Newsweek published an article that was titled:
Yeah, first of all, the headline ....
1. “Rest of Europe Suffers Spike” — that’s debatable.
2. “Never had Lockdown” —wrong. Sweden entered it’s second lockdown on June 19th, and if they hadn’t, they would go full Florida right now.
I don’t see any “spiking” of new cases in “the rest of Europe”. A comparison of new cases and new deaths in the following graph and a comparison of excess mortality throughout Europe shows that in all European countries the mortality is well below the spikes seen ealier this year. Especially new deaths, — the only viable statistic we can count on, see explanation below —are far from the numbers we saw this spring.
But still — even if the rest of Europe isn’t spiking: the second part of the headline claims that Sweden is running “without lockdown”, yet still manages to keep the numbers down.
How did a country without lockdown manage to reduce infections? The answer is simple: They didn’t.
No, of course, they did reduce cases and deaths. But they didn’t not lockdown.
2. How Lockdown is Sweden?
Here’s their mobility data: Since June 19th — this is as far as Google allows me to look — every metric of importance in Stockholm declined.
throughout the whole country, people were either at home or in parks. -59% for workplace mobility and -27% for transit stations is a lot.
If you use a moving seven day average, workplace mobility is -45% right now. That the same number they had at peak Covid-19 restrictions around April 20th and just 10 percentage points below the lowest number Germany had throughout the year.
The decline in workplace mobility, at transit stations and the increase in parks suggests that something lockdown-like happened around June 19th. We know this because the curves showed such a behaviour of and only if a lockdown was in place.
If it looks like a lockdown, it’s made of a lockdown
Mobility data is important.
The European Commission’s science and knowledge service says in their new report: “the reduction of mobility is strongly correlated to the decrease of Rt during the outset of the epidemic.” The same publication also conducted an analysis of the effect of confinement measures on the reproductive factor R. Interestingly: After the public was educated, wore masks and track & trace was implemented, most countries could scale back the majority of measures without an increase in R.
Here’s how Italy released restrictions
Now, some measures could be more effective than others. In order to find out, we need to analyse which change in a given mobility indicator could be explained by the introduction of the respective confinement measures. Unsurprisingly, the EU Commission pays statisticians to do just that.
In short: Closing schools did more than closing businesses, and both were only rivalled by “restrictions on internal movements”, i.e. regional borders.
Now, first of all: Sweden did lock something down, and second of all: we know by inferring from statistical cadabra that closing schools and businesses is the most effective means of lowering R. It would be really nice to have proof here — best would be a country with a strongly increasing caseload whose only response is closing schools and businesses.
Thankfully, Sweden did just that and saved the (incredibly long Swedish summer) day. The situation was not good at all: The first Swedish lockdown was just enough to keep the hospitals working (see a little further down), but it wasn’t enough to supress the pandemic.
Before their second lockdown on June 19th, they were going down in a similar way as the US did a little later. But unlike the States, Sweden locked down again, and it worked. They closed all schools for the Swedish summer break of eight weeks and the numbers went down. To be very clear: They needed to — Swedes were about to see a substantial increase in mortality, changing their reality from “uncomfortable outsider” to the “Florida Man of Europe”.
Sweden was about to become the “Florida Man of Europe”
If you wouldn’t know where to look, you wouldn’t even be able to distuingish the imminent increase of mortality from statistical noise: It’s because they were the first country on a downward trajectory to impose a lockdown just *before* mortality significantly went up again. Now, their numbers are describing a curve similar to that of other European lockdown countries, albeit a little behind.
They did something that worked, and we can prove it.
This allows us to draw conclusions for other countries as well. Here’s what is happening in a little more detail.
- Why look at excess mortality?
- Why Swedens strategy until now was adequate
- Why I still believe they shouldn’t have done it
- Why I believe that schools should remain closed
1. Excess mortality: A statistic I can believe in
Generally, when you look at case fatality rates (deaths per infections), you have a huge problem since the base number (confirmed infections) isn’t reliable at all. Different countries and different times mean different testing regimes that keep changing all the time. Consequently, in order to draw meaningful conclusions, I can only compare the change of the case fatality rate over time in a certain area if I can ensure that the testing regime stayed roughly the same. To make things worse: How the cause of death is labelled by the authorities is beyond my knowledge. The numbers can be inflated or deflated even without intent, possibly varying within a single hospital.
That being said: We still want to know if Covid-19 is more deadly in Sweden, right?
I think we have only one reliable statistic here: Dead people. They don’t lie.
That’s why I like to look at excess mortality: Given the numbers of the previous years, how many people do we expect to die in this week of the year and how many did actually pass away? The deviation creates the so called z-score* which allows to compare excess mortality of different populations over different time frames, i.e. summer and winter are de-trended as well.
In this, we can see that Sweden is clearly worse off than the other Nordic countries and Germany, clearly better off than the other rainy coast countries like UK, Belgium and the Netherlands and than the sunny coast countries like Spain or France (I know — the Normandy has a lot of rain as well. Sorry, I’m just an optimistic person). Sweden is roughly on par with Italy, Scotland, and a little better than Northern Ireland and Portugal.
Poland and Spain are the only countries showing first signs of spiking, Austria, Hungary, Germany and the Nordics except Sweden are doing extremely well.
Here’s a different source breaking it down along Covid-19 deaths and “other” excess causes that might be indirectly caused by the pandemic:
We see that there are almost no unexplained deaths in Sweden, but the comparison to Denmark and Norways suggests that there some unneccessary deaths if Sweden had done like the Danish.
This graph is using a slightly different scale, and it shows substantial unexplained deaths in Spain, Italy, and the Netherlands.
This one for some US States (has yet a different scale)
2. The Swedish strategy was adequate
Given all that, it’s absolutely OK to claim that the Swedish “lockdown by recommendation” strategy was adequate. For four reasons:
- Stockholm recommends, the Swedes follow
The Swedish largely followed the ‘recommendations’ of their government more disciplined than, say, the Spaniards the ‘orders’ of theirs. This became pretty obvious in those regions of Gotland and Stockholm with large immigrant communities: They treated the recommendations as actual recommendations and suffered heavily. The Swedish were so good at their voluntary lockdown that the economic fallout equalled the one in Denmark, which had an actual lockdown.
- Swedes live pre-distanced by default
Ca. 56% of all Swedes live in single households, and only a very low percentage in multi-generational homes. In Italy, those numbers are reversed (if my memory serves — I’d have to look those numbers up). I.e. Sweden can afford the occasional restaurant visitor be infected, because they don’t come home to granma and granpa. Intergenerational spreading outside nursing homes is a lot less likely, and not visiting those was ‘strongly recommended’ . The Swedish recommendation strategy took into account that Swedes live in that permanent state of social distancing throughout the year, with or without Covid-19.
- Hospitals stayed operational
Any sensible national Covid-19 strategy needs to focus first and foremost on hospitals not getting overwhelmed. Why? Because overwhelmed hospitals are tha one thing that is worse than a lockdown, by several orders of magnitude. The reason: When people start to believe that their lives won’t be saved if they have a severe health problem or become unlucky — an accident, an allergic shock or a food poisoning — any economic activity in a society comes to a grinding halt and anyone who can afford it stops doing anything while the poor are scrambling to survive in an accelerating downward spiral with increasing risk of death.
We’re talking Sweden here, so thihe following risk is not a real danger — but situations like those potentially breed revolutions. All you need are a few skilled leaders. If you add hunger and deny a silver lining on the horizon, you can replace ‘potentially’ with ‘almost certainly breed revolutions’. However, Sweden seems to have succeeded in keeping their hospitals functional.
> How do we know whether hospitals are failing?
There’s one way of telling: If hospitals are overwhelmed, we would expect an excess mortality among younger age groups. If the health care system keeps working, even in highly affected regions, the mortality of younger age groups remains largely unaffected. If hospitals are completely in ruins, Covid-19 cases that don’t have to be lethal turn out to be deadly, and other conditions that are generally treatable for younger people don’t get treated as needed for survival.
Here’s a breakdown of excess mortality measued by z-score throughout Europe according to age:
Unsurprisingly, we see that there’s a difference for different age groups. Plus: There’s a huge difference to other years. Here’s a comparison of 2020 with the preceding five years: Usually, mortality rises around new yeras, only in 2018 a deadly flu pandemic (yellow curve) shook Europe.
Here’s another way of comparing the impact of Covid-19 to seasonal flu seasons:
If we break down numbers according to countries, we can see which countries are responsible for the European spike seen among the 15–44 years old.
Hospitals in England failed so badly, they’re the only European country where the 14 to 44 years old suffered significant excess mortality
England alone is responsible for the staistically significant rise of mortality among the 14 to 44 years old. In the next age group, more countries were affected: England, Spain, France and Belgium suffered excess deaths, with Italy just above the dotted red line, while Sweden stayed just below that line.
Conclusion: Public health institutions continued to serve Swedes
4. Stockholm did it’s ethical duty.
It’s absolutely OK for a government to leave personal health prevention in the hands of their citizens, provided they are able to make an informed decision. It’s not OK for a government to let a pandemic overwhelm the healthcare system and push the society into an abyss, especially if it’s preventable. This includes mandatory measures to keep the healthcare system functional. But that wasn’t needed in Sweden.
Sweden seems to have found the exact amount of political measures to fullfill its protective duties towards its citizens.
This graph comparing excess mortalities shows that Sweden is on the same level as Switzerland, the US (first wave), Netherlands and France. Of course, they’re not performing like Norway, Denmark or Germany:
This shows that finding just the adequate amount of anti-Covid-19 measures is a feat other countries have achieved as well. But: The Swedish strategy works in Sweden and might already fail in jolly Denmark — given their obedience to recommendations and their pre-distanced society. Thus, the Swedish data does not allow us to draw direct conclusions for other countries — simply because we don’t know which factors matter and have to be taken into account and which don’t.
Sweden acted adequate, but not well
However, although Sweden did OK enough, in my personal opinion, Sweden should have locked it down like the rest of Europe. The reasons:
- The economic harm was done regardless (See comparison with Denmark)
- The disease lingers around, putting a strain on psychological wellbeing and limiting personal freedem, for example to travel.
- People died that didn’t have to, and there was nothing Sweden gained in return. Because
- herd immunity will be only be achievable long after the vaccine will be deployed (3+ years) and causes much harm in the meantime.
- If a big and connected country like Sweden is a constant source for new infections, neighbouring countries are constantly at risk.
In the graph below we see that Sweden behaved just like the US did, with a slowly exponentially rising curve. Until July 3rd.
In this graph below, we see the Swedish yellow curve bend and leave the US American trajectory around July 3rd.
Why are cases in Sweden declining since July 3rd?
We see that the cases started declining around July 3rd. Since there’s usually a statistical delay of 14 days until any anti-pandemic measure shows up in the data, we shall look at what happend on June 19th. Well, around midsummer, the Swedish summer break of eight weeks without school started.
If anything, than this plummeting is an incredibly strong case for closing schools as an effective means of stopping the spread of Covid-19.
Now I wonder what governments will make of this. But we have to be careful: It’s not necessarily true that the reverse of this proof is true. Opening schools in, say, Denmark, Germany or the United States might just do nothing. Or it might:
The Swedish and the Israeli example show that we should at least be cautious and maybe try opening a test region first and wait a month. However, in my personal opinion:
I believe that reopening schools will be the single most stupid political decision any world leader can make right now.
If the US open their schools, it will be equivalent to pouring gasoline into an oil fire. And if I were deciding Covid-19 policy in Stockholm, the schools would not reopen anytime soon.
Cheers and stay healthy :)