Excess deaths in Germany, 2022 update

Numbers 1:46

Smurf, bragging

Now that destatis has published monthly deaths data for 2022, I will follow up on my previous posts from March and July. I am trying to better understand if there is excess mortality at all, to quantify it, and to locate it in the population. If you haven’t done already, please also read fellow substacker Witzbold’s report.

In order to improve on my previous analysis, I did not aggregate the 0-70 age band. Starting point are the population figures as of end of the previous year, stratified by age:

These numbers are interesting in themselves. Overall, the population has grown by 1.5 million over an 11-year period but the effect does not scale to the age groups. For example, now a much larger proportion of people is 80 or older. Here’s the year-to-year changes, with decreases in yellow:

The yellow diagonals indicate certain waistlines making their way up (or, in my ordering, down) the age pyramid. Every single year has a lot more deaths than births; the increase in population is due to net migration into the country. I don’t know what happened in 2011, maybe this is a data issue. The “migration crisis” of 2015 is clearly visible, as well as the decrease in migration during 2020, the first year of the pandemic.

Now deaths, of which there were plenty in 2021, and even more in 2022:

Dividing numbers of deaths by population as of end of the preceding year, we get mortality rates. Note that migration introduces uncertainty into these rates because it affects the numerators but not the denominators. This effect must be considered largest for 2015 and for 2022 (when more than one million refugees from Ukraine entered the country).

Since these mortality rates are hard to grasp, I distilled them into ranks (1 = largest rate among the 12 years, 12 = smallest rate) and painted them (1 = red, 12 = white):

Ranks of total mortality rates (last row) are almost perfectly in decreasing order, reflecting the ageing population. In the single age groups, however, things are not that simple. The year 2021 was worst for three age groups, and 2022 for another three, but both years are far from completely red. The year 2019, on the other hand, is white as snow, making it all the more astonishing that the “pandemic year” 2020 did not burn all the dry tinder – yes, some very old people died (mostly towards the end of the year), but that was that.

I am a big fan of counterfactual thinking, so I applied mortality rates from previous years to current populations, and compared the results to actual deaths. Positive numbers (in yellow, or in red if larger than 1,000) indicate higher actual mortality. This is 2020:

For example, in the 70-74 age group, the actual number of deaths was about 3,245 larger than the projection of the 2019 mortality rate would have predicted. Next comes 2021:

And finally 2022:

The very red last rows for 2021 and 2022 indeed indicate excess mortality, but how much? In a way, 2013 was much worse (presumably influenza) but we simply did not care back then.

I am using the difference of the “total” rows to quantify excess mortality from one year to the next. These “total” rows contain numbers of the form “actual deaths(year) – hypothetical deaths(year)”, or “act(year) – hyp(year)”. For example, by subtracting 2020 from 2021, we get:

act(2021) – hyp(2021) – (act(2020) – hyp(2020)) =

act(2021) – act(2020) – (hyp(2021) – hyp(2020))

This way, we get rid of effects due to changing populations. By then averaging over the row, we eliminate some uncertainty from “natural” variation of mortality rates. The value we get for 2021 is around 17,000, with a range of less than 3,000. So, my estimate of actual excess mortality of 2021 over 2020 is 17,000 (not 19,000, as indicated in this previous post, where I was being slightly Simpsoned by aggregating the 0-70 age groups). The actual excess mortality for 2022 over 2021 is around 17,000 as well (again, with a range of less than 3,000), giving 3 * 17,000 or around 50,000 for the 2021/2022 period over 2020 – which is around half of the total increase in the number of deaths. Excess mortality of 2021 over 2020 is present in all age groups except 85+ (which had taken a Covid hit at the end of 2020), and excess mortality of 2022 over 2021 is mainly in the 75+ age groups.

Year-on-year comparison, however, is not the whole story. What is more striking, and hinting at changes in the causes of deaths, is the distribution of deaths within the year. Before the pandemic, waves of deaths occurred in February and March, mostly driven by influenza. Maybe the pandemic (virus plus vaccination, that is) has turned the universe of respiratory viruses upside down (Eugyppius has written about this repeatedly)? I can only hope that this is the case. Just imagine an influenza wave in March of 2023 on top of what we have seen in December of 2022.

To illustrate, here’s deaths by calendar week for Germany, 2016-2022 (dashed lines indicate deaths minus official Covid deaths):

Note how subtracting Covid deaths (shaky as the numbers might be) in 2020 brings the blue line down to (very stable) 2016-2019 levels. That doesn’t really work for (the second half of) 2021, and not at all for 2022.

Will the public wake up? I am not sure. There are too many narratives floating around (“Our immune systems lack training because masks worked so well.”; “There have been more Covid deaths than officially recorded.”; “That’s Long Covid!”; “The health system is underpowered.” – that one is actually true; “It’s climate change!”; “Population size is being underestimated due to migration.”). The next milestones to observe are what happens in the spring of 2023 (the period formerly known as influenza season), and the statistics on causes of death for 2022 (don’t expect them to be published in time; they already had “issues” with the 2021 figures).