How Many People Have Died of Covid-19?

Why the true number of deaths is probably higher than our official count

Pictured: Pretty grim. Source: Pexels

The Covid-19 pandemic has been fraught with unknowns, from the nature of the disease to broader societal impacts and everything in between. The last 15 months have often been more about the endless cavalcade of things we don’t really understand than the few facts we can be sure of.

Imagine each of these drummers is something we’re uncertain about. Source: Pexels

Nowhere is this more true than in the figures we take for granted as accurate. I wrote recently about how unsure we are of the true number of Covid-19 cases in the world. There’s another place where precise figures are much more fluid than you might imagine.

I’m talking about the number of people who have died from Covid-19.

Now, the initial response to the question of how many people have died is to simply go to one of the many excellent dashboards that collate worldwide Covid-19 data and simply read off the figures. According to most data collations, the number is somewhere around 3 million deaths as of April 2021, and still steadily climbing.

But the thing is, confirmed deaths are not a perfect representation of the true number of fatalities. With most diseases, we have to carefully estimate the true number of deaths, because it’s quite complex to know if someone died of a disease, especially when they weren’t tested for it.

There are, ultimately, two possibilities when it comes to Covid-19 deaths. We could be undercounting the death toll, or overcounting it (technically it could also be perfectly accurate, but as I’ll explain, this is very unlikely). However, it appears based on quite a bit of recent evidence that we are probably undercounting Covid-19 deaths, and potentially by quite a large margin. The true death toll from Covid-19 could be nearly double the official number of 3 million.

Let’s look at the data.

Overcounting

The first potential issue that we have to deal with is overcounting. The basic idea is that we may count deaths as Covid-19 fatalities when really they were caused by something else. The problem here tends to be that sometimes people have positive Covid-19 tests, yet die of something else, but are accumulated into the same coronavirus death dataset at a national level.

This famously happened early on in the pandemic in the U.K. After carefully reviewing their procedures for counting Covid-19 deaths, the U.K. government revised down its death toll by a bit over 10% in mid-2020. This prompted other governments around the world to review their death counting because having such a large overcount is a real worry when it comes to public health decision-making.

According to stock photo websites, decision-making is mostly about chess. Source: Pexels

What did these reviews find? Well, by and large, they showed that in most places, death overcounting was not much of an issue. In Belgium, for example, there is good evidence that the official death toll does not overcount deaths by much (if at all). In the United States, the CDC recently published a paper demonstrating that 91% of official Covid-19 deaths had Covid-19 as the underlying cause, with the remainder having Covid-19 listed as a contributing cause. In other words, the U.S. at most overcounted the death figures by 9%, but may not have overcounted Covid-19 deaths at all. A recent investigation by the U.K. Office for National Statistics found that similarly most official Covid-19 deaths were caused by the virus.

In general, overcounting appears to be fairly minimal. It’s mostly an issue in places that do a very large amount of Covid-19 testing, and even then most countries have refined their death reporting to make sure that the official figures are not substantially higher than the true number of deaths. We are probably counting some deaths in our figures that were not caused by Covid-19, but almost certainly not more than 10% of all deaths, and probably far less than that.

Undercounting

So overcounting is mostly a nonissue — it happens, but it probably doesn’t have much impact.

What about undercounting? This is an issue that happens in a few ways, but it basically boils down to a lack of testing — sometimes people have Covid-19 but don’t get tested before/after they die, and so they are not recorded in the official statistics for the disease.

Ah, the joy of Covid-19 tests. Source: Pexels

There are several ways to try and see if deaths are being missed in official reporting systems, but one commonly used one is to look at excess mortality. This is a calculation where we compare the expected number of deaths in a given year to the actual recorded number and see if there have been more deaths than anticipated.

For developed countries like the United States, these excess mortality calculations show that there has probably been some undercounting of Covid-19 mortality. In 2020, there were 373,000 officially recorded deaths, but 549,000–668,000 excess deaths. While some of these excess deaths were due to other causes, the CDC predicts that some portion of them were also due to people who died of Covid-19 being misclassified.

But when we look at less-developed nations, the problems with undercounting becomes far more severe. Mexico recently revised its death figures for Covid-19 upward by 60%, adding more than 100,000 deaths to their figures. A paper looking at excess mortality in Russia estimated that the true death figures may be two to three times the officially recorded figures. There’s evidence that some places in India may be missing half of all Covid-19 deaths. A recent study conducted in Zambia that tested people who had passed away from Covid-19 found that the true number of deaths in the country could be 10x higher than the currently reported estimate.

The visual abstract of the Zambia paper. It’s well worth a read. Source: BMJ

These numbers are vast, far exceeding the relatively modest issues with overcounting. Even if we assume that all countries overcount deaths by the highest margin seen — 10% — we’d be looking at a reduction in our global death toll of 300,000. Just the figures from potential undercounts in Mexico and Russia add up to nearly that much, and these two countries are probably just the tip of the iceberg.

Indeed, a preprint looking at excess deaths attempted to calculate the likely total global number of deaths caused by Covid-19 and estimated that it was probably at least 1.6x the reported figure. In other words, we’ve recorded 3 million total deaths from Covid-19 worldwide, but the true figure is likely to be at least 1.8 million higher, and potentially far more.

Bottom line

What does this all mean?

Well, firstly, it means that our death counts are quite uncertain. We have an official global figure for the number of Covid-19 deaths that puts it at about 3 million as of April 2021, but the true numbers could be very different and we aren’t sure by quite how much.

What we do know, however, is that we probably aren’t overcounting deaths on a global scale by very much. While there are almost certainly some deaths that are being inappropriately counted as being caused by Covid-19 that were really due to something else, the number is probably enormously outweighed by undercounting. It’s plausible that the true global toll of Covid-19 is 10% lower than we think, but it’s quite likely to be 50% higher or more. Given the data from some developing nations, it’s entirely possible that there are millions of uncounted Covid-19 deaths.

The slightly terrifying fact is that we won’t know for sure how many people died from Covid-19 for some time yet. These sorts of careful inferences—estimating deaths where the official data isn’t perfect — take years to carefully work through.

That being said, the best evidence suggests that, overall, we are probably undercounting Covid-19 deaths by quite a wide margin.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store