Skip to content

Raw data: Cumulative excess deaths from COVID-19

I haven't done a COVID-19 post for a little while, so I decided to use my dex-driven insomnia to do one tonight. Warning: I got pulled down a bit of a rabbit hole.

This time I calculated excess deaths per country, which is generally considered a more accurate metric than official figures for COVID death rates. Here are cumulative deaths per million since the start of the pandemic:

There are no big surprises here. The United States is highest and the Nordic countries are lowest. As usual, I'm comparing largeish, rich countries in Europe plus the US and Canada. The reason is that these countries are all culturally similar and economically similar, which means they could be expected to have fairly similar resources for responding to COVID.

But then I decided to look at all these countries by how far north they are. This is because there's a common belief that colder countries have lower natural rates of COVID spread. I used the population center of each country, not the geographical center, since viruses obviously spread among people, not trees or wheat fields. Here it is:

The trendline that showed the closest fit was a power curve, which had a surprisingly high R-squared of 0.58. This means that latitude explains 58% of the variability in COVID death rates.

As an aside, there was very little difference in geographical center and population center. In some cases this was because the countries were too small to allow much of a difference in the first place (Belgium, Switzerland) and in other cases it was because the population was fairly centrally dispersed (US, Germany). The country where it had the biggest impact was Canada, where the population center is near Toronto, which is almost literally the southernmost point in the country. The most surprising countries were Norway, Sweden, and Finland, which are all tall, thin countries with big populations in their southern climes but have a difference of only a degree or two between their population and population centers.

The first thing that comes to mind when you see this chart is how far each country is from the trendline. So I checked that out:

The winner (i.e., loser) is the UK, which has a death rate far higher than it should. The US, by contrast, is fairly average. We have a high death rate, but we're also the farthest south of every comparison country (even farther south than Greece).

Using this metric, Sweden looks worse than its raw count suggests, as does Spain. Canada is by far the best, along with Norway and Denmark. Germany and Switzerland, which usually get good marks, turn out to be only average considering their location.

50 thoughts on “Raw data: Cumulative excess deaths from COVID-19

  1. timdanner

    If you're willing to go further down the rabbit hole, the next dimension to analyze would be age distribution by country. Some of these European countries are notably older than the US, and age is a significant driver of covid mortality.

  2. Vog46

    Gee Kevin why not re-chart this into which countries use Standard Time versus Daylight savings time as well?
    By education level?
    How about by abortion access?

    Good grief

    1. Jasper_in_Boston

      Huh? I don't get this attitude at all. Kevin explains with considerably lucidity why he's chosen the various data points of this analysis (that is, why they might be relevant to the question of excess mortality).

      Which of the factors he's looked at do you find questionable, and why?

      1. bobsomerby

        As a possible way of explaining the attitude, here's Jim Bouton, quoting Gene Brabender in Ball Four:

        "Where I come from, we only talk so long. After that, we start hitting."

    2. kaleberg

      One big problem is that our host uses the word "explain" as opposed to "correlate". If you want latitude to explain the difference, one has to come up with an actual causal relationship. There are just too many things that correlate with latitude, many for historical, not geographical, reasons. For example, it probably isn't abortion laws, though they have been historically looser in the north than in the south, closer to Rome and the Vatican.

    3. azumbrunn

      All or almost all countries on the list have Daylight savings time, mostly with changes on the same dates as the US.

      Before making snide comments check your facts!

  3. Thomas Andrews

    If the correspondence is not linear, you probably need a weighted average of latitudes, especially for the US. But average latitude is probably good enough.

    Of course, US elderly population tends to move south, so weighting by age would possibly be an adjustment. But an interesting first pass.

  4. Pittsburgh Mike

    I thought that colder countries were supposed to have worse Covid results, because people were stuck inside more.

    This seems to be an example of being able to explain any result with some handwaving. I'm unconvinced that we, or any of the high death rate countries, are doing average or above.

    I don't know exactly why we suck, but we do. I'd guess it is our half-assed lockdowns early on, and a relatively large number of old unvaxed people. I doubt it is latitude, longitude, average internet speed, or vitamin C consumption.

    1. Jasper_in_Boston

      I don't know exactly why we suck, but we do.

      2020:

      A) Pandemic arrived to hit a population that suffered from more poverty and related health problems than other rich countries.
      B) A government that not only was uncommitted to robust mitigating measures, but in many cases was actively attempting to spread the virus.

      2021:

      A) The rich world's strongest and most vocal vaccine-hesitancy community, translating into mediocre vaccination levels, and near disastrous levels with respect to booster shots.
      B) The poverty and attendant health-problems haven't gone away, either.

      Anyway, that's my take.

    2. Mitch Guthman

      Apart from agreeing with Jasper’s analysis below, it seems to me that a portion of Kevin’s premise about colder climates is wrong since it fails to account for the fact that in hot and/or humid climates in the role of air conditioning. My educated guess would be that Arizona as much or more time indoors as anyone in the world and maybe more.

  5. Steve_OH

    Latitude is not a very good proxy for climate when you're comparing North America to western Europe. Europe is much warner than it "should" be for its latitude. For example, the climate in Stockholm (59.3°N) is comparable to that of Portland ME (45.5°N). This makes places like the UK fare worse than they would if you took actual climate into account.

    1. cephalopod

      Exactly.

      If you are trying to look at the impact of climate, it would make more sense to measure against something like average high in January.

      It would also be interesting to compare individual US states to European countries.

        1. cld

          Then in the first chart isn't it saying that Italy and the UK are huge outliers that latitude doesn't account for?

            1. Steve_OH

              Italy, yes, the UK less so. Using mean temperature rather than latitude either moves the Europe to the left or North America to the right. In either case, the outlier-ness of both Italy and the UK is reduced.

    2. azumbrunn

      Yeah, I too thought that the Gulf Stream should be somehow part of the equation. In Europe they teach it in primary school, in the US apparently not.

      If you move the US more South (maybe around Italy), it looks even worse than now. But it also destroys whatever significance the data have.

      BTW if we believe the temperature-COVID connection the UK looks even worse too: They are the biggest profiteer from the warmth of the Gulf Stream. What is wrong with Anglo-Saxons?

  6. golack

    Have you looked up average temperatures? Sunshine, maybe....but temps?
    Avg. US high temp: 53 F. UK: 56 F. It was harder to look up than I thought it would be...so I'll stop there...

  7. Bardi

    Don't forget that Sweden started out with some misunderstanding concerning "herd immunity" and their stance consisted of no initial response, to be changed when they realized how stupid they were.

    1. sturestahle

      Sweden going for herd immunity is only a truth in non-Swedish media (or more precisely,,it’s fake news). I have never heard a Swedish official saying we were .
      Sweden not having an initial response is also fake news .
      Greetings from your Swedish friend
      (don’t bother to post links to non-Swedish media, if you are interested try to check info from Sweden)
      Have a nice day

      1. kaleberg

        Sweden did take some measures, but they were not as draconian as in many other places. They did ban gatherings of more than 500 which effectively shut down a lot of things, though they did manage to get in a last Eurovision final.

        Judging from the Swedes I have heard from, directly and indirectly, one thing that helped is that Swedes frequently follow government advice even if it is not implemented as law. In some countries, advice is ignored and laws routinely flouted. The government discouraged face masks but encouraged social distancing. High schools and colleges went online. One blogger I follow reported that his LARPing groups shut down and in-person board gaming went virtual.

        Surveys and cell phone records show that Swedes stopped going out a much at the same rate as people in other countries with explicit shutdowns. Even in the US, people cut activity whether their state had an official shutdown or not. Swedes were encouraged to do their own contact tracing and to stay home when they had symptoms. Swedes have an excellent vaccination rate.

        How this correlates with latitude is another matter.

        1. sturestahle

          A good description!
          One has to understand Sweden to understand why we acted as we did.
          Sweden is always accused of having a “big government “ .
          We don’t!
          The actual Government Office is extremely small and the power of the Swedish government/PM is negligible compared to their colleagues elsewhere, definitely compared to the president of the United States of America.
          A Swedish government cannot restrict citizens movements, cannot set up roadblocks with personnel armed with automatic guns and cannot use drones to track people out in the streets. The government cannot order private shops to close down completely.
          A government can issue recommendations based upon recommendations from government run specialists/scientists and we Swedes are following scientifically well documented recommendations
          We are a little “exceptional “ (weird)
          Check this one ( just scroll down and check the upper right corner of the infogram)

          https://www.worldvaluessurvey.org/WVSContents.jsp?CMSID=Findings

          We made a number off mistakes , the first and most obvious one our scientists made was that they forgot that 20% of people living in Sweden wasn’t born here, many of them being refugees from countries where one sure doesn’t trust the government (it’s a proud Swedish tradition to give shelter to people coming from war zones and dictatorships)
          We should have used other channels and methods to inform these groups of the situation
          Well , it’s getting late over here, this is just a short answer but I am tired
          Have a nice day my friend

          1. KenSchulz

            Most of the government actions you cite are also forbidden to the President of the United States. Lockdowns, travel restrictions, closing of private businesses, mask mandates, etc. were imposed by governors or local authorities, who have powers not delegated by the central government. Unlike most Westminster systems, ours is truly federal, not unitary.

            1. sturestahle

              The powers of your president is unique in a country claiming to be a democracy. (And is in my opinion an abomination)
              .Head of state
              .Leader of the government but contrary to a PM in a parliamentary system is he able to totally disregard the rest of the government and run a one man show
              .Undisputed commander in chief
              .Executive orders!
              .Having the power to override the Congress using a veto and with the help of a few cronies in the Senate able to override the Congress if the Congress tries to override a veto
              This reminds me of the power of a sovereign monarch from times gone by
              In a parliamentary system , like the ones in the Nordic countries, are the power resting in the arms of the parliament

    2. Jasper_in_Boston

      ...Sweden started out with some misunderstanding concerning "herd immunity" and their stance consisted of no initial response, to be changed when they realized how stupid they were.

      Coronavirus: Swedish King Carl XVI Gustaf says coronavirus approach 'has failed'

      https://www.bbc.com/news/world-europe-55347021

      How Sweden botched its coronavirus response:

      https://foreignpolicy.com/2020/12/22/sweden-coronavirus-covid-response/

      Sweden acted too slowly as virus swept country:

      https://www.reuters.com/world/sweden-acted-too-slowly-pandemic-hit-commission-finds-2021-10-29/

      Scathing evaluation of Sweden's COVID response reveals 'failures' to control the virus

      https://abcnews.go.com/Health/scathing-evaluation-swedens-covid-response-reveals-failures-control/story?id=83644832

      1. sturestahle

        Only four?
        I could find an infinite number of them, all repeating the same and all of them giving an distorted version of it .
        The “best” (aka worst) ones are to be found in British the Times and Washington Post.
        … just a small advice

  8. realrobmac

    Where are these cumulative numbers coming from? It seems off by an order of magnitude. If total COVID deaths in the US were really 352 per million (a total of about 110,000) we wouldn't even be talking about it. Isn't the real number around 3000 deaths per million?

    1. Andrew G

      These are excess deaths related to Covid, not deaths directly caused by Covid. Covid killed many people that would have died anyways (or at least, we would have predicted some large fraction of Covid deaths would have died even if they didn't get Covid).

  9. sturestahle

    Mr Drum please!

    “…..there's a common belief that colder countries have lower natural rates of COVID spread. “
    “Common belief “ usually equals stereotypes , or “me and the boys discussed this in the bar over a couple of beers “
    Could you please verify it a little more scientific.
    Greetings from your Swedish friend

  10. denniscferguson

    Climate, rather than latitude, would certainly make Canada less of an outlier. I know from personal experience that winter weather in Toronto is significantly colder than Copenhagen and Stockholm (and Toronto is considered to have fairly mild winters by anyone living in Canada east of British Columbia), though with the summers being warmer the yearly averages are probably pretty similar.

    If climate is an important issue, though, one would think this would show up in comparisons between different parts of the US.

    1. golack

      There was discussion of seasonal outbreaks--winter outbreaks in the North, summer outbreaks in the South--seasons where people spend time indoors. But Omicron kind of hit everywhere around the holidays.

  11. typhoon

    This is an interesting exercise. But, I think arguing for northern latitudes being a prime driver of lower excess deaths is misguided. Europe, especially the Nordic countries, UK and Ireland are significantly warmer than the U.S. and Eastern Europe given their latitude. Also, as mentioned elsewhere, being stuck inside seems to have been associated with Covid surges.

    In my humble opinion, I think there are two main factors that best explain this. The first, which is easy to quantify, is per capita GDP. Of the 18 countries listed, the 4 with by far the lowest per capita GDP are Italy, Greece, Spain and Portugal, which are 4 of the 5 worst performing countries. The second factor is a more squishy factor which I’ll call the prevalence of freedumb, or perhaps the prevalence of lack of respect for the governmental. Obviously, the U.S. would be high on such a measure, explaining our poor performance. The best performing countries seem to me to be places (Nordic nations, Canada, Germany) where my impression is people are less likely to hate the government.

    Anyway, it’s interesting.

    1. illilillili

      Yes, but, that's all directly caused by latitude. Southern countries have low GDP because stuff just grows there and you can live off the low hanging fruit. Nordic nations are socialist because if you don't cooperate, you freeze to death.

      😉

        1. Crissa

          What's your definition, and how does it match up to socialist practices and demands of equal care and opportunity?

  12. skeptonomist

    First, as others point out, if temperature is really the thing to be tested for, then use the temperature at population center and not latitude. Second, some countries, the US in particular, have a wide range of both temperature and latitude - what is the internal variation? Northern parts of the US have similar temperature to the population center of Canada. How does Maine or New York compare to Canada and to Florida?

    Determining what geographic/physical aspects as well as national policies affect covid rates is is a very, very complex problem, and the pandemic is not over yet. As I pointed out before, there seems to be a tendency for countries to converge to a high and fairly narrow range of infection rates:

    https://ig.ft.com/coronavirus-chart/?areas=eur&areas=usa&areas=gbr&areas=kor&areas=aus&areas=sgp&areasRegional=usny&areasRegional=usnh&areasRegional=uspr&areasRegional=usdc&areasRegional=usfl&areasRegional=usmi&cumulative=1&logScale=0&per100K=1&startDate=2021-06-01&values=cases

    although countries which postponed infection and vaccinated seem to have avoided high death rates. China (for example) may be in for a huge wave of infection when they open up (like South Korea and Iceland).

    When the pandemic is really over, it might be possible to sort out the many possible influences, using data for *all* countries. So far nobody has isolated a single magic controlling factor. Latitude is probably not it.

  13. golack

    And somebody must.....
    Correlations does not mean causation. Trend is interesting, but could also be an artifact. "Excess deaths"--if the calculation includes "normal" deaths from the flu, and we had no flu outbreaks due to anti-covid measures, then places normally hit harder by flu outbreaks may look better than places normally not hit as hard.

  14. sturestahle

    I gave a somewhat sarcastic but at least fairly polite answer above.. but I decided to go all in .
    The latitude idea is batshit crazy.
    In my Sweden isn’t it just common knowledge but it’s statistically more than a truth that flue season is winter , when we are indoor and it always blows over , completely, in summer. This goes for seasonal flue and definitely for Covid . Check our curves for Covid
    A small reminder from a Swedish troll

  15. illilillili

    Plot to some sort of average temperature and not latitude. The UK is an island, so the temperature is moderated by the liquid oceans around it. Canada has a huge amount of land far from oceans and so will tend to be colder than its latitude suggests.

  16. Pingback: Cumulative excess deaths from COVID-19 | Later On

  17. Andrew G

    Really interesting stuff. But as some other commenters have mentioned:

    1. latitude isn't average temperature, and
    2. don't we expect cold to drive people indoors, thus making the flu and Covid worse during colder months?

    Regarding #2, it could be that people living in colder countries are more used to the cold, so they spend more time outdoors when it's cold out. That should also predict that seasonality of the flu and Covid would be lower. But that doesn't really explain why a colder country would have lower Covid excess deaths.

  18. rs

    I'm getting tired of this focus on rich, european countries plus US and Canada (the global "north") with the obviously problematic and incorrect explanation that "these countries are all culturally similar and economically similar, which means they could be expected to have fairly similar resources for responding to COVID". Just take a sample of countries around the world, and make some kind of correction for GDP if you will, and perhaps account for democracy vs. dictatorships. Culturally there is uniformity globally because COVID and the response to a large extent have a basis in a common shared scientific temper. Vaccines have been developed and produced in other countries, and people wear masks and wash their hands as well.

Comments are closed.