search

CSSEGISandData / COVID-19

Novel Coronavirus (COVID-19) Cases, provided by JHU CSSE
https://systems.jhu.edu/research/public-health/ncov/
29.13k stars 18.43k forks source link

Sweden Deaths #3594

Open CSSEGISandData opened 3 years ago

CSSEGISandData commented 3 years ago

Hello all,

Apologies for the several issues raised about this point that have not yet received a response. Today, we are overwriting our data for Sweden deaths from 04-04-2020 to 20-01-2021. This to ensure our data consistently represents "deaths by day of report". On their dashboard, the only time series file available for deaths is deaths by date of report. However, this data is missing the most recent days and thus can't be used for our purposes, which is the reason we use "deaths by day of report" in the first instance. As documented in our README, on August 27 we had to overwrite our Sweden data due to falsely identified cases and deaths and the only data accessible from the Swedish Ministry of Health was "deaths by date of death". We have since been able to access a repository of total deaths by day (that only extends to April 4, 2020), and are using this data to overwrite our deaths data such that it remains consistent.

An externality from this approach is that there will be a decrease in deaths in Sweden from April 3 to April 4.

davidjayjackson commented 3 years ago

The challenge Sweden "official" numbers don't have any gaps in the data for the Saturday - Monday. While your data and Our World In Data has the Sunday - Monday gap and a spike for Tuesday numbers....

David

On 1/21/21, CSSEGISandData notifications@github.com wrote:

Hello all,

Apologies for the several issues raised about this point that have not yet received a response. Today, we are overwriting our data for Sweden deaths from 04-04-2020 to 20-01-2021. This to ensure our data consistently represents "deaths by day of report". On their dashboard, the only time series file available for deaths is deaths by date of report. However, this data is missing the most recent days and thus can't be used for our purposes, which is the reason we use "deaths by day of report" in the first instance. As documented in our README, on August 27 we had to overwrite our Sweden data due to falsely identified cases and deaths and the only data accessible from the Swedish Ministry of Health was "deaths by date of death". We have since been able to access a repository of total deaths by day (that only extends to April 4, 2020), and are using this data to overwrite our deaths data such that it remains consistent.

An externality from this approach is that there will be a decrease in deaths in Sweden from April 3 to April 4.

-- You are receiving this because you are subscribed to this thread. Reply to this email directly or view it on GitHub: https://github.com/CSSEGISandData/COVID-19/issues/3594

love-lace commented 3 years ago

Thank you for updating this to be more consistent. Although can I ask, where is this new data repository you are using now coming from? The reason being I took a touble take at the 8th January which says it has 706 deaths reported. Being a Swede, I would remember the news outcry at that seeing as are largest reporting death we have had was 350 deaths at the 14th January and that was a big shock.

I assume this is because the data reported from week 1 on Tueday 5th January (249) and Thursday 7th January (276) has been combined with the numbers on Friday the 8th January? I see that they are reported as 0 deaths from the new repo I assume.

davidjayjackson commented 3 years ago

jhu_covid-19 sweden_offical

Ok, I just "git pull" on this repo and I'm still see the Saturday - Monday gap? With a spike on Tuesday? The attached plots show both Official daily Cases and your Daily cases. I did sent a email to the Swedish agency in charge of Covid response.

David Jackso

On 1/23/21, love-lace notifications@github.com wrote:

Thank you for updating this to be more consistent. Although can I ask, where is this new data repository you are using now coming from? The reason being I took a touble take at the 8th January which says it has 706 deaths reported. Being a Swede, I would remember the news outcry at that seeing as are largest reporting death we have had was 350 deaths at the 14th January and that was a big shock.

I assume this is because the data reported from week 1 on Tueday 5th January (249) and Thursday 7th January (276) has been combined with the numbers on Friday the 8th January? I see that they are reported as 0 deaths.

-- You are receiving this because you commented. Reply to this email directly or view it on GitHub: https://github.com/CSSEGISandData/COVID-19/issues/3594#issuecomment-765950005

davidjayjackson commented 3 years ago

Here is link to a video report comparing John Hopkins Covid19 daily cases, to official Swedish daily Covid19 cases.

https://youtu.be/kbI3Urtm8kQ

OneBobone commented 3 years ago

Systematic underreporting Covid data for deaths in Sweden

Thank you for a great site (Worldinfo)!

For various reasons, Sweden is studied more closely than other countries in the pandemic. For this reason it is of great importance that you report the correct numbers of deaths. Or rather, you must display and update the official numbers correctly.

Currently, you are under reporting deaths for systematic reasons.

A choice has to be made:

  1. either you take the reported totals, which come 3-4 times per week and always groups several calendar days of deaths, and average them over a 7-day period. (The average and reporting lag will be quite consistent.)
  2. you use the exact and periodized numbers reported by the gvt. They indicate the precise day of death. This is of course the best, but requires a continuous backtracking of updates up to a month back. The main problem is severe underreporting of recent deaths giving the illusion of an improving trend. Not good.

As you are doing now, your most recent data only reflects the first batch of reported deaths for a specific day. That under represents the truth by up to 50-60% at times.

https://ourworldindata.org/grapher/daily-covid-deaths-per-million-7-day-average?time=2020-11-01..&country=AUT+CAN+DNK+FRA+DEU+ITA+NOR+KOR+ESP+SWE+GBR+USA Please compare your 7 day average with e.g. the data displayed by the Swedish national television site. It has both an average of the reported deaths and the precise, and updated daily data.

https://www.svt.se/datajournalistik/har-sprider-sig-coronaviruset/ (It should be say to run G Translate on the site).

If you have any questions, I would appreciate your reply. Please don’t let incorrect data be spread from your site. Some would call it “fake news”. It is a threat to democracy and lives.

Many thanks, Bob Master of Science in Health Care