While the rank-and-file righteous were posting on Facebook their hopes for continued exponential growth through November in American deaths, journalists were publishing their hopes for dramatic near-term exponential growth in Swedish deaths. As infidels against the religion of “saving lives” via social and economic shutdown, the Swedes deserve to experience the full wrath of the coronavirus god:
- “After Rejecting A Coronavirus Lockdown, Sweden Sees Rise In Deaths” (Forbes, 3 days ago)
- “Sweden Continues With Controversial Coronavirus Strategy: Is It A Big Mistake?” (Forbes, 2 days ago)
- “Sweden Girds for Thousands of Deaths Amid Laxer Virus Policy” (Bloomberg, April 4)
After 8 days of unimpeded exponential growth since that April 4 article, how many thousands of deaths did 10 million Swedes experience in the last 24-hour period, according to the WHO? 17. Seventeen thousand? No: 17. Locked-down Denmark, with a population size half of Sweden’s, had 13.
My latest update to “Number of new COVID-19 cases worldwide is declining now?”:
4/12 comment: Sweden has 466 new cases; locked-down Denmark has 177. Sweden suffered 17 deaths; Denmark 13. I.e., adjusted for its 2X population, wide-open Sweden has a higher rate of new cases and a lower rate of deaths. Massachusetts has gone off the rails compared to Sweden. With 2,615 new cases, Massachusetts has 8X the new case rate of Sweden’s. With 70 deaths, locked-down Massachusetts has 6X the death rate of Sweden.
In other words, the exponential doom forecast for schools-/restaurants-open Sweden has actually happened, but to schools-/restaurants-/offices-closed Massachusetts. The University of Washington prophets are still showing that Massachusetts will suffer a higher death rate, adjusted for population, than New York (though Sweden’s death rate will be yet worse, say these lockdown advocates, roughly twice as high as New York’s).
Graphic from today’s NYT:
Separately, I still think it is an open question as to whether the number of worldwide new cases of COVID-19 is declining. Testing continues to ramp up, but the number of new cases is only 17 percent higher than it was on April 1.
Certainly the Swedish numbers will not convince Americans who bought into the shutdown religion, any more than a high death rate during the Black Plague sowed doubt in the minds of the Flagellants (if half the people in a town were killed, the Flagellants would say “if we hadn’t whipped ourselves, the death rate would have been nearly 100 percent”!).
[Separately, a friend asked privately “What is your ultimate Sweden prediction? It is nice we actually have a control that rational people can look on.” So that I can’t go back on this and deny him the opportunity to ridicule me later, I will put my answer here. Massachusetts and Sweden are fairly comparable in terms of percentage of the population that live in an urban area. Sweden’s population has a slightly older median age, but, perhaps healthier so that’s a wash. My prediction is based on the Massachusetts “curve flattening” efforts, including shutting down schools, restaurants, and offices, saving zero lives compared to Sweden’s shutting down mass gatherings. Neither Sweden nor Massachusetts will run out of health system capacity and therefore every patient will get whatever medical interventions are available. Sweden spends 11 percent of a lower per capita GDP on health care versus 17 percent here in the U.S., but our incompetence could easily soak up the difference in spending and therefore the systems should be equally effective (actually equally ineffective would be more accurate right now due to the lack of any drug or procedure that seriously impedes the virus). The American system will not let anyone die, regardless of how many organ systems are destroyed, so long as Medicare can still be billed, and this will be the basis of a differential. My prediction is that Sweden’s death rate from COVID-19 for all of 2020 will be 1.15X the Massachusetts rate.
(Where are they right now? Sweden has 887 deaths, or 90 per million inhabitants. Massachusetts has suffered 756 deaths so far, roughly 110 per million.)
As long as we’re making predictions, what about the total for Massachusetts and Sweden? The University of Washington right now says that doom is in store for Sweden. They’ll have 13,259 deaths through August 4, 2020. They’ll have 79 ICU beds and need 3,378(!).
Shut-down Massachusetts is forecast to suffer 6,739 deaths through August 4, 2020, out of a total U.S. death toll of 61.545. I.e., with 2% of the nation’s population, we will have 10% of the COVID-19 fatalities. This seems like too large a share, even for a population that spends most of its time reflecting with satisfaction on how stupid people in the south and Midwest are (imagine if we’d put that effort into stockpiling masks and coming up with a plan for a flu-like epidemic!). So let’s cut this back to 7% of the U.S. fatalities: 4,300 (if we believe the 61,545 prophecy). This implies that Sweden will lose 5,000 citizens by August 4, 2020. I have much less confidence in these numbers than in the idea that the death rates in Sweden and Massachusetts will be within 20% of each other.]
Update 4/14: The latest report for Sweden: 465 new cases and 20 deaths (Denmark was at 144 and 12). Massachusetts had 1,296 new cases and 113 deaths (8X the death rate).
Related:
- https://philip.greenspun.com/blog/2020/04/12/google-is-monitoring-our-social-distancing-failures/ (the Swedes are not, as some people have suggested, hiding in their apartments)
- urbanization in Sweden versus the U.S. (a higher percentage of Swedes live in cities; it is not that Swedes are isolated in their ice fishing huts)
- obesity in Sweden: “Around 50 % of adults in Sweden are overweight or obese.” (i.e., Swedes are not immune to coronavirus because they’re slender)
- countries ranked by smoking rate (the U.S. and Sweden are similar)
- “As the rest of Europe lives under lockdown, Sweden keeps calm and carries on” (Guardian, March 28), in which Sweden’s top epidemiologist explains his rationale
- An attempt to calculate Rt, the effective reproduction number, in each state: Which states have the epidemic least “under control?” To answer this, I plotted states where the best case (eg. low end of HDI) is above 1.0, indicating the true value of Rt is almost certainly above 1.0. Surprisingly, Rhode Island, Maryland and Massachusetts sit at the top. Part of this may be that Rhode Island is earlier in its infection curve, but seeing large states like Massachusetts and Texas above 1.0 is worrisome —especially because none of these states have hit the headlines as being trouble spots.
Huh? According to https://coronavirus.jhu.edu/map.html as of 4/12, Sweden has almost 10,500 cases and just under 900 deaths. And, on a per captia basis, has diverged from its Nordic neighbors that have chosen to lockdown.
js290, the worldometers.info dashboard has Sweden at 1038 cases/million, Denmark at 1066 cases/million, and Norway at 1204 cases/million – where’s the divergence?
Sweden’s chief epidemiologist has always said there were plans for a lockdown, but only when necessary, rather than jumping straight to a stage that can’t be maintained for very long.
js290: I don’t think the JHU page you cite is inconsistent with the original post (WHO data). I wrote “Sweden has 887 deaths”. You write “Sweden has … just under 900 deaths.” If you are willing to accept that 887 is “just under 900” there is no inconsistency. As for the comparison with other Nordic countries, well, the post compares Sweden to Massachusetts (I hope you will forgive my parochial concern!).
@Colin Suttie The divergence is that number of cases is very unreliable. It depends on how much countries test. For example, Norway has been testing at a much higher rate (more than double the rate of South Korea). Deaths are more reliable, although still prone to uncertainties in classifying what is exactly a COVID-19 death. And in deaths per capita, Sweden has 4x more than Norway and 9x more than Finland.
As an aside, I should say that while officially not in lockdown, many Swedes have chosen to voluntarily stay at home (and in fact the high schools and universities are closed). So it can’t be exactly compared with life as normal.
Why haven’t you done an analogous comparison between Sweden, Norway, Denmark and Finland?
The continuously updated comparison to Denmark is in the linked-to https://philip.greenspun.com/blog/2020/03/26/number-of-new-covid-19-cases-worldwide-is-declining-now/
From a purely parochial point of view, I am less interested in policy and outcomes in Denmark than I am in what happens here in Massachusetts!
OK, I’ll do it for you:
Sweden 919 deaths (87.83 deaths per million)
Denmark 285 deaths (44.89 deaths per million)
Norway 131 deaths (18.08 deaths per million)
Finland 59 deaths (8.84 deaths per million)
Source: Oxford University (April 13, 2020)
https://ourworldindata.org/grapher/total-covid-deaths-per-million
From ourworldindata’s charts, Sweden appears to be nearer its peak than the other Scandinavian countries. That might be consistent with Knut Wittkowski’s assessment that it’s better to avoid lockdown and let the not very severe epidemic come and go.
That is a pretty selective view of the data. You chose one 24h period when the deaths in Sweden were relatively low to say there were much better than Denmark, when in fact they have a per-capita death rate about twice as large as Denmark. A more apt comparison would be with Norway (~4x less deaths per capita) or Finland (~9 times less deaths per capita), which have a population density (17/km2 and 16/km2) much closer to Sweden (23/km2) than Denmark (135/km2). Therefore, the data show a clear effect of lockdown. Whether or not lockdowns should be considered in light of economic/social harm, is a different question.
Jono: What is the relevance of population density if a country’s population is 87% urban? Isn’t that like saying people in London had a lower chance of getting sick after the British Empire acquired Australia? Or New Yorkers experienced a sense of new spaciousness after the U.S. purchased Alaska?
As noted above, it is interesting to me what happens in Sweden v. Denmark and I have been tracking it in https://philip.greenspun.com/blog/2020/03/26/number-of-new-covid-19-cases-worldwide-is-declining-now/ , but I am much more interested in what happens here in Massachusetts. If our lockdown/shutdown does not result in any reduction of cases/deaths compared to Sweden, that is poor argument for us having closed the schools, restaurants, and offices.
(The “pretty selective view of the data” goes away if you look at the linked post. There is a comment for each day since I started tracking. So every 24-hour period is covered! The 4/12 comment is, of course, selective regarding data released on 4/12, just as the 4/11 comment is regarding data released on 4/11.)
@philg You have a point that population density is not the best metric. What is more relevant is the “population-weighted density”, which takes into account what is the typical density where people most people live. I find comparisons with Australia are not apt, since both Sweden, Norway, and Finland have vast areas where a non-negligible fraction of the population lives, in contrast with Denmark, which is a much more compact country. If you have a look at e.g. https://theconversation.com/think-your-country-is-crowded-these-maps-reveal-the-truth-about-population-density-across-europe-90345 , you will see in the first figure that the red population dots are much more closely spaced in Denmark than in Sweden (whose pattern is closer to Finland or Norway). In that author’s estimation (these things vary a bit), the population-weighted density metric in 183 in Denmark, 89 in Norway, 84 in Sweden, and 53 in Finland. Which again is evidence that they way most people live in Sweden is closer to Norway or Finland than Denmark, and may reflect on the likelihood of SARS-Cov2 contagion.
I understand your focus on Massachusetts. But I live in Sweden, so I find it more interesting what is going on here. And I feel it is really misleading to say Sweden is doing about as well as Denmark, when the data show one of the clearest examples of the opposite. I saw your linked post but don’t really see the relevance. Some days you compare new cases, others new deaths. New cases are very dependent on how much a country tests, and Sweden is testing a lot less than Norway or Denmark. You could have also mentioned April 7, when Sweden recorded 114 deaths, Denmark 16, and Norway 12. By April 7, those 114 deaths in Sweden were more than all of Norway’s COVID-19 deaths to that date. See how easy it is to be selective with data? The fact is that Sweden’s death totals are at the moment more than twice those on Norway + Denmark, which together have about as many people. If you compare Sweden (89 deaths/million) with its closest country, Norway (24 deaths/million), then the effects of non-lockdown vs lockdown are even more overwhelming.
Never make a prediction without confidence interval!
SK: That’s sort of at the end! “Within 20% of each other”! It wouldn’t help to say “I am 95% confident that they will be within 20% of each other,” would it? We have only one trial (I hope).
The difference between the public policy in Sweden and other European countries is probably not as high as described in the press based on the reading of regulations.
Essentially they do have a lockdown (people working from home, restaurants empty, no travel), it’s just voluntary and somewhat more porous than the partially enforced lockdown in Germany (note in most parts of Germany there were literally no fines for breaking the lockdown regulations until about a week ago). I’ve heard anecdotally there are fewer people on the streets of Stockholm than Berlin.
So in my book a positive development there will be a point against enforced lockdowns (agreeing with you on this) but not against the lockdowns in general (disagreeing with you on this).
ilya: I’m going to bet that you don’t have children who are K-12 age! Otherwise you would not say that there is minimal difference between children going to school (Sweden) and children not going to school (Germany, Massachusetts, et al.)!
@philg I don’t have children, but the picture I had in my mind was that most of the schools in Sweden have voluntarily (for the schools, not for parents, that is) closed. From the sources, though, it’s “all” high schools and universities but “some” primary schools; I couldn’t find detailed statistics, but yes, I was probably wrong.
Note Denmark plans to go into the same “school closure” state (primary schools open, high schools not) in two days according to https://www.schengenvisainfo.com/news/covid-19-eu-commission-suggests-gradual-return-to-normalcy/ and from what I know Germany is planning something similar within a month.
ilya: Also think about this from a generational equity point of view. Children are not at risk from coronavirus. School is considered so beneficial for children that a parent who denies a child the opportunity to go to school (or who allows an older child to skip school) can be imprisoned (as with everything else in the U.S., what we enjoy most is imprisoning poor people: https://newrepublic.com/article/121186/truancy-laws-unfairly-attack-poor-children-and-parents ). Now we are saying “If it can prevent even one adult from dying we should keep these children out of school for months or years because how could we think about putting a pricetag on human life.” At my relatively advanced age of 56, of course, I remain passionate about living forever. I can vote and children can’t so I will vote to keep them out of school until a cure for COVID-19 is developed, even if it takes until 2030. Oh yes, I also don’t want to die from flu so I will vote to keep children from going to school and spreading flu until a flu vaccine is developed that is 100% effective (rather than the current 10-50%, supposedly, depending on the season).
These votes to save my own skin I will characterize as selfless and virtuous! #SaveLives #CloseSchools #StayAtHome
Very fair points, and the approach of Sweden and now Denmark seems to be more and more validated.
For the same reasons there is talk in Germany about opening the primary schools next week. Note that (because we are supposed to be in the middle of long school holidays anyway) if that happens only 3 weeks of the school year will have been lost.
College Fix which I don’t normally read, has a bunch of articles from ‘credentialed scientist types’ saying the lockdown is stupid. Here is one https://www.thecollegefix.com/epidemiologist-coronavirus-could-be-exterminated-if-lockdowns-were-lifted/
Coincidence theory is Fauci(and his buddy Gates) don’t want herd immunity, they want social control, and vaccines of death for all.
Has any disease been eliminated without vaccination?
I should have written “disease caused by a virus”.
If the U.S. federal government is paying for all hospitalizations “from COVID-19,” I predict that that hospitalizations and deaths “from COVID-19” will be very high. According, I predict that Massachusetts will be much worse hit “by COVID-19” than is Sweden. Have you taken this factor into account in making your predictions above? Relative to baseline death rates, the difference is likely to be much less.
I did not factor that in, Bill. So probably I will get this wrong and be buying you lunch any time that you come to plague-central Massachusetts!
The latest WHO data are available (for today, Monday, April 13, 2020). Sweden suffered 12 deaths; Denmark suffered 13. (i.e., not very different from Sunday’s numbers). https://www.who.int/docs/default-source/coronaviruse/situation-reports/20200413-sitrep-84-covid-19.pdf?sfvrsn=44f511ab_2
(Both today and yesterday were holidays in Sweden and Denmark, so that might result in some missing data. Separately, when exactly do Europeans actually work?)
The latest Massachusetts data: 88 deaths, 1,392 cases. More deaths than Sweden and Denmark combined. More new cases than Sweden and Denmark combined. A death rate that is 10X the rate in Sweden, adjusted for population. https://www.mass.gov/doc/covid-19-cases-in-massachusetts-as-of-april-13-2020/download
The Systrom study trying to estimate Rt seems deeply flawed. He hasn’t plugged the numbers from Italy into his model, and I don’t think is model would predict Italy’s turning the corner.
There is very interesting retrospective data from Italy showing that the number of symptom onsets started dropping almost instantly when the lockdown started. But the reported confirmed cases and deaths and hospitalizations kept getting worse looking for 2 or 3 weeks. I would be very surprised if anyone could predict the reduction in true cases by looking only at reported cases (without knowing, for example, when the lockdown happened) https://www.epicentro.iss.it/en/coronavirus/bollettino/Infografica_13aprile%20ENG.pdf
And I’ve been ranting about the IHME model at UW. It’s terrible. Or at least it was terrible last week. It didn’t even predict the present or yesterday right (look at the number of beds is says we need in MA right (but they seem to have gotten this error under control.) I think its future predictions for MA are still way off: I suspect we’re already past peak new cases and peak deaths, but their model really wants the peak to be near May 1.