COVID-19: The Nordic Experiment

Note: This post is based on incomplete data sets, and will likely not age well.

How's that working out for you?

A few nations took a different approach to the global pandemic, but then changed course to fall in line with what most of the rest of the developed world is doing with social distancing and shutdowns. Great Britain tried it for awhile before they capitulated, then Japan did the same. From a soulless data analysis perspective, this is unfortunate, because both have large population centers and make good comparators to other nations in Europe and Asia.

There are a few holdouts remaining, including one I'd like to take a look at today: Sweden. A few caveats before we dive into the charts (which you can find at this source). First, I'll be making comparisons among Nordic countries, but that doesn't mean those nations are interchangeable. There could be reasons for the differences we're seeing beyond just their chosen approaches to curtailing the spread of SARS-CoV-2. If this were a randomized trial and we had a sufficient sample size we could treat the different groups the same. As it is, I still think there are conclusions we can tentatively draw from this data. Second, Sweden's population is about double the population of the other Nordic countries on this list. So while I'll show the total graphs, I'll pay more attention to the per capita graphs to make comparisons. This could also be a major confounder, since potential disease spread is heavily dependent on population density. Third, all these graphs suffer from the same inconsistencies in data collection and recording (sometimes driven by motivated reasoning) that I and others have discussed throughout the crisis.

On to the data! First, did Sweden do better or worse than comparable Nordic countries? We can start with total number of cases,



It looks like Sweden has double the number of cases as its neighbors. So the per capita chart shouldn't be surprising.


But that's entirely dependent on testing. Did Sweden do more or less testing than the other Nordic countries? Looks like less.



I hear people saying that despite Sweden's approach to COVID-19 they aren't seeing any more cases than comparable Nordic countries. I wouldn't go there based on the charts above. We have no idea what Sweden's actual rate is, because they're not looking as hard for those patients as other nations are. This is a similar pattern to what we've seen with other nations that haven't quarantined. They seem less interested in large scale testing to determine the magnitude of the viral spread in their societies.

We knew case numbers was a bad metric. On to deaths attributed to the virus. This is a much better measure, though still subject to reporting issues and other biases. Here we start to see Sweden pull away from the other countries.



In the area it matters most - per capita deaths - they still did poorly compared to their neighbors.


I think there are a couple of hypotheses we can assess based on these data:

  • Interventions will slow the increase in COVID-19
    • This happened. Especially when we look at the deaths data, it looks really good for the interventionists. This is good news for the next pandemic. If we saw something like a hybrid of MERS' case fatality rate (near 40%) and SARS-CoV-2's replication rate, we do have a validated tool in the toolbox with which to fight it.
  • Interventions will spread out the time which the pandemic happens (flatten the curve) so the same number of people will get sick, but over a longer period.
    • I think this is unlikely. I'm sure we'll get a more complete picture in the days ahead, but it looks to me like other Nordic countries are slowing down their rate of new cases/deaths, and that Sweden is either still going strong or also beginning to slow down. The 'flatten the curve' graph would predict still growing numbers for Norway, but that Sweden would hit a high peak and then decline rapidly. I don't think that's happening. I think at most Sweden could top out at around the same time as the other Nordics. This means the interventions weren't just about reducing the rate of spread, but also about curtailing the absolute number of people it spread to. In other words, it looks like we got the replication rate down below 1.0 without infecting everyone.
What should the lesson be for Sweden here? Should they be pariahs? Should they be ridiculed and mocked for not following what everyone else was doing? I think that's uncharitable and unwarranted. As I've maintained throughout the crisis, there's a high degree of uncertainty in situations like this one, and Sweden made the decisions they did based on the data they had available to them at the time. In fact, without them trying something different we wouldn't have any idea how effective our interventions can be in slowing the spread of a pathogen like this - especially in the modern day. Much of our prior experiences with true pandemics predate the information age.

Yes, a lot of people died as a consequence of their approach, but in the absence of a validated model - and in the context of the uncertainty present at the time they made their decisions - they couldn't have known exactly how many lives they were putting at risk.

Was it Worth it?

By now, lots of nations - especially in Europe - are starting to pass the plateau. It looks like Nordic countries are still at or near the plateau. You can see this most clearly when you look at semi-log graphs, though there's still a lot of noise in the daily rate, which would give us a better picture of what's really going on.


Now, let's assume Sweden tops out around the same time the other Nordic countries do, which looks like it will happen very soon. What's important here is not so much the place they top out at, but their position relative to the nations we're comparing them to. In other words, what's the ratio of deaths in the intervention countries versus Sweden as the control country. That should tell us how many lives the intervention saved. Currently, it looks like the difference is a little over twice as much as the next highest death rate nation, Denmark. Let's take a conservative estimate, and assume the ratio isn't 2:1, but far less favorable to Sweden. We'll instead use the average of Finland and Norway - around 20 deaths per million in the intervention group - and say that for every seven people who die in Sweden we could save six of them with the quarantine interventions.

To date, Sweden has 1,400 deaths attributed to COVID-19. If they end up with ten times that number of deaths, at 14,000, that represents about 0.14% of the population of the whole country. Here, we're assuming 12,000 of those people could have been saved with the right measures. How much did it cost to save each of those people's lives?

More importantly, who bears the brunt of that cost? The rate of unemployment in Norway was stable around 2.3% before the crisis, then rose to about 10.4% by March 24, and further to 14.7% by April 3 with the measures implemented. Let's say it tops out at around 20% and no higher. A similar intervention in Sweden would put almost two million people out of work. There's a strong link between unemployment and depression and suicide.  I don't think you'll have more people killing themselves than would die of the virus. However, most of the people dying of the virus aren't the people who are actually paying the price for it.

Students, too, are paying in the form of lost education. Children are our future and education is important and all, but students from college on down through grade school aren't getting the same education they would otherwise be getting. (Also, closing schools creates issues for nurses, but that's another story.)

It is true that stock prices have fallen dramatically throughout the crisis, but those will likely bounce back faster than other sectors. In the aftermath of the '08 crisis they certainly did, and since they've already made back more than half of what was lost since the nadir of the crisis they're likely to do the same this time.

I'm not going to do a Quality Adjusted Life Year analysis on how much was spent per life-year for COVID-19. Instead, let's talk absolute costs per person. Sweden's GDP is around $550 billion USD. It looks like the contraction is going to cause a loss in GDP of at least 5%, maybe more. That sets the overall cost of the response at no less than $27.5 billion for Sweden, in order to save around 12,000 people's lives, or about $2.3 million USD to save each person's life.

That's a lot of money per person. There are plenty of other places those resources could have been spent in order to save lives - including against other infectious diseases. It's reasonable to expect a nation to spend money on its own citizens first, so I won't invoke mosquito bed nets here, but what about smoking cessation campaigns? Or programs for the homeless? Maybe I'm not creative enough to figure out how this could be used to prevent more suicides, compulsive gambling, domestic abuse, etc.

I think the counterfactual isn't 'take a bunch of money from society in order to solve problems X, Y, and Z' since Sweden didn't do all that instead of responding to COVID-19 the same way their neighbors did. The point of this exercise is to ask whether we would spend similar sums of money to help prevent other people from dying sooner, and the answer to that is clearly not.

Of all the ways this money could have been spent, though, it was spent in a way that clearly transfers wealth and well-being from the younger generations to the older generations. I've watched the "Okay boomer" trend becoming more popular with the rising generation. The meme is entirely disrespectful in a way that previous generations wouldn't dream of treating their elders. But it's not just a dismissal of what the older generation has to say, it's also an indictment of the state of the world that generation left for the rising generation.

I remember many years ago my father told me, "your generation is going to be very upset when they realize what my generation has done to you." Before COVID-19 came along, I think that vision was beginning to be realized. I'm concerned that the current crisis will only intensify and deepen the divide and distrust between the generations.

What should the lesson be for the other Nordic countries and everyone else who paid sky high prices per life saved? Should they be ridiculed and mocked for taking such drastic and unprecedented steps? I think that's uncharitable and unwarranted. As I've maintained throughout the crisis, there's a high degree of uncertainty in situations like this one, and many countries made the decisions they did based on the data they had available to them at the time. In fact, without them trying something different we wouldn't have any idea how costly our interventions would be for each life saved. Much of our prior experiences with true pandemics predate modern economies.

Yes, a lot of people lost their livelihoods as a consequence of their approach, and are now struggling to feed their children. Yes, this will likely lead to increased inter-generational strife. But in the absence of a validated model - and in the context of the uncertainty present at the time they made their decisions - they couldn't have known exactly how many lives they were saving. If the death toll were an order of magnitude - or more! - higher the costs would be much easier to justify.

Comments

  1. My understanding is that the only way fewer people die in the other Nordic countries (on a per capita basis) is if they can keep their measures up long enough that some sort of medical intervention becomes possible. Either we get way better at knowing how to treat it or we get a vaccine. Or if their hospitals don't become overwhelmed but Sweden's do and people die in Sweden who wouldn't have died with adequate care. Did you come across any evidence on what's happening in Swedish hospitals?

    Absent a vaccine or a really good method for treatment that significantly lowers the fatality rate, we have to get to herd immunity, which requires a certain percentage of people to get infected and out of that a certain percentage are going to die and those numbers are going to be roughly the same for everyone.

    ReplyDelete
    Replies
    1. I'm not confident in a vaccine. I know there are a few different approaches they're taking, but I have little confidence they'll have an EFFECTIVE vaccine available within 3 years - which would still be quick. We'd have to be really lucky to get something sooner than that, so a public health strategy shouldn't rely on getting one sooner than later. Herd immunity is great and all, but the only justification for a 'flatten the curve' approach that relies on herd immunity (not fewer total infections) it would be if we prevent additional deaths from health care service overload. I have no idea how to measure THAT, though. Presumably, if you can scale out certain COVID-19 related services you'd be able to relax all those measures and let people infected, secure in the knowledge you can take care of them.

      All of that is assuming everyone eventually gets the virus. It comes in part from the 'flatten the curve' graph that has been making its way around, where if you integrate the area under the two curves (on a graph with no labeled axes) you get the same area in both scenarios.

      In order for that to happen, the other Nordic countries would have to plateau for many months, while Sweden gets a sharp spike that then dives down quickly. For example, if Norway goes the way of South Korea, Sweden would have to have already peaked and be on the decline to reach parity in total deaths. So far the evidence doesn't support that model.

      It looks instead like some countries hit a lower peak than others and the plateau isn't significantly different for each country. That implies the social distancing etc. is reducing the replication rate of the virus below 1.0, meaning that not everyone will eventually get the virus. As fewer and fewer new people get infected, the virus eventually gets isolated in the last infected individuals and doesn't find anyone new to infect. At that point, you still have a large susceptible population but no virus to infect them.

      Your biggest risk, then, is people coming from outside the country to begin the cycle again. This is why people are talking about keeping at least some social distancing measures going for not just months, but potentially years. If Sweden has more infected people, Norway needs to keep their measures in place longer to ensure the new infections that come to them (and they will come) hit in a country with a replication rate < 1.0. If they get five new cases in a month and R=0.5, then they end up with less than 10 new cases.

      (Note that R is different from R0. R is the actual rate of replication in a population based on factors like amount of social contact, hygiene, etc. R0 doesn't change unless the virus changes, but R is entirely susceptible to human behavior and public policies.)

      Delete

Post a Comment

Popular posts from this blog

Nutrition Graph: Cereal Comparisons

Covid-19: Epidemiology is useful

Open Questions Meet Settled Science