Excess Deaths 2020-2022

I read a blog post which looked at excess deaths in OECD countries from the beginning of the pandemic until present. The author made the claim, and showed with arithmetic, that Sweden, with its far less aggressive approach to the pandemic, actually did the best of all OECD countries.

I sent it to a colleague whose field is evidence based medicine, and while he agreed that excess deaths is the most relevant measure, the tone of the post made him question the results. So I told him I would double check the results using my own methods.

why excess deaths?

Why is excess deaths the most important result for judging the effectiveness of pandemic public health policy?

It’s hard to get correct, comparable numbers about COVID deaths. Testing is not consistent in different places and over different time periods. People can die from COVID without having been tested. If they were tested, there can be different standards for reporting if the person died from COVID or just coincidentally tested positive when he died from another cause. This latter could happen because hospitals, where a lot of deaths occur, test everyone even if COVID was not the admission diagnosis.

Furthermore, even if a person’s death was precipitated by COVID, and so legitimately a death from COVID, he might have been destined to die from his other health issues within a week or a month anyway.

Excess deaths also serves to normalize different populations somewhat by age and health status. countries with more elderly and more chronic disease would be expected to have more deaths per year even without a pandemic and so looking at excess deaths would help us not credit to policy that which was more due to the existing characteristics of the population. Policy should get credit for preventing death.

Finally, and relevant to public health policy, measures we take to control COVID deaths could lead to increased deaths from other causes. If a country reduced COVID deaths by 500k but increased heart attack deaths by 500K, that policy earned us net zero.

oecd data

The data are freely available from the OECD stats website and easy to download so you can check yourself. Of course one can ask questions about OECD methodology for determining excess deaths in the different countries, but for the purposes of this blog post and these calculations I am not going to do that.

https://stats.oecd.org/index.aspx?queryid=104676

OECD data had absolute numbers of excess deaths for every week from the beginning of 2020 (the approximate start of the pandemic from our point of view) until present. For each country I added all of these together to get the total number of excess deaths and then I divided that by the number OECD gave for the population of the country at the start of 2020. I can see arguments for using different numbers for normalizing by population, but this makes the most sense to me.

In this way we answer the question “of the people who were alive in your country at the start of 2020, what percentage of them were unexpectedly dead by the end of December 2022?”

data, calculation, & chart

The table below shows the results of my calculation and is sorted by excess deaths/starting population. I also show the percentage of population that was “fully vaccinated” by the end of the calculation period. I think this means two shots but not necessarily boosted but I’m not certain.

Country	Fully Vaccinated %	Excess Deaths (% of starting population) 2020-2022
Costa Rica	82.68	0.075
Sweden	72.35	0.133
Norway	74.61	0.171
Israel	65.16	0.180
Australia	82.73	0.263
Iceland	77.82	0.269
New Zealand	79.79	0.270
United Kingdom	75.19	0.272
Denmark	80.98	0.308
Luxembourg	71.42	0.314
Switzerland	68.78	0.411
Belgium	78.64	0.419
Canada	82.5	0.454
Finland	78.46	0.464
United States	68.97	0.503
France	78.34	0.505
Germany	76.19	0.543
Austria	76.32	0.546
Spain	85.49	0.596
Mexico	64.82	0.599
Netherlands	68.08	0.603
Italy	81.26	0.695
Portugal	86.5	0.712
Colombia	71.04	0.727
Slovenia	57.66	0.776
Estonia	63.95	0.783
Chile	90.26	0.813
Latvia	70.57	0.814
Hungary	62.27	0.819
Greece	73.59	0.888
Lithuania	68.36	0.892
Czech Republic	65.67	0.974
Slovak Republic	45.68	1.084
Poland	56.74	1.227

So by my calculation the actual champ is Costa Rica with Sweden at a respectable #2.

Why is this important? Earlier in the pandemic the true believers in lockdowns and more restrictive policies wanted to explain why Sweden, despite not having above average COVID deaths, was still kind of a disaster if you looked at it correctly. Sure, Sweden did very well compared to Europe as a whole(and the US), but that’s not the correct comparison they argued. What you really need to do is see how well their policies worked compared to their neighbors, the other Scandinavian countries, Denmark, Norway and uh…I’m not really good at geography, Finland? By that standard Sweden did poorly. They had many more COVID deaths than the other Scandinavians.

Why is it more informative to compare Scandinavians to other Scandinavians rather than Germans, English or even Italians? I agree you want to compare groups that are similar based on factors that might be medically relevant. On that basis there are a lot of factors that argue for the “Vikings to Vikings” comparison being a good idea but there are also many factors that argue for broader comparisons.

For example, while Viking heritage might make people inherently more immune to dying from COVID, it might be a smaller factor than age distribution in the population. It might be a smaller factor population density and urban vs rural distribution. And if the race (ethnicity? Not sure what Scandinavian is) is relevant, shouldn’t they also compare the distribution of ethnicities within those countries? My vague impression is that Sweden had more immigration from non Scandinavian people than Denmark.

That debate is no longer needed. None of its neighbors did better than Sweden on the most important measure of pandemic success, excess deaths per population.

vaccination stuff

And so I could make an original contribution, not just check someone else’s work, I am checking the association between very high vaccination rates and excess deaths. Some people say only at risk people (elderly or having some serious chronic condition) should get the shot. Some people say everyone should get the shot to help protect the vulnerable people. Originally the argument was that we needed to achieve herd immunity. You may have noticed that no one has said that in a long time. That’s because with Delta and especially Omicron it became clear that herd immunity was never possible.

But you don’t need to achieve herd immunity to see a benefit from there being, on the average, fewer people around who can potentially spread, or even just making everyone a slightly less effective spreader. It might show a benefit to make sure everyone is vaccinated all the time or at least get as close to that goal as possible.

So I thought it would be good to see if there is an obvious benefit to higher vaccination rates that we can see in the excess deaths results. See scatter plot below.

This is far too few data points to draw any definite positive conclusions but it is helpful because of what we don’t see- an obvious benefit to more vaccination all the way up to 100%. At most we can say lower rates of vaccination (eyeballing it maybe below 70 or 65%?) might be a problem, but above that level the results look pretty random.

This makes sense on an instinctive level if you assume that people are sensible and that people who are vulnerable know, on average, to take precautions, and that people who interact with vulnerable people know, on average, to take precautions. And that on average the people refusing the vaccine are in low risk of death groups and possibly don’t interact on average much with at risk people.

What you don’t see much evidence for here is the benefit of trying to get to 100%. It could be there is a marginal benefit, and it’s just not visible above all the noise of other differences between these countries, but I would argue if you want the government to force everyone to take a shot, you want to see an obvious benefit in the numbers.

The X- axis is percent of the population that was fully vaccinated by the end of the period. The Y-axis is the excess deaths as a percent of the beginning population.

conclusions

What can we conclude? We have few data points, so it’s more what we cannot say: 1) There is no evidence here that societies should try to enforce very high rate of vaccination. Above 70% or so we see no additional benefit. 2) When planning for next time we should consider Sweden to be more an example and less a cautionary tale.