22

Sections of the Australian media have been claiming recently that Sweden experienced lower excess deaths during the pandemic years compared with other countries, including Australia. The implication is that Australia was wrong to impose lockdowns during 2020 and 2021, because Sweden didn't.

Here is a non-paywalled article from news.com.au about it from late last year and here is a paywalled article from The Australian, a more right-leaning mainstream media source from yesterday.

I have been looking for evidence for this but hampered by paywalls and the non-paywalled article seems pretty vague on it. Honestly I find it unlikely that the extra covid deaths would be outweighed by other mortality, though I understand that people delayed other medical care during covid, were isolated, etc, so it is plausible. I did find one journal article comparing Sweden to Norway which seemed to be vaguely in line with my thoughts, concluding (hardly decisively) that:

All-cause mortality in 2020 decreased in Norway and increased in Sweden compared with previous years. The observed excess deaths in Sweden during the pandemic may, in part, be explained by mortality displacement due to the low all-cause mortality in the previous year.

I know individual studies don't necessarily reflect the whole picture. So my question, using the headline from the non-paywalled article:

Is the claim that "Sweden [had] the lowest excess mortality rate [during] the pandemic, despite refusing to lock down" true?

Note: I'm only adding the political leanings of the outlets to provide context for readers who don't live in Australia. I think the characterisations are broadly uncontroversial and I hope it is clear that this is not the focus of the question.

12
  • 4
    Q title doesn't match bold claim. Title says "during", the body says "after". Mar 23, 2023 at 10:55
  • 4
    @Stef The argument is something along the lines of, "P1: the point of lockdowns is to lower deaths (by reducing deaths from covid). P2: lockdowns cause other harms. P3: Sweden had less deaths and no lockdowns. C The harms of lockdowns may outweigh the benefits" Mar 24, 2023 at 1:48
  • 3
    @BugCatcherNakata But Sweden is so different from Australia?! Just because Sweden had less death than Australia doesn't say much about how many deaths Australia would have had if they had chosen a different policy. In fact those two countries are so different from each other, it looks like they were chosen with the intended purpose of making the argument look absurd.
    – Stef
    Mar 24, 2023 at 9:41
  • 5
    One might posit that Swedes would take precautionary measures (wearing masks, distancing, avoiding gatherings, etc.) without lockdown, and Australians wouldn't. In that case, lockdown may save many lives in Australia and would therefore be justified, but lockdown may not make much of a difference in deaths in Sweden. Sweden is just single one data point. If you use that alone to conclude that lockdowns are unnecessary, that is the definition of cherry picking. That's simply not how you draw conclusions from data.
    – NotThatGuy
    Mar 24, 2023 at 12:10
  • 3
    In Sweden public gatherings were limited to 8 people at its lowest. Restaurants had to have spacing between parties and had to close at 20:30. In shopping centres restaurants were limited to a party size of 1 to encourage people follow the recommendation to shop alone. The number of people allowed in a store was limited depending on the size of the store. Far from a lockdown but also far from full freedom.
    – Anders
    Mar 25, 2023 at 18:52

4 Answers 4

25

Sweden was not particularly good or bad on covid compared to Scandinavian peers, but it did better than some major developed countries

If we compare the cumulative covid death toll from trustworthy standardised sources (avoiding the possibility of using specific and unrepresentative time periods or other issues) then the claim appears to be wrong.

This is from Our World in Data (OWID—a trustworthy source) and includes the cumulative deaths from covid avoiding issues about arbitrary time period selections.

Cumulative covid deaths from OWID

(The picture is a screenshot but the link below should go directly to the live chart which will update as new data arrives.)

Link to live chart.

For those who think that inconsistencies in how deaths "from" covid are defined OWID also allows comparison of the excess deaths over the same period. This is arguably a better metric because it doesn't depend on attribution of deaths to covid. But it is harder to directly compare because the calculation depends on population demographics which often differ in significant ways that reduce the comparability.

But, in this case, the cumulative excess death chart shows roughly the same overall story at least to the most recent data.

It is also worth noting, though, that the timing of excess death differ by country even when they end up in the same place: Swedish excess deaths were higher early in the pandemic and then flattened; Finland and Norway were low early on but then rose, largely ending up close to those in Sweden.

OWID cumulative excess deaths

Link to live OWID chart.

While the two charts do not tell exactly the same story, neither has Sweden as an exceptional outlier over the whole pandemic period. Broadly, they seemed to do similarly to most of their neighbours, better than the the USA and UK, and worse than Australia. If anything Denmark, not Sweden is the Scandinavian outlier (doing better than Sweden).

Australia looks like a good performer on both metrics.

These charts might also explain why selective use of data gives a misleading picture in some reports

The two charts above show cumulative data over then entire pandemic period. But close examination (the charts are interactive on the OWID site) shows that both the rate of excess deaths and covid mortality varied a lot over time. This creates the possibility of selectively quoting the data over specific periods to give a deceptive view of the data. This may account for some of the more extreme claims (either through carelessness or a deliberate attempt to misrepresent the data).

0
9

It seems that the source of the recent news is a paywalled newspaper article in Swedish. It only compares some European countries so Australian media has probably added Australia themselves. The chart from the article in English(source): chart

The chart above shows 2020-2022 compared with 2017-2019.

If you compare with other years you get different numbers, Statistics Sweden has done a few other comparisons (source) chart

Y-Axis is excess mortality in percent. From left to right: Norway, Sweden, Denmark, Finland, Belgium, Netherlands.

  • Dark blue: 2020-2022 compared with 2015-2019
  • Light blue: 2020-2022 compared with 2017-2019
  • Green: 2020-2022 compared with 2019 alone

Excess mortality is a blunt measure, for example it does not take the age structure of the population into account, still it is likely the source of the recent media coverage.

9
  • 7
    Also, there are relevant differences between countries besides pandemic measures. You do not need to tell Swedes to stand 2m apart at bus stops, they've already been doing that before the pandemic. Mar 23, 2023 at 19:50
  • 2
    @SimonRichter plenty of Swedes were out partying when the rest of Europe was on lockdown: forbes.com/sites/davidnikel/2020/10/14/… (note that the 'warnings' were not enforced and many students ignored them). Mar 23, 2023 at 23:08
  • 3
    @JonathanReez, Europe was never in a "lockdown." There was no single day I wasn't expected to go to work in a non-essential job, and no single day I could not go shopping, but people balked at the idea of being asked to keep as much personal space as the Swedes do. Mar 24, 2023 at 1:34
  • 6
    @SimonRichter i.e. Italy had police chasing people away from empty beaches and plenty of countries had night curfews, not to mention the mask mandates. Sweden and the US were one of the few nations where a man could go anywhere at any time without being confronted by the police Mar 24, 2023 at 2:30
  • 2
    @SimonRichter: There were quite a few curfews, in France notably. Many students (across all age groups) studied from home, being outside was restricted to going to/from work and 1 daily "walk" within a given radius from domicile (at some points as low as 1 km). I would refer to such measures as lockdown. Mar 25, 2023 at 11:34
4

Regarding the SVD-based answer, there's slightly more to it:

Örjan Hemström, a statistician specialising in births and deaths at Sweden’s state statistics agency Statistics Sweden (SCB), put together the figures at the request of Svenska Dagbladet.

He told The Local that the numbers published in the newspaper came from him and had not been doctored in any way by the journalists.

He did, however, point out that he had produced an alternative set of figures for the Nordic countries, which the newspaper chose not to use, in which Sweden had exactly the same excess mortality as Denmark and Norway.

“I think they also could have published the computation I did for the Nordic countries of what was expected from the population predictions,” he said of the way SvD had used his numbers. “It takes into consideration trends in mortality by age and sex. The excess deaths were more similar for Denmark, Sweden, and Norway. Almost the same.”

Here are Hemström’s alternative numbers:

enter image description here

As you can see, only Finland significantly differs in the first column. Also that can be explained somewhat too

Another issue with the analysis is that the SvD graph compares deaths in the pandemic years to deaths over just three years, a mean of 2017-2019, and does not properly take into account Sweden’s longstanding declining mortality trend, or the gently rising mortality trend in some other countries where mortality is creeping upwards due to an ageing population, such as Finland.

Maybe someone else can look into the age structure of Finland vs its neighbors, because that's not too well detailed there. Anyhow, other expert were more cautious about drawing conclusions:

“It’s very difficult to compare countries and the longer the pandemic goes on for the harder it is, because you need a proper baseline, and that baseline depends on what happened before,” Karin Modig, an epidemiologist at Sweden’s Karolinska Institute whose research focuses on ageing populations, told The Local.

“As soon as you compare between countries, it’s more difficult because countries have different trends of mortality, they have different age structures, and in the pandemic they might have had different seasonal variations.”

Do the numbers match those provided by other international experts and media?

Sweden’s excess mortality over the three years of the pandemic is certainly below average worldwide, but it is only in the SvD/SCB figures that it beats Norway and Denmark.

enter image description here

The Economist newspaper also put together an estimate, using their own method based on projected deaths.

enter image description here

TLDR: the construction of the baseline matters quite a bit in this game. OTOH one could generally say that Sweden didn't significantly differ from its neighbors (minus Finland) in outcomes on that measure. As for whether their measures differed significantly... stay tuned. But the Local felt like pointing out that deaths from Covid did differ:

So if Sweden had similar excess mortality as the other Nordics over the period, does that mean it had a similar Covid-19 death rate?

Not at all. Sweden’s per capita death rate from Covid-19 over the period covered by the SvD/SCB figures, at 2,249 per million people, is more than double Norway’s 959 per million, 60 percent more than the 1,409 per million who died in Denmark, and more than 50 percent more than the 1,612 per million who died in Finland.

There are some graphs on that too, but I'm omitting them here as less relevant.

Interestingly however, there's this potential explanation:

“The most striking difference between Sweden and the other Nordic countries is that only Sweden had large excess mortality in 2020 and the winter of 2020-21,” [Preben] Aavitsland [the Director for Surveillance and the Norwegian Institute of Public Health] explained. “In 2022, the field levelled out as the other countries also had excess mortality when most of the population was infected by the omicron variant after all measures had been lifted.” [...]

“My interpretation is that in the first year of the pandemic, say March 2020 – February 2021, Sweden had several thousand excess deaths among the elderly, including nursing home residents,” he said. “Most of this was caused by Covid-19. In the other [Nordic] countries, more people like these survived, but they died in 2022. The other countries managed to delay some deaths, but now, three years after, we end up at around the same place.”

Alas there are no detailed graphs included to back up this interpretation. It may be a good question on its own, whether this explanation stands. On my quick check of it with The Economist's data as of now (warning it changes as it's updated regularly!) it appears to roughly check out, but one needs to be a little weary of the non-identical timeframes

enter image description here

Although OWID has somewhat different data, the (relative) trend claim can be easily seen in theirs as well:

enter image description here

That's the cumulative graph, by the way, the weekly one is much more noisy, but even in the latter you can see that the spikes become much more aligned in the past few weeks (unlike at the begging of the pandemic)

![enter image description here

This bit is more anecdotal, but I was curious for the reason for the spike in 2021 in Norway. Apparently they did relax mesures beforehand and then re-introduced them in mid Dec, but as it's been the case in many other countries that yo-yo'd the measures, governments were generally a bit late with reintroducing them.

Finally, I'm not so sure about the methodology/reliability of the OxCGRT stringency index, but according to that (also via OWID) there was not actually that much difference between these 4 countries that regard... I suspect that the weighting of the index's various components belies their relative effectiveness.

enter image description here

There are probably some [more] regression models for that stringency index vs Covid outcomes, but I could only find one for Latin American countries, and it uses Covid deaths, not total excess deaths, as the dependent variable.

Based on some slides I found (from an Oxford researcher), the effects of various measures were hardly equal, e.g. banning all gatherings >2 persons was much more significant than banning >10, and likewise closing all non-essential business much more impactful than a night curfew. So yeah, that OxCGRT index isn't too helpful as such.

2
  • The other countries managed to delay some deaths, but now, three years after, we end up at around the same place. => this maps well to statistical analysis showing that COVID speeds up death by 3-4 years on average: sciencedirect.com/science/article/pii/S1098301522000110 Mar 25, 2023 at 21:36
  • 1
    Thanks for this very detailed answer and exploration of the claim, I accepted the earlier answer because it is a more straightforward demonstration that the claim is false - for excess mortality Sweden is probably in the pack, not the best. But this answer was very interesting and informative. May 2, 2023 at 0:51
1

According to Reuters in February 2022, Sweden had an increased mortality rate and more deaths than other Nordic countries but not as many as some in Europe.

More than 17,000 people have died from or with COVID-19 in Sweden, far more per capita than among Nordic neighbours but fewer than in most European countries that opted for lockdowns.

Statistics agency Eurostat figures showed the country had 7.7% more deaths in 2020 than its average for the preceding four years, among the lowest excess mortality rates in Europe.

4
  • 9
    "have died from or with COVID-19" is different from "died from" and even more different from "excess deaths". The first is somewhat useless, as it would include people that died in car crashes, and in a country where the virus was allowed to spread one would expect such results from the "with COVID-19" statistic. Similarly "died from" is highly subjective to what the doctors and governments decide should go into that category. The "excess deaths" is used because it is objective and doesn't attempt to directly attribute cause and effect. Mar 23, 2023 at 16:07
  • 1
    @RayButterworth but it's also good to keep in mind that excess death stats are based on statistical prediction models and it's not straightforward how to interpret what is going on. For example, drug overdose deaths were about 40k per year for most of 2010's but that number went north of 100k in 2021(and it's possible that undercounting increased during covid, so that number could be off by more than you would expect previously).Traffic accident deaths also went up significantly. So when looking at excess deaths, you are looking at excess from not only covid, but also lockdown related effects
    – eps
    Mar 24, 2023 at 0:53
  • 3
    it also has fundamental issues when you try to answer the question "how well did this country handle covid". A country with a much older and overweight population is going to have more excess pandemic related deaths than a younger average weight pop, but that doesn't mean the younger country did "better", it's just that covid deaths were overwhelmingly deaths in older, obese, and sicker people.
    – eps
    Mar 24, 2023 at 0:57
  • @eps traffic accidents, overdoses and suicides are basically impossible to notice in overall mortality stats as there's too few of them. The vast majority of people die in old age, not from accidents. Mar 25, 2023 at 21:37

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .