally agreed-upon definition of what constitutes a ‘COVID death’.
This means that many people who die of COVID will not be officially recorded as such.
Excess mortality isn't a perfect measure: in some countries, such as India, for example,
even data on the total number of deaths from all causes can be unreliable.
Moreover, figures for recent dates are liable to be revised upwards,
even in countries that have reliable statistics.
Finally, some countries
take far longer
to report statistics than others.
Making international comparisons is a tricky business
whatever metric you use,
so these graphs can only tell you so much.
Nevertheless, excess mortality--where the statistics are available--
is generally a far better metric than confirmed COVID deaths.
These graphs are a pretty good place to start.
To read more on the pros and cons of excess mortality as a metric,
and the different ways of measuring excess mortality, I can recommend Our World in Data's
methodology guide.
But hang on. Surely if you use ‘excess mortality’, that's liable to be affected by all sorts of non-COVID factors?
It's true! Excess deaths could be affected by anything!
There are good reasons to think carefully about what these graphs are telling us.
For example, there are plausible reasons to think that lockdowns might have led
to some deaths in some countries.
Many people have suffered negative mental-health outcomes due to loneliness,
and still more have put off visits to the doctor that might have seemed inessential at the time.
There are also plausible reasons to think that lockdowns might have prevented
some non-COVID deaths that might otherwise have occurred.
Reducing the number of cars on the road, for example,
could plausibly have led both to less pollution and to fewer traffic accidents.
And overall mortality rates are impacted by other diseases, too.
If a country has experienced a major surge or decline in another infectious disease,
this will clearly also have a large impact on their mortality statistics for 2020-2021.
This point isn't just hypothetical. Flu rates have plummeted over the past year-and-a-bit,
due to lockdowns around the world.
Nonetheless, for the vast majority of countries,
a declining flu rate and fewer traffic accidents will not have been nearly enough
to have fully offset the death toll of COVID-19.
So it's important to see these graphs as a metric of a country's overall pandemic
response—rather than simply as ‘how many died of COVID’ over a given period.
However, they remain, arguably,
one of our most useful tools for making international comparisons regarding the impact of COVID-19.
Where are you getting your data from?
None of the data used on this site was collected by me.
All credit must go to the incredible team of data journalists at the Financial Times,
both for collecting the data and for making it
open-source
for the world to see. This site would not be possible, otherwise!
I would encourage anybody who enjoys this site to read the
FT's
coverage
on excess deaths, which they have made entirely free to read.
Why aren't all countries in the FT's dataset?
Not all countries report excess mortality statistic promptly, and some don't report figures at all! The FT can only give figures for excess mortality when they are available (although there has been a recent attempt by The Economist to model excess deaths in countries where official figures are not available, which is well worth a look).
How are you measuring ‘excess mortality’?
I take the FT's figure for the number of deaths in a given week for a given country, and divide that by the FT's figure for the number of deaths that country would "expect" to experience in that week of the year, based on the FT's analysis of the long-term mortality trends in that country.
Why do the data series for different countries start and end on different dates?
The FT's data series for a country starts at the point
at which that country hit 100 confirmed cases of COVID-19.
Prior to this point, although it is possible that COVID-19 may have been circulating
in any given country undetected, any fluctuations in mortality figures cannot reliably be said
to have been caused by COVID-19.
Each country's data series ends at the latest available data point in the FT's dataset.
Different countries report this data on different timetables,
so there may be much more up-to-date data for some countries than for others.
How are you making these graphs/this website?
All the HTML and CSS on this site has been manually coded by me. The backend runs entirely in Python, using the Flask library to handle requests to the server. The graphs are generated on the server side using the pandas and matplotlib Python libraries, while the lines on the graphs are smoothed using a cubic interpolation function from the Python SciPy library. The web app is hosted by Google App Engine.
Can we see the code behind this website and the graphs?
Sure! It's all right here.
Couldn't you have done [feature of this website] better/more effectively if you'd used a little more Javascript?
Probably! But I know Python very well, and don't really know Javascript much.
I should probably learn Javascript a little more, but that's a project for another day :)
Hey, you seem cool, can I employ you?!
Please do!!