Guest Blogger @HaraldofW – All you ever wanted to know about Corona Sweden

Introduction

This is a great posting that’s going viral on Twitter as we speak, about the Swedish Corona experience, written by a Guest Blogger, twitter handle @HaraldofW a person who’s deep insights and great analytic skills I’ve come to appreciate over the past few weeks.  If you are at all interested in Corona, and you are on Twitter, I’d strongly recommend you to follow @HaraldofW.

(A note on the stylistics & structure of the post : this blog post has been concatenated together basically by cut & paste of the 25 (!) individual Twitter posts comprising the entire message, so stylistically, it might appear a bit weird, due to the Blog Master’s bad cut & paste skills… 🙂

Over to @Harald and his post:

Corona and All Cause Deaths in Sweden

Third comprehensive thread compiling my graphs, but where did it begin? Does anyone remember when stores were all out of toilet paper & #covid19 panic started to build? At that point I wanted to put things into perspective quite like @InProportion2 has done with UK.

image26

Using IHME mean value. Model used to illustrate difference between reality and projection i.e. model is not representative of the best possible model produced
Data Source: https://www.icuregswe.org/data–resultat/covid-19-i-svensk-intensivvard/ and Downloaded IHME projection for Sweden 2nd May from https://covid19.healthdata.org/sweden

But first a few words on Sweden

We are a country with most of its inhabitants living in Stockholm and south of Stockholm. The north is not as densely populated as Stockholm has 5200 inhabitants per square km, Gothenburg at 1300 inhabitants per square km and Malmö at 2200 individuals per square km.

image9

Sweden’s population grew with ca 16 percent since year 2000. Ca 20 percent of people living in Sweden are born abroad. Largest immigrant population is 1) Syrians and 2) Iraqi and 3) Finnish people. In Stockholm, Gothenburg and Malmö some 20-25 percent are born abroad. Google ‘people that travel the most’ and you will find Sweden right at the top of most lists. The idea of who we are and the nature of our reality might not always be the same. Reality is more complex. Now back to #covid19 and my comparison of Sweden to… Sweden.

As shown, reality proved different than projections, but it was partly those projections that informed our response. Early on I stumbled upon @JenniferWegerup which provided data on Italy regarding for example age distribution which – as can be seen – is similar to the Swedish distribution of covid19 deaths.

Wegerups presentations showed that deaths primarily occurred in elderly and people with comorbidities. Now this should tell us something in contrast to the narrative of everyone at great risk.

image25

Year 2020 preliminary with figures likely to be somewhat adjusted upwards Data Source: https://www.scb.se/om-scb/nyheter-och-pressmeddelanden/scb-publicerar-preliminar-statistik-over-doda-i-sverige/

We can also look at week 16-26 in order to see the distribution even more pronounced since covid19 first made its mark in March of 2020.

image13

Year 2020 preliminary with figures likely to be somewhat adjusted upwards Data Source: https://www.scb.se/om-scb/nyheter-och-pressmeddelanden/scb-publicerar-preliminar-statistik-over-doda-i-sverige/

with ca 89 percent of covid19 deaths in age group above 70 and more than twice as many above 90 years old as under 70 years old we see that #covid19 is greater risk to elderly than the working & school population.

But further we see that within the age group 70+ almost half of individuals deceased have been care home residents and additional ca 20 percent of age group 70+ receiving home care service. This means that within a small percentage of the population we have more than half of #covid19 deaths.

According to a study, half of the individuals moving in to care homes only lived there two years before dying. It’s already a quite fragile population. But perhaps also individuals who need human connection and not distancing during this pandemic, next chart shows age 70+ and distribution between population in care home and non care home individuals.

image18

Data Source: https://www.socialstyrelsen.se/statistik-och-data/statistik/statistik-om-covid-19/statistik-over-antal-avlidna-i-covid-19/ and https://www.socialstyrelsen.se/globalassets/sharepoint-dokument/artikelkatalog/statistik/2019-2-20.pdf Aproximately what percentage of age group 70+ lives in care homes and of those deceased with covid19, how many lived in a care home as of July 1

As mentioned in the beginning, I wanted to gain some perspective. For this reason I compared deaths all causes in different ways. As can be seen in this chart trend is slightly downwards but year 2019 was very low. Actually, 2019 had fewest deaths since 1977.

I asked myself, in the longer perspective do we see what could be seen in 1919? Or how would it compare with deaths all causes during Hong Kong flu or Asian flu?

image8

Data: http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101__BE0101G/ManadFoddDod/ Comparing deaths all causes around the time of the spanis flu and the russian flu in order to get an idea of the impact these periods had on Sweden.

Events around Asian Flu and Hong Kong Flu are events that my mother don’t remember in any apocalyptic manner.

image12

Data: http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101__BE0101G/ManadFoddDod/ Adjusting number of deaths in order to account for population growth

But that was some time ago, so what does it look like in more recent history. Comparing period 1990-2020 deaths all causes we first look at October through June.

image24

Data: http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101__BE0101G/ManadFoddDod/ and https://www.scb.se/om-scb/nyheter-och-pressmeddelanden/scb-publicerar-preliminar-statistik-over-doda-i-sverige/ Population per 100 000, eg deaths 1990 compared with population end of 1989 according to SCB. Comparing from october since October normally marks the start of flu season (week 40-20), adding May and June

Next we look at comparison January through June. We should also keep in mind what message was broadcasted in February, so as to remember the context.

In this graph we look at comparison january through June but stratified using age (only week 1-22) but message the same.

image14

UPDATE 26/6 Adjusting figures to match population growth, eg ‘x’ percent of age group ‘y’ died Y2000, how many would that have been Y2020 given the population growth of each sub group. Data Source: Deaths – https://ec.europa.eu/eurostat/web/covid-19/data and Population – http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101/

When looking at March through June and individual months we we that April is one of the months with highest mortality since 1990 but still not unprecedented. Although one could argue that we have as of late been accustomed to a different reality.

image1

Data Source:http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101__BE0101G/ManadFoddDod/ May 2020 included, however not part of the months with highest mortality rates in Sweden. 2020 figures preliminary. Version update 6/7, adjusting eg deaths 2011 to pop. end of 2010.

We can also experiment with expected deaths based on average and trend. First looking at average it would look like this.

image17

Based on 20 years of data where ‘x’ percent of age group ‘y’ dies each year an average number of deaths per age group have been calculated showing estimated deaths 2020. Further, adding up deaths per month between year 2000-2019 in order to distribute expected deaths per month as they normally arent distributed equally over the year. Data Source: http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101/ and https://www.scb.se/om-scb/nyheter-och-pressmeddelanden/scb-publicerar-preliminar-statistik-over-doda-i-sverige/

image21

Based on 20 years of data where ‘x’ percent of age group ‘y’ dies each year an average number of deaths per age group have been calculated showing estimated deaths 2020. Further, adding up deaths per month between year 2000-2019 in order to distribute expected deaths per month as they normally arent distributed equally over the year. Data Source: http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101/ and https://www.scb.se/om-scb/nyheter-och-pressmeddelanden/scb-publicerar-preliminar-statistik-over-doda-i-sverige/

And then looking at trend it would look like this, hence not a lot above average but breaking with downward trend.

image7

Assumption: Based on 20 years of data where ‘x’ percent of age group ‘y’ dies each year a trend – using Forecast – have been estimated showing expected deaths year 2020. Eg. 25% of age group 90+ dies in year 2000 but 23% in year 2019, what does that annual trend look like? Further, adding up deaths per month between year 2000-2019 in order to distribute expected deaths per month as they normally arent distributed equally over the year. Data Source: http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101/ and https://www.scb.se/om-scb/nyheter-och-pressmeddelanden/scb-publicerar-preliminar-statistik-over-doda-i-sverige/

image27

Assumption: Based on 20 years of data where ‘x’ percent of age group ‘y’ dies each year a trend – using Forecast – have been estimated showing expected deaths year 2020. Eg. 25% of age group 90+ dies in year 2000 but 23% in year 2019, what does that annual trend look like? Further, adding up deaths per month between year 2000-2019 in order to distribute expected deaths per month as they normally arent distributed equally over the year. Data Source: http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START__BE__BE0101/ and https://www.scb.se/om-scb/nyheter-och-pressmeddelanden/scb-publicerar-preliminar-statistik-over-doda-i-sverige/

What has been apparent is how the impact have had different severity in different regions, here comparing, Stockholm, Skåne & Västra Götaland. And we see that Stockholm has ca 43 percent of all #covid19 deaths. Skåne, neighbouring with Denmark, has few deaths.

image2

Data Source: https://www.folkhalsomyndigheten.se/smittskydd-beredskap/utbrott/aktuella-utbrott/covid-19/bekraftade-fall-i-sverige/ Stockholm and Västra Götaland are the two regions worst impacted by Covid19 (updated 6/7)

This does not mean other regions have no spread of virus, seroprevalence shows they had spread as well, although less than Stockholm but Skåne has seen no higher mortality than previous year, Stockholm has higher mortality as can be seen in next chart.

But we can also see that #covid19 deaths are declining in all three compared regions.

Back to the whole country, when comparing excess deaths as the difference to prior 5 year average last ca 100 years would look like this (all of Sweden), one has to understand the numbers and context behind the charts.

As can be seen in next chart, peak covid19 deaths at week 15 and now declining back to normal levels (slightly above) regarding #covid19 & deaths all causes.

And whatever we collectively do in the future, let’s not go strict lockdown as some countries have done putting the whole world through this ordeal at high costs, not only purely economical costs.

End of Harald’s post.

Many thanks Harald for contributing this very insightful and important post to the blog.  Those that still believe that Covid-19 is the Bubolic Plague of our times, can hopefully benefit from even more information about Corona in Sweden, here.

 

 

About swdevperestroika

High tech industry veteran, avid hacker reluctantly transformed to mgmt consultant.
This entry was posted in Data Analytics, Epidemics and tagged , , . Bookmark the permalink.

2 Responses to Guest Blogger @HaraldofW – All you ever wanted to know about Corona Sweden

  1. Tamara Lombardo says:

    Thank you. The graph concerning the yearly month with highest death rates has a fault – it contains both April and May 2020. Furthermore, several other sources place April 2020 as less deadly only than the deadliest months of 93, 96 and 2000, and not many other years as presented in Harald’s graph: https://medium.com/@emanuelkarlsten/more-swedes-died-in-one-month-1993-and-2000-compared-to-april-2020-why-25b762d03a1f

    • Harald says:

      Dear Tamara,
      Thank you for your comment. I will leave a short reply.
      1) The reason May is included in the graph is briefly explained below graph. It is not because it is part of months with highest mortality.
      2) In my graph, deaths per months are adjusted for population growth. Further the result is replicated by the blog owner with similar result. When not considering population growth fewer months have had higher mortality (As Karlsten demonstrates) but nonetheless there have been months with higher mortality.
      Regards

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s