VoidSurf Profile picture
7 Oct, 20 tweets, 7 min read
We can't understand the present if we don't understand the past.

This thread will show you the history of death and disease outbreaks in Sweden. All source data is available for download from the Statistics Sweden website. Here are monthly DPM since 1851:
These are the values I used for 2020. I've tried to be conservative - where there's a chance deaths might be understated, I've methodically adjusted them upwards. Also - I started this project over a month ago, so newer data is now available. These are still good though.
Winter has the most months with high mortality due to disease and harsh conditions. So, to make annual figures, I'll be summing from July to June whenever possible. This will give a good idea of what a season looked like.
The yellow line in the first chart is a LOESS regression. It's like a moving average, but more robust and tracks better with the center. It measures our changing expectation of mortality. Here are only the positive differences between the monthly DPM and the yellow line:
Now, I'll sum these positive residuals from Jul-Jun for each 12 month period. This shows us how bad a season has been after removing the effect of the changing level of expected mortality.
Here are the overall deaths/million for each Jul-Jun period. Notice how variance plummeted in the 1950s with the introduction of advanced healthcare. Even with Covid-19, Sweden's absolute mortality rates are at all-time historical lows.
I've detrended the previous chart, showing just the positive differences. The meaning is very similar to that of the chart in tweet 5. I'm removing the effect of changing baseline mortality to only look at the spikes:
Sweden records monthly deaths since 1851, but annual data goes back to 1749. It shows just how brutal human history has been. In 1773, a staggering 1 in 19 Swedes died.

I also show what 2020 would look like if Sweden experienced a second Covid wave, doubling Jan-Jun deaths.
This is the previous chart, but detrended. We can clearly see how the advancement of civilization and technology has led to better conditions and drastically fewer events of extreme death.
Just adjusting for population doesn't give a full picture. Age adjustment shows what each year would have looked like if it happened to a population with the same age profile. I use the population of 2000. Sadly, I can't go back further than 1968.
If you want to see what earlier years might have looked like for Sweden, these are the age-adjusted death rates for the USA from 1900-2019.
Monthly figures aren't available by age group for the whole period, so I made an estimate by weighting by the proportion of all deaths in each month, per year. It's not perfect, but it should be close.
I've detrended the previous chart here. Clearly, Sweden has had several jumps that were significantly more severe than April 2020 in the past 52 years.
Finally, I'll sum the monthly age-adjusted death rates for each Jul-Jun period. We can now see 2019-2020 in a historical context. Sweden has gone back 2 years to the level seen in the 2017-2018 season.
I hope this information gives people some perspective on what we're dealing with today. The conditions our ancestors dealt with daily were much harsher than even the worst of Covid-19.
Sweden has taken a measured approach to the pandemic. It hasn't experienced an unprecedented disaster, even in the context of the past 50 years. While all covid deaths are tragic, the consequences of extreme measures are also tragic. It is best to find a balance.
As an extra, I've also played around with fitting probability distributions to some of these numbers. It's challenging since most aren't IID. Here's one I found. The Wakeby distribution pops up very frequently in relation to mortality.
Check out a similar analysis of the USA here:
I also want to thank @HaraldofW and @boriquagato for your great work. Your charts are what inspired me to do this analysis. I hope you find it informative and keep up the great work.
People have pointed out there's a typo in some of the captions. The caption should read 1857, not 1957. Apologies. I don't know if I can replace the images. If anyone knows how to do that, let me know.

• • •

Missing some Tweet in this thread? You can try to force a refresh

Keep Current with VoidSurf

VoidSurf Profile picture

Stay in touch and get notified when new unrolls are available from this author!

Read all threads

This Thread may be Removed Anytime!


Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @VoidSurf1

7 Oct
We can't understand the present if we don't understand the past.

In this thread, you'll see what the past 120 years looked like for the USA. The source data is all available on the CDC website, though it was a real pain to aggregate. This chart shows monthly DPM since 1900:
These are the values I used for 2020. They were implied from weekly counts retrieved on Oct 3, so the vast majority of deaths should be registered for Jan-Aug. Future adjustments shouldn't impact the overall picture.
The red line you saw is called a LOESS regression. It's like a moving average, but more robust and tracks better with the center. It measures our changing expectation of mortality. Here we see only the positive differences (residuals) between the DPM and the red line:
Read 13 tweets

Did Thread Reader help you today?

Support us! We are indie developers!

This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Too expensive? Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal Become our Patreon

Thank you for your support!

Follow Us on Twitter!