What can Explain Sweden’s Debacle?

It has been a while since my last post on the pandemic, so it might be time for an update of what I am thinking about it now. Debates on policy responses have somewhat settled, at least in the public spheres of social media and regular media. I expect the scientific community to discuss different aspects of the pandemic for many years to come, but the attention span of the public is a lot shorter and now other things are becoming more interesting in the eyes of the general public.

Despite the decreasing interest, I would like to discuss a working paper which I stumbled upon a few weeks ago. The paper asks whether there are any factors other than lockdown policy that may explain the difference in death rates between Sweden and the rest of the Nordic countries. This paper is especially interesting since the main conclusion of the paper has been discussed by chief epidemiologist Anders Tegnell, just the past few days.

The authors of the paper suggest 15 possible factors to Sweden’s higher death rate other than the less stringent restrictions on swedes compared to other Nordic countries. I am going to raise two points about some of the factors suggested, then I am going to venture into a discussion on timing and stringency of lockdowns.

The main factor that the authors suggest is what they call the “dry-tinder” situation in Sweden. They get the name “dry-tinder” from an analogy to forest fires. When there have been a few years with very few or no forest fires, then the next forest fire may be larger than it otherwise would have been because of all the dry tinder that did not get burnt down in the previous years. (I really do not like to compare elderly people to dry tinder, but here I am just using the same terminology as the authors of the paper) The authors of the paper provide evidence that Sweden experienced an unusually mild flu season in 2018/19 (see my figure 1, their figure 5 from the paper below), which may have left more weak and elderly swedes alive to, unfortunately, contract the coronavirus. This is the possible explanation that Tegnell has discussed the last few days. The authors suggest that this factor alone could account for 25-50% of the entire Swedish death toll. I am not sold on their estimate of the total effect of this factor, but there is a possibility that the relatively low death rates during the last flu season actually increased the total deaths in covid-19.

A close up of a map
Description automatically generated
Figure 1

Another bundle of factors concerns the fact that Sweden, and Stockholm especially, may have faced some circumstances that may have acted to increase the spread of covid-19. Most interesting for the coming discussion is the fact that Stockholm, where the spread of covid-19 has been the worst in all of the Nordics, had its “sportlov” in week 9, while other countries and regions had their “sportlov” in weeks 7 and 8. Many swedes travel to the Alps to ski during the “sportlov” and most of the early spread in Europe came originated in the alps. This factor is rather weakly supported in terms of data in the paper, but it may be a factor. More interesting though is that the metro system in Stockholm is much larger and much more used than its counterparts in Oslo and Copenhagen. This may also have increased the spread of the virus. The authors claim that these factors increased the spread in Stockholm, which lead to the worse death rate in Sweden.

It may be that the authors are correct. But I do believe that if one wanted one could have created a similar list of circumstances that made any country in the Nordics have a higher death rate than the others. The real difference between Sweden and the others come from the slower and less stringent adaptation of restrictions. As seen in figure 2 below, Finland and Norway imposed light restrictions in late January. This was before their respective “sportlov”. Sweden and Denmark imposed light restrictions later, Denmark on Feb 27 and Sweden on March 9. With a delay of a few weeks the death rates of Sweden and Denmark are higher than the death rates of Finland and Norway. See figure 3. This also seems to point against the authors’, whom are suggesting that Sweden had a higher number of imported cases which made Sweden have a higher number of cases. We must also remember that the testing capabilities are different in each country which matters a lot in terms of case counts. Relative to many other countries, Sweden has tested very little and many tests may have given false negatives etc.

A close up of a map
Description automatically generated
Figure 2

This also suggests, with very little data, that the timing of restrictions could have a huge effect on the resulting spread of the virus. And the coming analyses of stringency versus timing is what I am looking forward to in the scientific aftermath of the pandemic. Unfortunately, we cannot have any counterfactuals, but I still think that we can learn a lot about government responses to situations of crisis.

To add to the above evidence, it would be beneficial to look at the case count a while after restrictions were set in place. As we can see in figure 4, the case count in Norway and Denmark was higher than Sweden’s up until April 1. At this point both Denmark and Norway had imposed restrictions that gave them a score of above 60 in the Government Response Stringency Index, these restrictions were imposed around March 15-17 in both countries. Which shows that restrictions may need some time to gain effect.

It seems as if more stringent responses in combination with good timing decreases the death rate. Still, many critics of lockdowns argue that lockdowns impose costs on both the economy and the freedom of the populace. I agree, such costs are imposed by lockdowns. But it is definitely not obvious that less stringent restrictions avoid such costs to the economy. Again, Sweden get to pose as an example. Swedish GDP fell by 8.6 percent in Q2, while in Denmark the decline only amounted to 7.4 percent despite more stringent restrictions.

Why Sweden has such a high death rate compared to the other Nordic countries will likely be hard to exactly pinpoint and I am sure that the question will be debated for a long time after the pandemic ends. Hopefully we can learn from what has happened, and what will happen, to make sure that we can meet future pandemics in a more prepared and effective way. I think we owe that to all the victims of the virus.

A close up of a map
Description automatically generated
Figure 3
Figure 4

Leave a Comment

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.