SCB.se just released their weekly preliminary numbers on all cause deaths, up until July 24th. Since the numbers are preliminary, by previous experience we know that the last data point, representing July 17-24, will very likely be revised, slightly upwards.
Let’s start by looking at a graph showing weekly deaths up until July 21st, compared to baseline of 2015-2019, and the cumulative number of Excess Deaths:
We can see a clear trend that deaths now, since beginning of July, are initially at, and for the past 3 weeks below average 2015-2019. Even when discounting the last data point, the trend for weekly deaths is clear – it’s now consistently below average for the time of year. And the cumulative number of Excess Deaths is clearly going down since 3 weeks back.
Next, let’s look at deaths per million year to date (2020-07-24) over the past 30 years:
Despite the high number of deaths in April 2020, the year 2020 Y2D looks quite… “normal” thus far, even considering that the bar for July thus far only represents 3 of 4 weeks. Even if the bar for July would double during the last week of July, the sum total will not be anything remarkable – worst case about the same as 2012 (you all surely recall the enormous disaster of deaths we had back then, don’t you….? – No…? Me neither…)
Next, let’s make some predictions for the deaths full year 2020. Here, I’m assuming that the deaths rest of the year grows as they have done Jan 1st – July 17, that is, I’ve knocked off the last 7 days of uncertain data, and assume that the number of deaths for the reminder of the year will be proportionally the same as from Jan 1st to July 17th, which IMO is a *very* unlikely scenario, but can serve as a “worst case scenario”:
First, let’s look at the absolute, non-adjusted numbers, and compare to 2001-2019:
Under the “worst case scenario” assumption, it indeed looks like deaths 2020 will be “all time high” for 2001-2020 in terms of absolute number of deaths. However, over that period, there has been massive population growth as well as a significant change in age demographics, so let’s see what the data looks like when conditioning on those, let’s first condition on age demographics:
Now (and remember, we are still dealing with the “worst case scenario”) things look a bit different: The total number of deaths, age adjusted, is not very different from the past 10 or so years, except when compared to 2019, a year that had the lowest total death toll since 1990 (at least, I haven’t bothered looking further back).
We have one more thing to consider, and that is the continuous population growth, so let’s look at the data, now adjusted both for age as well as population size, in terms of deaths per million:
So there we have it, the “Swedish Covid-19 disaster”, under the “worst case scenario… about the same death rate as 2012 – 2013..
I’d suggest media outlets such as New York Times, USAToday and others, as well as WHO to pay us a visit to see the “disaster” live… I’m sure you’d get surprised…
[Graphs with better resolution to be found here ]
systems perestroika - éminence grise 