Let’s have a look at the year-to-date COVID-situation in Sweden, allegedly “The World’s Cautionary Tale” when it comes to COVID.
Actually, most of what follows is focused on All Cause Mortality, that is, deaths regardless of cause of deaths, that is, *all* deaths, not just those attributed to COVID.
Reason for this focus on all cause mortality is that Public Health >> COVID, that is, in my not so humble opinion, if you want a metric on Public Health for a nation, region or town, in order to intervene in that system, you simply can’t just focus on a single parameter or disease, you will want to look at the bigger picture. Why…? Simply because what we are dealing with here – society at large – is a great example of a COMPLEX SYSTEM, and any attempts to optimize a complex system by one single parameter, will inevitably lead to disaster or chaos.
Unfortunately, this path to disaster is what most of the world has now chosen, by placing almost all focus on COVID, while disregarding the multitude of other system parameters that clearly are going to be impacted by the COVID interventions, in ways that we are nowhere near understanding, and with likely huge consequences for a very long time.
In short, all cause deaths is IMO a good proxy for overall Public Health, and that’s what I’ve been using now for more than 6 months in order to build my own opinion on how serious this pandemic is. Or isn’t.
But let’s start by having a look at COVID-numbers, since these numbers are what’s been driving the panic, as well as Sweden’s reputation of being “The World’s Cautionary Tale”: below chart shows a normalized plot of COVID “cases” – that is, positive tests, and COVID-associated deaths. The data have been normalized so that “cases” as well covid-attributed deaths are scaled by the currently largest value for respective parameter.
As we can see from the plot above, “cases” are indeed going thru the roof now! The passed week Sweden had 4 times as many “cases” as during any week of the spring pandemic, so that does indeed look scary.
However, Covid-attributed deaths are now, during autumn, nowhere near where they were back in spring. So, despite the fact that “cases” (better described as number of positive tests) are growing “exponentially”, Covid-attributed deaths are not. So, by looking at the number of “cases”, it indeed looks like Sweden is having a terrifying “Second Wave“. And since Sweden per million has more Covid-attributed deaths than e.g. the rest of the Nordic Countries, case is closed, right…? Sweden simply must be the “World’s Cautionary Tale”…
Not so fast. Let’s have a look at overall mortality, that is, looking not only at COVID-attributed deaths, but deaths by any cause, for the months of January – October:
The graph above presents absolute (non-population adjusted) monthly all cause deaths. A couple of interesting observations:
– January and February of 2020 had significantly lower number of deaths that the baseline 2015-2019
– April, May and June 2020 were significantly above the baseline, of which April really had exceptionally high number of deaths.
– July, August, September and October 2020 were below baseline.
So April 2020 looks really bad. But how bad was it really, in a more historical context…? Let’s have a look.
But first, let’s have a look at how population size has evolved in Sweden over the past 160 years:
Since 1860, population has almost tripled, and since 2000, the growth has been exceptional, if not even exponential… So, we must be a bit careful not to do as media typically does (click baiting), and even our own Government Agency of Statistics (SCB) did a few months ago in a press release, using non-population adjusted (absolute) number of deaths and claiming that “Deaths in Sweden have never been higher in 100+ years!”. Talk about spinning data to suit their story… !
But let’s pretend for a minute that we want to be alarmist, whether we are media doing click-baiting, or a confused politician wanting to make a compelling point supporting the only story we have, that COVID is indeed the modern version of the Spanish Flu, and sound the alarm for the “Extreme Number of Deaths 2020 in Sweden”. Then we might want to present a chart like the one below:
In the above graph, absolute deaths are stacked by month.
Wow! 2020 Jan-Oct had a record number of deaths, more deaths than ever since 1900 (at least)….!!!!
Indeed, if you look at the bar for October 2020, and compare it to the other ~120 years, it really looks like that the only previous year with about the same number of deaths as 2020 for the period of Jan-Oct was 1918, the year of the Spanish Flu. And of course this is the way to present the data, if you want to fuel the panic and hysteria, whether it’s media click-baiting, or politicians, academics or the pharmaceuticals pursuing their own agendas.
But no one even semi-intelligent person would present the data in this way, without taking the population growth into account. Still, media et al does.
In fact, it’s not sufficient even to adjust for population size, you’d also have to take population demographics into account for a meaningful comparison of deaths between years, and we’ll do that further down this post, but for now, let’s adjust the above numbers for population size and see what we get:
Now, when taking population size into account, it’s harder to spin an alarmistic story around the deaths in Sweden: all of sudden deaths 2020 in Sweden look… VERY NORMAL…!
A couple of illustrations regarding population demographics, and its importance to mortality, first two plots on how the age structure of the population has changed over the years:
From the above two graphs, we can see that the number of elderly people in Sweden has grown over the years, in absolute as well as in relative terms, which clearly is a factor that impacts mortality rates.
Next, another aspect of changes in demography, the ratio of foreign born of the population. The reason that this matters for mortality during the pandemic will become clear a bit further down: the graph below shows the rapid increase since 2000 of foreign born people living in Sweden:
To see that also this aspect of demographics, ratio of foreign born, has an impact on overall mortality, let’s next look at the week-by-week difference in mortality for foreign born vs native born, comparing 2020 to baseline 2015-2019:
First, observe that during “normal” years, such as the baseline 2015-2019, mortality for foreign born is significantly lower than for native born. This reflects that most foreign born in Sweden are young, typically male migrants in late teens or early 20:ies. Secondly, observe also that despite the foreign born being a younger cohort, they got hit even harder by Covid. Let’s look closer at how much harder, by looking at the week-by-week growth factors for respective cohort, still comparing 2020 vs baseline 2015-2019:
If we look at the peak of the pandemic, back in April, we can see that while deaths grew by an astonishing 40% for native born, for foreign born deaths grew twice as much! So, despite the foreign born being generally a younger population than the native born, they got unproportionally hard hit by Covid.
Now, let’s leave demographics and historical data behind us for the moment and return looking at more recent years. Below monthly population adjusted deaths 1990 – 2020, presenting the Top-20 deadliest months for the period:
Turns out April 2020 ranks as the 14:th deadliest month since 1990.
Let’s now look at the Top-20 Least deadly months for the same period, 1990-2020:
Turns out 4 months of 2020 make it to the Top-20 Least Deadly Months since 1990, with September 2020 hitting the podium at second place!
So, by now, with the data above, we should start questioning the story we’ve been fed by media, academia and politicians, the fear mongering and panic – is COVID really the contemporary version of the Bubolic Plaque or the Spanish Flu, that media et al tries so hard to make us believe…?
Let’s take a deeper look at this “second wave” of Covid, and it’s impact on all cause mortality. First, let’s look at a week-by-week summary of Covid “cases”, ICU’s and Deaths, in relation to all-cause deaths:
Let’s start by looking at the blue plot, showing Covid “cases”: we can indeed see truly exponential growth since 4 weeks back. Looking at the green and red plots, showing Covid ICU’s and Covid attributed deaths, we can also see that both are now, since two weeks, increasing, after having been flat for some 12 weeks. And if we look really carefully at the black plot, showing all cause deaths, we can almost see a tiny increase.
So, by looking at this chart, we might deduce that we are indeed in a terrible second wave, yes…?
Well, let’s not jump into any conclusions, let’s look again at all cause deaths, together with excess deaths, ad do so this time on a more granular scale, week-by-week:
From this chart, we can see that indeed, all cause deaths – the red plot, has slowly started to rise since a few weeks back. However, look at the red plot in relation to the shaded grey area, which shows the range of weekly deaths for the baseline period of 2015-2018: deaths during the past 10 weeks have been consistently below the grey area, meaning that during those 10 weeks, all cause deaths have been at record low levels. That low death rate extend in fact further back in time for yet another 8 weeks, where all cause deaths all the way back to late June have been at or below average, the latest 10 weeks exceptionally low below average.
By looking at the orange plot, we can see that standardized, pop.adjusted excess deaths are now approaching zero, that is, taking population size into account, there’s now very few excess deaths compared to baseline 2015-2018. The black plot shows the number of non-population adjusted, absolute excess deaths.
But yes, the red line, showing all cause deaths is going up. Isn’t that a reason to worry…?
Well, we are now well into the annual flu season, so let’s take a look at how all cause deaths behave during normal years, and compare that to current year:
In the above graph, we have all cause weekly deaths plotted in red for the 6 years from 2015 to 2020 plotted, for the period of Jan 1:st to November 3:d. The orange plot shows the 2015-2018 baseline. Notice again that for the past 12 or so weeks, deaths have been significantly under the average for the period, as mentioned above, the past 10 weeks at record lows.
We can see what we should expect by looking at the monthly all cause deaths:
Now, we have the deaths for the 6 years 2015-2020 plotted by month, for each year. Look at the orange line that shows the average behavior: each autumn, deaths are starting to climb, after a lull during summer. Flu Season has arrived!
Next, look again at 2020: July, August, September and October are all well below average, in fact, the last 3 months are below the range, that is, at record lows!
We can look at the same data with an other type of graph, “Nightingale Plots”:
So, what’s happening in Sweden…? Is the country “The World’s Cautionary Tale”, or is 2020 going to be a pretty normal year, wrt Public Health and all cause mortality…?
How are we to understand the seemingly contradictory facts that, on the one hand, Covid “cases” seem to be going thru the roof at the moment, while excess mortality is almost non-existent..?
During the spring outbreak, COVID was mainly driven by the frail and elderly. Younger, healthy people did not really register in the “cases”. But lets look at what that looks like now:
Above wee see that now, during autumn, with flu season upon us, Covid is driven by the young, in particular, the 20-29 year olds are now driving the “cases”, but extremely few of that cohort is impacted in any other way than having been positively tested – ICU’s in that age groups are barely visible, and Covid-deaths are not visible at all. Thus, during autumn/winter, Covid seems to act like the normal flu does during “normal” years : everybody gets it sooner or later, but very few people outside of the normal risk groups are hospitalized, not to mention at risk of death.
We can also look at the numbers relative to cohort size:
Now, we see that the top age group is still, the 90+, relative to the sizes of the age groups, as well as that very few people under 70 are severely impacted by Covid.
Let’s next look at what ratio of deaths the Covid-attributed deaths constitute:
Above we see that during the peak in April-June, Covid-attributed deaths constituted a significant part of all deaths, but then, during summer and early autumn, dropped very low. In October, at arrival of the annual flu season, deaths attributed to Covid seems to begin rising again, which would be expected if indeed Covid has replaced other seasonal flu’s this year.
I’ve already mentioned the importance of demographics above. So let’s look at 2020 YTD age adjusted mortality, and compare it to full year age adjusted mortality some previous years:
So, in the above graph the blue bars depict full year age adjusted mortality, and the orange bar shows 2020 age adjusted mortality year-to-date, including data up until November 5:th. The interesting question is how high the orange bar will reach by Dec 31:st…
I’ll end this post with a prediction on 2020 full year mortality:
“The World’s Cautionary Tale”….?